N⁶-Substituted adenosine derivatives: selectivity, efficacy, and species differences at A₃ adenosine receptors

Abstract

The activation of the human A₃ adenosine receptor (AR) by a wide range of N⁶-substituted adenosine derivatives was studied in intact CHO cells stably expressing this receptor. Selectivity of binding at rat and human ARs was also determined. Among N⁶-alkyl substitutions, small N⁶-alkyl groups were associated with selectivity for human A₃ARs vs. rat A₃ARs, and multiple points of branching were associated with decreased hA₃AR efficacy. N⁶-Cycloalkyl-substituted adenosines were full (≤5 carbons) or partial (≥6 carbons) hA₃AR agonists. N⁶-(endo-Norbornyl)adenosine 13 was the most selective for both rat and human A₁ARs. Numerous N⁶-arylmethyl analogues, including substituted benzyl, tended to be more potent in binding to A₁ and A₃ vs. A_2AARs (with variable degrees of partial to full A₃AR agonisms). A chloro substituent decreased the efficacy depending on its position on the benzyl ring. The A₃AR affinity and efficacy of N⁶-arylethyl adenosines depended highly on stereochemistry, steric bulk, and ring constraints. Stereoselectivity of binding was demonstrated for N⁶-(R-1-phenylethyl)adenosine vs. N⁶-(S-1-phenylethyl)adenosine, as well as for the N⁶-(1-phenyl-2-pentyl)adenosine, at the rat, but not human A₃AR. Interestingly, DPMA, a potent agonist for the A_2AAR (K_i = 4 nM), was demonstrated to be a moderately potent antagonist for the human A₃AR (K_i = 106 nM). N⁶-[(1S,2R)-2-Phenyl-1-cyclopropyl]adenosine 48 was 1100-fold more potent in binding to human (K_i = 0.63 nM) than rat A₃ARs. Dual acting A₁/A₃ agonists (N⁶-3-chlorobenzyl- 29, N⁶-(S-1-phenylethyl)- 39, and 2-chloro-N⁶-(R-phenylisopropyl)adenosine 53) might be useful for cardioprotection.

1. Introduction

The structure–activity relationships for adenosine derivatives as agonists at the ARs have been studied in both binding and functional assays [1,2]. Several of the four subtypes of adenosine receptors have cytoprotective effects when activated. The activation of the A₃AR by selective agonists has been demonstrated to be both cardioprotective and cerebroprotective [3,4]. In previous studies, a number of structural determinants for A₃AR activation have been identified [5–9], leading to the general conclusion that the ability of an adenosine derivative to activate the A₃AR is highly structure sensitive. In this study, we further evaluated the binding affinity and functional properties of a wide range of N⁶-substituted adenosine derivatives at both human and rat A₃ARs stably expressed in CHO cells.

The affinity and efficacy of many of these of N⁶-substituted adenosine derivatives have been evaluated at rat A₁ and/or A_2AARs [10]. However, at A₃ARs, the structure–activity relationships of a wide variety of N⁶-substituted adenosine derivatives has not been fully evaluated. Recent studies [5,6,8,9] suggested that various adenosine derivatives, previously assumed to be full/partial agonists at A₃ARs, are indeed antagonists. Thus, the efficacy of adenosine derivatives at the A₃AR appears to be more dependent on small structural changes than at other subtypes. In an effort to develop potent and selective A₃ receptor agonists and antagonists, the affinity and/or functional potency of these compounds for human and rat A₃ARs were measured. In order to ascertain the selectivity within the same species, we also determined the binding affinity of selected adenosine derivatives at human and/or rat A₁ and A_2AARs.

2. Materials and methods

2.1. Materials

[¹²⁵I]N⁶-(4-Amino-3-iodobenzyl)adenosine-5′-N-methyluronamide ([¹²⁵I]I-AB-MECA; 2000 Ci/mmol), [³H]R-PIA (34 Ci/mmol), [³H]DPCPX (120 Ci/mmol), and [³H]cyclic AMP (40 Ci/mmol) were from Amersham Pharmacia Biotech. [³H]CGS21680 (47 Ci/mmol) was from Perkin-Elmer Life Sciences. ENBA, NECA, CPA, CHA, DPMA (N⁶-[2-(3.5-dimethoxyphenyl)-2-(2-methylphenylethyl)] adenosine), ADAC, R- and S-PIA, N⁶-benzyl-NECA, and N⁶-benzyladenosine (Bn-Ado) were purchased from Sigma-RBI. Other N⁶-substituted adenosine derivatives were the kind gift of Dr. Ray A. Olsson (University of South Florida) and Dr. John W. Daly (NIDDK). All other chemicals were from standard commercial sources and of analytical grade.

2.2. Cell culture and membrane preparation

The CHO cells expressing recombinant human and rat A₃ARs were cultured in DMEM and F12 (1:1) supplemented with 10% fetal bovine serum, 100 units/mL penicillin, 100 µg/mL streptomycin, 2 µmol/mL glutamine, and 800 µg/mL geneticin. After harvest and homogenization, cells were centrifuged at 500 g for 10 min, and the pellet was resuspended in 50 mM Tris–HCl buffer (pH 8.0) containing 10 mM MgCl₂, 1 mM EDTA. The suspension was homogenized with an electric homogenizer for 10 s, and was then re-centrifuged at 20,000 g for 20 min at 4°. The resultant pellets were resuspended in buffer in the presence of 3 U/mL adenosine deaminase, and the suspension was stored at −80° prior to the binding experiments. The membranes from rat forebrain and striatum were prepared as previously described [11]. Striatal and forebrain tissues from Wistar rats were homogenized in ice-cold 50 mM Tris–HCl buffer, pH 7.4, using an electric homogenizer. The homogenate was centrifuged at 20,000 g for 10 min at 4°, and the pellet was washed in fresh buffer. The final pellet was stored at −80° until the binding experiments. The protein concentration was measured using the Bradford assay [12].

2.3. Binding assay

For A₃AR binding experiments, the procedures used were similar to those previously described [13]. Briefly, each tube contained 100 µL of membrane suspension, 50 µL of [¹²⁵I]I-AB-MECA (final concentration 0.5 nM), and 50 µL of increasing concentrations of compounds in Tris–HCl buffer (50 mM, pH 8.0) containing 10 mM MgCl₂, 1 mM EDTA. Non-specific binding was determined using 10 µM Cl-IB-MECA. The mixtures were incubated at 25° for 60 min. Binding reactions were terminated by filtration through Whatman GF/B filters under reduced pressure using an MT-24 cell harvester (Brandell). Filters were washed three times with ice-cold buffer. Radioactivity was determined in a Beckman 5500B γ-counter. The binding of [³H]R-PIA to the rat forebrain and recombinant human A₁ARs and the binding of [³H]CGS21680 to rat striatal and recombinant human A_2AARs were performed as previously described [11].

2.4. Cyclic AMP accumulation assay

Intracellular cyclic AMP levels were measured with a competitive protein binding method [14]. CHO cells expressing recombinant human and rat A₃ARs were harvested by trypsinization. After centrifugation and resuspension in medium, cells were planted in 24-well plates in 1.0 mL medium/well. After 24 hr, the medium was removed and cells were washed three times with 1 mL/well of DMEM, containing 50 mM HEPES, pH 7.4. Cells were then treated with agonists and/or test compounds in the presence of rolipram (10 µM) and adenosine deaminase (3 U/mL).After 45 min forskolin (10 µM) was added to the medium, and incubation was continued for an additional 15 min. The reaction was terminated by removing the medium, and cells were lysed upon the addition of 200 µL/well of 0.1 M ice-cold HCl. The cell lysate was resuspended and stored at −20°. For determination of cyclic AMP production, protein kinase A (PKA) was incubated with [³H]cyclic AMP (2 nM) in K₂HPO₄/EDTA buffer (K₂HPO₄, 150 mM; EDTA, 10 mM), 20 µL of the cell lysate, and 30 µL 0.1 M HCl, or 50 µL of cyclic AMP solution (0–16 pmol/200 µL for standard curve). Bound radioactivity was separated by rapid filtration through Whatman GF/C filters and washed once with cold buffer. Bound radioactivity was measured by liquid scintillation spectrometry.

2.5. Statistical analysis

Binding and functional parameters were estimated using GraphPAD Prism software (GraphPAD). IC₅₀ values obtained from competition curves were converted to K_i values using the Cheng–Prusoff equation [15]. Data were expressed as means ± standard error.

3. Results

3.1. Activation of A₃ARs by N⁶-substituted adenosine derivatives

The activation of A₃ARs by a wide range of the N⁶-substituted, halo-containing, and other adenosine derivatives (Tables 1 and 2) was examined by measuring their effects on forskolin-stimulated cyclic AMP accumulation in CHO cells stably expressing the human A₃AR. The efficacy of each of these adenosine derivatives was evaluated at a fixed concentration of 10 µM, and in some cases full concentration–response curves were measured. As shown in Fig. 1A, the non-selective agonist NECA 56 and the selective A₃AR agonist Cl-IB-MECA 59 both inhibited maximally the forskolin-stimulated cyclic AMP production by approximately 50%. Bn-Ado 21 was demonstrated to be less efficacious, while benzyl-NECA 57 [16] showed similar maximal efficacy as the full agonist Cl-IB-MECA. The A₁AR agonist, CHA 10, was shown to be less efficacious than Cl-IB-MECA, while no significant activation by the A_2AAR agonist DPMA 46 [17] was demonstrated. 2-Cl-R-PIA 53 [18] was also less than fully efficacious.

Table 1.

Binding affinities of adenosine derivatives at rat A₁ and A_2AARs and at human and rat A₃ARs and maximal agonist effects at human A₃ARs expressed in CHO cells

Compound	N6 substitution	Ki (rA1AR) (nM)a	Ki (rA2AAR) (nM)a	Ki (rA3AR) (nM)a	Ki (hA3AR) (nM)a	Percent activation (hA3AR)
N6-Substituted analogues
Alkyl
1	CH3	60 ± 11	>10000	6390 ± 1630	9.3 ± 0.4	96 ± 3
2	CH3O	223 ± 32	>10000	997 ± 343	28.6 ± 4.7	107 ± 13
3	CH2CH3	4.9 ± 0.2	8900 ± 770	1050 ± 140	4.7 ± 1.9	102 ± 6
4	CH2C(CH3)3	17 ± 6	>10000	1870 ± 320	306 ± 80	76 ± 4
5	CH(CH3)2	1.9 ± 0.1	2030 ± 510	201 ± 38	18.3 ± 5.5	111 ± 4
6	CH(CH2CH3)2	0.8 ± 0.2	471 ± 200	147 ± 40	55.1 ± 9.5	99 ± 6
7	CH(CH((CH3)2)2	3.8 ± 0.8	2170 ± 490	753 ± 384	3760 ± 840	21 ± 2
Cycloalkyl
8	Cyclobutyl	0.7 ± 0.1	1740 ± 170	144 ± 64	6.4 ± 1.0	100 ± 7
9 CPA	Cyclopentyl	0.45 ± 0.04b	462c	240 ± 36c	72 ± 12	97 ± 4
10 CHA	Cyclohexyl	0.9 ± 0.2	514c	167 ± 26c	73 ± 23	76 ± 2
11	Cyclooctyl	1.7 ± 0.1	6200 ± 1120	498 ± 197	411 ± 55	49 ± 5
12	exo-2-Norbornyl	0.7 ± 0.2	3400 ± 140	253 ± 30	85 ± 31	114 ± 6
13 ENBA	(S)-endo-2-Norbornyl	0.34 ± 0.06	477 ± 72	282 ± 101	915 ± 299	23 ± 10
14	7-Norbornyl	0.48 ± 0.01	>10000	229 ± 76	112 ± 25	103 ± 1
15	1-Adamantyl	73 ± 3	14800 ± 3500	>10000	>10000	0
16	2-Adamantyl	46 ± 5	>10000	>10000	>10000	0
17	Cyclopropylmethyl	0.8 ± 0.3	1370 ± 410	608 ± 242	10.2 ± 4.1	108 ± 4
18	Dicyclopropylmethyl	0.8 ± 0.2	590 ± 30	772 ± 200	41.3 ± 5.3	31 ± 6
19	Cyclohexylmethyl	19 ± 7	>10000	2550 ± 1610	263 ± 73	38 ± 3
Aryl-containing
20	Phenyl	3.3 ± 0.3	663c	802 ± 279c	14.9 ± 3.1	102 ± 9
21	Benzyl	175 ± 20	285c	120 ± 20c	41.3 ± 5.3	55 ± 3
22	2-Phenylethyl	24.0 ± 8.8d	161c	240 ± 58c	2.1 ± 0.4	84 ± 5
23	2-Phenylethoxy	225 ± 45	>10000	8660 ± 2900	88.7 ± 7.4	73 ± 6
24 ADAC	4-[[[4-[[[(2-Aminoethyl)amino]-carbonyl]-methyl]aniline]-carbonyl]methyl]phenyl	0.85c	210c	185 ± 64	13.3 ± 3.0	103 ± 4
25 Metrifudil	2-Methylbenzyl	59.6 ± 14.3e	24.1 ± 1.8e	35 ± 15	47.2 ± 10.8	100 ± 3
26	2-Methoxybenzyl	36 ± 2	761 ± 460	29 ± 11	32.5 ± 4.6	81 ± 8
27	2-Fluorobenzyl	6 ± 1	551 ± 282	60 ± 8	339 ± 5	67 ± 7
28	2-Chlorobenzyl	17 ± 3	93 ± 16	13 ± 1	17.3 ± 3.2	95 ± 1
29	3-Chlorobenzyl	45 ± 10	>10000	35 ± 20	4.4 ± 1.7	80 ± 3
30	4-Chlorobenzyl	61 ± 3	5120 ± 1230	96 ± 38	47.5 ± 4.1	96 ± 2
31	2-Pyridylmethyl	225 ± 5	>10000	111 ± 32	115 ± 33	73 ± 15
32	3-Pyridylmethyl	115 ± 4	3220 ± 1210	288 ± 67	4.5 ± 1.1	100 ± 6
33	4-Pyridylmethyl	70 ± 5	3860 ± 1520	67 ± 9	80.1 ± 20.0	99 ± 12
34	2-Furanylmethyl	95 ± 9	>10000	301 ± 72	21.8 ± 3.9	54 ± 10
35	2-Thienylmethyl	36 ± 7	734 ± 60	112 ± 33	59.8 ± 25.7	92 ± 2
36	3-Thienylmethyl	45 ± 9	3300 ± 450	297 ± 53	26.3 ± 8.2	97 ± 8
37	1-Naphthylmethyl	13 ± 3	1120 ± 580	2.7 ± 0.5	25.2 ± 9.7	67 ± 8
38	R-1-Phenylethyl	3.4 ± 0.4	1300 ± 620	60 ± 25	113 ± 22	76 ± 10
39	S-1-Phenylethyl	195 ± 20	>10000	1110 ± 460	68.7 ± 17.3	83 ± 5
40	R-1-Indanyl	65 ± 10	2480 ± 740412f	79 ± 12	233 ± 27	82 ± 5
41 R-PIA	R-1-Phenyl-2-propyl	1.2 ± 0.1	124c	158 ± 52c	8.7 ± 0.9	102 ± 6
42 S-PIA	S-1-Phenyl-2-propyl	49.3c	1820c	920 ± 311c	68 ± 12	97 ± 3
43	R-1-Phenyl-2-pentyl	3.34 ± 0.66d	>10000	76 ± 18	70.9 ± 26.2	92 ± 9
44	S-1-Phenyl-2-pentyl	282 ± 17d	>10000	1810 ± 530	37 ± 13	101 ± 10
45	R-1-Phenyl-isopentyl	8.89 ± 1.97d	3250 ± 580	103 ± 43	96 ± 17	77 ± 4
46 DPMA	2-(3,5-Dimethoxy-phenyl)-2-(2-methylphenylethyl	142c	4.4c	3570 ± 1700c	106 ± 22	0
47	(1R,2S)-2-Phenyl-1-cyclopropyl	15.2 ± 3.2d	3040 ± 490	358 ± 33	24.1 ± 10.9	87 ± 4
48	(1S,2R)-2-Phenyl-1-cyclopropyl	11.8 ± 2.4d	560 ± 232	694 ± 157	0.63 ± 0.17	117 ± 9
49	cis-(1R,2R)-2-Phenylcyclohexyl	15 ± 6	>10000	1170 ± 30	1450 ± 241	73 ± 12
50	trans-(1R,2S)-2-Phenylcyclohexyl	6 ± 3	672 ± 51	279 ± 41	559 ± 96	72 ± 9
Compound	Substitution	Ki (rA1AR) (nM)	Ki (rA2AAR) (nM)	Ki (rA3AR) (nM)	Ki (hA3AR) (nM)	Percent activation (hA3AR)
Other halo analogues
51 CADO	2-Chloro	6.7 ± 1.0b	63e	1890 ± 900e	87 ± 24	100 ± 7
52 FADO	2-Fluoro	68.3 ± 18.9d	28c	4590 ± 2410	99 ± 13	31 ± 3
53	2-Chloro-R-PIA	0.86 ± 0.14d, 1.4 ± 0.1g	1070 ± 250	34 ± 10	13.1 ± 0.9	76 ± 13
54	5′-Chloro-5′-deoxyadenosine	20 ± 1b	62.7 ± 14.4	4590 ± 2410	107 ± 6	9 ± 4

^aAll A₃AR experiments were performed using adherent CHO cells stably transfected with cDNA encoding the human or rat A₃ receptor. Percent activation of the human A₃AR was determined at 10 µM. Unless otherwise noted, K_i values at A₁AR are from Daly et al. [10]. Binding at A₁ and A_2AARs was carried out as described in Section 2. Values from the present study are means ± SEM, N = 3–5.

^bData from Daly and Padgett [19].

^cData from Van Galen et al. [16].

^dK_i values at the A₁AR determined in the present study.

^eData from Siddiqi et al. [20].

^fData from Trivedi et al. [17].

^gData from Thompson et al. [18].

Table 2.

Binding affinities of reference adenosine agonists at rat A₁ and A_2A receptors and at human and rat A₃ receptors and maximal agonist effects at human A₃ receptors expressed in CHO cells

Compound	Ki (rA1AR) (nM)a	Ki (rA2AAR) (nM)a	Ki (rA3AR) (nM)a	Ki (hA3AR) (nM)a	Percent activation (hA3AR)
55 CCPA	0.6b	950b	237 ± 71b	38 ± 6c	0d
56 NECA	6.3b	10.3b	113 ± 34b	35 ± 12	103 ± 6
57 N6-Bn-NECA	87 ± 14b	95 ± 25b	6.8 ± 2.5b	10.8 ± 1.5	107 ± 8
58 CGS21680	2600b	15b	584 ± 32b	114 ± 16d	98 ± 5
59 Cl-IB-MECA	820e	470e	0.33 ± 0.08e	1.4 ± 0.3d	100
60 MRS 542	18.5e	38.5e	1.41 ± 0.17e	1.8 ± 0.1d	0d

^aAll A₃AR experiments were performed using adherent CHO cells stably transfected with cDNA encoding the human or rat A₃ receptor. Binding at A₁ and A_2AARs was carried out as described in Section 2. Percent activation of the human A₃AR was determined at 10 µM. Values from the present study are means ± SEM, N = 3.

^bData from Van Galen et al. [16].

^cData from Gao and Jacobson [9].

^dData from Gao et al. [8].

^eData from Kim et al. [21].

Fig. 1.

Inhibition of forskolin-stimulated cyclic AMP production in CHO cells stably transfected with the human A₃AR, induced by various agonists (A) and by N⁶-benzyl derivatives (B). All experiments were performed in the presence of 10 µM rolipram and 3 U/mL adenosine deaminase. Forskolin (10 µM) was used to stimulate cyclic AMP levels. The level of cyclic AMP corresponding to 100% was 220 ± 30 pmol/mL. The data shown were from one experiment performed in duplicate and are typical of three independent experiments giving similar results. IC₅₀ values were (in nM): 10, CHA, 354 ± 116; 21, Bn-Ado, 120 ± 34; 28, 2-Cl-Bn-Ado, 81 ± 19; 29, 3-Cl-Bn-Ado, 103 ± 22; 30, 4-Cl-Bn-Ado, 146 ± 31; 53, Cl-R-PIA, 76± 21; 56, NECA, 62 ± 17; 57, Bn-NECA, 18.2 ± 3.4; 59, Cl-IB-MECA, 3.2 ± 1.4.

Various fluoro and chloro substitutions of the N⁶-benzyl group may increase the efficacy of Bn-Ado 21. Chloro substitutions at the 2- or 4-position converted Bn-Ado 21 into full agonists 28 and 30, while substitution at the 3-position increased efficacy only to a limited extent (Fig. 1B). The 2-fluorobenzyl substituent of 27 [19] also significantly increased the efficacy. In contrast, the chloro substituent at the 2-position of the adenine moiety might convert the full or partial agonists into antagonists, which have been demonstrated earlier [8,9].

Other N⁶-substituents that significantly decreased the efficacy (Tables 1 and 2) included: dicyclopropylmethyl 18, 2-phenylethyl 22, 2-phenylethoxy 23, 2-methoxybenzyl 26, 3-chlorobenzyl 29, 2-pyridylmethyl 31, 2-furanylmethyl 34, 1-naphthylmethyl 37 R- and S-1-phenylethyl 38 and 39, R-phenyl-isopentyl 45, and (1R,2S)-2-phenyl-1-cyclopropyl 47. The 3-pyridylmethyl derivative 32 was a potent, full agonist. 5′-Chloro-5′-deoxyadenosine 54 displayed low efficacy at the human A₃AR. It should be noted that some compounds tested in this study could not be clearly identified as full or partial agonists due to low affinity for the human A₃AR, although the efficacy observed at 10 µM was reduced. These include the following N⁶ substitutions: 2,2-dimethylpropyl 4, 2,4-dimethyl-3-pentyl 7, cyclooctyl 11, (S)-endo-2-norbornyl 13, 1-adamantyl 15, 2-adamantyl 16, cyclohexylmethyl 19, cis- and trans-(1R,2S)-2-phenyl-cyclohexyl 49 and 50.

3.2. Binding of the N⁶-substituted adenosine derivatives to A₁, A_2A, and A₃ARs

We initially measured the affinity and potency of the adenosine derivatives for the human A₃AR. We also examined the selectivity and species differences by determining the binding affinity of these compounds for rat A₁, A_2A, and A₃ARs.

Fig. 2 shows that two compounds were found to be more potent at human than at rat A₃ARs. N⁶-Methyladenosine 1 and N⁶-[(1S,2R)-2-phenyl-1-cyclopropyl]adenosine 48 were 687- and 1100-fold more potent, respectively, in binding to human A₃ARs vs. rat A₃ARs. A number of other compounds, including the simplest substituted adenosine derivatives CADO and FADO, were also found to be more potent for human than for the rat A₃AR (Table 1). Small N⁶-substituents (1–3) were consistently associated with selectivity for human A₃ARs vs. rat A₃ARs.

Fig. 2.

Competition for radioligand binding ([¹²⁵I]I-AB-MECA) at A₃ARs in two species, by two representative adenosine analogues 1 and 48, which were more potent at human than at rat A₃ARs. The procedures were described in “Section 2”. The data points are from a representative experiment performed in duplicate. The mean K_i values calculated from three independent experiments are listed in the text.

The reverse species dependence was also noted. Fig. 3 shows that 1-naphthylmethyladenosine 37 was approximately 10-fold more potent for rat than for human A₃ARs. Some other compounds, including ENBA 13 and 2-fluorobenzyladenosine 27, were also shown to be more potent for rat than for human A₃ARs (Table 1).

Fig. 3.

Competition for radioligand binding ([¹²⁵I]I-AB-MECA) at A₃ARs in two species, by an adenosine analogue which was more potent at rat than at human A₃ARs. The procedures were described in “Section 2”. The data points are from a representative experiment performed in duplicate. The mean K_i values calculated from three independent experiments are listed in the text.

Stereoselectivity of binding of adenosine derivatives has been established at A₁ and A_2AARs [10], but not fully investigated at A₃ARs. Fig. 4 shows compounds 38 and 39 N⁶-(R- and S-1-phenylethyl), as well as compounds 43 and 44 N⁶-(R- and S-1-phenyl-2-pentyl) showed stereoselectivity for rat but not for human A₃ARs. In the case of compounds 38 and 39, the potency of the R-isomer was similar for rat and human, while the S-isomer was 16-fold more potent at the human than at the rat A₃AR. Similarly, in the case of compounds 43 and 44, the potency of the R-isomer was similar for rat and human, while the S-isomer was 48-fold more potent for human than for rat A₃ARs (Table 1).

Fig. 4.

Competition for radioligand binding ([^125I]I-AB-MECA) at A₃ARs by pairs of R- and S-diastereoisomers (N⁶-(1-phenyl-2-pentyl)adenosines (A) and (B); N⁶-(1-phenylethyl)adenosines (C) and (D) at rat (A) and (C) and at human (B) and (D) A₃ARs, demonstrating that stereoselectivity of binding was species dependent. The procedures were described in “Section 2”. The data points are from a representative experiment performed in duplicate. The mean K_i values calculated from three independent experiments are listed in the text.

Several compounds, including N⁶-cyclooctyl 11, N⁶-(exo-norbornyl) 12, ENBA 13, and N⁶-(7-norbornyl) 14, were found to be over 200-fold selective for rat A₁ vs. rat A_2A and A₃ARs (Table 3). We further tested the possible selectivity of these compounds for the human A₁AR. However, as shown in Table 3, most of these compounds only showed less than 100-fold selectivity for human A₁ vs. human A_2A and A₃ARs. Surprisingly, ENBA was demonstrated to be the most potent and most selective compound for both human and rat A₁ARs. Similar to CGS21680, DPMA, shown to be selective for the rat A_2AAR, was shown to be non-selective at human A₁, A_2A, and A₃ARs (Tables 1 and 3). Compound 29 (3-chlorobenzyl) was demonstrated to be over 100-fold selective for both human and rat A₁ and A₃ vs. A_2AARs. Compounds 38 (R) and 39 (S), as well as 43 (R) and 44 (S), diastereoisomers which lost stereospecificity of binding at the human A₃AR, were further evaluated at the human A₁AR. In contrast to the human A₃AR, stereospecific binding was preserved at the human A₁AR (Table 3).

Table 3.

Comparison of binding affinity of selected adenosine derivatives at human A₁, A_2A, and A₃ARs (N⁶-substituent or compound abbreviation in parentheses)

	Ki (nM)
	hA1	hA2A	hA3
11 (cyclooctyl)	6.4 ± 1.4	>10000	411 ± 55
12 (exo-norbornyl)	2.4 ± 0.8	>10000	85 ± 31
13 ENBA	0.38 ± 0.19	>10000	915 ± 299
14 (7-norbornyl)	2.1 ± 0.5	>10000	112 ± 25
29 (3-chlorobenzyl)	34.9 ± 9.6	>10000	4.4 ± 1.7
38 (R-1-phenylethyl)	28.6 ± 9.8	1750 ± 540	113 ± 22
39 (S-1-phenylethyl)	94.5 ± 29.5	>10000	68.7 ± 17.3
43 (R-1-phenyl-2-pentyl)	16.9 ± 5.2	1120 ± 320	70.9 ± 26.2
44 (S-1-phenyl-2-pentyl)	113 ± 18	6840 ± 2350	37 ± 13
46 DPMA	168 ± 29	153 ± 26	106 ± 22
58 CGS21680	ND	43 ± 18	114 ± 16
59 Cl-IB-MECA	1240 ± 320	5360 ± 2470	1.4 ± 0.3

Values from the present study are means ± SEM, N = 3.

4. Discussion

4.1. Efficacy

It has been previously demonstrated that N⁶-(3-iodobenzyl) adenosine was a partial agonist for the human A₃AR [8]. However, it was not known whether the decrease in efficacy was due to the benzyl group or its iodo substitution. Here, we further demonstrated the N⁶-benzyl group itself decreased the maximal efficacy. Furthermore, it was demonstrated in this study that a number of other structural modifications of adenosine (Fig. 5) contributed variously to affinity and efficacy. The full agonist of the A₁AR, CHA, was shown to be a partial agonist of the A₃AR. A potent and full agonist for the A_2AAR, DPMA, was demonstrated to be a moderately potent antagonist of the A₃AR.

Fig. 5.

Structures of representative adenosine derivatives and their efficacy (% of inhibition of adenylate cyclase at 10 µM compared to the effect of 10 µM Cl-IB-MECA) as agonists of the human A₃AR.

In previous studies, it has been demonstrated that 2-chloro substitution of the adenine ring may both increase the affinity and decrease the efficacy of the adenosine derivatives for the human A₃AR [6,8,9]. Here, we further demonstrated that the chloro substituent might alternately decrease, increase, or have no effect on the efficacy of the adenosine derivatives depending on the position of substitution. For example, R-PIA was a full agonist, while a chloro group substituted at the 2-position of the adenine moiety (2-Cl-R-PIA) significantly decreased the efficacy. In contrast, a chloro substituent at the 2- (28) or 4- (30) positions of the benzyl group of Bn-Ado (21, itself having 55% of maximal efficacy) significantly increased the efficacy, converting the partial agonist 21 into a full agonist. Chloro substitution at the 3-position of the benzyl group (29) also significantly increased the efficacy, although to a lesser extent. Similarly, a fluoro substituent at the 2-position of the benzyl ring (27) also induced a modest increase of the efficacy.

In contrast to the N⁶-benzyl group, which decreased the efficacy, the N⁶-phenyl group did not significantly influence the efficacy, thus N⁶-phenyladenosine 20 was still a full agonist. Also, similar to Cl-IB-MECA [8,21], the benzyl group did not influence the efficacy of NECA, and thus, N⁶-benzyl-NECA [16] was a full agonist.

The appending of an additional cyclopropyl group to compound 17, obtaining compound 18, which only caused a slight change of its affinity, dramatically diminished its maximal A₃AR efficacy. Similarly, bridging the methylene group in compound 25 to give 40, thus introducing a ring constraint, also reduced A₃ efficacy. A comparison of compounds 17 and 19 suggested that the enlargement of a ring attached to the N⁶-methyl group appeared to lower both affinity and efficacy at the human A₃AR, while lengthening the chain between N⁶ and the phenyl group seemed to decrease efficacy, with minimum efficacy observed with compound 21. Compared with compound 22, the introduction of an oxygen (hydroxylamine linkage of compound 23) seemed to induce a larger decrease in its affinity for all three AR subtypes, and a smaller decrease of its maximal A₃AR efficacy. This is in contrast to compound 2, which showed enhanced potency at the rat A₃AR as a result of the oxygen inclusion [22]. Branching at the terminal alkyl position (compound 45 vs. compound 43) did not change the A₃ affinity but diminished the A₃ efficacy. A 5′-Cl-5′-deoxy substitution (compound 54), already reported for adenosine agonists [5,22], greatly reduced A₃AR efficacy.

4.2. Selectivity

N⁶-(2-Methylbenzyl)adenosine (25) [20] was found to have comparable affinity at rat A₁, A_2A, and A₃ARs, while N⁶-(2-methoxybenzyl)adenosine 26 was found to be selective for both A₁ and A₃ but not A_2AARs. N⁶-(2-Fluorobenzyl) adenosine (27) was >10-fold selective for A₁ over A_2A or A₃ARs, while N⁶-(2-chlorobenzyl)adenosine (28) was shown to have similar affinity for A₁ and A₃ARs. N⁶-(3-Chlorobenzyl)adenosine (29) was found to be extremely selective for A₁ and A₃ vs. A_2AARs. N⁶-(Cyclooctyl)adenosine (11), the N⁶-norbornyladenosines (12–14), and the N⁶-(1- and 2-adamantyl)adenosine derivatives (15 and 16) were demonstrated to be selective for the A₁AR.

Some compounds shown to be selective agonists for a particular adenosine receptor subtype in the rat, were less selective among human ARs. For example, DPMA was selective for the A_2AAR only in the rat. R-PIA was selective for the rat A₁AR, but had similar affinity for human A₁ and A₃ARs [2]. Compared with R-PIA (41), the substitution of the propyl group by a pentyl group (43) induced a dramatic decrease in A_2A affinity, but only minor change in A₁ and A₃AR affinity. CGS21680 58, originally shown to be a selective agonist for rat A_2AAR, was demonstrated to be equipotent at human A_2A and A₃ARs [2].

4.3. Species difference: human and rat A₃ARs

The identity in amino acid sequence between human and rat A₃ARs is only 72% [23], which is much lower that the interspecies homology of A₁, A_2A, and A_2BARs. It is known that almost all non-nucleoside antagonists of the human A₃AR showed extremely low affinity for rat A₃ receptors. For example, MRS 1220 [24], MRE 3008F20 [25], and (5-[[(4-pyridyl)amino]carbonyl]amino-8-methyl-2-(2-furyl)-pyrazolo[4,3-e]1,2,4-triazolo[1,5-c]pyrimidine hydrochloride) (K_i = 0.01 nM) [26] displayed exceptionally high affinity for the human A₃AR, but extremely low affinity for the rat A₃AR (K_i > 1000 nM). Adenosine derivatives were presumed to display less variability of affinity than non-nucleoside antagonists between human and rat A₃ARs. However, as shown in the present study, species differences in binding were characterized for many of these compounds. For example, N⁶-[(1S,2R)-2-phenyl-1-cyclopropyl]adenosine 48 was demonstrated to be over 1000-fold more potent for human than rat A₃AR. A small N⁶-substituent was demonstrated to be extremely important for the species difference between human and rat A₃ARs. For example, N⁶-methyladenosine 1 was demonstrated to be approximately 700-fold more potent at human A₃ARs vs. rat A₃ARs. A number of other compounds, including 3 and 22, were also demonstrated to be over 100-fold more potent at human than at rat A₃ARs. Interestingly, several compounds were found to be more potent at the rat A₃AR (compounds 7, 13, 27, 37, and 40). Compounds 38 and 39 (N⁶-R- and S-1-phenylethyl) and compounds 43 and 44 (N⁶-R- and S-1-phenyl-2-pentyl) showed stereoselectivity for rat but not for human A₃ARs. In each case the R-isomer was equipotent at rat and human A₃ARs, while the S-isomer was more potent at human by 20- and 50-fold for 39 and 44, respectively.

In conclusion, it was demonstrated in this study that a number of substitutions of adenosine contributed differently to affinity and efficacy. A number of full agonists for A₁ and A_2AARs were demonstrated to be partial agonists or antagonists for the hA₃AR. Among N⁶-alkyl substitutions, small N⁶-alkyl groups were associated with selectivity for human A₃ARs vs. rat A₃ARs, and multiple points of branching were associated with decreased hA₃AR efficacy. N⁶-Cycloalkyl-substituted adenosines were full (≤5 carbons) or partial (≥6 carbons) hA₃AR agonists. N⁶-(endo-Norbornyl)adenosine 13 was the most selective for both rat and human A₁ARs. Numerous N⁶-arylmethyl analogues, including substituted benzyl, tended to be more potent in binding to A₁ and A₃ vs. A_2AARs (with variable degrees of partial to full A₃AR agonism). A chloro substituent decreased the efficacy depending on its position on the benzyl ring. The A₃AR affinity and efficacy of N⁶-arylethyl adenosines depended highly on stereochemistry, steric bulk, and ring constraints. The identification of dual acting A₁/A₃ agonists, such as 29, 39, and 53, might be useful for cardioprotection [3].

Abbreviations

ADAC

N⁶-[4-[[[4-[[[(2-aminoethyl)amino]carbonyl]-methyl]aniline]carbonyl]methyl]phenyl]adenosine

Bn-Ado

N⁶-benzyladenosine

Cl-IB-MECA

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

CGS21680

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

CHA

N⁶-cyclohexyladenosine

DBXRM

N-methyl-1,3-dibutylxanthine-7-β-d-ribofuronamide

CPA

N⁶-cyclopentyladenosine

DPMA

N⁶-[2-(3.5-dimethoxyphenyl)-2-(2-methylphenylethyl)]adenosine

CADO

2-chloroadenosine

ENBA

(S)-endo-2-norbornyladenosine

FADO

2-fluoroadenosine

GPCR

G protein-coupled receptor

Tris

tris(hydroxymethyl)aminomethane

I-AB-MECA

N⁶-(4-amino-3-iodobenzyl)-5′-N-methylcarboxamidoadenosine

MRS 542

2-chloro-N⁶-(3-iodobenzyl)adenosine

MRS 1220

N-[9-chloro-2-(2-furanyl)[1,2,4]triazolo[1,5-c]quinazolin-5-yl]benzene-acetamide

NECA

5′-N-ethylcarboxamidoadenosine

PIA

N⁶-[phenylisopropyl]adenosine

RDT-39

N⁶-[(1S,2R)-2-phenyl-1-cyclopropyl]adenosine