Cells of the cardiovascular system generate and release purine nucleoside adenosine in increasing quantities when constituent cells are “stressed” or subjected to injurious stimuli. This increased adenosine can interact with surface receptors in myocardial, vascular, fibroblast, and inflammatory cells to modulate cellular function and phenotype. Additionally, adenosine is rapidly reincorporated back into 5′-AMP to maintain the adenine nucleotide pool. Via these receptor-dependent and independent (metabolic) paths, adenosine can substantially modify the acute response to ischemic insult, in addition to generating a more sustained ischemia-tolerant phenotype (preconditioning). However, the molecular basis for acute adenosinergic cardioprotection remains incompletely understood and may well differ from more widely studied preconditioning. Here we review current knowledge and some controversies regarding acute cardioprotection via adenosine and adenosine receptor activation.
- adenosine receptor activation
- reperfusion injury
the heart possesses its own intrinsic protective responses, including the adenosine receptor system, which enhance resistance to ischemic insult. An understanding of the mechanisms involved in these responses not only informs us of how the heart reacts to injurious stimuli, but may provide avenues for developing novel protective strategies applicable in the setting of ischemia-reperfusion. The purine nucleoside adenosine was attributed with cardioregulatory functions almost 75 years ago (64), and since then has emerged as a crucially important control substance in essentially every tissue of the body. In the heart adenosine not only plays a role in regulating growth and differentiation, angiogenesis, coronary blood flow, cardiac conduction and heart rate, substrate metabolism, and sensitivity to adrenergic stimulation (23, 67, 68, 73, 74, 260, 271, 283), but may play a role as an endogenous determinant of ischemic tolerance (189, 225, 245, 259, 311, 312, 318). From a therapeutic viewpoint, adenosine-based therapies protect against ischemic injury in a variety of animal models (72, 177, 208, 265, 288) and in human cardiac tissue (34, 36, 37, 240, 296). However, much remains unclear regarding the roles and mechanisms of action of adenosine in producing acute protection against ischemia and reperfusion.
Generation of Endogenous Adenosine During Insult
Endogenous adenosine levels increase rapidly with ischemic insult (104, 109, 285, 286) to mediate a retaliatory response. This protective pool of adenosine can be formed via dephosphorylation of 5′-AMP by intra- and extracellular 5′-nucleotidases and from S-adenosylhomocysteine (SAH) via SAH hydrolase. Extracellular adenosine is rapidly taken into cells via a facilitative transporter (131). Within cells it is either deaminated by adenosine deaminase or rephosphorylated to 5′-AMP via adenosine kinase. Owing to the relatively high “activities” of adenosine transport and catabolism, the compound has a short biological half-life, and its physiological actions are highly localized. Thus adenosine is considered a regulatory autacoid, with local formation responding to a variety of physiological or pathological stressors.
In terms of cellular sources of adenosine, the majority originates in cardiomyocytes. However, early work by Schrader and Gerlach (252) revealed alternate cardiovascular formation sites. It became apparent that the endothelium was not only a highly active metabolic barrier to adenosine but could also generate adenosine. Subsequent investigations, based on measures of release from prelabeled endothelial pools, revealed an endothelial contribution of 15–25% to basal vascular adenosine levels (21, 154). This proportion declines during stress (ischemia, hypoxia, catecholamine stimulation, and acidosis) (19, 55). Thus myocardial release is of primary importance with ischemic insult.
Studies documenting elevations in adenosine formation with reduced O2 delivery (23, 242) are consistent with the idea that myocardial hypoxia is a prerequisite for stimulated adenosine formation (56). However, it has also been shown that elevations in O2 consumption under normoxic conditions can markedly enhance adenosine formation (18, 25, 106), leading to the concept that formation is triggered by an imbalance between myocardial energy supply and demand (reflected by the so-called O2 supply-to-demand ratio) (18, 106). Adenosine formation is enhanced when reductions in ATP supply surpasses the ability of adaptive responses to concomitantly reduce ATP consumption. With net ATP degradation, ADP and 5′-AMP increase, the latter providing substrate for adenosine formation via 5′-nucleotidase. From the stoichiometry and equilibrium of the myokinase reaction, [5′-AMP] should increase as a function of free [ADP]2, providing some level of metabolic “amplification.” Consistent with the substrate-dependent control of adenosine formation, numerous studies document correlations between free cytosolic [5′-AMP] and myocardial adenosine levels (103–107, 116, 144). However, adenosine formation generally increases much more than 5′-AMP (48, 90, 107, 116), indicating formation via additional paths, allosteric activation of 5′-nucleotidase (e.g., by Mg2+ and ADP) (105), and/or inhibition of adenosine catabolism.
In terms of alternate sources, extracellular formation via ectonucleotidases on myocardial and vascular cells contribute to adenosine formation (88, 89). Additionally, cardiac fibroblasts may contribute to extracellular formation via the so-called “cAMP-adenosine pathway,” involving release of cAMP into the extracellular compartment with subsequent conversion to 5′-AMP by ectophosphodiesterase and hydrolysis of AMP by ecto-5′-nucleotidase (66). Adenosine generated via this latter path may be involved regulating cardiac remodeling. However, studies employing inhibitors of ectonucleotidases reveal that extracellular pathways contribute ∼10% to the overall myocardial formation under baseline conditions, and this proportion declines during periods of metabolic stress (28, 114, 236), as does intracellular endothelial formation (19, 55). Thus ectoenzymatic formation is not a key player. Nonetheless, the importance of this pool should not be underestimated because “site-specific” formation at the surface of vascular cells could specifically protect against injury within this compartment (e.g., modification of inflammatory processes). This is consistent with the findings of Vinten-Johansen and colleagues (134, 277, 278) regarding postischemic adenosine-mediated protection within the vascular compartment. The transmethylation (SAH) pathway also contributes relatively little to adenosine formation. Estimates in the guinea pig and dog place rates at 10–25% of total formation under baseline conditions (55, 180), and this relatively small proportion also declines with stress. Thus neither extracellular formation nor formation via SAH can explain disproportionate elevations in adenosine relative to [5′-AMP] during ischemia or hypoxia.
There is some support for allosteric activation of 5′-nucleotidase because cytosolic isoforms have been identified possessing appropriate regulatory properties: activation by ADP and Mg2+ and inhibition by ATP (45). Changes in these modulators during ischemia-hypoxia may be sufficient to amplify the effects of the increased substrate (105), contributing to nonlinear relationships between adenosine generation and [5′-AMP]. However, more compelling evidence demonstrates that modulation of adenosine salvage via adenosine kinase is the primary mechanism by which formation is markedly enhanced during periods of deenergization.
In 1978, Arch and Newsholme (3) first addressed the potential physiological importance of the cycle between adenosine and 5′-AMP. However, at that time it was thought adenosine kinase was saturated by high intracellular adenosine levels. Subsequently, it was found that >90% of intracellular adenosine is bound to proteins such as SAH hydrolase and myosin ATPase, ensuring very low cytosolic concentrations under resting conditions, and a net inward concentration gradient (57) reversed during ischemia or hypoxia. Work by Decking et al. (49) indicates that the flux through adenosine kinase is ∼80% under baseline conditions and falls dramatically to only 35% during hypoxia. This corresponds to >90% inhibition of enzyme activity. Studies in the rat (32) and mouse (227) verify that the majority of adenosine is normally cycled back to 5′-AMP and that salvage is markedly inhibited by ischemia. Reduced flux through adenosine kinase has also been reported in deenergized myocytes (295). The inhibition of adenosine kinase has been suggested to occur via accumulated Pi (90) and substrate and product inhibition by adenosine (205) and ADP (220), respectively. Despite the suggested inhibitory role of Pi, there is evidence that Pi actually relieves substrate inhibition and enhances activity at high substrate levels existing during ischemia (190). Thus the precise nature of ischemic inhibition is not clear. Potential allosteric effects may be facilitated by the action of adenosine at A1 receptors, initiating PKC-dependent inhibition of adenosine kinase (261). A positive feedback cycle may therefore exist by which receptor activation enhances adenosine generation, amplifying associated protection. On the other hand, there is evidence for adenosine-mediated inhibition of 5′-nucleotidase activity (142). Precisely how these opposing actions might ultimately modulate adenosine formation during ischemia has not been directly addressed. One can speculate that A1 receptor-mediated inhibition of kinase activity might play a role in adenosine-mediated protection in a manner similar to that thought to arise from enhanced 5′-nucleotidase activity with preconditioning (143). Kitakaze and colleagues (143) propose that enhanced adenosine formation as a result of a PKC-dependent activation of 5′-nucleotidase is a key element in observed protection. Interestingly, with prolonged ischemic or hypoxic insult, adenosine formation via 5′-nucleotidase may be impaired by the accumulation of H+, preventing excess purine washout and thereby enhancing postischemic repletion of the nucleotide pool (15, 16, 95). This pattern of changes coincides with the phasic release measured during prolonged ischemia or hypoxia (104).
Whereas the above discussion focuses on formation of adenosine via constituent cells of the heart, noncardiac adenosine formation via inflammatory cells may represent a source of adenosine relevant to cardioprotection. Additionally, the endothelial release of adenosine discussed above may be selectively manipulated by inflammatory elements. There is now considerable support for primarily ectoenzymatic formation of adenosine on the surface of neutrophils, mast cells, and activated endothelium during inflammatory responses (169, 177, 211, 262). This manner of highly localized adenosine release may be of particular relevance in reperfused hearts when inflammation is activated. The data of Narravula et al. (211) reveal a paracrine loop whereby adenosine release and subsequent A2B receptor agonism upregulates endothelial ecto-5′-nucleotidase expression and activity. Locally, generation of adenosine in response to inflammation modifies the inflammatory process (169, 177, 211, 216, 262) and potentially limits resultant cardiovascular injury and cell death (177, 288, 310, 313).
The issue of adenosine formation in relation to protective responses to endogenous adenosine is of interest. Sufficient endogenous generation will maximally activate receptors, whereas reductions in release (e.g., with inhibition of 5′-nucleotidase during prolonged insult) may limit receptor activation and permit responses to exogenous agonist. Presumably, if 5-nucleotidase activation enhances ischemic tolerance (143), receptor activation must normally be submaximal, in agreement with the protection with adenosine agonists. Extracellular adenosine increases up to an order of magnitude within minutes of ischemia, and whereas it is problematic to estimate relevant interstitial adenosine concentrations, a variety of measures suggest they are more than sufficient to substantially activate cardiovascular receptors (104, 109, 115, 285, 286). This is pharmacologically validated by the effects of adenosine receptor antagonists in different models (225, 245, 259, 311, 312), keeping in mind the potential for adenosine formation to overcome effects of applied antagonism via “opening” the regulatory feedback loop between stimulus (deenergization) and signal (adenosine formation and protection) (111, 123). Collectively, studies with both agonists and antagonists support substantial yet submaximal receptor activation by endogenously generated adenosine during ischemiareperfusion (see Adenosine Receptor-Mediated Cardioprotection: Which Subtypes Are Involved?). The varied pathways potentially involved in regulating adenosine generation in the heart are depicted in Fig. 1.
Adenosine Receptor-Mediated Cardioprotection: Which Subtypes Are Involved?
Currently four adenosine receptor subtypes encoded by distinct genes have been characterized: A1,A2A,A2B, and A3 receptors (81, 177, 219, 283). The A1, A2A, and A3 receptors have high affinities for adenosine, whereas the A2B has a relatively low affinity. All are G-coupled receptors (145) and appear to be expressed within cells of the cardiovascular system, although direct and unequivocal evidence of expression of some subtypes is lacking. Cardioprotective actions have been associated with all subtypes, although the precise roles of each in protecting against ischemia-reperfusion remains controversial. It is worth noting here that much of the controversy regarding receptor identities stems from two chief problems: 1) many receptor agonists and antagonists previously thought to be highly selective are in fact capable of interacting with multiple receptor subtypes at concentrations typically employed; and 2) this problem is compounded by significant species differences in the pharmacology of adenosine receptors. These problems have been particularly evident in identifying the function of the A3 receptor but also apply to the other subtypes in different models and species.
Much earlier work supported a central role of the A1 adenosine receptor (A1AR) in cardioprotection (159–161, 164), in part, because of the lack of knowledge at that time of the A3 adenosine receptor (A3AR). Although these pioneering studies may be clouded by the lack of selectivity of then-available agents [e.g., R-phenylisopropyladenosine (R-PIA), 8-sulfophenyltheopylline, and BW-1433U], they nonetheless established that the A1AR is of major importance in mediating anti-ischemic actions of adenosine. Numerous subsequent investigations support A1AR agonist-mediated protection (75, 80, 136, 181, 182, 207, 210, 267). Expression of A1ARs also confers ischemic tolerance in different models (64, 197), and a relatively small number of studies document impaired tolerance with A1AR antagonists (225, 312), supporting protection via endogenous adenosine. Signaling via the A1AR has been shown to provide protection not only in animal tissues but in the human myocardium (36, 240).
Confounding findings do exist with respect to A1AR-mediated cardioprotection. For example, protection with A1AR agonists is not always observed (197, 225, 312) and, as already noted, only a limited number of studies demonstrate exacerbation of ischemic injury with A1AR antagonism. Lasley and colleagues (162) have forwarded an intriguing hypothesis to explain some of the inconsistencies in A1AR-mediated responses. They observed translocation of A1ARs from caveolae upon agonist stimulation (162), a form of compartmentalization that may explain lack of direct effects of A1AR agonism under certain conditions. From the effects of ischemia and endogenous adenosine on caveolae translocation, this process could regulate A1AR responsiveness before and during ischemia. This possibility awaits further testing.
The A2 adenosine receptors are classified as subtypes A and B (A2AARs and A2BARs, respectively). These receptors were thought to primarily exist in vascular and blood cells, where they mediated vasodilatory and anti-inflammatory actions, respectively. However, expression of mRNA encoding the A2AAR in rat ventricular myocytes (303), together with A2AAR agonist-mediated changes in cAMP and contraction of chick (172) and rat myocytes (56, 239, 303), support the expression of A2AARs on myocytes. This indirect evidence of myocardial expression is supported by a single report of immunological evidence of a canine A2AAR-like protein in human and porcine ventricular tissue (199).
Irrespective of cellular location, a number of studies support A2AR-mediated cardioprotection during ischemia-reperfusion. In particular, the work of Vinten-Johansen and colleagues (277, 278, 288, 311, 313) has revealed A2AR-dependent protection mediated primarily postischemia and involving modulation of inflammatory processes and vascular function. A more recent study from this group supports A2AAR-mediated protection against myocardial apoptosis (310). Another study has revealed functionally beneficial effects of A2AAR agonism in postischemic dog hearts in situ (158). However, the authors suggest the response may reflect positive inotropic effects of A2AAR-mediated elevations in adenosine 3′,5′-cyclic monophosphate (because PDE inhibition and elevations in cAMP exert positive inotropic actions) as opposed to a direct protective action. Two other investigations in isolated perfused hearts do provide support for direct A2AAR-mediated protection in the rat (182) and rabbit (35) myocardium. These latter studies contrast others that find no protection with A2 agonism in isolated hearts (75, 224, 317). Moreover, A2AAR activation fails to protect isolated cardiomyocytes (24). A key issue relating to the two in vitro studies supporting A2AAR protection is that hearts were perfused in a constant-flow mode (35, 182). Recent work from our laboratory indicates that reduced coronary perfusion pressure during A2AAR agonism modifies myocardial contractility (224), rendering interpretation of functional changes in such models problematic.
Whereas a number of studies support A2AAR-mediated protection in vivo, there is little evidence for acute A2BAR-mediated cardioprotection, partially due to the lack of selective and potent A2BAR agonists/antagonists. Given the low affinity of the A2BAR, it is probable that the protein is only activated by pathophysiological elevations in adenosine. However, it is interesting to note A2BARs expressed in Chinese hamster ovary cells have an EC50 for mitogen-activated protein kinase (MAPK) activation of only 25 nM, in contrast to the low potency for endogenous receptors in HEK-293 cells (254). This suggests that the potency may be substantially modified by relative levels of receptor and G-protein expression. In any case, evidence that the A2BAR is expressed in cardiovascular tissue remains indirect, and signaling involved in A2BAR responses in cardiac tissue remains to be delineated. There is evidence, however, that A2BARs may regulate postischemic remodeling via inhibition of cardiac fibroblast growth (67), providing a level of chronic protection.
The A3AR was relatively recently identified and has been cloned in a variety of species, including humans (175). Shortly after its identification, newly synthesized A3AR agonists were shown to exert beneficial actions in ischemic-reperfused myocardium (4, 9, 267, 281), including human tissue (36). More recent studies verify cardioprotection via A3AR agonism (98, 133, 173, 224, 276, 282). However, key issues relate to the cellular location of the receptor and the specificity of pharmacological agents employed. In ascribing cardioprotective functions to the A3AR, it is important to recognize that the A3AR protein itself has to date only been identified in human eosinophils (150) and a rat basophil leukemia cell line (218), despite detection of A3AR gene transcript in a variety of tissues (61). One explanation for A3AR-dependent cardioprotection is inhibition of resident mast cell degranulation. However, even this is questionable because the mast cell response may actually be mediated via A2BARs in humans and in dogs (10). On the other hand, the degranulation response is refractory to A3AR agonists in A3AR knockout mice (247), supporting a role for the receptor in this species. In terms of agonists and antagonists, there is some question regarding specificity, particularly between species (126, 176, 177). Agents thought to possess high specificity may interact significantly with other subtypes. As discussed by Hill and colleagues (126), this may stem in part from the greater diversity of A3 receptors in different species compared with the greater homology observed between A1 receptors. Highlighting the issue, Kilpatrick et al. (140) recently acquired evidence that cardioprotection via A3AR agonism was abrogated by selective A1AR antagonism. These findings, however, are equivocal because effects of A1AR antagonism alone were not assessed, and this has been shown to impair ischemic tolerance in other models (225, 312). Recent work has verified A3AR agonist-mediated protection is unrelated to A1AR activation in the rabbit (149).
Given issues regarding selectivity at the A3AR, transgenic or knockout approaches might appear promising in identifying roles of this receptor. In this respect, increased expression of A3ARs in myocytes and intact hearts does generate tolerance to ischemichypoxic stress (26, 43, 64). However, because myocardial expression in wild-type tissue has yet to be verified, one must interpret these studies cautiously. A paradoxical ischemia-tolerant phenotype has been observed in A3AR knockout mice, supporting a detrimental rather than protective function of A3ARs (94, 98). Unfortunately, knockout models should be considered with even more caution than transgenic models, because phenotypic results of the life-long absence of a protein do not simply reflect the effect of acute deletion or inhibition of the protein or response. As discussed recently (98), this phenotype may reflect the impact of numerous “compensatory” mechanisms and/or result from expression of associated background or “hitchhiker” genes. Indeed, studies of compensatory processes in knockout animals may prove to be a more fruitful avenue of research in these models.
Endogenous Versus Exogenous (Pharmacological) Cardioprotection
The vast majority of work assessing adenosinergic cardioprotection focuses on exogenous or “pharmacological” activation of the adenosine receptor system as opposed to the protective responses to endogenously generated adenosine. However, a number of studies verify that endogenously generated adenosine does enhance intrinsic ischemic tolerance in different species and models (225, 245, 258, 311, 312). These studies generally support cardioprotective roles for A1 and A2 receptors, whereas intrinsically activated A3 receptors do not appear to modify ischemic tolerance. Indeed, this is consistent with the data of Hill and colleagues (127) who demonstrated in rabbits that the A1AR is more than an order of magnitude more sensitive to adenosine than the A3AR. Thus the A1AR is more likely to be intrinsically (and substantially) activated during ischemia-reperfusion than the A3. The protective function of intrinsically activated A1ARs has been verified by a small number of studies employing A1-selective antagonists (225, 312, 259) or antagonists displaying greater selectivity for A1 versus A3 receptors (245). In contrast, selective A3 antagonism appears to have little effect on intrinsic tolerance to ischemia or hypoxia in isolated myocytes (173, 267) or hearts (188). Nonetheless, overexpression of the A3 has been shown to enhance intrinsic ischemic tolerance (26, 43), as does A1AR overexpression (197). Whereas this implies an intrinsic role in cardioprotection, functional amplification due to markedly increased numbers of effector-coupled receptors may exaggerate sensitivity to adenosine and therefore also the role of the intrinsically activated receptor in wild-type hearts. There is also some support for intrinsic A2 receptor-mediated protection via endogenous adenosine. For example, A2 antagonism limits functional outcome and exacerbates platelet adhesion in vivo (259), and mixed A1/A2 antagonism is also shown to be much more effective than selective A1AR antagonism in worsening ischemic outcomes in vivo, implicating a role for A2 receptors (312). Effects of A2 antagonists are restricted to in situ models consistent with A2 receptors modulating inflammatory processes and platelet aggregation, requiring the presence of blood-borne elements. As already noted A2AAR activation fails to protect isolated cardiomyocytes (24) or blood-free perfused hearts (75, 224, 317).
Collectively, studies employing adenosine antagonists provide support for a role for endogenously generated adenosine in determining intrinsic ischemic tolerance. These studies, while limited in number, tend to support a primary role for A1ARs, with little contribution from the A3AR. Whereas even less information is available regarding intrinsically activated A2 receptors, there is support for a role in determining ischemic tolerance, potentially mediated via control of inflammation and vascular interactions.
Molecular Basis of Receptor-Mediated Cardioprotection
The A1 and A3ARs couple to pertussis-sensitive Gi and Go family proteins. In cardiac tissue the A1AR inhibits adenylyl cyclase activity, enhances an inwardly rectifying K+ current in atrial and nodal cells (22), is coupled to phospholipase activation and membrane-associated PKC in ventricular myocytes (124), activates sarcolemmal ATP-sensitive K+ channels (141), and can also modify the activity of mitochondrial ATP-sensitive K+ channels (mitoKATP) (207). The A3AR has been shown to couple to a number of G proteins, including Giα2, Giα3, and Gqα (222), and also apparently modifies mitoKATP function (276, 282). The A2A and A2BARs are thought to activate adenylyl cyclase via Gs protein, with the A2BAR additionally stimulating phospholipase C (PLC) via Gq protein (73). Through these and related interactions the adenosine receptor family modifies a broad range of signaling pathways (145). Identifying the paths involved in acute cardioprotection has been the focus of a large number of recent investigations. Here we consider several of the molecular pathways implicated [e.g., protein kinase cascades, mitoKATP channels, phosphatidylinositol-3 (PI3)-kinase, nitric oxide (NO) synthase, and Na+/H+ exchange (NHE) inhibition], and specific cellular responses (vascular protection, modulation of neutrophil function) are addressed in subsequent sections. A synthesis of varied pathways implicated in acute adenosinergic cardioprotection is provided in Fig. 2.
Phospholipase C and D signaling in A1- and A3AR-mediated protection. Studies from Liang and colleagues (167, 223) demonstrate (in chick myocytes) that A1AR- and A3AR-mediated protection involve unique signaling pathways. The A1AR-mediated response was shown to involve Gi-dependent modulation of PLC activity, whereas the A3AR response involves selective RhoA-dependent activation of phospholipase D (PLD). These different coupling mechanisms may contribute to differing temporal profiles of A1AR- and A3AR-mediated protection, with the PLC-dependent path providing a short-lived response, whereas the RhoA/PLC path generates longer-lasting protection (167, 223). In both responses, there is evidence the paths converge on common mediators, including mitoKATP channels and PKC.
PKC and mitoKATP channels. The chief mechanistic pathway recently associated with adenosinergic protection involves activation and translocation of PKC and mitoKATP channel activation, although the order of these events remains controversial. These elements are implicated in both acute protection with adenosine receptor activation (113, 207, 210, 228, 275, 282) and adenosinergic preconditioning (226, 248, 280). Though earlier studies implicated sarcolemmal KATP channels, subsequent evidence favors a role for the mitoKATP channel in these responses (93, 248). Even this scheme remains controversial because recent studies question the role of mitoKATP channels in cardioprotection (see below), and there remains support for some role for sarcolemmal KATP channels in adenosine-mediated cardioprotection (175, 280).
In the conventional signaling cascade, PKC activation/translocation is thought to occur proximal to mitoKATP channels, although again much of this evidence stems from studies of preconditioning in which inhibitors of PKC fail to modulate protection with mitoKATP channel openers (e.g., 207). An upstream effect of PKC on mitoKATP channels is supported by studies of sarcolemmal channels, which are activated via PKC-mediated phosphorylation of a channel domain in the pore-forming unit (the PKC consensus site is conserved in different channels), limiting the inhibition of channel opening by ATP (174). Moreover, protective responses to PKC activators can be inhibited by mitoKATP channel blockers (304). In contrast, there is evidence that inhibition of PKC can abrogate protection with mitoKATP channel openers (228, 293, 294). These observations place PKC distal to the channels. The order of these events is a nontrivial topic of considerable debate, particularly within the preconditioning field.
In addition to confusion regarding upstream or downstream locations of PKC and mitoKATP channels, there are conflicting findings regarding A1AR-mediated activation versus inhibition of myocardial PKC isoforms (5, 82, 124). There is even evidence that PKC and MAPK inhibitors protect rather than worsen injury in ischemic hearts (163, 184). Additionally, PKC may represent one of several kinase pathways activated by adenosine receptors, as evidenced by recent studies of myocardial preconditioning. Finally, with respect to studies employing chelerythrine to block PKC, recent data indicate this agent potently inhibits adenosine receptors in concentrations typically employed to block PKC (256). Thus studies documenting impaired adenosinergic protection in the presence of chelerythrine may not reflect a protective contribution from PKC.
Evidence of mitoKATP involvement primarily revolves around the protection with diazoxide (reportedly a mitoKATP opener) and inhibition of the protective responses with 5-hydroxydecanoate (5-HD, a mitoKATP inhibitor) (179). There is now an emerging view, however, that these agents modify ischemic tolerance via nonspecific actions unrelated to mitoKATP channel activity (70, 97). Furthermore, a recent study by Das et al. (47) argues against the existence of diazoxide/5-HD-sensitive mitoKATP channels in the rat. Protection with diazoxide may stem from inhibition of adenine nucleotide-requiring enzymes such as mitochondrial succinate dehydrogenase and cellular ATPases (70) and/or modulation of reactive oxygen species (ROS) generation and nucleotide degradation by inhibition of mitochondrial respiration (97). Inhibitory effects of 5-HD may reflect antagonism of these actions and/or activation to 5-HD-CoA with modulation of fatty acid oxidation and mitochondrial metabolism (97). Given the common effects of adenosine and diazoxide (77, 207, 210, 228) and inhibition of adenosinergic protection by 5-HD (113, 118, 207, 210, 228), it is possible acute adenosinergic protection involves the above-mentioned inhibitory actions on ROS generation, mitochondrial metabolism, and nucleotide-dependent enzymatic activity as opposed to mitoKATP activation. In this respect, a number of investigations verify that adenosine receptor activation modifies mitochondrial energy metabolism (2, 43, 79, 112) and inhibits ROS generation (210) in ischemic-reperfused myocardium. The role of mitoKATP channels in adenosinergic protection requires further validation beyond the use of agents such as diazoxide and 5-HD.
Tyrosine kinases, phosphatidylinositol-3-kinase, and Akt-dependent signaling. Whereas the PKC-mitoKATP path has received considerable attention, studies of acute and delayed preconditioning with adenosine and other stimuli have unmasked contributions from alternate or additional kinase cascades (44, 171, 272, 284, 306). Our own recent studies (226) indicate that mitoKATP channels and PKC exert additive effects in adenosine-mediated preconditioning, consistent with parallel signaling pathways. Although the genesis of protection with preconditioning may differ from acute adenosinergic protection, these findings hint at roles for multiple signaling processes in cardioprotection. Considering the importance of intrinsic cardioprotection, it is not surprising that some level of signaling redundancy would exist. If this is the case, treatment with individual and specific inhibitors may fail to identify roles for individual signaling paths in adenosinergic protection.
In terms of alternate kinases potentially involved in acute adenosinergic protection, adenosine receptor activation can activate PI3-kinase through transactivation of tyrosine kinase activity in myocardial tissue (151). A downstream target of PI3-kinase is the serine-threonine kinase Akt (protein kinase B), which has also been implicated in cardioprotection with preconditioning (171). This kinase is modulated by products of PI3-kinase activity (phosphatidylinositol 3,4,5-trisphosphate and phosphatidylinositol 3,4-diphosphate), which trigger 3′-phosphoinositide-dependent kinase-1 and -2-dependent phosphorylation of the serine and threonine residues of Akt. Activation of Akt promotes cell survival via phosphorylation of Bad, permitting dissociation of the Bad/Bcl-Xl complex and binding of the 14-3-3 protein and also by ultimate activation of nuclear factor (NF)-κB (Fig. 2). It has been revealed that A3ARs may specifically trigger phosphorylation of Akt (151). Indeed, A3AR agonism inhibits ultraviolet light-induced apoptotic death in noncardiac cells (84) and protects against endothelial apoptosis (178) via PI3-kinase/Akt-dependent pathways. In contrast, there is evidence the path is not obligatory in A1AR-mediated protection against irreversible injury in the ischemic mouse heart (237) or in adenosine-mediated preconditioning (232). Thus it remains to be established that PI3-kinase-dependent phosphorylation of Akt plays a key role in acute adenosinergic cardioprotection.
MAPK/Erk. The MAPK family has been implicated in cardioprotective responses, including early and delayed preconditioning with adenosine (44, 272, 306) and other stimuli (210). There is, however, limited information regarding the role of MAPKs in acute adenosinergic cardioprotection. There is certainly ample evidence from noncardiac tissues that A1ARs couple to p42/p44 and p38 MAPK signaling (238), the former involving PI-3 kinase and the latter apparently pertussis toxin sensitive. Vascular A2AARs couple to this pathway in regulating NO generation (299). All adenosine receptor subtypes have been shown to mediate Erk phosphorylation in Chinese hamster ovary cells, supporting a role for this path in responses to each receptor (254). A more recent study indicates that A3AR activation of Erk involves Gi/o-coupled modulation of PI3-kinase, Ras, and MAPK kinase (MEK), independently of Ca2+, PKC, and c-Src (255). In cardiac tissue, preliminary reports support a role for the stress-activated protein kinase p38 MAPK in A1AR-mediated protection (132, 157). In terms of the A2 receptors, studies in neuronal PC12 cells suggest that A2AARs mediate protection via elevated cAMP and activation of Rap1a (6, 146). Thus A2AAR-mediated protection against injury and apoptosis (309, 313) could involve this protective path, although direct evidence from the cardiac tissue is still lacking. Additionally, there is also evidence from the same cells that A2AR-mediated protection is mediated via reduced cellular [Ca2+] via activation of PKA (148).
NO bioavailability. There is indirect evidence to suggest adenosine may protect via enhancing NO bioavailability. For example, delayed protective effects of adenosine-mediated preconditioning involve changes in NO synthase activity (305). A more acute effect is supported by the observation that NO synthase inhibition abrogates A1AR-mediated actions in rabbit atrioventricular nodal cells (195). Additionally, changes in NO availability parallel changes in adenosine-mediated cardioprotection in young versus aged myocardium (86). However, there is also evidence against a role for NO in adenosinergic cardioprotection (35), and NO may actually worsen ischemic tolerance via an adenosine-dependent process (298). The role of NO in adenosine-mediated acute cardioprotection therefore awaits adequate testing, despite evidence for its involvement in delayed protective responses to adenosine.
Energy and substrate metabolism. As noted above, there is evidence that A1 and A3ARs may enhance ischemic tolerance by modifying energy metabolism. Overexpression of either A1 or A3ARs significantly limits ATP depletion during ischemic insult (43, 112), consistent with improvement of the myocardial energy state and ionic homeostasis by exogenous adenosine (78), and endogenously generated adenosine (2, 196), in ischemic and hypoxic myocardium. Preservation of the energy state and ATP may contribute to improved outcome by limiting H+, Ca2+, and Na+ overload and effects of elevated Pi and depressed ΔGATP on contractile function.
Additional to energy metabolism, the work of Clanachan and colleagues (75, 80) has established a protective function of A1ARs via modulation of substrate metabolism. Specifically, A1AR activation limits glycolysis and improves coupling between glycolytic metabolism and glucose oxidation, thereby reducing the generation of H+ (75, 80). In this respect, we have also found that substituting pyruvate for glucose as exogenous carbon substrate can eliminate protection associated with A1AR overexpression (unpublished findings), supporting a key role for glycolysis upstream of pyruvate in cardioprotection.
NHE inhibition. Inhibition of NHE is one of the most effective means of enhancing myocardial ischemic tolerance. Recent work shows selective A1AR activation inhibits adrenoceptor-stimulated sarcolemmal NHE activity (11), although the effect appears related to inhibition of adrenoceptor activation rather than direct NHE inhibition. However, because A1AR agonism also reduces thrombin-stimulated NHE activity, the response may reflect a broad interaction between A1ARs and G-protein-dependent control of NHE. Studies in renal tissue reveal A1- and A2AR-mediated inhibition of both NHE activity and membrane protein levels (59, 60). Curiously, the more recent of these studies employed high levels of an A1-selective agonist to examine the putative A2AR-mediated response. Because cardioprotective NHE inhibition is mediated by mitoKATP channels (206), because adenosine receptors also protect via mitoKATP channel-dependent (or diazoxide/5-HD-dependent) processes (113, 118, 207, 210, 228, 282), and because adenosine receptor signaling is coupled to NHE modulation in cardiac and noncardiac tissues (11, 59, 60), it seems probable that NHE modulation may be involved in adenosinergic protection. This possibility deserves more direct attention.
Modulation of oxidants. There is some support for the notion that A1AR activation protects via enhancing antioxidant defense (208, 233). Adenosine receptor activation reduces mitochondrial radical formation (210) and oxidant injury (136, 275) and increases cellular antioxidant capacity (189). Moreover, there is evidence that cardioprotection mediated by diazoxide and 5-HD-sensitive processes (including adenosinergic protection) may involve inhibition of ROS generation (97). Given the central role of oxidant damage in both reversible and irreversible injuries, these effects may contribute to the resistant phenotype observed with adenosine receptor activation. Conversely, triggered formation of ROS, via mitoKATP-dependent processes, has been implicated in the protection afforded by preconditioning (13, 208, 265). Thus on the one hand acute adenosinergic cardioprotection may limit ROS generation and injury to mediate protection, whereas signaling involved in adenosinergic preconditioning could involve selective generation of ROS. These paradoxical findings may be reconciled by the recent observation that adenosine-dependent preconditioning activates a unique path distinct from other G-protein-coupled stimuli that does not apparently involve ROS generation (40).
Sarcoplasmic reticulum Ca2+ handling. The data of Zucchi et al. suggest that modulation of sarcoplasmic reticulum (SR) Ca2+ handling may be involved in both A1- (316) and A3AR (317)-mediated cardioprotection. This group has shown that A1/A3AR activation is associated with a reduction in the density of the SR Ca2+ channel (ryanodine receptor), akin to that observed in preconditioned hearts. This may limit the development of Ca2+ overload, protecting from Ca2+-dependent injury. Modulation of SR Ca2+ handling is conceivably an “end effector” sought in recent studies, although how it might be coupled to signaling elements such as mitoKATP channels, PKC, or other protein kinase cascades remains obscure.
Mitochondrial permeability transition pore. In terms of potential end effectors, the mitochondrial permeability transition pore (MPTP) appears to be a good candidate. This nonspecific inner mitochondrial membrane pore causes apoptosis/oncosis on opening (29, 102, 152). Pore opening uncouples mitochondria to collapse the mitochondrial membrane potential and permits entry of H2O and solutes, increasing the mitochondrial matrix volume, and ultimately rupturing the outer membrane. The resultant cytochrome c release initiates apoptosis. The pore opens during ischemia or early reperfusion, likely due to ATP depletion, accumulation of Pi and mitochondrial [Ca2+], and changes in matrix [H+]. Importantly, inhibition of pore opening inhibits cytochrome c release and apoptosis in ischemic myocardium (29). There are currently no data on the MPTP in acute adenosinergic cardioprotection, although the fact of A1AR-mediated preconditioning limits MPTP opening and resultant postischemic cell death (102) supports some role for the pore in adenosine signaling. Moreover, antiapoptotic effects of adenosine receptor activation in noncardiac cells involve inhibition of the MPTP (178). Again, direct assessment of the function of this pore in adenosinergic cardioprotection is warranted.
Are Acute Effects of Adenosine Receptor Activation Entirely Posttranslational?
It is generally considered that acute cardioprotective effects of stimuli, including preconditioning and adenosine receptor activation, involve posttranslational modifications (e.g., protein phosphorylation and KATP channel modulation). However, acute changes in gene expression occur during ischemia-reperfusion (120, 183) and with protective stimuli (46, 51, 198), and there is support for roles for de novo protein synthesis and/or gene transcription in acute protection with preconditioning (199, 241, 268). Several studies provide intriguing, if not compelling, support for a transcriptional component in acute adenosinergic protection (120, 266). Recently, Asakura et al. (7) observed adenosine-dependent changes in gene expression during ischemia, and we have documented substantial transcriptional modification by overexpression of A1ARs in both normoxic (156) and postischemic hearts (8). Although not entirely surprising, many of these changes are consistent with the observed ischemia-tolerant phenotype (e.g., modulation of pro- and antiapoptotic genes and genes for structural and contractile proteins). In more direct analysis, putative inhibitors of transcription and translation can modify acute cardioprotective actions of adenosine (118). Collectively, these studies support a role for gene transcription and/or protein synthesis in the acute effects of adenosine receptor activation. This form of control agrees with adenosine-mediated inhibition of transcriptionally regulated apoptosis postischemia (237, 310), possibly via modulating transcription and expression of Bcl-2 and Bax (310).
Nonreceptor-mediated protective responses to adenosine are evidenced by the inability of receptor antagonism to eliminate protection with adenosine, despite elimination of protection with selective receptor agonists (75, 224). Because adenosine represents a diffusible product of the adenine nucleotide pool, its salvage is important in maintaining or replenishing the pool following injury. Early studies focused on manipulating purine catabolism with the aim of enhancing metabolic tolerance to ischemia (27, 130). Bolling and colleagues (27) first presented the notion of adenosine as “substrate” versus “signal” and provided evidence that metabolic effects might be equally important in mediating ischemic tolerance. Differing purine metabolism was forwarded as a key mechanism underlying enhanced tolerance of neonatal/immature hearts to ischemic or hypoxic stress (220), and this provided credence to the notion that selective manipulation of nucleotide depletion/repletion might enhance tolerance in mature hearts (27, 130).
Recently, studies employing modulators of adenosine salvage reveal that bypassing phosphorylation enhances ischemic tolerance (226, 227). Moreover, protective effects of adenosine are limited by blockade of adenosine kinase, supporting a role for phosphorylation in cardioprotection with adenosine itself (228). Inhibition of adenosine deamination has been known for some time to also enhance ischemic tolerance (58, 130, 200, 315). The latter response has been attributed to reduced generation of xanthine oxidase-derived radicals (301) or enhanced purine salvage (27, 130). However, protective actions of adenosine deaminase inhibition are abrogated by adenosine receptor activation, supporting enhanced receptor agonism as the underlying mechanism (227). Indeed, the notion that inhibition of adenosine deaminase substantially limits ROS generation is flawed because the vast majority of postischemic purine catabolism occurs via inosine 5'-monophosphate versus AMP catabolism (38, 117, 227), the kinetics of the xanthine oxidase reaction indicate the path will be saturated by levels of hypoxanthine-xanthine generated during ischemia (301, 302), and uric acid generated in the same process is an antioxidant. Thus substantial reductions in deamination should have limited effects on xanthine oxidase-derived ROS generation. This is borne out by the lack of effect of adenosine catabolites such as inosine and hypoxanthine on postischemic recovery (13, 227).
It is interesting to consider the relative importance of individual pathways of adenosine metabolism versus receptor activation in modifying ischemic tolerance. As already noted, inhibition of adenosine phosphorylation yields an ischemia-tolerant phenotype (224, 227, 228), indicating enhanced adenosine receptor activation at the expense of phosphorylation exerts net benefit. From the relative effects of adenosine receptor antagonism and adenosine kinase inhibition, the major proportion of protection with adenosine is receptor mediated, with a lesser but still significant contribution from adenosine phosphorylation (228). When deamination is blocked, the net result is a similar degree of cardioprotection as that observed with adenosine kinase blockade (58, 130, 200, 227, 315). However, when both processes are simultaneously inhibited, effectively trapping formed adenosine, postischemic recovery is impaired (227). These findings collectively indicate that whereas a single path of purine salvage is operative (adenosine phosphorylation or incorporation of hypoxanthine into inosine 5'-monophosphate then 5′-AMP), enhanced endogenous adenosine proves protective. When salvage is eliminated, enhanced receptor activation is insufficient to counter the decline in tolerance. This agrees with observations of impaired viability in isolated cells in which both paths are simultaneously blocked for prolonged periods (263). Under these conditions, provision of ribose moiety can improve viability by enhancing nucleotide repletion.
Although benefit from enhanced 5′-AMP is assumed to involve repletion of the nucleotide pool and maintenance of ΔGATP, any process dependent on 5′-AMP, ADP, or ATP levels might be involved. In this respect, an elevation in 5′-AMP will impact on AMP-activated protein kinase (AMPK) activity, and this enzyme may play a role in the adaptive and protective responses to ischemic insult (52, 138, 229). Cardioprotection via AMPK appears related to glucose metabolism (52), as is adenosinergic protection (75, 80), and AMPK stimulates glucose transport and metabolism (244), potentially via modulation of ERK activity. Blockade of adenosine kinase inhibits the effects of activated AMPK on glucose metabolism (244), inhibits ERK activation (78), and reduces the protective effects of adenosine (228), consistent with a role for AMPK and ERK activity. Nonreceptor effects of adenosine may therefore involve both simple substrate effects in addition to modulation of nucleotide-dependent signaling paths.
Protection Against Reversible Versus Irreversible Injuries Mediated During Ischemia Versus Reperfusion?
Injury from ischemia-reperfusion can be broadly classified as reversible (i.e., stunning) or irreversible (oncosis and/or apoptosis). Adenosinergic protection has been clearly established as a mechanism for reducing both oncosis (9, 207, 224, 228, 277, 314) and apoptosis (237, 310), leading to substantial reductions in infarct size and injury. Whether adenosinergic responses protect against reversible injury is more difficult to ascertain because both forms of injury often coexist in the same model. In this situation, a reduction in irreversible injury will impact on conventional measures of stunning (myocardial contractility and arrhythmogenesis). Nonetheless, postischemic contractility is enhanced by exogenous and endogenous adenosinergic stimuli under conditions of limited or no detectable cell death (149, 161, 165, 200, 201, 210, 215, 234, 235, 246, 258, 318). In addition, recent work has identified differential effects of adenosine (77, 228) and adenosinergic preconditioning (226) on oncosis versus contractile function, supporting amelioration of both injuries via different pathways.
The temporal characteristics of adenosinergic protection (ischemic vs. postischemic) against stunning and cell death are also emerging, although contradictory findings exist. Initial studies by Lasley and Mentzer (159, 161, 164) supported the notion that A1AR cardioprotection occurred during the ischemic insult. These early conclusions, largely focussing on the limitation of contractile dysfunction, are supported by more recent studies. For example, overexpression of A1ARs limits diastolic contracture yet appears to exert limited additional benefit postischemia (196), and postischemic receptor activation exerts little if any effect on indexes of reversible and irreversible myocardial damage (91, 245, 312). In contrast, a number of investigators reveal beneficial actions of both exogenous (86, 188) and endogenous adenosine (225, 259, 311) in the postischemic heart.
The explanation for these discrepancies may lie in the type of injury assessed. Contractile dysfunction appears more effectively limited by pre- and intraischemic adenosine treatment, with evidence that stunning is unaltered by postischemic treatment (245). In contrast, irreversible injury (oncosis and apoptosis) is effectively limited by both pre- or postischemic treatment. Additionally, ischemic effects appear mediated by A1 and A3ARs (77, 159, 164, 223–225, 227, 228, 267, 281), whereas postischemic effects are largely ascribed to A2 and A3ARs (133, 188, 309). However, even these patterns are a generalization because there is evidence for no reduction in infarct size with postischemic adenosine (91). Another key factor is the presence of blood. Postischemic protection observed with A2 agonism appears to involve attenuation of neutrophil activity, oxidant stress, and inflammation. However, again there is also evidence that postischemic adenosine protects in blood-free models (75, 86, 225, 317). Additionally, whereas recent studies indicate A3AR activation on reperfusion limits neutrophil-mediated reperfusion injury in vivo (133) and apoptotic death and injury in isolated hearts and cardiomyocytes (188), there is also evidence for minimal postischemic protection in isolated hearts (77).
It is interesting to speculate, based on current evidence, that a “hierarchy” of protection exists in which the A1AR reduces acute injury primarily during ischemia and via processes unrelated to blood cell-mediated injury (e.g., inflammation), whereas A2ARs exert substantial protection during reperfusion (via distinct mechanisms involving limitation of inflammatory processes). Such a scheme is consistent with the majority of (but not all) data. From a clinical viewpoint, the timing of adenosinergic protection is not trivial, because effective treatment of myocardial infarction (for example) commonly requires postischemic intervention. In our own recent work (77), we show the greatest degree of protection is observed with continuous adenosine treatment during both ischemia and early reperfusion.
Importance of Intravascular Versus Interstitial Adenosine
The above discussion alludes to tissue compartmentation of adenosinergic cardioprotection. Whereas it is generally held that cardioprotection involves activation of A1 and possible A3ARs on myocardial cells, there is substantial support for an intravascular action of both exogenous and endogenous adenosine. There is evidence for expression of A1, A2A, A2B, and A3ARs in vascular cells (271), and adenosine receptors modulate function in multiple blood cell types, notably those involved in inflammatory responses (31, 41, 288). The powerful metabolic barrier represented by the endothelium can effectively separate changes in interstitial adenosine from changes in intravascular adenosine (115, 235, 285). A number of studies identify protective responses to relatively low levels of adenosine infusion, which should theoretically fail to modify interstitial adenosine (213, 230, 235, 258, 279). Additionally, protection with preconditioning can be associated with reduced interstitial adenosine and thereby myocardial receptor activation (285). One might therefore conclude that intravascular adenosine is critical in mediating cardioprotection. This is further supported by studies demonstrating reduced infarct size or contractile dys-function with adenosine-regulating agents (155, 314), which fail to modify interstitial adenosine. Large-molecular-weight adenosine agonists restricted to the vascular compartment also elicit cardioprotection (277). These varied observations are all consistent with the proposal of Rubio et al. (17) that intravascular adenosine is crucial in regulating myocardial function. This proposal was recently verified by studies revealing modification of myocardial function via intravascular A1, A2A, and A3ARs, potentially through actions of NO and/or prostaglandins (243). It is clear from such studies that the notion of myocyte receptor activation as the key mediator of adenosinergic cardioprotection requires reassessment.
In terms of intravascular targets, adenosine potently inhibits neutrophil, platelet, and mononuclear leukocyte function. Because neutrophils are proposed as key determinants of injury during ischemia-reperfusion, contributing to endothelial dysfunction, ROS generation, and myocardial injury, inhibition is likely to benefit the postischemic heart. Inhibitory effects in neutrophils involve A2AR activation (41, 134, 278) limiting ROS generation and endothelial adherence. Adenosine receptor activation inhibits neutrophil expression of CD11/CD18 (297), and McPherson et al. (202) have shown that selective A2AAR activation significantly limits neutrophil and macrophage recruitment and vascular cell adhesion molecule-1, intercellular adhesion molecule-1, and P-selectin expression. Thus A2AR activation may modify a range of inflammatory processes, contributing to intravascular protection. Work by Jordan et al. (133) also supports A3AR-mediated protection against neutrophil-mediated injury postischemia (133), with an associated reduction in myocardial apoptosis.
Despite considerable support for a neutrophil-dependent response to adenosinergic stimuli, the role of the “anti-neutrophil” effects of adenosine in acute versus chronic cardioprotection may be questioned. For example, the time course of neutrophil invasion is not generally appropriate in explaining acute injury. The elegant study of Mallory and colleagues (192) was among the first to establish that neutrophil invasion developed during the initial 24-h postischemia, peaked at 4 days, and then declined gradually over the next 10 days. Other studies (274) around the same time documented significant neutrophil invasion within 2 h of reperfusion. More recent work (264) suggests briefer periods of ischemia may actually accelerate neutrophil invasion. Nonetheless, the temporal characteristics of neutrophil invasion and activation appear inappropriate for mediating acute cellular oncosis and stunning within the first minutes of reperfusion. As reviewed recently by Baxter (20), evidence for an injurious role for neutrophils is mixed; infarct size has been shown to be unaltered by neutropenia in some studies, anti-inflammatory treatments do not always limit injury, and a number of anti-neutrophil approaches have been unsuccessful in limiting postischemic injury. Finally, blood-free models display many of the hallmarks of ischemic and reperfusion injuries evident in the in situ myocardium. It may well be that the extent of neutrophil activation and invasion reflects rather than determines the severity of myocardial injury. Nonetheless, the work of Vinten-Johanssen and colleagues (133, 134, 278, 288, 313) in particular provides considerable support for a protective anti-neutrophil effect of adenosine in reperfused myocardium.
Cardioprotection Through Modulation of Inflammation
There is an emerging view that many of the protective actions of adenosine in reperfused tissues may stem from modification of inflammatory responses, additional to the above-mentioned anti-neutrophil effects of adenosine. As outlined by Sitkovski (262), adenosine may serve as the intrinsic “metabolic switch” that senses excess tissue injury and acts to downregulate the release and activity of proinflammatory agents. All adenosine receptor subtypes are attributed with one or more effects on the inflammatory process. It has emerged that monocytes/macrophages are key cellular targets for adenosinergic regulation of inflammation, together with leukocytes and endothelium (30, 32, 100, 147, 174, 191, 203, 212, 216, 262). Adenosine receptor stimulation in lipopolysaccharide (LPS)-stimulated monocytes/macrophages reduces production of tumor necrosis factor-α (TNF-α) (30, 203, 245), interleukin-12 (101), and macrophage inflammatory protein (MIP)-1a (270), limits induction of inducible NO synthase (iNOS) (300), and represses expression of the membrane protein major histocompatibility complex II (71, 300). Adenosine also inhibits endothelial cytokine secretion and expression of adhesion molecules (30) and NF-κb activation and signaling stimulated by TNF-α in myeloid, lymphocytic, and epithelial cells (191). However, a recent study suggests NFκb signaling is unimportant in the anti-inflammatory effects of adenosine receptors in macrophages (212).
With respect to adenosine receptor subtypes, our own recent work employing cDNA microarrays reveals multiple gene transcripts involved in the modulation of inflammatory responses are selectively modulated by A1 adenosine receptor expression in the normoxic heart (156). Furthermore, postischemic changes in transcription of various inflammatory elements are also modulated by A1 expression (8). In terms of the A2AAR, this subtype is generally considered of chief importance in modulating cytokine generation and macrophage activity. However, there are contradictory findings regarding the A2AAR. For example, there is recent evidence against expression of the A2A subtype in specific macrophage lines (212), and selective A2AAR agonism is not particularly potent in inhibiting MIP-1a production in these same macrophages (270). However, the latter may reflect the importance of receptor density and coupling in dictating functional sensitivity to receptor agonism in different cell types. Countering such issues, genetic deletion studies verify an important role for the A2AAR in suppressing inflammation (216). An interesting study from Platts et al. (231) also indicates that A2AAR activation specifically inhibits structural changes to the glycocalyx, which are an early event in postischemic inflammation. This glycocalyx effect may reflect a novel pathway by which A2AAR activation limits postischemic inflammatory injury in organs including the heart. Another study (50) has demonstrated that A2AAR (and A3AR) activation represses while A1 receptors enhance endothelial tissue factor expression induced by TNF-α, thrombin, or PKC activation. Thus adenosine may modulate procoagulant actions of inflammatory cytokines in this manner.
Activation of the A2BAR has been shown to reduce vascular endothelial growth factor release from neutrophils and limit transendothelial migration (292) and to modify endothelial junctions and barrier function (169). Additionally, an interesting feedback loop involving the A2BAR has been identified in which interferon-γ (IFN-γ) upregulates A2B receptors (by enhancing expression), and activation of the A2B receptors inhibits the IFN-γ-triggered expression of MHC class II genes, NO synthase, and proinflammatory cytokines (300). Induction of A2BARs has also been observed in LPS-stimulated macrophages (212).
Activation of the A3AR inhibits TNF-α secretion from macrophages (32), inhibits eosinophil chemotaxis (147), and represses endothelial tissue factor expression (50). A stimulatory action of adenosine on monocyte interleukin-10 secretion (which inhibits TNF-α) may also be attributed to the A3AR because the response was insensitive to receptors activated by R-PIA, 5′-N-ethylcarboxamidoadenosine, and 2-chloroadenosine (166).
It is clear that responses to adenosine receptors are quite varied, and in some cases effects of one receptor subtype may actually oppose those of another, with ultimate responses representing an amalgam of concentration-dependent effects of the different functional receptors present. For example, neutrophil adherence to endothelium can be activated by A1 receptors and inhibited by A2 receptors (42). In addition, quite selective effects of different subtypes on similar processes have been reported. For example, whereas data supports A3-but not A2A-mediated inhibition of LPS-induced TNF-α secretion from macrophages (203, 246), there is also evidence for A2A-but not A1- or A3-mediated inhibition of interleukin secretion from monocytes or macrophages (30, 101). There are also contradictory findings regarding the action of adenosine. For example, recent work indicates that the reduced release of TNF-α from macrophages does not involve changes in gene transcription, whereas Sajjadi et al. (246) show that adenosine does decrease TNF-α mRNA in stimulated human monocytes.
Formation of adenosine by inflammatory and endothelial cells may also be specifically modulated during injury. Activated polymorphonuclear leukocytes release increased quantities of 5′-AMP, which can be rapidly hydrolyzed by ecto-5′-ectonucleotidase (CD73) to locally generate adenosine (Fig. 1). This adenosine may then play a key role in modulating local inflammatory responses. Additionally, Naravula et al. (211) provides evidence for a paracrine feedback loop whereby locally generated adenosine may transcriptionally control ecto-5′-nucleotidase expression (and thereby adenosine generation) and endothelial barrier function (Fig. 1). This endothelial junction organization and barrier function appears regulated by the A2B receptor (169). Other studies reveal that endothelial ectoenzymatic generation of adenosine limits platelet activation and adherence to inhibit inflammation (137). The inflammatory process itself also targets ecto-5′-nucleotidase activity because TNF-α reduces endothelial ecto-5′-nucleotidase levels (135) and adherent leukocytes inhibit endothelial ecto-5′-nucleotidase activity and adenosine formation (125). These inhibitory responses, in turn, may limit adenosine formation and contribute to endothelial activation during inflammation.
In cardiac tissue, adenosine has been shown to limit TNF-α secretion from both normal and failing myocardium (289, 291). The same research group has shown that A2 receptors are involved in the inhibition of cardiac TNF-α secretion, whereas the A3 receptor enhances cardiac secretion of interleukin-6 (290). Interestingly, whereas these and other studies (34), reveal that adenosine limits cytokine expression following myocardial ischemia, adenosine-mediated protection of human hearts (during coronary artery bypass graft) is not associated with changes in varied indicators of inflammation (leukocyte and neutrophil numbers and cytokine responses). Thus protection against injury did not appear to be related to the suppression of inflammation (296). However, because systemic indicators were assessed, changes in localized inflammatory responses within the heart could not be excluded. Whereas it is clear that modulation of inflammation might provide long-term benefit following ischemia-reperfusion, the precise role of anti-inflammatory actions of adenosine in acute cardioprotection remains to be assessed.
Protection Against Coronary Vascular Injury
In the earlier sections we introduced evidence of compartmentation of adenosinergic protection and potential effects on coronary endothelial function. Although research into myocardial ischemia-reperfusion has somewhat logically focused on the myocyte, there is increasing support for a key role for coronary injury in generating the postischemic cardiovascular phenotype (92, 121). Vascular dysfunction predicts risk of cardiovascular events in patients with coronary artery disease (122) and is linked to worsened clinical outcomes in treatment of cardiovascular disease (251, 269). As already noted, a number of in vitro and in vivo studies support vascular sites of action of adenosine, particularly in postischemic tissue (155, 277, 278, 311, 313). Other studies verify specific protection against postischemic coronary vascular dysfunction via adenosinergic mechanisms. Maczewski and Beresewicz (185, 186) have presented evidence that adenosine receptors are involved in vascular protection with preconditioning, and Maddock et al. (187) found A2AAR activation protects against postischemic vascular injury in perfused hearts. Our own findings (76) reveal substantial protection against coronary dysfunction via A1AR activation by endogenous adenosine. These in vitro studies are consistent with evidence of vascular protection in vivo (12).
The mechanism of adenosinergic “vasoprotection” is unclear. Maczewski and Beresewicz (186) argue PKC-dependent reductions in endothelin formation are involved, consistent with the observation that PKC inhibition prevents vascular dysfunction in some models. On the other hand studies of vascular protection with preconditioning support signaling common to myocardial protection (185), with PKC and mitoKATP channel activation protecting against vascular insult. Vascular injury may also stem from substantial postischemic endothelial apoptosis documented recently by Scarabelli and colleagues (249, 250). In this respect, adenosinergic protection is associated with the limitation of apoptosis in other cell types. Thus “vasoprotection” may also involve reduced vascular apoptosis. The extent to which protection of the vasculature impacts on myocardial outcome from ischemia-reperfusion remains to be defined.
“Dysregulation” of Adenosinergic Cardioprotection
If the adenosinergic system is a determinant of ischemic tolerance and postischemic phenotype, abnormalities in this system may be associated with impaired tolerance. There is evidence for this in several models and conditions. For example, hypertrophic myocardium is intolerant of ischemic and other forms of insult, likely for multiple reasons. Not only is adenosine potentially involved in regulating cardiac and vascular hypertrophic growth (66–69), but also adenosinergic signaling is modified in hypertrophic tissue (204, 273, 307). Whether such changes contribute to the intolerant phenotypes remains to be directly tested. However, adenosinergic therapy has been shown to counter detrimental phenotypic changes associated with hypertrophy (39), dysregulation of adenosine formation may occur in hypertrophy (53), and dysregulation of adenosine formation with hypertension is involved in modifying vascular proliferation (69). There is therefore good evidence that cardiovascular responses to these disease states are associated with altered adenosine signaling and/or protection.
Ischemic heart disorders are primarily a disease of the elderly, with coronary artery disease effecting at least 50% of those over the age of 65 yrs of age. It is therefore the aged myocardium that is at greatest risk of ischemic insult, and unfortunately the aged myocardium appears intrinsically less tolerant of this insult (83, 110, 170, 193), although this is not a universal observation. A number of studies show altered adenosinergic signaling in aged tissues (85, 86, 108, 128). Recently, Gao et al. (86) and our own group (119) have shown that adenosine-mediated cardioprotection is absent in the aged myocardium. Similarly, preconditioning responses and PKC-mediated protection have also been shown to be absent or impaired (253). It is thus tempting to speculate that impaired adenosinergic protection contributes to the aging-related decline in ischemic tolerance. This is supported by recent studies in the mouse heart (86, 119). The age-related failure in adenosinergic protection does not involve a reduction in A1AR transcription or expression (119) but appears to involve a failure in signal transduction distal to the receptor itself. Because the putative mitoKATP activator diazoxide does protect aged hearts (119, 168), the failure lies proximal to this mediator. Moreover, because amplification of the stimulus (i.e., A1AR expression) can overcome the failure, it is still possible to generate a normal protective response in the aged myocardium.
It is clear that the adenosine receptor system and adenosine metabolism play important roles in determining the cardiovascular outcome from ischemic insult. In the last decade our understanding of the role of adenosine in modulating ischemic tolerance has developed enormously. Nonetheless, our knowledge is incomplete regarding the identity of the receptor subtypes involved, the locations and relative contributions to protection, precise effects on different forms of ischemic and postischemic injury, and the molecular bases of observed protective responses. Clarification of these processes will undoubtedly provide important insights into the intrinsic response of the heart to acute ischemic insult, and whereas “adenosinergic therapy” itself may prove problematic, this knowledge may aid in developing novel therapeutic strategies targeting selected aspects of ischemic and reperfusion injuries in man.
Dr. Headrick is the recipient of a Career Research Fellowship from the National Heart Foundation of Australia.
- Copyright © 2003 by the American Physiological Society