Adenosine A₁ and A₃ Receptors: Distinct Cardioprotection

Abstract

Adenosine is released in large amounts during myocardial ischemia and exerts potent cardioprotective effects in the heart. Although these observations on adenosine have been known for a long time, how adenosine acts to achieve its antiischemic effect remains incompletely understood. Recent advances in the chemistry and pharmacology of adenosine receptor ligands have provided important and novel information on the function of adenosine receptor subtypes in the cardiovascular system. The development of model systems for the cardiac actions of adenosine has yielded important insights into its mechanism of action and have begun to elucidate the sequence of signaling events from receptor activation to the actual exertion of its cardioprotective effect. The goal of the current article is to review recent advances on the cellular and molecular mechanisms that mediate the cardiac actions of adenosine and to show the cardioprotective effect of novel adenosine ligands.

INTRODUCTION

It is well documented that adenosine can exert potent cardioprotective effects in the heart [Ely and Berne, 1992; Babbitt et al., 1989]. An important cardioprotective effect of adenosine is known as the preconditioning effect of adenosine, which has been well described in virtually all species studied [Downey, 1992; Gross, 1992; Yao and Gross, 1994b], including humans [Leesar et al., 1997]. Two types of preconditioning have been described. Classical or early preconditioning, first described more than 14 years ago [Murry et al., 1986], has a duration of cardioprotection that lasts 1–2 h after the preconditioning stimulus. In a more recently described type of preconditioning, the protection reappears 24 h after the initial preconditioning stimulus. This is termed the “second window of protection” or late preconditioning [Kuzuya et al., 1993; Marber et al., 1993; Baxter et al., 1994; Baxter and Yellon, 1997, 1999]. The advantage of the preconditioning-like effect of adenosine or adenosine receptor agonist is that the heart can be protected without being first exposed to ischemia. While these studies clearly demonstrated a number of potent cardioprotective actions of adenosine in the heart, how adenosine acts to achieve its protective effects is not well understood. The goal of the current article is to review recent advances in the cellular and molecular mechanisms that mediate the cardiac actions of adenosine and to show the cardioprotective effect of novel adenosine ligands.

ADENOSINE A₁ AND A₃ RECEPTORS

Adenosine receptors belong to the superfamily of G-protein-coupled receptors and consist of four subtypes: A₁, A_2A, A_2B, and A₃ [Linden et al., 1998; Stiles, 1992]. As G-protein-coupled receptors, all four are characterized by seven hydrophobic membrane-spanning domains. The adenosine A₁ and A₃ receptors are coupled to the inhibitory G-protein G_i. Classically, activation of these G_i-coupled receptors result in inhibition of adenylyl cyclase activity [Matherne et al., 1998]. However, recent studies showed that these receptors can also modulate other effectors in the heart. For example, the A₁ receptor is also coupled to the atrial cardiac myocyte potassium channel (Kir) [Liang, 1992] and to phospholipase C [Parsons et al., 2000]. While the A₃ receptor is known to couple to phospholipase C, a recent study shows that this receptor can also mediate phospholipase D activation [Parsons et al., 2000]. In fact, in the cardiac cell the A₃ receptor is preferentially coupled to phospholipase D, whereas the A₁ receptor appears to couple selectively to phospholipase C. The specificity of the A₁ receptor–PLC and A₃ receptor–PLD linkages raised the possibility that such receptor-effector coupling may underlie distinct biological effects in the heart (see below).

CHEMISTRY AND PHARMACOLOGY OF ADENOSINE RECEPTOR LIGANDS

A variety of adenosine agonists (Fig. 1) [Ely and Berne, 1992; Babbitt et al., 1989; Downey, 1992; Gross, 1992; Yao and Gross, 1994a; Leesar et al., 1997; Murry et al., 1986; Kuzuya et al., 1993; Marber et al., 1993; Baxter et al., 1994; Baxter and Yellon, 1997, 1999; Linden and Jacobson, 1998; Stiles, 1992; Matherne et al., 1998; Liang, 1992; Parsons et al., 2000; Kitakaze et al., 1997; Musser et al., 1999; Jacobson et al., 1993] and antagonists (Fig. 2) [Fozard et al., 1996; Hill et al., 1997, 1998; McVey et al., 1999; Klotz et al., 1989; Liang and Jacobson, 1998; von Lubitz et al., 1993, 1996, 1999a; Snowdy et al., 1999; Jacobson et al., 1987; Zhang et al., 1997; Tracey et al., 1997; Olsson and Pearson, 1990; Schwabe and Trost, 1980; Sheldrick et al., 1999; Knutsen et al., 1999; Ji et al., 1992] have been used in recent studies of cardiac function. There are highly selective agonists for A₁-, A_2A-, and A₃-receptors. There are now antagonists selective for all four subtypes of adenosine receptors. To truly define subtypes involved in a particular response requires the use of both agonists and antagonists (including negative controls, keeping in mind that the selectivity of these agents is often dependent on species). Furthermore, some of the classical pharmacological tools defined previously as selective with respect to A₁- and A_2A-receptors must now be evaluated fully at A_2B- and A₃-receptors.

Fig. 1.

Agonists at the various adenosine receptor subtypes. a: Nonselective and A₁ receptor-selective. b: A_2A receptor- and A₃ receptor-selective agonists. K_i values, in nM, for A₁, A_2A, and A₃ receptors (rat, unless indicated) are shown in that order below the structure and name for each compound.

Fig. 2.

Antagonists at the various adenosine receptor subtypes. a: Nonselective, A₁ receptor- and A_2A receptor-selective. b: A₃ receptor- and A_2B receptor-selective antagonists. The structure and K_i values for adenosine receptor antagonists are shown. K_i values, in nM, for A₁, A_2A, and A₃ receptors (rat, unless indicated) are shown in that order below the structure and name for each compound.

In addition to the use of agonists it is feasible to modulate levels of stimulation of the receptors indirectly by altering the adenosine levels through inhibition of its metabolism, e.g., inhibitors of adenosine kinase, adenosine deaminase, and other enzymes, or its uptake via nucleoside transporters [Kitakaze et al., 1997]. A third category of adenosine receptor modulators consists of selective allosteric enhancers of the action of agonists at A₁-receptors [Musser et al., 1999].

Adenosine Agonists

Nonselective agonists

NECA (1, 5′-N-ethylcaboxamidoadenosine) is among the most potent nonselective adenosine agonists and is still the most potent agonist reported at the A_2B adenosine receptor subtype, for which there are currently no selective agonists. Previously, NECA had been used erroneously as an A_2A-selective agonist. As are most of the synthetic adenosine agonists, NECA is stable toward deamination by adenosine deaminase. NECA is also not subject to phosphorylation by adenosine kinase, since a 5′-hydroxyl group is absent.

Many N⁶-substituted adenosine derivatives display selectivity for A₁ receptors, although some are nonselective. Several of these relatively nonselective agonists have been used in cardiac studies. For example, APNEA (2, N⁶-p-aminophenylethyladenosine) has been used to activate cardiac and other A₃ receptors [Fozard et al., 1996]. The selectivity of APNEA is also dependent on species, e.g., differences exist between canine and rabbit [Hill et al., 1997]. Similarly, I-ABA (3, N⁶-p-iodobenzyladenosine) has been used as a radio-iodinated ligand for both A₁ or A₃ receptors and in cardioprotection studies [Hill et al., 1998]. An A₁-receptor selective antagonist must be coadministered in order to draw conclusions concerning the selective activation of A₃ receptors using these nonselective agonists.

The carbocyclic nucleoside AMP579, 4, which contains both 5′-uronamide and N⁶-substitutions, has been shown to be cardioprotective [Budde et al., 2000].

A₁ receptor agonists

Analogs substituted at the N⁶-exocyclic amine with aryl, alkyl, or cycloalkyl groups (6–13) generally activate selectively the A₁ adenosine receptor subtype. Among the more highly selective A₁ agonists are CHA (5, N⁶-cyclohexyladenosine) and CPA (6, N⁶-cyclopentyladenosine) and its 2-chloro analog CCPA, 7 [Klotz et al., 1989]. Both CPA and CCPA have been used in various models of cardioprotection [Liang and Jacobson, 1998] and cerebroprotection [von Lubitz et al., 1993]. CVT510 (8, N⁶-((S)-3-tetrahydrofuranyl)adenosine) is an ether derivative structurally related to CPA, which is highly selective for the A₁ receptor and has been proposed for the treatment of reentrant tachycardias that involve the A-V node [Snowdy et al., 1999].

ADAC (9, N⁶-[4-[[[4-[[[(2-aminoethyl)amino]carbonyl]methyl]-anilino]carbonyl]methyl]phenyl-adenosine) was introduced in the mid-1980s as a functionalized congener (i.e., it contains a primary amino group at the terminal position of a strategically placed elongated chain, thus allowing conjugation to carriers without losing receptor binding) with high affinity and selectivity for the A₁ receptor subtype [Jacobson et al., 1987]. ADAC is cardioprotective in a chick cardiac myocyte model of ischemia [Stambaugh et al., 1997], and also cerebroprotective in a model of global ischemia [von Lubitz et al., 1999a]. In the CNS, acute administration of either CPA or the A₁-selective adenosine amine congener ADAC was protective in a model of global ischemia in gerbils [von Lubitz et al., 1996]. Acute administration of ADAC at a very low dose range (25–100 mg/kg) was found to limit postischemic hippocampal damage in gerbils following bilateral carotid artery occlusion. Side effects of this adenosine agonist on heart rat and blood pressure were minimal within the above-mentioned dose range [von Lubitz et al., 1996]. ADAC has been used as a chemical precursor for spectroscopic, e.g., fluorescently labeled, probes of the A₁ receptor. Also, following conjugation with bifunctional crosslinking reagents, m- and p-phenylene diisothiocyanate (DITC), the resulting conjugates, such as 10, p-DITC-ADAC, serve as irreversibly binding agonists at the A₁ receptor in the heart [Zhang et al., 1997].

R-PIA (11, (R)-N⁶-phenylisopropyladenosine) is only moderately selective for A₁ receptors. It has been used as a selective A₁ agonist for studies of cardioprotection [Tracey et al., 1997]. Thus other more selective agonists are generally preferred as pharmacological probes, especially in the heart, in which the selectivity of A₁ adenosine agonists may be diminished with respect to binding affinities at A₁/A_2A receptors [Olsson and Pearson, 1990]. In the late 1960s, prior to the discovery of its mechanism of action, i.e., adenosine receptors, this compound was reported to have a hypotensive effect in humans, presumably due to activation of A_2A receptors, but the clinical development was curtailed due to side effects. [³H] R-PIA is a satisfactory radioligand for binding to A₁ adenosine receptors, but only when they occur in very high density such as in the cerebral cortex [Schwabe and Trost, 1980].

GR79236 (12, N⁶-[(1S, trans]-2-hydroxycyclopentyl)-adenosine) is an agonist selective for A₁ adenosine receptors that is closely structurally related to CPA. GR79236 has been shown to be cardioprotective in rabbits [Sheldrick et al., 1999]. In contrast, another A₁ agonist, NNC 21–0136 (13, N⁶-[(R)-1-(2-benzothiazolyl)thio-2-propyl]-adenosine) has a dose window of selective action at A₁ receptors in the CNS [Knutsen et al., 1999] and has been demonstrated to be neuroprotective.

A_2A receptor agonists

Derivatization of the 2-position of adenosine is the class of modifications most commonly associated with selectivity for the A_2A-receptor. Perhaps the most widely used A_2A-receptor selective agonist in pharmacological studies is CGS 21680 (14, 2-[4-[(2-carboxyethyl)-phenyl]ethylamino]-5′-N-ethylcarboxamidoadenosine), which is modified at both 2- and 5′-positions (an NECA-like derivative). CGS 21680 has a K_i value of approximately 15 nM at both rat and human [Ji et al., 1992] A_2A-receptors. It is 140-fold selective for A_2A vs. A₁ receptors [Hutchison et al., 1989], and does not readily cross the blood–brain barrier, making it useful in cardiovascular studies. CGS 21680 was developed originally for the treatment of hypertension through its peripheral action. CGS 21680 is also highly selective for the A_2A vs. A_2B receptor [Hide et al., 1992], although its affinity at rat A₃ receptors is greater than its rat A_2A receptor affinity [Kim et al., 1994]. [³H] CGS 21680 is a useful high affinity agonist radioligand for the A_2A receptor displaying low non-specific membrane binding.

The ethylenediamine derivative of CGS 21680, APEC, is an A_2A-selective amine-functionalized congener (see ADAC, above), which does readily cross the blood–brain barrier [Nikodijevic et al., 1990]. Thus, APEC is suitable for the study of the function of A_2A receptors in the CNS, following its administration peripherally. The covalent conjugate of APEC with p-amino-phenylacetic acid (PAPA-APEC) is useful as a high-affinity radioligand for A_2A receptors and was used for photoaffinity crosslinking in the first study to distinguish A₁ and A_2A receptors as distinct molecular species [Barrington et al., 1989]. APEC has been used as a chemical precursor for a fluorescent probe of the A_2A receptor [McCabe et al., 1992], and its conjugation with a bifunctional crosslinking reagent, DITC, has resulted in an irreversibly acting A_2A agonist with coronary vasodilator properties [Niiya et al., 1993].

HENECA (15, adenosine 2-(hexynyl)-5′-ethyluron-amide), which is modified with an alkyne group at the 2-position as well as at the 5′-position, is a highly potent vasodilator and A_2A receptor agonist with a K_i of 2.2 nM in binding assays [Cristalli et al., 1995]. HENECA is not as selective as CGS 21680 for A_2A vs. either A₁ or A₃ receptors, although it has greater binding affinity, and has been shown to protect rabbit heart when given during reperfusion [Cargnoni et al., 1999].

While the doubly substituted adenosine analogs CGS21680 and HENECA are both highly potent and selective for A_2A receptors, several singly substituted analogs, CPCA (2, adenosine 5′-cyclopropyluronamide) and DPMA (16, N⁶-[(2-(3,5-dimethoxyphenyl)-2-(2-methylphenyl)ethyl]adenosine), have been used for the activation of A_2A receptors in cardiac studies. DPMA inhibits lipopolysaccharide-induced cardiac expression of tumor necrosis factor-a [Wagner et al., 1998]. CPCA, a NECA analog used in a study of blood flow effects during reperfusion [Nomura et al., 1997], is of low A_2A-selectivity, while DPMA is approx. 30-fold selective for A_2A vs. A₁ receptors.

A₃ receptor agonists

The first highly potent and selective A₃ agonists introduced contain substitution at both 5′- and N⁶-positions. One of the most widely used A₃ receptor agonists is IB-MECA (17, N⁶-(3-iodobenzyl)-adenosine-5′-N-methyluronamide), which is approximately 50-fold selective [Gallo-Rodriguez et al., 1994] for rat A₃ vs. either A₁ or A_2A receptors in vitro and appears to be A₃-selective in vivo in several models [Jacobson, 1998; von Lubitz et al., 1999b]. The corresponding 3-chlorobenzyl derivative (18, CB-MECA, MRS 452) has been demonstrated to be cardioprotective via the selective activation of A₃ receptors [Tracey et al., 1998]. The corresponding derivative containing a 4-amino group in the benzyl ring is I-ABMECA (19), which is used widely as a high-affinity radioligand for recombinant A₃ receptors [Olah et al., 1994], although in autoradiographic studies of rat brain sections it was found to also label A₁ and A_2A receptors.

Substitution of adenosine analogs at the 2-position in combination with modifications at N⁶ and 5′-positions was found to further enhance A₃ selectivity [Kim et al., 1994]. Cl-IB-MECA (2 0, 2-chloro-N⁶-(3-iodobenzyl)-adenosine-5′-N-methyluronamide) displayed a K_i value of 0.33 nM at A₃ receptors and is selective for A₃ vs. A₁ and A_2A receptors by 2,500- and 1,400-fold, respectively. A₃ agonists have been shown to both inhibit adenylyl cyclase and activate phosphatidylinositol-4,5-bis-phosphate-specific phospholipase C [Kim et al., 1994; Jacobson et al., 1997]. There may be differences in the relative activation of various second messenger systems based on the concentration of the A₃ agonists.

Low concentrations of A₃ agonists such as Cl-IB-MECA (100–200 nM) have been shown to have protective effects in various models. High concentrations of A₃ agonists have been found to induce apoptosis [Kohno et al., 1996]. A high concentration of Cl-IB-MECA (≥10 mM) was lethal to cultured rat cerebellar granule neurons and the toxic effects of glutamate were also augmented. In contrast, cytoprotective effects of A₃ agonists have been observed. In human ADF cells of astroglial lineage, 100 nM Cl-IB-MECA caused a marked reorganization of the cytoskeleton, with appearance of stress fibers and numerous cell protrusions (which became enriched in the antiapoptotic protein Bcl-x_L), accompanied by induction of the expression of Rho, a small GTP-binding protein [Abbracchio et al., 1997]. Thus the activation of the A₃ receptor may have a dual role, i.e., damage-inducing at high levels of activation and protective at low levels.

Adenosine Antagonists

A large number of xanthines have been synthesized as adenosine receptor agonists. It should be noted that, in general, the A₃ receptor affinity of most xanthines is highly species-dependent, with the affinity at human receptors typically >100-fold greater than that at rat receptors. Also, many of the 8-substituted xanthines display considerable affinity at A_2B receptors (see below). Xanthines are often of low aqueous solubility; for example, the maximal solubility of DPCPX in pH 7.4 buffer is 30 mM.

Recently, numerous classes of chemically diverse nonxanthine antagonists have been reported.

Nonselective antagonists

The xanthines comprise the largest structural class of adenosine receptor antagonists. Theophylline (21, 1,3-dimethylxanthine) and caffeine, its 7-methyl analog, are the classical nonselective antagonists of micromolar affinity at A₁/A_2A/A_2B receptors. Enlarging the 1,3-dialkyl substituents to propyl and other groups or adding a phenyl ring at the 8-position generally enhances receptor affinity. The two sulfonate-containing analogs, SPT (22, 8-p-sulfophenyltheophylline) and DPSPX (23, 8-p-sulfophenyl-1,3-dipropylxanthine), are water-soluble antagonists, thus overcoming the typical low solubility of adenosine antagonists, although they are relatively nonselective [Forman et al., 2000].

BWA 1433 (24, 8-p-(carboxyethenyl)phenyl-1,3-dipropylxanthine) has been used as an antagonist for both A₁ and A₃ receptors, but is highly selective at the A₃ vs. the A₁ receptor in rabbit [Hill et al., 1997]. The xanthine amine congener, XAC (25, 8-[4-[[[[(2-amino-ethyl)-amino]carbonyl]methyl]oxy]phenyl]-1,3-dipropylxanthine) displays moderate selectivity for A₁ receptors in the rat, while in other species it is less selective. For example, the affinity of XAC (K_i = 3.0 nM) and other 8-phenylxanthines at human A_2A receptors is greatly increased [Ji et al., 1992].

A₁ receptor antagonists

DPCPX, also known as CPX, (26, 8-cyclopentyl-1,3-dipropylxanthine) was the first highly selective A₁ antagonist to be introduced [Jacobson et al., 1996]. Tritiated DPCPX has been used widely as a high-affinity radioligand for the A₁ receptor (K_D ~0.5 nM). Although it is roughly 500-fold selective for the rat A₁ vs. A_2A receptor, its use as an absolutely selective adenosine antagonist must be tempered by studies at the other two subtypes. The potency of DPCPX at A_2B receptors and at human (but not rat) A₃ receptors is intermediate (K_i values of approximately 30 and 75 nM, respectively) [Ji and Jacobson, 1999; Salvatore et al., 1993]. Curiously, DPCPX was found to have an unanticipated action in the promotion of chloride efflux in CFPAC cells, which contain a mutation of the CFTR chloride transporter (the F508) that occurs in cystic fibrosis [Guay-Broder et al., 1995]. DPCPX is in Phase II clinical trials related to the treatment of cystic fibrosis, an action apparently not resultant from its A₁ receptor antagonist properties.

A conformationally constrained refinement of the 8-cyclopentyl group of DPCPX is the 2-(5,6-epoxy)-norbornyl group, leading to extremely high selectivity for A₁ receptors in binding and functional assays [Pfister et al., 1997]. The pure S-enantiomer, BG9719, also known as CVT 124 (27, S-1,3-dipropyl-8[2-(5,6-epoxynorbornyl)]xanthine), the more potent isomer, is selective for A₁ receptors even with respect to the A_2B receptor. It is under development by Biogen Corp. as a diuretic for the treatment of congestive heart failure and renal failure.

Among nonxanthine adenosine antagonists which have been explored are both purine and nonpurine heterocycles. 9-Alkyladenine derivatives, such as WRC-0571 (28, 8-(N-methylisopropyl)amino-N⁶-(5′-endohydroxyendonorbornyl)-9-methyl-adenine) [Martin et al., 1996], displays a selectivity for A₁ vs. A₃ receptors. FK352 (29, (2R)-1-[(2E)-1-oxo-3-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)-2-propenyl] 2-piperidineacetic acid) is also a nonxanthine A₁ receptor antagonist that was shown to be cardioprotective through the stimulation of antioxidants [Katori et al., 1999].

A_2A receptor antagonists

Although most 8-substituted alkylxanthine derivatives reported are selective for the A₁ subtype, 8-styrylxanthines have been demonstrated to be A_2A-selective antagonists. For example, CSC (30, 8-(3-chlorostyryl)caffeine) displays selectivity for rat A_2A vs. rat A₁, human A_2B, and rat A₃ receptors of 520-fold, 62-fold, and >1,800-fold, respectively [Jacobson et al., 1996]. Selective A_2A receptor antagonists, such as KW6002 (31) and SCH 52861 (see below), have potential as novel treatments for Parkinson’s disease and stroke.

Among highly potent and selective nonxanthine antagonists for the A_2A subtype is SCH 58261 (32, 5-amino-7-(2-phenylethyl)-2-(2-furyl)pyrazolo[4,3-e]-1,2,4-triazolo[1,5-c]pyrimidine) [Zocchi et al., 1996], which is highly selective for A_2A vs. both A₁ and A_2B receptors and has been tritiated for use as a radioligand [Dionisotti et al., 1996]. Another potent nonxanthine antagonist, ZM 241385 (33, 4-(2-[7-amino-2-(²-furyl)[1,2,4]triazolo-[2,3a][1,3,5]triazinyl-amino]ethyl)-phenol) [Poucher et al., 1995] is not as selective as SCH 58261 for A_2A vs. A_2B receptors. In fact, [³H] ZM 241385, while an excellent radioligand for A_2A receptors, is of sufficiently high affinity at recombinant A_2B receptors for use as a radioligand when these receptors are expressed in systems lacking A_2A receptors [Ji and Jacobson, 1999].

A_2B receptor antagonists

Recently, a series of potent and A_2B receptor-selective xanthines was reported solving a long-elusive need among adenosine receptor ligands [Kim et al., 2000]. A p-cyanoanilide derivative, MRS 1754 (34, 8-[4-[[(4-cyano)phenylcarbamoylmethyl]oxy]phenyl]-1,3-di-(npropyl)xanthine) was 200–300-fold selective for hA_2B receptors vs. hA₁/A_2A/A₃ receptors. Although less selective vs. rA₁/A_2A receptors, MRS 1754 (100 nM) inhibited NECA-stimulated calcium mobilization in HEK cells expressing the human A_2B receptor.

A₃ receptor antagonists

Since xanthines tended to bind only weakly to A₃ receptors, an alternate strategy, i.e., screening diverse molecules in chemical libraries for leads, was adopted in the design of selective A₃ receptor antagonists. The Merck group [Jacobson et al., 1996], using high-throughput screening of heterocyclic compounds, has identified antagonists which are highly selective for human A₃ receptors.

1,4-Dihydropyridines, known as potent blockers of L-type calcium channels and used widely in treating coronary heart disease, were found to bind appreciably to human adenosine A₃ receptors [Jacobson et al., 1997]. Common dihydropyridine drugs typically bound either nonselectively (for example, nifedipine, with a K_i value of 8.3 mM) or in some cases with selectivity for the A₃ vs. other adenosine receptor subtypes (for example, S-niguldipine, with a K_i value of 2.8 μM).

1,4-Dihydropyridines (and later pyridines, see MRS 1523, 37) were used as templates, in which it was possible to select for affinity at adenosine receptors and completely deselect for affinity at L-type Ca²⁺-channels. For example, the dihydropyridines MRS 1191 (36, 3-ethyl 5-benzyl 2-methyl-6-phenyl-4-phenylethynyl-1,4-(±)-dihydropyridine-3,5-dicarboxylate) and MRS 1097 (35, 3-ethyl 5-benzyl 2-methyl-6-phenyl-4-phenylethynyl-1,4-(±)-dihydropyridine-3,5-dicarboxylate) were found to inhibit radioligand binding at the human A₃ receptor with K_i values of 31 and 108 nM, respectively. MRS 1191 was nearly inactive in binding at A₁ and A_2A receptor sites (i.e., >1,300-fold selective). MRS 1191 is moderately selective at rat A₃ receptors, with a K_i value of 1.42 μM. It competitively antagonized the effects of IB-MECA, an A₃ receptor-selective agonist, on inhibition of adenylyl cyclase mediated by the recombinant human A₃ receptor. Dihydropyridine A₃ receptor antagonists have also proven A₃ receptor-selective in chick cardiac myocytes, in which the activation of A₃ receptors induces protective antiischemic effects. MRS 1191 was also shown to be A₃ receptor-selective in the rat hippocampus, in which it was demonstrated that A₃ receptor activation suppresses the effects of activation of presynaptic A₁ receptors on inhibition of neurotransmitter release [Dunwiddie et al., 1997]. MRS 1191 was also utilized to demonstrate that presynaptic A₃ receptor activation antagonizes group III metabotropic glutamate autoreceptors [Macek et al., 1998].

Antagonism at human A₃ receptors by MRS 1191 was demonstrated in functional assays consisting of agonist-induced inhibition of adenylate cyclase and the stimulation of binding of [³⁵S]GTP-γ-S to the associated G-proteins, with a K_B value of 92 nM [Jacobson et al., 1997].

A pyridine derivative, MRS 1523 (37, 5-ethyl 2-ethyl-3-(ethylsulfanylcarbonyl)-4-propyl-6-phenylpyridine-5-carboxylate) is highly potent at human A₃ receptors [Li et al., 1998]. It is one of the few A₃ antagonists reported to be selective and potent at both human and rat A₃ receptors, with K_i values of 18.9 and 113 nM, respectively.

MRS 1220, 38, is an N⁵-p-phenylacetyl derivative of the nonselective nonxanthine antagonist CGS 15943 (9-chloro-2-(2-furyl)[1,2,4]triazolo[1,5-c]quinazolin-5-amine). While MRS 1523 is an A₃ antagonist of broad application to various species, the triazoloquinazoline MRS 1220 is extremely potent in binding to the human (K_i = 0.65 nM) but not the rat A₃ receptor [Jacobson et al., 1997]. MRE 3008-F20, (39, 5-[[(4-methoxyphenyl)-amino]carbonyl]amino-8-propyl-2-(2-furyl)-pyrazolo[4,3-e]triazolo[1,5-c]pyrimidine) is highly potent and selective at the human A₃ receptor, and its use as a selective, high-affinity radioligand was recently reported [Varani et al., 2000]. The use of MRS 1220 or MRE 3008-F20 as an A₃ receptor antagonist in nonprimate species is not recommended.

CARDIAC MYOCYTE MODELS FOR STUDYING THE CARDIOPROTECTIVE EFFECTS OF ADENOSINE

Intact heart preparations, including isolated buffer-perfused heart and blood-perfused in situ heart, have been the main models for investigating cardioprotection during ischemia and reperfusion [see Ely and Berne, 1992; Downey, 1992; Gross, 1992]. Studies using intact heart models demonstrate the potent antiischemic effect of adenosine and adenosine analogs.

Studies on the intact heart model of cardioprotection provided not only the first demonstrations of the various cardioprotective effects of adenosine, but also suggest that adenosine or adenosine analogs can serve as novel antiischemic agents in vivo. However, recent investigations have begun to focus on the molecular and cellular signaling pathways that mediate the cardioprotective effect of preconditioning. The intact heart is composed of cardiac myocytes, vascular smooth myocytes, endothelial cells, neural tissues, and in the case of the in situ heart, circulating immune cells and platelets as well. As each of these individual cell types has one or more adenosine receptor subtypes, it is unclear as to the cellular localization or the importance of a specific receptor-effector when the intact heart is used as the model. Because the target cells protected by preconditioning are the cardiac myocytes, it has been hypothesized that the salutary effect of preconditioning is exerted, at least in part, at the level of cardiac myocytes in the intact heart. The strongest support for this hypothesis comes from studies in which cardiac myocytes can be directly preconditioned by brief simulated ischemia or adenosine. A number of cardiac myocyte models have been characterized to investigate the intracellular signaling pathways, from receptors to effectors, that mediate the preconditioning effect.

The advantages of the cardiac myocyte model include the relative homogeneity of the cells, the precise titration of the concentrations of receptor agonists and antagonists in the media in which the cells are incubated, and the ability to selectively manipulate the cells with pharmacological agents during ischemia, as well as to genetically manipulate the cells through transduction of cDNAs, antisense deoxynucleotides, or RNA. The disadvantages include a number of uncertainties such as whether the receptor and the cellular signaling machinery that mediate preconditioning in the isolated myocyte differ from those of the myocytes in intact heart, and whether simulated ischemia and reoxygenation reflect ischemia/reperfusion in vivo.

ADENOSINE A₁ AND A₃ RECEPTORS CAN MEDIATE CARDIOPROTECTION

Recent studies showed that activation of either adenosine A₁ or A₃ receptor can cause a preconditioning-like effect [Armstrong and Ganote, 1994; Strickler et al., 1996; Liu et al., 1994; Lasley and Mentzer, 1992; Lasley et al., 1990; Thornton et al., 1992; Tsuchida et al., 1993; Yao and Gross, 1994a; Auchampach et al., 1997; Thourani et al., 1999a,b; Takano et al., 1999; Carr et al., 1997]. It is not unexpected that the first identified and the most well-characterized adenosine receptor in the myocardium, the A₁ subtype, is capable of mediating such protection. The cardioprotective effect of A₃ receptor activation suggests that this receptor may also mediate the protective action of adenosine. Recent evidence indicates that activation of the A₃ receptor can initiate a preconditioning-like effect in isolated cardiac myocytes from chick [Liang and Jacobson, 1998; Strickler et al., 1996], rat [Shneyvays et al., 2000], and rabbit [Armstrong and Ganote, 1994], in isolated rat [Thourani et al., 1999a,b] and rabbit [Hill et al., 1998; Tracey et al., 1997; Auchampach et al., 1997] hearts, and in human myocardial trabeculae [Carr et al., 1997]. Evidence is emerging that cardiac myocytes express the A₃ receptor and that its activation can directly precondition the cardiac myocytes. It is likely that the myocyte A₃ receptor plays an important role in mediating the cardioprotective response in the intact heart.

A major question relates to the apparent redundancy of the cardioprotection by both A₁ and A₃ receptors. One possible answer to this question is that the relative importance of A₁ and A₃ receptors in triggering the protective effect of preconditioning ischemia is species-dependent. In support of this notion, a recent study shows that while the A₃ receptor can exert a potent preconditioning-like effect in isolated rabbit heart, the A₁ receptor appears to mediate largely the initiation of protection during preconditioning ischemia [Hill et al., 1998]. The same A₃ receptor agonist may have different cardiovascular hemodynamic effects in rat, rabbit, and pig [Lasley et al., 1999]. Another possibility is that A₃ receptor appears to mediate a more sustained cardioprotective response than does the A₁ receptor [Liang and Jacobson, 1998]. The human adenosine A₃ receptor cDNA can mediate such a sustained protective effect following its transfection in the chick atrial myocyte, which lacks an endogenous A₃ receptor. Further, in the chick cardiac myocyte model the A₁ and A₃ receptors can interact in a synergistic manner to achieve a protective response greater than that induced by stimulation of either receptor alone [Olafsson et al., 1987]. Such synergistic interaction may provide the basis for novel therapeutic development.

SIGNAL TRANSDUCTION MECHANISM: FROM ADENOSINE RECEPTOR ACTIVATION TO CARDIOPROTECTION

Studies using the intact heart model clearly demonstrated an important role of the adenosine receptor [Downey, 1992; Liu et al., 1994; Lasley and Mentzer, 1992; Lasley et al., 1990; Thornton et al., 1992; Tsuchida et al., 1993; Yao and Gross, 1994b; Auchampach et al., 1997; Thourani et al., 1999a,b; Takano et al., 1999; Hu et al., 1998], PKC [Baxter et al., 1995; Cohen et al., 1995; Mitchell et al., 1995; Ytrhus et al., 1994; Ping et al., 1999a,b; Dorn et al., 1996], and K_ATP channel [Gross, 1992; Grover et al., 1992] in mediating the cardioprotective effect of preconditioning. Although the exact temporal and vectorial relationship between the adenosine receptors, PKC and K_ATP channels remain to be elucidated, the protective role of these signaling molecules have been demonstrated in a number of cardiac myocyte models of preconditioning [see Marber, 2000; Gray et al., 1997]. Recent studies also implicate a role of some of the isoforms of mitogen activated protein kinase (MAPK) in protecting against ischemia in intact heart [Ping et al., 1999a,b; Haq et al., 1998; Nakano et al., 2000]. In isolated cardiac myocytes, the p38 isoform is activated during lethal hypoxia and its inhibition by SB203580 results in reduced injury [Mackay and Mochly-Rosen, 1998; Saurin et al., 1999]. The molecular mechanisms underlying the activation of adenosine receptors and the subsequent stimulation of downstream effectors such as PKC, K_ATP channel, and MAPK during preconditioning remain an area of intense interest. Current data suggest that adenosine released during ischemia activates its receptors, which in turn stimulate PKC. PKC then interacts with the K_ATP channel and some of the MAPK. It is unclear whether the K_ATP channel and MAPK represent divergent effectors downstream of PKC or whether the channel and MAPKs represent signaling components in the same pathway leading to cardioprotection.

Transmembrane Signaling Mediated by Adenosine A₁ and A₃ Receptors

The cellular and molecular mechanism by which adenosine receptors mediate PKC and the K_ATP channel activation is not well understood. Attempts to understand such mechanisms have been facilitated by the development of a cardiac myocyte model [Linden and Jacobson, 1998; Strickler et al., 1996; Liang and Jacobson, 1998; Olafsson et al., 1987]. The model, based on cultured embryonic chick cardiac myocytes, recapitulates all the basic characteristics of adenosine-elicited preconditioning in the intact heart. Data obtained with the myocyte model show that the adenosine A₁ receptor is coupled via phospholipase C to produce an increased diacylglycerol level, which in turn activates PKC. The A₃ receptor is coupled via phospholipase D to stimulate DAG accumulation and PKC activation. The PLC-mediated elevation of DAG has a short duration and correlates with a more abbreviated protective effect of A₁ receptor agonists. The PLD-derived DAG, in response to A₃ receptor agonists, is more sustained and correlates with a more prolonged preconditioning effect of an A₃ receptor agonist. A PLC inhibitor blocks only the A₁ receptor-mediated protection, whereas a PLD inhibitor selectively abolishes the protective effect of an A₃ agonist. Further evidence for the cardioprotective role of the A₃ receptor-PLD linkage comes from studies on atrial cardiac cells. Atrial cardiac cells express little or no endogenous A₃ receptor and exhibit an abbreviated cardioprotective effect. Transfection of atrial cells with the human adenosine A₃ receptor cDNA results in acquisition of an A₃ receptor-mediated PLD response and the appearance of a more sustained protection. These effects are blocked by PLD inhibitors.

Role of K_ATP Channel

An increasing body of literature, based on selective K_ATP channel openers and inhibitors, suggests that the K_ATP channel acts downstream of the adenosine receptors in mediating the cardioprotective effect of preconditioning. The channel may be involved in initiating, as well as in actually exerting, the protective effect of preconditioning.

Evidence is accumulating to support the hypothesis that the K_ATP channel involved is the mitoK_ATP channel rather then the sarcolemmal K_ATP channel [Garlid et al., 1997; Liu et al., 1998]. Thus, there is little correlation between cardioprotective effect and the action potential duration-shortening effect of the K_ATP channel openers bimakalim and cromakalim. The mitoK_ATP channel-selective opener diazoxide [Hu et al., 1998] is a potent cardio-protective agent [Garlid et al., 1997]. On the other hand, 5-hydroxydecanoic acid, a mitoK_ATP channel-selective inhibitor [Hu et al., 1998], completely abolished the cardioprotective effect of diazoxide [Garlid et al., 1997]. The development of sarcolemmal K_ATP channel-selective opener and inhibitor and demonstration of a lack of their effects on cardioprotection should provide further evidence for this hypothesis.

The current data suggest that PKC may act upstream of the K_ATP channel in mediating the cardio-protective effect of preconditioning. However, it has been suggested the K_ATP channel can also contribute to PKC activation in a positive feedback loop via the release of reactive oxygen species (ROS) [Finkel, 1998; Zhang and Yao, 2000].

PKC and MAPK

PKC is known to activate p44/p42 MAPK, which is suggested to mediate the protective effect of PKC activation in adult rabbit cardiac myocytes and in conscious rabbits [Ping et al., 1999a,b]. The role of the other two MAPKs, the p38 and the JNK isoforms, in ischemia and cardioprotection is unclear. While the p38a isoform is suggested to contribute to the extent of ischemia-induced damage [Mackay and Mochly-Rosen, 1998; Saurin et al., 1999], MAPKAPK2, a kinase downstream of the p38-MAPK, is activated in heart preconditioned by prior ischemia, by the nonselective adenosine agonist R-PIA or by the p38/JNK activator anisomycin [Gray et al., 1997]. An inhibitor of p38 and JNK isoforms of MAPK, SB203580, blocks the activation of MAPKAPK2 in rabbit heart preconditioned by prior exposure to ischemia. These data are consistent with a potential protective role of the p38 and/or JNK in ischemic preconditioning. However, it is unknown whether the p38 inhibitor can block the preconditioning effect.

Mediators of Second Window of Cardioprotection

The signaling molecules downstream of the adenosine receptor appear to act in a posttranslational manner, likely via phosphorylation, to initiate and mediate the protective effect of classical or early preconditioning. In late preconditioning, new protein synthesis is required [Rizvi et al., 1999] and appears to be mediated via nitric oxidedependent activation of NF-kappa B. PKC, ROS, and tyrosine kinase are also implicated [Xuan et al., 1999; Bolli et al., 1998]. Oxidative stress can increase adenosine A₁ receptor expression by activating NF-kappa B [Bolli et al., 1998]. This may provide a positive feedback mechanism to augment the protective effect of adenosine. It is not unexpected that both types of preconditioning share similar downstream signaling molecules. Activation of either adenosine A₁ or A₃ receptors can elicit both types of preconditioning effects [Takano et al., 1999]. The link between adenosine receptors and nitric oxide synthesis and the vectorial/temporal relationship among the various downstream signaling molecules remains incompletely understood. This important area deserves further study.