Modulation of adenosine receptor affinity and intrinsic efficacy in adenine nucleosides substituted at the 2-position

Abstract

We studied the structural determinants of binding affinity and efficacy of adenosine receptor (AR) agonists. Substituents at the 2-position of adenosine were combined with N⁶-substitutions known to enhance human A₃AR affinity. Selectivity of binding of the analogues and their functional effects on cAMP production were studied using recombinant human A₁, A_2A, A_2B, and A₃ARs. Mainly sterically small substituents at the 2-position modulated both the affinity and intrinsic efficacy at all subtypes. The 2-cyano group decreased hA₃AR affinity and efficacy in the cases of N⁶-(3-iodobenzyl) and N⁶-(trans-2-phenyl-1-cyclopropyl), for which a full A₃AR agonist was converted into a selective antagonist; the 2-cyano-N⁶-methyl analogue was a full A₃AR agonist. The combination of N⁶-benzyl and various 2-substitutions (chloro, trifluoromethyl, and cyano) resulted in reduced efficacy at the A₁AR. The environment surrounding the 2-position within the putative A₃AR binding site was explored using rhodopsin-based homology modeling and ligand docking.

1. Introduction

Receptors for the extracellular local modulator adenosine consist of four subtypes: A₁, A_2A, A_2B, and A₃.¹ The selective activation of the A₃AR (adenosine receptor) is both cardioprotective and cerebroprotective in a variety of ischemic models.^2,3 The activation of this receptor subtype has also been associated with a cytostatic, anticancer effect in several tumor models.⁴ Thus, selective A₃AR agonists have therapeutic potential.

Previous medicinal chemical studies demonstrated that an adenosine derivative’s ability to activate the A₃AR is more structure sensitive than at other AR subtypes.^5–7 Changes in various regions of the adenosine molecule have been shown to reduce A₃AR efficacy, which leads to nucleoside analogues that are partial agonists or antagonists. A structure–efficacy relationship at this subtype, separate from the structure–affinity relationship derived from receptor binding experiments, was analyzed.⁷ Substitution with N⁶-benzyl groups, 2-chloro substitution of the adenine moiety, and conformational constraint of the ribose moiety all contribute to a reduction of efficacy. In the present study, we evaluated the binding affinity and functional properties of adenosine derivatives modified at the N⁶-position to achieve high A₃AR affinity and at the 2-position with simple substitutions, such as cyano, carbonyl, aminomethyl, and trifluoromethyl. Previously, the effects of such substitutions at AR subtypes had not been fully evaluated.⁸

2. Results and discussion

2.1. Chemistry

Adenosine agonists 2–17 and 19–21 were prepared (Schemes 1–3) to study the effects of 2-position substitution in interactions with ARs. Key synthetic intermediates 24 and 25 contained the 2-iodo and 2-cyano group, respectively, in combination with 6-chloro substitution. The introduction of the 2-cyano group was carried out by the reaction of zinc cyanide on the corresponding 2-iodo analogue with palladium chemistry (Scheme 1). We chose the protected 2-amino nucleoside 23 as the synthetic intermediate because of the simplicity of its synthesis from (−)-2-amino-6-chloropurine riboside 22. An attempted Sandmeyer-type cyanation (t-BuONO, CuCN, MeCN, 65 °C) on 23 resulted in decomposition without isolation of the desired 2-CN compound. We then applied Pd chemistry to this cyanation. The 2-amino group of 23 was converted to 2-iodo using a neutral purinyl radical, which was generated transiently from the thermal homolysis of the 2-diazonium intermediate.⁹ Stille-type conditions (PdCl₂(PPh₃)₂, Bu₃SnCN) or use of potassium cyanide (PdCl₂(PPh₃)₂, KCN)¹⁰ on the iodo derivative 24 resulted in decomposition. Only use of zinc cyanide and a Pd(0) complex generated in situ from tris-2-furylphosphine and Pd₂(dba)₃ (tris(dibenzylideneacetone)dipalladium(0) chloroform adduct)¹¹ effected cyanation of 24. Use of tetramethylurea as solvent gave a higher yield than N-methyl-2-pyrrolidinone, which was recommended in the literature.

Scheme 1.

Scheme 3.

Intermediate 25 was coupled with the selected amines (methylamine, 5-chloro-2-methoxybenzylamine, or trans-2-phenyl-1-cyclopropylamine) at room temperature, and the acetyl groups were hydrolyzed to give the final target compounds 2, 15, and 19 (Schemes 1 and 2). The 2-cyano group could be reduced with BH₃ to give the 2-aminomethyl derivative 5 (Scheme 1) and in other cases was hydrolyzed to give two 2-carboxylic acid derivatives 16 and 20 (Scheme 2). In the case of the N⁶-methyl derivative 2, the 2-cyano group also reacted with methylamine at elevated temperature and provided a N-methyl-carboxyamidine (C(=NH)NHMe), which was hydrolyzed to give 6. In the case of the 3-iodobenzylamine derivative 26, the methanol solvent reacted with the 2-cyano group and upon hydrolysis produced the 2-methoxycarbonyl derivative 12. In methanol-free hydrolysis conditions, the desired N⁶-(3-iodobenzyl) derivative 10 was obtained. Adding excess 3-iodobenzylamine to 25 and heating resulted in amine addition at the 2-cyano group to give the 2-(3-iodobenzylamide) derivative 13 upon hydrolysis.

Scheme 2.

Trifluoromethylation was performed on the intermediate 24 by treatment with ‘CF₃Cu’ species, which was generated in situ from CF₃ZnBr and CuI (Scheme 3).^12,13 Intermediate 27 was coupled with methylamine or 3-iodobenzylamine, and the acetyl groups were hydrolyzed to give the final target compounds 3 and 11.

A 2-iodoadenosine derivative 9 was also synthesized from 24 by treatment with 3-iodobenzylamine followed by hydrolysis. A 2-amino-N⁶-methyl derivative 4 was synthesized by treatment of 22 with methylamine with-out the use of a protecting group. Each compound obtained was purified by column chromatography, preparative thin-layer chromatography, or HPLC.

2.2. Biological activity

We measured the binding affinities of the adenosine derivatives examined in this study (1–21) at hA₁, hA_2A, and hA₃ (human) ARs and at the rA₃AR (rat) and their degree of activation at the four hAR subtypes (Table 1). The efficacy of each of these adenosine derivatives in activation of the hARs was evaluated at a fixed concentration of 10 μM. Four different N⁶-substitutions were included: methyl, 3-iodobenzyl, 5-chloro-2-methyloxybenzyl, and trans-2-phenyl-1-cyclopropyl. The choice of these four N⁶-substituents was based on either their prior use as groups that enhance hA₃AR affinity or selectivity^6,7,14 or structural similarity to such a moiety.¹⁵ Stereoselectivity of the N⁶-functional group of adenosine derivatives in receptor binding has been established at A₃ARs.⁷ Although the trans-2-phenyl-1-cyclopropyl analogues 18–21 shown in Table 1 are diastereomeric mixtures, it was previously established that the principal biological activity at the A₃AR for this series is associated with the (1S,2R) isomer.⁷ These N⁶-substitutions were combined with various 2-modifications, including chloro, iodo, cyano, trifluoromethyl, carboxylic acid, amide, ester, or aminomethyl groups.

Table 1.

Binding affinities and maximal agonist effects of adenosine derivatives (upon structural variation at N⁶ and C2 positions) at human A₁, A_2A, A_2B, and A₃ARs and at rat A₃ARs expressed in CHO (Chinese hamster ovary) cells^a

#	N6-R1	C-2	Ki (hA1AR)a or % displ.	% Act. (hA1AR)b	Ki (hA2AAR)a or % displ.	% Act. (hA2AAR)b	% Act. (hA2BAR)b	Ki (hA3AR)a	% Act. (hA3AR)b	Ki (rA3AR)a
1	CH3	H	5970±2030	38±10	17%	5±5	16±3	9.3±0.4	96±3	6390
2	CH3	CN	69.8±4.4	60±7	23%	0	0	3.4±0.8	101±7	>10,000
3	CH3	CF3	4650±580	4±3	9%	20±3	0	64.5±3.8	93±3	>10,000
4	CH3	NH2	484±22	13±2	15%	29±6	11±2	39.0±2.4	98±3	>10,000
5	CH3	CH2NH2	27%	32±4	25%	30±3	4±2	719±37	57±4	>10,000
6	CH3	CH(=NH)– NHCH3	10%	8±5	9%	12±3	13±8	2730±268	7±1	ND
7	IB	H	7.4±1.7	78±6	135±22	90±7	58±1	5.8±0.4	46±8	9.5±1.4c
8	IB	Cl	16.8±2.2	8±3	197±34	99±2	16±3	1.8±0.1	0	2.7±1.2
9	IB	I	191±12	0	1910±320	92±8	9±6	24.3±3.1	0	ND
10	IB	CN	1750±290	1±1	25%	32±5	0	119±19	0	256±30
11	IB	CF3	612±131	0	35%	61±3	15±2	138±7	0	257±19
12	IB	CO2CH3	73.1±22.9	38±4	390±160	91±9	18±2	3.21±0.17	0	6.05±0.16
13	IB	CONH(3-IB)	500±160	66±16	3990±150	52±18	5±2	133±19	0	ND
14	CMB	H	9.2±0.5	98±8	399±6	93±6	56±4	1.31±0.16	53±3	ND
15	CMB	CN	63.2±16.9	36±13	1260±190	90±15	16±3	2.76±0.51	29±6	12.8±2.3
16	CMB	COOH	41%	33±10	9%	31±6	6±7	211±36	0	698±100
17	CMB	CO2CH3	6200±790	30±4	35%	74±3	6±4	8.04±0.4	35±4	ND
18	PC	H	124±30	101±13	2530±720	88±23	20±15	0.86±0.09	101±5	399±28
19	PC	CN	1750±280	28±14	5640±780	51±8	6±4	8.7±2.1	0	ND
20	PC	COOH	30%	6±2	0%	5±5	3±1	1730±540	28±1	ND
21	PC	CONH2	295±25	98±7	41%	94±4	16±3	19.7±0.4	74±6	ND

^aAll AR binding experiments were performed using adherent CHO cells stably transfected with cDNA encoding the human or rat ARs (unit: nM). Radioligands: 2nM [³H]R-PIA (A₁), 15nM [³H]CGS21680 (A_2A), and 0.5nM [¹²⁵I]I-AB-MECA (A₃). % displacement at 10 μM.

^bPercent activity at 10 μM, relative to 10 μM CPA (A₁), 10 μM NECA (A_2A, A_2B), or 10 μM Cl-IB-MECA (A₃). Values are expressed as mean±sem, n = 3–5.

^cData from Kim et al.²⁵ ND, not determined. IB, 3-iodobenzyl; CMB, 5-chloro-2-methyloxybenzyl; PC, trans-2-phenyl-1-cyclopropyl.

The codependence of N⁶ and C-2 regions in AR binding and activation was evident. With a small group at the N⁶ position (methyl), the affinity at the hA₃AR was reduced by some groups at C-2 (trifluoromethyl 3, amino 4, aminomethyl 5, and N-methyl-carboxamidine 6) and enhanced by another group, cyano 2. In the case of N⁶-methyl derivatives, the 2-cyano, 2-trifluoromethyl, and 2-amino analogues 2–4 were full agonists, and 2 also showed moderate hA₃AR selectivity (A₁AR/A₃AR 20-fold, A_2AAR/A₃AR > 1000-fold). Aminomethyl 5 and N-methyl-carboxamidine 6 analogues, being positively charged at physiological pH, decreased the affinity at ARs significantly. Also, by virtue of a charged group, 5 and 6 may prove to be neoligands for engineered ARs.¹⁶

In contrast to N⁶-methyl, with a larger N⁶-substituent, the 2-cyano substitution did not enhance the affinity, but rather resulted in a reduction in the ability to activate the receptor. For example, for N⁶-(5-chloro-2-methoxybenzyl), the efficacy of the 2-cyano analogue 15 was some-what reduced compared to the 2-H analogue 14 of similar A₃AR affinity. Nevertheless, both were partial agonists. Similarly, a loss of efficacy was seen for the corresponding N⁶-(3-iodobenzyl) analogues, that is 10 compared with 7. The 2-cyano analogue 15 showed improved hA₃AR selectivity (A₁AR/A₃AR 23-fold, A_2AAR/A₃AR 457-fold); 14 behaved as an = ′ A₁AR full agonist and an A₃AR partial agonist. For a pair of trans-2-phenyl-1-cyclopropyl derivatives, at the hA₃AR the addition of the 2-cyano group transformed a selective full agonist 18 into a selective full antagonist 19.

Other 2-substitutions had varied effects. Curiously, when substituted with 2-COOMe, the high affinity of these N⁶-substituted adenosine analogues at the hA₃AR was maintained or even increased (12 compared with 7 with N⁶-(3-iodobenzyl)). Also, 12 showed favorable hA₃AR selectivity (A₁AR/A₃AR = 23-fold, A_2AAR/A₃AR = 130-fold). Compound 12 appeared to be an A₃AR antagonist and an A₁AR weak, partial agonist. Another 2-COOMe derivative, 17, was a highly selective A₃AR partial agonist.

At the hA₁AR, our compounds showed variable efficacy. In a previous report, 2-halo analogues showed full agonist activity at the A₁AR.¹⁷ In our results, 14, 18, and 21 showed full agonism, and 9–11 were antagonists. Compounds 2, 7, 8, 12, and 15 appeared to be partial agonists at the hA₁AR. hA_2AAR affinities were moderate to weak, and there was no indication of partial agonism. In a previous report, 2-halo substituents reduced the efficacy at the A_2AAR.¹⁷ At the hA_2AAR, 2-Cl (8) and 2-I (9) analogues showed full agonism. Compounds 7, 12, 14, 15, and 21 were also full A_2AAR agonists. At the hA_2BAR, only 7 and 14 substantially activated at 10 μM.

The loss of A₃AR affinity of 13, which was substituted with a 2-CONHBn group, could be explained by the bulkiness of this group. The less bulky 2-iodo analogue 9 was four-fold less potent in binding to the hA₃AR than the 2-H, while the corresponding methyl ester 12 was twice as potent. Another electron withdrawing group, 2-CF₃, was similar to 2-CN in its effect on A₃AR affinity (11 compared with 10) when N⁶-(3-iodobenzyl) was present. However, with N⁶-methyl there was a 20-fold reduction (3 compared with 2). 2-COOH, which is also charged at physiological pH, decreased the affinity at several ARs significantly (16 compared with 14 and 20 compared with 18). However, 2-CONH₂ 21 retained A₃AR affinity. Although compound 21 fully activated the A₁AR, it was only a partial agonist at the A₃AR.

The usefulness of the compounds may also be limited by species differences. For example, with the N⁶-methyl group, the affinity at the rA₃AR was dramatically reduced (e.g., 1 and 2); this is similar to previous observations.^7,18 Species differences at the A₃ARs were also observed for larger N⁶-substituent derivatives, for example, 18. However, for substituted N⁶-benzyl derivatives, binding affinity at the rA₃AR tended to be preserved. For example, 12 displayed a K_i value of 6 nM at the rA₃AR, which was almost the same as its hA₃AR affinity. Compound 12 also displayed substantial A₃AR selectivity. Thus, 12 may prove to be an A₃AR-selective antagonist in the study of rat models. In general, for N⁶-(3-iodobenzyl)–derivatized compounds, the affinity at the rA₃AR was no more than three-fold weaker than at hA₃AR, making these compounds attractive for testing in nonprimate animals.

2.3. Molecular modeling

To locate energetically favorable binding orientations of 2-H and 2-CN derivatives 1, 2, 7, 10, 14, and 15, the previously reported Cl-IB-MECA hA₃AR complex was used as the starting geometry for the ribose-binding position.⁶ For the optimal interaction, various t₀ angles for the N⁶-substituent and χ₁ angles for the adenine ring were generated, and the resulting conformations were compared energetically in the putative binding site.

The hA₃AR docking result of the high-affinity nucleoside 15 is shown in Figure 1. Residues in the putative binding site that were within 5-Å proximity to the ligand were L91 (3.33), T94 (3.36), H95 (3.37), Q167 (EL2), F168 (EL2), S181 (5.42), M177 (5.38), V178 (5.39), F182 (5.43), W243 (6.48), L246 (6.51), S247 (6.52), N250 (6.55), I268 (7.39), S271 (7.42), and H272 (7.43). Molecular modeling results were correlated with results of point mutation in TMs (transmembrane helical domains) 3, 6, 7, and EL2.^6,19–21

Figure 1.

The complexes of the A₃AR with 2-cyano ligands (A) the overall binding site of 15 and (B) the 2-cyano binding site of 2 with distances in Å. All ligands are displayed as ball-and-stick models in magenta color, and the side chains of the hA₃AR are shown as stick models. The H bonding between the ligand and the hA₃AR is displayed in yellow. The A₃AR is represented by a tube model, with a different color for each TM (TM3 in yellow, TM5 in green, TM6 in cyano, TM7 in purple).

The purine ring was surrounded by a hydrophobic pocket, which was defined by L91 (3.33) and L246 (6.51). In addition, H-bonds formed between the exocyclic amine and the hydroxyl group of S247 (6.52), which was also in proximity to N250 (6.55), and between the purine N³ atom and the side chain of Q167 (EL2) (Fig. 1A). The 2′-OH group of the ribose moiety formed an H bond with the backbone carbonyl group of I268 (7.39), and the 3′-OH group H bonded with the backbone carbonyl group of S271 (7.42) and with the side chain of H272 (7.43), consistent with our A₃ neoceptor model.¹⁹ The 5′-hydroxyl group also formed an H bond with T94 (3.36). The phenyl moiety showed an additional hydrophobic interaction with F168 (EL2), and the methyl of the methoxy group was stabilized through hydrophobic interaction with L91 (3.33) and I186 (5.47), consistent with its higher binding affinity.

In comparison to N⁶-methyl adenosine 1, the more potent 2-cyano-N⁶-methyl derivative 2 showed a more favorable nonbonding van der Waals interaction between the cyano group and the side chains of L90 (3.32) and L91 (3.33) and between the methyl group of the side chain of T87 (3.29) and the aromatic ring of F168 of EL2 (Fig. 1B). The distances between the N atom of the cyano group and the closest carbon atoms of the surrounding hydrophobic residues were 3.4–4.3 Å. The C-2 substituent was also proximal to the hydrophilic residue Q167 (EL2), with a distance of only 3.9 Å between the N atom of the cyano group and the side chain carbonyl O atom of Q167. There are some variable amino acids surrounding the binding site of the 2-cyano group. The amino acid corresponding to L90 (3.32) was Val in other ARs. T87 (3.29) was Ala in other human AR subtypes and Ser for the rA₃AR. Q167 was a highly variable residue among the different subtypes and species, corresponding to Glu for A₁AR, Leu for A_2A and A_2BAR, and His for rat A₃AR. Thus, the hA₃AR preference of 2-cyanoadenosine derivatives might be explained by an optimal van der Waals interaction. Another possibility is that a different binding preference for χ₁ angle at each AR subtype would vary the residues of interaction with the C-2 substituents.

The preferred t₀ and χ₁ angles of the energetically favorable bound conformation, derived from hA₃AR docking of various compounds, depended on the particular N⁶ and C-2 substituents. In the lowest-energy complexes of 1 and 2, the t₀ and χ₁ angles were −134.7° and −102.1° and −162.7° and −97.3°, respectively. For larger N⁶ groups, such as 3-I-benzyl, the preferred t₀ and χ₁ angles were −113.8° and −111.5° for 7 and −117.5° and −108.3° for 10. Compared to 2, the − binding of the 2-cyano group in 10 was directed more toward the methyl group of T87 (3.29), resulting in unfavorable van der Waals interactions, consistent with decreased binding affinity. The difference in 2-CN orientation between 2 and 10 was a consequence of additional hydrophobic interaction of the phenyl ring of the N⁶-substituent. Thus, the modeling has demonstrated how the interactions at the N⁶ and C-2 regions of the hA₃AR might be interdependent.²²

3. Conclusions

In previous studies, it was demonstrated that 2-chloro and other 2-substitutions of the adenine ring may increase the affinity and decrease the efficacy of the adenosine derivatives for the hA₃AR.^5–7 The 2-chloro modification has been incorporated into many of the more highly potent AR probes. At the A₁ARs, 2-chloro has also been observed to enhance affinity of selective agents. Here we further demonstrated that the 2-substitution with the cyano group and other small groups has variable effects.

Although the SAR (structure–activity relationship) of the many 2-substitutions of adenosine at ARs has been thoroughly explored,^23–26 the cyano, trifluoromethyl, simple ester, and other groups included here have not been among them. However, various 2-alkyne derivatives,^18,27 such as 2-hexynyladenosine-5′-N-ethyluron-amide, have been found to bind potently to both A_2A and A₃ARs. The 2-iodo group was found previously to reduce efficacy at the A_2AAR and retain efficacy at the A₁AR.¹⁷ In our study, 2-iodo analogue 9 showed full agonism at the hA_2AAR and antagonism at the hA₁AR.

We have studied the effects of mainly sterically small groups substituted at the 2-position on the affinity and intrinsic efficacy at A₁, A_2A, A_2B, and A₃ receptors. The previously reported compound 8 appears to be a mixed A₁/A₃AR antagonist. Compound 12 appears to be an A₃AR antagonist and an A₁AR weak, partial agonist. Compound 14 behaves as an A₁AR full agonist and an A₃AR partial agonist, and compound 17 is a highly selective A₃AR partial agonist. Compound 19 is a potent A₃AR antagonist, and the related compound 21 is an A₁AR full agonist and an A₃AR partial agonist. Thus, we have modulated both the AR affinity and efficacy upon substitution at the 2-position in combination with various N⁶-substituents. This has resulted in a collection of adenosine analogues having widely differing pharmacological activities, both quantitatively and qualitatively. The dependence of intrinsic efficacy at the A₃AR has recently been emphasized,^5–7 and with this study we extend the analysis to other AR subtypes.

In conclusion, we have expanded the range of structures suitable as nucleoside antagonists of the A₃AR. Such compounds include 12 and 16. A₃AR antagonists are of interest for treatment of glaucoma and possibly asthma.¹ Partial A₁AR agonists, such as 2, 12, and 15, may be of interest for treatment of cardiac arrhythmias.²⁶ Whether the presence of A₃AR agonism or antagonism in these compounds might provide a therapeutic advantage is unknown. Compound 2 as a selective hA₃AR agonist could be examined for activity as a cardioprotective and cerebroprotective agent.^2,3,28 These compounds may also be useful as receptor probes for mutagenesis and for the design of neoceptors (e.g., moderately potent ligands that have either negative, for example, 16, or positive, for example, 5, charges near the 2-position).^16,19–21 The expanded analysis of the SAR of nucleoside binding at the A₃AR may now be applied to other agonists series, to probe the additivity with ribose ring modifications.^6,7

4. Experimental materials

[¹²⁵I]N⁶-(4-Amino-3-iodobenzyl)adenosine-5′-N-methyl-uronamide ([¹²⁵I]I-AB-MECA; 2000 Ci/mmol), N⁶-[(R)-phenylisopropyl]adenosine ([³H]R-PIA; 34 Ci/mmol), and [³H]cyclic AMP (40 Ci/mmol) were obtained from Amersham Pharmacia Biotech (Piscataway, NJ). 2-[p-(2-carboxyethyl)phenyl-ethylamino]-5′-N-ethylcarbox-amido-adenosine ([³H]CGS21680; 47 Ci/mmol) was supplied by Perkin–Elmer Life Sciences (Boston, MA). Several N⁶-substituted adenosine derivatives 1, 7, 8, and 18 were prepared as reported (Ref. 7 and references therein). All other chemicals were of analytical grade from standard commercial sources.

4.1. Chemistry

Nucleosides and synthetic reagents were purchased from Sigma Chemical Co. (St. Louis, MO) and Aldrich (Milwaukee, WI). ¹H NMR spectra were obtained with a Varian Gemini-300 spectrometer (300 MHz) with D₂O, CDCl₃, CD₃OD, and DMSO-d₆ as a solvent. Purity of the nucleosides was checked using a Hewlett–Packard 1100 HPLC equipped with a Luna 5 μ RP-C18(2) analytical column (250 × 4.6 mm; Phenomenex, Torrance, CA). System A: linear gradient solvent system: H₂O/CH₃CN/AcOH from 90/10/0.05 to 50/50/0.05 in 20 min and 10/90/0.05 in 50 min, flow rate 0.5 mL/min. System B: linear gradient solvent system: H₂O/CH₃CN/AcOH from 90/10/0.05 to 50/50/0.05 in 40 min and 0/100/0.05 in 60 min, flow rate 0.5 mL/min. System C: linear gradient solvent system: CH₃CN/TBAP from 5/95 to 80/20 in 20 min, flow rate 1.0 mL/min. System D: linear gradient solvent system: H₂O/CH₃CN/AcOH from 80/20/0.05 to 60/40/0.05 in 5 min and 30/70/0.05 in 20 min, flow rate 1.0 mL/min. System E: linear gradient solvent system: H₂O/CH₃CN from 95/5 to 0/100 in 30 min, flow rate 1.0 mL/min. Peaks were detected by UV absorption with a diode array detector. All derivatives tested for biological activity showed ≥97% purity in the HPLC systems. Low-resolution mass spectra were measured with a Finnigan–Thermoquest LCQ with APCI (Atmospheric Pressure Chemical Ionization) interface. Vaporizer heater was 480 °C interfaced with HP1100 LC apparatus (System A). Low-resolution and high-resolution FAB (fast atom bombardment) mass spectrometry was performed with a JEOL SX102 spectrometer with 6-kV Xe atoms following desorption from a glycerol matrix.

4.2. (2R, 3S, 4S, 5R)-2-(2-Cyano-6-methylamino-purin-9-yl)-5-hydroxy-methyl-tetrahydrofuran-3,4-diol (2)

A solution of 2.0 M MeNH₂ in THF (2.0 mL) was added to a solution of 25 (42 mg, 0.096 mmol) in THF (0.80 mL), and the mixture was stirred at rt (room temperature) for 26 h. MeOH (2.0 mL) was added, and the reaction mixture was stirred at rt for 3 h. The solvent was removed under reduced pressure and the residue was purified by preparative TLC (silica gel, thickness 0.25 mm, solvent: CHCl₃/MeOH = 10/1) to give 2 (19 mg, 66%). ¹H NMR (DMSO-d₆) δ 8.62 (s, 1H), 8.48 (br, 1H), 5.90 (d, 1H, J = 5.7 Hz), 5.52 (d, 1H, J = 6.1 Hz), 5.23 (d, 1H, J = 5.1 Hz), 5.05 (t, 1H, J = 5.7 Hz), 4.52 (m, 1H), 4.15 (m, 1H), 3.95 (m, 1H), 3.75–3.60 (m, 1H), 3.60–3.50 (m, 1H), 2.97 (d, 3H, J = 4.2 Hz); MS (m/e) (CI) 307 (M + H)⁺, HRMS (positive-FAB) calcd for C₁₂H₁₅N₆O₄ (M + H)⁺ 307.1155, found 307.1158; HPLC (System B) 24.6 min (99%) (System C), 9.9 min (99%).

4.3. (2R, 3S, 4S, 5R)-2-(6-Methylamino-2-trifluoromethyl-purin-9-yl)-5-hydroxymethyl-tetrahydrofuran-3,4-diol (3)

To a solution of 28 (2.6 mg, 0.0055 mmol) in MeOH (2.0 mL) was added 3 N NaOH (0.02 mL) and the mixture was stirred at rt for 4 h. The solvent was removed under reduced pressure and the residue was purified by preparative TLC (silica gel, thickness 0.25 mm, solvent AcOEt), to give 3 (0.8 mg, 43%). ¹H NMR (CD₃OD) δ 8.38 (s, 1H), 6.00 (d, 1H, J = 6.0 Hz), 4,72 (t, 1H, J = 5.4 Hz), 4.34 (dd, 1H, J = 3.3, 5.1 Hz), 4.14 (dd, 1H, J = 3.3, 6.3 Hz), 3.88 (dd, 1H, J = 3.0, 12.3 Hz), 3.75 (dd, 1H, J = 3.6, 12.3 Hz), 3.13 (br = s, 3H); MS (m/e) (positive-FAB) 350 (M + H)⁺, 372 (M + Na)⁺. HRMS (positive-FAB) calcd for C₁₂H₁₅N₅O₄F₃ (M H)⁺ 350.1076, found 350.1069; HPLC (System C) 11.8 + min (99%) (System E), 4.3 min (99%).

4.4. (2R, 3S, 4S, 5R)-2-(2-Amino-6-methylamino-purin-9-yl)-5-hydroxy-methyl-tetrahydrofuran-3,4-diol (4)

A solution of 22 (19 mg, 0.063 mmol) in 30% methylamine in H₂O (2.0 mL) was stirred at 60 °C for 7 h. The solvent was removed under reduced pressure and the residue was purified by C18 reverse-phase silica gel chromatography (eluted by a linear gradient of H₂O/MeCN = 100/0 to 0/100), to give 4 (18 mg, 94%). ¹H NMR (DMSO-d₆) δ 7.94 (s, 1H), 5.90 (d, 1H, J = 6.6 Hz), 4.77–4.72 (m, 1H), 4.41 (dd, 1H, J = 3.0, 5.4 Hz), 4.27 (dd, 1H, J = 3.0, 6.0 Hz), 3.91 (dd, = 1H, J = 2.7, 12.9 Hz), 3.82 (dd, 1H, J = 3.3, 12.9 Hz), 3.04 (s, 3H); MS (m/e) (positive-FAB) 297 = (M + H)⁺. HRMS (positive-FAB) calcd for C₁₁H₁₇N₆O₄ (M H)⁺ 297.1311, found 297.1317; HPLC (System C) 5.5 + min (99%) (System E), 3.3 min (99%).

4.5. (2R, 3S, 4S, 5R)-2-(2-Aminomethyl-6-methylaminopurin-9-yl)-5-hydroxymethyl-tetrahydrofuran-3,4-diol (5)

To a solution of 2 (2.4 mg, 0.0079 mmol) in THF (0.50 mL) was added 1.0 M BH₃–THF complex THF solution (1.5 mL), and the mixture was stirred at 60 °C for 20 h. MeOH (2.0 mL) was added to the reaction mixture, and the solvent was removed under reduced pressure. The residue was purified on an Amberlite CG-50 column (eluted by a linear gradient of ammonium bicarbonate 0–1.0 M) and HPLC (Luna 5 μ RP-C18(2) column, eluted by a linear gradient of H₂O/MeCN/AcOH = 100/0/0.05 to 80/20/0.05 in 15 min), to give 5 as acetic acid salt (1.1 mg, 36%). ¹H NMR (D₂O) δ 8.23 (s, 1H), 6.10 (d, 1H, J = 3.9 Hz), 4.56 (m, 1H), 4.44 (m, 1H), 4.26 (m, 1H), 4.20–4.12 (br, 2H), 3.92–3.73 (m, 2H), 3.12 (br s, 3H) 1.90 (s, 3H); MS (m/e) (CI) 311 (M + H)⁺ (positive-FAB), 311 (M + H)⁺. HRMS (positive-FAB) calcd for C₁₂H₁₉N₆O₄ (M + H)⁺ 311.1468, found 311.1464; HPLC (System B) 5.1 min (99%) (System C), 2.4 min (99%).

4.6. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-methylamino-9H-purine-2-carboxamidine (6)

To a solution of 25 (21 mg, 0.047 mmol) was added to a 2.0 M MeNH₂ THF solution (2.0 mL), and the mixture was stirred at rt for 2 h. MeOH (0.5 mL) was added, and the reaction mixture was stirred at 50 °C for 2 h. The solvent was removed under reduced pressure and the residue was purified on a short column (silica gel 2.0 mL, eluent: CHCl₃/MeOH = 10/1) and an Amberlite CG-50 column (eluted by linear gradient of ammonium bicarbonate 0–1.0 M), to give 6 (2.8 mg, 18%). 1H NMR (D₂O) δ 8.61 (s, 1H), 6.20 (d, 1H, J = 6.9 Hz), 4.99 (t, 1H, J = 6.5 Hz), 4.49 (m, 1H), 4.21 (m, 1H) 3.82 (d, 2H, J = 3.6 Hz), 3.17 (s, 3H), 2.87 (s, 3H); MS (m/e) (CI) 338 (M + H)⁺ (positive-FAB), 338 (M + H)⁺. HRMS (positive-FAB) calcd for C₁₃H₂₀N₇O₄ (M + H)⁺ 338.1577, found 338.1571; HPLC (System B) 12.1 + min (99%) (System C), 2.1 min (99%).

4.7. (2R, 3S, 4S, 5R)-2-(2-Iodo-6-(3-iodo-benzylamino)-purin-9-yl)-5-hydroxymethyl-tetrahydrofuran-3,4-diol (9)

To a solution of 24 (18.5 mg, 0.034 mmol) in DMF (0.10 mL) was added 3-iodobenzylamine (25 mg) and i-Pr₂NEt (0.10 mL), and the mixture was stirred at rt for 20 h. The solvent was removed under reduced pressure and the residue was purified by column chromatography (silica gel, eluent: AcOEt/petroleum ether = 2/1), to give triacetate (21 mg). To a solution of this = triacetate (21 mg) in MeOH (1.0 mL) was added 3 N NaOH (0.10 mL), and the mixture was stirred at rt for 3 h. The solvent was removed under reduced pressure, and the residue was purified by short-column chromatography (silica gel, eluent: AcOEt) and preparative TLC (silica gel, thickness 0.25 mm, solvent AcOEt), to give 9 (11 mg, 53%). ¹H NMR (CD₃OD) δ 8.15 (s, 1H), 7.79 (s, 1H), 7.60 (d, 1H, J = 7.8 Hz), 7.39 (d, 1H, J = 7.8 Hz), 7.09 (t, 1H, J = 7.8 Hz), 5.89 (d, 1H, J = 6.3 Hz), 4.65 (m, 1H), 4.30 (dd, 1H, J 3.3, 5.1 Hz), 4.13 (m, 1H), 3.88 (dd, 1H, J = 2.7, 12.6 ′Hz), 3.74 (dd, 1H, J = 3.3, 12.6 Hz); MS (m/e) (positive-FAB) 610 (M + H)⁺. HRMS (positive-FAB) calcd for C₁₇H₁₈N₅O₄I₂ (M + H)⁺ 609.9448, found 609.9424; HPLC (System C) 16.0 min (99%) (System E), 6.9 min (99%).

4.8. (2R, 3S, 4S, 5R)-2-(2-Cyano-6-(3-iodo-benzylamino)-purin-9-yl)-5-hydroxymethyl-tetrahydrofuran-3,4-diol (10)

To a solution of 26 (7.4 mg, 0.012 mmol) in MeCN (1.00 mL) and H₂O (0.030 mL) was added 3 N aqueous NaOH (0.030 mL), and the mixture was stirred at rt for 6 h. Aqueous citric acid (5%, 0.050 mL) was added to neutralize the reaction and the solvent was removed under reduced pressure. The residue was purified by column chromatography (silica gel, eluent: AcOEt then CHCl₃/MeOH = 5/1) first and then purified again by HPLC (Luna 5 μ RP-C18(2) column, eluted by a linear gradient of H₂O/MeCN/AcOH = 70/30/0.05), to give 10 (1.2 mg, 20%). ¹H NMR (DMSO-d₆) δ 8.51 (s, 1H), 7.34 (br s, 1H), 7.68 (m, 1H), 7.60 (m, 1H), 7.35 (m, 1H), 7.12 (m, 1H), 5.95 (m, 1H), 4.65 (br, 2H), 4.51 (m, 1H), 4.13–4.10 (m, 2H), 3.92 (m, 2H); MS (m/e) (CI) 509 (M + H)⁺, HRMS (positive-FAB) calcd for C₁₈H₁₈N₆O₄I (M + H)⁺ 509.0434, found 509.0429; HPLC (System B) 49.2 min (99%) (System C), 16.1 min (99%) (System D), 11.6 min (99%).

4.9. (2R, 3S, 4S, 5R)-2-(6-(3-Iodo-benzylamino)-2-trifluoromethyl-purin-9-yl)-5-hydroxymethyl-tetrahydrofuran-3,4-diol (11)

To a solution of 29 (9.2 mg, 0.014 mmol) in MeOH (1.0 mL) was added potassium carbonate (18.5 mg, 0.13 mmol), and the mixture was stirred at rt for 4 h. The solvent was removed under reduced pressure, and the residue was purified by column chromatography (silica gel, eluent: CHCl₃/MeOH = 10/1) and preparative TLC (silica gel, thickness 0.25 mm, solvent AcOEt), to give 11 (3.4 mg, 45%). ¹H NMR (CDCl₃) δ 7.98 (s, 1H), 7.76 (s, 1H), 7.63 (d, 1H, J = 8.1 Hz), 7.37 (d, 1H, J = 7.2 Hz), 7.08 (t, 1H, J = 7.8 Hz), 6.65 (br s, 1H), 5.84 (d, = 1H, J = 6.9 Hz), 4.99 (m, 1H), 4.80 (m, 2H), 4.49 (d, 1H, J = 4.5 Hz), 4.35 (s, 1H), 3.95 (dd, 1H, J = 1.8, 12.6 Hz), = 3.77 (d, 1H, J = 12.6 Hz); MS (m/e) (positive-FAB) 552 (M + H)⁺. HRMS (positive-FAB) calcd for C₁₈H₁₈N₅O₄F₃I (M + H)⁺ 552.0356, found 552.0336; HPLC (System C) 17.8 min (99%) (System E), 8.2 min (99%).

4.10. (2R, 3S, 4S, 5R)-2-(2-Methoxycarbonyl-6-(3-iodo-benzylamino)-purin-9-yl)-5-hydroxymethyl-tetrahydrofuran-3,4-diol (12)

To a solution of 26 (10.5 mg, 0.017 mmol) in i-PrOH (0.60 mL), MeOH (0.05 mL) and THF (0.20 mL) was added 3 N aqueous NaOH (0.040 mL), and the mixture was stirred at rt for 5 h. The solvent was removed under reduced pressure, and the residue was purified by column chromatography (silica gel, eluent: CHCl₃/MeOH = 10/1), to give 12 (1.3 mg, 15%). ¹H NMR (DMSO-d₆) δ 8.74 (br s, 1H), 8.56 (s, 1H), 7.82 (br s, 1H), 7.58 (m, 1H), 7.39 (m, 1H), 7.10 (m, 1H), 5.93 (d, 1H, J = 6.0 Hz), 5.46 (d, 1H, J = 6.3 Hz), 5.22 (d, 1H, J = 4.5 Hz), 5.03 (t, 1H, J = 4.5 Hz), 4.65 (m, 1H), 4.58 (br, 2H), 4.12 (m, 1H), 3.94 (m, 1H), 3.85 (s, 3H), 3.65 (m, 2H); MS (m/e) (CI) 542 (M + H)⁺, HRMS (positive-FAB) calcd for C₁₉H₂₁N₅O₆I (M + H)⁺ 542.0537, found 542.0540; HPLC (System A) 28.6 min (99%) (System C), 14.5 min (99%).

4.11. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(3-iodo-benzylamino)-9H-purine-2-carboxylic acid 3-iodo-benzylamide (13)

To a solution of 25 (9 mg, 0.020 mmol) in THF (1.50 mL) was added 3-iodobenzylamine (19 mg, 0.082 mmol), and the mixture was stirred at 50 °C for 21.5 h. MeOH (2.0 mL) and potassium carbonate (24 mg, 0.174 mmol) were added, and stirring continued at rt for 3 h. The solvent was removed under reduced pressure and the residue was purified by preparative TLC (silica gel, thickness 0.25 mm, solvent CHCl₃/MeOH = 5/1), to give 13 (11 mg, 76%). ¹H NMR (DMSO-d₆) d 9.05 (br s, 1H), 8.84 (s, 1H), 7.80 (d, 2H, J = 5.4 Hz), 7.58 (d, 2H, J = 8.4 Hz), 7.39 (d, 2H, J = 8.7 Hz), 7.09 (t, 2H, J = 8.0 Hz), 6.10 (d, 1H, J = 5.4 Hz), 5.49 (d, 1H, J = 6.0 Hz), 5.24 (d, 1H, J = 4.2 Hz), 4.89 (s, 2H), 4.76 (s, 2H), 4.55 (m, 1H), 4.17 (m, 1H), 3.98 (m, 1H), 3.65–3.50 (m, 2H); MS (m/e) (positive-FAB) 743 (M + H)⁺. HRMS (positive-FAB) calcd for C₂₅H₂₅N₆O₅I₂ 742.9976, found 742.9972; HPLC (System A) 24.6 min (99%) (System B), 35.0 min (99%) (System C), 12.9 min (99%).

4.12. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(5-chloro-2-methoxy-benzylamino)-9H-purine (14)

Compound 14 (240 mg, 0.57 mmol) was prepared in one step¹⁹ from 6-chloropurine riboside (180 mg, 0.64 mmol), 5-chloro-2-methoxy-benzylamine hydrochloride²⁹ (200 mg, 0.96 mmol) and triethylamine (120 μL, 0.83 mmol) in 88% yield. ¹H NMR (DMSO-d₆) δ 8.41 (s, 1H), 8.35 (br s, 1H), 8.18 (s, 1H), 7.24 (dd, 1H, J = 2.5, 8.7 Hz), 7.04 (s, 1H), 7.00 (d, 1H), 5.89 (d, 1H, J = 5.9 Hz), 5.37 (t, 1H, J = 5.6 Hz), 5.20 (d, 1H, J = 4.7 Hz), 4.63 (br, 3H), 4.14 (d, 1H, J = 2.9 Hz), 3.95 (d, 1H, J = 2.9 Hz), 3.83 (s, 3H), 3.66 (m, ′2H), 3.54 (m, 2H); HRMS (ESI) calcd for C₁₈H₂₁ClN₅O₅ (M + H)⁺: 422.1231, found 422.1223; Anal. (C₁₈H₂₀ClN₅O₅) C, H, N.

4.13. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(5-chloro-2-methoxy-benzylamino)-9H-purine-2-carbonitrile (15)

To a solution of triacetate 25 (6.2 mg, 0.014 mmol) in DMF (0.10 mL) was added 5-chloro-2-methoxybenzylamine hydrochloride (7.0 mg, 0.034 mmol) and i-Pr₂NEt (0.10 mL), and the mixture was stirred at rt for 24 h. The solvent was removed under reduced pressure, and the residue was purified by short-column chromatography (silica gel, eluent: AcOEt). The obtained material was dissolved in MeOH (1.0 mL) and added potassium carbonate (15 mg, 0.108 mmol), and the mixture was stirred at rt for 6 h. The solvent was removed under reduced pressure, and the residue was purified by column chromatography (silica gel, eluent: CHCl₃/MeOH = 10/1 then 5/1), to give 15 (5.2 mg, 82%). ¹H NMR (DMSO-d₆) δ 8.49 (s, 1H), 7.28 (d, 1H, J = 2.7 Hz), 7.23 (dd, 1H, J = 2.7, 8.5 Hz), 6.97 (d, 1H, J = 8.5 Hz), 6.01 (d, 1H), 4.65 (t, 1H, J = 5.4 Hz), 4.32 (t, 1H, J = 4.7 Hz), 4.13 (s, 2H), 3.88 (m, 3H), 3.63 (s, 3H); MS (m/e) (CI) 447, 449 (M + H)⁺ (relative peak height ratio was 3:1), HRMS (positive-FAB) calcd for C₁₉H₂₀N₆O₅Cl (M + H)⁺ 447.1184, found 447.1188; HPLC (System A) 31.2 min (99%) (System C), 15.6 min (99%).

4.14. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(5-chloro-2-methoxy-benzylamino)-9H-purine-2-carboxylic acid (16)

To a solution of 15 (3.1 mg, 0.0069 mmol) in EtOH (0.50 mL) was added 3 N NaOH (0.100 mL) and 30% H₂O₂ (0.040 mL), and the mixture was stirred at rt for 4 h. The solvent was removed under reduced pressure, and the residue was purified by Sephadex column chromatography (eluted by a linear gradient of ammonium bicarbonate 0–0.5 M), to give 16 (1.1 mg, 33%). ¹H NMR (D₂O) δ 8.32 (s, 1H), 7.40 (d, 1H, J = 3.0 Hz), 7.33 (dd, 1H, J = 3.0, 9.0 Hz), 7.04 (d, 1H, J = 9.0 Hz), 6.12 (d, 1H, J = 5.7 Hz), 4.88 (m, 1H), 4.73 (s, 2H), 4.44 (m, 1H), 4.30 (m, 1H), 3.93 (dd, 1H, J = 2.7, 12.9 Hz), 3.88 (s, 3H), 3.83 (dd, 1H, J = 2.7, 12.9 ′Hz); MS (m/e) (CI) 466, 468 (M + H)⁺ (relative peak height ratio was 3:1) (negative-FAB), 464, 466 (M − H)⁺ (relative peak height ratio was 3:1). HRMS (negative-FAB) calcd for C₁₉H₁₉N₅O₇Cl (M − H)⁺ 464.0973, found 464.0969; HPLC (System B) 36.0 min (99%) (System C), 13.1 min (99%).

4.15. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(5-chloro-2-methoxy-benzylamino)-9H-purine-2-carboxylic acid methyl ester (17)

To a solution of 16 (1.0 mg, 0.0021 mmol) in MeOH (0.30 mL) and H₂O (0.10 mL) was added (trimethylsilyl)diazomethane (0.100 mL, 2.0 M in ether), and the mixture was stirred at rt for 5 h. The solvent was removed under reduced pressure, and the residue was purified by preparative TLC (silica gel, thickness 0.25 mm, solvent CHCl₃/MeOH 10/1), to give 17 (0.5 mg, 50%). ¹H NMR (CD = 3OD) δ 8.41 (s, 1H), 7.54 (br s, 1H), 7.21 (dd, 1H, J = 3.0, 8.7 Hz), 6.95 (d, 1H, J = 8.7 Hz), 6.01 (d, 1H, J = 6.0 Hz), 4.70 (m, 1H), 4.58 (s, 2H), 4.33 (m, 1H), 4.16 (m, 1H), 3.98 (s, 3H), 3.88 (s, 3H), 3.75 (m, 1H), 3.55 (m, 1H); MS (m/e) (CI) 480, 482 (M + H)⁺ (relative peak height ratio was 3/1). HRMS (positive-FAB) calcd for C₂₀H₂₃N₅O₇Cl (M + H)⁺ 480.1286, found 480.1279; HPLC (System A) 25.5 min (99%) (System C), 14.1 min (99%).

4.16. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(2-phenyl-cyclopropyl-amino)-9H-purine-2-carbonitrile (19)

To a solution of 25 (8.1 mg, 0.019 mmol) in DMF (0.10 mL) was added trans-2-phenyl cyclopropylamine hydrochloride (10.0 mg, 0.059 mmol) and i-Pr₂NEt (0.10 mL), and the mixture was stirred at rt for 26 h. The solvent was removed under reduced pressure, and the residue was purified by short-column chromatography (silica gel, eluent: AcOEt). The obtained material (5.7 mg) was dissolved in MeCN (1.00 mL) and H₂O (0.030 mL), and the mixture was treated with 3 N NaOHaq (0.030 mL) and was stirred at rt for 4 h. 5% citric acid aq (0.050 mL) was added to neutralize the reaction, and the solvent was removed under reduced pressure. The residue was purified by column chromatography (silica gel, eluent: AcOEt), to give 19 (1.2 mg, 15%). ¹H NMR (CDCl₃) δ 8.02 (s, 1H), 7.34–7.02 (m, 5H), 6.44 (br s, 1H), 5.85 (d, 1H, J = 6.9 Hz), 4.95 (m, 1H), 4.70 (m, 1H), 4.52 (m, 1H), 4.35 (m, 1H), 3.99 (m, 1H), 3.83 (m, 1H), 3.44 (br, 1H), 3.15 (br, 1H), 2.87 (m, 1H), 2.22 (m, 1H), 0.88 (m, 2H); (DMSO-d₆) δ 8.93 (br s, 1H), 8.64 (s, 1H), 7.32–7.10 (m, 5H), 5.90 (d, 1H, J = 5.7 Hz), 5.52 (d, 1H, J = 5.7 Hz), 5.23 (d, 1H, J = 5.4 Hz), 5.04 (t, 1H, J = 5.1 Hz), 4.50 (m, 1H), 4.14 (m, 1H), 3.95 (m, 1H), 3.65 (m, 1H), 3.57 (m, 1H), 3.10 (m, 1H), 2.16 (m, 1H), 1.45–1.30 (m, 2H); MS (m/e) (CI) 409 (M + H)⁺, HRMS (positive-FAB) calcd for C₂₀H₂₁N₆O₄ (M + H)⁺ 409.1624, found 409.1627; HPLC (System A) 30.2 min (99%) (System B), 46.7 min (99%) (System C), 15.3 min (99%).

4.17. (2R, 3S, 4S, 5R)-9-(3,4-Dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(2-phenyl-cyclopropyl-amino)-9H-purine-2-carboxylic acid (20) and 9-(3,4-dihydroxy-5-hydroxymethyl-tetrahydrofuran-2-yl)-6-(2-phenyl-cyclopropyl-amino)-9H-purine-2-carboxamide (21)

To a solution of 19 (4.4 mg, 0.0108 mmol) in EtOH (0.50 mL) was added 3 N NaOH (0.100 mL) and 30% H₂O₂ (0.050 mL), and the mixture was stirred at rt for 28 h. The solvent was removed under reduced pressure, and the residue was purified by Sephadex column chromatography (eluted by a linear gradient of ammonium bicarbonate 0–0.5 M), to give 20 (1.0 mg, 21%) and 21 (1.2 mg, 26%). For 20: ¹H NMR (D₂O) δ 8.31 (s, 1H), 7.45–7.32 (m, 5H), 6.13 (d, 1H, J = 5.7 Hz), 5.05–5.00 (m, 1H), 4.43 (t, 1H, J = 4.8 Hz), 4.30 (m, 1H), 3.96–3.81 (m, 2H) 3.35 (m, 1H), 2.34 (m, 1H), 1.51–1.41 (m, 2H); MS (m/e) (CI) 428 (M + H)⁺. HRMS (positive-FAB) calcd for C₂₀H₂₂N₅O₆ (M + H)⁺ 428.1570, found 428.1571; HPLC (System B) 34.4 min (99%) (System C), 13.5 min (99%). For 21: ¹H NMR (D₂O) δ 8.36 (s, 1H), 7.43–7.23 (m, 5H), 6.14 (d, 1H, J = 5.7 Hz), 5.00–4.95 (m, 1H), 4.44 (t, 1H, J = 4.5 Hz), 4.29 (m, 1H), 3.96–3.80 (m, 2H) 3.07 (m, 1H), 2.15 (m, 1H), 1.47 (m, 2H); MS (m/e) (CI) 427 (M + H)⁺. HRMS (positive-FAB) calcd for C₂₀H₂₃N₆O₅ (M + H)⁺ 427.1730, found 427.1733; HPLC (System B) 35.0 min (99%) (System C), 12.1 min (99%).

4.18. (2R, 3S, 4S, 5R)-2-(2-Amino-6-chloro-purin-9-yl)-3,4-diacetoxy-5-acetoxymethyl-tetrahydrofuran (23)

To a solution of (−)-2-amino-6-chloropurine riboside 22 (995 mg, 3.30 mmol) in DMF (15 mL) was added pyridine (2.0 mL, 25 mmol) and acetic anhydride (2.0 mL, 21 mmol), and the mixture was stirred at 80 °C for 2 h. The solvent was removed under reduced pressure and the residue was purified by silica gel column chromatography (eluent: AcOEt/petroleum ether 1/1 then 1/0), to give 23 (1.26 g, 89%). ¹H NMR (CDCl = 3) δ 7.87 (s, 1H), 6.01 (d, 1H, J = 4.8 Hz), 5.96 (t, 1H, J = 5.2 Hz), 5.75 (t, 1H, J = 5.0 Hz), 5.19 (br s, 2H), 4.49–4.36 (m, 3H), 2.15 (s, 3H), 2.10 (s, 3H), 2.09 (s, 3H); MS (m/e) (positive-FAB) 428, 430 (M + H)⁺ (relative peak height ratio was 3:1).

4.19. (2R, 3S, 4S, 5R)-2-(6-Chloro-2-iodo-purin-9-yl)-3,4-diacetoxy-5-acetoxymethyl-tetrahydrofuran (24)

To a solution of 23 (206 mg, 0.48 mmol) in MeCN (0.50 mL) and CH₂I₂ (2.0 mL) was added t-butyl nitrite (0.200 mL, 2.22 mmol), and the mixture was stirred at 80 °C for 3 h. The reaction mixture was directly purified by silica gel column chromatography (eluent: chloroform then AcOEt/petroleum ether 2/1), to give 2-iodo compound 24 (178 mg, 69%). ¹H NMR = (CDCl₃) δ 8.20 (s, 1H), 6.21 (d, 1H, J = 5.4 Hz), 5.78 (t, 1H, J = 5.6 Hz), 5.59 (t, 1H, J = 5.0 Hz), 4.48 (t, 1H, J = 3.9 Hz), 4.42 (m, 2H), 2.18 (s, 3H), 2.15 (s, 3H), 2.11 (s, 3H); MS (m/e) (positive-FAB) 539, 541 (M + H)⁺ (relative peak height ratio was 3:1).

4.20. (2R, 3S, 4S, 5R)-2-(6-Chloro-2-cyano-purin-9-yl)-3,4-diacetoxy-5-acetoxymethyl-tetrahydrofuran (25)

A solution of Pd₂ (dba)₃–CHCl₃ (50 mg, 0.048 mmol) and tri-2-furylphosphine (150 mg, 0.646 mmol) in tetramethylurea (0.50 mL) was stirred at 80 °C for 30 min. The solution was cooled to rt and treated with 24 (110 mg, 0.204 mmol) and Zn(CN)₂ (125 mg, 1.06 mmol) in tetramethyl urea (3.50 mL) and stirred at 80 °C for 22 h. The reaction mixture was passed through a short column (SiO₂ 3.0 mL and eluent AcOEt), and the eluted fraction was evaporated to dryness. The residue was purified twice by silica gel column chromatography (first column eluent: AcOEt/petroleum ether 2/1, second column eluent: AcOEt/petroleum ether 1/1), = to give 2-cyano compound 25 (52 mg, 59%). = 1H NMR (CDCl₃) δ 8.51 (s, 1H), 6.28 (d, 1H, J = 5.7 Hz), 5.77 (t, 1H, J = 5.7 Hz), 5.54 (dd, 1H, J = 5.7, = 3.9 Hz), 4.53 (dd, 1H, J = 3.9, 7.5 Hz), 4.44 (m, 2H), 2.19 (s, 3H) 2.17 (s, 3H), 2.11 (s, 3H); MS (m/e) (CI) 438 (M + H)⁺.

4.21. (2R, 3S, 4S, 5R)-2-(2-Cyano-6-(3-iodo-benzyl-amino)-purin-9-yl)-3,4-diacetoxy-5-acetoxymethyl-tetrahydrofuran (26)

To a solution of 25 (8.3 mg, 0.019 mmol) in DMF (0.10 mL) was added 3-iodobenzylamine hydrochloride (9.0 mg, 0.033 mmol) and i-Pr₂NEt (0.10 mL), and the mixture was stirred at rt for 25 h. The solvent was removed under reduced pressure, and the residue was purified by short-column chromatography (silica gel, eluent: AcOEt), to give triacetate 26 (10.5 mg, 87%). ¹H NMR (CDCl₃) δ 8.06 (s, 1H), 7.73 (s, 1H), 7.64 (d, 1H, J = 7.8 Hz), 7.36 (d, 1H, J = 7.8 Hz), 7.09 (t, 1H, J = 7.8 Hz), 6.36 (br s, 1H), 6.18 (d, 1H, J = 5.7 Hz), 5.78 (t, 1H, J = 5.6 Hz), 5.57 (t, 1H, J = 5.1 Hz), 4.80 (br, 2H), 4.45 (m, 1H), 4.40 (m, 2H), 2.17 (s, 3H) 2.16 (s, 3H), 2.10 (s, 3H); MS (m/e) (CI) 635 (M + H)⁺.

4.22. (2R, 3S, 4S, 5R)-Acetic acid 4-acetoxy-2-acetoxymethyl-5-(6-chloro-2-trifluoromethyl-purin-9-yl)-tetrahydrofuran-3-yl ester (27)

To a suspension of activated Zn powder (140 mg, 2.20 mmol) in DMF (0.50 mL) was added CF₂Br₂ (0.10 mL, 1.08 mmol), and the mixture was stirred at rt for 2.5 h. HMPA (0.40 mL) was added to the reaction mixture, and it was cooled to 0 °C. CuI (76 mg) was added to the reaction mixture. A solution of 24 (76 mg, 0.141 mmol) in DMF (0.50 mL) was added at 0 °C. The reaction mixture was stirred at 50 °C for 26.5 h and filtered. The solvent was removed under reduced pressure and the residue was purified by column chromatography (silica gel, solvent AcOEt/petroleum ether = 2/1), to give 27 (52 mg, 77%). ¹H NMR (CDCl₃) δ 9.09 (br s, 1H), 6.34 (d, 1H, J = 5.4 Hz), 5.84 (t, 1H, J = 5.4 Hz), 5.62 (t, 1H, J = 5.0 Hz), 4.55 (m, 1H), 4.45 (m, 2H), 2.20 (s, 3H), 2.15 (s, 3H), 2.09 (s, 3H); MS (m/e) (positive-FAB) 481 (M + H)⁺.

4.23. (2R, 3S, 4S, 5R)-Acetic acid 4-acetoxy-2-acetoxymethyl-5-(6-methylamino-2-trifluoromethyl-purin-9-yl)-tetrahydrofuran-3-yl ester (28)

A solution of 27 (8.8 mg, 0.019 mmol) in 2 N methylamine in THF (2.0 mL) was stirred at rt for 20 h and at 70 °C for 24 h. The solvent was removed under reduced pressure, and the residue was purified by column chromatography (silica gel, eluent: CHCl₃/MeOH = 5/1) and preparative TLC (silica gel, thickness 0.25 mm, solvent AcOEt/petroleum ether = 1/1), to give 28 (2.6 mg, 29%). ¹H NMR (CDCl₃) δ 7.98 (s, 1H), 6.16 (d, 1H, J = 4.8 Hz), 6.04 (br s, 1H), 5.83 (t, 1H, J 5.3 Hz), 5.72 (t, 1H, J = 5.1 Hz), 4.50–4.35 (m, 3H), 3.22 = (s, 3H), 2.17 (s, 3H), 2.10(s, 3H), 2.09 (s, 3H); MS (m/e) (positive-FAB) 476 (M + H)⁺, 498 (M + Na)⁺.

4.24. (2R, 3S, 4S, 5R)-Acetic acid 4-acetoxy-2-acetoxymethyl-5-(6-(3-iodo-benzylamino)-2-trifluoromethyl-purin-9-yl)-tetrahydrofuran-3-yl ester (29)

To a solution of 27 (16.2 mg, 0.034 mmol) in DMF (0.40 mL) was added 3-iodobenzylamine hydrochloride (30 mg, 0.111 mmol) and i-Pr₂NEt (0.10 mL), and the mixture was stirred at rt for 24 h. The solvent was removed under reduced pressure, and the residue was purified by column chromatography (silica gel, solvent AcOEt) and preparative TLC (silica gel, thickness 0.25 mm, solvent AcOEt/petroleum ether 1/1), to give 29 (9.5 mg, 42%). ¹H NMR (CDCl₃) δ 8.00′(s, 1H), 7.78 (s, 1H), 7.63 (d, 1H, J = 8.0 Hz), 7.38 (d, 1H, J = 8.0 Hz), 7.08 (t, 1H, J = 8.0 Hz), 6.31 (br s, 1H), 6.16 (d, 1H, J = 5.1 Hz), 5.84 (t, 1H, J = 5.3 Hz), 5.71 (t, 1H, J = 5.1 Hz), 4.80 (br, 2H), 4.50–4.30 = (m, 3H), 2.16 (s, 3H), 2.10 (s, 3H), 2.09 (s, 3H); MS (m/e) (positive-FAB) 678 (M + H)⁺, 700 (M + Na)⁺.

4.25. Biological assays

4.25.1. Cell culture and membrane preparation

CHO cells expressing recombinant human and rat A₃ARs were cultured in DMEM (Dulbecco’s modified Eagle’s medium) 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, the cells were centrifuged at 500g for 10 min. The pellet was resuspended in 50 mM Tris–HCl buffer (pH 8.0) containing 10 mM MgCl₂ and 1 mM EDTA. The suspension was homogenized with an electric homogenizer for 10 s, and was then recentrifuged at 20,000g for 20 min at 4 °C. The resulting pellets were resuspended in buffer containing 3 units/mL of adenosine deaminase, and the suspension was stored at −80 °C prior to the binding experiments. The rat A₃AR was expressed recombinantly via transfection in CHO cells, and the procedure was the same as for the human subtype. The protein concentration was measured using the Bradford assay.³⁰

4.25.2. Binding assay

For the A₃AR binding experiments, the procedures used were similar to those previously described.⁶ Briefly, each tube contained 100 lL of membrane suspension, 50 μL of [¹²⁵I]I-AB-MECA (final concentration 0.5 nM), and 50 lL of increasing concentrations of compounds in Tris–HCl buffer (50 mM, pH 7.4) containing 10 mM MgCl₂, 1 mM EDTA. Non-specific binding was determined using 10 μM NECA (5′-N-ethyluronamidoadenosine). 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 (Brandel, Gaithersburg, MD). 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 recombinant hA₁AR and the binding of [³H]CGS21680 to the recombinant hA_2AAR was performed as previously described.^7,31

4.25.3. Cyclic AMP accumulation assay

Intracellular cyclic AMP levels were measured with a competitive protein binding method.³² CHO cells expressing recombinant human³³ and rat³⁴ ARs were harvested by trypsinization. After resuspension in the medium, cells were plated in 24-well plates in 0.5 mL medium/well. After 24 h, the medium was removed and cells were washed three times with 1 mL/well of DMEM, containing 50 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, pH 7.4. Cells were then treated with agonists and/or test compounds in the presence of roli-pram (10 μM) and adenosine deaminase (3 units/mL) and incubated at 37 °C. For A_2A and A_2BARs, incubation was carried out for 1 h. For A₁ and A₃ARs, after 45 min forskolin (10 μM) was added to the medium, and incubation was continued for an additional 15 min. The reaction was terminated upon removal of the medium, and the cells were lysed with 200 μL/well of 0.1 M ice-cold HCl. The cell lysate was resuspended and stored at −20 °C. 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. Bound radioactivity was separated by rapid filtration through Whatman GF/C filters under reduced pressure and washed once with cold buffer. Bound radioactivity was subsequently measured by liquid scintillation spectrometry.

4.26. Statistical analysis

Binding and functional parameters were estimated with GraphPAD Prism software (GraphPAD, San Diego, CA). IC₅₀ values obtained from competition curves were converted to K_i values using the Cheng–Prusoff equation.³⁵ Data were expressed as mean ± standard error.

4.27. Molecular modeling

All calculations were performed on a Silicon Graphics Octane workstation (300 MHz MIPS R12000 (IP30) processor). All ligand structures were modified from the lowest-energy conformation of Cl-IB-MECA⁶ with the Sketch Molecule of SYBYL 6.9.³⁶ In all cases, MMFF force field³⁷ and charge were applied using distance-dependent dielectric constants and conjugate gradient method until the gradient reached 0.05 kcal mol⁻¹ Å⁻¹.

A hA₃AR model (PDB code: 1o74) constructed by homology to the X-ray structure of bovine rhodopsin with 2.8-Å resolution³⁸ was used for the docking study. Compounds 1, 2, 7, 10, 14, and 15 in Table 1 were docked within the hA₃AR model. The atom types of all ligands were manually assigned with Amber all-atom force field,³⁹ and their charges were calculated before docking. The starting geometry of ligand conformation was chosen from the hA₃AR complex model with Cl-IB-MECA, which was already validated by point mutation.⁶ The ribose-binding position of this series was fixed, using an atom-by-atom fitting method for the carbon atoms of the ribose ring. To determine the binding region of N⁶ derivatives and C-2 substituents, the flexible bonds, χ₁ (O–C₁′ –N₉–C₄) and t₀ (C₅–C₆–N₆–C) angles were variously searched in the putative binding cavity. χ₁ was rotated by −60°, −110°, and −160° within the anti-conformational range and was rotated in 30° increments. Several conformations without any steric bump were selected for further optimization. The initial structures of all complexes were optimized using the Amber force field with fixed dielectric constant of 4.0 and terminating gradient of 0.1 kcal mol⁻¹ Å⁻¹.