Search for New Purine- and Ribose-Modified Adenosine Analogues as Selective Agonists and Antagonists at Adenosine Receptors

Abstract

The binding affinities at rat A₁, A_2a, and A₃ adenosine receptors of a wide range of derivatives of adenosine have been determined. Sites of modification include the purine moiety (1-, 3-, and 7-deaza; halo, alkyne, and amino substitutions at the 2- and 8-positions; and N⁶-CH₂-ring, -hydrazino, and -hydroxylamino) and the ribose moiety (2′-, 3′-, and 5′-deoxy; 2′- and 3′-O-methyl; 2′-deoxy 2′-fluoro; 6′-thio; 5′-uronamide; carbocyclic; 4′- or 3′-methyl; and inversion of configuration). (−)- and (+)-5′-Noraristeromycin were 48- and 21-fold selective, respectively, for A_2a vs A₁ receptors. 2-Chloro-6′-thioadenosine displayed a K_i value of 20 nM at A_2a receptors (15-fold selective vs A₁). 2-Chloroadenin-9-yl(β-L-2′-deoxy-6′-thiolyxofuranoside) displayed a K_i value of 8 μM at A₁ receptors and appeared to be an antagonist, on the basis of the absence of a GTP-induced shift in binding vs a radiolabeled antagonist (8-cyclopentyl-1,3-dipropylxanthine). 2-Chloro-2′-deoxyadenosine and 2-chloroadenin-9-yl(β-D-6′-thioarabinoside) were putative partial agonists at A₁ receptors, with K_i values of 7.4 and 5.4 μM, respectively. The A_2a selective agonist 2-(1-hexynyl)-5′-(N-ethylcarbamoyl)adenosine displayed a K_i value of 26 nM at A₃ receptors. The 4′-methyl substitution of adenosine was poorly tolerated, yet when combined with other favorable modifications, potency was restored. Thus, N⁶-benzyl-4′-methyladenosine-5′-(N-methyluronamide) displayed a K_i value of 604 nM at A₃ receptors and was 103- and 88-fold selective vs A₁ and A_2a receptors, respectively. This compound was a full agonist in the A₃-mediated inhibition of adenylate cyclase in transfected CHO cells. The carbocyclic analogue of N⁶-(3-iodobenzyl)adenosine-5′-(N-methyluronamide) was 2-fold selective for A₃ vs A₁ receptors and was nearly inactive at A_2a receptors.

Introduction

The adenosine receptors are members of the super-family of receptors coupled to guanyl nucleotide-binding proteins (G-proteins). The are composed of seven transmembrane helical domains.¹ Adenosine mediates a wide variety of physiological functions including vasodilatation, vasoconstriction in the kidney, cardiac depression, inhibition of lipolysis, inhibition of platelet aggregation, inhibition of lymphocyte functions, inhibition of insulin release and potentiation of glucagon release in the pancreas, inhibition of neurotransmitter release from nerve endings, stimulation of steroidogenesis, and potentiation of histamine release from mast cells.^1–5 The A₁, A_2a, A_2b, and A₃ adenosine receptors have been cloned from several species, including in each case rat, dog, mouse, and human.⁶ Activation of A₁ and A₃ receptors causes the inhibition of adenylate cyclase, activation of phospholipase C, activation of potassium channels, and inhibition of calcium channels.^4,5 The A_2a and A_2b receptors activate adenylate cyclase via G-protein coupling.⁵

Numerous structure–activity relationship studies of adenosine responses have been published (see ref 1). The objectives of these studies have been to discover highly potent analogues of adenosine, long-acting and nonmetabolizable agonists and antagonists, and analogues which are specific for a subtype of adenosine receptors. The recent discovery of the novel rat A₃ adenosine receptor subtype⁷ prompted us to undertake a detailed examination of a variety of nucleoside analogues as ligands for A₃ and other adenosine receptors. Agents selective for A₃ receptors^8–10 have promise as agents for treating ischemia of the brain and heart,^11,12 inflammation, and asthma.¹³ This study is a survey of some known and some novel modifications of adenosine to identify structural changes that result in selectivity.

Since some adenine and adenosine derivatives have been shown to be adenosine receptor antagonists,^15,26 we have examined potential new leads for adenosine derivatives as selective antagonists. A selective antagonist at the rat A₃ receptor is lacking. Several xanthine analogues that act as the potent antagonists at rat, rabbit, and human A₁ and A₂ receptors only weakly displaced the binding of radioligand from cloned rat A₃ receptors.¹⁶

Results

Chemistry

Adenosine analogues containing a variety of structural modification (compounds 1–44) were examined in this study. The affinity of the adenosine analogues at A₁, A_2a, and A₃ receptors was established in binding assays^17–19 (Table 1). Physical data for new adenosine analogues are listed in Table 2. Compounds 10b^24b and 12b⁵² were prepared according to literature-known procedures.

Table 1.

Affinities of Adenosine Derivatives in Radioligand Binding Assays at Rat Brain A₁, A_2a, and A₃ Receptors^a–^c,^g

a. Non-N6-substituted Derivatives
Compound	Structure	R2=	Ki (μM) or % displacement at 10−4 Mg
			Ki(A1)a	Ki(A2a)b	Ki(A3)c	Reference
1	A-Rib	Cl	0.0093	0.063	1.89d	9
2		-	0.226 ±0.028	0.163 ±0.040	2.48±0.36	42
3	A-Rib	F	0.0059	0.028	10.4±2.3	43
4		-	41%	22.7	31%d	9
5		-	53%h	24.1	26%	23,38
6		-	11.4h	11.5	31%	23,38
7		H	3±1%	0%	200±8	-
8		H	0.269	0.596	2.83d	9
9		H	12±4%	4.30±0.41	66±2%	28
10a		H	33.0±6.8	0.692 ±0.096	68±1%	20,24
10b		H	37±7%	41.4± 11.0	122± 17	24b
11		φ-(CH2)2 NH-	0.98e (IC50)	0.068f (IC50)		25
12a		NH2	30±1%	20.8±3.7	199±26	-
12b		NH2	0%	16±5%	198±33	52
12cj		φ-(CH2)2NH-	0.946 ±0.179	1.82±0.20i	59.2±9.2	-
13		-	21.5	59.8	61.7d	9
14j		-	4%	8%	20%	44
15a		-	>100	48%	39%d	9
15b		-	29±6%	11 ±2%	70.4±2.2	-
16j		-	0%	8%	19%	20,44
17		H	29.0	25%	10%d	9
18		H	180±40	8.54±1.77	24±1%	24
19		Cl	0.300 ±0.053	0.0198 ±0.0046	1.09±0.11	29
20		H	20%	26%	24%d	9
21		Cl	5.36±0.87	7.12±1.36	49.3±2.2	29
22		Cl	7.99±1.38	12.8±2.4	32%	29
23		Cl	0.640 ±0.107	0.897 ±0.178	82.4±5.2	30
24		H	31%	39%	28%d	9
25		Cl	7.32±0.96	20.4±3.5	207±8	46
26		-	29.0±4.4	35±3%	21%	31
27		-	101±19	55.5±2.8	16%	32
28a		H	29%	49%	43%d	9
28b		H	0%	8%	11%	-
29a		H	51.0±5.5	13%	9%	49
29b		H	3%	0%	8%	50
30j		H	1.03±0.05	5.00±1.63	11.0±1.9	34
31		H	0.063	0.0103	0.113d	9
32		-	0.051	0.58f	0.703 ±0.035	35
33		C≡C-(CH2)3CH3	0.130e	0.0022	0.0256 ±0.0032	36

b. N6-Substituted Derivatives
Compound	Structure	R6 =	Ki(A1)a	Ki(A2a)b	Ki(A3)c	Reference
34	A-Rib	H2N-	29.7±6.6	7.34±1.12	16.9±1.3	-
35		-	9.4h	10	56% (10−4)	48
36	A-Rib	φCH2-	0.120	0.285	0.120d	9
37	A-Rib		0.0596 ±0.0143	0.0241 ±0.0018	0.36±0.21	37
38	A-Rib		0.236 ±0.005	0.129 ±0.039	0.742 ±0.312	45
39	A-Rib		0.042 ±0.013	0.293 ±0.077	1.48 ±0.41	-
40		φCH2-	0.898	0.597	0.016d	9
41		φCH2-	62.4±6.1	53.6±14.6	0.604 ±0.143	21
42			8.61 ±2.55	6.22 ±2.71	0.720 ±0.250	-
43			0.054	0.056	0.0011	8
44			35.9 ± 8.3	28±5%	19.5 ±4.7	-

^aDisplacement of specific [³H]PIA binding, unless noted, in rat brain membranes expressed as K_i ± SEM in μM (n = 3–6), or as a single average value if from a literature report.

^bDisplacement of specific [³H]CGS 21680 binding, unless noted, in rat striatal membranes expressed as K_i ± SEM in μM (n = 3–6), or as a single average value if from a literature report.

^cDisplacement of specific binding of [¹²⁵I]-N⁶-(4-amino-3-iodobenzyl)adenosine-5′-(N-methyluronamide),¹⁹ unless noted, in membranes of CHO cells stably transfected with the rat A₃-cDNA expressed as K_i ± SEM in μM (n = 3–7).

^dDisplacement of specific binding of [¹²⁵I]-N⁶-(4-amino-3-iodophenethyl)adenosine⁹ in membranes of CHO cells stably transfected with the rat A₃-cDNA expressed as K_i ± SEM in μM (n = 3–5).

^eDisplacement of specific [³H]-N⁶-cyclohexyladenosine binding in rat brain membranes,

^fDisplacement of specific [³H]-5′-(N-ethylcarbamoyl)adenosine binding in rat striatal membranes.

^gWhen a percentage is given, it refers to the percent displacement of radioligand binding at 10⁻⁴ M.

^hDisplacement of specific [³H]-8-cyclopentyl-1,3-dipropylxanthine binding in rat brain membranes.

ⁱRadioligand binding increased at 10⁻⁴ M.

^jDMSO stock solution made fresh before assay.

Table 2.

Characterization of Newly Synthesized Compounds

comp no.	formula	analysis	mp (°C)
12a	C10H14N6O	C, H, N	214–215
12c	C18H22N6O3	C, H, N	200
15b	C22H29N5O3·2 H2O	C, H, N	210–213
21	C10H12ClN5O3S	C, H, N	207
22	C10H12ClN5O2S	C, H, N	171–173
39	C19H22N5O6·0.5 H2O	C, H, N	176–178
41	C19H22N6O4	C, H	125–130 (dec)
42	C16H17N6O5	C, H, N	200–203
44	C19H21IN6O3	C, H, N	foam
46	C10H8Cl2N4O	C, H, N	foam
48	C10H12ClN5O3	C, H, N	180
50	C11H19NO5	C, H	oil
β-51	C12H19NO7	C, H	oil
52	C16H18ClN5O6	C, H	foam
57	C12H14ClN5O3	C, H, N	foam
58	C17C21N3O2·H2O	C, H, N	211–212
60	C17H19Cl2N3	C, H, N	70–72
61	C16H17Cl2N3O	C, H, N	80
62	C18H23Cl2N3O2	C, H, N	52–55
64	C22H27ClN4O3·H2O	C, H, N	172–175
65	C9H13ClN4O	C, H, N	171–173
66	C15H16Cl2N6O	C, H, N
68	C10H12ClN5O	C, H, N	197–198

The synthesis of the carbocyclic derivative 12c is shown in Scheme 1. Using a procedure that has been employed to couple cyclopentyl acetates with nucleophiles in the presence of Pd(0) catalysis to give cis-oriented products,²⁰ reaction of (+)-45 with 2,6-dichloropurine gave (+)-46. Standard vicinal glycolization of (+)-46 to (−)-47 was followed by exchanging the 6-chloro group with NH₄OH to give 48. Heating 48 with phenylethylamine gave 12c.

Scheme 1.

Synthesis of Compound 12c^a

^aReaction conditions: (a) (i) NaH, 2,6-dichloropurine, THF, 0 °C; (ii) Pd(PPh₃)4, PPh₃, 45, THF, 50 °C; (b) OsO₄, N-methylmorpholine N-oxide, THF/H₂O; (c) NH₄OH, MeOH; (d) phenylethylamine, EtOH, Et₃N, reflux.

3′-C-Methyladenosine, 29a, and 4′-C-methyladenosine, 29b, were prepared according to known procedures.^49,50 The 4′-C-methyl modification was combined with other substitutions known to favor A₃ selectivity (Scheme 2). The methyl 4-methyl-D-ribofuranoside uronate 49 was prepared previously as an intermediate in the chemoenzymatic synthesis of methyl 4-methyl-D-ribofuranoside from cyclopentadiene.²¹ This methyl uronate was transformed into the corresponding nucleoside. Direct amidation proceeded smoothly with methylamine in MeOH to give the methyl amide 50, which was converted to the diacetate 51 by treatment with HCl/MeOH and acetylation with acetic anydride and pyridine. When the β-methyl glycoside 51 was subjected to Vorbrüggen conditions²² (N⁹-TMS-6-chloropurine, TMS-OTf, CH₃CN, 50 °C), a significant amount of the α-nucleoside was initially formed even though the 2-acetyl group is β-directing. Apparently the carbonyl of the uronamide can participate as an α-director. However, upon heating to 80 °C for 6–12 h, the thermodynamically more stable β-nucleoside 52 was formed in good yield as the sole product. When the displacement of the chloro group with benzylamine was conducted in methanol at 70 °C, 1–2% of the 6-methoxy derivative contaminated the product. The use of tert-butyl alcohol instead of methanol as cosolvent suppressed this type of reaction and gave clean N-benzyl-adenosinuronamide 41.

Scheme 2.

Synthesis of Uronamide 41^a

^aReaction conditions: (a) MeNH₂, MeOH; (b) (i) HCl, MeOH, (ii) Ac₂O, py; (c) N-TMS-6-Cl-purine, TMS-OTf, CH₃CN; (d) (i) NH₃, MeOH, 0 °C, (ii) BnNH₂, t-BuOH.

An N⁶-substituted 5′-uronamide, 42, was synthesized following the approach of Gallo-Rodriguez et al.⁸ via the action of furfurylmethylamine on 6-chloropurine-5′-(N-methyluronamido)riboside 2′,3′-dimethylacetal, 53. Deblocking the 2′,3′-dimethylacetal protecting group was sluggish due to the acid-sensitive nature of the furfuryl group.

The synthesis of the carbocyclic analogue (±)-44 (Scheme 3) began by heating (±)-54 with methylamine to provide (±)-55. Reaction of 55 with 5-amino-4,6-dichloropurine gave intermediate 56. Heating 56 with methyl dimethoxyacetate followed by acidic treatment gave 57. The 6-chloro group of 57 was displaced by 3-iodobenzylamine to give 44.

Scheme 3.

Synthesis of Carbocyclic IB-MECA, 44^a

^aReaction conditions: (a) MeNH₂, THF, heat; (b) 5-amino-4,6-dichloropyrimidine, n-BuOH, Et₃N; (c) (i) (MeO)₂CHOAc, reflux, (ii) 1 N HCl; (d) 3-I-PhCH₂NH₂·HCl, Et₃N, EtOH, heat.

Synthesis of 12a (Scheme 4) was accomplished by employing a standard procedure for preparing carbocyclic nucleosides.⁵⁶ Thus, reaction of (±)-trans-2-aminocyclopentanol with 2-amino-4,6-dichloropyrimidine yielded 65, which was subjected to C-5 diazo coupling with 4-(chlorophenyl)diazonium chloride to result in 66. Completion of the preparation of 12a followed a routine sequence:⁵⁶ reduction of 66 to amine 67, ring closure with trimethyl orthoformate to prepare precursor 68, and, finally, displacement of chlorine by ammonia to give 12a.

Scheme 4.

Synthesis of Compound 12a^a

^aReaction conditions: (a) (±trans-2-aminocyclopentanol,⁵⁷ reflux in 1-BuOH; (b) (4-chlorophenyl)diazonium chloride; (c) Zn in AcOH/EtOH/H₂O, reflux; (d) (MeO)₃CH, concentrated HCl; (e) NH₃ in MeOH.

The synthesis of the carbocyclic 15b began with the reaction of N-[4-(1-butyl)phenyl]guanidine nitrate with diethyl allylmalonate to give 58 (Scheme 5) plus a small amount of the isomeric 59.⁵³ To achive successful chlorination of 58 to 60, it was necessary to add tetraethylammonium chloride to the phosphorus oxychloride. In our hands, the standard ozonolysis approach for converting the allyl side chain of 60 to the acetaldehyde substituent of 61 could not be accomplished, apparantly due to susceptibility of the arylamino side chain to the reaction conditions. Thus, treatment of 60 with osmium tetraoxide in the presence of sodium periodate yielded 61 that was, in turn, readily converted into its dimethyl acetal derivative 62. Reaction of 62 with (±)-4α-amino-2β,3α-dihydroxy-1α-cyclopentanemethanol⁵⁴ provided 63. The latter product was not fully characterized but was ring closed under acidic conditions to 64 that was then converted into the target derivative 15b by heating with methanolic ammonia.

Scheme 5.

Synthesis of 2-[[4-(1-Butyl)phenyl]amino]-ara-aristeromycin, 15b^a

^aReaction conditions: (a) NaOEt, EtOH; (b) POCl₃, Et₄NCl, N,N-diethylaniline in MeCN, 70 °C then 100 °C; (c) OsO₄, NaIO₄ in MeOH, acetone, H₂O; (d) NH₄Cl, pyridinium p-toluenesulfonate in absolute MeOH; (e) (±)-4a-amino-2β,3α-dihydroxy-1α-cyclopentanemethanol⁵⁴ in n-BuOH containing ET₃N, heat; (f) 2 N HCl in 1,4-dioxane; (g) NH₃-MeOH, heat.

Biology

2-Chloroadenosine, 1, was much more potent than 1-deaza-2-chloroadenosine, 2, at A₁ receptors and nearly equipotent at A₃ receptors. Thus, the 1-deaza modification is well tolerated at rat A₃ receptors. Among 2-halo analogues, the difference in affinity between 2-chloro- and 2-fluoroadenosine was most pronounced at A₃ receptors, at which the 2-fluoro analogue 3 was 5.5-fold less potent. As shown previously, certain 8-position substitution, such as bromo, 4, was not well tolerated at A₃ receptors, and the potencies of the 8-amino and 8-butylamino analogues²³ 5 and 6, respectively, were consistent with this observation. There was a slight tendency toward A_2a selectivity among compounds 4–6.

Carbocyclic nucleosides have found wide application in the antiviral area.²⁴ At adenosine receptors, there are few reports of the affinity of carbocyclic derivatives of adenosine. Francis et al.²⁵ reported that the carbocyclic analogue of CGS 21680, an A_2a selective agonist, retained that selectivity. Racemic aristeromycin was reported to have activity at A_2b receptors.²⁶ Enantiomerically pure aristeromycin, 7, isolated from Streptomyces citricolor,²⁷ showed essentially no displacement of radioligand from A₁ or A_2a receptors but was weakly potent at A₃ receptors. The 5′-deoxy modification of adenosine, shown previously to be tolerated at A₁, A_2a, and A₃ receptors, was somewhat compatible with the carbocyclic modification at A_2a and A₃ receptors. 5′-Deoxyaristeromycin,²⁸ 9, was only 7-fold less potent than 5′-deoxyadenosine, 8, at A_2a receptors and of comparable potency to aristeromycin, 7, at A₃ receptors. Curiously, by deleting the 5′-methylene group of aristeromycin, as in compound 10a ((−)-5′-noraristeromycin), the potency was greatly enhanced at A₁ and A_2a receptors. The K_i value of 10a at A_2a receptors was 0.69 μM, and there was also selectivity vs A₁ (48-fold) and A₃ receptors. This prompted us to synthesize an analogue which contained a 2-(2-phenylethyl)amino group, found previously to be conducive to A₂ selectivity,^18,25 as in compound 11. The two sets of modifications however were not additive at A_2a receptors, and the hybrid compound 12c was nearly equipotent at A₁ and A_2a receptors. With the 2-(2-phenylethyl)amino group, there was no decrease in affinity at A₁ receptors from the 5′-norcarbocyclic modification. The same carbocyclic modification nearly abolished affinity at all adenosine receptors when combined with the 3-deaza modification (cf. 13 and 14). Didehydrodideoxy and dideoxy analogues of carbocyclic nucleosides 10b^24b and 12a were inactive at A₁, and they were marginally active at A_2a and A₃ receptors. Dideoxy-3-oxa derivative 12b⁵² was nearly inactive at A₁ receptors, but it showed very weak binding affinity to A_2a and A₃ receptors. 3-Deazaadenosine, 13, itself, was only weakly potent at the three subtypes. Similarly 7-deazaadenosine, 15a, was only marginally potent, and in combination with the 5′- and carbocyclic modifications was essentially inactive in binding. 7-Deaza-2-(phenylbutyl)-ara-aristeromycin, 15b, was only weakly potent at the three sybtypes, and there was also selectivity for A₃ vs A₁ and A_2a receptors.

β-L-Adenosine, compound 17, was shown to bind weakly to A₁ receptors.⁹ Curiously, compound 18 ((+)-5′-noraristeromycin), the L-analogue of compound 10a, showed considerable potency at A_2a (>A₁ and A₃) receptors and was 21-fold selective for A_2a vs A₁ receptors. The 6′-thio modification of adenosine²⁹ was well tolerated in receptor binding. (The position corresponding to the carbohydrate 4′-oxygen is here numbered 6′, as in carbocyclic derivatives.) Compound 19 was 3.2-fold more potent than the corresponding oxygen analogue 1 at A_2a receptors, while at A₁ receptors affinity diminished by 32-fold. Compounds 1 and 19 were of similar potency at A₃ receptors. Thus, this thio modification, which provided 15-fold selectivity in 19, may serve as a means of increasing A_2a selectivity in general. The 6′-thio and 2-chloro modifications of the arabinoside resulted in a substantial gain in potency at all three subtypes (cf. 21 and 20). Surprisingly, even the 2′-deoxy-6′-thio-L-xylofuranoside analogue 22 had potency comparable to 21 at A₁ and A_2a receptors.

Compounds 23–29 contain 2′- and 3′-ribose modifications. The 2′-deoxy-2′-fluoro modification of 2-chloroadenosine,³⁰ 23, had K_i values at A₁ and A_2a receptors in the micromolar range and was 2 orders of magnitude less potent at A₃ receptors. The degree of tolerance of the 2′-fluoro substituent (cf. compound 1) was greatest at A_2a receptors. The corresponding 2′-deoxy analogue 25⁴⁶ was less potent than 23 at A₁ and A_2a receptors by factors of 11 and 23. Thus at the 2′-position substituents in the order OH > F > H are favored in receptor binding at all three subtypes. For purposes of this comparison, it was essential to have the potency-enhancing 2-chloro substituent in order to obtain measurable K_i values (cf. compound 24, which was nearly inactive).

The 1-deaza modification of adenosine was already shown to be tolerated in receptor binding (see compound 2).⁴² Deletion of the 2′-hydroxy group of 2, resulting in compound 26, was poorly tolerated in receptor binding.^31,32 Only at A₁ receptors was there a measurable K_i value (29 μM). Deletion of both 2′- and 3′-hydroxyl groups, as in compound 27, resulted in a 3-fold loss of potency at A₁ receptors (vs 26) and an apparent slight gain in affinity at A_2a receptors.

The 2′-methoxy derivative of the A₁-selective agonist N⁶-cyclohexyladenosine (CHA) has been reported by Wagner et al.³³ to have substantial in vivo activity, perhaps as a prodrug of CHA. We have evaluated methoxy derivatives of adenosine at the three subtypes and found the 3′-derivative 28a to be nonselective and of very weak affinity and the 2′-derivative 28b to be inactive. The inclusion of methyl groups in place of hydrogen in the ribose ring was examined in compounds 29a,b. The 3′-methyl analogue 29a bound weakly with A₁ selectivity, while the 4′-methyl analogue 29b was virtually inactive.

Limited modification of the 5′-position of adenosine is tolerated at adenosine receptors. NECA, 31, has long been known as a highly potent, nonselective agonist.¹ A cytotoxic aminosulfonate derivative 30³⁴ was much less potent than NECA, with K_i values in the 1–10 μM range. 1-Deaza-NECA, 31, as reported previously,³⁵ has substantial activity at A₁ receptors, and at A_2a and A₃ receptors, the potencies vs those of NECA are diminished by factors of 56- and 6-fold. Thus, at A_2a and A₃ receptors, the 1-deaza substitution is tolerated better in the 2-chloro series (cf. compounds 1 and 2), while at A₁ receptors, the 1-deaza modification is tolerated better in the uronamide series. The addition of the 2-hexynyl group to NECA, resulting in the A_2a selective agonist compound 33 (HE-NECA), previously studied in platelets and on the coronary artery,³⁶ enhanced the A₃ potency. Thus, the order of selectivity of HE-NECA (with K_i values in nM) is A_2a (2.2) > A₃ (25.6) > A₁ (130).

Numerous modifications of the N⁶-position of adenosine have been introduced and shown to result generally, but not exclusively, in A₁ selectivity.¹ Table 1b shows a comparison of affinities of a variety of N⁶-substituted analogues. 6-Hydrazinopurine riboside, 34, had K_i values in the 10⁻⁵ M range and was nearly nonselective for A₁/A_2a/A₃ receptors. 1-Deaza-6-(hydroxylamino)-purine riboside, 35, also bound weakly but was 1 order of magnitude selective for A₁ and A_2a vs A₃ receptors. The affinity of adenosine, itself, for comparison with these compounds is difficult to evaluate, due to the presence of adenosine deaminase in the medium.

At A₃ receptors, N⁶-benzyl substituents, as in 36, have been shown to be favored. Several other known N⁶-arylmethylene analogues were studied. Metrifudil, 37, a potent hypotensive agent, which had entered human trials in the early 1970’s,³⁷ was found to be 2.5-fold selective for rat A_2a vs A₁ receptors and 15-fold selective for rat A_2a vs A₃ receptors. The affinity of 37 was enhanced 12-fold and diminished 3-fold at A_2a and A₃ receptors, respectively, by the presence of the 2-methyl group on the benzyl ring. (−)-Kinetin riboside,⁴⁵ 38, was also slightly selective for A_2a receptors, although 5-fold less potent than metrifudil. An N⁶-bicyclic methylene analogue, 39, was slightly selective for A₁ receptors.

N⁶-Benzyl-NECA was the first A₃ selective agonist identified,⁹ although it was only 14-fold selective verses either A₁ or A_2a receptors. The corresponding N-methyl analogue 40 had enhanced potency and selectivity.⁸ A 4′-methyl analogue, 41, displayed even greater selectivity, although it was 38-fold less potent than 40 at A₃ receptors. Compound 41 was 100- and 89-fold selective for A₃ vs A₁ and A_2a receptors, respectively. Curiously, the affinity of 41 is in contrast to the inactivity of the simple 4′-methyladenosine derivative 29b. The combination of the 5′-N-methyluronamide group with the N⁹-furfuryl group of kinetin riboside resulted in an A₃-selective compound, 42, of micromolar affinity.

A selective A₃ agonist that has been studied in vivo is IB-MECA, 43.^10,11 The carbocyclic analogue of IB-MECA, 44, was weakly binding and more potent at A₁ and A₃ vs A_2a receptors. IB-MECA, 43, was 18 000-fold more potent than this carbocyclic analogue at A₃ receptors.

Since many of the adenosine derivatives diverge greatly in structure from adenosine itself, it was not to be assumed that all of the compounds found to bind to the receptors were agonists. Previously, removal of the 2′ and 3′-hydroxyl groups of N⁶-substituted adenosine derivatives was shown to result in partial agonist or antagonist properties.^15,38 Therefore we tested selected analogues for A₁ agonist efficacy using a straight forward binding method. GTP shifts in the displacement curves for agonist verses antagonist have been shown to be indicative of agonist.³⁸ Thus, we examined the ability of selected analogues to displace the antagonist radioligand [³H]CPX and the degree of shift of the displacement curve in the presence of 1 mM GTP, and the results are given in Table 3. For agonists a typical shift in the K_i value is > 2-fold. Full agonists, such as (R)-PIA and CPA, under the same conditions give shifts of 5.5 and 6.0, respectively. For pure antagonists, no shift or only a very small shift is expected. For example, the value for the A₁ adenosine antagonist 9-methyladenine¹⁴ is 1.2. Compound 22 appears to be in the antagonist range, which is consistent with its highly unnatural carbohydrate moiety. Compounds 21 and 25 are predicted to be partial agonists at A₁ receptors. The 2′-deoxy 2′-fluoro analogue, compound 23, is likely a full agonist.

Table 3.

Displacement of the Antagonist Radioligand [³H]CPX and the Degree of Shift of the Displacement Curve in the Presence of 1 mM GTP

compd	Ki value (μM) or % displacement at 10−4 Ma		GTP Shiftb
	A1 (−GTP)	A1 (+GTP)
10a	40	nd
12c	35	nd
18	36	nd
21	13	30	2.3
22	8.4	12	1.4
23	1.7	8.3	4.9
25	12	21	1.7
26	31	nd
27	37	nd
28b	12	nd

^aAverage of two determinations.

nd = not determined.

^bRatio of K_i values in the presence and absence of GTP.

Compound 41, which was A₃ selective, was tested in a functional assay at rat A₃ receptors. Compound 41 inhibited adenylate cyclase (Figure 1), using membranes from CHO cells stably transfected with rat A₃ receptors. We previously demonstrated that compounds with high affinity in binding were very active in this assay of A₃-mediated inhibition of adenylate cyclase.³⁹ Compound 41 appeared to be a full agonist, with an inhibition of forskolin-stimulated adenylate cyclase of 32.7% ± 3.2% at 10⁻⁴ M and an IC₅₀ value of 10 μM. (based on a maximal inhibition of 40%). Thus it was less potent than NECA, 31, which had an IC₅₀ value of 5.6 μM, but was more selective for A₃ receptors.

Figure 1.

Inhibition of adenylate cyclase in membranes from CHO cells stably transfected with rat A₃ receptors. The assay was carried out as described in the Experimental Section in the presence of 1 mM forskolin. Each data point is shown as the mean ± SEM for four determinations. Adenosine derivatives were (number of separate experiments in parentheses) triangles, 41, N⁶-benzyl-4′-methyladenosine-5′-(N-methyluronamide); circles, 2-chloro-N⁶-(3-iodobenzyl)adenosine-5′-(N-methyluronamide); and squares, NECA. IC₅₀ values were Cl-IB-MECA, 66.8 ± 9.0 nM; NECA, 3 μM; and 41, 10 μM.

Discussion

The results of this study imply the existence of receptor binding sites for adenosine with very strict structural requirements. 1-Deaza modification of adenosine derivatives is well tolerated at A₁, A_2a, and A₃ receptors. The marked difference in affinity between 2-chloro- and 2-fluoroadenosine at rat A₃ receptor suggests preference for hydrophobic interaction near the 2-position of the purine ring. Substitution at the 8-position of adenosine resulted in agonists having decreased affinity for all the receptor subtypes.

Carbocyclic nucleosides were in general weakly A_2a selective ligands and showed poor affinity for A₃ receptors. Therefore it can be assumed that the 6′-oxygen of the ribose moiety is beneficial for binding to A₁ and A₃ receptors. Interestingly deletion of the 5′-methylene group of carbocyclic nucleosides (5′-noraristeromycin derivatives) resulted in enhanced affinity at A₁ and A_2a receptors. However, substitution at the 2-position of the purine ring of carbocyclic nucleosides did not enhance their affinity at A_2a receptors.

Replacement of the 6′-oxygen by sulfur was well tolerated at all the adenosine receptors. Adenine-β-D-arabinofuranoside, 20, was weakly potent at all the receptor subtypes.⁹ Curiously, the 2′-deoxyxylose modification combined with the 6′-thio modification resulted in enhanced binding at A₁ and A_2a receptors and a moderate affinity for A₃ receptors. In general 6′-thio and carbocyclic modifications of agonists were more suitable for increasing A_2a selectivity.

Various 2′-deoxy- and 2′,3′-dideoxyadenosine analogues were inactive in binding at rat A₃ receptors, but they showed considerable binding affinity at A₁ and A_2a receptors. Compounds 21 and 25 were putative partial agonists. 2-Chloro-2′-deoxy-1-deazaadenosine, 26, was active, possibly as a partial agonist at A₁ receptors only. Recently Sipe et al.⁴⁰ reported that therapeutic use of 2-chloro-2′-deoxyadenosine, 25, an antileukemic drug, slowed the progression of multiple sclerosis.

β-L-Adenosine, 17, bounds appreciably A₁ receptors only.⁹ However, β-L-5′-noraristeromycin, 18, was an A_2a selective ligand. Surprisingly, when 6′-thio and 2′-deoxyxylo modifications were applied to the β-L-nucleoside enantiomer, the resultant nucleoside, 22, behaved as an antagonist at A₁ receptors.

The 2-chloro 1-deaza modification was well tolerated at A_2a and A₃ receptors. Combination of the 1-deaza modification with a 5′-uronamide group leads to A₁ selectivity. Replacement of the 6-amino group by hydrazine or hydroxylamine was tolerated. Substitution at N⁶ by benzyl, furfuryl, and benzopyryl groups was well tolerated at A₁, A_2a, and A₃ receptors. N⁶-Substitution modification combined with 5′-uronamide derivatives increased A₃ selectivity. Substitution of 4′-hydrogen by a methyl group (compound 41) maintained agonist activity at A₃ receptors and selectivity. This methyl group is designed to diminish metabolic hydrolysis of the amide group in vivo to provide a longer biological half-life. The 5′-uronamide N⁶-benzyl-modified carbocyclic nucleoside 44 displaced binding at A₁ and A₃ receptors. It is to be noted that the inactivity of a singly modified derivative of adenosine, e.g., 29b, does not preclude activity in multiple modified versions of the same, e.g., 41.

Most of the compounds in this study are not substrates for adenosine deaminase. Those analogues having the N⁶-alkyl, 5′-carboxamido, 2-chloro, 3- and 7-deaza, and 8-position modifications, and the carbocyclic “nor” analogues (e.g., 9, 10a, 12c, and 16),^24a are expected to be stable to adenosine deaminase. It is possible that some of the compounds (20, 24, 28, and 29) may be substrates for adenosine deaminase and that any residual deaminase present during the binding assay may raise the observed K_i value. Compound 7 is reported to be an adenosine deaminase substrate.⁵¹ Compound 35 is an inhibitor of adenosine deaminase in the micromolar range.⁴⁸

Conclusions

The most significant new compound, N⁶-benzyl-4′-methyladenosine-5′-(N-methyluronamide), 41, was identified as an A₃ selective agonist, and it displayed a K_i value of 604 nM at A₃ receptors and was 103- and 88-fold selective vs A₁ and A_2a receptors, respectively. Compounds 1,3, and 25 were A₁ selective. Compounds 17,35, and 39 were somewhat selective for A₁ receptors. Compounds 4, 5, 22, 37, 38, and 39 were identified as slightly A_2a selective ligands. The moderately A_2a selective agonists were 9, 11, 18, 19, and 33. On the basis of intermediate GTP shifts in binding versus an antagonist radioligand, 2-chloro-2′-deoxyadenosine, 25, and 2-chloroadenine-9-(β-D-6′-thioarabinoside), 21, are putative partial agonists at A₁ receptors, with K_i values of 7.4 and 5.4 μM, respectively. Compound 22 was identified as an A₁ antagonist from this search for new leads for ligand selectivity at adenosine receptors. We are synthesizing derivatives of 22 to enhance its potency as antagonists at adenosine receptors subtypes.

Although the corresponding simple 4′-methyladenosine derivative 29b was inactive at adenosine receptors, compound 41, containing two other modifications as well, was an A₃ selective, full agonist. Thus, this study emphasizes that inactivity or low binding affinity of an adenosine derivative does not preclude activity in optimally modified derivatives of the same ligand.

Experimental Section

All chemicals were purchased from Aldrich (Milwaukee, WI) unless otherwise noted. Compounds 28b and 34 were the gift of the Drug Synthesis & Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment, NCI. Compounds 1, 7, 31, and 38 were purchased from Sigma Chemicals, St. Louis, MO. Cylinders of ammonia and methylamine were obtained from Union Carbide and Matheson, respectively. Anhydrous methyl alcohol and tert-butyl alcohol were used from freshly opened bottles as received. Column chromatography was carried out with silica gel 60 from EM Science (230–400 mesh). ¹H and ¹³C NMR spectra were recorded in the solvent and field strength noted. Chemical shifts are reported relative to the solvent standard. Elemental analyses were performed by Midwest Microlab, Indianapolis, IN, and M-H-W Laboratories, Phoenix, AZ.

(±)-(1α,2β)-2-(2,6-Diamino-9H-purin-9-yl)cyclopentan-1-ol (12a)

Compound 68 (220 mg, 0.87 mmol) in MeOH (30 mL) saturated with anhydrous NH₃ was heated in a sealed cylinder at 100 °C for 12 h. The solvent was removed under reduced pressure and the residue purified by column chromatography (CH₂Cl₂:MeOH, 9:1) to give 12a as white crystals following recrystallization from MeOH (140 mg, 69.7%): mp 214–215 °C; ¹H NMR (DMSO-d₆) δ 1.5–2.2 (m, 6 H, 3 × CH₂ of cyclopentyl), 4.20 (m, 2 H, H-1 and H-2 of cyclopentyl), 5.79 (s, 2 H, NH₂), 6.69 (s, 2 H, NH₂), 7.74 (s, 1 H, H-8 of purine); ¹³C NMR (DMSO-d₆) δ 19.90, 28.97, 31.99, 62.15, 74.93, 113.57, 136.30, 151.88, 156.04, 159.77.

(1S,2R,3S,4R)-4-[6-Amino-2-[(phenylethyl)amino]-9H-purin-9-yl]cyclopentane-1,2,3-triol (12c)

To a suspension of 48 (300 mg, 1.05 mmol) in absolute EtOH (10 mL) was added phenylethylamine (300 mg, 2.45 mmol). This mixture was heated at 90 °C for 3 days. The solvent was removed by rotary evaporation, and the residue was subjected to column chromatography on silica gel (eluent CH₂Cl₂–MeOH) to give 12c (272 mg, 70%) as a yellow solid: mp 200 °C dec; ¹H NMR (DMSO-d₆) δ 1.77–1.90 (m, 1 H, H-5′), 2.50–2.60 (m, 1 H, H-5′), 2.8 (m, 4 H, 2 × CH₂), 3.75 (m, 1 H, H-2′), 3.89 (m, 1 H, H-1′), 4.40–4.52 (m, 1 H, H-3′), 4.62–4.81 (m, 1 H, H-4′), 4.87 (d, 1 H, J = 3.6 Hz, OH), 5.01 (d, 1 H, J = 6.6 Hz, OH), 5.36 (d, 1 H, J = 4.8 Hz, OH), 7.22 (s, 2 H, NH₂), 7.48 (m, 5 H, Ph), 8.48 (s, 1 H, H-8).

(±)-1α,2β,3β,5β)-3-[2-[[4-(1-Butyl)phenyl]amino]-4-amino-7H-pyrrolo[2,3-d]pyrimidin-7-yl]-5-(hydroxymethyl)-1,2-cyclopentanediol (15b)

Into an ice-cooled solution of 64 (500 mg, 1.16 mmol) in dry MeOH (50 mL) was bubbled anydrous NH₃ until a saturated solution was obtained. The mixture was transfered to a sealed tube and heated at 125 °C for 55 h. After cooling to 0 °C, the sealed tube was opened and the volatiles were removed by evaporation in vauo. The residual solid was chromatographed on silica gel (CH₂-Cl₂–MeOH, 9:1). The product-containing fractions were evaporated to dryness, and the residue was rechromatographed on 5 g of Norit-A (CH₂Cl₂–MeOH, 6:4) to give pure 15b (454 mg, 95%): mp 210–213 °C; ¹H NMR (DMSO-d₆) δ 0.89 (t, 3 H, Me), 1.42 (m, 5 H, 2 × CH₂ and H-4′), 1.9 (m, 2 H, H-4′ and H-5′), 2.5 (m, 2 H, CH₂), 3.0–4.1 (m, 7 H, CH₂OH, H-1′, H-2′, H-3′, and 2 × OH), 5.0 (m, 1 H, OH), 6.4 (d, J = 3.8 Hz, 1 H, H-7), 7.0 (d, J = 8.5 Hz, 2 H, ArH). 7.4 (br, 3 H, H-8 and NH₂), 7.7 (d, J = 8.5 Hz, 2 H, ArH), 8.8 (br, 1 H, NH); ¹³C NMR (DMSO-d₆) δ 13.76, 21.67, 30.06, 33.23, 34.13, 46.38, 55.15, 63.38, 77.14, 78.07, 100.38, 101.09, 118.43, 120.59, 128.56, 135.65, 136.89, 147.41, 148.82, 158.35.

N⁶-(Benzodioxanylmethy)adenosine Hemihydrate (39)

6-Chloropurine riboside (200 mg, 0.70 mmol) was refluxed in 10 mL of ethanol with 162 mg (0.77 mmol) of racemic benzdioxan-2-methylamine and 2 g (2.1 mmol) of triethylamine solution (5.3 g of triethylamine in 50 g of ethanol) for 18 h. After 48 h at −20 °C, a white crystalline product was formed that was collected and dried. Further workup afforded a second crop of the product: total yield 180 mg (62%); mp 176–178 °C; MS (CI) MH⁺ = 416.

1-[6-(Benzylamino)-9H-purin-9-yl]-1-deoxy-N,4-dimethyl-β-D-ribofuranosiduronamide (41)

The diacetyl nucleoside 52 (195 mg, 0.474 mmol) was selectively deacylated by treatment with a methanolic solution of NH₃ at 0 °C for 10 min. The solution was evaporated to dryness, and to the residue was added a 1:1 solution of t-BuOH and benzylamine (4 mL). This solution was heated at 70 °C for 16 h and concentrated under reduced pressure (0.1 Torr, 40 °C). Chromatography of the residue on silica gel (gradient 30:1–10:1 CH₂Cl₂:MeOH) gave 159 mg (84%) of N-benzyladenine nucleoside 41 as a white solid: [α]²³_D–25.7° (c 1.02, CH₃OH); ¹H NMR (500 MHz, CD₃OD, prior H–D exchange) δ 8.29 (s, 1 H), 8.13 (s, 1 H), 7.35 (d, 2 H, J = 7.0 Hz), 7.28 (t, 2 H, J = 7.5 Hz), 7.21 (t, 1 H, J = 7.5 Hz), 5.97 (d, 1 H, J = 8.5 Hz), 4.83 (dd, 1H, J = 8.5, 5.0 Hz), 4.78 (br s, 2 H), 4.30 (d, 1 H, J = 5.0 Hz), 2.82 (s, 3 H), 1.50 (s, 3 H); ¹³C NMR (125 MHz, CD₃OD) δ 174.98, 154.73, 152.48, 148.33, 140.77, 138.78, 128.13, 127.10, 126.81, 120.10, 88.16, 87.74, 73.75, 71.75, 43.62, 24.90, 18.79.

1-[6-(Furfurylamino)-9H-purin-9-yl]-1-deoxy-N-methyl-β-D-ribofuranosiduronamide (42)

To a solution of 2′,3′-isopropylidene-1-(6-chloro-9H-purin-9-yl)-1-deoxy-N-methyl-β-D-ribofuranosiduronamide⁸ (50 mg, 0.16 mmol) in absolute EtOH (5 mL) was added furfurylamine (20 mg, 0.21 mmol). This mixture was heated at 90 °C for 20 h. After cooling to room temperature, solvent was removed by rotary evaporation and the residue was dissolved in 0.5 N HCl and heated at 60 °C for 1 h. After cooling to 0 °C in an ice bath, the reaction mixture was neutralized with concentrated NH₄OH and evaporated to dryness. The residue was purified by preparative thin-layer chromatography on silica gel to give 42 (21 mg, 40%) as white solid: ¹H NMR (DMSO- d₆) δ 3.32 (d, J = 4.3 Hz, 3 H, Me), 4.12 (m, 1 H, H-3′), 4.33 (s, 1 H, H-4′), 4.60 (dd, J = 4.6, J = 4.3 Hz, 1 H, H-2′), 4.70 (br s, 2 H, N⁶-CH₂-), 5.53 (d, J = 6.4 Hz, 1 H, OH-2′), 5.56 (d, J = 7.4 Hz, 1 H, H-1′), 5.71 (d, J = 4.1 Hz, 1 H, OH-3′), 5.60 (d, J = 7.4 Hz, 1 H, H-1′), 7.3 (m, 3 H), 8.42 (s, 1 H, H-8), 8.56 (br s, 1 H, H-N^e), 8.55 (br s, 1 H, NH-Me).

(±)-9-[2α,3α-Dihydroxy-4β-(N-methylcarbamoyl)cyclopent-1β-yl)]-N⁶-(3-iodobenzyl)adenine (44)

To a solution of 57 (100 mg, 0.32 mmol) in absolute EtOH (10 mL) was added 3-iodobenzylamine hydrochloride (90 mg, 0.34 mmol), and the resulting mixture was heated at 90 °C for 24 h, under a nitrogen atmosphere. The solvent was removed by evaporation in vacuo, and the residue was purified by column chromatography on silica gel (eluent CH₂Cl₂–MeOH, 10:0.5) to give 44 (140 mg, 85%) as colorless foam: ¹H NMR (DMSO- d₆) δ 2.71 (s, 2 H, CH₂), 3.31 (m, 1 H, H-1′), 3.42 (d, J = 4.3 Hz, 3 H, Me), 4.32 (m, 1 H, H-3′), 4.35 (s, 1 H, H-4′), 4.70 (s, 2 H, CH₂-Ph), 4.74 (dd, J = 4.0, J = 4.3 Hz, 1 H, H-2′), 5.45 (d, J = 6.4 Hz, 1 H, OH-2′), 5.60 (d, J = 4.1 Hz, 1 H, OH-3′), 5.60 (d, J = 7.4 Hz, 1 H, H-1′), 7.13 (t, J = 7.1 Hz, 1 H), 7.40 (d, J = 7.7 Hz, 1 H), 7.60 (d, J = 7.6 Hz, 1 H), 8.21 (s, 1 H, H-8), 8.50 (br s, 1 HN⁶), 8.60 (br s, 1 H, NH-Me).

(1R,4S)-4-(2,6-Dichloro-9H-purin-9-yl)cyclopent-2-en-1-ol (46)

To a solution of 2,6-dichloropurine (2 g, 10.64 mmol) in dry DMSO (25 mL) was added sodium hydride (60% suspension in mineral oil, 0.42 g, 10.64 mmol). The reaction mixture was stirred at the room temperature for 30 min followed by the addition of tetrakis(triphenylphosphine)-palladium (0.5 g, 0.22 mmol), triphenylphosphine (0.25 g, 0.95 mmol), and a solution of (+)-45²⁰ (1.66 g, 11.70 mmol) in dry THF (25 mL). This mixture was stirred at 50 °C for 20 h. The volatiles were removed by rotary evaporation in vacuo at 50 °C. The residue was slurried in CH₂Cl₂ (50 mL) and filtered to remove insoluble solids. The filtrate that resulted was washed with brine (2 × 50 mL), dried (MgSO₄), and evaporated to dryness. The residual oil was purified by flash chromatography on silica gel by eluting first with AcOEt to remove the nonpolar impurities and then with AcOEt–MeOH (9:1). The product containing fractions were evaporated to dryness to give 46 (3.55 g, 89%) as a colorless foam: ¹H NMR (DMSO- d₆) δ 1.62–2.7 (m, 2 H, H-5′), 3.26 (m, 3 H, OH), 3.65 (m, 1 H, H-1′), 4.58 (m, 3 H, H-2′, H-3′, H-4′), 8.34 (m, 1 H, H-8).

(1S,2R,3S,4R)-4-(6-Amino-2-chloro-9H-purin-9-yl)cyclopentane-1,2,3-triol (48)]

To a solution of 46 (1 g, 3.70 mmol) in THF-H₂O (10:1, 50 mL) was added a 60% aqueous solution of N-methylmorpholine N-oxide (1.2 mL, 1.14 mmol) and then osmium tetraoxide (30 mg). The reaction mixture was stirred at room temperature for 24 h. The solvent was removed by rotary evaporation, and the residue was coevaporated with EtOH (3 × 50 mL) to give a gummy material. This residue was dissolved in MeOH presaturated with anhydrous ammonia (50 mL) and stirred in a sealed tube at room temperature for 5 days. Volatiles were removed by rotary evaporation, and the residue was subjected to column chromatography on silica gel (eluent MeOH–CH₂Cl₂, 9:1) to give 48 (800 mg, 80%) as a white solid: mp 180 °C dec; ¹H NMR (DMSO- d₆) δ 1.77–190 (m, 1 H, H-5′), 2.50 (m, 1 H, H-5′) 3.75 (m, 1 H, H-2′), 3.89 (m, 1 H, H-1′), 4.40–4.52 (m, 1 H, H-3′), 4.62–4.81 (m, 1 H, H-4′), 4.87 (d, 1 H, J = 3.6 Hz, OH), 5.01 (d, 1 H, J = 6.6 Hz, OH), 5.36 (d, 1 H, J = 4.8 Hz, OH), 7.22 (s, 2 H, NH₂), 8.11 (s, 1 H, H-8).

Methyl 2,3-O-Isopropylidene-N,4-dimethyl-β-D-ribofuranosiduronamide (50)

A solution of methyl ester 49²¹ (502 mg, 2.04 mmol) in MeOH (30 mL) was saturated with gaseous methylamine and stirred until TLC (2:1 hexane: EtOAc) showed the reaction was complete. The solution was concentrated and the residue chromatographed on silica gel (1.5:1 hexane:EtOAc) to give 502 mg (100%) of 50 as a clear oil: [α]²³_D–45.0° (c 1.05, CH₂Cl₂); ¹H NMR (300 MHz, CDCl₃) δ 6.56 (br s, 1 H), 5.10 (d, 1 H, J = 5.9 Hz), 4.93 (s, 1 H), 4.51 (d, 1 H, J = 5.9 Hz), 3.39 (s, 3 H), 2.79 (d, 3 H, J = 5.0 Hz), 1.47 (s, 3 H), 1.44 (s, 3 H), 1.30 (s, 3 H); ¹³C NMR (75 MHz, CDCl₃) δ 174.55, 112.53, 110.25, 89.33, 85.15, 82.35, 55.96, 26.16, 25.98, 24.76, 20.96.

Methyl 2,3-O-Diacetyl-N,4-dimethyl-β-D-ribofuranosiduronamide (51)

To a solution of acetonide 50 (502 mg, 2.04 mmol) in MeOH (60 mL) was added concentrated HCl (0.1 mL). The solution was stirred for 18 h and then concentrated under reduced pressure. The residue was chromatographed on silica gel (gradient EtOAc–30% MeOH in EtOAc) to give 105 mg (21%) of starting acetonide and 310 mg of diols as a mixture of glycosides which were acetylated with Ac₂O (0.48 mL) and pyridine (0.85 mL) in CH₂Cl₂ (30 mL) containing a catalytic amount of DMAP. Toluene (10 mL) was added, and the solution was concentrated to dryness. The residue was chromatographed on silica gel (1:1 hexane:EtOAc) to give 380 mg (64%) of β-51 and 40 mg (7%) of α-51 (81% and 8.5% yields, respectively, based on recovered acetonide). β-51: oil; [α]²³_D –27.5° (c 1.2, CH₂Cl₂); ¹H NMR (300 MHz, CDCl₃) δ 6.77 (br s, 1 H), 5.54 (d, 1 H, J = 4.8 Hz), 5.19 (dd, 1 H, J = 3.2, 4.8 Hz), 5.01 (d, 1 H, J = 3.2 Hz), 3.48 (s, 3 H), 2.82 (d, 3 H, J = 5.0 Hz), 2.10 (s, 3 H), 2.05 (s, 3 H), 1.46 (s, 3 H); ¹³C NMR (75 MHz, CDCl₃) δ 173.26, 169.19 (2), 106.54, 85.69, 74.71, 73.75, 56.87, 26.11, 20.56 (2), 20.45. Anal. Calcd for C₁₂H₁₉NO₇: C, 49.82; H, 6.62. Found: C, 49.75; H, 6.52. α-51: mp 112–113 °C (CH₂Cl₂/hexane); ¹H NMR (300 MHz, CDCl₃) δ 6.71 (br s, 1 H), 5.64 (d, 1 H, J = 6.3 Hz), 5.12 (d, 1 H, J = 4.8 Hz), 4.94 (dd, 1 H, J = 6.3, 4.8 Hz), 3.42 (s, 3 H), 2.81 (d, 3 H, J = 5.0 Hz), 2.16 (s, 3 H), 2.08 (s, 3 H), 1.48 (s, 3 H); ¹³C NMR (75 MHz, CDCl₃) δ 172.92, 169.81, 169.58, 101.35, 85.59, 72.08, 70.78, 55.83, 26.08, 21.14, 20.65, 20.45.

2,3-O-Diacetyl-1-(6-chloro-9H-purin-9-yl)-1-deoxy-N,4-dimethyl-β-D-ribofuranosiduronamide (52)

A suspension of 6-chloropurine (590 mg, 3.82 mmol) in hexamethyldisilazane (6 mL) was heated to 100 °C until dissolution was complete, ca. 1 h. Toluene (2 mL) was added, and the solution was concentrated under an inert atmosphere. To remove final traces of HMDS, toluene (2 × 4 mL) was again added and the solution was concentrated in a similar manner. To the silylated 6-chloropurine in dry CH₃CN (3 mL) was added 307 mg (1.05 mmol) of β-51 (dried by azeotropic distillation with toluene under reduced pressure) in dry CH₃CN (5 mL) and trimethylsilyl trifluoromethanesulfonate (0.75 mL). This solution was heated to reflux for 12 h. The two initial nucleoside products detected by TLC gave way to a single thermodynamic product during this time. The reaction mixture was cooled, the reaction quenched by the addition of saturated aqueous NaHCO₃ (1 mL), and the mixture partitioned between CH₂-Cl₂ (40 mL) and H₂O (10 mL). The aqueous layer was extracted with CH₂Cl₂ (2 × 40 mL). The combined organics were dried (MgSO₄), filtered, and concentrated under reduced pressure. Column chromatography on silica gel (1:2 CH₂Cl₂: EtOAc) gave 270 mg (63%) of 6-Cl-purine nucleoside 52 as a faint yellow foam: [α]²³_D 2.48° (c 1.45, CH₂Cl₂); ¹H NMR (300 MHz, CDCl₃) δ 8.68 (s, 1 H), 8.27 (s, 1 H), 7.61 (br q, 1 H, J = 4.9 Hz), 6.15 (d, 1 H, J = 7.5 Hz), 6.00 (dd, 1 H, J = 7.5, 5.0 Hz), 5.82 (d, 1 H, J = 5.0 Hz), 2.77 (d, 3 H, J = 4.9 Hz), 2.13 (s, 3 H), 1.88 (s, 3 H), 1.49 (s, 3 H); ¹³C NMR (75 MHz, CDCl₃) δ 171.13, 168.78 (2), 151.80, 151.56, 150.96, 144.67, 133.39, 86.80, 86.07, 73.03, 71.15, 25.99, 20.20, 19.96, 19.75.

(±)-9-[2α,3α-Dihydroxy-4β-(N-methylcarbamoyl)cyclopent-1β-yl)]-6-chloropurine (57)

To a solution of compound 54 (1 g, 7.04 mmol, prepared according to the procedure reported by Cermak and Vince⁴⁷) in dry MeOH (20 mL) was bubbled anhydrous methylamine for 10 min. The resulting solution was heated at 90 °C in a sealed tube for 20 h. After cooling to room temperature, solvent was removed by rotary evaporation in vacuo and the residue was used in the next step without further characterization. To this residue were added 5-amino-4,6-dichloropyrimidine (1.00 g, 6.13 mmol), triethylamine (2 mL), and n-BuOH (20 mL). The resulting mixture was heated at 100 °C, under an N₂ atmosphere, for 24 h. Volatiles were evaporated in vacuo, and the residue was dissolved in methyl diethoxyacetate (10 mL). This mixture was heated at 100 °C for 2 h and then evaporated to dryness. The residue was dissolved in 1 N HCl (10 mL) and stirred at room temperature for 3 h. The ice-cold reaction mixture was neutralized with concentrated NH₄OH and evaporated to dryness. The residue was subjected to column chromatography on silica gel (eluent CH₂Cl₂–MeOH, 9.5:0.5) to give 57 (1.3 g, 59% yield based upon 54) as a yellow foam: ¹H NMR (DMSO- d₆) δ 2.75 (s, 2 H, CH₂), 3.33 (m, 1 H, H-1′), 3.40 (d, J = 4.3 Hz, 3 H, Me), 4.22 (m, 1 H, H-3′), 4.33 (s, 1 H, H-4′), 4.75 (dd, J = 4.0, J = 4.3 Hz, 1 H, H-2′), 5.45 (d, J = 6.4 Hz, 1 H, OH-2′), 5.60 (d, J = 4.1 Hz, 1 H, OH-3′), 5.60 (d, J = 7.4 Hz, 1 H, H-1′), 8.21 (s, 1 H, H-8), 8.50 (br s, 2 H, HN⁶), 8.60 (br s, 1 H, NH-Me).

5-Allyl-2-[[4-(1-butyl)phenyl]amino]-4,6-dihydroxypyrimidine (58)

N-[4-(1-Butyl)phenyl]guanidine nitrate⁵⁶ (16 g, 62.99 mmol) was added to cold NaOEt in EtOH (250 mL prepared by dissolving 25 g of Na in 250 mL of absolute EtOH). This mixture was stirred at 5 °C for 10 min before diethyl allylmalonate (12.59 g, 62.95 mmol; Aldrich Chemical Co.) was added. The reaction mixture was then refluxed for 3 h. Concentrated HCl (40 mL) was added to the cold reaction mixture, which was allowed to stand for several hours. The crude product and inorganic salts precipitated. After filtration and washing the crude mixture with water and recrystallization of this material using EtOH, 12.5 g (66%) of pure 58 was obtained: mp 211–212 °C; ¹H NMR (DMSO- d₆) δ 0.81 (t, 3 H, CH₃), 1.39 (m, 4 H, 2 × CH₂), 2.44 (t, 2 H, CH₂), 2.90 (d of t, 2 H, CH₂), 3.03 (br, 1 H, OH), 4.86 (m, 2 H, CH₂), 5.61 (m, 1 H, CH), 7.09 (d, 2 H of C₆H₅), 7.62 (d, 2 H of C₆H₅), 8.57 (s, 1 H, side chain NH), 10.50 (br, 1 H, ring OH/NH); ¹³C NMR (DMSO- d₆) δ 13.77, 21.73, 26.39, 33.32, 34.25, 90.75, 113.66, 119.89, 128.51, 136.20, 136.80, 149.58, 165.08.

5-Allyl-2-[[4-(1-butyl)phenyl]amino]-4,6-dichloropyridimidine (60)

To a suspension of dried 58 (10 g, 33.44 mmol) in dry acetonitrile (150 mL) was added tetraethylammonium chloride (dried over P₂O₅) (2 g, 17.09 mmol) followed by N,N-diethylaniline (2.5 mL) and phosphorus oxychloride (30 mL). The reaction mixture was kept in an oil bath preheated at 70 °C for 2 h. An additional amount of phosphorus oxychloride (20 mL), containing N,N-diethylaniline (1.5 mL), was added, and the temperature was raised to 100 °C. After 2 h, the volatiles were evaporated under reduced pressure and the residue was dissolved in CHCl₃ (100 mL). The organic solution was added to ice H₂O, and the layers were separated. The aqueous layer was extracted with CHCl₃ (3 × 50 mL), and the combined organic extracts were washed with saturated aqueous NaHCO₃ solution (2 × 50 mL) and then H₂O (2 × 50 mL). The CHCl₃ solution was dried over MgSO₄ and filtered, and the filtrates were evaporated on a rotary evaporator to yield crude 60. Recrystallization of this material using MeOH afforded 7.72 g (69%) of pure 60: mp 70–72 °C; ¹H NMR (CDCl₃) δ 0.89 (t, 3 H, CH₃), 1.25 (m, 4 H, 2 × H₂), 2.48 (m, 2 H, CH₂), 3.97 (m, 2 H, CH₂), 5.03 (m, 2 H, CH₂), 6.00 (m, 1 H, CH₂), 7.36 (m, 5 H, C₆H₅ and NH); ¹³C NMR (CDCl₃) δ 13.89, 22.23, 33.23, 33.61, 34.96, 116.71, 119.58, 128.79, 132.42, 135.67, 138.22, 156.91, 161.78.

[2-[[4-(1-Butyl)phenyl]amino]-4,6-dichloropyrimidin-5-yl]acetaldehyde (61)

To a suspension of 60 (37.12 g, 110.81 mmol) in MeOH (300 mL), acetone (200 mL), and H₂O (200 mL) were added OsO₄ (230 mg) and NaIO₄ (94.8 g, 443.21 mmol). After mechanically stirring for 48 h, the reaction mixture was evaporated to remove the organic solvent. The suspension was diluted with H₂O (500 mL) and extracted several times with EtOAc. The EtOAc layer was separated, dried over MgSO₄, and evaporated to dryness. The crude product was recrystallized from a small amount of MeOH to give 61 (29.83 g, 79%) as white needles: ¹H NMR (DMSO- d₆) δ 0.91 (t, 3 H, CH₃), 1.40 (m, 4 H, 2 × CH₂), 1.99 (t, 2 H, CH₂), 3.90 (s, 2 H, CH₂CHO), 7.16 (d, J = 8.5 Hz, 2 H, ArH), 7.4 (d, J = 8.5 Hz, 2 H, ArH), 9.7 (s, 1 H, NH), 10.25 (s, 1 H, CHO); ¹³C NMR (DMSO- d₆) δ 13.87, 21.71, 33.20, 34.24, 43.58, 112.65, 119.66, 128.38, 136.36, 137.01, 158.81, 161.37, 197.79.

2-[[4-(1-Butyl)phenyl]amino]-4,6-dichloro-5-(2,2-dimethoxyethyl)pyrimidine (62)

To a solution of 61 (29.25 g, 86.79 mmol) in dry MeOH (250 mL) was added pyridinium p-toluensulfonate (2.18 g, 8.67 mmol) and NH₄Cl (460 mg, 8.68 mmol, freshly dried over P₂O₅), and this solution was refluxed for 4 h. After cooling to room temperature, the reaction mixture was treated with Norit-A and filtered through a pad of Celite, and the filtrate was evaporated to dryness. The residue was dissolved in AcOEt–hexane (2:8) and filtered through a plug of Florisil (150 g). Removal of the solvent and chromatography on Florisil (AcOEt–hexane, 2:8) gave 62 (29.71 g, 89%) as a yellow oil, which crystallized slowly at room temperature and was recrystallized from hexane to give pure 62 as colorless needles: mp 52–55 °C; ¹H NMR (CDCl₃) δ 0.93 (t, 3 H, Me), 1.3 (m, 4 H, 2 × CH₂), 2.5 (t, 2 H, CH₂), 3.0 (d, 2 H, CH₂), 3.3 (s, 6 H, OMe), 4.5 (t, 1 H, (MeO)₂CH), 7.1 (d, J = 8.5 Hz, 2 H, ArH), 7.2 (s, 1 H, NH), 7.4 (d, J = 8.5 Hz, 2 H, ArH); ¹³C NMR (CDCl₃) δ 13.83, 22.39, 33.55, 33.72, 35.12, 54.14, 103.38, 116.93, 119.74, 129.01, 135.73, 138.43, 157.02, 162.43.

(±)-(1α,2β,3β,5β)-3-[2-[[4-(1-Butyl)phenyl]amino]-4-chloro-7H-pyrrolo[2,3-d]pryimidin-7-yl]-5-(hydroxymethyl)-1,2-cyclopentanediol (64)

A mixture of (±)-4α-amino-2β,3β-dihydroxy-1α-cyclopentanemethanol (obtained by the acidic hydrolysis of 3.5 g (11.18 mmol) of its tetraacetate derivative⁵⁴), 62 (3.0 g, 8.9 mmol), and Et₃N (20 mL) in 1-BuOH (100 mL) was heated under reflx for 38 h. The reaction mixture was evaporated to dryness to give crude 63. A small amount of 63 was purified by column chromatography (CH₂Cl₂–MeOH, 95:5): ¹H NMR (DMSO- d₆) δ 0.89 (t, 3 H, Me), 1.42 (m, 5 H, 2 × CH₂ and H-4), 2.0 (m, 2 H, H-4 and H-5), 2.5 (t, 2 H, CH₂), 2.8 (d, 1 H, H-3), 3.3 (s, 6 H, 2 × OMe), 3.4–4.0 (m, 6 H, (MeO)₂CHCH₂-, CH₂OH, H-1, and H-2), 4.2–5.2 (m, 4 H, (MeO)₂CH amd 3 × OH), 6.5 (d, 1 H, NH), 7.0 (d, J = 8.5 Hz, 2 H, ArH), 7.6 (d, J = 8.5 Hz, 2 H, ArH), 9.2 (s, 1 H, NH); ¹³C NMR (DMSO- d₆) δ 12.57, 20.53, 30.50, 32.18, 33.35, 45.51, 51.57, 52.82, 53.20, 62.41, 75.74, 77.09, 99.09, 103.53, 117.34, 126.93, 133.59, 137.17, 156.38, 161.22.

The crude 63 from the above step was dissolved in 1,4-dioxane (200 mL), and 2 N HCl (20 mL) was added to this. After stirring for 20 h at room temperature, the reaction mixture was neutrallized with concentrated NH₄OH followed by evaporation to dryness. The residue was purified by medium pressure chromatography (CH₂Cl₂–MeOH, 95:5) to give 64 (1.6 g, 42% based upon 62) as a yellow foam, which crystallized from CH₂Cl₂: mp 172–175 °C; ¹H NMR (DMSO- d₆) δ 0.89 (t, 3 H, Me), 1.41 (m, 5 H, 2 × CH₂ and H-4′), 2.0 (m, 2 H, H-4′ and H-5′), 2.5 (m, 2 H, CH₂), 3.1 (d, 1 H, H-3′), 3.2–4.0 (m, 6 H, CH₂OH, H-1′, H-2′, and 2 × OH), 5.1 (d, 1 H, OH), 6.36 (d, J = 3.8 Hz, 1 H, H-7), 7.0 (d, J = 8.5 Hz, 2 H, ArH), 7.38 (d, J = 3.5 Hz, 1 H, H-8), 7.7 (d, J = 8.5 Hz, 2 H, ArH), 9.5 (br, 1 H, NH); ¹³C NMR (DMSO- d₆) δ 13.97, 21.88, 29.74, 33.53, 34.40, 45.72, 55.31, 63.44, 77.14, 78.23, 98.00, 110.14, 118.48, 127.25, 128.45, 135.00, 138.58, 150.61, 152.88, 154.83.

(±)-(1α,2β)-2-(2-Amino-6-chloro-9H-purin-9-yl)cyclopentan-1-ol (68)

A solution of 18.64 g (114 mmol) of 2-amino-4,6-dichloropyrimidine and 11.50 g (114 mmol) of (±)-trans-2-aminocyclopentanol⁵⁷ in BuOH was heated at reflux under Ar. Following the reflux period, the solvent was removed by distillation under reduced pressure and the residue treated with H₂O. The resulting beige solid was collected by filtration and washed well with H₂O followed by CHCl₃. The solid that remained was dried at 60 °C under vacuum to give 18.13 g (70%) of (±)-(1α,2β)-2-[(2-amino-6-chloropyrimidin-4-yl)amino]-cyclopentan-1-ol, 65, as a tan solid, which was suficiently pure for the following step. An analytical sample was prepared by recrystallization from MeCN: mp 171–173 °C.

A solution of 3.42 g (26.85 mmol) of 4-chloroaniline in 95% EtOH (25 mL) was treated with concentrated H₂SO₄ (2.75 mL). The solution was cooled to 27 °C and treated dropwise with 1-butyl nitrite (3.2 g). The solution was then warmed carefully to 35–40 °C for 10 min and then chilled in ice. The crystals of diazonium salt that separated were collected and washed with a minimum amount of cold EtOH and used immediately in the next step.

The diazonium salt prepared in the last step was dissolved in MeOH (30 mL) and the resulting solution treated with 65 (3.0 g, 13.12 mmol). Yellow crystals began to separate immediately. After 20 min, a solution of anhydrous AcONa (2.2 g) in H₂O (20 mL) was added and the reaction mixture then stirred for 90 min. The thick yellow paste was collected and washed well with hot MeOH. The bright yellow sample (4.10 g, 85%) of (±)-(1α,2β)-2-[[2-amino-6-chloro-5-[(4-chlorophenyl)azo]pyrimidin-4-yl]amino]cyclopentan-1-ol, 66, was dried under high vacuum and found to be suffiently pure to use in the next step.

A solution of 66 (1.6 g, 4.37 mmol) and Zn (2.5 g) in a mixture of AcOH (1 mL), EtOH (60 mL), and H₂O (20 mL) was refluxed under N₂ until the yellow color disappeared. After filtration, the solvent was evaporated under reduced pressure to give a residue of (±)-(1α,2β)-2-[(2,5-diamino-6-chloropyrimidin-4-yl)amino]cyclopentan-1-ol, 67, which was placed in a mixture of cooled DMF (20 mL) and trimethyl orthoformate (50 mL) containing concentrated HCl (0.5 mL). This mixture was then stirred at room temperature overnight and the solvent mixture evaporated under reduced pressure. The residue was stirred in 0.5 N HCl (50 mL) for 3 h. After adjusting the pH to 9, the mixture was evaporated to dryness under reduced pressure to provide a residue that was purified by column chromatography (CH₂Cl₂:MeOH, 10:1) to give 68 as white crystals following recrystallization from MeOH (770 mg, 68.8%): mp 197–198 °C; ¹H NMR (DMSO- d₆) δ 1.5–2.2 (m, 6 H, 3 × CH₂ of cyclopentyl), 4.45 (m, 2 H, H-1 and H-2 of cyclopentyl), 6.88 (s, 2 H, NH₂), 8.21 (s, 1 H, H-8 of purine); ¹³C NMR (DMSO- d₆) δ 19.64, 28.53, 31.78, 62.55, 74.71, 123.78, 142.15, 149.27, 154.23, 159.45.

Cell Culture and Radioligand Binding

CHO cells stably expressing the A₃ receptor^9,19 were grown in F-12 medium containing 10% FBS and penicillin/streptomycin (100 U/mL and 100 μg/mL, respectively) at 37 °C in a 5% CO₂ atmosphere, and membrane homogenates were prepared as reported.⁹

Binding of [¹²⁵I]-N⁶-(4-amino-3-iodobenzyl)adenosine-5′-(N-methyluronamide) ([¹²⁵I]AB-MECA) to the CHO cells membranes was performed essentially as described.^16,19 Assays were performed in 50/10/1 buffer in glass tubes and contained 100 μL of the membrane suspension, 50 μL of [¹²⁵I]AB-MECA (final concentration 0.3 nM), and 50 μL of inhibitor. Inhibitors were routinely dissolved in DMSO and then diluted with buffer; final DMSO concentrations never exceeded 1%. Incubations were carried out in duplicate for 1 h at 37 °C and terminated by rapid filtration over Whatman GF/B filters, using a Brandell cell harvester (Brandell, Gaithersburg, MD). Tubes were washed three times with 3 mL of buffer. Radioactivity was determined in a Beckman γ-5500B counter. Nonspecific binding was determined in the presence of 40 μM. (R)-PIA. K_i values were calculated according to the Cheng-Prusoff equation,⁴¹ assuming a K_d for [¹²⁵I]AB-MECA of 1.55 nM.¹⁶

Binding of [³H]PIA (Amersham, Arlington Heights, IL) to A₁ receptors from rat brain membranes and [³H]CGS 21680 (DuPont NEN, Boston, MA) to A₂ receptors from rat striatal membranes was performed as described previously.^17,18 Adenosine deaminase (3 U/mL) was present during the preparation of brain membranes, in which an incubation at 30 °C for 30 min is carried out, and during the incubation with radioligand. At least six different concentrations spanning 3 orders of magnitude, adjusted appropriately for the IC₅₀ of each compound, were used. IC₅₀ values, computer-generated using a nonlinear regression formula on the InPlot program (Graph-PAD, San Diego, CA), were converted to apparent K_i values using K_d values of 1.0 and 14 nM for [³H]PIA and [³H]CGS 21680 binding, respectively, and the Cheng–Prusoff equation.⁴¹

GTP shifts in the displacement of the binding of [³H]-8-cyclopentyl-1,3-dipropylxanthine (CPX; DuPont NEN) were determined as described.³⁸ Adenylate cyclase measurements in A₃-transfected CHO cells were carried out as described.⁹