Development of Bicyclo[3.1.0]hexane-Based A3 Receptor Ligands: Closing the Gaps in the Structure–Affinity Relationships

The adenosine A3 receptor is a promising target for treating and diagnosing inflammation and cancer. In this paper, a series of bicyclo[3.1.0]hexane-based nucleosides was synthesized and evaluated for their P1 receptor affinities in radioligand binding studies. The study focused on modifications at 1-, 2-, and 6-positions of the purine ring and variations of the 5′-position at the bicyclo[3.1.0]hexane moiety, closing existing gaps in the structure–affinity relationships. The most potent derivative 30 displayed moderate A3AR affinity (Ki of 0.38 μM) and high A3R selectivity. A subset of compounds varied at 5′-position was further evaluated in functional P2Y1R assays, displaying no off-target activity.


Introduction
The G protein-coupled adenosine (P1) receptors A 1 , A 2A , A 2B and A 3 play a central role in the complex mechanisms of purinergic signaling. In general, adenosine, the endogenous agonist at P1 receptors, exhibits protective functions as a response to organ stress and release of damage-associated-molecular pattern (DAMP) molecules such as e.g., ATP and S100 proteins [1][2][3]. Various P1 receptor agonists have been in clinical trials; to name a few, capadenoson (A 1 AR agonist) for the treatment of atrial fibrillation (NCT00568945) [4], apadenoson (A 2A AR agonist) for the SPECT-myocardial perfusion imaging (NCT01313572), the A 3 receptor agonists namodenoson in phase III for liver cancer (NCT04697810) [5,6], and piclidenoson (IB-MECA) for the treatment of psoriasis (NCT03168256), rheumatoid arthritis, and most recently, COVID-19 infections (NCT04333472) [7]. We are particularly interested in targeting the A 3 receptor due to its high overexpression in inflammatory and cancer cells compared to its low expression levels in healthy cells, thus making it a potentially promising therapeutic and diagnostic target [8][9][10]. The introduction of the bicyclo[3.1.0]hexane scaffold, also known as (N)-methanocarba (N for North), in place of the furanose ring of nucleoside agonists is known to increase the A 3 receptor (A 3 AR) potency and selectivity in comparison to other adenosine receptor subtypes [11,12]. In 2005 Jacobson et al. reported compounds 1a and 1b as highly potent A 3 receptor agonists [13] and most recently, the synthesis of S-thioether (N)-methanocarba adenosine derivatives such as compound 2 ( Figure 1) [14]. We were interested in exploring these scaffolds further through various substitutions at 6-position of the purine ring (purine numbering), the introduction of the 1-deazapurine scaffold, and variations of the 5 -position (ribose numbering) at the methanocarba moiety ( Figure 1, general structure I). Jacobson et al. have already established the methyl and ethyl carboxamides as highly efficient substituents at the 5 -position. There are only a few reports on introducing other functional moieties at the 5 -position of adenosine receptor ligands, one of them being the tetrazole compound 3 as a highly potent dual A 1 AR and A 3 AR ligand [15]. However, the introduction of other, in particular acidic, functional groups at the 5 -position was never investigated. Therefore, we decided to combine the (N)-methanocarba moiety (providing A 3 AR preference [12]) with various functional groups at the 5 -position to develop novel adenosine receptor ligands.

Results and Discussion
The synthesis of the bicyclo[3.1.0]hexane scaffold followed the reported procedure by Michel et al. [16], starting with D-ribose and leading to the TBDPS-protected bicyclo[3.1.0]hexan alcohol 4 as a central building block in 9 consecutive steps (see Supplementary Materials Scheme S1). First, we decided to explore the role of the nitrogen atom at the 1-position of the purine ring. Nitration of 6-chloro-1-deazapurine has led selectively to the formation of 2-nitro derivative 7. Mitsunobu reaction of either 2,6-dichloro-1-deazapurine (6) or 6-chloro-2-nitro-1-deazapurine (7) with the methanocarba building block 4 had led to the formation of the protected nucleoside derivatives 8 and 9, respectively, that were subsequently varied further at the 2-position of the purine ring through the introduction of either amino or methylthio groups (Scheme 1). The exocyclic amine at 6-position was introduced in a reaction of the 6-chloro derivative 9 and benzylamine (for compounds 13 and 14) or para-methoxybenzyl amine (PMB, for compound 15). Cleavage of the PMB group led to the derivatives 16 and 17 bearing a free exocyclic amine. The attempt of introducing the nitro group at the Boc-protected 2-chloro-1-deazapurine (18), in order to introduce the electron-withdrawing nitro group at 6-position, has led to the formation of one single compound, the 2-chloro-1-nitro-1-deazapurine (19), in 76% yield and not the desired 6-nitro derivative (purine numbering). The position of the nitration was additionally proven by an X-ray structure of compound 19 (Scheme 2). The reaction of the protected 6-chloro-2-nitro nucleoside 8 with dibenzylamine was sluggish; therefore, to synthesize the N,N-dibenzyl-1-deaza derivatives, we envisaged the introduction of the dibenzyl group at the tosyl-protected 6-chloro-2-nitro-1-deazapurine 20 followed by subsequent cleavage of the tosyl group and a Mitsunobu reaction with compound 4. Interestingly, the reaction of dibenzylamine with deazapurine 20 provided selectively the ring-opened product 21 in 78% yield. Due to the strong electron-withdrawing effect of the tosyl group, the dibenzylamine was able to perform a nucleophilic attack at the 8-position of the purine scaffold. The structure of compound 21 was additionally confirmed by X-ray crystal structure analysis (Scheme 3). Since the reaction of nucleoside 8 bearing a nitro group at the 2-position with dibenzylamine has led to the formation of various side-products, the synthesis of dibenzyl derivatives was skipped, and the nitro group was subsequently reduced to the primary amine function leading subsequently to the nucleoside 11. Scheme 3. Reaction of dibenzyl amine with compound 20. Reagents and conditions: (a) dibenzylamine, CH 2 Cl 2 , RT. Molecular structure of compound 21. Thermal ellipsoids are depicted at 30% probability. CCDC number: 2157453.
The purine derivatives 24 and 25, bearing two benzyl groups, were prepared through the reaction of 6-chloropurines 22 and 23 with dibenzyl amine and subsequent Mitsunobu reaction of the methanocarba building block 4, respectively. The methylthio group was introduced by reacting the protected 2-chloropurine nucleoside 25 with NaSCH 3 . Additionally, the 5 -hydroxy group was replaced by a chloride using cyanuric chloride. Cleavage of the acetonide and TBDPS groups has led to the formation of the respective nucleosides 26, 27, 30 and 31 in high yields (Scheme 4).
Intrigued by the high A 3 AR affinity of compound 1b and moderate affinity of compound 2, we selected the 2-methylthio substituted adenine scaffold for the evaluation of the modifications at the 5 -position. Also, (N)-methanocarba adenine 36 should be prepared as a reference compound for the SAR studies. Hereby adenine (32) or 2-chloro adenine (33) were subjected to the Mitsunobu reaction. Subsequent cleavage of the protecting groups of compound 34 furnished (N)-methanocarba adenosine 36, while the protected nucleoside 35 was used for the introduction of the methylthio group at the 2-position. Selective cleavage of the TBDPS protecting group and subsequent tosylation of the free alcohol and nucleophilic substitution of the tosylate led to the formation of the azide 38 as a central intermediate. Huisgen cycloaddition of the azide 38 with various alkynes and subsequent cleavage of the acetonide provided the triazole nucleosides 39-46 bearing neutral (39-42), basic (43), or acidic (44-46) functional groups. Reduction of the azide function using Pd/C, H 2 led to the formation of an amine suitable for the reaction with squaric acid dimethyl ester or ethyl 2-(chlorosulfonyl)acetate to provide compounds 47 and 48, respectively (Scheme 5). The compounds were evaluated for their P1 receptor affinity in A 1 , A 2A , A 2B , and A 3 receptor binding studies (Table 1). From all synthesized compounds, the (N)-methanocarba adenosine 36 is the only derivative displaying affinity to more than one P1 receptor. Compound 36 shows a preference for the A 3 receptor subtype with a Ki of 960 nM, 2-to 6-fold lower affinity towards A 2A and A 1 receptors, respectively, and no affinity at the A 2B subtype. The nitrogen atom at 1-position is not required for A 3 receptor affinity, as receptor binding appears to highly depend on substituents at 2-and 6-position. The derivatives 11 and 12 bearing a chloro substituent at 6-position show no P1 receptor affinity. The introduction of an amino group at the 6-position of the adenine ring as in compound 16 significantly increases the A 3 receptor affinity (Ki = 1.60 µM) while not showing any binding at other subtypes. Replacing the chloro with a methylthio group as in 17 leads to a loss of P1 receptor affinity. Interestingly, benzylation of the exocyclic amine as in compounds 13 and 14 restores the A 3 R affinity irrespective of the substituent at the 2-position. Extending the benzyl to a para-methoxybenzyl group as in 15 has no effect on A 3 R binding (Ki = 0.50 µM). In the purine series, the dibenzylation of the exocyclic amine appears to work only in combination with the methylthio group (30, Ki (A 3 R) = 0.38 µM); derivatives 26, 27 and 31 were not potent at the A 3 receptor. Most variations at the 5 -position were not tolerated. Only the triazole ester 42 displays a low A 3 R affinity of Ki 6.35 µM. Considering the potential of the introduced moieties in compounds 39-48 to serve as potential bioisosteres of mono-and diphosphate groups, the compounds 39-48 were tested for their functional activity (agonistic and antagonistic) at P2Y 1 receptors; none of the derivatives displayed any functional activity at P2Y 1 receptors up to a concentration of 10 µM.

Conclusions
With the aim to further explore the SAR of (N)-methanocarba nucleosides at A 3 receptor, a series of derivatives 11-17, 26, 27, 30, 31, 36, 39-48 varied at 1-, 2-, 6-and 5 -positions were prepared and evaluated for their affinity across all P1 receptor subtypes. The (N)-methanocarba adenosine 36 displayed affinity at A 1 , A 2A , and A 3 receptors combined with only moderate A 3 AR preference. The most potent compound 30, bearing dibenzylamino group at the 6-position and methylthio at the 2-position, displayed high A 3 R selectivity. The presence of the nitrogen atom at the 1-position of the purine ring was not required for the A 3 AR affinity, consistent with a recent report on hypermodified (N)-methanocarba derivatives [15]. The introduction of larger moieties at the 5 -position led to a complete loss of A 3 AR affinity, except for the triazole ester 42 displaying low A 3 AR affinity. Further structural modifications such as e.g., benzylation of the exocyclic amine function might restore the affinity of the 5 -triazoles at the A 3 receptor.
In conclusion, based on the multiple potential applications of (N)-methanocarba nucleosides as therapeutic agents [17,18], we have introduced new lead compounds that bind to the A 3 AR and can be further elaborated to increase affinity and selectivity.  13 C NMR (100 MHz): Unity Mercury plus 400 spectrometer (Varian ®® ) JEOL JNM-ECA-400; δ in ppm relative to tetramethylsilane; coupling constants are given with 0.5 Hz resolution, the assignments of 13 C and 1 H NMR signals were supported by 2D NMR techniques; MS: APCI = atmospheric pressure chemical ionization, EI = electron impact, ESI = electrospray ionization: MicroTof (Bruker Daltronics, Bremen, Germany), calibration with sodium formate clusters before measurement. All solvents were of analytical grade quality and demineralized water was used. HPLC solvents were of gradient grade quality, and ultrapure water was used. All HPLC eluents were degassed by sonication prior to use. Thin-layer chromatography was conducted with silica gel F 254 on aluminum plates in a saturated chamber at room temperature. The spots were visualized using UV light (254 nm) or reagents such as cerium molybdate dipping bath with additional heating using a standard heat gun. The retention factor values strongly depend on the temperature, the chamber saturation, and exact ratio of components of the eluent (highly volatile); the given retention factor values represent just approximate values. Flash column chromatography was conducted with silica gel 600 (40-63 µm, Macherey-Nagel). X-ray crystal structures: Equipment: Bruker APEX II CCD diffractometer (Bruker, Bremen, Germany): four circle diffractometer, Cu X-ray tube, graphite monochromator, APEX II CCD surface detector, Oxford Cryosystem 700 series (Oxford, UK) (N 2 flow: 100-300 K).
X-ray crystal structure analysis of 19: a colorless, plate-like specimen of C 6 H 3 ClN 4 O 2 , approximate dimensions 0.060 mm × 0.200 mm × 0.260 mm, was used for the X-ray crystallographic analysis. The X-ray intensity data were measured on a rotating anode Nonius FR591 system equipped with a Mo rotating anode Mo rotating anode (Mo Kα, λ = 0.71073 Å) and a Montel mirror monochromator. The integration of the data using an orthorhombic unit cell yielded a total of 3297 reflections to a maximum θ angle of 28.13 • (0.75 Å resolution), of which 1804 were independent (average redundancy 1.828, completeness = 99.1%, R int = 1.94%, R sig = 2.27%) and 1633 (90.52%) were greater than 2σ(F 2 ). The final cell constants of a = 11.3385(3) Å, b = 6.5407(2) Å, c = 20.0740(7) Å, volume = 1488.72(8) Å 3 , are based upon the refinement of the XYZ-centroids of reflections above 20 σ(I). Data were corrected for absorption effects using the multi-scan method (SADABS). The calculated minimum and maximum transmission coefficients (based on crystal size) are 0.8850 and 0.9720. The structure was solved and refined using the Bruker SHELXTL Software Package, using the space group Pbca, with Z = 8 for the formula unit, C 6 H 3 ClN 4 O 2 . The final anisotropic full-matrix least-squares refinement on F 2 with 122 variables converged at R1 = 3.16% for the observed data and wR2 = 8.40% for all data.
The goodness-of-fit was 1.101. The largest peak in the final difference electron density synthesis was 0.293 e − /Å 3 and the largest hole was −0.263 e − /Å 3 with an RMS deviation of 0.050 e − /Å 3 . On the basis of the final model, the calculated density was 1.772 g/cm 3 and F(000), 800 e − . The hydrogen at N2 atom was refined freely.
X-ray crystal structure analysis of 21: A pale yellow, prism-like specimen of C 27 H 24 ClN 5 O 4 S, approximate dimensions 0.070 mm × 0.160 mm × 0.200 mm, was used for the X-ray crystallographic analysis. The X-ray intensity data were measured on a rotating anode Nonius FR591 system equipped with a Mo rotating anode (Mo Kα, λ = 0.71073 Å) and a Montel mirror monochromator. The integration of the data using a monoclinic unit cell yielded a total of 9321 reflections to a maximum θ angle of 26.73 • (0.79 Å resolution), of which 5490 were independent (average redundancy 1.698, completeness = 98.2%, R int = 2.98%, R sig = 3.82%) and 4655 (84.79%) were greater than 2σ(

Cell Culture and Membrane Preparation
Chinese hamster ovary (CHO) cells stably expressing the human adenosine A 1 receptor (CHOhA1R) were kindly provided by Prof. S. J. Hill and CHO cells stably expressing the human adenosine A 3 receptor (CHOhA 3 R) were a gift from Dr. K.-N. Klotz (University of Würzburg, Germany). Chinese hamster ovary cells stably expressing the human A 1 -receptor (CHOhA 1 R) or the human A 3 -receptor (CHOhA 3 R) were grown in Dulbecco's Modified Eagle's Medium (DMEM) and Ham's F12 medium (1:1) supplemented with 10% (v/v) newborn calf serum, 50 µg/mL streptomycin, 50 IU/mL penicillin, and 200 µg/mL G418 at 37 • C and 5% CO 2 . CHOhA 1 R cells were subcultured twice a week at a ratio of 1:20 on 10 cm Ø plates and 15 cm Ø plates. CHOhA 3 R cells were subcultured twice a week at a ratio of 1:8 on 10 cm Ø plates and 15 cm Ø plates.
All cells were grown to 80-90% confluency and detached from plates by scraping them into 5 mL phosphate-buffered saline. Detached cells were collected and centrifuged at 200 g for 5 min. Pellets derived from 100 15 cm Ø plates were pooled and resuspended in 70 mL of ice-cold 50 mM tris(hydroxymethyl)aminomethane (Tris)-HCl buffer, pH 7.4. A Heidolph Diax 900 homogenizer was used to homogenize the cell suspension. Membranes and the cytosolic fraction were separated by centrifugation at 100,000 g in a Beckman Optima LE-80 K ultracentrifuge (Beckman Coulter, Fullerton, CA, USA) at 4 • C for 20 min. The pellet was resuspended in 35 mL of the Tris-HCl buffer, and the homogenization and centrifugation steps were repeated. Tris-HCl buffer (25 mL) was used to resuspend the pellet, and adenosine deaminase (ADA) was added (0.8 U/mL) to break down endogenous adenosine. Membranes were stored in 250 µL and 500 µL aliquots at −80 • C. Total protein concentrations were measured using the bicinchoninic acid (BCA) method. At these concentrations, total radioligand binding did not exceed 10% of that added to prevent ligand depletion. Nonspecific binding was determined in the presence of 100 µM N 6 -cyclopentyladenosine (CPA). Incubations were terminated by rapid vacuum filtration to separate the bound and free radioligand through prewetted 96-well GF/B filter plates using a PerkinElmer Filtermate-harvester (Perkin Elmer, Groningen, the Netherlands).

yl-amino]ethyl]phenol ([ 3 H]ZM241385
). At these concentrations, total radioligand binding did not exceed 10% of that added to prevent ligand depletion. Nonspecific binding was determined in the presence of 100 µM adenosine-5-N-ethyluronamide (NECA). Incubations were terminated by rapid vacuum filtration to separate the bound and free radioligand through prewetted 96-well GF/B filter plates using a PerkinElmer Filtermate-harvester (Perkin Elmer, Groningen, The Netherlands). Filters were subsequently washed 12 times with ice-cold 50 mM Tris-HCl, pH 7.4.
A 2B Receptor: Membrane aliquots containing 30 µg (CHO-spap-hA 2B R) total protein were incubated in a total volume of 100 µL assay buffer (0.1% CHAPS in 50 mM TrisHCl, pH 7.4) at 25 • C for 2 h. Radioligand displacement experiments were performed using six concentrations of competing ligand in the presence of 1.5 nM . At these concentrations, total radioligand binding did not exceed 10% of that added to prevent ligand depletion. Nonspecific binding was determined in the presence of 10 µM ZM241385. Incubations were terminated by rapid vacuum filtration to separate the bound and free radioligand through prewetted 96-well GF/B filter plates using a PerkinElmer Filtermateharvester (Perkin Elmer, Groningen, the Netherlands). Filters were subsequently washed 12 times with ice-cold 0.1% BSA in 50 mM Tris-HCl, pH 7.4. A 3 Receptor: Membrane aliquots containing 15 µg (CHOhA 3 R) total protein were incubated in a total volume of 100 µL assay buffer (50 mM Tris-HCl, pH 8.0, supplemented with 10 mM MgCl 2 , 1 mM EDTA and 0.01% (w/v) CHAPS) at 25 • C for 2 h. Radioligand displacement experiments were performed using six concentrations of competing ligand in the presence of 10 nM ). At these concentrations, total radioligand binding did not exceed 10% of that added to prevent ligand depletion. Nonspecific binding was determined in the presence of 100 µM NECA. Incubations were terminated by rapid vacuum filtration to separate the bound and free radioligand through prewetted 96-well GF/B filter plates using a PerkinElmer Filtermate-harvester (Perkin Elmer, Groningen, The Netherlands). Filters were subsequently washed 12 times with ice-cold 50 mM Tris-HCl supplemented with 10 mM MgCl 2 , and 1 mM EDTA, pH 8.0 for CHOhA 3 R.
The plates of all four adenosine receptor assays were dried at 55 • C after which MicroscintTM-20-cocktail was added (Perkin Elmer, Groningen, The Netherlands). After 3 h the filter-bound radioactivity was determined by scintillation spectrometry using a 2450 MicroBeta Microplate Counter (Perkin Elmer, Groningen, The Netherlands).

Data Analysis
All experimental data were analyzed using the non-linear regression curve fitting program GraphPad Prism 7.0 (GraphPad Software Inc., San Diego, CA, USA). IC 50 values obtained from competition displacement binding data were converted into K i values using the Cheng-Prusoff equation.

Data Analysis
The activation or inhibition curves of three independent measurements, each done in duplicates, were fitted to Hill equation using GraphPad Prism software version 9.3.1 (GraphPad Software Inc. San Diego, CA, USA).