Design and Synthesis of 2,6-Disubstituted-4′-Selenoadenosine-5′-N,N-Dimethyluronamide Derivatives as Human A3 Adenosine Receptor Antagonists

A new series of 4′-selenoadenosine-5′-N,N-dimethyluronamide derivatives as highly potent and selective human A3 adenosine receptor (hA3AR) antagonists, is described. The highly selective A3AR agonists, 4′-selenoadenosine-5′-N-methyluronamides were successfully converted into selective antagonists by adding a second N-methyl group to the 5′-uronamide position. All the synthesized compounds showed medium to high binding affinity at the hA3AR. Among the synthesized compounds, 2-H-N6-3-iodobenzylamine derivative 9f exhibited the highest binding affinity at hA3AR. (Ki = 22.7 nM). The 2-H analogues generally showed better binding affinity than the 2-Cl analogues. The cAMP functional assay with 2-Cl-N6-3-iodobenzylamine derivative 9l demonstrated hA3AR antagonist activity. A molecular modelling study suggests an important role of the hydrogen of 5′-uronamide as an essential hydrogen bonding donor for hA3AR activation.


Introduction
Adenosine, which is the endogenous ligand of the adenosine receptors (ARs), is an important neuromodulator and mediates through activation of its four receptors, consisting of A 1 , A 2A , A 2B , and A 3 subtypes. These receptors are widely distributed in tissues and involved in various physiological activities [1]. Each subtype couples to a preferred type of G protein; A 1 and A 3 ARs primarily couple to the G i/o proteins, and A 2A and A 2B ARs couple to G s proteins. A 2B and A 3 ARs are also known to be coupled to G q proteins. AR signaling and their physiological roles have been extensively studied [2,3]. Among them, the A 3 AR is an important receptor to regulate cardioprotection in cardiac ischemia [4], degranulation of neutrophils [5], and cell proliferation [6]. These results led to the development of A 3 AR agonists as anticancer agents [7]. Selective A 3 AR antagonists are also promising ligands to modulate inflammation [8] and cerebroprotection [9,10]. Some studies showed that A 3 AR antagonists could enhance cancer treatment via the inhibition of HIF-1α and VEGF protein accumulation in hypoxia and in tumors [11] and are potential anti-glaucoma therapeutics as they reduce intraocular pressure in mouse and monkey [12].
In the past decades, a variety of approaches have been followed to discover novel drug candidates targeting A 3 AR. 2-Chloro-N 6 -(3-iodobenzyl)adenosine-5 -N-methyluronamide (Cl-IB-MECA, 1) and its 4 -thio analogue 2 were discovered as potent and selective A 3 AR agonists from the extensive structure-activity relationships based on the structure of adenosine [13] (Figure 1). The 5 -uronamide hydrogen of 1, required for full agonism, forms a putative hydrogen bond with T94 (3.36) at hA 3 AR as modeled, suggesting that this interaction is essential for receptor activation by adenosine agonists [14]. Consistent with these agonism, forms a putative hydrogen bond with T94 (3.36) at hA3AR as modeled, sugge ing that this interaction is essential for receptor activation by adenosine agonists [14]. Co sistent with these findings, the 4′-truncated analogues, such as 3 and 4, lacking this h drogen bond donor were discovered to act as A3AR antagonists or low-efficacy agonis demonstrating that a hydrogen bond donating ability of the 5′-uronamide promotes A3A activation [15,16]. The removal of the hydrogen bond donor ability by appending anoth methyl group to the 5′-uronamide, e.g., 5′-N,N-dimethyluronamide derivatives 5 and similarly reduced A3AR efficacy. These compounds were characterized as potent and s lective A3AR antagonists [17,18]. On the basis of a bioisosteric rationale, we recently reported that 4′-seleno analogues 1 and 2, i.e., 7 and 8, were discovered as potent and selective hA3AR agonists [19] (Figure They exhibited comparable A3AR binding affinity as the corresponding 4′-oxo-and 5′-th nucleosides 1 and 2. However, X-ray analysis indicated that in the pure crystalline state th preferred a syn nucleobase orientation and a South sugar conformation, unlike 1 and 2. A mentioned above, removal of the amide hydrogen of the 5′-uronamide of 1 and 2 by methylation, resulting in 5 and 6 successfully converted A3AR agonists into A3AR antag nists. Based on these findings, we hypothesized that the 4′-seleno analogue of 5 or 6, beari a 5′-N,N-dimethyluronamide moiety might be an A3AR antagonist ( Figure 2). Thus, we a alyzed the structure-activity relationship as A3AR ligands of this series by modifying N and C2 positions, by synthesizing novel 4′-selenonucleosides 9a-l. Herein, we report t synthesis and biological evaluation of 2,6-disubstituted-4′-selenoadenosine-5′-N,N-dim thyluronamide derivatives 9a-l as potent and selective A3AR a3ntagonists. On the basis of a bioisosteric rationale, we recently reported that 4 -seleno analogues of 1 and 2, i.e., 7 and 8, were discovered as potent and selective hA 3 AR agonists [19] (Figure 2). They exhibited comparable A 3 AR binding affinity as the corresponding 4 -oxo-and 5 -thio nucleosides 1 and 2. However, X-ray analysis indicated that in the pure crystalline state they preferred a syn nucleobase orientation and a South sugar conformation, unlike 1 and 2. As mentioned above, removal of the amide hydrogen of the 5 -uronamide of 1 and 2 by N-methylation, resulting in 5 and 6 successfully converted A 3 AR agonists into A 3 AR antagonists. Based on these findings, we hypothesized that the 4 -seleno analogue of 5 or 6, bearing a 5 -N,N-dimethyluronamide moiety might be an A 3 AR antagonist ( Figure 2). Thus, we analyzed the structure-activity relationship as A 3 AR ligands of this series by modifying N 6 -and C2 positions, by synthesizing novel 4 -selenonucleosides 9a-l. Herein, we report the synthesis and biological evaluation of 2,6-disubstituted-4 -selenoadenosine-5 -N,N-dimethyluronamide derivatives 9a-l as potent and selective A 3 AR a3ntagonists.

Chemistry
For the synthesis of final compounds 9a-l, key intermediates, 4′-seleno purine nucle osides 15a-b were synthesized from D-ribose following the previously reported proce dures [18,19] (Scheme 1). Briefly, D-ribose was converted to L-lyxonolactone derivative 1 in three steps (oxidation to lactone, conversion of D-ribo configuration to L-lyxo configu ration, and tert-butyldiphenylsililyl (TBDPS) protection). Reduction of 10 with NaBH ring cyclization of resulting diol with selenide ion, and a Pummerer rearrangement of 4 seleno sugar afforded the glycosyl donor 11. A Vorbrüggen condensation of 11 with 6 chloropurine and 2,6-dichloropurine produced the N 9 -β-anomers 12a and 13a with con comitant formation of their corresponding N 7 -β-anomers 12b and 13b, respectively. Con version of N 7 isomers 12b and 13b to their corresponding N 9 isomers 12a and 13a wa achieved by using TMSOTf at high temperature. Removal of the TBDPS group of 12a an 13a yielded the 5′-CH2OH derivatives 14a and 14b, respectively. Conversion of the 5′-hy droxymethyl group of 14a and 14b into a 5′-N,N-dimethyluronamide was successfull achieved, but the final deprotection of the acetonide group under strongly acidic cond tions resulted in decomposition, instead of giving the desired final products. Thus, w exchanged the acetonide protecting group of 14a and 14b with a TBS group in four step giving 15a and 15b, respectively. Firstly, a PNB protecting group was attached to the 5 position of 14a and 14b, followed by acetonide group deprotection with 50% aqueous TF to give diols. The diols were then protected with a TBS group using TBSOTf followed b deprotection of the PNB group with sodium hydroxide in 1,4-dioxane to give 15a and 15b respectively. The final deprotection with sodium hydroxide required mild reaction con ditions (room temperature, overnight) because of the possible hydrolytic conversion of 6 chloropurine to hypoxanthine.

Chemistry
For the synthesis of final compounds 9a-l, key intermediates, 4 -seleno purine nucleosides 15a-b were synthesized from D-ribose following the previously reported procedures [18,19] (Scheme 1). Briefly, D-ribose was converted to L-lyxonolactone derivative 10 in three steps (oxidation to lactone, conversion of D-ribo configuration to L-lyxo configuration, and tert-butyldiphenylsililyl (TBDPS) protection). Reduction of 10 with NaBH 4 , ring cyclization of resulting diol with selenide ion, and a Pummerer rearrangement of 4-seleno sugar afforded the glycosyl donor 11. A Vorbrüggen condensation of 11 with 6-chloropurine and 2,6-dichloropurine produced the N 9 -β-anomers 12a and 13a with concomitant formation of their corresponding N 7 -β-anomers 12b and 13b, respectively. Conversion of N 7 isomers 12b and 13b to their corresponding N 9 isomers 12a and 13a was achieved by using TMSOTf at high temperature. Removal of the TBDPS group of 12a and 13a yielded the 5 -CH 2 OH derivatives 14a and 14b, respectively. Conversion of the 5 -hydroxymethyl group of 14a and 14b into a 5 -N,N-dimethyluronamide was successfully achieved, but the final deprotection of the acetonide group under strongly acidic conditions resulted in decomposition, instead of giving the desired final products. Thus, we exchanged the acetonide protecting group of 14a and 14b with a TBS group in four steps, giving 15a and 15b, respectively. Firstly, a PNB protecting group was attached to the 5 -position of 14a and 14b, followed by acetonide group deprotection with 50% aqueous TFA to give diols. The diols were then protected with a TBS group using TBSOTf followed by deprotection of the PNB group with sodium hydroxide in 1,4-dioxane to give 15a and 15b, respectively. The final deprotection with sodium hydroxide required mild reaction conditions (room temperature, overnight) because of the possible hydrolytic conversion of 6-chloropurine to hypoxanthine. Synthesis of the final nucleosides, 9a-l from the key intermediates, 15a and 15b is shown in Scheme 2. The direct oxidation of the alcohols of 15a and 15b to the carboxylic acids have been tried with many oxidizing reagents, but none of them could give the desired acid. Thus, we employed a sequential oxidation method via aldehyde instead of direct oxidation to the carboxylic acid. Albright-Goldman oxidations of 15a and 15b, using DMSO as an oxidizing agent under mild condition afforded the aldehydes 16a and 16b, respectively. Tollens′ oxidation converted the aldehydes 16a and 16b to the corresponding carboxylic acids smoothly, which without purification underwent the amide coupling reaction with dimethylamine in the presence of DIPEA, and HATU to yield the 5′-N,N-dimethyluronamides 17a and 17b, respectively. The TBS deprotection of 17a and 17b with TBAF and acetic acid gave the diols 18a and 18b, respectively. The key intermediates 18a and 18b were treated with various amines such as ammonia, alkylamines and 3-halobenzylamines to yield 2-H derivatives 9a-f and 2-Cl derivatives 9g-l, respectively. Synthesis of the final nucleosides, 9a-l from the key intermediates, 15a and 15b is shown in Scheme 2. The direct oxidation of the alcohols of 15a and 15b to the carboxylic acids have been tried with many oxidizing reagents, but none of them could give the desired acid. Thus, we employed a sequential oxidation method via aldehyde instead of direct oxidation to the carboxylic acid. Albright-Goldman oxidations of 15a and 15b, using DMSO as an oxidizing agent under mild condition afforded the aldehydes 16a and 16b, respectively. Tollens oxidation converted the aldehydes 16a and 16b to the corresponding carboxylic acids smoothly, which without purification underwent the amide coupling reaction with dimethylamine in the presence of DIPEA, and HATU to yield the 5 -N,Ndimethyluronamides 17a and 17b, respectively. The TBS deprotection of 17a and 17b with TBAF and acetic acid gave the diols 18a and 18b, respectively. The key intermediates

Binding Affinity
The binding affinities of all the final compounds 9a-l were evaluated, using radioligand binding assays at four human AR subtypes (Table 1), by reported methods [20]. All of the final compounds 9a-l exhibited medium to high binding affinity at the hA3AR, while no binding affinity at other subtypes such as hA1, hA2A, and hA2BARs was observed. Among the tested compounds, compound 9f exhibited the highest affinity (Ki = 22.7 nM) at hA3AR, which is comparable to the corresponding 4′-oxo-and 4′-thio analogues 5 (Ki = 29.0 nM) and 6 (Ki = 15.5 nM). The introduction of a 3-halobenzyl group at the N 6 position increased the hA3AR binding affinity when compared to the N 6 -unsubstituted adenine compounds 9a or 9g, indicating that a favorable hydrophobic interaction exits at the hA3AR binding site. In the 2-H series, the binding affinity of 3-halobenzyl derivatives 9cf was decreased in the following order: 3-I-benzyl 9f > 3-Br-benzyl 9e > 3-Cl-benzyl 9d > 3-F-benzyl 9c. The halogen size correlated with hA3AR binding affinity, whereas in the 2-Cl series, the binding affinity of 3-halobenzyl derivatives 9i-l was almost same within the range of 180-250 nM. In general, the 4′-seleno-5′-N,N-dimethyluronamide derivatives 9al exhibited lower binding affinity than the 4′-seleno-5′-N-methyluronamide derivatives 7 and 8. It is interesting to note that 2-Cl-N 6 -3-iodobenzyl analogue 9l exhibited much

Binding Affinity
The binding affinities of all the final compounds 9a-l were evaluated, using radioligand binding assays at four human AR subtypes (Table 1), by reported methods [20]. All of the final compounds 9a-l exhibited medium to high binding affinity at the hA 3 AR, while no binding affinity at other subtypes such as hA 1 , hA 2A , and hA 2B ARs was observed. Among the tested compounds, compound 9f exhibited the highest affinity (K i = 22.7 nM) at hA 3 AR, which is comparable to the corresponding 4 -oxo-and 4 -thio analogues 5 (K i = 29.0 nM) and 6 (K i = 15.5 nM). The introduction of a 3-halobenzyl group at the N 6 position increased the hA 3 AR binding affinity when compared to the N 6 -unsubstituted adenine compounds 9a or 9g, indicating that a favorable hydrophobic interaction exits at the hA 3 AR binding site. In the 2-H series, the binding affinity of 3-halobenzyl derivatives 9c-f was decreased in the following order: 3-I-benzyl 9f > 3-Br-benzyl 9e > 3-Cl-benzyl 9d > 3-F-benzyl 9c. The halogen size correlated with hA 3 AR binding affinity, whereas in the 2-Cl series, the binding affinity of 3-halobenzyl derivatives 9i-l was almost same within the range of 180-250 nM. In general, the 4 -seleno-5 -N,N-dimethyluronamide derivatives 9a-l exhibited Pharmaceuticals 2021, 14, 363 6 of 14 lower binding affinity than the 4 -seleno-5 -N-methyluronamide derivatives 7 and 8. It is interesting to note that 2-Cl-N 6 -3-iodobenzyl analogue 9l exhibited much weaker binding affinity than the corresponding 2-H analogue 9f. This tendency was also found in the 5 -N-methyluronamide 4 -seleno derivatives 7 and 8. weaker binding affinity than the corresponding 2-H analogue 9f. This tendency was also found in the 5′-N-methyluronamide 4′-seleno derivatives 7 and 8.

CAMP Functional Assay
In a cAMP functional assay at hA3AR expressed in CHO cells, compound 9l behaved as an antagonist, like compounds 5 and 6, with KB value of 114.5 nM (Figure 3). Like 5 and 6, an additional methyl group on the 5′-N-methyluronamide converted an agonist into an antagonist, indicating that amide hydrogen is essential for receptor activation in this series, as well. However, the fact that the 5′-N,N-dimethyluronamide derivatives exhibited weaker binding affinity than the corresponding 5′-N-methyluronamide derivatives demonstrates that steric effects induced by 5′-N,N-dimethyluronamide reduce the binding affinity at the A3AR.

Compound
Affinity, K i , nM ± SEM a,b (or % Inhibition at 10 uM)

CAMP Functional Assay
In a cAMP functional assay at hA 3 AR expressed in CHO cells, compound 9l behaved as an antagonist, like compounds 5 and 6, with K B value of 114.5 nM (Figure 3). Like 5 and 6, an additional methyl group on the 5 -N-methyluronamide converted an agonist into an antagonist, indicating that amide hydrogen is essential for receptor activation in this series, as well. However, the fact that the 5 -N,N-dimethyluronamide derivatives exhibited weaker binding affinity than the corresponding 5 -N-methyluronamide derivatives demonstrates that steric effects induced by 5 -N,N-dimethyluronamide reduce the binding affinity at the A 3 AR.

Molecular Modelling Studies
To investigate how 5′-N,N-dimethyluronamide 4′-selenonucleoside derivatives bind at hA3AR, we docked our compounds into the reported homology model of hA3AR [21] using Autodock Vina [22]. The most potent compound 9f bound well at the orthosteric binding site with a South ring conformation (2′-endo/3′-exo), displaying H-bonds with Ser271 and His272 (Figure 4). Compared to the 5′-N-methyluronamide derivative 5, the adenine ring still maintained π−π interaction with Phe168 and the iodobenzene ring had interactions with Val169, Ile253 and Leu264 [19]. The glycosidic bond was in an anti conformation. However, either H-bonding of 5′-N-uronamide with Thr94 or the adenine with Asn250 was not observed (marked as a red circle in Figure 4), suggesting that this Hbonding plays a key role in discriminating an agonist from an antagonist.

Molecular Modelling Studies
To investigate how 5 -N,N-dimethyluronamide 4 -selenonucleoside derivatives bind at hA 3 AR, we docked our compounds into the reported homology model of hA 3 AR [21] using Autodock Vina [22]. The most potent compound 9f bound well at the orthosteric binding site with a South ring conformation (2 -endo/3 -exo), displaying H-bonds with Ser271 and His272 (Figure 4). Compared to the 5 -N-methyluronamide derivative 5, the adenine ring still maintained π−π interaction with Phe168 and the iodobenzene ring had interactions with Val169, Ile253 and Leu264 [19]. The glycosidic bond was in an anti conformation. However, either H-bonding of 5 -N-uronamide with Thr94 or the adenine with Asn250 was not observed (marked as a red circle in Figure 4), suggesting that this H-bonding plays a key role in discriminating an agonist from an antagonist.

Molecular Modelling Studies
To investigate how 5′-N,N-dimethyluronamide 4′-selenonucleoside derivatives bind at hA3AR, we docked our compounds into the reported homology model of hA3AR [21] using Autodock Vina [22]. The most potent compound 9f bound well at the orthosteric binding site with a South ring conformation (2′-endo/3′-exo), displaying H-bonds with Ser271 and His272 (Figure 4). Compared to the 5′-N-methyluronamide derivative 5, the adenine ring still maintained π−π interaction with Phe168 and the iodobenzene ring had interactions with Val169, Ile253 and Leu264 [19]. The glycosidic bond was in an anti conformation. However, either H-bonding of 5′-N-uronamide with Thr94 or the adenine with Asn250 was not observed (marked as a red circle in Figure 4), suggesting that this Hbonding plays a key role in discriminating an agonist from an antagonist.

Chemical Synthesis
Proton ( 1 H) and carbon ( 13 C) NMR spectra were obtained on a Jeol JNM-ECA 300 (JEOL Ltd. Tokyo, Japan; 300/75 MHz), Bruker AV 400 (Bruker, Billerica, MA, USA; 400/100 MHz), and AMX 500 (Bruker, Billerica, MA, USA; 500/125 MHz) spectrometer. The 1 H NMR data were reported as peak multiplicities: s for singlet; d for doublet; dd for doublet of doublets; t for triplet; td for triplet of doublet; q for quartet; quin for quintet; bs for broad singlet and m for multiplet. Coupling constants were reported in hertz. The chemical shifts were reported as ppm (δ) relative to the solvent peak. All reactions were routinely carried out under an inert atmosphere of dry nitrogen. IKA RCT basic type heating mantle was used to provide a constant heat source. Microwave-assisted reactions were carried out in sealed vessels using a Biotage Initiator + US/JPN (Biotage, Uppsala, Sweden; part no. 356007) microwave reactor, and the reaction temperatures were monitored by an external surface IR sensor. High-resolution mass spectra were measured with electrospray-ionization quadrupole time-of-flight (ESI-Q-TOF) techniques. Melting points were recorded on a Barnstead electrothermal 9100 instrument and are uncorrected. Reactions were checked by thin layer chromatography (Kieselgel 60 F254, Merck, Kenilworth, NJ, US). Spots were detected by viewing under a UV light, and by colorizing with charring after dipping in a p-anisaldehyde solution. The crude compounds were purified by column chromatography on a silica gel (Kieselgel 60, 70−230 mesh, Merck). All the anhydrous solvents were redistilled over CaH 2 , or P 2 O 5 , or sodium/benzophenone prior to the reaction.

Conclusions
On the basis of potent and selective antagonist 5 and 6 at the human A 3 AR, N 6substituted-5 -N,N-dimethylcarbamoyl-4 -selenonuceloside derivatives (9a-l) were synthesized from D-ribose and evaluated for their binding affinity toward hARs. All final compounds exhibited medium to high binding affinity toward A 3 AR with high selectivity compared to other subtypes. Among these derivatives, compound 9f was found to show the highest binding affinity (K i = 22.7 nM) at hA 3 AR, comparable to the corresponding 4 -oxo-and 4 -thio analogues 5 (K i = 29.0 nM) and 6 (K i = 15.5 nM). As in the case of 4 -oxo-and 4 -thio analogues 5 and 6, addition of another methyl group to the 5 -Nmethyluronamide converted an A 3 AR agonist into an A 3 AR antagonist, demonstrating the importance of amide hydrogen for receptor activation, which was supported by the molecular modelling study.
We believe that this study helps to define the pharmacophore needed for receptor activation or inactivation and will aid in the design of selective A 3 AR ligands by medicinal chemists.