Synthesis, Structure–Activity Relationships, Radiofluorination, and Biological Evaluation of [18F]RM365, a Novel Radioligand for Imaging the Human Cannabinoid Receptor Type 2 (CB2R) in the Brain with PET

The development of cannabinoid receptor type 2 (CB2R) PET radioligands has been intensively explored due to the pronounced CB2R upregulation under various pathological conditions. Herein, we report on the synthesis of a series of CB2R affine fluorinated indole-2-carboxamide ligands. Compound RM365 was selected for PET radiotracer development due to its high CB2R affinity (Ki = 2.1 nM) and selectivity over CB1R (factor > 300). Preliminary in vitro evaluation of [18F]RM365 indicated species differences in the binding to CB2R (KD of 2.32 nM for the hCB2R vs KD > 10,000 nM for the rCB2R). Metabolism studies in mice revealed a high in vivo stability of [18F]RM365. PET imaging in a rat model of local hCB2R(D80N) overexpression in the brain demonstrates the ability of [18F]RM365 to reach and selectively label the hCB2R(D80N) with a high signal-to-background ratio. Thus, [18F]RM365 is a very promising PET radioligand for the imaging of upregulated hCB2R expression under pathological conditions.


■ INTRODUCTION
−4 CBRs are primarily located in the central and peripheral nervous systems and are involved in a variety of physiological processes, including pain, appetite, memory, and immune function. 3,5,6There are two main types of cannabinoid receptors: CB1R and CB2R.−13 The most well-known cannabinoid is (−)-Δ 9 trans-tetrahydrocannabinol (THC), which is responsible for the psychoactive effects of cannabis. 1,14Up to date, more than 200 cannabinoids from the cannabis plant were reported, including cannabidiol (CBD), which does not have psychoactive effects but has been shown to have therapeutic potential for a variety of conditions, such as anxiety, pain, and inflammation. 15,16hile psychoactive effects are transmitted via the activation of CB1R, a selective targeting of CB2R was proposed as a safe and promising therapeutic approach for a wide range of medical applications, including cancer, 17 neurological disorders, and metabolic disorders. 18When activated, CB2R leads to a decrease in cAMP production by inhibiting the activity of adenylyl cyclase, resulting in a decrease in the downstream effects of cAMP, such as the activation of protein kinase A (PKA) and the inhibition of the formation of cAMP-dependent protein kinases. 19The impact of the intracellular level of the secondary messenger cAMP on a great variety of cellular processes, 20 and the location of CB2R in the immune system of both the brain and the periphery, as well as in microglia and endothelial cells, makes the receptor an interesting pharmacological target for the treatment of pain syndromes, neuroinflammation, and neurodegenerative processes. 21−24 While early studies failed to detect CB2R in brain tissue, 25 the presence and localization of CB2R in the brain were demonstrated by, e.g., RT-PCR and immunohistochemical analysis. 26Given the low level of expression of CB2R along with the high level of CB1R under physiological conditions, for the neuroimaging of CB2R with PET, a radioligand binding to CB2R with high affinity with K D in low nanomolar range and high selectivity versus CB1R is needed. 22 number of PET radioligands were developed for CB2R to date and excellently summed-up in several recent reviews, 22,23,27 with the most recent CB2R radioligands depicted in Figure 1.Compound [ 18 F]DM102 was reported by Modemann and coworkers and proved by autoradiography with spleen tissue to bind to CB2R. 28The most recently 18 F-labeled CB2R ligand reported by the group of Ametamey is [ 18 F]RoSMA-18-d 6 , 29,30 which besides the subnanomolar CB2R affinity and selectivity showed excellent imaging properties as demonstrated by in vitro autoradiography with rat spleen and spinal cord tissue from ALS patients as well as in vivo imaging in rat with PET.Gundel and co-workers reported [ 18 F]LU13, a radioligand with high CB2R  binding affinity and selectivity. 31Despite its inability to label CB2R in the spleen by autoradiography in vitro, compound [ 18 F]LU13 is suitable to visualize intracranially overexpressed human CB2R(D80N) in a rat model. 32Recently, Ueberham and co-workers reported [ 18 F]LUZ5-d 8 , the first carborane-based CB2R radioligand setting the stage toward theranostic application. 33Despite the large number of CB2R-targeting radioligands reported to date, little progress has been made in imaging this receptor in pathological conditions, which may be related either to the insufficient affinity of the radioligands or the too low expression of CB2R to be detected by PET.
Besides quinolinones, 34−45 naphthyridones, 46−48 and thiazoles, 49−54 indole 55−65 is one of the most widely investigated scaffolds in the medicinal chemistry of CB2R-targeted small molecules.As the most broadly used substitution pattern, the Nalkyl-3-carbonyl indole plays a central role; however, the cannabinoid receptor ligands based on this scaffold most often suffer from low selectivity against CB1R as, e.g., JWH-018 65 and the more recently reported compound GBD-003 59 (Figure 2).On the other hand, a few structurally related ligands lacking the N-alkyl residue and bearing a 2-carboxamide instead possess high CB2R affinity and selectivity (e.g., 6, Figure 2). 60In our continuous efforts to develop an 18 F-labeled radioligand for the imaging of CB2R in the brain with PET, we carried out a medicinal chemistry study on the structure of compound 6 (Figure 2) 60 due to its low nanomolar affinity for CB2R (CB2R K i = 0.26 nM 60 ).For efficient synthesis and easy access to a library of fluorinated derivatives and a precursor compound for radiofluorination based on the structure of 6, an enantioselective synthesis was envisaged.The medicinal chemistry was carried out aiming at a fluorinated analogue of 6 with (i) high affinity toward CB2R, (ii) high selectivity toward CB1R, and (iii) an easily accessible position for aliphatic radiofluorination.The best candidate according to affinity and selectivity has been radiofluorinated and subsequently investigated by PET regarding the ability to image CB2R in the brain of a wellestablished and characterized rat model of local hCB2R overexpression. 32,66RESULTS AND DISCUSSION Chemistry.The synthesis of compound (±)-6 was performed by a modified 4-step sequence starting from the commercially available 5-bromo-1-indanone (8), which is converted into the racemic amine 11 within three steps and 37% overall yield.In the first step, 5-bromo-1-indanone (8) was selectively double-methylated with MeI at position 2 in deprotonative (NaH) reaction conditions (Scheme 1).For the aromatic nitrilation, the Rosenmund−von Braun reaction 67−69 was then employed after which the molecule was reductively aminated with ammonium acetate in the presence of sodium cyanoborhydride.In the last step, amine 11 was coupled with the commercially available acid 12 in the presence of the BOP reagent.Previously, 60 the enantiomeric separation of amine 11 was performed by subcritical fluid chromatography using a chiralpack AD column, and only for (R)-6 (optical rotation not mentioned), the CB2R affinity was reported. 60In the present study, we synthesized (±)-6 and resolved the racemic mixture at the last step by chiral HPLC.This strategy allowed us to perform the organic synthesis in an easy-to-handle scale with final access to both enantiomers which could be investigated regarding their binding affinity toward CB1R and CB2R.The quantitative separation of the enantiomers was achieved by HPLC using a chiral column.A total of 3 mg of each enantiomer was separated, and the enantiopurity was determined by analytical chiral HPLC coupled with a chiral detector measuring the optical rotation (Figure S1 in the Supporting Information).Thus, we can assign the R-enantiomer to a minus rotation and the S-enantiomer to a plus rotation.Moreover, both enantiomers were investigated Scheme 1. Synthesis of (±)-6 and Chiral HPLC Resolution of (−)-6 and (+)-6 using another chiral detector, which is based on circular dichroism, a feature allowing optically active compounds to differently absorb right and left circularly polarized light.This difference is recorded and the value is referred to as ellipticity, often given in mdeg (millidegrees).As the ellipticity value depends on the absorbance wavelength and thus on the chromophores of a compound, the resulting CD spectrum is a typical characteristic for a chiral compound and the two enantiomers are usually giving mirror image spectra.The selection of the most suitable wavelength for CD monitoring of (+)-6 and (−)-6 was the first procedure to be performed, and the respective CD spectra were recorded with chiral HPLC in stopped-flow mode.As shown in Figure 3A, the enantiomers have two strong CD bands in the ranges of 230−250 and 270− 290 nm.At ∼240 nm, a positive CD band is related to (−)-6 (Rconfiguration) and a negative CD band to (+)-6 (Sconfiguration).In contrast, at ∼285 nm, a positive CD band is related to (+)-6 (S-configuration) and a negative CD band to (−)-6 (R-configuration).As a consequence, the amplitudes in the respective chromatograms of both enantiomers recorded with the CD detector are dependent on the wavelength used, as can be seen in Figure 3B.
The investigation of the binding assays revealed that only (−)-6 (RM365) possesses CB2R affinity with K i (CB2R) = 2.1 nM (0.24 nM reported 60 ) and (+)-6 proved to be inactive.Compound (−)-6 also possesses high CB2R selectivity with a K i (CB1R) of 827 nM.To further investigate the potential of this scaffold to bind CB2R by developing fluorinated derivatives of (−)-6, an enantioselective synthesis was needed.Several highly efficient methods for enantioselective reductive amination were reported to date. 70In our efforts to synthesize enantiopure RM365 ((−)-6), the method developed by Ellman 71 using sulfinamide 13 as chiral auxiliary was selected.
By reacting 10 with (R)-tert-butanesulfinamide (13), compound 14 was formed as a single enantiomer.Although examples in the literature show that the NaBH 4 -mediated reduction of such tert-butanesulfinyl imines of type 14 can lead to the formation of sulfonamides of type 15 in high  diastereomeric excess, in our hands, a 1:1 mixture of diastereomers 15 and 16 was formed.Attempts to increase the diastereomeric excess by strictly controlling the reaction temperature (ranging from −40 to 0 °C) did not improve the diastereoselectivity of the reaction.However, compounds 15 and 16 were readily separated by flash chromatography on silica.TFA-mediated sulfonamide cleavage leads to compound 17, which was coupled with the carboxylic acid 12 to give (−)-6 (RM365, Scheme 2).With large amounts of enantiomeric pure amine 17 in our hands, a series of fluorinated derivatives of (−)-6 was developed by coupling amine 17 with various fluorinated bioisosters of 12 (Scheme 3).A further series of derivatives was synthesized by keeping 5-fluoro-N-methyl-1H-indole-2-carboxamide intact and modifying the dihydroindene subunit (Scheme 4).First, to evaluate the need for a 5-indole substitution pattern, the regioisomers 19A−C of RM365 were synthesized.Compounds bearing 5-fluoroethoxy and 5fluoropropoxy residues (19D-E, 21A-B) were synthesized bearing an aliphatic fluorine.Compounds 19F−H were synthesized to evaluate the importance of the pyrrole ring for CB2R binding.Compounds 21A−D were synthesized as racemic mixtures by the same procedure (Scheme 4).Compound 21C was obtained by the hydrolysis of a nitrile group of 6.Furthermore, the dihydroindene subunit or RM365 was replaced by fenchone (21D) and adamantine (21E)

Journal of Medicinal Chemistry
residues, as these are frequently used scaffolds in the medicinal chemistry of CB2R ligands. 31,72 Vitro Binding Assay.The binding affinity of the herein described compounds toward the human CB2R was determined by a competitive radioligand binding assay according to the established in-house protocol using cell homogenates obtained from CHO cells stably transfected with hCB2R (hCB2R-CHO). 53,73For compounds binding with high affinity to hCB2R, the affinity toward CB1R was determined as well by radioligand displacement studies performed either with homogenates obtained from the mouse brain or hCB1R-CHO cells.
According to the results summarized in Table 1, all derivatives with an (R)-configuration at the carboxamide-indenyl subunit bind with low nanomolar to nanomolar affinity toward CB2R and with micromolar affinity toward CB1R.As aforementioned, compound (−)-6 (Scheme 1, Table 1) obtained by chiral HPLC separation of (±)-6 is characterized by a binding affinity of 2.1 nM, while for the opposite enantiomer (+)-6, no binding to the CB2R has been measured in the investigated concentration range.The enantioselectively synthesized compound RM356 (Scheme 2) presented a binding affinity in the same range.Compound's RM365 regioisomers, 19A−C, showed low nanomolar CB2R affinity.The introduction of fluoroethoxy and fluoropropoxy residues at the 5-indole position slightly decreased the affinity in comparison to (R)-6 (compounds 19D and 19E, Table 1).The introduction of a methyl group at the 3indole position also had a deleterious effect on the binding affinity (19F).The substitution of the indole subunit with benzimidazole in 19G or benzothiophene in 19H gave compounds with a high CB2R affinity, comparable to that of the starting compound RM365.Compounds 21A and 21B were obtained by substituting the nitrile function at the indane subunit with fluoroethoxy and fluoropropoxy, respectively, and showed a slightly reduced affinity toward CB2R.However, the substitution of the indane subunit by fenchone (compound 21D) and adamantane (compound 21E) resulted in a loss of CB2R binding.
Radiochemistry.For the radiosynthesis of [ 18 F]RM365, a precursor compound bearing a leaving group at the indole 5position had to be designed.Attempts to radiosynthesize [ 18 F]RM365 starting from the 5-nitroindole precursor 22 by using either [ 18 F]K 222 /K 2 CO 3 or [ 18 F]TBAF under both thermal and MW irradiation conditions failed.The low reactivity of 22 toward S N Ar can be explained by the electron-rich nature of the benzyl subunit of the indole.The S N Ar radiofluorination is generally challenging and applicable only for electron-deficient substrates.The introduction of 18 F at aromatic positions which are not activated for S N Ar usually implies multistep radiosynthesis or chemical modifications of the target compounds.Recent advances using transitional metal for the introduction of 18F are however not dependent on the electron density of the aromatic ring, providing good radiolabeling yields for a large variety of substrates. 74−77 The boronic acid pinacol ester precursor 24 was synthesized in enantiomeric pure form in one step from the corresponding bromo-derivative 23 via the Miyaura borylation reaction 78 in high yield (85%) as shown in Scheme 5. To optimize the yield of the radiofluorination reaction, several reaction conditions were tested by varying (a) the amount of precursor (1−5 mg), (b) the Cu reagent− precursor molar ratio ( 25), (c) the order of addition of reagents, (d) the atmosphere (air or argon), and (e) the temperature (110−140 °C) using the [ 18 F]TBAF system in DMA and t-BuOH as solvents.While the atmosphere used seemed to play no or only a marginal role, the stepwise addition of the Cu reagent to [ 18 F]TBAF, followed by the addition of the precursor seemed to greatly improve the formation efficiency of [ 18 F]RM365.The time span between the addition of the Cu reagent and the precursor did not influence the labeling yield and a 10 s interval was found to be sufficient.Furthermore, by using an optimal (3 mg) amount of precursor at 120 °C, the critical factor determining the rate of the reaction proves to be the Cu reagent−precursor molar ratio (25/24) with the best result obtained by using a 25/24 ratio of 2 and 3, as shown in Figure 4. To keep the Cu reagent load in the system as low as possible, the following experiments were carried out by using a molar ratio 25/24 of 2.
The most efficient protocol for the manual radiosynthesis of [ 18 F]RM365 was transferred to an automated synthesis module (Elysia-Raytest radiosynthesizer).For the biological evaluation, [ 18 F]RM365 was obtained with a moderate radiochemical yield of about 5% and high radiochemical purities (>99%), molar activity of about 80 GBq/μmol in a total synthesis time of 60 min (n = 6) starting with activities ranging from 8 to 13 GBq.A comparable decrease of the radiolabeling efficiency for the automated compared to the manual radiosynthesis for this type of Cu-mediated radiofluorination has been previously reported and extensively discussed in the literature. 79RP and chiral HPLC analysis of the final product coeluted with the corresponding reference compound RM365 confirmed the identity and enantiopurity of the radioligand (Figure 5B).A log D 7.4 of 3.0 ± 0.1 was determined by the shake flask method. 80,81n Vitro Evaluation of [ 18 F]RM365.To complement the indirect determination of the affinity of RM365 toward the CB2R in vitro, we performed a direct binding study with the radiofluorinated ligand using preparations of CHO cells stably transfected with the cloned human receptor and one spleen from a female SPRD rat expressing high levels of the receptor in the leukocytes. 12The results of the homologous competition study shown in Figure 6 confirm the good affinity of [ 18 F]RM365 toward the human CB2R with a K D value of about 2 nM.By  contrast, with rat spleen homogenates, no detectable interaction of [ 18 F]RM365 with a specific binding site could be observed.Furthermore, the measurement signal obtained from the incubation of [ 18 F]RM365 with the homogenate of rat spleen was in the range of the values obtained for the nonspecific binding of the radioligand toward the preparation of CB2Rtransfected CHO cells.We conclude that RM365 binds the rat CB2R with lower affinity than the human CB2R but expect no significant interactions of the radioligand with nonspecific binding sites. 82Because of the available rat model to assess the potential of [ 18 F]RM365 to cross the blood−brain barrier and to specifically label human CB2R�the hCB2R(D80N) model with high expression of human CB2R in the brain 31,32,66,83 �we intended to confirm the initial findings from the rat homogenate by an autoradiographic study.The homogeneous and nondisplaceable distribution pattern of [ 18 F]RM365 in cryosections of spleens from two male F433 rats (Figure S2) confirms the absence of a specific interaction of this radioligand with rat CB2R under physiological conditions.Molecular Docking.The unexpectedly high human−rat species differences in the binding affinity of [ 18 F]RM365 prompted us to investigate the interaction mode of the ligand in the CB2R binding pocket by molecular modeling.In silico modeling of the interactions between (−)-6 and hCB2R and between (−)-6 and rat CB2R was done by docking of the optimized (at B3LYP D3BJ/def2-TZVP level of theory) structures of RM365 in the binding pockets of the hCB2R and rCB2R, respectively.The crystallographic structure of hCB2R is available from PDB under the code 3ZTY. 84The structure of rCB2R was generated by folding the primary sequence in I-TASSER 85 and optimizing the structure of the best model with the molecular dynamic (see details in the SI).
The calculated binding energies (ΔG) demonstrated that the best docked poses of (−)-6 have a higher affinity to hCB2R (−10.45 kcal mol −1 ) than to rCB2R (−6.64 kcal mol −1 ).The computed inhibitory constants (EK i ) of (−)-6 calculated based on the equation: ΔG = RTln(K i ) made up of 27 nM for hCB2R and 14 μM for rCB2R, which are in agreement with the experimental data.The explanation of the observed difference could be given based on the binding mode between (−)-6 and the amino acid residues of the ligand-binding domains (LBDs, Figure S6).
In contrast to this, the best docked pose of (−)-6 forms a lower number of interactions with F87, W258, L262, M265, C288, and A261 in the LBD of rCB2R.These noncovalent bonds are dominated by the pi−alkyl interactions.The binding energy indicates the lower stabilizing influence of these contacts on the docked compound in rCB2R compared to multiple pi− pi−T-shaped interactions formed between (−)-6 and hCB2R.Interestingly, in both receptor subtypes, the best docked poses interact with the toggle switch W258, which was reported as the crucial residue for the activity of CB2R 86 (Figure 7).
We also investigated the correlation between the binding energies of the best ranked position of the S-isomer and the experimental affinity.Docking suggests a decrease of the binding energy of (+)-6 inside the LBDs of both hCB2R and rCB2R (−9.71 and −6.15 kcal mol −1 , respectively), which decreases the affinity of this isomer to both receptors.In general, we observed a different binding mode of (+)-6 with less noncovalent contacts between (+)-6 and the LBD amino acids of the receptors.Consequently, this has a lower stabilizing impact on the best ranked poses of (+)-6 in hCB2R and in rCB2R (see details in the SI).
In Vivo Metabolism.The good in vitro results obtained for [ 18 F]RM365 justified its further evaluation in vivo, and we proceeded with the investigation of radioligand metabolism in mouse.The brain as well as blood samples were obtained from two female CD-1 mice 30 min after intravenous administration of [ 18 F]RM365.The HPLC radiochromatograms obtained from the extracts of the samples are shown in Figure 8.About 90% of the radioactivity detected in the brain samples accounted for intact radioligand (Figure 8), while in blood plasma, the fraction of the intact radioligand represented only 55% of the detected radioactivity (Figure 8B).Accordingly, the absence of a significant fraction of radiometabolites of [ 18 F]RM365 in the brain indicates the suitability of this radioligand for neuroimaging.
Regarding the structure of the slightly more polar radiolabeled metabolite M1 of [ 18 F]RM365 detected in both plasma and brain samples (Figure 8), we assume that the metabolic degradation might take place at four different positions: (a) hydrolysis of the carboxamide group, (b) hydrolysis of the nitrile group to an amide and/or further to a carboxylic acid, (c) Nacylation at the indole, or (d) oxidation of the indole-3-position (Figure 8C).By analyzing commercially available 5-fluoro-1Hindole-2-carboxylic acid by HPLC, we could exclude an in vivo hydrolysis of the amide group of RM365.However, by hydrolyzing the nitrile group of RM365 with NaOH/H 2 O 2 , we obtained the corresponding amide (21C, Table 1) and identified M1 as 21C from the same R f values (not shown).According to the previously determined K i value of 21C (Table 1), the confounding interaction of [ 18 F]M1 with hCB2R in PET studies is unlikely.It is, however, unclear why the radiometabolite [ 18   PET Imaging.The biodistribution of [ 18 F]RM365 was investigated in vivo in mice (n = 4) and rats (n = 2) by dynamic PET imaging studies.As shown in Figure 9, a constant SUV of 0.7 ± 0.1 between 1 and 60 min was observed in the mouse spleen (Figure 9A,B), which was comparable to rat, although the low number of animals and difference of the maximal SUVs in the plateau phase between 1 and 60 min are drawbacks of the study (Figure 9C,D).The normalized time−activity curve (TAC) of the rodent spleen to the blood TAC (Figure 9B,9D) suggests a low specific enrichment in both species and confirms the results of the in vitro binding studies.
The time-dependent uptake into the other tissues is presented in Figures S9 and S11 and was comparable for both investigated species.Among others, increasing activity in the duodenum over time and accumulation in the brown adipose tissue (BAT) were observed (Figure S10).In contrast, a slow washout from the salivary glands, muscle, bones, and kidneys could be observed.An initial comparably high TAC peak value with a SUV mean > 4 in the liver, left ventricle (blood), heart wall, and lung followed by a fast washout and a subsequent slower washout phase could be observed.Furthermore, we conclude a mainly hepatic excretion of the activity, as shown by the liver and jejunum TACs in both species and a bladder activity of 0.24 ± 0.19% of injected dose at 60 min in mice (Figures S8 and S10).
In a proof-of-concept study, the specific binding to the human CB2R and the ability to cross the blood−brain barrier of [ 18 F]RM365 in vivo was investigated by PET imaging experiments using an established rat model with a local overexpression of hCB2R(D80N) in the right striatum (target region). 32,66xemplary images of the averaged time frames of the PET study, as well as the TACs of the target region and both reference regions on the contralateral site and the cerebellum, are shown in Figure 10.The noncompartmental (semiquantitive) analysis of the TACs is presented in Table 2. Compared to the CB2R-directed radioligands previously evaluated by us in the same animal model (TAC peak SUV of 3.3 and 3.6 for [ 18 F]LU14 83 and [ 18 F]LU13, 31 respectively), the uptake of [ 18 F]RM365 into the target region was more than two times higher, with a SUV of 8.6 ± 3.2 at 60 min postinjection, whereby, further activity accumulation after 60 min could be assumed.However, the TAC peak value was four to five times lower in the reference regions, which was reflected by the seven to eight times lower accumulated activity over time expressed as the area under the curve (AUC) of [ 18 F]RM365 (Table 3).The high signal-tobackground ratio is reflected by a SUVr of 15 ± 5 and 20 ± 5 for the target region-to-contralateral and -cerebellum ratios, respectively, at 60 min p.i.Hence, the results confirm the blood−brain barrier permeability and the high CB2R specific binding as well as the low signal-to-background ratio of [ 18 F]RM365 in the brain of this rat model.Additionally, we investigated the reversibility of the binding toward the human CB2R receptor in vivo in displacement studies by administering the CB2R-specific agonist GW405833 into the tail vein 20 min after the radioligand injection into the hCB2R(D80N)-overexpressing rat model (Figure 10B, Table 3).The signal in the target region was reduced 15 min after the injection of GW405833 (p < 0.05, Bonferroni's multiple comparison test) compared to the control group, resulting in a reduction of the normalized AUC by 64 and 70% between 20 and 60 min p.i., using the contralateral site or cerebellum as the reference region, respectively, indicating a reversible binding toward the human CB2R (D80N) receptor.The signal in the contralateral side (pvalue = 0.1621, two-way ANOVA) and in the cerebellum (pvalue = 0.6177, two-way ANOVA) was not affected by the treatment over time with GW405833 (Figure S6), underlining the eligibility of these reference regions.
For the semipreparative chiral separation of (±)-6, a Merck-Hitachi Model D-6000 with an L-6100 pump, a Rheodyne injection valve with a 4 mL sample loop, and an L-6400 UV detector (monitoring at 254 nm) were used.The separation was performed on a CHIRALPAK IA column (250 × 10 mm) using the same eluent conditions as for the analytical separation.When 3.5 mg of the racemate dissolved in 500 μL of MeCN was injected at a flow rate of 3.0 mL/min, the plusenantiomer eluted with a t R = 27 min and the minus-enantiomer with a t R = 29 min.They could be collected without impurities of the other enantiomer.Conveniently, this sample was injected several times every 30 min, without stopping the run.
N- (14).Ti(OEt) 4 (500 μL, 2.1 mmol, 2 equiv) and compound 13 (130 mg, 2 mmol, 1 equiv) were added to a solution of compound 10 (200 mg, 1 mmol, 1 equiv) in toluol (20 mL) and the whole was heated at 70 °C for 2 h.After cooling to room temperature, H 2 O (20 mL) was added, the precipitate was removed by filtration, and the resulting aqueous phase was washed three times with EtOAc (3 × 10 mL).The combined organic solutions were dried over MgSO 4 , and the solvent was evaporated under reduced pressure.The resulting residue was dissolved in THF (20 mL) and cooled to −40 °C, and NaBH 4 (82 mg, 2.1 mmol, 2 equiv) was added.The reaction temperature was kept at −40 °C after which it was allowed to slowly Mean; 95% confidence interval (CI 95% ).

Journal of Medicinal Chemistry
warm up to rt overnight.An aqueous NaHCO 3 solution (2% aq, 20 mL) and DCM (20 mL) were added, vigorously stirred, and separated.The resulting aqueous phase was washed one more time with DCM (10 mL).The combined organic solutions were dried over MgSO 4 , the solvent was evaporated under reduced pressure, and the resulting residue was purified by flash chromatography on silica (SiO 2 , EA/Hex, 1:4 to 1:1) to give compounds 15 (first eluting fraction, 80 mg, 0.03 mmol, 31%) and 16 (second eluting fraction, 75 mg, 0.03 mmol, 29%).(R)-3-amino-2,2-dimethyl-2,3-dihydro-1H-indene-5-carbonitrile (17).A solution of compound 14 (50 mg, 0.2 mmol) in DCM (0.5 mL) was then treated with TFA (100 μL) at room temperature for 1 h.The solvent was evaporated, and the residue (17, 37 mg, 0.2 mmol) was used in the next step without purification. 1 General Procedure for the Synthesis of the Target Compounds 19A−H and 21A−E.A solution of the corresponding amine (1 equiv) and carboxylic acid (1 equiv) in DCM was treated with BOP (1.5 equiv) and Et 3 N (3 equiv) at rt for 16 h.Upon completion of the reaction, NaHCO 3 (2% aq) was added, vigorously stirred, and separated, and the aqueous phase was washed one more time with DCM.The combined organic phases were dried over MgSO 4 , the solvent was evaporated under reduced pressure, and the resulting residue was purified by flash chromatography on silica. (
Radiochemical yield, radiochemical purity, and analyses of plasma and brain samples were assessed via reversed-phase HPLC (RP-HPLC) in gradient mode Molar activity was determined using analytical HPLC with a Reprosil-Pur C18-AQ column (250 × 4.6 mm, 5 μm) and 50% MeCN/ 20 mM NH 4 OAc aq. as an eluent at a flow rate of 1 mL•min −1 and UV detection at 254 nm.
Determination of Lipophilicity (LogD 7.4 ).Log D 7.4 of [ 18 F]RM365 was experimentally determined in n-octanol/PBS, 0.01 M, pH 7.4, at room temperature by the shake flask method.The measurement was performed twice in triplicate. 87iological Experiments.All studies involving animals were carried out according to the national law on the protection of animals and were approved by the responsible authorities (Landesdirektion Sachsen, No. DD24.1-5131/446/19;TVV 18/18).
Determination of Binding Affinities by Homogenate Assays.The binding affinities toward hCB1R and hCB2R were determined according to a previously published protocol. 22 Homologous radioligand displacement studies investigating the potential of RM365 to displace [ 18 F]RM365 from binding sites in membrane homogenates of rat spleen or hCB2R-CHO cells were performed according to the same protocol.
The nonspecific binding of the respective radioligand was determined by coincubation with CP55,940 (final concentration 10 −5 M).The normalized values of bound activity (% specific binding) were calculated and plotted vs the logarithm of the concentration of the respective test compound.The IC 50 values of the resulting inhibition curves were estimated by nonlinear regression analysis (GraphPad Prism 2.01).To calculate the K i , the equation of Cheng and Prusoff was used.For homologous competition experiments, K i = K D . 38n Vitro Autoradiography.The in vitro autoradiographic experiments were performed according to a previously published protocol. 88n brief, 10 μm cryosections of CD-1 mouse brain rat spleen (female SPRD rat,10−12 weeks) were incubated in binding buffer (50 mM Tris-HCl, pH 7.4, 5% bovine serum albumin (BSA), 5 mM MgCl 2 , 1 mM EDTA) with [ 18 F]RM365 alone (total binding) or with coadministered RM365 at 1 μM or 10 nM (self-blocking), or of 1 μM CP55940 (CB1R/CB2R nonselective agonist), SR141716A (CB1R-selective antagonist), or SR144528 (CB2R-selective antagonist) for 1 h at room temperature.Afterward, the samples were washed and exposed to imaging plates (Fuji Photo Film, Co. Ltd., Tokyo, Japan) and eventually scanned using a HD-CR 35 scanner (Raytest Isotopenmessgeraete GmbH, Straubenhardt, Germany).The scan data were visualized and processed by computer-assisted microdensitometry (Aida version 2.31, Raytest Isotopenmessgeraẗe GmbH, Straubenhardt, Germany).
Quantification of Radiometabolites.About 30 MBq of [ 18 F]RM365 dissolved in about 150 μL of isotonic saline was administered intravenously as a bolus in the tail vein of awake female CD-1 mice weighing about 33 g (n = 3).At 30 min p.i., the animals were anesthetized and blood was withdrawn by retrobulbar bleeding using glass capillaries.Immediately afterward, the animals were euthanized by cervical dislocation and released urine sampled.Blood plasma was obtained from the whole blood sample by centrifugation (2 min, 8000 rpm, room temperature).In addition, the brain was isolated and homogenized in 1 mL of demineralized water on ice (1000 rpm, 10 strokes; glass vessel, PTFE plunger; Potter S, B. Braun Biotech International, Goettingen, Germany).
The samples were further processed for subsequent radio-chromatographic analyses.Two consecutive extractions were performed in duplicate for plasma and brain determinations.Plasma and brain samples were added to an ice-cold MeOH/H 2 O mixture (9:1, v/v).The samples were vortexed for 3 min, incubated on ice for 5 min, and centrifuged at 10,000 rpm for 5 min.Supernatants were collected, the precipitates were redissolved in 100 μL of extraction solvent, and the extraction procedure was repeated.The activities of supernatants and precipitates were measured in an γ-counter (1480 WIZARD, Fa.PerkinElmer), and the extraction efficiencies were calculated as the ratio of the radioactivity in the supernatant to the radioactivity in the original sample (supernatant + precipitate).The supernatants from both extractions were combined, concentrated at 70 °C under an argon stream up to a remaining volume of 100 μL, and subsequently analyzed by analytical radio-HPLC with a gradient system as applied in Section 5.2.2.

= ×
where c (radioactivity) is expressed as the standardized uptake value normalized to the bodyweight in g (SUV).Data are shown in mean ± standard deviation (SD).Group differences were tested by Student's ttest, whereas the TACs of reference tissues with and without displacement by GW405833 were tested by a two-way ANOVA, and additionally, the target region by Bonferroni's multiple comparison test, with p < 0.05 designated as significant.Area under the curves (AUCs) and corresponding 95% confidence intervals (CIs, 95%) were calculated with GraphPad Prism (v.8.2) following the assumptions described by Gagnon et al. 39 ■ ASSOCIATED CONTENT * sı Supporting Information

Figure 2 .
Figure 2. Rational design of a novel high affinity and selectivity 18 F-labeled CB2R radioligand.

Scheme 2 .
Scheme 2. Enantioselective Synthesis of RM365 and Its Derivatives a

Scheme 3 .
Scheme 3. Synthesis of Target Compounds 19A−H with the Modified 5-Fluoroindole Subunit a

Scheme 4 .
Scheme 4. Synthesis of Target Compounds with Modified Indanone Subunit a

Figure 4 .
Figure 4. Radiosynthesis of [ 18 F]RM365.(A) The influence of the Cu reagent−precursor molar ratio (25/24) on the yield of the formation of [ 18 F]RM365 (yield calculated from the crude reaction mixture, nondecay-corrected). (B) Representative radio-HPLC chromatogram from aliquots of the reaction mixture using 3 mg precursor 24 and a 25/24 molar ratio of 2.

Figure 6 .
Figure 6.Determination of the equilibrium dissociation constant K D of [ 18 F]RM365 in vitro.A homologous competition binding experiment was performed by incubating membrane preparations obtained from CHO cells stably transfected with the human CB2R or from the spleen of a female SPRD rat, expressing high levels of CB2R in leukocytes under physiological conditions with 0.078 nM [ 18 F]RM365 and increasing concentrations of RM365.Data are obtained from a single experiment performed in triplicate to give IC 50 , hCB2R 2.40 nM (CI 95% : 1.09−5.26nM) and K D , hCB2R 2.32 nM by the simplified Cheng−Prusoff equation K D = IC 50 − concentration radioligand.
F]M1 penetrates brain to a much lower extent as compared to [ 18 F]RM365 and if species differences regarding the [ 18 F]M1/[ 18 F]RM365 ratio in the brain are to be expected.Therefore, prior to a potential human application of [ 18 F]-RM365, the passive diffusion of [ 18 F]M1 and [ 18 F]RM365 should be tested in vitro using the tissue of both rodent and human origin.

Figure 9 .
Figure 9.In vivo uptake of [ 18 F]RM365 into the spleen of CD-1 mice (A, B) and Wistar rats (C, D). (A) and (C) Coronal section of a representative PET image (left), showing the averaged time frames from 0 to 60 min after intravenous administration of [ 18 F]RM365 and PET merged with the T1w MR image (right), where the spleen region is marked with a dotted ellipse.(B) and (D) Time−activity curve (TAC) of the radiotracer uptake into the spleen in mean standardized uptake values (SUV mean ) and the corresponding normalized TAC to the blood activity in the left ventricle (SUVr).Mice: n = 3, rats: n = 2, mean ± SD.

Figure 10 .
Figure 10.PET imaging of [ 18 F]RM365 in a rat model with local hCB2R overexpression in the right striatal region of the brain.Representative coronal planes showing merged MR and PET images of averaged time frames of the control group (A) and displacement study (B) with the corresponding time−activity curves (TACs) of the right (target region, hCB2R D80N) and the reference regions (contralateral and cerebellar regions) expressed in mean standardized uptake values (SUV mean ), as well as the normalized TACs to the reference regions (SUVr).Compared to the control group, in the displacement study, GW405833 (5 mg/kg bodyweight) was injected i.v.20 min after administration of [ 18 F]RM365.Mean ± SD, n = 3.

Table 2 .
Noncompartmental (Semiquantitative) Analysis of TACs, Obtained from the Regions of Local Overexpression of hCB2R(D80N) in the rRight Striatum of Rats, the Contralateral Region in the Left Striatum, and the Cerebellum After i.v.Injection of [ 18 F]RM365 (n = 3)

Table 1 .
Binding Affinities of Compound 6 and New DerivativesDetermined in-house according to protocols previously reported by us.53 bReported in ref60.cObtained by chiral HPLC separation (Scheme 1).Results are the K i values determined in independent experiments, each performed with technical triplicates.CB2R K i values were estimated from inhibition curves obtained by displacement of [ 3 H]WIN55.2212−2from hCB2R acquired by homogenization of stably transfected CHO cells.CB1R K i values were estimated from inhibition curves obtained by displacement of [ 3 H]SR141716A from hCB1R obtained by homogenization of stably transfected CHO cells, respectively.Single K i value was obtained by a single experiment.Mean of K i values of two independent experiments is reported with the single values of each experiment in brackets.Mean of K i values of ≥ three independent experiments is reported with the standard deviation.
a *