Medicinal Chemistry of the A₃ Adenosine Receptor: Agonists, Antagonists, and Receptor Engineering

Abstract

A₃ adenosine receptor (A₃AR) ligands have been modified to optimize their interaction with the A₃AR. Most of these modifications have been made to the N⁶ and C2 positions of adenine as well as the ribose moiety, and using a combination of these substitutions leads to the most efficacious, selective, and potent ligands. A₃AR agonists such as IB-MECA and Cl-IB-MECA are now advancing into Phase II clinical trials for treatments targeting diseases such as cancer, arthritis, and psoriasis. Also, a wide number of compounds exerting high potency and selectivity in antagonizing the human (h)A₃AR have been discovered. These molecules are generally characterized by a notable structural diversity, taking into account that aromatic nitrogen-containing monocyclic (thiazoles and thiadiazoles), bicyclic (isoquinoline, quinozalines, (aza)adenines), tricyclic systems (pyrazoloquinolines, triazoloquinoxalines, pyrazolotriazolopyrimidines, triazolopurines, tricyclic xanthines) and nucleoside derivatives have been identified as potent and selective A₃AR antagonists. Probably due to the “enigmatic” physiological role of A₃AR, whose activation may produce opposite effects (for example, concerning tissue protection in inflammatory and cancer cells) and may produce effects that are species dependent, only a few molecules have reached preclinical investigation. Indeed, the most advanced A₃AR antagonists remain in preclinical testing. Among the antagonists described above, compound OT-7999 is expected to enter clinical trials for the treatment of glaucoma, while several thiazole derivatives are in development as antiallergic, antiasthmatic and/or antiinflammatory drugs.

1 Introduction

The four subtypes of adenosine receptors (ARs), designated A₁, A_2A, A_2B, and A₃, are all seven-transmembrane spanning (7TM) receptors that couple to G proteins. The A₃AR inhibits adenylate cyclase through coupling to G_i. A₃AR activation may lead to an activation of the phospholipase C pathway through the β, γ subunit. The A₃AR is found at a high receptor density in the lungs, liver, and in immune cells such as neutrophils and macrophages, as well as at lower densities in the heart and brain (Fredholm et al. 2001) The A₃AR is expressed in neurons in the brain (Lopes et al. 2003; Yaar et al. 2002).

ARs in general, and the A₃AR in particular, are involved in many of the body’s cytoprotective functions. Recently, agents that act at the A₃AR have been targeted for pharmaceutical development based on their anti-inflammatory, anticancer, and cardioprotective effects. For example, activation of the cardiac A₃AR preconditions cardiac myocytes against ischemic damage (Strickler et al. 1996; Tracey et al. 2003) and protects against apoptosis. Selective A₃AR agonists have been shown to protect cardiac muscle in various ischemic models and are protective against the cardiotoxic effects of the anticancer drug doxorubicin (Shneyvais et al. 2001). A₃AR antagonists are of interest as potential antiglaucoma agents (Yang et al. 2005) and as anticancer agents (Gessi et al. 2008).

Agonist ligands for the ARs, including the A₃AR, are almost exclusively nucleoside derivatives. The search for antagonists of the A₃AR in the early 1990s initially encountered an unanticipated difficulty: the lack of an obvious lead structure. Previously, efforts to develop antagonist ligands for the A₁ and A_2A ARs focused on xanthine derivatives. However, at the A₃AR, the prototypical AR antagonists (i.e., the xanthines) are typically much weaker in binding than at the other AR subtypes. This observation stimulated the screening of structurally diverse heterocyclic molecules as potential antagonists (Moro et al. 2006). Chemically diverse leads were discovered in this process that were subsequently optimized to achieve high antagonist selectivity for the A₃AR.

While the A₃AR may be activated by orthosteric agonists that are competitive with adenosine, the action of nucleosides at this receptor may also be enhanced by allosteric modulators. Several heterocyclic classes of positive allosteric modulators of the A₃AR, including 1H-imidazo-[4,5-c]quinolines such as LUF6000 (N-(3,4-dichloro-phenyl)-2-cyclohexyl-1H-imidazo[4,5-c]quinolin-4-amine) and pyridinylisoquinolines, have been reported (Gao et al. 2005; Göblyös et al. 2006).

The structure–activity relationships (SARs) of nucleoside derivatives in binding to the A₃AR and other ARs have been extensively studied, leading to the development of both selective agonists and, more recently, antagonists. Most of the useful modifications of adenosine 1 (Fig. 1) to achieve high A₃AR affinity and selectivity have been made at the N⁶ or C2 positions of adenine or on the ribose group of adenosine. The systematic probing of SAR of both adenosine derivatives and nonpurine antagonists is frequently guided by molecular modeling (Kim et al. 2003), in which the receptor protein is modeled based on structural homology to the light receptor, rhodopsin. The effects of substitution at various sites (i.e., on the nucleobase and ribose moiety) on both the affinity and relative efficacy of nucleoside derivatives at the A₃AR have been extensively probed (Gao et al. 2003, 2004). The approach initially taken to identify agonists for the newly cloned A₃AR was to screen known AR ligands in binding assays. The agonist NECA 2 (adenosine 5′ -N-ethyluronamide) was found to be highly potent but nonselective for this receptor (Zhou et al. 1992). The structural features that promoted A₃AR potency were combined, leading to the first selective A₃ agonist, IB–MECA (N⁶-(3-iodobenzyl)-5′ -N-methylcarboxamidoadenosine), developed in 1993 at the National Institutes of Health (Jacobson et al. 1993). This potent A₃AR agonist IB–MECA 3 and its more selective 2-chloro analog, Cl–IB–MECA 4 (2-chloro-N⁶-(3-iodobenzyl)-5′ -N-methylcarboxamidoadenosine), are used widely as pharmacological tools. A related derivative 5 is widely used as an iodinated radioligand for the A₃AR. IB–MECA and Cl–IB–MECA have entered clinical trials for the treatment of rheumatoid arthritis and cancer (Baharav et al. 2005; Ohana et al. 2001).

Fig. 1.

Structures of adenosine and widely used agonist probes of the A₃AR

One problem encountered in refining selective A₃AR ligands into pharmaceutically useful agents has been the species dependence of binding. This difference in affinity reflects the difference in sequence between the rodent and the human receptors, with only a 74% sequence identity between the rat (r) and human (h) A₃ARs (Fredholm et al. 2001). The species-dependence of A₃AR affinity is particularly pronounced for agonists that contain small alkyl N⁶ substituents and for various heterocyclic antagonists, both of which are more potent in binding to the human than to the rat A₃AR (Yang et al. 2005). The first report of a cloned receptor sequence to be later identified as an A₃AR was that of the rat (Meyerhof et al. 1991; Zhou et al. 1992), and this species was initially used for screening purposes. Nevertheless, many of the nucleoside analogs that were shown to be rat A₃AR agonists, including Cl–IB–MECA and IB–MECA, were later found to be moderately selective for the human A₃AR after it was cloned (Jacobson and Gao, 2006; Salvatore et al. 1993).

The ligand recognition within the putative binding site of the ARs has also been probed through extensive mutagenesis to confirm the predictions concerning ligand recognition made using molecular modeling (Kim et al. 2003). The hydrophobic environment surrounding the purine ring of AR agonists, as found in the putative A_2AAR model, is defined mainly by residues of TM5 and TM6 (Kim et al. 2003). This region is very similar to the putative binding region of hydrophobic heterocyclic (e.g., triazolopyrimidine) antagonists. An exocyclic amino group is common to both adenosine agonists and to typical heterocyclic antagonists, and this amine is generally required to donate a hydrogen bond to the receptor protein. Amino acid residues involved in the ligand recognition in the putative A_2A and A₃ AR binding sites have been reviewed (Kim et al. 2003).

2 A₃AR Agonists

The subtype selectivity of adenosine derivatives as AR agonists has been probed extensively, principally through modification of the N⁶-amine moiety (where large hydrophobic groups tend to produce A₁AR and A₃AR selectivity, Table 1) and the C2 position (where large hydrophobic groups tend to produce A_2AAR selectivity, but have also been shown to enhance A₃AR selectivity, Table 2). The ribose moiety is less amenable than the adenine moiety to the addition of steric bulk, although substitution of the 5′ -CH₂OH moiety with certain amides, ethers, or other hydrophilic groups has resulted in enhancement of A₃AR selectivity.

Table 1.

Binding affinities of monosubstituted adenosine derivatives (N⁶-substituted) at the human A₃AR expressed in CHO cells and at A₁ and A_2A ARs, and maximal A₃AR agonist effect

Compound	Substitution R1 =	pKi at A1ARa	pKi at A2AARa	pKi at A3ARa	%Activation, A3ARa
6	CH3	7.22b	<5b	8.03	96
7	CH2CH3	8.31b	5.05b	8.34	102
8	OCH3	6.65b	<5b	7.55	107
9		9.15b	5.76b	8.19	100
10		9.35b	6.34b	7.14	97
11		9.05b	6.28b	7.14	76
12		8.48b	6.18b	7.83	102
13		6.76b	6.55b	7.38	55
14		7.35	<5b	8.36	80
15		7.89, 7.89b	7.17, 7.17b	8.68, 6.62b	84
16		6.91	5.60	9.06	101
17		6.98	5.60	8.72	101
18		7.82b	5.52b	9.20	100
19		7.93b	5.60b	7.62	87
20		8.15	6.24	7.04	100
21		7.30	6.29	8.41	0
22		6.31	<5	5.48	87
23		7.85	6.84	9.04	99

^aA₃AR binding experiments were performed with membranes prepared from adherent CHO cells stably transfected with cDNA encoding the human A₃AR, using as radioligand [¹²⁵I]N⁶-(4-amino-3-iodobenzyl)adenosine-5′-N-methyluronamide ([¹²⁵I]I–AB–MECA; 2000 Ci/mmol) at a final concentration of 0.5 nM, in Tris·HCl buffer (50 mM, pH 8.0) containing 10mM MgCl₂, 1mM EDTA. Nonspecific binding was determined using 10 μM Cl–IB–MECA. The mixtures were incubated at 25°C for 60 min. Maximal A₃AR agonist effect is the inhibition of forskolin-stimulated adenylate cyclase at 10 μM using a reference value for Cl–IB–MECA of 100%. (Gao et al. 2003; Tchilibon et al. 2004).

^bIn rat brain (Gao et al. 2003; Tchilibon et al. 2004).

Table 2.

Binding affinities of monosubstituted adenosine derivatives (2-ether-substituted) at the human A₃AR expressed in CHO cells and at A₁ and A_2A ARs, and maximal A₃AR agonist effect

Compound	Substitution R1 =	pKi at A1ARa	pKi at A2AARa	pKi at A3ARa	%Activation, A3ARa
29	Cl	8.12	6.20	7.06	100
30		5.29	7.36	6.44	32
31		6.19	6.23	6.93	17
32		6.66	8.03	7.27	71
33		5.43	6.23	5.71	ND
34		6.28	7.21	6.51	72
35		6.66	7.75	6.85	1
36		6.54	7.19	6.98	91
37		5.32	7.57	6.76	0
38		<5	6.51	7.27	0

^aA₃AR binding experiments were performed with membranes prepared from adherent CHO cells stably transfected with cDNA encoding the human A₃AR, using as radioligand [¹²⁵I]N⁶-(4-amino-3-iodobenzyl)adenosine-5′-N-methyluronamide ([¹²⁵I]I–AB–MECA; 2000 Ci/mmol) at a final concentration of 0.5 nM, in Tris·HCl buffer (50 mM, pH 8.0) containing 10mM MgCl₂, 1mM EDTA. Nonspecific binding was determined using 10 μM Cl–IB–MECA. The mixtures were incubated at 25°C for 60 min. ND, not determined. Maximal A₃AR agonist effect is inhibition of forskolin-stimulated adenylate cyclase at 10 μMusing a reference value for Cl–IB–MECA of 100% (Gao et al. 2004)

The binding of a nucleoside to the A₃AR and its activation of the receptor are separate processes that appear to have distinct structural requirements. There is no general correlation between the affinity of a given nucleoside derivative in binding to the A₃AR and its ability to fully vs. partially activate the receptor (Table 1). Specific functionality on the nucleoside structure that lowers efficacy relative to that of a full agonist (e.g., NECA) has been identified. For example, N⁶-benzyl and certain 2-position substituents on the adenine moiety reduce the relative efficacy at the A₃AR. 2-Chloro alone does not reduce A₃AR efficacy, but, in combination with a substituted N⁶-benzyl moiety, it leads to a further reduction (Gao et al. 2002). Other N⁶ substitutions have been studied using the same criteria. For example, the relative efficacy of N⁶-(2-phenylethyl) derivatives is extremely sensitive to substitution of the phenyl ring and the β-methylene carbon (Tchilibon et al. 2004).

Modifications of the ribose moiety have also been explored for effects on both A₃AR binding affinity and efficacy (Gao et al. 2004; van Tilburg et al. 2002). SAR studies also indicate that flexibility in the ribose 5′ region is a prerequisite for A₃AR activation, in concert with a proposed required rotation of TM6 (Kim et al. 2006). Thus, with proper manipulation of groups at the N⁶ and/or ribose moieties, a high-affinity agonist may be converted into a selective A₃AR antagonist. Conversely, agonist function may be maintained fully with proper derivatization of the ribose moiety. A flexible 5′ -uronamide moiety is particularly well suited to maintaining efficacy, and even overcomes the reduction of efficacy induced by various adenine substituents at the N⁶ and C2 positions.

2.1 Substitution of the Adenine Moiety of Adenine Nucleosides

2.1.1 N⁶ Position

Multiple studies have been undertaken to optimize the N⁶ position of adenosine in order to design selective A₃AR agonists (Table 1, 6–23). Addition of small groups such as methyl (6) and oxymethyl (8) to the N⁶ amine gave at least a tenfold increase in potency over adenosine and increased the selectivity of the ligand for the human A₃AR over other human ARs (Volpini et al. 2007). However, increasing the alkyl chain length to an ethyl group (7) increased the affinity of the ligand for both the A₃ and A₁ ARs, thus, decreasing the selectivity (Gao et al. 2003). Larger alkyl chains were not well tolerated at the N⁶ position, and increased branching of the chain caused a decrease in A₃AR affinity and efficacy. Various cycloalkyl groups were also appended to the N⁶-amino group. Adding an N⁶-cyclobutyl (9) or cyclopentyl (10) ring resulted in agonists that had greater affinity to the A₁AR than the A₃AR (Gao et al. 2003) but were full agonists at the A₃AR. Analogs bearing larger N⁶-cycloalkyl rings such as 11 were only partially efficacious as A₃AR agonists. When a benzyl ring was attached to the N⁶ amine (13), the compound was three- to four-fold selective in binding to the human A₃AR in comparison to the A₁ and A_2A ARs, but only displayed a 55% relative efficacy at the A₃AR. N⁶-Phenyladenosine (12) was fully efficacious as an A₃AR agonist. N⁶-(2-Phenylethyl)adenosine (15) was the most potent in binding to the A₃AR among a series of arylalkyl-substituted homologs. However, the N⁶-benzyl and the N⁶-phenyl substituents provided greater selectivity than 2-phenylethyl for the A₃AR. Generally, halogen substitution at the 3 position of the N⁶-benzyl ring caused an increase in A₃AR affinity and selectivity. For example, N⁶-(3-chlorobenzyl)adenosine (14) showed a tenfold selectivity for the A₃AR and a nanomolar affinity. Halogen substitution at other positions of the ring frequently decreased the A₃AR affinity.

Addition of certain larger N⁶ substituents also increased the potency and affinity of the ligands at the A₃AR. For instance, N⁶-(trans-2-phenylcyclopropropyl)adeno sine (16) was a full agonist with high selectivity and a subnanomolar potency (Gao et al. 2003). Further variations on this substituent were prepared, and the importance of conformational factors in the relative efficacy was demonstrated. The addition of one bond to bridge the phenyl rings could change an antagonist into an agonist. Thus, while N⁶-(2,2-diphenylethyl)adenosine (21) was an antagonist at the A₃AR, adding a bond between the phenyl groups to create N⁶-(9-fluorenylmethyl)adenosine (23) restored the efficacy. This compound also had a subnanomolar A₃AR affinity but was less selective than N⁶-(trans-2-phenylcyclopropyl)adenosine (Tchilibon et al. 2004). The most selective compound of the series was N⁶-(trans-2-(3-trifluoromethyl)phenyl)-1-cyclopropyl adenosine (17), which had a 100-fold selectivity at the A₃AR in comparison to the A₁AR.

Various labs have combined sterically bulky N⁶ groups with a 5′ -uronamide moiety on the ribose group to make potent, selective A₃AR agonists. The first A₃AR-selective compounds combined a 5′ -N-alkyluronamide with an N⁶-benzyl group (Gallo-Rodriguez et al. 1994; Jacobson et al. 1993; van Galen et al. 1994). One of the most common A₃ agonists, Cl–IB–MECA (Fig. 1), has an N⁶-iodobenzyl group, a 2-chloro group, and a 5′ -methylcarboxamido group. This compound has a K_i of 0.33 nM at the rat A₃AR, but K_i values of only 2,500 and 1,400 nM at rat A₁ and A_2A ARs, respectively (Kim et al. 1994a). At the human ARs, the binding affinities of Cl–IB–MECA are (nM): A₁AR 220, A_2AAR 5400, and A₃AR 1.4. Thus, Cl–IB–MECA is more selective for the rat A₃AR than the human A₃AR (Melman et al. 2008a). The ¹²⁵I form of I–AB–MECA (N⁶-(4-amino-3-iodobenzyl)-5′ -N-methylcarboxamidoadenosine, 5, Fig. 1) is commonly used as a high-affinity radioligand for characterizing binding to the A₃AR of various species.

Baraldi et al. (1998) prepared a series of N⁶-substituted-aminosulfonylphenyl derivatives of NECA (e.g., compound 24). Among these compounds, the most favorable substituents of the sulfonamido group for increasing affinity at the A₃AR were small alkyl groups, such as ethyl or allyl moieties, and disubstitution of thesulfonamido group. The A₃AR selectivity was increased by the addition of a saturated heterocyclic ring, such as piperidine or morpholine, to the sulfonamido moiety.

Finally, A₃ selective fluorescent probes have also been made by attaching 7-nitrobenzofurazan fluorophores to NECA derivatives using an alkyl spacer (e.g., compound 25). These compounds displayed 500-fold selectivity at the A₃AR and bound in the low nanomolar range (Cordeaux et al. 2008).

2.1.2 Adenine 2 Position

Many modifications at the 2 position of adenosine (Table 2, 29–38) tend to increase A_2AAR potency, but some additions have been found to contribute to A₃AR selectivity. Adding a simple 2-chloro group (29) increased the A₃AR affinity in comparison to adenosine, but it also significantly increased the potency at the A₁AR (Gao et al. 2004). Generally, 2-ether modifications decreased A₃AR affinity, with certain exceptions. For example, adding a 2-i-pentyloxy moiety increased A₃AR affinity threefold, and the compound was slightly selective. 2-Benzyloxy substitution (31) decreased the efficacy to 17% of the full agonist Cl–IB–MECA. 2-Phenylethyloxy substitution (32) often increased affinity at both the A₃ and A_2A ARs, but many such analogs displayed a decreased efficacy as A₃AR agonists. Other 2-ethers, such as 2-(2,2-diphenylethyloxy)adenosine (38), were A₃ AR antagonists, in curious parallel to the effect of the same group when placed at the N⁶ position (Tchilibon et al. 2004).

Many other substitutions at the 2 position of adenosine were combined with previously introduced substitutions at the N⁶ position of adenosine. For instance, adding a 2-cyano group to N⁶-(3-iodobenzyl)adenosine created an A₃AR antagonist, but when the 2-cyano group was added to N⁶-methyladenosine, the compound was a full agonist that was 30-fold selective for the human A₃AR in comparison to the A₁AR (Ohno et al. 2004). However, when other small modifications were made at the 2 position of N⁶-methyladenosine, such as an amino or a trifluoromethyl group, there was a decrease in selectivity and affinity toward the A₃AR. Elzein et al. (2004) synthesized a series of 2-pyrazolyl-N⁶-substituted adenosine derivatives that were very potent and selective for the A₃AR. Cosyn et al. (2006b) found that several 2-triazol-1-yl substitutions of N⁶-methyladenosine increased affinity at the A₃AR. However, in order to maintain efficacy, a 5′ -ethyluronamide was necessary. 9-(5-Ethylcarbamoyl-β-D-ribofuranosyl)-N⁶-methyl-2-(4-pyridin-2-yl-1,2,3-triazol-1-yl)adenine 26 (LC257, Fig. 2) was a full agonist with a K_i of 1.8 nM at the A₃AR and a minimum of 900-fold selectivity over other ARs.

Fig. 2.

Structures of a novel, multiply substituted A₃AR agonists

Additions at the 2 position of NECA often increased potency and/or selectivity. For instance, 2-(3-hydroxy-3-phenyl)propyn-1-yl-NECA 27 (PHPNECA) (Fig. 2) exhibited a subnanomolar affinity at the A₃ receptor (Volpini et al. 2002). Also, Zhu et al. (2006) made a series of N⁶-ethyl-2-alkynyl-NECA derivatives which had sub- to low nanomolar affinities and were very selective in comparison to the A_2A and A_2B ARs, with some selectivity over the A₁AR. The most potent compound in that series (28) had a (p-(methoxy)phenyl)alkynyl substituent at the 2 position.

2.2 Ribose Modifications

2.2.1 Modification of Ribose Hydroxyl Groups

Many modifications have been made to the ribose ring. As mentioned above, the 5′ -N-alkyluronamide modification has been particularly fruitful. Gallo-Rodriguez et al. (1994) initially found that adding a 5′ -N-methyluronamide group to N⁶-benzyl derivatives increased the binding affinity at all three ARs examined and resulted in several of the compounds gaining selectivity for the A₃AR. They also found that adding a 5′ -N-ethyluronamide more than doubled the potency of several N⁶-benzyl derivatives of adenosine. Other modifications at the ribose 5′ position, such as alkylthioethers (van Tilburg et al. 2002) have been found to modulate affinity and efficacy at the A₃AR.

Both the 2′ - and the 3′ - hydroxyl groups contribute to the binding process, since replacing either of these groups in Cl–IB–MECA with a fluoro group caused a significant drop in both affinity and efficacy (Gao et al. 2004). A less drastic decrease in binding and efficacy was seen when the 3′ -hydroxyl of the adenosine analogs was replaced with an amino group (DeNinno et al. 2003). When a methylene spacer was added between the 3′ -amino and ribose groups, there was a total loss of affinity (Van Rompaey et al. 2005). Also, 3′ -deoxy-3′ -acetylamino analogs were weak at the A₃AR. However, DeNinno et al. (2003, 2006) found that the 3′ -amino substitution was tolerated and gave high selectivity when the 5′ and N⁶ positions of adenosine were also appropriately modified in compounds 39 and 40 (Fig. 3). Replacement of the 3′ -hydroxyl with an azido group generally abolished A₃AR activation. The 2′ -hydroxyl group appeared to be more important than the 3′ -hydroxyl group, because when it was replaced with the fluoro group there was no binding or activation of the A₃AR (Gao et al. 2004).

Fig. 3.

Structures of ribose ring-modified selective A₃AR agonist probes

2.2.2 Modification of the Pentose Ring

4′ -Thio derivatives were usually equipotent or slightly more potent at ARs than their oxygen equivalents (Jeong et al. 2006a). Many 4′ -thio derivatives of adenosine have been found to be full agonists. For example, LJ-529 41 (2-chloro-N⁶-(3-iodobenzyl)-4′ -thioadenosine-5′ -methyluronamide) (Fig. 3) is a highly potent ligand (K_i = 0.38 nM against ¹²⁵I]–AB–MECA binding to the human A₃AR expressed in CHO cells). In the same 4′-thio-modified series, a wide variety of ribose 5′ -alkyluronamides have shown that there is tolerance for groups larger than N-ethyl (Jeong et al. 2006a). For example, compounds 42 and 43 were full agonists with K_i values of 3.6 and 18 nM at the hA₃AR, respectively. The nature of the N-alkyl or N-arylalkyl group can modulate affinity and efficacy at the A₃AR (Jeong et al. 2008).

However, when the thio modification was combined with shifting the adenine moiety of Cl–IB–MECA from the 1′ to the 4′ position of the ribose ring, the compound was curiously transformed into a potent antagonist (Gao et al. 2004).

Ring-constrained nucleosides have been used to define conformational preferences at the A₃AR. Medicinal chemists frequently utilize the approach of conformationally constraining otherwise flexible molecules to probe the “active” conformation(s) and to increase ligand affinity by overcoming the energy barriers needed to attain this preferred conformation. Nucleoside analogs containing novel rigid ring systems in place of the ribose ring have been explored as ligands for the ARs. The focus on conformational factors of the ribose or ribose-like moiety allows the introduction of general modifications that lead to enhanced potency and selectivity at certain subtypes of these receptors. One ring system selected for this purpose is the methanocarba (bicyclo[3.1.0]hexane) ring system, which has been incorporated in either of two isomeric forms that adopt either a North (N) or South (S) envelope conformation (Jacobson et al. 2000; Marquez et al. 1996). These ribose modifications were combined with known enhancing modifications at other positions on the molecule to explore the resulting SARs. (N)-Methanocarba-adenosine was favored in binding at the A₃AR by 150-fold over the (S) conformation and by 2.5-fold over adenosine. Doubly modified nucleoside derivatives containing the (N)-methanocarba ring system have confirmed that this conformation of the pseudoribose ring is highly preferred over the (S) conformation for agonists at the A₃AR in general.

Introducing an (N)-methanocarba modification to adenosine 5′ -ethyluronamide increased the human A₃AR binding affinity by sixfold. This modification also demonstrated that the ring oxygen is not required for binding or activation of the receptor (Lee et al. 2001).

Highly selective ring-constrained agonists of the A₃AR have been designed and synthesized based on the (N)-methanocarba ring system (Fig. 3). This led to the introduction of MRS3558 44 ((1′ R,2′ R,3′ S,4′ R,5′ S)-4-{2-chloro-6-[(3-iodophenylmethyl)amino]purin-9-yl}-1-(methylaminocarbonyl)bicyclo-[3.1.0]-hexane-2,3-diol) as a full agonist with subnanomolar potency at the A₃AR and its congeners (e.g., 45 and 46) as full agonists with nanomolar potency at the A₃AR (Tchilibon et al. 2005). The SAR of MRS3558 and related congeners as A₃AR agonists (Melman et al. 2008a) was recently explored in detail. The utility of MRS3558 in treating lung injury was shown in a model of ischemia reperfusion lung injury (Matot et al. 2006). In this series of (N)-methanocarba nucleosides, a 5′ -uronamide moiety is needed in order to achieve full efficacy at the A₃AR. The corresponding 5′ -alcohol is an antagonist of the A₃AR. The 5′ -uronamide moiety overcomes the loss of efficacy associated with substitution of the N⁶ and ribose ring moieties. Thus, in the (N)-methanocarba series, as in the ribose series, a freely rotating 5′ -uronamide that is able to make and break multiple hydrogen bonds provides a necessary degree of flexibility during the receptor activation step.

2.3 Nonadenine Nucleosides and Nonnucleosides as A₃AR Agonists

Occasionally, nonadenine nucleotides are also found to activate the A₃AR. For instance, xanthines such as caffeine are generally found to act as antagonists, but N-methyl-1,3-dibutylxanthine 7-β-D-ribofuronamide 48 (DBXRM) acted as a moderately selective A₃AR agonist (Kim et al. 1994b).

A series of atypical, nonnucleoside agonist ligands that activated various ARs were reported (Chang et al. 2005). In addition to compounds in this family of pyridine-3,5-dicarbonitriles that were selective agonists of the A₁AR, various members of this series substantially activated the A₃AR.

2.4 Further Optimization of A₃AR Agonists Using Multiple Modifications

Interestingly, certain modifications (such as a 5′ -alkylamide or an N⁶-methyl group) can restore efficacy to previously modified compounds. For instance, adding a 2-chloro group to N⁶-cyclopentyladenosine creates an A₃AR antagonist (Gao et al. 2002), but activation is restored by the 5′ -methylcarboxamide and 4-thio substitutions. This is particularly interesting since 4′ -thioadenosine is also an A₃AR antagonist, and 2-chloro-4-thioadenosine is only a partial agonist (Jeong et al. 2006b).

A series of (N)-methanocarba-2,N⁶-disubstituted adenine nucleosides were made by Tchilibon et al. (2004), who found that adding the (N)-methanocarba, 2-chloro, and 5′ -methyluronamido groups significantly improved the selectivity and efficacy of several compounds. For instance, N⁶-(2,2-diphenylethyl)adenosine was an A₃AR antagonist with 12-fold and 130-fold selectivity over A₁ and A_2A ARs, respectively. However, by adding the above substitutions, the compound became a full agonist with a K_i of 0.69 nM and a selectivity of close to 2,000-fold over A₁ and A_2A AR (Tchilibon et al. 2004). Also, the 2-cyano derivative of N⁶-methyl adenosine was a full agonist whereas the 2-cyano derivative of N⁶-(2-phenylcyclopropyl) adenosine was an A₃AR antagonist (Ohno et al. 2004).

Adding several substitutions may also improve selectivity for the A₃AR. Adding an N⁶-methyl group and 2-chloro group to 4′ -thioadenosine-5′ -methyluronamide created a compound with a K_i of 0.28 nM and a nearly 5,000-fold selectivity for the A₃AR (Jeong et al. 2006a). A series of these compounds was made by varying the N⁶ and 5′ groups. While none of these derivatives could match the potency and selectivity of the original compound, it was found that 4′ -thioadenosine derivatives were often more potent than their oxy counterparts. The most potent compound was 9-(3-amino-3-deoxy-5-methylcarbamoyl-β-D-ribofuranosyl)-2-amino-N⁶-methylpurine. Another highly substituted yet extremely potent N⁶-methyl derivative is 2-chloro-N⁶-methyl-4-thioadenosine-5-methyluronamide, which has a K_i of 0.28 nM (Jeong et al. 2006a). N⁶-Methylation also seems to improve human A₃AR selectivity, as N⁶-methyl-2-(2-phenylethyl)-adenosine is much more selective than 2-(2-phenylethyl)-NECA (Volpini et al. 2002). While large 2-position substitutions are not always tolerated, (2R,3S,4R)-tetra hydro-2-(hydroxymethyl)-5-(6-(methylamino)-2-(4-pyridin-2-yl)-1H-pyrazol-1-yl) -9H-purin-9-yl) furan-3,4-diol had a K_i of 2 nM and was extremely selective (Van Rompaey et al. 2005).

Recently, new potent and A₃-selective N⁶,2-disubstituted adenosine derivatives have been reported. Volpini et al. (2007) made a series of N⁶-methoxy-2-alkyladenosine derivatives, of which N⁶-methoxy-2-p-acetylphenylethylMECA was the most potent and selective. This compound had a K_i of 2.5 nM at the human A₃AR and selectivities of 21,000 and 4,200 against A₁ and A_2A ARs, respectively. Recently, a series of water-soluble A₃AR agonists were synthesized (DeNinno et al. 2006). Of these compounds, (2S,3S,4R,5R)-3-amino-5-{6-[5-chloro-2-(2-oxo-2-piperazin-1-yl-ethoxy)-benzylamino]-purin-9-yl}-4-hydroxy-tetrahydro-furan-2-carboxylic acid methylamide was the most potent/selective derivative, with a K_i of 10 nM. Van Rompaey et al. (2005) found that adding additional substitutions to 3-amino-3-deoxyadenosine increased the potency, but these compounds were only partial agonists. 9-[3-Amino-3-deoxy-5-(methylcarbamoyl)-β-D-ribofuranosyl]-N⁶-(5-chloro2-methoxybenzyl)adenine had a K_i of 27 nM and was extremely selective for the A₃AR, but had an efficacy of only 51%. Cosyn et al. (2006a) made a series of 3′ -amino-3′ -deoxy congeners that were highly selective for the A₃AR.

The selectivity at the mouse A₃AR of analogs containing the (N)-methanocarba ring system was reduced due to an increased tolerance of this ring system at the mouse A₁AR (Melman et al. 2008a). Substitution of the 2-chloro atom with iodo or hydrophobic alkynyl groups tended to increase the A₃AR selectivity (up to 430-fold) in mouse and preserve it in human. Extended and chemically functionalized alkynyl chains attached at the C2 position of the purine moiety preserved A₃AR selectivity more effectively than similar chains attached at the 3 position of the N⁶-benzyl group. For example, the carboxylic acid congener MRS5151 47 (Fig. 3) is a highly potent agonist (K_i 2.38 nM at hA₃AR) and is selective in binding at human (6,260-fold) and mouse (431-fold) A₃ARs in comparison to A₁ARs in the same species.

3 A₃AR Antagonists

Initial attempts at obtaining potent and highly selective A₃AR antagonists focused on wide pharmacological screening of different heterocyclic compounds (Jacobson et al. 1995; Ji et al. 1996; Siddiqi et al. 1995). One of the first nonxanthine heterocyclic derivatives (Fig. 4) found to be selective for the human A₃AR (K_i 0.65 nM) was MRS1220 (N-[9-chloro-2-(2-furanyl)[1,2,4]triazolo[1,5-c]quinazolin-5-yl]benzeneacetamide) 49, which was based on appropriate acylation of the exocyclic amino group of this class of known AR antagonists (Kim et al. 1996). During subsequent evaluations, different classes of nonxanthine nitrogen-containing molecules were identified as potent A₃AR antagonists: flavonoids, 1,4-dihydropyridines and pyridines, triazoloquinazolines, isoquinolines and quinazolines (Baraldi et al. 2003a; Mullër et al. 2003). The 1,4-dihydropyridine (DHP) derivative MRS1191 (1,4-dihydro-2-methyl-6-phenyl-4-(phenylethynyl)-3,5-pyridinedicarboxylic acid, 3-ethyl 5-(phenylmethyl) ester) 51 was structurally optimized for binding to the A₃AR (K_i 31 nM) from library screening that identified various DHP calcium channel blockers as weak A₃AR antagonists (Jacobson et al. 1997). The pyridine derivative MRS1523 (5-propyl-2-ethyl-4-propyl-3-(ethylsulfanylcarbonyl)-6-phenylpyridine-5-carboxylate) 50 was the first heterocyclic A₃AR antagonist to display considerable potency and selectivity for the rat A₃AR (K_i 113 nM), as well as the human (18.9 nM) and mouse A₃AR (Li et al. 1998). In this section, the recent advancements in this field have been summarized, with particular attention paid to the most important reports of the last five years.

Fig. 4.

Structures of heterocyclic derivatives that are widely used as selective human A₃AR antagonists

3.1 Recent Developments in Nonpurine Heterocycles

3.1.1 Thiazole and Thiadiazole

IJzerman and coworkers investigated a series of 3-(2-pyridinyl)-isoquinoline derivatives for their affinity at the A₃AR (Van Muijlwijk-Koezen et al. 2000). The effect of an additional nitrogen atom was valued by synthesizing bioisosteric quinazoline derivatives. The compounds VUF 8504 (4-methoxy-N-(3-(2-pyridinyl)-1-isoquinoli nyl)benzamide, 52) and VUF 5574, (N-(2-methoxyphenyl)-N′ -(2-(3-pyridyl)quina zolin-4-yl)urea, 53) (Fig. 4) display considerable A₃AR affinity and appreciable selectivity versus A₁ and A_2A AR subtypes.

The bicyclic system of isoquinoline and quinazoline has been replaced by several monocyclic rings (Van Muijlwijk-Koezen et al. 2001). Some thiazole and thiadiazole derivatives were shown to be most promising candidates for the identification of new A₃AR ligands.

The derivative N-[3-(4-methoxy-phenyl)-[1,2,4]thiadiazol-5-yl]-acetamide (54, Fig. 5) has been claimed to be the most potent A₃AR antagonist of the series, exhibiting a K_i value of 0.79 nM at hA₃AR and antagonistic properties in a cAMP functional assay (Jung et al. 2004). A series of potent and selective A₃AR antagonists have been obtained via an optimization study of compound 55 that revealed that a 5-(pyridine-4-yl) moiety on the 2-aminothiazole ring was optimal for enhanced receptor potency and selectivity (Press et al. 2004). Of particular note, N-[4-(3,4,5-trimethoxyphenyl)-5-pyridin-4-ylthiazol-2-yl]-acetamide 56 showed subnanomolar affinity at the human A₃AR as a competitive antagonist of [¹²⁵I]I–AB–MECA, binding with 1,000-fold selectivity versus the other ARs.

Fig. 5.

Thiazole and thiadiazole derivatives as human A₃AR antagonists

Binding affinity data on thiazole and thiadiazole derivatives at the hA₃AR have been subjected to QSAR analysis (Bhattacharya et al. 2005). This study disclosed the importance of the molecular electrostatic potential surface (Wang–Ford charges) in relation to atoms C2, C5, C7, X8 and S9 (Fig. 5), the last two playing the most important roles. Furthermore, the A₃AR binding affinity increases with decreasing lipophilicity of the compounds and in the presence of short alkyl chains—methyl (Me) or ethyl (Et)—at the R position.

3.1.2 Pyrazoloquinolines

The binding affinities at bovine A₁ and A_2A ARs and at human cloned A₃ARs of some 2-arylpyrazolo[3,4-c]quinolin-4-ones along with their corresponding 4-amines and 4-substituted-amino derivatives were reported by Colotta et al. (2000). The 4-benzoylamido derivative 57 (Fig. 6) displayed one of the best binding profiles of the series of A₃AR antagonists. The same group recently reported an extension of the SAR study of this class of compounds (Colotta et al. 2007) which high-lighted that bulky and lipophilic acyl–amino groups at the 4 position seemed able to promote hA₃AR potency and selectivity. Selected compounds of these series were tested in an in vitro rat model of cerebral ischemia and prevented the irreversible failure of synaptic activity induced by oxygen and glucose deficiency in the hippocampus, thus confirming that potent and selective A₃AR antagonists may substantially increase the tissue resistance to ischemic damage.

Fig. 6.

Pyrazoloquinoline derivatives as human A₃AR antagonists

The synthesis and the affinity profile at ARs of a series of 2-phenyl-2,5-dihydro-pyrazolo[4,3-c]quinolin-4-ones, conceived as structural isomers of the parent 2-arylpyrazolo[3,4-c]quinoline derivatives, have also been reported (Baraldi et al. 2005a). Some of the synthesized compounds showed A₃AR affinities in the nanomolar range and good selectivities, as evaluated in radioligand binding assays at hARs. In particular, substitution at the 4 position of the 2-phenyl ring with methyl, methoxy, or chlorine and the presence of a 4-oxo functionality gave good activity and selectivity (58).

3.1.3 Triazoloquinoxalines

Triazolo[4,3-a]quinoxalines

Interesting studies performed in the last decade by Colotta and coworkers highlighted that the 1,2,4-triazolo[4,3-a]quinoxalin-1-one moiety is an attractive scaffold for obtaining potent and selective hA₃AR antagonists (Colotta et al. 2004; Lenzi et al. 2006). Intensive efforts in the chemical synthesis of compounds based on the systematic substitution of the 2, 4 and 6 positions of the tricyclic template, along with molecular modeling investigations performed to rationalize the experimental SAR findings, led to the identification of optimal structural requirements for A₃AR affinity and selectivity. In particular, the introduction into the triazoloquinoxaline moiety of a 4-oxo (59) or a 4-N-amido (60, Fig. 7) function affords selective and/or potent A₃AR antagonists, indicating that a C = O group (either extranuclear or nuclear) is necessary for A₃AR affinity. This suggested that the probable engagement of this site of the molecule is a hydrogen bond with the A₃AR binding site. Hindering and lipophilic acyl–amino moieties at the 4 position showed enhanced A₃AR affinity (60). Substitution of the 4 position of the 2-phenyl ring with a methoxy or a nitro group and 6-nitro substitution, as well as the combination of these substituents, afforded nanomolar A₃AR affinity and better A₃AR selectivity. 1-Oxo, 6-nitro, and 4-amino groups have been proposed to be involved in hydrogen bonds that anchor the antagonists to the binding site.

Fig. 7.

Triazoloquinoxaline derivatives as A₃AR antagonists

Triazolo[1,5-a]quinoxalines

Some 2-aryl-8-chloro-1,2,4-triazolo[1,5-a]quinoxaline derivatives have been synthesized and tested in radioligand binding assays at bovine (b) A₁ and bA_2AARs and at hA₁ and hA₃ARs (Catarzi et al. 2005a, b). The SAR of these compounds are in agreement with those of previously reported for 2-aryl-1,2,4-triazolo[4,3-a]quinoxalines and 2-arylpyrazolo[3,4/4,3-c]quinolines, thus suggesting a similar AR-binding mode. These studies provided some interesting compounds; among them, 2-(4-methoxyphenyl)-1,2,4-triazolo[1,5-a]quinoxalin-4-one (61, Fig. 7) is the most potent and selective hA₃AR antagonist of this series.

3.1.4 Pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidines

The first example of an AR antagonist containing the pyrazolo-triazolo-pyrimidine scaffold (Cacciari et al. 2007) was reported by Gatta and coworkers (Gatta et al. 1993).

A wide number of compounds (MRE series) originated from the structure– activity optimization work based on systematic substitution at the C2, C5, C9, N7, and N8 positions (Baraldi et al. 2002a; 2003b; 2006). The N⁷-substituted derivatives were found to bind principally to the hA_2AAR (Baraldi et al. 2002b), while the most potent and selective hA₃AR antagonists in this series were derived from the combination of a small alkyl chain at the N⁸-pyrazole position with a (substituted)phenylcarbamoyl chain at the N5 position (Baraldi et al. 2003a). The compound designated MRE-3008-F20, (5-N-(4-methoxyphenylcarbamoyl)amino-8-propyl-2-(2-furyl)pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine, 62) (Fig. 8), one of several high-affinity antagonists of this series, is a highly potent A₃AR ligand (K_i = 0.29 nM against [¹²⁵I]I–AB–MECA binding to human AR receptors expressed in HEK293 cells) with good selectivity over the other hARs. It showed antagonist activity in a functional assay blocking the effect of IB–MECA on cAMP production in CHO cells with an IC₅₀ value of 4.5 nM. [³H] MRE 3008-F20 shows a K_d value of 0.82 ± 0.08 nM and a B_max value of 297 ± 28 fmol mg⁻¹ protein (Varani et al. 2000).

Fig. 8.

A₃AR antagonists based on a pyrazolo-triazolo-pyrimidine scaffold

An important problem with the pyrazolo-triazolo-pyrimidine series was the low water solubilities typically observed, which could limit their use as pharmacological and diagnostic ligands. The bioisosteric replacement of the phenyl ring of the 5-phenylcarbamoyl moiety with a 4-pyridyl moiety (Maconi et al. 2002) provided high water solubility while enhancing hA₃AR affinity. Compound MRE-3005-F20, (5-N-(4-methoxyphenylcarbamoyl)amino-8-ethyl-2-(2-furyl)pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine, 63) and the corresponding HCl salt, which showed very high affinities and good selectivities at the hA₃ receptor subtype, with K_i values in the picomolar range (40 and 10 pM, respectively), can be considered ideal candidates for pharmacological and clinical investigations of the hA₃AR subtype. Receptor modeling ascribed this increase in affinity, compared to neutral arylcarbamate derivatives, to strong electrostatic interactions between the pyridinium moiety and the side chain carbonyl oxygen atoms of Asn274 and Asn278, both located on TM7. Additional studies suggested that involvement of the residue Tyr254 in a hydrogen bond with the pyridyl ring was responsible for both enhanced receptor affinity and selectivity (Tafi et al. 2006). The replacement of the N⁵-pyridine moiety with several N⁵-heteroaryl rings produced a general loss of affinity and selectivity at the hA₃AR (Pastorin et al. 2006).

In order to rationally design and synthesize hA₃AR antagonists with improved binding and/or absorption, distribution, metabolism, and excretion (ADME) profiles, and as suitable clinical candidates, different molecular modeling investigations have been carried out in the last years. Particular attention has been paid to the pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine family, the most potent class of A₃AR antagonists ever reported (Tafi et al. 2006). A combined target-based (highthroughput molecular docking) and ligand-based (CoMFA) (comparative molecular field analysis) drug design approach has recently been performed by Moro and coworkers (Moro et al. 2005), which defined a novel “Y-shaped” binding motif for pyrazolo-triazolo-pyrimidines and rationally delineated some key ligand–receptor interactions for this class of molecules as follows: (1) steric control around the 3 and 4 positions of the N⁵-phenyl ring justifies the decrease in affinity of 3- or 4-substituted-phenyl derivatives; (2) an important π–π interaction takes place between the 2-furyl ring and two phenylalanine residues of the binding site; (3) a hydrophobic pocket, bordered by two hydrophilic amino acids, surrounds the N8 interaction area; and (4) strong hydrogen bonding is possible between a residue of Asn and the N4 of the triazolo ring.

3.1.5 Various Heterocycles

In the last few years, other classes of heterocyclic compounds have been identified as A₃AR antagonists, but large structural dissimilarities meant that none of these could be classified into particular family groups. The quinoxaline derivative 64 (Fig. 9) deserves to be mentioned here, not only because of its good binding profile as an A₃AR antagonist, but also (especially) due to the novelty of the strategy applied to its design, which was based on a 3D database-searching approach (Novellino et al. 2005). There is increasing evidence of the importance of 2D/3D database searching as a valuable tool to discover novel lead compounds for the A₃AR and for other G-protein-coupled receptors (GPCRs) (Costanzi et al. 2008).

Fig. 9.

A₃AR antagonists based on quinoxaline and triazolobenzotriazinone scaffolds

The structural manipulation of a series of phenyltriazolobenzotriazindiones, previously described as ligands at the central benzodiazepine receptor, led Da Settimo and coworkers to the identification of a series of aminophenyltriazolobenzotriazinones. Among these, compound 65, a result of a systematic lead optimization, stands out for its remarkable potency and selectivity at the A₃AR (K_i values at the A₁, A_2A, A₃ARs of 2,700 > 10,000, 1.6 nM, respectively, and IC₅₀ value from cAMP assay at the A_2B > 1,000 nM) (Da Settimo et al. 2007). Interestingly, the triazolobenzotriazinone nucleus is isomeric with that of the triazoloquinoxalinone series described above (compounds 59–61, Fig. 7).

3.2 Purine Derivatives

3.2.1 Adenines

The first class of A₃AR-selective antagonists with a bicyclic structure strictly correlated to the adenine nucleus was claimed in 2005 by Biagi and coworkers (Biagi et al. 2005). The authors described the synthesis of a series of N⁶-ureidosubstituted-2-phenyl-9-benzyl-8-azaadenines whose adenine-like structure was responsible for the antagonist activity and whose phenylcarbamoyl group ensures selectivity at the A₃AR. The structure–activity relationship studies were performed based on the systematic optimization of substituents at the 2, 6 and 9 positions of the bicyclic scaffold, and led to the desired enhancement of A₁/A₃ selectivity (compound 66, Fig. 10).

Fig. 10.

A₃AR antagonists based on nonnucleoside adenine scaffolds

Basing on the finding that the known differentiation agent “reversine” (2-(4-morpholinoanilino)-N⁶-cyclohexyladenine) exerted a moderate antagonist activity at the hA₃AR (K_i value of 0.66 μM), Jacobson and coworkers developed a series of reversine analogs, focusing their attention on the substitution pattern at the 2 and N⁶ positions of the adenine scaffold (Perreira et al. 2005). One of most interesting compounds in terms of hA₃AR affinity and selectivity, MRS3777, (2-(phenyloxy)-N⁶-cyclohexyladenine, 67), combines the N⁶-cyclohexyl moiety of reversine with a 2-phenyloxy group. A few derivatives tested in binding assays to the rat A₃AR seemed to reflect the species dependence of the affinity typical of most known nonnucleoside A₃AR antagonists, and were shown to be inactive at 10 μM in this species.

3.2.2 Triazolopurines

Okamura et al. (Okamura et al. 2002, 2004a) recently reported the study of a new series of 1,2,4-triazolo[5,1-i]purines. This research group highlighted the structural similarity between the new class of compounds and the triazoloquinazoline derivatives and consequently evaluated the corresponding A₃AR affinities. These investigations led to potent and selective hA₃AR ligands, the most potent of which are reported in Fig. 11 (68,69). In particular, 5-n-butyl-8-(4-n-propoxyphenyl)-3H-[1,2,4]triazolo[5,1-i]purine (69) exhibited the best selectivity profile of this series (affinity ratios vs. other AR subtypes > 19,600). Compound (70), 5-n-butyl-8-(4-trifluoromethylphenyl)-3H-[1,2,4]triazolo-[5,1-i]purine (OT-7999), significantly reduced intraocular pressure in cynomolgus monkeys at 2–4 h following topical application (500 mcg) (Okamura et al. 2004b).

Fig. 11.

A₃AR antagonists based on triazolopurine scaffolds

3.2.3 Tricyclic Xanthines

Natural antagonists for ARs such as caffeine and theophylline show, in general, low affinity for the A₃AR subtype (Baraldi et al. 2003a; van Galen et al. 1994). In a recent study, the approach based on the ring annelation of xanthine derivatives for the development of AR antagonists was considered in depth (Drabczyńska et al. 2003).

Some pyrido[2,1-f ]purine-2,4-dione derivatives, which could be considered tricyclic xanthine derivatives, have been reported to exert subnanomolar affinity at the hA₃AR (Priego et al. 2002). The most potent compound of this recent series is the 1-benzyl-3-propyl-1H, 3H-pyrido[2,1-f ]purine-2,4-dione derivative (71, Fig. 12), which presents a K_i value of 4.0 ± 0.3 nM at hA₃AR. The replacement of the benzyl nucleus at the 1 position with a methyl moiety caused dramatic losses of both affinity and selectivity. The effect of the replacement of the pyridine ring of the pyrido[2,1-f ]purine-2,4-dione core with different five-membered heterocycles was examined. In particular, the synthesis and the SAR profile at the ARs of a series of 1-benzyl-3-propyl-7-aryl/alkyl-1H,6H-pyrrolo[2,1-f ]purine-2,4-dione and 1-benzyl-3-propyl-7-aryl/alkyl-1H,8H-imidazo[2,1-f ]purine-2,4-dione derivatives were recently reported (Baraldi et al. 2005b). Among the examined tricycles, the imidazo[2,1-f ]purine-2,4-dione derivatives were two-to tenfold more potent than the corresponding pyrrolo[2,1-f ]purine-2,4-dione derivatives. The best results were obtained with the introduction of small alkyl chains at the 7 position (1-benzyl-7- methyl-3-propyl-1H,8H-imidazo[2,1-f ]purine-2,4-dione 73, 1-benzyl-7-ethyl-3-propyl-1H,6H-pyrrolo[2,1-f ]purine-2,4-dione 74, Fig. 12). Compound 73 shows a subnanomolar affinity towards the target A₃AR, with noteworthy selectivity with respect to the other AR subtypes (K_i (hA₃) = 0.8 nM, K_i (hA₁/hA₃) = 3, 163, K_i (hA_2A/hA₃) > 6,250, IC₅₀ (hA_2B)/K_i (hA₃) = 2,570).

Fig. 12.

A₃AR antagonists based on tricyclic xanthine scaffolds

The synthesis and biological evaluation of a series of fused xanthine derivatives was investigated by Müller and coworkers (Müller et al. 2002a). In particular, the (R)-4-methyl-8-ethyl-2-phenyl-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]purin-5-one (PSB-11 (76), Fig. 12) exhibited a K_i value of 2.3 nM for the A₃AR and good selectivity vs. all other AR subtypes. The radiolabeled derivative of this compound ([³H]PSB-11) exhibited a K_d value of 4.9 nM and a B_max value of 3,500 fmol mg⁻¹ of protein in human A₃AR binding in transfected CHO cells (Müller et al. 2002b). An important innovation of such a series, in comparison with xanthines, is a significant increase in water solubility due to the introduction of a basic nitrogen atom, which can be protonated in physiological conditions. Compound PSB-10, bearing a 2,3,5-trichlorophenyl moiety at the 2 position, showed inverse agonist activity in binding studies in CHO cells expressing recombinant hA₃ARs (IC₅₀ = 4 nM) (Ozola et al. 2003). The 2-(4-bromophenyl)-derivative named KF-26777 (77) with subnanomolar affinity at the hA₃AR (K_i = 0.2 nM) and high selectivity over A₁, A_2A and A_2B ARs (9,000-, 23,500-, 31,000-fold, respectively) was considered a potential lead molecule for development for the treatment of brain ischemia and inflammatory diseases such as asthma (Saki et al. 2002).

3.3 Nucleoside-Derived A₃AR Antagonists

Based on the observation that the relative efficacy of purine nucleosides depends on structural features (see Sect. 2), new subtype-selective nucleoside antagonists of the A₃AR have been designed. One of the first such antagonists was the rigid spirolactam MRS1292 (78) (Fig. 13, (2R,3R,4S,5S)-2-[N⁶-3-iodobenzyl)adenos-9′ -yl]-7-aza-1-oxa-6-oxospiro[4.4]-nonan-4,5-diol) (Gao et al. 2002), which binds potently and selectively to the rat and human A₃ARs but does not activate these receptors, and thus acts as an antagonist.

Fig. 13.

A₃AR antagonists based on nucleoside scaffolds

Modeling/mutagenesis of ARs has focused on distinct residues related to ligand binding and the relative efficacy of adenosine derivatives, and on a conserved Trp residue (6.48) which is involved in the activation process (termed a “rotamer switch,” Shi et al. 2002). Docking studies of agonists suggest that the activation pathway of the A₃AR involves a characteristic anticlockwise rotation of this residue, as viewed from the exofacial side (Kim et al. 2006). The docking of MRS1292 (78) to the A₃AR model is not accompanied by rotation of this residue, as occurs with nucleoside agonists, consistent with its action as an antagonist (Kim et al. 2006). Moreover, the affinity and selectivity of MRS1292 occurs across species, unlike most other heterocyclic antagonists for the A₃AR reported. This allows its use in nonprimate (e.g., murine) experimental animals used as clinical models. For example, MRS1292 applied directly to the eye in mouse has been shown to be effective in reducing intraocular pressure, which may be predictive of its utility as an antiglaucoma agent (Yang et al. 2005).

The removal of the ability of the 5′ -N-alkyluronamide to donate a hydrogen bond was found to convert agonists into selective antagonists (Gao et al. 2006a). In both the 4′ -oxo and the 4′ -thio series, N-methylation of an N-methylamide (i.e., to form a dimethylamide) resulted in potent and selective A₃AR antagonists. Recently, nucleosides that are truncated at the 4′ position were found to act as A₃AR antagonists. For example, (2R, 3R, 4S)-2-(2-chloro-6-(3-chlorobenzylamino)-9H -purin-9-yl)tetrahydrothiophene-3,4-diol (LJ-1416, 80) and (2R, 3R, 4S)-2-(2-chloro-6-(3-iodobenzylamino)-9H-purin-9-yl)tetrahydrothiophene-3,4-diol (LJ-1251, 81) (Fig. 13) (Jeong et al. 2007) displayed K_i values of 1.66 and 4.16 nM, respectively, at the human A₃AR, with > 600-fold selectivity in comparison to the A₁AR. LJ-1251 was shown to have neuroprotective properties in an ischemia model in the rat hippocampus (Pugliese et al. 2007). Truncation at the 4′ position of A₃AR agonist in the (N)-methanocarba series produces potent and selective A₃AR antagonists (Melman et al. 2008b), such as the 3-bromo derivative 1′ R, 2′ R, 3′ S, 4′ R, 5′ S)-4′ -[2-chloro-6-(3-bromobenzylamino)-purine]-2′, 3′ -O-dihydroxybicyclo-[3.1.0]hexane (MRS5147) (83, Fig. 13) (2,900-fold selective for hA₃ vs. hA₁AR) or its 3-iodo analog, MRS5127 (84) (2,400-fold selective for hA₃ vs. hA₁AR). MRS5127 (84) displayed a K_B (Schild constant) value of 8.9 nM as an antagonist of the human A₃AR in a functional assay.

4 Engineering of the A₃AR to Avoid Side Effects of Conventional Synthetic Agonists

Although selective agonists of several of the ARs have been known for years, their use as pharmaceutical agents has been impeded by undesirable side effects of exogenously administered adenosine derivatives. In spite of the clinically useful protective properties of adenosine agonists observed in experimental animals, such as protection against ischemic damage and suppression of excessive inflammation, none of the selective synthetic agonists have yet been approved for human therapeutic use. The A_2AAR-selective agonist Lexiscan (regadenoson, CV Therapeutics, Palo Alto, CA, USA) (CVT-3146, 1-{9-[(4S, 2R, 3R, 5R)-3,4-dihydroxy-5-(hydroxymethyl)oxolan-2-yl]-6-aminopurin-2-yl pyrazol-4-yl)-N-methylcarboxamide) was recently approved for cardiac imaging} in patients. The only other adenosine agonist currently in clinical use is adenosine itself, for the treatment of supraventricular tachycardia and as an aid in cardiac imaging.

Since ARs are widespread in the body, in order to overcome inherent nonselectivity of activating the native ARs using synthetic agonists, we have introduced the concept of neoceptors, by which the putative ligand binding site of a 7TM receptor is re-engineered for activation by synthetic agonists (neoligands) that are built to have a structural complementarity. This is a molecular modeling approach to receptor engineering by which a mutant receptor (neoceptor) is designed for selective activation by a novel synthetic ligand (neoligand) at concentrations that do not activate the native receptor. An amino acid residue of the receptor and a functional group of the ligand moiety thought to be in close proximity can be modified in a complementary fashion so that the two groups exhibit a novel mode of interaction (e.g., reversing the polarity in a salt bridge or introducing unique hydrogen-bonding sites). If a stabilizing interaction exists between these two groups, an increase in affinity is expected at the mutant receptor relative to the native receptor. This strategy is intended for eventual use in gene therapy and may also be useful in mechanistic elucidation, using neoceptor–neoligand pairs that are pharmacologically orthogonal with respect to the native species. Neoceptors have so far been applied successfully to A_2A and A₃ ARs (Gao et al. 2006b; Jacobson et al. 2001, 2005). Compounds 85–87 (Fig. 14) were found to interact selectively with the H272E mutant hA₃AR. All three compounds activated this neoceptor.

Fig. 14.

Compounds that interact selectively with the H272E mutant hA₃AR neoceptor

5 Conclusions

A₃AR ligands have been modified to optimize their interaction with the A₃AR. Most of these modifications have been made to the N⁶ and 2 positions of adenine as well as the ribose moiety, and using a combination of these substitutions leads to the most efficacious, selective, and potent ligands. A₃AR agonists such as IB–MECA and Cl–IB–MECA are now advancing into Phase II clinical trials for treatments targeting diseases such as cancer, arthritis, and psoriasis.

Also, a wide number of compounds exerting high potency and selectivity in antagonizing the hA₃AR have been discovered. These molecules are generally characterized by a notable structural diversity, taking into account that aromatic nitrogen-containing monocyclic (thiazoles and thiadiazoles), bicyclic (isoquinoline, quinozalines, (aza)adenines), tricyclic systems (pyrazoloquinolines, triazoloquinoxalines, pyrazolotriazolopyrimidines, triazolopurines, tricyclic xanthines) and nucleoside derivatives have been identified as potent and selective A₃AR antagonists. Probably due to the “enigmatic” physiological role of A₃AR, whose activation may produce opposite effects (for example, concerning tissue protection in inflammatory and cancer cells) and may produce effects that are species dependent, only a few molecules have reached preclinical investigation. Indeed, the most advanced A₃AR antagonists remain in preclinical biological testing. Among the antagonists described above, compound OT-7999 is expected to enter clinical trials for the treatment of glaucoma, while several thiazole derivatives are in development as antiallergic, antiasthmatic and/or anti-inflammatory drugs.

Abbreviations

ADME

Absorption, distribution, metabolism, and excretion

AR

Adenosine receptor

b

Bovine

cAMP

Cyclic adenosine monophosphate

CHO cells

Chinese hamster ovary cells

Cl–IB–MECA

2-Chloro-N⁶-(3-iodobenzyl)-5′-N-methylcarboxamido-adenosine

CoMFA

Comparative molecular field analysis

CVT-3146

1-{9-[(4S, 2R, 3R, 5R)-3,4-Dihydroxy-5-(hydroxymethyl) oxolan-2-yl]-6-aminopurin-2-yl}pyrazol-4-yl)-N-methylcarboxamide

DBXRM

7-β-D-Ribofuronamide

DHP

1,4-Dihydropyridine

Et

Ethyl

GPCR

G-protein-coupled receptor

h

Human

HEK293 cells

Human embryonic kidney 293 cells

I–AB–MECA

N⁶-(4-Amino-3-iodobenzyl)-5′-N-methylcabroxamidoa-denosine

IB–MECA

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

KF-26777

2-(4-Bromophenyl)-7,8-dihydro-4-propyl-1H-imidazo[2,1-i]purin-5(4H)-one

LJ-529

2-Chloro-N⁶-(3-iodobenzyl)-4′-thioadenosine-5′-methyluronamide

LJ-1251

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

LJ-1416

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

LUF6000

N-(3,4-Dichloro-phenyl)-2-cyclohexyl-1H-imidazo[4,5-c] quinolin-4-amine

Me

Methyl

MRE-3005-F20

5-N-(4-Methoxyphenylcarbamoyl)amino-8-ethyl-2-(2-furyl) pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine

MRE-3008-F20

5-N-(4-Methoxyphenylcarbamoyl)amino-8-propyl-2-(2-furyl) pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine

MRS1191

1,4-Dihydro-2-methyl-6-phenyl-4-(phenylethynyl)-3, 5-pyridinedicarboxylic acid, 3-ethyl 5-(phenylmethyl) ester

MRS1220

N-[9-Chloro-2-(2-furanyl)[1,2,4]triazolo[1,5-c]quinazolin-5-yl]benzeneacetamide

MRS1292

(2R, 3R, 4S, 5S)-2-[N⁶-3-Iodobenzyl)adenos-9′ -yl]-7-aza-1-oxa-6-oxospiro[4.4]-nonan-4,5-diol

MRS1523

5-Propyl-2-ethyl-4-propyl-3-(ethylsulfanylcarbonyl)-6-phenylpyridine-5-carboxylate

MRS3558

(1′ S, 2′ R, 3′ S, 4′ R, 5′ S)-4-{2-Chloro-6-[(3-iodophenylmethyl) amino]purin-9-yl}-1-(methylaminocarbonyl)bicyclo-[3.1.0]-hexane-2,3-diol

MRS3777

2-(Phenyloxy)-N⁶-cyclohexyladenine

MRS5127

(1′ R, 2′ R, 3′ S, 4′ R, 5′ S)-4′ -[2-chloro-6-(3-iodobenzylamino)-purine]-2′, 3′ -O-dihydroxybicyclo-[3.1.0]hexane

MRS5147

(1′ R, 2′ R, 3′ S, 4′ R, 5′ S)-4′ -[2-chloro-6-(3-bromobenzylamino)-purine]-2′, 3′ -O-dihydroxybicyclo-[3.1.0]hexane

MRS5151

(1′ S, 2′ R, 3′ S, 4′ S, 5′ S)-4′ -[6-(3-chlorobenzylamino)-2-(5-hydroxycarbonyl-1-pentynyl)-9-yl]-2′, 3′-dihydroxybicyclo[3.1.0]hexane-1′ -carboxylic acid N-methylamide

NECA

adenosine 5′ -N-ethyluronamide

OT-7999

5-n-Butyl-8-(4-trifluoromethylphenyl)-3H-[1,2,4]triazolo-[5,1-i]purine

Pr

Propyl

PSB-10

8-Ethyl-1,4,7,8-tetrahydro-4-methyl-2-(2,3,5-trichlorophenyl)-5H-imidazo[2,1-i]purin-5-one

PSB-11

(R)-4-Methyl-8-ethyl-2-phenyl-4,5,7,8-tetrahydro-1H-imidazo[2,1-i]purin-5-one

QSAR

Quantitative structure–activity relationships

r

Rat

SARs

Structure–activity relationships

TM

Transmembrane domain

VUF 5574

N-(2-Methoxyphenyl)-N′ -(2-(3-pyridyl)quinazolin-4-yl)urea

VUF 8504

4-Methoxy-N-(3-(2-pyridinyl)-1-isoquinolinyl)benzamide