Structure-Activity Relationships of Truncated D- and L-4′-Thioadenosine Derivatives as Species-Independent A₃ Adenosine Receptor Antagonists

Abstract

Novel D- and L-4′-thioadenosine derivatives lacking the 4′-hydroxymethyl moiety were synthesized, starting from D-mannose and D-gulonic γ-lactone, respectively, as potent and selective species-independent A₃ adenosine receptor (AR) antagonists. Among the novel 4′-truncated 2-H nucleosides tested, a N⁶-(3-chlorobenzyl) derivative 7c was the most potent at the human A₃ AR (K_i = 1.5 nM), but a N⁶-(3-bromobenzyl) derivative 7d showed the optimal species-independent binding affinity.

Introduction

On the basis of the structure of adenosine, an endogenous cell signaling molecule that binds to four specific subtypes (A₁, A_2A, A_2B, and A₃) of adenosine receptors (ARs)², a number of nucleoside analogues have been synthesized and evaluated as adenosine receptor ligands.³ Among these, IB-MECA⁴ 1 and Cl-IB-MECA⁵ 2 were discovered as potent and selective A₃ AR full agonists (K_i = 1.0 and 1.4 nM, respectively, at the human A₃ AR) and are being developed as antiinflammatory and anticancer agents. Based on the bioisosteric rationale, we reported the 4′-thionucleosides 3 and 4, derivatives of compounds 1 and 2, to also be highly potent and selective A₃ AR full agonists.⁶ Compound 4 exhibited potent in vitro and in vivo antitumor activities,⁷ resulting from the inhibition of Wnt signaling pathway (Chart 1).

Chart 1.

The rationale for the design of the target nucleosides 7.

However, because of the structural similarity to adenosine, most of these adenosine analogues were found to be A₃ AR agonists. Only a few nucleoside derivatives⁸ have been reported to be A₃ AR antagonists, but these generally exhibit weaker and less selective human A₃ AR antagonism than nonpurine heterocyclic A₃ AR antagonists. Although these nonpurine heterocyclic A₃ AR antagonists⁹ bound with high affinity at the human A₃ AR, they were weak or ineffective at the rat A₃ AR, indicating that they were not ideal for evaluation in small animal models and thus as drug candidates.¹⁰ Therefore, it is highly desirable to develop A₃ AR antagonists that are independent of species. The fact¹⁰ that nucleoside analogues show minimal species-dependence at the A₃ AR prompted us to search for novel potent and selective A₃ AR antagonists, derived from nucleoside templates.

A molecular modeling study of the A₃ AR indicated that hydrogen of the 5′-uronamides of compounds 1–4 serves as a hydrogen-bonding donor in the binding site of the A₃ AR, which is essential for the induced-fit required for the activation of the A₃ AR.¹¹ On the basis of these findings, we appended extra alkyl groups on the 5′-uronamides of compounds 1–4 to remove hydrogen-bonding ability at this site, thus precluding the conformational change required for activation of the A₃ AR. As expected, these 5′-N,N-dialkyl amide derivatives¹² displayed potent and selective A₃ AR antagonism, in which steric factors were crucial for affinity in binding to the A₃ AR. Within this class, 5′-N,N-dimethylamide derivative 5 was discovered to be the most potent full A₃ AR antagonist. Encouraged with these results, we designed and synthesized another new template to remove the 5′-uronamide group of compound 4 in order to minimize the steric repulsion at the binding region of the 5′-uronamide group and to abolish its hydrogen-bonding ability. This led to the discovery of compounds 6a – 6e as highly potent and selective human A₃ AR antagonists, which was more potent and selective than compound 5.¹ Among these, compound 6e also showed species-independent binding affinity, as indicated by its high affinity at the rat A₃ AR.¹ On the basis of these results, it is of interest to systematically establish structure-activity relationships by modifying the C2 and N⁶ positions of the purine moiety of compounds 6a – 6e in order to develop novel A₃ AR antagonists. In this article, we extend previous observations that truncated D-4′-thioadenosine derivatives 6a – 6e containing 2-Cl substitution are selective A₃ AR antagonists.¹ A series of 2-H analogues were prepared and characterized biologically. The binding affinities at the human A₃ AR were compared with those at the rat A₃ AR to develop species-independent A₃ AR antagonists. We also compared the binding affinities of D-4′-thionucleosides with those of the corresponding L-4′-thionucleosides to determine a stereochemical preference. Thus, here we report a full account of truncated D- and L-4′-thioadenosine derivatives 7 as highly potent and species-independent A₃ AR antagonists.

Results and discussion

The D-glycosyl donor 8 was subjected to the Lewis acid-catalyzed condensation for the synthesis of the final D-4′-thionucleosides lacking a 4′-hydroxymethyl group, as shown in Scheme 1. The D-glycosyl donor 8 was condensed with 6-chloropurine in the presence of TMSOTf as a Lewis acid to give β-6-chloropurine derivative 9 as a single diastereomer. The anomeric configuration of compound 9 was easily confirmed by ¹H NOE experiment between 3′-H and H-8. Removal of the isopropylidene group of 9 was achieved with 2 N HCl in THF to give 10. The 2-H intermediate 10 was converted to the novel N⁶-methyl derivative 7a and N⁶-3-halobenzyl derivatives 7b – 7e by treating with methylamine and 3-halobenzylamines, respectively. This route parallels the synthesis of the 2-chloro-N⁶-susbtituted-4′-thiopurine analogues 6a – 6e that we reported earlier¹.

Scheme 1.

Synthesis of truncated D-4′-thioadenosine derivatives 7a–7e.

Reagents and conditions: a) 6-chloropurine, ammonium sulfate, HMDS, 170 °C, 15 h, then TMSOTf, DCE, rt to 80 °C, 3 h; b) 2 N HCl, THF, rt, 15 h; c) RNH₂, Et₃N, EtOH, rt, 1–3 d.

In order to determine whether a stereochemical preference exists in the binding to the A₃ AR, the L-enantiomers, 7f and 7g of D-4′-thionucleosides were synthesized as illustrated in Scheme 2. D-Gulonic γ-lactone was converted to the diol 11 according to our previously published procedure.¹³ One-step conversion of the diol 11 into the L-glycosyl donor 12 was achieved using excess Pb(OAc)₄, indicating that oxidative diol cleavage, oxidation of the resulting aldehyde to the acid, and oxidative decarboxylation occurred simultaneously.¹³ Using the same synthetic strategy shown in Scheme 1, L-4′-thioadenosine derivatives 7f and 7g were synthesized from L-glycosyl donor 12.

Scheme 2.

Synthesis of truncated L-4′-thioadenosine derivatives 7f and 7g.

Initial binding experiments were performed using adherent mammalian cells stably transfected with cDNA encoding the appropriate human ARs (A₁ AR and A₃ AR in CHO cells and A_2A AR in HEK-293 cells).^14,15 Binding was carried out using 1 nM [³H]CCPA, 10 nM [³H]CGS-21680, or 0.5 nM [¹²⁵I]I-AB-MECA as radioligands for A₁, A_2A, and A₃ ARs, respectively. As shown in Table 1, most of the synthesized compounds exhibited high binding affinity at the human A₃ AR with low binding affinities at the human A₁ AR and human A_2A AR. Among the novel 2-H truncated adenosine derivatives tested, compound 7c (R = 3-chlorobenzyl) showed the highest binding affinity (K_i = 1.5 ± 0.4 nM) at the human A₃ AR with high selectivities versus the A₁ AR (570-fold selective) and the A_2A AR (290-fold selective). Compound 7e (R = 3-iodobenzyl) was also very potent (K_i = 2.5 ± 1.0 nM), with selectivities of 210- and 92-fold versus the A₁ and A_2A AR, respectively. N⁶-Substituted adenosine derivatives 7a – 7e without a 2-chloro substituent showed a very similar pattern to the corresponding 2-chloro derivatives 6a – 6e in the binding affinity at the human A₃ AR but showed less selectivity versus the other subtypes of ARs. In the 3-halobenzyl series, the order of binding affinity for 2-H analogues was as follows: Cl > I > Br > F, indicating that the size of halogen alone does not determine the binding affinity at the human A₃ AR. It is interesting to note that 2-H derivatives are less lipophilic than the corresponding 2-Cl derivatives, conferring more water solubility on the molecules for further biological evaluation. For example, the cLogP values of corresponding structures 6c and 7c are 1.84 and 1.12, respectively. In order to determine a stereochemical preference, the binding affinities of D-series were compared with those of L-series. As shown in Table 1, L-type nucleosides, 7f and 7g were totally devoid of binding affinities at all subtypes of ARs, indicating that the D-series induced optimal interaction with all subtypes of ARs.

Table 1.

Binding affinities of known A₃ AR agonists, 1 – 4 and antagonist 5, and truncated 4′-thioadenosine derivatives 6a – 6e and 7a – 7g at three subtypes of ARs.

Compound	Affinity, Ki, nM ± SEM (or % Inhibition at 10−5M)a,b
	hA1	hA2A	rA3	hA3c
1(IB-MECA)	51	2900	1.1	1.0
2 (Cl-IB-MECA)	222 ± 22	5360 ± 2470	0.33	1.4 ± 0.3
3 (thio-IB-MECA)	17.3	ND	1.86 ± 0.36	0.25 ± 0.06
4 (thio-Cl-IB-MECA)	193 ± 46	223 ± 36	0.82 ± 0.27	0.38 ± 0.07
5	6220 ± 640	> 10,000	321 ± 74	15.5 ± 3.1
6a (R1 = Cl, R2 = methyl)	55.4 ± 1.8	45.0 ± 1.4	658 ± 160	3.69 ± 0.25
6b (R1 = Cl, R2 = 3-fluorobenzyl)	(20%)	(48%)	36.2 ± 10.7	7.4 ± 1.3
6c (R1 = Cl, R2 = 3-chlorobenzyl)	(38%)	(18%)	6.2 ± 1.8	1.66 ± 0.90
6d (R1 = Cl, R2 = 3-bromobenzyl)	(34%)	(18%)	6.1 ± 1.8	8.99 ± 5.17
6ed (R1 = Cl, R2 = 3-iodobenzyl)	2490 ± 940	341 ± 75	3.89 ± 1.15	4.16 ± 0.50
7a (R1 = H, R2 = methyl)	1070 ± 180	(22 ± 5%)	(28 ± 10%)	4.8 ±1.7
7b (R1 = H, R2 = 3-fluorobenzyl)	1430 ± 420	1260 ± 330	98 ± 28	7.3 ± 0.6
7c (R1 = H, R2 = 3-chlorobenzyl)	860 ± 210	440 ± 110	17 ± 5	1.5 ± 0.4
7d (R1 = H, R2 = 3-bromobenzyl)	790 ± 190	420 ± 32	6.3 ± 1.3	6.8 ± 3.4
7e (R1 = H, R2 = 3-iodobenzyl)	530 ± 97	45.0 ± 1.4	658 ± 160	3.69 ± 0.25
7f (R1 = Cl, R2 = 3-bromobenzyl)	(6.1%)	(45.7%)	ND	(12.6%)
7g (R1 = Cl, R2 = 3-iodobenzyl)	(−8.0%)	(−0.95%)	ND	(18.4%)

ND: Not determined.

^aAll binding experiments were performed using adherent mammalian cells stably transfected with cDNA encoding the appropriate human AR (A₁ AR and A₃ AR in CHO cells and A_2A AR in HEK-293 cells) or the rat A₃ AR (CHO cells). Binding was carried out using 1 nM [³H]CCPA, 10 nM [³H]CGS-21680, or 0.5 nM [¹²⁵I]I-AB-MECA as radioligands for A₁, A_2A, and A₃ ARs, respectively. Values are expressed as mean ± sem, n = 3–4 (outliers eliminated), and normalized against a non-specific binder, 5′-N-ethylcarboxamidoadenosine (NECA, 10 μM). Data for compounds 6a – 6e at the human ARs and compound 6e at the rat A₃ AR were reported in ref. 1.

^bWhen a value expressed as a percentage refers to percent inhibition of specific radioligand binding at 10 μM, with nonspecific binding defined using 10 μM NECA.

^cA functional assay was also carried out at this subtype: percent inhibition at 10 μM forskolin-stimulated cyclic AMP production in CHO cells expressing the human A₃ AR, as a mean percentage of the response of the full agonist 3 (n = 1 – 3). None of the analogues 5 – 7 activated the hA₃AR (>10% of full agonist effect) by this criterion.

^dCompound 6e at 10 μM displayed <10% of the full stimulation of cyclic AMP production, in comparison to 10 μM NECA; no inhibition of the stimulatory effect of 150 nM NECA in CHO cells expressing human A_2B AR (ref. 1).

In order to determine if all final nucleosides show species-independent binding affinity at the A₃ AR, their binding affinity at the rat A₃ AR expressed in CHO cells was also measured (Table 1). As expected, most of compounds exhibited species-independent binding affinity, indicating that they are suitable for evaluation in small animal models or for further drug development Among the 2-H nucleoside analogues tested, a N⁶-(3-bromobenzyl) derivative 7d exhibited the most potent binding affinity at the rat A₃ AR (K_i = 6.3 ± 1.3 nM) followed by N⁶-(3-chlorobenzyl) derivative 7c, N⁶-(3-iodobenzyl) derivative 7e, and N⁶-(3-fluorobenzyl) derivative 7b. N⁶-Methyl derivative 7a was totally devoid of A₃AR binding affinity in this species. In the 2-Cl-N⁶-substituted adenosine series, the binding affinity was in the following order: I > Br ≈ Cl > F > Me. The 2-Cl derivatives generally showed more potent and species-indepndent binding affinity than the corresponding 2-H analogues. Compound 6e exhibited the highest binding affinity at the rat A₃ AR (K_i = 3.89 ± 1.15 nM) among all compounds tested and was inactive as agonist or antagonist in a cyclic AMP functional assay^16,17 at the hA_2B AR. It is interesting to note that N⁶-methyl derivatives 6a and 7a showing high binding affinities (K_i = 3.69 ± 0.25 nM and 4.8 ± 1.7 nM, respectively) at the human A₃ AR lost their binding affinities at the rat A₃ AR, indicating that there must be a larger N⁶ substituent for species-independent binding affinity at the A₃ AR.^18,19

In a functional assay, percent inhibition at 10 μM forskolin-stimulated cyclic AMP production in CHO cells expressing the human A₃ AR was measured as a mean percentage of the response of the full agonist 3 (n = 1 – 3). None of the analogues 6 and 7 activated the human A₃ AR (> 10% of full agonist effect) by this criterion.

Conclusion

We have established structure-activity relationships of novel truncated D- and L-4′-thionucleoside analogues as potent species-independent A₃ AR antagonists. The glycosyl donors 8 and 12 were efficiently synthesized from D-mannose and D-gulonic γ-lactone, respectively, using ring closure of dimesylate with sodium sulfide and one step conversion of the diol into the acetate with lead tetraacetate as key steps. Among the novel 4′-truncated 2-H nucleosides tested, D-N⁶-(3-halobenzyl) derivatives 7b – 7e exhibited high binding affinities at the human A₃ AR as well as at the rat A₃ AR with very low binding affinities at the human A₁ and A_2A ARs and a N⁶-(3-chlorobenzyl) derivative 7c was the most potent at the human A₃ AR, but at the rat A₃ AR 3-bromobenzyl derivative 7d was the most potent. Among both 2-H and 2-Cl analogues tested, 2-chloro-N⁶-(3-iodobenzyl) derivative 6e was found to exhibit the most potent binding affinity at the rat A₃ AR. Since this class of potent nucleoside human A₃ AR antagonists showed species-independence in interaction at this AR subtype, they are regarded as good candidates for efficacy evaluation in small animal models and for further drug development.

Experimental Section

General methods

Melting points are uncorrected. ¹H NMR (400 MHz) and ¹³C NMR (100 MHz) spectra were measured in CDCl₃, CD₃OD or DMSO-d₆, and chemical shifts are reported in parts per million (δ) downfield from tetramethylsilane as internal standard. Column chromatography was performed using silica gel 60 (230–400 mesh). Anhydrous solvents were purified by the standard procedures. cLogP values were calculated using ChemDrawUltra, version 11.0 (CambridgeSoft).

Synthesis

6-Chloro-9-((3aR,4R,6aS)-2,2-dimethyltetrahydrothieno[3,4-d][1,3]dioxol-4-yl)-9H-purine (9)

6-Chloropurine (3.91 g, 25.3 mmol), ammonium sulfate (84 mg, 0.63 mmol) and HMDS (50 mL) were refluxed under inert and dry conditions overnight. The solution was evaporated under high vacuum. The resulting solid was re-dissolved in 1,2-dichloroethane (20 mL) cooled in ice. The solution of 8¹ (2.76 g, 12.6 mmol) in 1,2-dichloroethane (20 mL) was added to this mixture dropwise. TMSOTf (4.6 mL, 25.3 mmol) was added dropwise to the mixture. The mixture was stirred at 0 °C for 30 min, at rt for 1 h, and then heated at 80 °C for 2 h. The mixture was cooled, diluted with CH₂Cl₂, and washed with saturated NaHCO₃ solution. The organic layer was dried with anhydrous MgSO₄ and evaporated under reduced pressure. The yellowish syrup was subjected to a flash silica gel column chromatography (CH₂Cl₂:MeOH = 50:1) to give 9 (3.59 g, 90%) as a foam: [α]^23.6_D-157.63 (c 0.144, DMSO); FAB-MS m/z 313 [M+H]⁺; UV (MeOH) λ_max 265.0 nm; ¹H NMR (CDCl₃) δ 8.67 (s, 1 H), 8.23 (s, 1 H), 5.88 (s, 1 H), 5.25-5.19 (m, 1 H), 3.69 (dd, 1 H, J = 4.0, 13.2 Hz), 3.18 (d, 1 H, J = 12.8 Hz), 1.51 (s, 3 H), 1.28 (s, 3 H). ¹³C NMR (CDCl₃) δ 152.0, 151.4, 151.1, 144.3, 132.6, 111.9, 89.6, 84.3, 70.3, 40.8, 26.4, 24.6. Anal. (C₁₂H₁₃ClN₄O₂S) C, H, N, S.

(2R,3R,4S)-2-(6-Chloro-9H-purin-9-yl)-tetrahydrothiophene-3,4-diol (10)

2 N Hydrochloric acid (12 mL) was added to a solution of 9 (2.59 g, 8.28 mmol) in THF (20 mL), and the mixture was stirred at room temperature overnight. The mixture was neutralized with 1 N NaOH solution, and then the volatiles were carefully evaporated under reduced pressure. The mixture was subjected to a flash silica gel column chromatography (CH₂Cl₂:MeOH = 20:1) to give 10 (1.79 g, 79%) as a white solid: [α]^23.5_D-109.14 (c 0.164, DMSO); FAB-MS m/z 273 [M+H]⁺; mp 192.3–192.8 °C; UV (MeOH) λ_max 264.5 nm; ¹H NMR (DMSO-d₆) δ 9.02 (s, 1 H), 8.81 (s, 1 H), 6.02 (d, 1 H, J = 7.2 Hz), 5.62 (d, 1 H, J = 6.0 Hz, D₂O exchangeable), 5.43 (d, 1 H, J = 4.1 Hz, D₂O exchangeable), 4.74-4.70 (m, 1 H), 4.40-4.36 (m, 1 H), 3.47 (dd, 1 H, J = 4.0, 11.2 Hz), 2.83 (dd, 1 H, J = 2.8, 11.2 Hz). ¹³C NMR (DMSO-d₆) 152.1, 151.6, 149.2, 146.6, 131.3, 78.6, 72.1, 62.4, 34.7. Anal. (C₉H₉ClN₄O₂S) C, H, N, S.

General procedure for the synthesis of 7a – 7e

To a solution of 10 in EtOH (5 mL) was added appropriate amine (1.5 equiv) at room temperature and the mixture was stirred at rt for a time period ranging from 2 h to 3 d and evaporated. The residue was purified by a flash silica gel column chromatography (CH₂Cl₂:MeOH = 20:1) to give 7a – 7e.

(2R,3R,4S)-Tetrahydro-2-(6-(methylamino)-9H-purin-9-yl)thiophene-3,4-diol (7a)

83% yield; [α]^22.8_D-175.60 (c 0.123, DMSO); FAB-MS m/z 268 [M+H]⁺; mp 223.9–224.8 °C; UV (MeOH) λ_max 266.0 nm; ¹H NMR (DMSO-d₆) δ 8.40 (s, 1 H), 8.23 (s, 1 H), 7.72 (br s, 1 H, D₂O exchangeable), 5.89 (d, 1 H, J = 7.2 Hz), 5.51 (d, 1 H, J = 6.4 Hz, D₂O exchangeable), 5. 32 (d, 1 H, J = 4.4 Hz, D₂O exchangeable), 4.70-4.64 (m, 1 H), 4.37-4.33 (m, 1 H), 3.40 (dd, 1 H, J = 4.0, 10.8 Hz), 2.95 (s, 3 H), 2.79 (dd, 1 H, J = 3.2, 10.8 Hz). ¹³C NMR (DMSO-d₆) δ 154.9, 152.5, 148.8, 139.5, 119.5, 78.3, 72.2, 61.5, 34.6, 27.0. Anal. (C₁₀H₁₃N₅O₂S) C, H, N, S.

(2R,3R,4S)-2-(6-(3-Fluorobenzylamino)-9H-purin-9-yl)tetrahydrothiophene-3,4-diol (7b)

82% yield; [α]^23.7_D-141.22 (c 0.114, DMSO); FAB-MS m/z 362 [M+H]⁺; mp 180.5–180.7 °C; UV (MeOH) λ_max 273.5 nm; ¹H NMR (DMSO-d₆) δ 8.45 (s, 1 H), 8.43 (br s, 1 H, D₂O exchangeable), 8.21 (s, 1 H), 7.36-7.30 (m, 1 H), 7.18-7.11 (m, 2 H), 7.03 (dt, 1 H, J = 2.4, 8.4 Hz), 5.90 (d, 1 H, J = 7.2 Hz), 5.53 (d, 1 H, J = 6.4 Hz, D₂O exchangeable), 5.35 (d, 1 H, J = 4.0 Hz, D₂O exchangeable), 4.70-4.66 (m, 2 H), 4.36-4.33 (m, 1 H), 3.41 (dd, 1 H, J = 4.0, 10.8 Hz), 3.17 (d, 1 H, J = 5.2 Hz), 2.79 (dd, 1 H, J = 2.8, 10.8 Hz). ¹³C NMR (DMSO-d₆) δ 163.4, 160.9, 152.4, 143.2, 140.0, 130.2, 130.1, 123.1, 123.1, 113.8, 113.6, 113.4, 113.2, 78.3, 72.2, 61.6, 48.6, 34.4. Anal. (C₁₆H₁₆FN₅O₂S) C, H, N, S.

(2R,3R,4S)-2-(6-(3-Chlorobenzylamino)-9H-purin-9-yl)-tetrahydrothiophene-3,4-diol (7c)

85% yield; [α]^23.9_D-162.5 (c 0.096, DMSO); FAB-MS m/z 378 [M+H]⁺; mp 165.0–165.3 °C; UV (MeOH) λ_max 274.5 nm; ¹H NMR (DMSO-d₆) δ 8.46 (s, 1 H), 8.44 (br s, 1 H, D₂O exchangeable), 8.22 (s, 1 H), 7.39-7.24 (m, 4 H), 5.90 (d, 1 H, J = 10.4 Hz), 5.53 (d, 1 H, J = 6.4 Hz, D₂O exchangeable), 5.35 (d, 1 H, J = 4.0 Hz, D₂O exchangeable), 4.71-4.67 (m, 2 H), 4.38-4.33 (m, 1 H), 3.47-3.31 (m, 2 H), 2.80 (dd, 1 H, J = 3.2, 10.8 Hz). ¹³C NMR (DMSO-d₆) δ 154.3, 152.4, 142.8, 140.0, 132.8, 130.1, 126.9, 126.6, 125.8, 78.3, 72.2, 61.6, 56.0, 34.4. Anal. (C₁₆H₁₆ClN₅O₂S) C, H, N, S.

(2R,3R,4S)-2-(6-(3-Bromobenzylamino)-9H-purin-9-yl)-tetrahydrothiophene-3,4-diol (7d)

71% yield; [α]^23.7_D-100.71 (c 0.139, DMSO); FAB-MS m/z 422 [M]⁺; mp 183.0–184.0 °C; UV (MeOH) λ_max 270.0 nm; ¹H NMR (DMSO-d₆) δ 8.46 (s, 1 H), 8.43 (br s, 1 H, D₂O exchangeable), 8.21 (s, 1 H), 7.53 (s, 1H) 7.42-7.24 (m, 3 H), 5.90 (d, 1 H, J = 7.2 Hz), 5.53 (d, 1 H, J = 6.4 Hz, D₂O exchangeable), 5.35 (d, 1 H, J = 4.0 Hz, D₂O exchangeable), 4.71-4.66 (m, 2 H), 4.37-4.34 (m, 1 H), 3.41 (dd, 1 H, J = 4.0, 10.8 Hz), 3.06 (q, 1 H, J = 7.2 Hz). 2.79 (dd, 1 H, J = 2.8, 10.8 Hz). ¹³C NMR (DMSO-d₆) δ 154.2, 152.4, 143.0, 140.0, 130.4, 129.8, 129.4, 126.2, 121.5, 78.3, 72.2, 61.6, 45.5, 34.5. Anal. (C₁₆H₁₆BrN₅O₂S) C, H, N, S.

(2R,3R,4S)-2-(6-(3-Iodobenzylamino)-9H-purin-9-yl)-tetrahydrothiophene-3,4-diol (7e)

88% yield; [α]^23.8_D-97.08 (c 0.137, DMSO); FAB-MS m/z 370 [M+H]⁺; mp 198.8–199.8 °C; UV (MeOH) λ_max 271.5 nm; ¹H NMR (DMSO-d₆) δ 8.45 (s, 1 H), 8.43 (br s, 1 H, D₂O exchangeable), 8.21 (s, 1 H), 7.72 (s, 1 H), 7.56 (d, 1 H, J = 7.2 Hz), 7.35 (d, 1 H, J = 7.6 Hz), 7.10 (merged dd, 1 H, J = 7.6 Hz), 5.90 (d, 1 H, J = 7.2 Hz), 5.53 (d, 1 H, J = 6.4 Hz, D₂O exchangeable), 5. 35 (d, 1 H, J = 4.4 Hz, D₂O exchangeable), 4.71-4.66 (m, 2 H), 4.37-4.34 (m, 1 H), 3.41 (dd, 1 H, J = 2.8, 10.8 Hz), 3.15 (d, 1 H, J = 5.2 Hz), 2.79 (dd, 1 H, J = 2.8, 10.8 Hz). ¹³C NMR (DMSO-d₆) δ 154.2, 152.4, 149.2, 142.9, 140.0, 137.0, 135.7, 135.3, 130.4, 126.6, 94.7, 78.3, 72.2, 61.6, 42.2, 34.4. Anal. (C₁₆H₁₆IN₅O₂S) C, H, N, S.

L-4-Thiosugar acetate 12 was synthesized from D-gulonic acid γ-lactone according to a similar procedure^1,13 used for the preparation of 8 (Scheme 1). Then L-4-thiosugar acetate 12 was converted to 13 according to a similar procedure used for the preparation of 9. The final L-4′-thio nucleosides 7f and 7g were synthesized from 12 according to the described general procedure for the synthesis of 7a – 7e.

The ¹H, ¹³C NMR, UV, and mp data of L-series compounds were the same as for the D-series of compounds as described above, except that the specific optical rotations were in the opposite direction. Yields of the L-series compounds were comparable with those of the D-series of compounds.

Binding assays^1,6

Human A₁ and A_2A ARs

For binding to human A₁ AR, [³H]CCPA (1 nM) was incubated with membranes (40 μg/tube) from CHO cells stably expressing human A₁ ARs at 25 °C for 60 min in 50 mM Tris·HCl buffer (pH 7.4; MgCl₂, 10 mM) in a total assay volume of 200 μL. Nonspecific binding was determined using 10 μM of NECA. For human A_2A AR binding, membranes (20 μg/tube) from HEK-293 cells stably expressing human A_2A ARs were incubated with 15 nM [³H]CGS21680 at 25 °C for 60 min in 200 μL 50 mM Tris·HCl, pH 7.4, containing 10 mM MgCl₂. NECA (10 μM) was used to define nonspecific binding. Reaction was terminated by filtration with GF/B filters.

Human and Rat A₃ ARs

For competitive binding assay, each tube contained 100 μL of membrane suspension (from CHO cells stably expressing the human or rat A₃ AR, 20 μg protein), 50 μL of [¹²⁵I]I-AB-MECA (0.5 nM), and 50 μL of increasing concentrations of the nucleoside derivative in Tris·HCl buffer (50 mM, pH 7.4) containing 10 mM MgCl₂. Nonspecific binding was determined using 10 μM of NECA in the buffer. The mixtures were incubated at 25 °C for 60 min. Binding reactions were terminated by filtration through Whatman GF/B filters under reduced pressure using a MT-24 cell harvester (Brandell, Gaithersburgh, MD, USA). Filters were washed three times with 9 mL ice-cold buffer. Radioactivity was determined in a Beckman 5500B γ-counter.

For binding at all three subtypes, K_i values are expressed as mean ± sem, n = 3–4 (outliers eliminated), and normalized against a non-specific binder, 5′-N-ethylcarboxamidoadenosine (NECA, 10 μM). Alternately, for weak binding a percent inhibition of specific radioligand binding at 10 μM, relative to inhibition by 10 μM NECA assigned as 100%, is given.

ABBREVIATIONS

AR

adenosine receptor

CCPA

2-chloro-N⁶-cyclopentyladenosine

CHO

Chinese hamster ovary

IB-MECA

N⁶-(3-iodobenzyl)-5′-N-methylcarboxamidoadenosine

Cl-IB-MECA

2-chloro-N⁶-(3-iodobenzyl)-5′-N-methylcarboxamidoadenosine

I-AB-MECA

2-[p-(2-carboxyethyl)phenyl-ethylamino]-5′-N-ethylcarboxamidoadenosine

NECA

5′-Nethylcarboxamidoadenosine