Adenosine receptors as therapeutic targets

Abstract

Adenosine receptors are major targets of caffeine, the most commonly consumed drug in the world. There is growing evidence that they could also be promising therapeutic targets in a wide range of conditions, including cerebral and cardiac ischaemic diseases, sleep disorders, immune and inflammatory disorders and cancer. After more than three decades of medicinal chemistry research, a considerable number of selective agonists and antagonists of adenosine receptors have been discovered, and some have been clinically evaluated, although none has yet received regulatory approval. However, recent advances in the understanding of the roles of the various adenosine receptor subtypes, and in the development of selective and potent ligands, as discussed in this review, have brought the goal of therapeutic application of adenosine receptor modulators considerably closer.

Extracellular adenosine acts as a local modulator with a generally cytoprotective function in the body¹. Its effects on tissue protection and repair fall into four categories: increasing the ratio of oxygen supply to demand; protecting against ischaemic damage by cell conditioning; triggering anti-inflammatory responses; and the promotion of angiogenesis².

There are four known subtypes of adenosine receptors (ARs) – referred to as A₁, A_2A, A_2B and A₃ – each of which has a unique pharmacological profile, tissue distribution and effector coupling (FIG. 1). All four subtypes are members of the superfamily of G-protein-coupled receptors (GPCRs), and are most closely related to the receptors for biogenic amines. Among the human ARs, the most similar are the A₁ and A₃ ARs (49% sequence similarity) and the A_2A and A_2B ARs (59% similarity).

Figure 1. Adenosine receptor signalling pathways.

Activation of the A₁ and A₃ adenosine receptors (ARs) inhibits adenylyl cyclase activity through activation of pertussis toxin-sensitive G_i proteins and results in increased activity of phospholipase C (PLC) via G_βγ subunits. Activation of the A_2A and A_2B ARs increases adenylyl cyclase activity through activation of G_s proteins. Activation of the A_2AAR to induce formation of inositol phosphates can occur under certain circumstances, possibly via the pertussis toxin-insensitive Gα15 and Gα16 proteins. A_2BAR-induced activation of PLC is through G_q proteins. All four subtypes of ARs can couple to mitogen-activated protein kinase (MAPK), giving them a role in cell growth, survival, death and differentiation. CREB, cAMP response element binding protein; DAG, diacylglycerol; IP₃, inositol 1,4,5-trisphosphate; PI3K, phosphatidylinositol 3-kinase; PIP₂, phosphatidylinositol-4,5-bisphosphate; PK, protein kinase; PLD, phospholipase D; NF-κB, nuclear factor-κB.

Extracellular adenosine levels are quite variable, depending on the tissue and the degree of stress experienced, and so the basal levels of stimulation of the four subtypes by the endogenous agonist vary enormously. The sources of adenosine are either release through an equilibrative transporter or as a result of cell damage³, or nucleotidase-mediated hydrolysis of extracellular adenine nucleotides⁴, which have their own signalling properties that are mediated by purinergic P2 receptors. Ectonucleotidases, of which the apyrase CD39 and the 5′-nucleotidase CD73 are prominent examples, are present on the extracellular surface of many tissues and are crucially involved in numerous important functions⁴. For example, in the brain, CD73 is known as a marker of astrocytes (but not neurons). These enzymes rapidly and effectively shift signalling by released adenine nucleotides and their products to signalling through ARs. Astrocytederived adenosine, acting on A₁ARs, has a central role in the integration of synaptic activity by astrocytes that leads to widespread coordination of synaptic networks⁵. Adenosine itself is rapidly metabolized by adenosine kinase⁶ and, to a lesser degree, adenosine deaminase to AMP and inosine, respectively, both of which are less active than adenosine at the ARs.

The development of potent and selective synthetic agonists and antagonists of ARs has been the subject of medicinal chemistry research for more than three decades. In addition, allosteric enhancers of agonist action could allow the effects of endogenous adenosine to be selectively magnified in an event-responsive and temporally specific manner, which might have therapeutic advantages compared with agonists⁷. AR action might also be modulated not only by direct-acting ligands, but by inhibition of the metabolism of extracellular adenosine⁶ or its cellular uptake³.

Although the basic science suggests that selective AR modulators have promise for numerous therapeutic applications, including cardiovascular, inflammatory and neurodegenerative diseases, in practice this goal has been elusive. One reason for this is the ubiquity of ARs and the possibility of side effects. In addition, species differences in the affinity of putatively selective ligands complicate preclinical testing in animal models. However, there has been a recent impetus towards novel clinical targets, in part as a result of the discovery of the A₃AR subtype in the early 1990s and of the elucidation of new roles for adenosine. In this review, we first present an overview of AR signalling, and summarize progress in the development of selective AR modulators. We then discuss the roles of the AR subtypes in disease, and preclinical and clinical results with AR modulators in various conditions.

AR signalling pathways and regulation

Classically, AR signalling is thought to occur through inhibition or stimulation of adenylyl cyclase (also known as adenylate cyclase), although it is now apparent that other pathways, such as phospholipase C (PLC), Ca²⁺ and mitogen-activated protein kinases (MAPKs), are also relevant (FIG. 1).

Activation of the A₁AR inhibits adenylyl cyclase activity through activation of pertussis toxin-sensitive G proteins^8,9 and results in increased activity of PLC^10,11. In cardiac muscle and neurons, A₁ARs can activate pertussis toxin-sensitive K⁺ channels, as well as K_ATP channels, and inhibit Q-, P- and N-type Ca²⁺ channels¹. Coupling to K⁺ channels in supraventricular tissue is responsible for the bradycardiac effect of adenosine on heart function¹². In the heart, A₁AR and A_2AAR agonistinduced preconditioning has been suggested to occur via modulation of p44/42 extracellular signal-regulated protein kinase (ERK) signalling¹³.

Activation of the A_2AAR increases adenylyl cyclase activity. G_s seems to be the major G-protein associated with A_2AARs in the peripheral systems but not in the striatum, where A_2AAR density is the highest. It has been shown that striatal A_2AARs mediate their effects predominantly through activation of G_olf¹⁴, which is similar to G_s and also couples to adenylyl cyclase. In rat tail artery, facilitation of noradrenaline release by activation of the A_2AAR triggers the PLC and adenylyl cyclase pathways¹⁵. Activation of the A_2AAR also induces formation of inositol phosphates to raise intracellular calcium and activate protein kinase C in COS-7 cells via pertussis toxin-insensitive Gα15 and Gα16 proteins¹⁶, which have limited tissue distribution and interact with most GPCRs.

The A_2BAR is positively coupled to both adenylyl cyclase and PLC^17–20. Results indicate that the activation of PLC, through G_q proteins, mediates many of the important functions of A_2BARs^21,22. Activation of the A_2BAR by the non-selective agonist NECA increased inositol phosphate formation in human mast cell line HMC-1 (REF. 20), which is not sensitive to cholera or pertussis toxin but is antagonized by the slightly A_2BAR-selective antagonist enprofylline (3-propylxanthine)²⁰. The arachidonic acid pathway was also recently demonstrated to be involved in A_2BAR activation²³.

The A₃AR couples to classical second-messenger pathways such as inhibition of adenylyl cyclase²⁴, stimulation of PLC²⁵ and calcium mobilization^26–29. In cardiac cells, A₃AR agonists induce protection through the activation of K_ATP channels³⁰. RhoA–phospholipase D1 signalling has been demonstrated to mediate the antiischaemic effect of A₃ARs³¹. The WNT signalling pathway is involved in A₃AR agonist-mediated suppression of melanoma cells³². In addition, like other ARs, the A₃AR couples to MAPK, which could give it a role in cell growth, survival, death and differentiation^33,34. An A₃AR agonist inhibits proliferation in A375 human melanoma cells via the phosphatidylinositol 3-kinase–protein kinase B–ERK1/2 pathway³⁵.

Phosphorylation and subsequent desensitization of ARs have been studied for all four subtypes. The rapidity of the desensitization depends on the subtype, with the A₃AR being more rapidly desensitized than the other subtypes³⁶. GPCR kinase-mediated mechanisms are thought to have a crucial role in the rapid desensitization of A_2A and A_2BARs³⁶.

AR agonists and antagonists

The main approach for discovering AR agonists has been modification of adenosine itself, and the structure–activity relationships of adenosine at ARs have been extensively probed³⁷. Most of the useful analogues are modified in the N⁶- or 2-position of the adenine moiety and in the 3′-, 4′- or 5′-position of the ribose moiety. Highly selective agonists of the various receptor subtypes (FIGS 2,3; TABLE 1) have been designed through both empirical approaches and a semi-rational approach based on molecular modelling^38,39.

Figure 2. Adenosine receptor agonists.

Adenosine receptor (AR) agonists acting at the A₁AR. K_i values for binding to ARs are given in TABLE 1.

Figure 3. Adenosine receptor agonists.

a ∣ Adenosine receptor (AR) agonists acting at the A_2AAR. LUF5835 (EC₅₀ of 10 nM) is an atypical A_2BAR agonist⁵². b ∣ AR agonists selective for the A₃AR. K_i values for binding to ARs are given in TABLE 1.

Table 1.

Affinity of selected adenosine receptor agonists and antagonists at the four receptor subtypes

Adenosine receptor subtype	Compound	Ki value for AR (nM)				Refs
		A1AR*	A2AAR*	A2BAR*	A3AR*
Agonists
A1	CPA	2.3	794	18,600¶	72	37,43,207
	CCPA	0.83	2,270	18,800¶	38	37,43,207
	S(–)-ENBA	0.38	>10,000	>10,000¶	915	43
	ADAC	0.85‡	210‡	N.D.	13.3	43
	AMP579	5.1‡	56‡	N.D.	N.D.	37
	NNC-21-0136	10‡	630‡	N.D.	N.D.	135
	GR79236	3.1‡	1,300‡	N.D.	N.D.	135
	CVT-510 (Tecadenoson)	6.5§	2,315§	N.D.	N.D.	37,66
	SDZ WAG 994	23§	25,000§	>10,000§,¶	N.D.	37,67
	Selodenoson	1.1‡	306‡	N.D.	N.D.	37,65
A2A	NECA	14	20	140¶	25	43,207
	CGS21680	289	27	>10,000¶	67	43,207
	DPMA	168	153	>10,000¶	106	43,46
	Binodenoson	48,000	270	430,000¶	903	37
	ATL-146e	77	0.5	N.D.	45	37,45
	CV-3146	>10,000	290	>10,000¶	>10,000	208
A2B	LUF5835	4.4	21	10¶	104	52
A3	IB-MECA	51	2,900	11,000¶	1.8	43,56,207
	Cl-IB-MECA	220	5,360	>100,000¶	1.4	39,43,207
	LJ568	193	223	N.D.	0.38	58
	CP-608039	7,200	N.D.	N.D.	5.8	78
	MRS3558	260	2,300	>10,000¶	0.29	39
	MRS1898	136	784	N.D.	1.5	39
Antagonists
A1	DPCPX	3.9	129	56	3,980	40
	WRC-0571	1.7	105	N.D.	7,940	44
	BG 9719	0.43	1,051	172	3,870	40
	BG 9928	29	4,720	690	42,110	161
	FK453	18	1300	980	>10,000	194
	FR194921	2.9	>10,000	N.D.	>10,000	107
	KW3902	1.3‡	380‡	N.D.	N.D.	40
A2A	KW6002	2,830	36	1,800	>3,000	124
	CSC	28,000‡	54‡	N.D.	N.D.	40
	SCH 58261	725	5.0	1,110	1,200	124
	SCH 442416	1,110	0.048	>10,000	>10,000	152
	ZM241,385	774	1.6	75	743	40,124
	VER 6947	17	1.1	112	1,470	124
	VER 7835	170	1.7	141	1,931	124
	‘Schering compound’	82	0.8	N.D.	N.D.	125
A2B	MRS1754	403	503	2.0	570	53
	MRE 2029-F20	245	>1,000	3.0	>1,000	54
	OSIP-339391	N.D.	N.D.	0.5	N.D.	55
A3	OT-7999	>10,000	>10,000	>10,000	0.61	200
	MRS1292	12,100‡	29,800‡	N.D.	29.3	42
	PSB-11	1640	1,280	2,100¶	3.5	61
	MRS3777	>10,000	>10,000	>10,000¶	47	62
	MRS1334	>100,000	>100,000	N.D.	2.7	56
	MRE 3008-F20	1,200	141	2100	0.82	40
	MRS1220	305‡	52.0‡	N.D.	0.65	56
	MRS1523	15,600‡	2,050‡	N.D.	18.9	40
	‘Novartis compound’	197	1,670	3.0¶	10.0	175

^*Binding experiments at recombinant human A₁, A_2A, A_2B and A₃ adenosine receptors (ARs), unless noted.

^‡Binding experiments at rat ARs.

^§Binding or functional experiments at porcine ARs.

^¶Data are from a cyclic AMP functional assay. N.D., not determined or not disclosed.

Similarly, the main approach for the discovery of AR antagonists (FIGS 4,5) has been modification of xanthines such as caffeine and theophylline⁴⁰. A modelling approach combining quantitative models of receptor and ligand has been demonstrated to accurately predict the potency of antagonists⁴¹. Molecular modelling of ARs using information from the crystal structure of the seventransmembrane protein rhodopsin, supported by mutagenesis studies, has also aided in understanding ligand recognition and provided insights into conformational dynamics^38,42.

Figure 4. Adenosine receptor antagonists.

Antagonists acting at the A₁ adenosine receptors (A₁ARs) and A_2AARs. K_i values for binding to the ARs are given in TABLE 1.

Figure 5. Adenosine receptor antagonists.

Antagonists acting at the A_2B and A₃ ARs⁴⁰. Novartis compound has high affinity at human A_2B and A₃ ARs¹⁷⁵. K_i values for binding to the ARs are given in TABLE 1.

The rest of this section summarizes the development of selective agonists and antagonists for each of the receptor subtypes.

A₁ ARs

Agonist selectivity for A₁ARs is typically accomplished through substitution at the adenosine N⁶-position, giving rise to compounds such as CPA. The 2-chloro analogue CCPA displays slightly greater A₁AR affinity than the parent compound CPA. The affinities of these N⁶-substituted derivatives for A₃ARs are often intermediate between their respective A₁AR and A_2AAR affinities. Agonists CPA and CCPA are highly selective for the rat A₁AR compared with the A₃AR subtype, but less selective in human tissue. S(–)-ENBA is an even more potent and selective agonist for human and rat A₁ARs compared with the three other AR subtypes⁴³

The classical, nonselective xanthine antagonists of ARs, theophylline (1,3-dimethylxanthine) and caffeine (1,3,7-trimethylxanthine), display micromolar affinity at A₁, A_2A and A_2B ARs¹. Potent and selective antagonists for A₁ARs have been derived by modification of the xanthines, including many 8-aryl and 8-cycloalkyl derivatives⁴⁰. One such derivative is DPCPX, which is highly selective for rat A₁AR compared with the A_2AAR (~500-fold) and less selective at the human A₁ARs compared with human A_2A and A_2B ARs. Certain non-xanthine antagonists, such as the inverse agonist WRC-0571, derived from adenine, are A₁AR selective, even compared with the A_2BAR⁴⁴.

A_2A ARs

Substitution with small alkyl amide groups at the 5′-position of adenosine, as in the non-selective agonist NECA, provides increased potency at all the ARs, and this approach was also used to generate CGS21680, which is a moderately A_2AAR-selective agonist in rats (140-fold selectivity for the A_2AAR compared with the A₁AR) but not humans^1,43. The selective agonist ATL-146e has much greater affinity for the A_2AAR than CGS21680⁴⁵. Although most N⁶-substituted adenosine agonists are A₁AR- or A₃AR-selective, the agonist DPMA (diastereomeric pair) is more than 30-fold selective for the rat A_2AAR compared with the rat A₁AR and A₃AR subtypes, but has similar affinity at human A₁, A_2A and A₃ ARs^43,46.

ZM241,385 and SCH 58261 are highly potent and selective A_2AAR antagonists^47,48, although ZM241,385 has been shown to bind with intermediate affinity at the human A_2BAR^46,47. The phenolic group of ZM241,385 can be radio-iodinated to provide an A_2AAR-selective radioligand⁴⁹. KW 6002, CSC and other 8-styrylxanthines are selective for A_2AARs compared with the A₁, A_2B and A₃ARs⁵⁰. However, in dilute solution, these compounds suffer from sensitivity to photoisomerization.

A_2B ARs

The A_2BAR is the least studied subtype of the AR family⁵¹. Selective antagonists have been reported; however, adenosine derivatives as selective A_2BAR agonists remain to be developed. Nearly all the known agonists are derivatives of adenosine. A notable exception is the development, on the basis of recent patents, of a series of pyridine-3,5-dicarbonitrile derivatives by IJzerman and colleagues⁵². These compounds act as AR agonists or partial agonists, with varying degrees of AR selectivity⁵², with LUF5835 being the most potent activator of the A_2BAR (EC₅₀ of 10 nM, FIG. 3a).

Xanthines that have been developed as selective antagonists of the A_2BAR include MRS1754 and MRE 2029-F20, both of which have been prepared as radioligands^53,54. Another selective antagonist, [³H]OSIP339391, has also recently been radiolabelled for the study of A_2BARs⁵⁵

A₃ ARs

The prototypical A₃AR agonist IB-MECA (CF101) and the more selective agonist Cl-IB-MECA have been widely used as pharmacological probes in the elucidation of the physiological role of the most recently identified AR subtype, A₃ (REF. 56). IB-MECA has a ~50-fold selectivity for rat A₃ARs over other subtypes in vitro. The related 4-aminobenzyl derivative can be radio-iodinated, giving rise to [¹²⁵I]-I-AB-MECA, which is widely used as a high-affinity radioligand for A₃ARs⁵⁷

The 4′-thio modification of adenosine derivatives, explored for its effect on AR selectivity, has produced several highly potent and selective A₃AR agonists, such as LJ568 (REF. 58). Conformational studies of the ribose moiety and its equivalents indicate that the ring oxygen is not required and that the North (N) ring conformation is preferred in binding to the A₃AR. One means of locking the conformation of the ribose-like ring is through use of the bicyclo[3.1.0]hexane ring system, which assumes an (N)-envelope conformation^39,42. Highly selective A₃AR agonists were recently reported in the series of (N)-methanocarba-5′-uronamide derivatives, including MRS3558, which has a K_i value of 0.29 nM for the human A₃AR³⁹

The search for A₃AR antagonists began with the discouraging observation that xanthines – such as caffeine and theophylline, the classical antagonists of A₁, A_2A and A_2B ARs – typically have low binding affinities for the A₃ AR²⁴. Initial findings were reported for rat A₃AR (before the report of the cloning of the human homologue) and many common xanthines have binding affinities of around 100 μM for this receptor. The cloning of the A₃AR from other species facilitated further investigation of this receptor⁵⁹. For sheep and human A₃ARs, xanthines have intermediate affinity (typically 100 nM for 8-phenylxanthine analogues). There is a marked species dependence of antagonist affinity at the A₃AR, with human affinity typically greatly exceeding that at the rat receptor for xanthine and other non-purine A₃AR antagonists^24,60. Therefore, the search for A₃AR antagonists turned towards more novel heterocyclic systems⁵⁶.

The screening of diverse chemical libraries resulted in the identification of new high-affinity hits for the human A₃AR, including dihydropyridines, flavonoids, pyridines, thiazoles and others, which were then optimized, typically by substitution of aromatic rings^40,41,56. The dihydropyridine derivative MRS1334 (not active at L-type calcium channels) and the pyridylquinazoline VUF5574 (not selective in rat) are both relatively potent A₃AR antagonists, with K_i values of 2.7 and 4.0 nM, respectively, at the human subtype. The pyridine derivative MRS1523 is a selective A₃AR antagonist in both rat and human. PSB-11, which is a selective antagonist for human A₃ARs, was tritiated for characterization of this receptor⁶¹. Numerous adenine derivatives have been studied as selective antagonists for A₁ or A_2A ARs, and the adenine derivative MRS3777 was recently reported to be highly selective for the human A AR⁶²

An alternative approach to designing A₃AR antagonists is to start with high-affinity adenosine derivatives and truncate the molecule in stages to remove the capacity to activate the receptor without compromising high-affinity binding. An initial attempt to find adenine derivatives, such as 9-alkyl-N⁶-iodobenzyladenines, that displayed these characteristics was unsuccessful⁵⁶. A more successful approach either added substituents to adenosine derivatives or rigidified the nucleosides, to reduce their intrinsic efficacy⁴². With more systematic studies of structure–efficacy relationships on substitution of adenosine at the N⁶, ribose and C2 adenine positions, it became apparent that the efficacy at A₃ARs is more easily diminished by structural modification than it is at the other AR subtypes^42,43. In some cases, N ⁶-substitution of adenosine 5′-OH derivatives with large groups (for example, substituted benzyl groups or large cycloalkyl rings) reduced the maximal efficacy, leading to decreased efficacy at the A₃AR. For example, CCPA and DPMA, which are full agonists at A₁ and A_2A ARs, respectively, are A₃AR antagonists. This structural insight was used advantageously to obtain the conformationally constrained nucleoside MRS1292, which proved to be a selective A₃AR antagonist in both rat and human^42,60.

ARs as targets in cardiovascular disease

Arrhythmia

A₁AR activation has a number of effects in the cardiovascular system, including a reduction in heart rate and atrial contractility, and the attenuation of the stimulatory actions of catecholamines on the heart^63,64. Activation of the A₁AR by intravenous infusion of adenosine (Adenocard; Astellas Pharma) is used to restore normal heart rhythm in patients with paroxysmal supraventricular tachycardia (PSVT). However, for more long-term indications, selective A₁AR modulators are needed to avoid side effects related to other AR subtypes such as hypotension.

Selodenoson (DTI0009, GR 56072, RG 14202; Aderis Pharmaceuticals) is a potent and selective A₁AR agonist with the potential to control heart rate without lowering blood pressure⁶⁵. It has been tested in Phase II clinical trials for its capacity to slow heart rate in atrial fibrillation, although it has been reported to have renal toxicity. Tecadenoson (CVT-510) is a potent A₁AR agonist with a dose-dependent negative dromotropic effect on the AV node⁶⁶. In patient trials (now Phase III), it terminated PSVT without the side effects associated with other AR subtypes, such as hypotensive effects. CVT-2759 is a partial agonist of the A₁AR, which in guinea pig heart seems to be useful in slowing down AV nodal conduction and thereby ventricular rate without causing AV block, bradycardia, atrial arrhythmias or vasodilation⁶³. SDZ WAG 994 was extensively characterized in clinically relevant models⁶⁷ and has been in Phase I clinical trials for the potential treatment of PSVT. However, development of this agent was discontinued in 1999.

Ischaemia

Adenosine is released in large amounts during myocardial ischaemia, resulting in effective pre-conditioning in cardiomyocytes through the activation of A₁ and A₃ ARs^29–31,68. Administration of a synthetic AR agonist to activate either or both of these receptors might therefore be beneficial to the survival of the ischaemic heart.

Ischaemic cardiac preconditioning by the A₁AR involves a series of intracellular events that begin with the activation of the receptor and end at the sensitive K +-ATP channels of the mitochondria. Most of the pharmacological data indicate that A₁ARs have an important role in protection of the heart and brain from ischaemia–reperfusion injury¹. The A₁AR-agonist-induced decreases in blood pressure and heart rate are mediated by the peripheral A₁ARs, with little or no contribution of central A₁ARs, which is consistent with the observation that typical agonists cross the blood–brain barrier to only a small degree⁶⁹. An important role of the A₁AR in protection of the murine heart by remote, delayed adaptation has been demonstrated⁷⁰.

Cardiac overexpression of A₁ARs in mouse heart results in substantial protection from ischaemia–reperfusion injury^71–73. Hearts isolated from transgenic animals with overexpression of A₁ARs have a lower basal rate than those of control mice⁷¹. Cardioprotection in wild-type hearts and hearts overexpressing A₁ARs was mediated by mitochondrial K⁺-ATP channel activation⁷⁴. A₁AR overexpression also improves myocardial tolerance to anoxia reoxygenation, in addition to protecting hearts from ischaemia–reperfusion injury⁷⁵. The cardiovascular effects of the A₁AR agonist CPA also convey protection against Sarin poisoning⁷⁶.

It is well documented that adenosine can protect tissues against hypoxia or ischaemia through A₁ARs. In the A₁AR-knockout mouse heart, baseline contractile function and heart rate were unaltered, but intrinsic myocardial resistance to ischaemia was limited⁷⁷. In addition, non-selective receptor agonism induced by 2-chloroadenosine was cardioprotective in A₁AR-knockout (albeit to a lesser extent) and wild-type hearts, indicating additional protective mechanisms through other AR subtypes.

Various lines of evidence indicate that the A₃AR has a role in protecting the heart^68,78. Overexpression of A₃ARs decreases heart rate, preserves energetics and protects ischaemic hearts⁷⁹, and low-level expression of A₃ARs in the heart provides effective protection against ischaemic injury without detectable adverse effects, although higher levels of A₃AR expression lead to the development of a dilated cardiomyopathy⁸⁰. Paradoxically, global deletion of the A₃AR in mice also confers resistance to myocardial ischaemic injury and does not prevent early preconditioning⁸¹. In an isovolumic Langendorff perfusion model, A₃AR-knockout mice also had improved functional recovery and tissue viability during reperfusion after ischaemia when compared with control mice⁷⁸. In one study, administration of an A₃AR agonist in wild-type and A₃AR-knockout mice showed similar cardioprotective effects: post-ischaemic recovery was enhanced in A₃AR-knockout mice⁸², implying action of the agonist at a non-A₃AR, probably the A_2AAR. It has been demonstrated that, in addition to the role of A₁ and A₃AR agonists, the A_2AAR activation is protective against ischaemia–reperfusion injury in mice, which has been proposed to be mainly due to its actions on lymphocytes⁸³.

An A₃AR-mediated direct cardioprotective effect has been evident³¹. However, the A₃AR expression level is low in cardiomyocytes. So, an indirect protective effect has also been proposed, for example, through modulation of the function of mast cells and neutrophils, in which A₃ARs are abundant. A₃AR signalling in rodent mast cells may be detrimental to the myocardium because of a pro-inflammatory mechanism, but this response might only be observed in rodents (that is, mouse A₃AR-knockout experiments) because mast cell degranulation seems to be A_2BAR-dependent in humans and dogs¹. This could at least in part explain the paradoxical effects observed in various experiments^75,81,82. Another explanation of the paradoxical effect could be the numerous compensatory mechanisms observed in receptor-knockout mice⁸⁴.

Some studies have shown that activation of either A₁ or A₃ ARs could trigger protection of function in preconditioned rat hearts, although maximal preconditioning requires activation of both A₁ and A₃ ARs^85,86. The selective agonists CCPA and Cl-IB-MECA have been used to discern two separate A₁ and A₃AR-mediated pathways leading to cardioprotection, either as early or late preconditioning or during prolonged ischaemia. Unlike the case of A₁AR-mediated cardioprotection, A₃AR-mediated cardioprotection is achieved in vivo in the absence of haemodynamic side effects, such as hypotensive effects, and could therefore be therapeutically more promising⁶⁸. Stimulation of A₃ARs could also be advantageous over A₁AR activation because it may be less likely to induce bradycardia⁸⁵. The A₃AR agonist CP608039 is in development for use in perioperative cardioprotection⁷⁸, and the A₃AR agonist Cl-IB-MECA protects rat cardiac myocytes from the toxicity induced by the cancer chemotherapeutic agent doxorubicin⁸⁷.

Vasodilation

The A_2AAR is involved in vasodilation in the aorta and coronary artery¹. It was suggested that the tachycardic effect of A_2AAR activation is mediated by centrally located receptors, whereas its hypotensive effect is mediated by the peripheral A_2AAR⁶⁹. In the late 1960s and 1970s, metabolically stable AR agonists were tested clinically as antihypertensives, and this was an intended use of the A_2AAR agonist CGS21680; however, its clinical path was curtailed following canine haemodynamic studies due to in vivo non-selectivity related to spare receptors. In platelets, an A_2AAR agonist was shown to inhibit aggregation by increasing intracellular cAMP levels, suggesting that adenosine agonists might have utility as antithrombotic agents⁸⁸.

Recently, there has been an effort to further improve subtype-selectivity of A_2AAR agonists for novel therapeutic applications, including imaging. Adenosine, under the name Adenoscan (Astellas Pharma), is used in myocardial stress imaging to evaluate coronary artery disease by achieving vasodilation in patients unable to exercise adequately. Regadenoson (CVT-3146), a potent and selective A_2AAR agonist, is being evaluated in Phase III studies for the same purpose during myocardial perfusion imaging⁸⁹. The selective A_2AAR agonist binodenoson (WRC-0470) has entered Phase III clinical trials and seems to be well tolerated as a short-lived coronary vasodilator and acts as an adjunct to radiotracers in imaging⁹⁰. ATL-146e, the most selective of these A_2AAR agonists, has also entered Phase III clinical trials for coronary imaging.

Activation of the A_2BAR induces vasodilation in some vascular beds^91,92, such as the main pulmonary artery of guinea pigs, and induces human chorionic vasoconstriction and signals through the arachidonic acid cascade²³. The A_2BAR is selectively upregulated by hypoxia, and A_2BAR antagonists effectively neutralize ATP-elicited reduction in post-hypoxic endothelial permeability⁹³. Inhibition of mitosis of rat aortic smooth-muscle cells has been achieved through selective A_2B AR activation⁹⁴. The A_2BARs are also important for adenosine-mediated inhibition of cardiac fibroblast functions⁹⁵ and the stimulation of nitric oxide production during Na⁺-linked glucose or glutamine absorption⁹⁶. Activation of the A_2BAR promotes angiogenesis by increasing the release of angiogenic factors^2,97.

Cutaneous vasopermeability, which is associated with activation and subsequent degranulation of mast cells, is completely lost in mice lacking functional A₃ARs⁹⁸. One of the well-known actions of adenosine is to dilate vascular beds. Interestingly, the concentration of cAMP is higher in the aortae of A₃AR-deficient mice, with no significant change in the amount of A₁ or A_2A ARs, than it is in control mice. The hypotensive effect observed after intravenous adenosine injection in mice lacking the A₃AR was significantly larger than in control mice⁹⁹. Genetic deletion of the A₃AR or antagonism of the A₃AR augmented coronary flow induced either by adenosine or by the A_2AAR agonist CGS21680 (REF. 100). However, A₃ARs do not regulate atherogenesis; the development of atherosclerosis and response to injury of the femoral artery were similar to those in wild-type mice¹⁰¹.

It has been clearly demonstrated that both agonist- and antagonist-binding profiles for the murine and human A₃ARs are different. The marked species difference, together with the paradoxical protection in A₃AR-knockout hearts despite A₃AR-mediated protection in wild-type hearts, could reflect limitations of gene-knockout studies, and the A₃AR-knockout data from mice should be interpreted with caution. Also, it should be noted that the selective ligands currently available are only relatively selective for a certain AR subtype. At relatively high concentrations, these ligands may also activate or block other AR subtypes. As such, cautious and thoughtful interpretation of pharmacological data is necessary.

ARs as targets in nervous system disorders

Observations of the effects of caffeine — a classical AR antagonist — on the nervous system, such as enhancement of awareness and learning, have encouraged the investigation of selective AR antagonists in the nervous system. Indeed, the A₁AR was recently shown to be involved in the discriminative-stimulus effects of caffeine¹⁰². However, it is the A_2AAR, which was the first AR to be genetically deleted¹⁰³, that is the primary mediator of the behavioural stimulatory effects of caffeine^103–105, and this receptor also has an important role in sleep regulation (see below)¹⁰⁶. Although high concentrations of caffeine can also block phosphodiesterases, which can have numerous behavioural consequences¹⁰⁴, most of the varied stimulant and other behavioural effects of caffeine are thought to result from antagonism of ARs in the nervous system. This suggests that modulation of ARs may provide therapeutic targets in nervous system disorders, in such diverse conditions as dementia and other neurodegenerative diseases, hyperactivity, anxiety, schizophrenia and sleep disorders.

Dementia and anxiety disorders

Recently, the novel A₁AR-selective antagonist FR194921 (which did not show any species differences in its high A₁AR affinity) was reported to have potential in the treatment of dementia and anxiety disorders¹⁰⁷. FR194921 was orally active and centrally available, ameliorated scopolamineinduced memory deficits, and also showed anxiolytic effects in the elevated plus maze test, without influencing general behaviour or having antidepressant activity¹⁰⁷.

Pain

Adenosine exerts multiple influences on pain transmission at peripheral and spinal sites. At peripheral nerve terminals in rodents, A₁AR activation produces antinociception by decreasing cAMP levels in the sensory nerve terminal¹⁰⁸. Mice lacking functional A₁AR show signs of increased anxiety and hyperalgesia, and the analgesic effects of adenosine observed in wild-type mice are lost¹⁰⁹. A recent study using A₁AR-knockout mice suggested that A₁ARs might be more important in chronic pain than in acute pain¹¹⁰. In humans, infusion of adenosine in the spinal cord was effective in decreasing post-operative pain¹¹¹.

The A₁AR agonist GR79236 administered in cats had a dose-dependent inhibitory effect on trigeminovascular nociceptive transmission, which is otherwise associated with the initiation of local increases in blood flow and enhanced protein permeability — that is, factors that contribute to vascular headaches¹¹². A₁AR activation leads to neuronal inhibition without concomitant vasoconstriction, indicating that this might be an effective treatment of migraine and cluster headache. Another A₁AR agonist, GW-493838, was evaluated in Phase II clinical trials for the treatment of pain and migraine, and was found to inhibit electrically induced nociceptionspecific blink reflex responses¹¹³. The A₁AR-selective allosteric enhancer T-62 (FIG. 6) was also shown to reduce hypersensitivity in carageenin-inflamed rats by a central mechanism¹¹⁴. T-62 has entered Phase I clinical trials as a treatment for neuropathic pain.

Figure 6. Examples of allosteric enhancers of the activity of adenosine receptor agonists.

PD81,723 and T-62 enhance the activity of agonists acting at the A₁ adenosine receptor (A₁AR) and VUF5455 and DU124183 enhance the activity of agonists acting at the A₃ AR⁷.

Parkinson’s disease

Some of the symptoms of Parkinson’s disease are thought to be caused by a deficit in dopamine release in the striatum. Interestingly, in this respect, various lines of evidence indicate that interaction between the A_2AAR and dopamine D₂ receptors (D₂Rs) in the striatum is antagonistic. The A_2AAR is co-expressed with D₂Rs in the striatum and heterodimerization of A_2AAR and D R subtypes inhibits D₂R function^1,115. In SH-SY5Y human neuroblastoma cells, A_2AAR–D₂R heteromeric complexes undergo co-aggregation and co-internalization as a result of long-term exposure to A_2AAR or D₂R agonists¹¹⁶. Independent action of the A_2AAR and D₂R has also been proposed in A_2AAR-knockout studies¹¹⁷. In D₂R-knockout mice, A_2AAR agonists and antagonists produce behavioural and cellular functions that are similar to those of their wild-type counterparts, suggesting a D₂R-independent mechanism^118,119.

The possible antagonistic relationship between A_2AARs and D₂Rs in the striatum has provided a rationale for evaluating A_2AAR antagonists in Parkinson’s disease. In addition, epidemiological evidence shows an inverse relationship between caffeine consumption and risk of developing Parkinson’s disease^120,121. A_2AAR antagonists not only provide symptomatic relief but also decelerate the neurodegeneration of dopaminergic cells in patients with Parkinson’s disease¹¹⁹. An A_2AAR antagonist, KW-6002 (istradefylline), has shown potential in a recently completed Phase II clinical trial (now in Phase III trials) as a novel treatment for Parkinson’s disease^123,124, and other A_2AAR antagonists, such as V2006 (a derivative of the antimalarial drug mefloquine), are in development^124–126. V2006, which is beginning Phase II clinical trials, is well tolerated in high doses, and its pharmacokinetic properties suggest that it should be suitable for daily dosing.

Ischaemia and neuroprotection

Pharmacological characterization of A_2AAR-knockout mice has shown that A_2AAR inactivation protects against neuronal cell death induced by ischaemia^127,128 and the mitochondrial toxin 3-NP (an animal model of Huntington’s disease)¹²⁹. Yu et al.¹³⁰ recently created a chimeric mouse model in which A_2AAR knockout is combined with bone marrow transplantation. Selective reconstitution of A_2AARs in bone marrow cells of A_2AAR-knockout mice abolished the neuroprotection against ischaemic brain injury in global A_2AAR-knockout mice. Conversely, selective inactivation of A_2AARs by transplantation of bone marrow cells-derived cells from A_2AAR-knockout mice into wild-type mice attenuated infarct volumes and ischaemia-induced expression of several pro-inflammatory cytokines in the brain. The finding indicated the possible use of A_2AARs on bone marrow-derived cells for the treatment of ischaemic brain injury. In contrast to adult mice, in newborn A_2AAR-knockout mice, brain damage is aggravated after hypoxic ischaemia¹³¹.

The seemingly paradoxical protective effects of both A_2AAR agonism and antagonism indicate the degree of complexity of the system, and the dependence of the results on developmental stage and the specific mechanism of injury. Studies have also shown that A_2AAR agonists result in neuroprotection in some experimental conditions, including cerebral haemorrhagic injury¹³² and ischaemia–reperfusion injury in the spinal cord¹³³.

In the brain, adenosine released under stress conditions counteracts the release and damaging effects of excitatory neurotransmitters, such as glutamate, by activation of the A₁AR. ADAC, an A₁AR agonist, was potent in cerebroprotection in a model of global ischaemia in gerbils¹³⁴. The A₁AR agonist NNC-21-0136 was neuroprotective in both global and focal rodent ischaemia models and had diminished cardiovascular effects in rats compared with reference A₁ AR agonists, such as CPA¹³⁵. However, it was recently shown that deletion of the gene that encodes the A₁AR does not alter neuronal damage that occurs after ischaemia in vivo or in vitro¹³⁶, although A₁ARs have been shown to be relevant to hypoxia protection in newborn mice¹³⁷. Glial cells express all the ARs, and A_2AAR activation was found to promote myelination in Schwann cells, suggesting that a selective agonist could be useful in treating demyelinating diseases such as multiple sclerosis¹³⁸. Although A₃AR expression levels are low in all regions of the brain^139–141, an A₃AR agonist (IB-MECA) depresses locomotor activity in mice¹⁴², suggesting a role for the A₃AR in depression of motor activity. Chronic administration of the A₃AR agonist IB-MECA was highly effective in a gerbil model of cerebroprotection against global ischaemia¹⁴³, and deletion of the A₃AR has a detrimental effect in a model of mild hypoxia, suggesting the possibility of using A₃AR agonists to treat cerebral ischaemia.

Sleep

Adenosine has been found to be an important endogenous sleep-promoting substance^144,145. It mediates the somnogenic effects of prior wakefulness, and also seems to have an important role in the regulation of the duration and depth of sleep after wakefulness¹⁴⁴.

Pharmacological data suggest that A₁ARs are involved in the regulation of sleep¹⁴⁵, but lack of A₁ARs does not prevent the homeostatic regulation of sleep¹⁴⁶. Therefore, it is possible that although the A₁AR is an important factor for sleep regulation in normal animals, other factors, such as the A_2AAR, could compensate for the role of A₁AR when it is deleted. Indeed, it was recently shown that the A_2AAR has a key role in adenosine-mediated sleep-promoting effects¹⁴⁷. It has been suggested that the cholinergic basal forebrain is an essential area for mediating the sleep-inducing effects of adenosine by inhibition of wakefulness-promoting neurons via the A₁AR, and the A_2AAR in the subarachnoid space below the rostral forebrain could have a role in the prostaglandin D₂R-mediated somnogenic effects of adenosine¹⁴⁸. The arousal effect of caffeine was recently shown to be dependent on the A_2A AR¹⁰⁵. However, the locomotor stimulatory effect of high doses of caffeine is not the result of the blockade of either the A₁AR or the A_2AAR, and an effect that is independent of AR activity is probable¹⁰⁴. Although the concept of using AR agonists as modulators for sleep disorders is intriguing, in practice this would be dependent on brain-selective receptor activation.

Other potential applications

Adenosine is important in mediating at least some of the neuronal responses to ethanol^149,150. Ethanol increases brain levels of adenosine by inhibiting adenosine reuptake, which activates A_2AARs and thereby raises cAMP concentrations. The resulting activation of protein kinase A leads to activation of cAMP response element (CRE)-mediated gene expression. Alcohol and adenosine therefore interact synergistically with the activation of D₂Rs in median spiny neurons of the striatum/nucleus accumbens, unlike the otherwise antagonistic relationship between dopamine and adenosine. This is thought to involve the G-protein β,γ dimers, the inhibition of which reduces voluntary alcohol consumption. Therefore, drugs that antagonize the synergism of A_2AAR and D₂R effects might be useful in controlling alcohol abuse.

There has also been recent progress in the imaging of ARs in the brain. An ¹⁸F analogue of DPCPX has been developed as a positron-emission tomographic imaging agent¹⁵¹. The highly potent and selective A_2AAR antagonist SCH442416 in ¹¹C-labelled form has been established as an in vivo receptor-imaging agent in the rat and primate brain, which may eventually strengthen the link between the A_2AAR and disease states, as well as aid in identifying those patients that are likely to benefit from adenosine-related therapeutics¹⁵².

Finally, blockade of A_2AAR has a clear antidepressant effect¹⁵³, suggesting that selective A_2AAR antagonists could be pursued as antidepressant drugs. A₁AR activation has also been a target in the development of antiepileptic therapy⁶, and an inhibitor of adenosine kinase was shown to inhibit seizure activity in animals⁶.

ARs as targets in renal system disorders

Activation of A₁AR protected against ischaemia– reperfusion-induced kidney injury¹⁵⁴. Pretreatment of C57BL/6 mice with an A₁AR antagonist caused a significant deterioration in renal function¹⁵⁴. In addition, A₁AR expression protects renal proximal tubular epithelial cells against cisplatin-mediated apoptosis¹⁵⁵. Mice lacking A₁ARs showed a completely blocked renal glomerular filtration rate by a tubuloglomerular feedback mechanism^156,157, and A₁AR-knockout mice had increased renal injury after ischaemia and reperfusion¹⁵⁸. Therefore, the kidney protective effect of A₁AR agonists has been evident, which provides the basis for their future development as drugs for the treatment of renal failure.

Unlike A₁AR agonists, A₁AR antagonists are effective diuretic agents that are useful in treating fluid-retention disorders, including congestive heart failure, although antagonism of A₁ARs is potentially a concern when using these agents in patients with ischaemic heart and kidney injury. An A₁AR antagonist, BG9719 (CVT-124), was in Phase II clinical trials (now discontinued) for the treatment of acute renal disorders in patients with congestive heart failure^159,160. However, BG9719 contains an epoxide group, which is potentially chemically reactive. However, a non-epoxide-containing antagonist in the same series, BG9928, improves renal function and congestive heart failure without exacerbating cardiac injury¹⁶¹. BG9928, now in Phase II clinical trials, was found to improve sodium excretion in heart failure patients.

A_2AAR agonist-mediated cellular protection is particularly evident in peripheral tissues, including the kidney². The A_2AAR in bone marrow-derived cells is also responsible for protection against ischaemic injury in the kidney¹⁶². A recently developed A_2AAR agonist, ATL-146e, was shown to protect against ischaemic renal injury¹⁶³, and the mixed A₁/A_2AAR agonist AMP579 was initially tested in patients with end-stage renal insufficiency; however, further clinical testing of the compound is impossible due to its inhibition of HERG channels¹⁶⁴.

The expression of the A AR in the kidney¹⁶⁵ 2B and the presence of endogenous A_2BARs in the HEK-293 cell line¹⁶⁶ suggested a potential role in the kidney for this subtype, which is known to regulate cell growth and pro-liferation¹⁶⁷. Finally, mice lacking the A₃AR or wild-type mice in which the A₃AR was blocked pharmacologically had significant renal protection¹⁶⁸, suggesting that A₃AR antagonists might have general renal-protective properties. Recent evidence suggested that both A₁AR agonists and A₃AR antagonists protect the kidney. Ligands possessing dual acting and opposite properties at these AR subtypes could therefore be effective therapeutic agents for renal protection.

ARs as targets in pulmonary disorders

A protective role for the A₁AR in adenosine-dependent pulmonary injury has been proposed¹⁶⁹. Genetic removal of the A₁AR from adenosine deaminase-deficient mice caused enhanced pulmonary inflammation along with increased mucus metaplasia and alveolar destruction. The expression of T_H2 cytokines interleukin-4 (IL-4) andIL-13 in the lungs, as well as chemokines and matrix metalloproteinases, was upregulated. These findings imply that A₁AR agonists have potential for the therapeutic treatment of pulmonary injury.

In the asthmatic lung, adenosine acts as an irritant and bronchoconstrictor, and so a synthetic AR antagonist, perhaps selective for the A_2BAR, could have therapeutic potential in asthma treatment¹⁷⁰. In mouse mast cells, A₃AR activation has been shown to induce mast-cell degranulation¹⁷¹. It has been suggested that this effect is mediated by the A_2BAR in human and canine mast cells^172–174. A bioavailable thiazole derivative that acts as a mixed antagonist at A_2B and A₃ ARs has been proposed to be a candidate therapeutic agent for the treatment of asthma¹⁷⁵; however, this has not been tested in animal models. In addition, the bronchodilating, anti-asthmatic effects of theophylline and other xanthines might involve A_2BAR blockade, although the evidence for this is controversial due to the lack of an adequate animal model. The lack of A_2BAR-selective agonists and, until recently, A_2BAR-knockout mice have hampered further clarification of the functional significance of this receptor.

Activation of the A_2AAR by CGS21680 produces broad-spectrum anti-inflammatory activity in a model of allergic asthma in the Brown Norway rat, suggesting that A_2AAR agonists could be useful alternatives to glucocor-ticosteroids in the treatment of asthma¹⁷⁶. GW328267, an A_2AAR agonist designed for intranasal administration, was also evaluated in Phase II clinical trials for upper respiratory inflammatory disease, chronic obstructive pulmonary disease and asthma; however, the results at the dose used were negative and the compound has been withdrawn from clinical testing¹⁷⁷.

In preclinical testing, the A₃AR agonist IB-MECA was shown to protect against lung injury and apoptosis in cats after reperfusion¹⁷⁸, and this protection was antagonized by MRS1191, a dihydropyridine that is a selective A AR antagonist⁵⁶

As described earlier, prominent species differences in the structure and function of ARs, especially the A₃AR, have been observed. For example, human and rat A₃ARs only share 72% overall identity at the amino-acid level; mast-cell degranulation is induced by the A_2BAR in dogs and humans, but by the A₃AR in mice. It is therefore plausible that some of the functions observed with one animal model might not be obtained in other animal models and in humans. Cross-species testing systems are necessary to validate the receptor function or effects of agonists and antagonists.

ARs as targets in inflammatory disorders

Adenosine-mediated activation of the A_2AAR, which is found in almost all immune cells, including lymphocytes, monocytes, macrophages and dendritic cells¹⁷⁹, seems to attenuate inflammation and reperfusion injury in a variety of tissues. Through A_2AAR activation, adenosine can inhibit T-cell activation, proliferation and production of inflammatory cytokines while enhancing the production of anti-inflammatory cytokines. Activation of the A_2AAR in T_H and cytotoxic T lymphocytes directly inhibits IL-2secretion in vitro and IL-2-driven expansion in vivo¹⁸⁰. In murine CD4⁺ T cells, the A_2AAR agonists ATL-146e and ATL-313 were shown to reduce T cell-receptor (TCR)-mediated production of interferon-γ (IFNγ). Rapid-induction TCR signalling of the mRNA for the A_2AAR suggests that this is a mechanism for limiting T-cell activation and secondary macrophage activation in inflamed tissues¹⁸¹.

Ohta and Sitkovsky¹⁸² reported that A_2AARs are crucially involved in the limitation and termination of prolonged inflammation. Knockout of the A_2AAR in mice showed that no other mechanism for inflammation could compensate fully for the loss of the A_2AAR (which has been referred to as a brake for inflammation¹⁸³) on immune cells. Interestingly, in knockout mouse models, the A_2AAR together with the A₃AR mediated the anti-inflammatory effect of methotrexate, which is used as a treatment of arthritis¹⁸⁴. ATL-146e, a selective agonist of the A_2AAR, profoundly protects mouse liver from reperfusion injury, and the protection is blocked by the A_2AAR antagonist ZM241385. In mice lacking the A_2AAR, protection by ATL-146e is lost and ischaemic injury of short duration is exacerbated, which contrasts with the results obtained in wild-type mice, suggesting a protective role for endogenous adenosine¹⁸⁵. The A_2AAR agonist ATL-146e is also of interest for the treatment of sepsis¹⁸⁶, inflammatory bowel disease¹⁸⁷ and for inclusion in drug-eluting stents to prevent restenosis after angioplasty. Selective activation of the A_2AAR has also been shown to reduce skin pressure, ulcer formation and inflammation¹⁸⁸, and wound healing is accelerated¹⁸⁹.

The A₃AR has been implicated in mediating allergic responses in mice; it facilitates the release of allergic mediators, such as histamine, in mast cells¹⁹⁰. Systemic infusion of IB-MECA causes scratching in mice that is prevented by co-administration of histamine antagonists¹⁴². The potentiation by Cl-IB-MECA of antigen-dependent degranulation of mast cells, as measured by hexosaminidase release, was lost in mice lacking A₃ ARs¹⁷¹. Attenuation of lipopolysaccharide-induced tumour-necrosis factor-α (TNFα) production was lower in mice lacking the A₃ AR than in control mice¹⁷¹.

Finally, adenosine has been implicated in arthritis treatment, and the possibility of administration of A_2A AR agonists for this purpose remains open¹⁹¹. The A₃AR agonist IB-MECA also showed beneficial effects in early human trials¹⁹².

ARs as targets for endocrine disorders

A₁AR agonists are under consideration as therapeutic candidates for obesity-related insulin resistance and type 2 diabetes¹⁹³. In mice overexpressing the A₁AR in adipose tissue, lower concentrations of plasma free fatty acids were observed than in litter-matched controls, and the transgenic mice did not develop insulin resistance. This supports a significant physiological role for adipocyte A₁AR in the control of lipolysis. GR79236 has been tested in humans for adjuvant therapy in insulin resistance (type 2 diabetes). This A₁AR agonist ameliorates the hypertriglyceridaemia induced by fructose feeding, and the reduction in fatty acid levels is associated with secondary improvements in glucose tolerance.

A_2BAR antagonists are also under consideration for diabetes treatment. A 2-alkynyl-8-aryl-9-methyladenine derivative developed by Eisai showed hypoglycaemic activity in an animal model of type 2 diabetes, suggesting that adenosine agonist-induced glucose production in rat hepatocytes is mediated through the A_2B AR¹⁹⁴. The A_2AAR agonist MRE-0094 is in Phase I clinical trials as a treatment for chronic diabetic neuropathic foot ulcers due to the anti-inflammatory and wound healing effects of A_2AAR agonists.

ARs as targets in cancer

A₃AR agonists can induce or attenuate apoptosis depending on the range of agonist concentrations used, which might have important implications for their therapeutic use in disorders in which the aim is to either attenuate apoptosis, such as arthritis (see above), or induce it, as in cancer. In human eosinophils and human promyelocytic HL-60 cells^195,196 the A₃AR agonist Cl-IB-MECA seems to induce apoptosis at relatively high concentrations (>10 μM), but in RBL-2H3 cells, a lower concentration (1 μM) of the A₃AR agonist IB-MECA block apoptosis that is induced by ultraviolet irradiation¹⁹⁷.

A role for the A₃AR in mediating control of the cell cycle has been reported¹⁹⁸. Activation of the A₃AR by adenosine triggers a cell survival response; by contrast, activation of A_2AARs induces an apoptotic signalling pathway that involves protein kinase C and MAPKs¹⁹⁹. Overexpression of the A₃AR in transgenic mice resulted in embryonic lethality¹⁴¹, suggesting the possible use of selective A₃AR agonists in anticancer therapy.

A₃AR activation has been implicated in inhibition of tumour growth both in vitro and in vivo²⁰⁰, and IB-MECA is in clinical trials for colon carcinoma. However, the novel anticancer effect discovered by Fishman and colleagues is caused by a cytostatic effect on tumours related to the WNT pathway³², rather than by induction of apoptosis. Recently, it was shown that the A₃AR is more highly expressed in tumour than in normal cells, which may justify A₃AR as a potential target for tumour growth inhibition²⁰¹. In human breast cancer cell lines, IB-MECA downregulated the oestrogen receptor and completely inhibited cell growth²⁰².

ARs as targets in visual disorders

The A₃AR-knockout mouse had significantly lower intraocular pressure, suggesting that A₃AR antagonists have potential in the treatment of glaucoma²⁰³. Most reported A₃AR antagonists are selective only at the human A₃AR, and so are not suitable for use in rodent models, but the selective A₃AR antagonist OT-7999 reduced intraocular pressure in the monkey²⁰⁴. Encouragingly, the cross-species A₃AR antagonist MRS1292 was recently found to reduce mouse intraocular pressure and also inhibited adenosine-triggered human non-pigmented ciliary epithelial cell fluid release⁶⁰.

Conclusions

The medicinal chemistry of ARs is well developed, and selective agonists and antagonists have been generated for most of the receptor subtypes. In addition to the potential of directly acting orthosteric ligands, allosteric modulation of ARs is a promising approach. The application of genetic therapy with neoceptors could also potentially achieve organ or tissue selectivity in the future (BOX 1). With suitable pharmacological probes and the availability of knockout mice for three of the four subtypes, the basic science of ARs has progressed to the identification of novel therapeutic targets (FIG. 7). It is hoped that new agents in development will avoid the undesirable side effects that have impeded the clinical development of AR ligands in the past. Selective agonists are well advanced in clinical trials for the treatment of atrial fibrillation, pain, neuropathy, pulmonary and other inflammatory conditions, and cancer. Selective antagonists have entered clinical trials for the treatment of Parkinson’s disease and congestive heart failure. Both in the case of diseases such as stroke, where there is an unmet medical need, and for diseases that already have pharmacological intervention options, the introduction of adenosine-based drug therapy will provide novel mechanisms for therapy. With the maturing of AR science, it is time for some of the myriad of selective agents synthesized to be implemented in the fight to improve human health.

Figure 7. Novel disease targets for selective adenosine receptor ligands.

Most promising prospects exist for treatment of arrhythmias, ischaemia of the heart and brain, pain, neurodegenerative diseases, sleep disorders, inflammation, diabetes, renal failure, cancer and glaucoma, and in cardiovascular imaging. High and intermediate levels of A₁ adenosine receptor (AR) expression were found in the brain, heart, adipose tissue, stomach, vas deferens, testis, spleen, kidney, aorta, liver, eye and bladder¹⁴⁰. The A_2AAR is highly expressed in the striatum, nucleus accumbens and olfactory tubercle¹⁴⁰. High and intermediate expression levels were also found in immune cells, heart, lung and blood vessels. The A_2B AR was generally expressed at low levels in almost all tissues¹⁴⁰. Rat testis has particularly high concentrations of A₃AR mRNA, with moderate levels in lung. The highest levels of human A₃AR mRNA have been found in lung and liver. A₃ARs have been detected in various tissues including testis, lung, kidney, placenta, heart, brain, spleen, liver, uterus, bladder, jejunum, aorta, proximal colon and eyes.

Box 1. Neoceptors.

Because of the widespread distribution of native adenosine receptors (ARs), their activation is inherently nonselective. To address this issue, efforts have been made to reengineer ARs into ‘neoceptors’ that can recognize uniquely modified nucleosides that are inactive at the native ARs^38,205,206. This neoceptor strategy, which is intended for eventual use in organ-targeted gene therapy, is made possible by modelling of the putative ligand-binding site of a G-protein-coupled receptor, leading to identification of sites for mutagenesis, and incorporation of a complementary functional group in a synthetic agonist (neoligand). A novel electrostatic or H-bonding pair formed between the neoceptor and neoligand allows receptor activation that is orthogonal with respect to the native species²⁰⁹. So far, neoceptors have been developed for A_2A and A₃ ARs.

Glossary

Angiogenesis

The growth of new blood vessels — for example, in pathology, the generation of a blood supply to a tumour.

Allosteric site

A modulatory binding site on a receptor that is topographically distinct from the agonist binding site.

Pertussis toxin

A compound that inhibits the guanine nucleotide binding proteins G_i and G_o via ADP-ribosylation.

Bradycardiac effect

An arrhythmia typified by an abnormally slow heart rate.

Mast cell

A type of leukocyte that has large secretory granules that contain histamine and various protein mediators.

Photoisomerization

A conversion between structural isomers caused by light-induced excitation.

Paroxysmal supraventricular tachycardia (PSVT)

A regular, abnormally fast heart beat caused by rapid firing of electrical impulses from a focus above the AV (atrioventricular) node.

Atrial fibrillation

A condition in which disorganized electrical conduction in the atrial walls results in ineffective pumping of blood into the ventricle and an irregular heart rhythm.

Dromotropic

Refers to velocity of AV nodal conduction in the heart.

Discriminative stimulus

In instrumental conditioning, the external stimulus that signals a particular relationship between the instrumental response and the reinforcer.

T_H2 cytokines

Cytokines such as interleukin (IL)-3, -4, -5, -6, -10 and -12 secreted by T_H2 helper T lymphocytes to control various aspects of the antibody response.

Bioavailability

The fraction or percentage of an administered drug or other substance that becomes available to the target tissue after administration.

Somnogenic

Sleep-inducing