Purine Derivatives as Ligands for A₃ Adenosine Receptors

Abstract

Selective agonists and antagonists for A₃ adenosine receptors (ARs) are being explored for the treatment of a variety of disorders, including brain and heart ischemic conditions, cancer, and rheumatoid arthritis. This review covers both the structure activity relationships of nucleoside agonist ligands and selected antagonists acting at this receptor and the routes of synthesis. Highly selective agonists have been designed, using both empirical approaches and a semi-rational approach based on molecular modeling. The prototypical A₃ agonists IB-MECA 10 and the more receptor-subtype-selective Cl-IB-MECA 11, both of which have affinity in binding to the receptor of ~ 1 nM, have been used widely as pharmacological probes in the elucidation of the physiological role of this receptor. In addition to the exploration of the effects of structural modification of the adenine and ribose moieties on A₃AR affinity, the effects of these structural changes on the intrinsic efficacy have also been studied in a systematic fashion. Key structural features determining A₃AR interaction include the N⁶-benzyl group, 2-position substitution such as halo, substitution of ribose (e.g., the (N)-methanocarba ring system, various 2′- and 3′-substitutions and 4′-thio substitution of oxygen). Conformational studies of the ribose moiety and its equivalents indicate that the ring oxygen is not required and the North (N) ring conformation is preferred in binding to the A₃AR. Using these observations, a series of ring constrained (N)-methanocarba 5′-uronamide derivatives was recently reported to be highly selective A₃AR agonists, the most notable amongst them was MRS3558 113 having a K_i value in binding to the human A₃ receptor of 0.3 nM.

INTRODUCTION

Extracellular adenosine 1 acts as a local modulator with a generally protective function in the body. Its effects include preconditioning the heart against ischemia, counteracting the damaging effects of excitotoxicity and seizure activity in the brain, and suppressing an excessive immune and inflammatory response [1]. From basic science considerations, there are numerous possible therapeutic modalities for adenosine agonists and antagonists. There are four subtypes of adenosine receptors (A₁, A_2A, A_2B, and A₃); of which the A₃ receptor was most recently identified as a result of its cloning from various species [2]. Adenosine may be released from cells as such or generated by the action of ectonucleotidases on extracellular adenine nucleotides. Extracellular adenosine levels are quite variable, depending on the tissue and the degree of stress experienced, thus the stimulation levels by the endogenous agonist of the four subtypes vary enormously. Depending on receptor density, the A₃ receptor may be activated endogenously by higher concentrations of adenosine than are required for activation of the A₁ /A_2A receptors. The effector mechanisms for the A₃ receptor are inhibition of adenylate cyclase and stimulation of phospholipase C [1, 3]. Both protective (usually at nM concentrations) and damaging effects (usually at μM concentrations) of A₃ agonists have been studied [3]. The receptor is distributed at low, diffuse levels in the brain and in the human periphery where it is present in lungs, liver, heart, and immune and inflammatory cells such as neutrophils, macrophages, and eosinophils [2].

Highly selective A₃ receptor agonists have been designed, using both empirical approaches and a semi-rational approach based on molecular modeling [4]. Present review will summarize the structure activity relationships at A₃ receptors upon altering the chemical modification of adenosine and also provide synthetic routes used to prepare the most important nucleoside ligand probes of this receptor. A table of K_i values from binding experiments for the agonists and antagonists described will be found at the end of the paper.

NUCLEOSIDES AS A₃ RECEPTOR AGONISTS

With the exception of derivatives of pyridine and xanthine-7-ribosides, nearly all of the known adenosine A₃-agonists are derivatives of adenosine [3]. The 5′-alkyluronamide NECA 2 is a nonselective, highly potent adenosine agonist, which was initially found to be among the most potent in binding to the receptor [5]. NECA was used as an agonist in the initial deorphanization and characterization of the A₃ receptor and was also used subsequently as a radioligand of moderate affinity for the human A₃ receptor [6].

It was also observed that among N⁶-substituted adenosine derivatives, the benzyl moiety provided favorable deselection of high affinity at the A₁ and A_2A receptors, thereby increasing the A₃ receptor selectivity. N⁶-benzyl substituted adenosine derivatives tend to have similar potency for human and rat A₃ARs while N⁶-methyl substitution increased affinity only for the human A₃AR. Other small N⁶ substitution (e.g., methoxy, ethyl) provides high affinity at the human but not the rat A₃ receptor [7]. The N⁶-(2-phenylethyl) substitution of adenosine produced considerably higher affinity for the A₃ subtype, although the affinities at the A₁ and A_2A receptors were also high, thus there was less selectivity than for N⁶-benzyl substitution. The first radioligand used by Stiles and coworkers to study the rat A₃AR was [¹²⁵I]APNEA, 3 which is an iodination product of N⁶-[2-(4-aminophenyl)ethyl] derivative of adenosine, but which also binds tightly to A₁ receptors [8].

In the same context Ji et al. designed a chemically-reactive receptor probe, an adenosine derivative 4 bearing an isothiocyanate group and have shown that it is selective for A₃ receptors in models of reversible and irreversible binding in the membranes from CHO cells stably transfected with rat brain A₃ receptors and in RBL-2H3 cell membranes [9].

The first entry into A₃ selective agonists was accomplished by Gallo-Rodriguez et al. with adenosine derivatives modified at the 5′- position as a uronamide followed by substitution of the N⁶-amino functionality with benzyl substituents [5]. Thus, the combination of the N⁶-benzyl and NECA-like 5′-N-alkyluronamido groups provided Bn-NECA 9, which was 7-fold selective for the rat A₃ receptor in comparison to rat A₁ and A_2A receptors. The initial synthesis of A₃ selective 5′-uronamide-N⁶-benzyl derivatives was achieved as shown in Scheme 1 by first oxidizing the 2′, 3′-isopropylidene adenosine to the corresponding acid 5 and subsequently converting it to the 5′-uronamide derivative 6. The N⁶-benzyl-5′-N-alkyluronamides and related derivatives were obtained by a Dimroth rearrangement of N¹-alkylated adenosine derivatives 7. The first systematic study of the effects of adenosine substitution on affinity and selectivity at an A₃ receptor was done by the same authors, which also dealt with the molecular modeling of the rat A₃ receptor protein based on a bacteriorhodopsin template [10].

Scheme (1).

Reagents: (a) (i)Acetone, TsOH; (ii) CrO₃; (b) SOCl₂, R′-NH_2; (c) R-Halide; (d) NH₄OH.

Refinement of combined adenosine 5′,N⁶-substitution of the first marginally selective A₃ agonist Bn-NECA 9 led to adenosine derivatives of greater selectivity for the A₃ receptor. A systematic study of substitution at the N⁶-position identified the 3-iodobenzyl group as one particularly suited for A₃ selectivity [5]. Methyl 1-[N⁶-(3-iodobenzyl)-adenin-9-yl]-β-D-ribofuronamide, IB-MECA 10 was identified in that study as being ~50-fold selective for the rat A₃ in comparison to the rat A₁ and A_2A receptors. The prototypical A₃ agonists IB-MECA and the more receptor-subtype selective Cl-IB-MECA 11, both of which have affinity in binding to the receptor of ~ 1 nM, have been used widely as pharmacological probes in the elucidation of the role of this receptor in the body [11]. Another important derivative related to IB-MECA is the radioiodinated ligand (¹²⁵I-AB-MECA), [¹²⁵I]N⁶-(4-amino-3-iodobenzyl)-5′-N-methylcarboxamidoadenosine 12 with a K_d value of 0.59 nM at the human A₃ subtype, which is used widely for screening of new ligands at the receptor [12]. This iodinated agonist derivative, in unlabeled form, was found in binding competition experiments to be a potent A₃ ligand but not highly selective in comparison to A₁ receptors.

Knutsen and colleagues at Novo Nordisk A/S replaced the bulky purine 6-amino substituents with the smaller N-methoxy group and also introduced a new range of 5′-ribose modifications by introducing the bioisosteric isoxazole moiety [13]. The target molecules retained the potent binding to the recombinant human A₃ receptor. They synthesized novel isoxazole derivatives NNC 53–0083, (18, K_i = 7.8 nM) and NNC 53–0082, (19, K_i = 31 nM) as shown in the Scheme 2. The isoxazole was assembled at the ribose 4-position and was subsequently coupled with 2,6-dichloro-9H-purine by employing the Vorbrüggen procedure.

Scheme (2).

Reagents : (a) NH₂OH.HCl, pyridine; (b) NBS, (CH₃)₃Si-acetylene, Et₃N, 0°C (c) 1N NaOH, used in (CH₃)₃Si case (d) Dowex 50 H⁺, BzCl, pyridine, DMAP, 0°C (e) Ac₂O, AcOH, H₂SO₄ (f) 2,6-dichloro-9H-purine, 140°C (g) CH₃ONH₂.HCl, Et₃N, or 3-I-PhCH₂NH₂.HCl, Et₃N (h) 10% NH₃ in MeOH.

Acylation of the N⁶-amine (urea derivatives 22–24) of adenosine 5′-uronamides has also been shown to enhance agonist potency at the A₃ receptor but with only moderate selectivity in comparison to other subtypes [14]. These derivatives were prepared readily by reaction of 2′,3′-isopropylidene-NECA or MECA with appropriate aryl isocyanate in the presence of triethylamine as catalyst as illustrated in Scheme 3. In a later study these modifications were combined with substitution at the 2-position.

Scheme (3).

Reagents: (a) R′-N=C=O, Dioxane, heat, Et₃N; (b) HCl 1N, Dioxane, heat.

Kim et al. have studied extensively the substitution of adenine with a xanthine moiety, and have prepared 1,3-dialkylxanthine 7-riboside analogues modified at 1-,3-and 8-purine positions as well as at the ribose 5′-position [15]. These compounds were prepared by condensing silylated xanthines 26 with 1-acetyl-2,3,5-tri-O-benzoylribofuranoside 25 using potassium nonaflate and trichlorosilane as Lewis acid catalyst (Scheme 4). They observed that the 1,3-dibutylxanthine 7-riboside was well tolerated at the receptor site, however A₃ efficacy was diminished. The efficacy-promoting effects of the 5′-uronamide group are sufficiently strong such that one of these derivatives, DBXRM 30 was a selective and moderately potent, full A₃-agonist [15].

Scheme (4).

Reagents: (a) potassium nonaflate, SiHCl₃, CH₃CN; (b) NH₃/MeOH; (c) p-TsOH,acetone; (d) RuCl₃, NaIO₄, CHCl₃-CH₃CN-H₂O (2:2:3); (e) EDAC, DMAP, MeOH; (f) 88% HCO₂H; (g MeNH₂, MeOH, 85°C,

The effects of substitution at the 2-position of the adenine moiety on A₃ receptor affinity have been explored extensively. In this context Tilburg et al. prepared 2-(N-3-methyl-1-butylidenehydrazino)-adenosine 33 starting from 2-iodoadenosine 31 as depicted in Scheme 5 [16]. Interestingly, they found that this 2-substituted derivative with an intact 5′-hydroxyl group showed partial agonism toward the A₃ as well as the A_2A receptor.

Scheme (5).

Reagents: (a) 2-propanol, hydrazine monohydrate, reflux; (b) MeOH, isovalderaldehyde, reflux.

IJzerman and colleagues also studied 2-nitro substitution and prepared several analogues with combination of a nitro group at the 2-position with several N⁶ substituents, such as cyclopentyl and m-iodobenzyl, etc. 2-Nitroadenosine was prepared from tri-O-acetyl-protected 6-chloropurine-9-riboside, 34 which is readily available from inosine [17].

Nitration was accomplished using tetrabutylammonium nitrate/trifluoroacetic anhydride followed by treatment with amines at lower temperature and removal of the acetyl protection to afford 2-nitroadenosine analogues as outlined in Scheme 6. They observed that the A₃ receptor seemed to accommodate the nitro substituent very well. For example, 2-nitroadenosine analogue 36 with a benzyl group at N⁶-amine was shown to have a K_i value of 163 nM versus 550 nM for the analogous derivative without nitro at the 2-position. Introduction of a 3-iodobenzyl moiety enhanced both affinity as well as selectivity for A₃ receptor.

Scheme (6).

Reagents: (a) TBAN/TFAA, DCM, 0°C; (b) RNH₂,TEA, DMF, 0°C; (c) KCN, MeOH.

Cristalli et al. have prepared a number of 2-alkynyl derivatives of 5′-N-ethylcarboxamidoadenosine (NECA), which were investigated for their affinity and selectivity at human A₃ adenosine receptors [18,65]. They prepared these compounds by a modification of the palladium-catalysed cross-coupling reaction as depicted in Scheme 7. In binding studies they found that the most potent compound, 2-(3-hydroxy-3-phenyl)propyn-1-yl-NECA (PHPNECA) 39, exhibited subnanomolar affinity for the human A₃ receptor with a K_i value of 0.4 nM. Diastereomers in this series have now been resolved [65].

Scheme (7).

Reagent s: (a) CuI, Ph₃P-PdCl₂,RC≡H

Ohno et al. introduced several small substituents, such as cyano, trifluoromethyl, or a simple ester at the 2-position of adenosine, and combined these groups with various A₃AR affinity-enhancing N⁶-substituents [19]. They found that nucleoside 44 having a 2-cyano functionality was a selective hA₃AR agonist with K_i of 3.4 nM. Nucleosides 45 and 47 with methyl ester or trifluoromethyl functionality, respectively, at the 2-position were also full agonists of lower potency. These nucleosides were prepared as shown in the Scheme 8 in which the cyano group was introduced by applying palladium chemistry. The trifluoromethyl group was introduced by treatment with a ‘CF₃Cu’ species, which was generated in situ from CF₃ZnBr and CuI (Scheme 8).

Scheme (8).

Reagents: (a) Ac₂O, DMF, pyridine; (b) t-BuONO, CH₂I₂; (c) Pd₂(dba)₃, P(2-Fur)₃, Zn(CN)₂, tetramethylurea; (d) CH₃NH₂, MeOH, THF; (e) NaOH, H₂O₂, H₂O, TMSCHN₂, MeOH; (f) Zn, CF₂Br₂, DMF, CuI, HMPA; (g) MeNH₂, NaOH, MeOH.

A 2-chloro substitution generally improved the binding affinity at A₃ receptors. For example, early in the study of A₃ agonists, it was found that Cl-IB-MECA 11 is a highly selective, full agonist with K_i value of 0.33 nM at the rat A₃ receptor [11]. Cl-IB-MECA was synthesized by condensation of the sugar 5′-uronamide moiety 53 with a purine moiety 54. The key sugar moiety 53 was synthesized starting from commercially available methyl β-ribofuranoside 48 as depicted in Scheme 9. The key step involved the oxidation of the 5′-position of 50 with ruthenium oxide to afford 51 after methylation. The glycosidic bond was formed on treatment of the 1′-O-acetyl riboside derivative 53 and the silylated adenine derivative 54. Subsequent treatment with ammonia gave Cl-IB-MECA 11 in good yield.

Scheme (9).

Reagents : (a) TBDPSiC1, DMAP, DMF, room temperature; (b) Bz₂O,Py; (c) n-Bu₄NF, THF; (d) RuO₂, NaIO₄, CHCl₃-CH₃CN-H₂O (2:2:3); (e) EDAC, DMAP, MeOH; (f) MeNH₂, THF, 75°C; (g) BzCl, Py-CH₂Cl₂; (h) Ac₂O, H₂SO₄, AcOH; (i) TMSOTf,CH₂Cl₂; (j) Ammonia, MeOH.

Recently, Zablocki and colleagues at CV Therapeutics reported the synthesis of a novel class of 2-pyrazolyl-N⁶-substituted adenosine analogues as high affinity and selective adenosine A₃ receptor agonists [20]. The pyrazole moiety at 2-position was introduced by condensing the 2-hydrazino-N⁶-substituted adenosine 58 with appropriate malonaldehyde depicted in Scheme 10. They found that one of the N⁶-methyl analogues 60 showed good binding affinity at A₃AR with K_i 73 nM. The replacement of carboxamide group in 60 with different heteroaryl groups resulted in several analogues with high binding affinities and selectivity for A₃AR. For example, compound 62 (K_i = 2 nM) with 2-pyridyl substitution on the pyrazoyl ring showed extremely high selectivity for A₃ AR versus A₁ and A₂ AR.

Scheme (10).

Reagents: (a) CH₃NH₂, EtOH, reflux; (b) NH₂.NH₂; (c) 2,2-diformylacetate, EtOH, reflux; (d) CH₃NH₂, reflux.

In the context of ribose modifications as discussed above, the 5′-hydroxyl group of the ribose ring may be replaced with a limited set of moieties. The 2′ and 3′-hydroxyl groups are generally required for affinity and/or the ability to fully activate the receptor, i.e. to achieve full efficacy. For example, 3′-hydroxyl group may be replaced with 3′-amino [21], but not with bioisosteric fluorine [22]. A 3′-fluoro analogue 63 of Cl-IB-MECA showed decreased binding affinity and complete loss of efficacy, indicating the essential role of 3′-hydroxy group in receptor activation as hydrogen bonding donor. The 3′-deoxy-3′-amino analogues are expected to maintain affinity at the A₃ receptor when combined with the appropriate N⁶-substitution [21]. For example, DeNinno et al. at Pfizer reported CP608039 68, which displayed full agonist activity at human A₃ receptor with an EC₅₀ of 3.4 nM and is intended for use in perioperative cardioprotection [21]. The 3′-amino group, which is mainly charged at physiological pH, also enhances water-solubility when formulated in acidic buffers.

The synthesis of CP608039 as outlined in Scheme 11, started with the glucose diacetonide 64. Selective synthetic manipulations provided the amide 66. Hydrolysis of the remaining acetonide followed by Vorbrüggen glycosidation afforded intermediate 67. Reaction of 67 with the appropriately substituted benzylamine and the reduction of the azide group furnished the requisite CP608039 68 [21]. It is important to mention here that 3′-amino-3′-deoxy analogues are also useful for the design of neoligands in relation to the concept of neoceptors. This refers to a means of engineering the binding site of a seven transmembrane-spanning (7TM) receptor for unique recognition of synthetic ligands [23].

Scheme (11).

Reagents : (a) Tf₂O, pyridine, CH₂Cl₂, −20° C; (b) NaN_3, DMF, room temp; (c) HIO_4, THF-H₂O; (d) RuO_2, NaIO₄, CHCl₃, CH₃CN, H₂O; (e) (COCl)₂, CH₂Cl₂; (f) MeNH₂, CH₂Cl₂; (g) HOAc, Ac₂O, H₂SO₄; (h) TMS-6-chloropurine, TMSOTf, DCE, 60° C; (i) Et₃N, MeOH; (j) 5-Chloro-2-(3-methylisoxazol-5-yl methoxy) benzylamine, EtOH, 60° C; (k) Ph₃P, NH₄OH, THF-H₂O.

Along similar lines Van Rompaey et al. have synthesized a series of adenosine derivatives having the 3′-amino-3′-deoxy or 3′-aminomethyl-3′-deoxy modification, with the result being moderate selectivity in binding to the human A₃ receptor [24]. They concluded that 3′-amino function resulted in partial agonist activity, where as introduction of methylene spacer between amino functionality and the ribofuranose ring has overall efficacy- and affinity- lowering effect. The requisite aminomethylene compounds were prepared as depicted in Scheme 12. The 3-C-azidomethyl sugar 72 was prepared in six steps from 1,2-O-isopropylidene-D-xylofuranose 69 by the reported procedures [24]. Vorbrüggen-type coupling of 72 with silylated 6-chloropurine followed by the displacement of the chloro atom with 3-iodobenzylamine, deprotection of the acetyl group, and reduction of the azido moiety gave access to the 3′-C-aminomethyl nucleoside 74.

Scheme (12).

Reagents : (a) (i) TBAF, THF, rt; (ii) toluoyl chloride, pyridine, rt; (b) (i) 50% CH₃COOH, 50°C (ii) (CH₃CO)₂O, pyridine, rt; (c) silylated 6-chloropurine, TMSOTf, 1,2-dichloroethane, reflux; (d) (i) 3-iodobenzylamine HCl, Et₃N, EtOH, reflux (ii) 7 N NH₃ in methanol, rt; (e) Ph₃P, NH₄OH, pyridine, rt.

Jeong et al. have designed and synthesized novel 4′-thio analogues of aryl and alkyl uronamides and explored their effects on adenosine receptor selectivity [25]. One of the most potent compounds was 2-chloro-N⁶-methyl-4′-thioadenosine-5′-methyluronamide 87, which had a K_i value of 0.28 nM at the human A₃ receptor. This was prepared starting from D-gulono-g-lactone 75 via 4-thioribosyl acetate 83 as a key intermediate as outlined in Scheme 13. Thus, glycosyl donor 83 was condensed with 2,6-dichloropurine followed by reaction with 3-iodobenzylamine and subsequent oxidation and esterification. Finally, methylamine treatment yielded the requisite nucleoside 87.

Scheme (13).

Reagents: (a) CH₃COCH₃, H₂SO₄, CuSO₄, room temp; (b) NaBH₄, MeOH; (c) MsCl, Et₃N, CH₂Cl₂; (d) Na₂S, DMF, heat; (e) 30% AcOH; (f) Pb(OAc)₄, EtOAc; (g) NaBH_4, MeOH; (h) BzCl, pyridine; (i) mCPBA, CH₂Cl₂; (j) Ac₂O; (k) silylated 2,6-dichloropurine, TMSOTf; (l) CH₃NH_2; (m) BzCl, pyridine; (n) 80% AcOH; (o) TBSOTf, pyridine; (p) NaOMe, MeOH; (q) (i) PDC, DMF, (ii) K₂CO₃, (CH₃O)₂SO₂; (r) 2 N MeNH₂, THF; (s) n-Bu₄NF, THF.

An important ribose substitution that has enhanced A₃ selectivity in a general manner is its replacement with the (N)-methanocarba ring system, as studied extensively by Jacobson and colleagues [26]. By this approach, the ribose moiety is replaced by a fused cyclopentane-cyclopropane (bicyclo[3.1.0]hexane) ring system. The effect of this substitution on A₃ receptor affinity and selectivity is superior to simple carbocyclics, such as the mixed A₁/A_2A adenosine agonist 89 AMP579 1S-[1a,2b,3b,4a(S*)]-4-[7-[[1-[(3-chloro-2-thienyl)methyl]propylamino]-3H-imidazo[4,5-b] pyridin-3-yl]-N-ethyl-2,3-dihydroxy cyclopentanecarboxamide [27].

Molecular modeling studies and receptor docking of the ribose moiety and its equivalents indicate that the ring oxygen is not required for receptor interaction, and the North (N) ring conformation is preferred in binding to the A₃ receptor. Use of the bicyclo[3.1.0]hexane ring system, which assumes a (N)-envelope conformation, is a highly effective means of locking the conformation of the ribose-like ring [26]. Initially these complex pseudonucleosides were prepared (Scheme 14) from the optically active cyclopentenone 93, which in turn was obtained from D-ribose [28] following a rather lengthy procedure, was reduced to the allylic alcohol 94. Simmons-Smith cyclopropanation of 94 produced the requisite bicycle [3.1.0]hexane intermediate 95. Mitsunobu reaction followed by amine displacement and the deprotection of the benzyl group afforded the requisite (N)-methanocarba nucleosides such as 98 and 99.

Scheme (14).

Reagents: (a) NaBH₄, CeCl₃; (b) Et₂Zn, CH₂I₂, 0° C; (c) DEAD, Ph₃P, 2,6-dichloropurine or 6-chloropurine; (d) 3- Iodobenzylamine, MeOH; (e) BCl₃.

Alternately, (N)-methanocarba 5′-uronamides were prepared [29] using the common intermediate 96, which was oxidized using NaIO₄/RuO₄. The resulting acid derivatives were substituted at the 6-position by amine treatment, converted to the acid chloride and reacted immediately with methylamine to form 5′-uronamides as shown in Scheme 15. The (N)-methanocarba, 5′-uronamides (e.g., MRS 2346 90, MRS 1898 91) are full agonists, while the 5′-OH analogues MRS 1743 98 and MRS 1760 99 are low efficacy partial agonists [27].

Scheme (15).

Reagents: a) (i) BCl₃, CH₂Cl₂, −78 °C; (ii) p-TsOH, DMP, acetone; (b) NaIO₄, RuO₂, K₂CO₃, MeCN/CHCl₃/H₂O=2:2:3; (c) (i) EDAC, DMAP, MeNH₂, CH₂Cl₂/DMF=1:1; (ii) 10% CF₃CO₂H/MeOH, H₂O; (d) 3-I-benzylamine·HCl, TEA, MeOH; (e) (i) (COCl)₂, 50°C, then MeNH₂, CH₂Cl_2; (ii) 10% CF₃CO₂H/MeOH, H₂O.

Recently, Joshi et al. developed a practical and robust synthetic route for the above (N)-methanocarba nucleosides starting from readily accessible 2,3-O-isopropylidene-D-erythronolactone 102 as depicted in Scheme 16 [30]. The key step involves intramolecular cyclopropanation of the diazo derivative 107 to afford the requisite bicyclo[3.1.0]hexan-2-one derivative 108 as the major isomer. Isomerization of the acetonide 109 followed by a Mitsunobu coupling reaction gave the ester 111. The esters were converted into the requisite (N)-methanocarba nucleosides by following a standard set of reactions in good yields. By following this synthetic approach and employing ester 111 as the building block, Tchilibon et al. have recently reported a highly selective series of (N)-methanocarba 5′-N-methyluronamide derivatives, prominent among them were MRS 3558 113 and MRS 3602 114. MRS 3558 has a K_i value in binding to the human A₃ receptor of 0.3 nM [31]. Based on these observations many of the previously known groups that enhance A₃AR affinity in the 9-riboside series, including those that reduce intrinsic efficacy, may be adapted to the (N)-methanocarba 5′-uronamide nucleoside series of full agonists.

Scheme (16).

Reagents:(a) DIBAL-H, CH₂Cl₂, −78°C; (b) methyltriphenylphosphonium bromide, KOBu^t, THF, −78°C to room temperature; (c) DMSO, (COCl)₂, CH₂Cl₂, −78°C; (d) N₂CHCOOEt, SnCl₂, CH₂Cl₂, room temperature; (e) TsN₃, CH₃CN, TEA, room temperature; (f) CuI, toluene, reflux; (g) NaBH₄, MeOH, room temperature; (h) p-TsOH, acetone, reflux; (i) 2,6-dichloropurine, DIAD, TPP, THF, room temperature; (j) 3-chlorobenzylamine.HCl, TEA (k) 40% aq. MeNH₂; (l) 10% CF₃COOH in MeOH, H₂O, 70°C.

NUCLEOSIDE DERIVATIVES AS A₃ RECEPTOR ANTAGONISTS AND PARTIAL AGONISTS

The search for antagonists of the A₃ receptor began with the discouraging observation that xanthines, such as caffeine and theophylline, the classical adenosine antagonists of A₁, A_2A, and A_2B receptors, are typically very weak in binding to the A₃ receptor [1–3]. The initial observation was made for the rat A₃ receptor (prior to the report of the cloning of the human homologue), at which the common xanthines bind only in the range of 100 μM, but the situation became slightly more hopeful with the cloning of the receptor from other species. At the sheep and human A₃ receptors, the xanthines displayed intermediate affinity (typically 100 nM for 8-phenylxanthine analogues). The species-dependence of affinity at the A₃ receptor, with human greatly exceeding that at the rat receptor, has also been found typically for non-purine A₃ antagonists. Nevertheless, in spite of the intermediate affinity of xanthines at the human A₃ receptor, the search for A₃ antagonists turned toward more novel heterocyclic systems. Recently, the search for antagonists has returned to the fused xanthine scaffold with the design of A₃-selective antagonists consisting of pyridopurine-2,4-dione derivatives like 117 which showed A₃ receptor antagonism with affinities in the low nanomolar range [32]. These compounds were prepared in two steps, starting with reaction of NBS with 6-aminouracil derivatives 115 to generate 5,5-dibromo-6-imino intermediate that reacts in situ with various pyridines to afford 1H, 3H-pyrido[2,1-f]purine-2,4-diones 116 as depicted in Scheme 17. One of these derivatives, a 1-benzyl-3-propyl-1h,3h-pyrido[2,1-f]purine-2,4-dione 117 derivative, had a K_i value of 4.0 nM. In fact, these derivatives can be considered xanthines that have been cyclized between the 7- and 8-positions.

Scheme (17).

Reagents: (a) (i) NBS, CH₃CN, 80°C; (ii) substituted pyridine, 80°C; (b) alkyl halide, DBU, CH₃CN

An alternate approach to designing A₃ antagonists was to start with adenosine derivatives known to bind with high affinity to the receptor, and simply truncate the molecule in stages in an effort to remove the ability to activate the receptor without losing high affinity of binding. An initial attempt to find adenine derivatives, such as 9-alkyl-N⁶-iodobenzyladenines, that displayed these characteristics was unsuccessful [33]. Instead, affinity at the A₃ receptor was greatly diminished.

In a rather similar approach recently Perreira and Jiang et al. have prepared relatively simple 2,6-disubstituted purine derivatives as highly selective A₃ AR antagonists reaching high affinity [34]. Among these compound, 120 was most potent and selective in comparison to A₁ and A2A ARs with K_i value of 0.051 mM. These compounds were prepared starting from 6-chloro-2-fluoropurine and condensing with series of amines in sequential manner as shown in Scheme 18.

Scheme (18).

Reagents : (a) R¹-NH₂ (0.9 equiv.), Hünig’s base, n-BuOH, 80°C, 24 h; (b) R²-NH_2, (2 equiv), EtOH, 110° C, 48 h.

Replacement of 2′- and 3′-hydroxyl groups of known A₃ agonists with H also failed to provide A₃ selective antagonists [35]. The formalistic removal of the 3′-hydroxyl group from IB-MECA resulted in compound 130 which was a full agonist at the rat A₃ receptor and moderately selective for this subtype [33]. This nucleoside 130 was prepared by first condensing the 3′-deoxy sugar moiety 127 with the silylated adenine base by a modified Vorbrüggen method to afford 129 (Scheme 19). Deprotection of 129 with ammonia afforded 3′-deoxy-2-chloro-IB-MECA 130, as shown in Scheme 19. Replacement of the 2’-hydroxyl group with F was highly detrimental to receptor binding, but replacement of the 3′-hydroxyl with F provided antagonists [63].

Scheme (19).

Reagents: (a) i. CS₂, NaH, MeI, THF, ii. Bu₃SnH, Et₃B, benzene; (b) NH₃, MeOH; (c) RuO₂, NaIO₄, CHCl₃:CH₃CN:H₂O (2:2:3); (d) MeOH, EDAC, DMAP; (e) CH₃NH₂/THF; (f) H₂SO₄, Ac₂O, AcOH; (g) TMSOTf, Cl(CH)₂Cl; (h) NH₃/MeOH.

A more successful approach consisted of either: 1) adding substituents to adenosine derivatives or 2) rigidifying the nucleosides, to reduce their intrinsic efficacy. Gradually, with more systematic studies of structure-efficacy relationships upon substitution of adenosine at the N⁶-, ribose [7], and C2 adenine [36] positions, it became apparent that the efficacy at A₃ receptors is more easily diminished upon structural modification than at the other subtypes. In some cases N⁶-substitution of adenosine 5′-OH derivatives with large groups (e.g., substituted benzyl groups or large cycloalkyl rings) reduced the maximal efficacy, leading to partial agonist activity at the A₃-receptor. For example Cl-R-PIA 134 is a partial agonist and N⁶-[2-(3,5-dimethoxyphenyl)-2-(2-methylphenylethyl)] adenosine (DPMA) 135 is an antagonist at human A₃AR (Fig. 5) [7]. It is interesting to note that 2-chloroadenosine was a full agonist, while 2-fluoroadenosine was a partial agonist [35].

Fig. (5).

Based on the SAR data over the years it was concluded that the efficacy of the adenosine derivatives appears to be more dependent on smaller structural changes at A₃ than other subtypes. For example the chloro substitution at 2-position of known A₁ receptor agonists 131 CPA and 133 R-PIA led to A₃ antagonism for their chloro counterparts 132 (CCPA) and partial agonism for 134 (Cl-R-PIA) [35]. Both 2′- and 3′-nucleoside substitutions had pronounced effects on the efficacy of A₃ receptor ligands, although the effect of 2′-substitution was more dramatic. The 4′-thio substitution of oxygen may also diminish efficacy, depending on other substitutions. Both N⁶-methyl and N⁶-benzyl groups may contribute to the A₃ affinity and selectivity; however, an N⁶-benzyl group, but not an N⁶-methyl group, diminishes A₃AR efficacy [7]. The combination of 2-chloro and N⁶-benzyl substitutions appeared to reduce efficacy further than either modification alone. Other examples of simple derivatives of adenosine that have fully reduced efficacy at the A₃ receptor, while perhaps remaining full agonists at other subtypes, are again 132 CCPA (also a potent A₁ agonist), and 135 DPMA (also a moderately selective A_2A agonist) [7,37]. While steric constraint of the ribose moiety, especially near the 5′-position, is associated with loss of A₃ efficacy, for hydrophobic N⁶-substitutions, there are examples in which the steric rigidification is associated with restoration of agonist efficacy. In this context the SAR at the hA₃ receptor of the N⁶-phenylethyl derivatives, including sterically constrained N⁶-(2-phenylcyclopropyl) analogues, has been thoroughly explored by Tchilibon et al. [38]. In their studies a N⁶-cyclopropyl analogue 138 was found to be A₃AR antagonist, and adding one or two phenyl rings at 2-position of the cyclopropyl moiety restored efficacy. Also N⁶-2,2-diphenylethyl adenosine 141 was found to be an antagonist at the human A₃ receptor, while the corresponding N⁶-fluorenylmethyl analogue 142 is a full agonist. They found that affinity and selectivity of these nucleosides is highly dependent on the species examined and on the substitution of distal aryl substitution. Upon probing of the SAR in this series, several novel nucleoside antagonists of A₃AR were identified [38]. Their synthetic route is shown in Scheme 20. The appropriately substituted amines were reacted directly with 6-chloropurines to give the desired adenosine analogues.

Scheme (20).

Reagents: (a) RNH₂, EtOH, TEA, reflux.

Also, Gao et al. [36] found that adenosine derivatives 143 and 144 having sterically bulky substitution of 2,2-diphenylethyloxy or S-2-phenylbutyloxy group at the 2-position of adenosine were found be antagonists of A₃AR (Fig. 6).

Fig. (6).

Rigidification of the ribose ring moiety containing a 5′-uronamide moiety also was shown to reduce A₃ efficacy. For example, the spiro-lactam derivative MRS 1292 154 is relatively a potent and highly selective nucleoside-derived A₃ receptor antagonist [37]. This spiral analogue was prepared in 10 steps by Gao et al. starting from brominated cyclopentenone derivative 145 as depicted in Scheme 21. The preparation of 154 mainly involved the insertion of functionalized side chain in the cyclopentyl moiety and the ozonolysis followed by intramolecular cyclization to afford ribose derivative 148. Reductive elimination of bromine, followed by ozonolysis, and reductive amination gave the lactam 150. Reaction of 150 with ceric ammonium nitrate and treatment with K₂CO₃ and introduction of acetyl groups afforded the lactam 152. Finally, treatment of 152 with silylated 6-chloropurine and trimethylsilyl triflate gave 153. Brief exposure of 153 to ammonia followed by treatment with 3-iodobenzylamine gave the target MRS 1292 154.

Scheme (21).

Reagents: (a) allylmagnesium chloride, THF, −78°C. (b) Br₂, Et₃N. (c) O₃, MeOH/pyridine, −78°C, then dimethyl sulfide. (d) MeOH, TsOH, 65°C. (e) Zn, MeOH, 65°C. (f) (1) O₃, CH₂C1₂, then dimethyl sulfide; (2) p-MeOC₆H₄CH₂NH₂, NaCNBH₃. (g) ceric ammonium nitrate, MeCN/H₂O. (h) K₂CO₃, MeOH, 18-crown-6. (i) (1) HCl (3%), MeOH, 65°C; (2) Ac₂O, Et₃N. (j) hexamethyldisilazane, 6-chloropurine, TMSOTf, MeCN, 85°C. (k) (1) NH₃, MeOH, 20 min; (2) 3-iodobenzylamine hydrochloride, Et₃N, t-BuOH, 85°C, 6 days.

A thorough summary of A₃ antagonists and their synthesis, has been provided in detail recently [39]. Other high affinity antagonists of note include MRE-3008-F20 155 and PSB-11 156, and their congeners, both of which classes have led to high affinity antagonist radioligands for use in characterizing the A₃ receptor (Fig. 7) [40,41].

Fig. (7).

BIOLOGICAL IMPORTANCE OF A₃ AGONISTS AND ANTAGONISTS

Potential therapeutic applications related to A₃ receptor activation are: anti-inflammatory (possibly through depression of TNF-α levels) [42], cardioprotective [43], cerebroprotective [44], and anticancer [45]. IB-MECA 10 is currently in clinical trials for treatment of rheumatoid arthritis and colon carcinoma. The novel anti-cancer effect discovered by Fishman and colleagues is due to a cytostatic effect on tumors related to the Wnt pathway, rather than induction of apoptosis [45]. Expression of A₃ receptors appears to be upregulated in cancer cells. A mouse line lacking the A₃ receptor demonstrated that this is a non-lethal mutation that has inflammatory, cardiovascular, and behavioral consequences [46–48]. The lack of the A₃ receptor in this mouse increased the hypoxic damage to the brain, suggesting that an A₃ agonist might be useful in neurodegenerative diseases, as was indicated previously in ischemic models in gerbils. A functional A₃ receptor has been demonstrated in microglial cells.

Antagonists of the A₃ receptor are of interest for the treatment of glaucoma [49]. Selective A₃ receptor antagonists, including OT-7999 157, (Fig. 8) have been reported to reduce intraocular pressure in nonhuman primates [50,51]. Recently, the nucleoside A₃ antagonist 154 MRS 1292 [37,64], which is more generally applicable across species, was reported to reduce intraocular pressure when administered to mice. Initially, the A₃-modulated release of histamine from RBL basophilic cells suggested that an A₃ antagonist might be useful in treating asthma, however, the situation has been complicated by the fact that release of inflammatory mediators from mast cells upon A₃ receptor activation may be specific to rodent species [52].

Fig. (8).

A₃ receptors appear to be involved in a number of disorders of the central nervous and the cardiovascular systems, as well as in the inflammatory processes. Early in the characterization process, the selective A₃ receptor agonist 10 IB-MECA was demonstrated to induce behavioral depressant effects in mice [53], which was ascribed to a central mechanism. In cerebral ischemia models, chronic pretreatment with 10 IB-MECA resulted in improved postischemic cerebral blood circulation, survival, and neuronal preservation, while the opposite effects were observed when 10 IB-MECA was given acutely [54]. In the heart, A₃ receptor activation induced late preconditioning against infarction in conscious rabbits [55]. In rats, 11 Cl-IB-MECA administered peripherally resulted in a short-lasting hypotension as a result of histamine release [56]. The A₃ receptors are also involved in vasoconstriction in vivo [57]. A₃ receptor activation may be involved both pro- and anti-inflammatory responses in mice [2]. Stimulation of A₃ receptors both enhanced degranulation in vitro and directly produced degranulation of rat mast cells in vivo [58,59]. A₃ receptor agonists inhibited murine macrophage tumor necrosis factor-α production both in vitro and in vivo [60,61]. Mucociliary transport in rabbit trachea in vivo was enhanced by IB-MECA 10 but not by an A₁ or A_2A agonist. Also, the effect of IB-MECA 10 was blocked by the A₃ antagonist MRS 1220 158, but not by an A₁ or A_2A antagonist [62].

In conclusion, soon after the cloning of the A₃ receptor in the early 1990’s, modification of the structure of adenosine and combination of known potency enhancing functionality led to highly selective agonists. A major step forward was the discovery of an A₃ receptor-preferred conformation of the ribose or ribose-like ring. By freezing the nucleosides in this (N)-conformation using the bicyclo[3,1.0]hexane (methanocarba) ring system, higher affinity and selectivity was achieved. Nevertheless, the synthesis of these methanocarba derivatives was arduous, and more improved preparative methods are needed. It became evident that the intrinsic efficacy of the nucleosides may vary widely, depending on substitution of the adenosine scaffold and flexibility of the ribose moiety. This resulted in the design of nucleoside derivatives as selective agonists, antagonists or partial agonists of the A₃ receptor. The further structural refinement of these selective nucleoside ligands is anticipated to provide new substances suitable for therapeutic use.

Fig. (1).

Fig. (2).

Fig. (3).

Fig. (4).

Table 1.

Affinity of Selected A₃ Adenosine Receptor Ligands at Three Receptor Subtypes, and the Maximal Efficacy Observed at the A₃ Adenosine Receptor. The Compound Numbers Correspond to Numbers Used in the Text and Schemes

Ki (nM)
No.	Compound	Reference:	hA1ARa	hA2AARa	hA3ARa	% Max. A3 effectc
2	NECA	6,10,18	6.8±2.4	2.2±0.6	16.0±5.4	100
4	NCS derivative	9	145 ± 41b	272 ± 93b	32.9 ± 7.8b
9	Bz-NECA	10	87.3 ± 13.9b	95.3 ± 24.6b	6.8 ± 2.5b
10	IB-MECA	5,31	51 ± 5	2900 ± 600	1.8 ± 0.7	100
11	Cl-IB-MECA	11,31,56	1240 ± 320	5360 ± 2470	1.4 ± 0.3	100
18	NNC 53–0083	13			7.8
19	NNC 53–0082	13			31
23	N6-urea derivative	14	110 ± 8b	5360 ± 790b	39 ± 16b
30	DBXRM	15	37,300 ± 4600b	19%b	816 ± 206	96
33	2-hydrazino derivative	16		92 ± 2	24 ± 17
37	2-NO2 analogue	17	138 ± 30	1440 ± 790	12.0 ± 3.7
39	PHP-NECA	18	2.7	3.1	0.42	100
44	2-cyano-N6-Me Ado	19	69.8 ± 4.4	23%	3.4 ± 0.8	101
60	N6-Me-2-pyrazole analogue	20	>6000	>6000	73.0
62	N6-Me-2-pyrazole analogue	20	3800	>5000	2.0
63	3′-F-3′-deoxy-Cl-IB-MECA	22,63	6%	0%	460 ± 60	0
68	CP608039	21	7200		5.8
74	3′-deoxy-3′-CH2NH2 analogue	24	72%	22%	8700	9
87	4′-thio N6-Me analogue	25	1330 ± 240	20%	0.28 ± 0.09	119
88	LJ-568	25	193 ± 46	223 ± 36	0.38 ± 0.07	114
90	MRS2346	29,31	2100 ± 1700	6%	2.2 ± 0.6	104
91	MRS1898	29,31	136 ± 22	784 ± 97	1.5 ± 0.2	100
98	MRS1743	26,31	35 ± 3	860 ± 70	9.2 ± 0.7	13
99	MRS1760	26,31	65 ± 17	1600 ± 400	1.9 ± 0.7	3
113	MRS3558	31	260 ± 60	2300 ± 100	0.29 ± 0.04	103
114	MRS3602	31	1600 ± 200	52%	1.4 ± 0.2	107
120	MRS3777	34	26%	16%	47 ± 12	0
130	3′-deoxy- Cl-IB-MECA	33	1030 ± 150b	4660 ± 740b	32.9 ± 7.8b	100b
132	CCPA	35,36	0.83	2270	38 ± 6	0
135	DPMA	10,38	168±29	153±26	106 ± 22	0
140	N6-(2-phenyl-1-cycloPr)	7,38	124 ± 30	2530 ± 720	0.86 ± 0.09	101
141	N6-(2,2-diphenylethyl)	38	49.9 ± 16.2	510 ± 49	3.9 ± 0.7	0
142	N6-(9-fluorenylmethyl)	38	14.0 ± 4.0	145 ± 26	0.91 ± 0.38	99
143	2-(2,2-diphenylethyl)-O	36	38.5	310±119	53.6±10.4	0
144	2-(S-2-phenylbutyl)-O	36	4780±990	26.9±6.9	175±31	0
154	MRS1292	37,64	12,100 ± 2400b	29800 + 400b	29.3 ± 1.6	0
157	OT-7999	51	> 10, 000	> 10,000	0.61	0
158	MRS1220	39	305 ± 51b	52.0 ± 8.8b	0.65 ± 0.25	0
155	MRE 3008-F20	40	1200 ± 100	141 ± 15	0.82 ± 0.08	0
156	PSB-11	41	1640 ± 50	1280 ± 430	3.5 ± 0.4	0

^a)Binding experiments at recombinant human A₁, A_2A, and A₃ARs, unless noted. Values expressed as Ki (nM) ± SEM, except when a percent is indicated, which means % inhibition of binding at 10 μM.

^b)Binding and functional experiments at rat ARs.

^c)as agonist, in inhibition of forskolin-stimulate adenylate cyclase in CHO cells expressing the human A₃AR.