Adenosine and adenosine receptor-mediated action in coronary microcirculation

Adenosine is an ubiquitous extracellular signaling molecule and plays a fundamental role in the regulation of coronary microcirculation through activation of adenosine receptors (ARs). Adenosine is regulated by various enzymes and nucleoside transporters for its balance between intra- and extracellular compartments. Adenosine-mediated coronary microvascular tone and reactive hyperemia are through receptors mainly involving A2AR activation on both endothelial and smooth muscle cells, but also involving interaction among other ARs. Activation of ARs further stimulates downstream targets of H2O2, KATP, KV and KCa2+ channels leading to coronary vasodilation. An altered adenosine-ARs signaling in coronary microcirculation has been observed in several cardiovascular diseases including hypertension, diabetes, atherosclerosis and ischemic heart disease. Adenosine as a metabolite and its receptors have been studied for its both therapeutic and diagnostic abilities. The present review summarizes important aspects of adenosine metabolism and AR-mediated actions in the coronary microcirculation.


Introduction
The coronary microcirculation supplies oxygen and nutrients by determining blood flow to the myocardium through the regulation of vascular resistance. The regulation of coronary microcirculation is essential but complex and is accomplished by changes in coronary microvascular tone, i.e. in contraction and relaxation of vascular smooth muscle, through integration of factors and multiple signals from the perivascular nerves, the myocardium, the endothelium as well as circulating cells [47,88]. Coronary microvascular dysfunction, resulting in impaired oxygenation and lowgrade inflammation, likely contributes to the pathogenesis of coronary microvascular angina [70,94]. These patients with signs and symptoms of ischemia and non-obstructive coronary artery disease are associated with elevated risk for adverse outcomes [70,94]. However, the diagnosis of coronary microvascular dysfunction is limited, the disease mechanisms are not fully understood and the patients with non-obstructive coronary artery disease remain undertreated [6,80].
Adenosine plays a crucial role in the regulation of coronary microvascular tone and coronary blood flow in both physiology and coronary vascular diseases [31,36,47]. Adenosine is an autacoid produced by the action of ecto-5′-nucleotidase on extracellular adenine nucleotides released from the parenchymal tissues including endothelium, myocardium and erythrocytes [60,71]. Extracellular adenosine exerts its vascular effect via interaction with specific cell-surface receptors located on the smooth muscle and endothelial cells of the coronary vasculature. There are four adenosine receptor (AR) subtypes, namely A 1 R, A 2A R, A 2B R, and A 3 R. A 1 R and A 3 R are negatively coupled to adenylyl cyclase through the Gi/o protein alpha-subunits and activation of those receptors decreases cAMP levels, whereas A 2A R and A 2B R are positively coupled to adenylyl cyclase through Gs and enhance cAMP levels [119]. All four AR subtypes are found in coronary smooth muscle and endothelial cells [3,31,67,76]. The distribution of ARs along the branches of coronary arteries also varies. For instance, in the porcine heart, expression of A 1 R and A 2A R proteins has been documented in the left anterior descending artery, while A 1 R, A 2A R and A 2B R are expressed in coronary arterioles [35,108]. Despite the fact that the A 2B R expression is suggested to be restricted to coronary microvascular origin [27,63], findings from studies using A 2A R knockout (KO) mice suggested a functional role of A 2B R in regulating larger coronary arteries than previously thought [96]. The primary effect of adenosine in coronary microcirculation is to induce vasodilation and hyperemia [31,47]. This property of adenosine to modify coronary microvascular function has been used for diagnostic effects for many years and is widely adopted as the gold-standard method of diagnosing ischemia invasively and noninvasively. The therapeutic potential of adenosine and its ARs has also been studied.
This review summarizes important aspects of adenosine and AR-mediated actions in the coronary microcirculation. The main focus is on the evidence addressing the role of adenosine and involvement of ARs in regulation of coronary microvascular function in physiology. We also discuss the pathophysiology of coronary microvascular regulation in several cardiovascular diseases. Finally, this review briefly touches upon the possible therapeutic potential of adenosine and AR modulation. Considering the differences in heart anatomy and metabolism among different species [89], coronary arteries with the diameter below 200 µm in human and large animal models are included as microvessels in the present review [85], while the changes in flow measured in vivo and ex vivo are regarded as the vasomotor control of the resistance vessels in human, large animal models and rodents.

Adenosine generation and metabolism
Adenosine is released in coronary microcirculation from tissues including endothelium, myocardium and erythrocytes at times of cellular stress such as hypoxia, ischemia and inflammation [60]. Adenosine can be formed intracellularly from ATP, ADP or adenosine monophosphate (AMP) by cytoplasmic 5′-nucleosidase activity. The conversion of cAMP to AMP by phosphodiesterase (PDE) is responsible for adenosine production referred to as the cAMP-adenosine pathway [73]. In addition, adenosine can be produced from S-adenosylhomocysteine (SAH) via SAH hydrolase [21,82]. Once being released extracellularly, ATP is degraded to ADP and AMP through the continuous action of NTPDase 1 (CD39) or possibly other NTDPases [119,124]. Adenosine is then generated from AMP derived from both ATP and cAMP pathways via CD73 [73] (Fig. 1).
Alteration of the regulatory enzyme activity under (patho) physiological conditions or in response to pharmacological stimuli can affect adenosine levels and subsequently the ARmediated vascular responses. Physical training may increase cytoplasmic 5′-nucleosidase and adenosine deaminase activity, thereby affecting adenosine concentration [46]. 5′-nucleosidase activity was thought to be inhibited during ischemia or hypoxia [29]. However, the net adenosine concentration was not measured. In a setting of reduced tissue oxygenation, the adenosine level can be elevated more than the AMP level likely via decreased activity of adenosine kinase [19]. Upon β-adrenergic stimulation, SAH-hydrolase was inhibited via a calcium-dependent mechanism [90], while CD73 activity was increased [32]. Hypoxia can also increase CD73 activity resulting in increased extracellular cardiac adenosine production [33]. α1-adrenergic stimulation, nitric oxide (NO)-donors and 8-bromo-cGMP could stimulate PKC leading to increased activity of CD73 [4,65]. Pharmacological inhibition of adenosine deaminase and kinase in perfused mouse hearts resulted in a significant increase in coronary flow [93].
Adenosine can diffuse across cell membranes to maintain the balance between intracellular and extracellular adenosine concentrations. Extracellular adenosine is rapidly taken up by the cells via both sodium-dependent (concentrative nucleoside transporter: CNT) and sodium-independent transporters (equilibrative nucleoside transporter: ENT) for subsequent metabolism [50,51,53]. Further, adenosine can pass through the plasma membrane of these cells and be used intracellularly [73]. Once taken up, e.g., by the endothelial cells, adenosine is phosphorylated by adenosine kinase to form AMP or degraded to inosine by adenosine deaminase (ADA) [73] (Fig. 1). Both ENT and CNT are expressed in the heart and vessel [53]. However, ENT1 and ENT2 are the best-characterized transporters for adenosine uptake in the cardiovascular system [50]. Existing evidence has shown that targeting ENT contributes to coronary vasodilation [7,39]. ENT1 and ENT2 are the predominant nucleoside transporters of the vascular endothelium with an approximate expression of ENT1 twice as high as that of ENT2 [51]. The human ENT1 and ENT2 differ in their sensitivities to inhibition by coronary vasodilators such as dipyridamole, dilazep and draflazine, with ENT1 being ≈100-to 1000-fold more sensitive than ENT2 [104].
However, there are several limitations in our current understanding of adenosine metabolism. The mechanisms underlying the regulation of these enzymes and transporters are not fully understood, which deserves further investigations. In addition, more studies are needed using human tissues, as there are species differences with respect to adenosine metabolism [20]. Finally, how altered enzyme activity and adenosine concentration affects sensitivity and activation of ARs in coronary microcirculation remains poorly elucidated. For more details on adenosine metabolism, the reader is referred to several excellent review articles [20,81].

Involvement of ARs in coronary microvascular tone control
Adenosine is a potent coronary vasodilator in all species studied, including human [15,60,69,109,126]. It can arise directly from cardiomyocytes after intracellular breakdown of ATP and after extracellular breakdown of ATP released from endothelial cells and erythrocytes [71]. The involvement of ARs in adenosine-mediated coronary vasodilation is species dependent. Several lines of evidence have shown that both A 2A R and A 2B R mediate exogenous adenosineinduced coronary vasodilation in mice [59,86,93], while A 2A R is the predominant receptor subtype contributing to coronary vasodilation in swine and dogs [9,35,52,125]. Involvement of ARs in human coronary vascular tone is not consistent. Activation of A 2A R has been shown to regulate human coronary vascular tone [79], whereas another study indicates an involvement of A 2B R in adenosine-induced relaxation in small arteries isolated from human [42]. ARs also interact with each other to regulate coronary vascular tone. Both A 1 R and A 3 R have been found to negatively modulate coronary vasodilation induced by A 2A R and/or A 2B R activation [92,95]. The A 2B R expression is upregulated in coronary arteries isolated from mice lacking A 2A R. As a functional consequence, the A 2B R-mediated increase in coronary flow is enhanced in mice lacking A 2A R [96]. Further, the A 2A R-mediated increase in coronary flow is enhanced in mice lacking A 2B R [78] (Fig. 2). Whether ARs play a significant role in the regulation of coronary basal tone remains controversial. In isolated rat hearts, the coronary baseline flow is significantly reduced by non-selective AR inhibition [49]. A 2A R activation has been observed to contribute to coronary basal NO release and basal tone in isolated hearts of mice [28,117,120] (Fig. 2). In contrast, the effect of AR blockade on coronary blood flow in vivo is rather small in human [25,26] and swine [24], and even absent in dogs and mice [5,99,115].
As mentioned earlier, upon induction of hypoxia or ischemia in various tissues, adenosine together with ATP and ADP is released from cells or tissues, all of which significantly contribute to reactive hyperemia [68,81]. It has been proposed that adenosine and adenosine-mediated ARs predominantly account for the mid-to late-phase of reactive hyperemia [68]. Existing evidence demonstrates that Adenosine generation and metabolism. Adenosine can be formed intracellularly from ATP, ADP or adenosine monophosphate (AMP) by cytoplasmic 5′-nucleosidase activity. The conversion of cAMP to AMP by phosphodiesterase is responsible for adenosine production referring as the cAMP-adenosine pathway. In addition, adenosine can be produced from S-adenosylhomocysteine (SAH) via SAH hydrolase. Once ATP is released extracellularly through pannexin 1 channels or ATP binding cassette transporter, ATP is degraded to ADP and AMP mainly through the continuous action of CD39. Adenosine is then generated from AMP derived from both ATP and cAMP pathways via CD73. Extracellular adenosine is rapidly taken up by the cells via nucleoside transporters for subsequent metabolism. Adenosine is then phosphorylated in the cells by adenosine kinase to form AMP or degraded to inosine by adenosine deaminase activation of A 2A R plays a pivotal role in reactive hyperemia in mice and dogs [9,86,117,122]. Other receptors play a lesser role. For instance, A 1 R has been shown to negatively modulate coronary reactive hyperemia mediated by A 2A R [122] (Fig. 2). A 2B R seems not to be involved in coronary reactive hyperemia [86,122]. There is also evidence showing that adenosine is unlikely to be involved in coronary reactive hyperemia [10,22].
Adenosine levels (calculated) do not increase enough to reach the concentration threshold to cause coronary vasodilation with increasing exercise intensity in human, swine and dogs [24,100], and there is no evidence for increased myocardial interstitial levels of adenosine following adenosine receptor blockade [100,115]. No involvement of adenosine or A 2A R has also been observed in a mouse model with pacing-induced coronary hyperemia [121]. These findings indicate that adenosine is not mandatory for coronary metabolic hyperemia.
Existing evidence demonstrated divergent effects induced by activation of A 1 R and A 3 R on coronary microvascular function. Vasodilator effect mediated by A 1 R and A 3 R has been evidenced by the A 1 R-induced vasodilation in canine coronary microcirculation [16] and the A 3 R-produced coronary vasodilation in isolated rat hearts [40,76]. In contrast, A 1 R antagonism augments the sensitivity to adenosine in isolated human coronary arterioles [79]. Further, both A 1 R and A 3 R have been found to negatively modulate coronary vasodilation induced by A 2A R and/or A 2B R activation in isolated mouse hearts [92,95], and A 1 R counteracts the A 2A R-mediated coronary reactive hyperemia [122] (Fig. 2).

Endothelium-dependent and -independent regulation
It has been suggested that both A 2A R and A 2B R mediate endothelium-dependent coronary relaxation and NO release from coronary artery endothelium [1]. Indeed, adenosine-5′-N-ethylcarboxamide (NECA), a nonselective adenosine agonist, and 2-[p-(2-carboxyethyl)] phenylethyl-amino-5′-Nethylcarboxamidoadenosine (CGS-21680), a selective A 2A R agonist, produced relaxation in isolated porcine coronary small arteries, which were attenuated by the endotheliumdenudation or NO synthase inhibition [34]. Using two different NO synthase inhibitors L-NAME and L-NMA in isolated hearts from wild-type (WT) and A 2A R KO mice, both inhibitors attenuated the NECA-or CGS-21680-induced increases in coronary flow in WT, but not A 2A R KO mice, indicating a role for NO in the A 2A R-mediated coronary vasodilation [96]. NO blockade or endothelium denudation also attenuated adenosine-induced vasodilation in porcine coronary arterioles [44]. Interestingly, adenosine-A 2A R pathway has been shown to regulate coronary basal tone through NO release in isolated mouse hearts [120]. There is also evidence showing that NO release is in part triggered by A 2A R accounting for reactive hyperemia in mice [117]. The role of A 2B R in NO release remains to be determined.
In contrast, many other studies have observed that adenosine mediates endothelium-independent relaxation in coronary microvasculature. Thus, adenosine-induced vasodilation in human coronary small arteries was not affected by endothelium denudation [79] or NO blockade [42]. NO synthase inhibition failed to affect adenosine-induced vasodilation in canine coronary arterioles [41]. Endogenous adenosine and NO work in a parallel manner to regulate vascular tone in isolated canine coronary small arteries [116]. NO does not contribute to the A 2A R-mediated increase in reactive hyperemia in A 1 R KO mice [122]. Numerous pieces of evidence obtained in denuded porcine coronary small arteries clearly demonstrated that A 2A R plays a predominant role in endothelial-independent coronary vasodilation, while A 2B R may play a minor role [91,98,125]. The discrepancies on the role of endothelium in the adenosine-mediated coronary microvascular regulation are not readily explained, but may be determined by the different expression and distribution of ARs between the endothelium and smooth muscle cells in the different vascular segments of the microcirculation. It may also depend on different species studied, as NO seems to be involved in adenosine-induced coronary vasodilation in swine, but not dogs [41]. Further studies are warranted to address this issue.

Post-receptor pathways and end-effectors
The coronary microvascular tone is ultimately determined by the interaction between actin and myosin in the vascular smooth muscle cells. This is regulated by the intracellular Ca 2+ concentration. Opening status of one of the important modes voltage-operated Ca 2+ channels in vascular smooth muscle is regulated by membrane potential, which in turn is determined by the activation of K + channels [24]. Many vasoactive substances including H 2 O 2 influence coronary microvascular function through K + channels [64,75,78,86,118]. The limited evidence regarding the mechanisms is pointed to a possible activation of both transcription and translation of K + channels located at the plasma membrane of the coronary smooth muscle cells [64]. The three main types of K + channels that have been investigated in relation to regulation of coronary vasomotor tone are K ATP , K Ca2+ and K V channels [23]. Despite information indicating that adenosine receptors and K ATP work as parallel vasodilator pathways to control coronary blood flow in swine [56], both A 2A R-and A 2B R-mediated increase in coronary flow in isolated mouse hearts have been observed to be through activation of K ATP channels [78]. The adenosine/A 2A R stimulation-or the adenosine analogue-induced relaxation in isolated porcine coronary arterioles or the A 2A R-induced increase in coronary blood flow in open-chest dogs is mediated via activation of K ATP channels [8,9,34]. Adenosine has been shown to potentiate the flow-mediated dilation in porcine coronary arterioles via activation of K ATP channels in endothelium [44]. There is NO and K ATP channel-dependent effects of A 2A R contributing to reactive hyperemia in mouse [117]. Of further interest, recent evidence has shown that A 2A R activation promotes NADPH oxidase 2-derived reactive oxygen species and subsequently leads to H 2 O 2 production contributing to the increase in coronary flow in isolated mouse hearts [126].  [86]. H 2 O 2 can activate K ATP current in smooth muscle cells [86]. Further, adenosine-mediated increase in coronary flow is blunted by catalase, while H 2 O 2 increases coronary flow which is attenuated by the K ATP blocker glibenclamide [86]. Finally, both H 2 O 2 and K ATP activation are involved in A 2A R-mediated coronary reactive hyperemia [86,122]. Altogether, these observations indicate that adenosine-mediated A 2A R is coupled to smooth muscle K ATP channels in coronary reactive hyperemia in mice via the production of H 2 O 2 as a signaling intermediate. Earlier studies have also shown an involvement of K ATP channels in hypoxia-induced coronary vasodilation as well as dipyridamole-mediated increase in coronary vasodilation in perfused guinea pig hearts [18,106], suggesting that adenosine could hyperpolarize smooth muscle cell membrane by opening K ATP channels under hypoxic condition.
There is also evidence showing the involvement of K V and K Ca2+ channels in adenosine-or the A 2A R agonist-mediated coronary vasodilation. Adenosine-mediated increases in coronary blood flow in dogs and relaxation in isolated canine coronary arteries, the adenosine analogue-induced relaxation in isolated porcine coronary arterioles, as well as the adenosine/the A 2A R agonist-induced relaxation in coronary arteries isolated from rats are attenuated by K V channel inhibition [8,9,22,43]. Moreover, adenosine-mediated vasodilation in pressurized human and canine coronary small arteries and in perfused rat hearts were blunted by K Ca2+ channel inhibition [11,58,79].
Collectively, adenosine-mediated coronary microvascular tone and reactive hyperemia are through complex mechanisms mainly involving A 2A R activation on both endothelial and smooth muscle cells, but also involving the interaction of different ARs (Fig. 2). Regarding the post-receptor mechanism, K ATP , K V and K Ca2+ channels appear to act as final effectors for the adenosine-mediated coronary microvascular tone regulation. However, the mechanisms underlying how adenosine or the A 2A R-mediated coronary vasodilation activates H 2 O 2 -K ATP axis remains incompletely understood. Moreover, whether adenosine-H 2 O 2 -K ATP axis can be extrapolated to human condition deserves further investigations. Table 1 summarizes important evidence regarding the role of adenosine-and AR-mediated actions in coronary microcirculation in physiology.

Hypertension
Hypertension is associated with structural and functional abnormalities in coronary microcirculation including coronary endothelial cell dysfunction, coronary microvascular remodeling and an impaired coronary flow reserve induced by adenosine observed in both human and animals [62,105]. Arterial hypertension can lead to an increase in the vascular pericyte coverage, which is interestingly not accompanied by a gain in capillary density [127]. In addition, this cell type also undergoes a transformation into a more vascular smooth muscle cell like phenotype showing a more contractile property [127].
Both adenosine and the selective A 2A R agonist produce concentration-dependent relaxation of coronary arteries isolated from control rats via activation of Kv 7 channels but not hypertensive rats [43]. There is downregulation of A 3 R expression and the A 3 R-mediated coronary vasodilation in perfused hearts from spontaneously hypertensive rats [38]. In hypertensive swine, the transmural spatial density of microvessels is twice as much as in normotensive animals, and myocardial levels and expression of endotheliumderived growth factors, e.g., FGF (in vascular smooth muscle cells and myocytes) and VEGF (in endothelial cells) are significantly increased. Functionally, the increase in blood volume and myocardial blood flow in response to intravenous adenosine application was blunted in these animals [74]. It seems that activation of downstream potassium channels plays a role. In high-salt diet-induced hypertensive rats, application of nicorandil, an activator of K ATP channels, restores NO synthase and attenuates enhanced VEGF and FGF gene expression resulting in coronary capillary and arteriolar growth [114]. These findings suggest that A 2A R, A 3 R and potential downstream potassium channels might play a crucial role in coronary microvascular dysfunction in hypertension and potentially set a new strategy to pharmacological manipulation of coronary microvascular function by applying agonists stimulating these components (Table 2).

Diabetes
Diabetes is an important risk factor for the development of cardiovascular disease including atherosclerosis and ischemic heart disease. The increased morbidity and mortality are significantly attributed to diabetes-induced microvascular dysfunction in the heart [119].
The coronary flow in hearts isolated from type 1 diabetic mice is observed to be significantly increased by the stimulation of the non-selective agonist of ARs and the selective A 2A R agonist. In addition, in vivo injection of the A 2A R agonist enhances the efficiency in increasing coronary flow in type 1 diabetic mice [45]. The functional observations are in accordance with the increased A 2A R expression in coronary arteries as compared to non-diabetic control mice [45]. In contrast, the coronary flow in response to adenosine is significantly blunted in isolated hearts of type 2 diabetic Goto-Kakizaki (GK) rats as compared to age-matched control rats [58]. The impaired adenosine-induced coronary flow in GK rats can be restored by endothelial K Ca2+ opening [58]. In obese rats with insulin resistance, the coronary microvascular perfusion is impaired in response to adenosine infusion [102]. A similar clinical observation was found in a recent study where the adenosine-induced coronary flow reserve is blunted in type 2 diabetic patients without obstructive coronary artery disease [48] ( Table 2). The different responses to adenosine stimulation may be due to different etiologies of diabetes, which warrants further investigations. In a swine model with early-stage metabolic syndrome and hyperglycemia, despite both adenosine-induced increase in coronary blood flow in vivo and the adenosine analogue-mediated relaxation in isolated coronary arterioles did not differ from control swine [8], there was a shift from the A 2A R-mediated coronary relaxation to enhanced A 2B R-mediated coronary relaxation in swine with earlystage metabolic syndrome [8] (Table 2). However, the A 2B R expression level was lower in coronary arterioles isolated from swine with metabolic syndrome. This may suggest that the sensitivity of A 2B R upon stimulation by the adenosine analogue is increased thereby maintaining the coronary blood flow [8]. Moreover, the involvement of K v channels in AR-mediated coronary relaxation was not affected by earlystage metabolic syndrome, whereas there was a reduced K ATP channel function [8]. Activation of A 2A R has been shown to be coupled to K ATP channel to regulate coronary

Atherosclerosis
Atherosclerosis is generally predominant in large coronary arteries. However, long-term exposure to hypercholesterolemia can activate endothelial cells and thus induce leukocyte recruitment, oxidative stress and loss of pericytes in the microcirculation [12]. This alteration may lead to capillary rarefication due to a decrease in capillary surface area resulting in a dysfunctional downstream vessel system and a drastic decrease in overall capillary diameter [127].
Moreover, in atherosclerotic areas, relative anoxia, inflammation and oxidative stress promote release of angiogenic factors resulting in angiogenesis and vasculogenesis [57]. It has been estimated that local adenosine may mediate 50-70% of the angiogenic response to hypoxia/ischemia [2]. A 1 R, A 2B R and A 3 R were involved in angiogenesis surroundings and downstream vessels of the atherosclerotic plaque, and A 1 R and A 2B R were reported to promote endothelial progenitor cell homing to coronary microvascular endothelium for the genesis of capillary networks [77].
Regarding the action of adenosine on vascular tone regulation, the coronary microvascular responses to adenosine are not consistent in atherosclerosis. Low coronary flow reserve after intracoronary adenosine infusion was observed in patients with atherosclerosis risk [72]. Reactive hyperemia-induced increase in coronary flow was lower in female atherosclerotic mice, and the less increase in coronary flow was inhibited by the A 2A R antagonist to a greater extent in atherosclerotic than control groups [120]. In contrast, an enhanced response in coronary flow to A 2A R stimulation in hyperlipidemic/atherosclerotic mice was reported [97]. It has been suggested that upregulation of A 2A R is a compensatory mechanism to maintain NO-dependent endothelial function as evaluated by coronary vasodilation in a mouse model of atherosclerosis [120]. One study indicates that responses of isolated coronary arterioles to adenosine are identical in atherosclerotic and control monkeys [84] ( Table 2). The experimental evidence may suggest that the diagnosis of coronary artery disease in patients using adenosine as a stimulator can be underestimated.

Ischemic heart disease
Among complex pathophysiological components, e.g. obstructive coronary atherosclerosis, more and more evidence has shown that coronary microvascular dysfunction significantly plays a role in the etiology of ischemic heart disease [54,70,94]. On the other hand, the coronary vasculature itself is also a victim of ischemia-reperfusion injury and myocardial infarction [30,37]. The majority of experimental and clinical studies have focused on the effects of adenosine more on cardiomyocytes as compared to the coronary vasculature, as dissecting the effects of adenosine or AR activation on the coronary microcirculation from cardiomyocytes is challenging, given the causal relationship between injuries to the coronary vasculature and cardiomyocytes following the myocardial infarction. Existing data demonstrated that there seems to be a reduced sensitivity to adenosine in the coronary microvasculature in ischemic heart [103,128]. the A 2 R agonist-induced coronary vasodilation was attenuated by the ischemia-reperfusion in anesthetized dogs [16]. The AR-, likely A 2B R-mediated relaxation to the novel dinucleotide Up 4 A in isolated coronary small arteries was found to be blunted in swine with myocardial infarction [123] (Table 2). More studies are needed to further elucidate the specific AR involvement in coronary microvascular function following myocardial infarction and how alteration of AR sensitivity is associated with ischemic heart disease.

Perspective on indirect adenosine modulation as therapeutic strategy
Adenosine and AR modulations may serve as therapeutic strategy in cardiovascular medicine in two manners. First, both endogenous and exogenous adenosine and adenosine-activated ARs per se have been evaluated in various preclinical and clinical settings. However, the effect of adenosine and AR modulation in myocardial injury and heart failure has shown inconsistent effects on cardiac function and myocardial perfusion. The exact mechanisms are not readily explained, but one possibility may rely on where the modulation takes place. For instance, endogenous generation of interstitial, but not venous adenosine, is critical to protect myocardium against infarction [83,87], which could be induced by ischemic preconditioning, but not coronary microembolization [87]. Moreover, none of the pharmacological tools targeting ARs that entered clinical trials have emerged as drug candidates due to lower efficacy, kinetics issues or adverse events reported. Better rational design and development of other agonists and antagonists may lead to successful clinical drug candidates in the future. Readers are referred to several review articles on this topic for more details [14,61,111]. Second, other drugs can initiate secondary effects through generation of adenosine and activation of AR-mediated signaling. It is important to note that AR-mediated actions may affect both coronary vasculature and cardiomyocytes, making it difficult to separate vascular effect from cardio-protection. This section focuses on the discussion of the indirect adenosine modulation for a potential therapeutic strategy.
An indirect, but clinically important, effect on AR-mediated signaling was recently postulated for ticagrelor [17]. Ticagrelor is the P2Y 12 R antagonist primarily targeting platelets and its application is clinically wellestablished to prevent thromboembolic complications after acute coronary syndrome [107]. Of note, ticagrelor can induce substantial amount of ATP release from erythrocytes via anion channels and target the ENT1 transporter in erythrocytes which inhibits adenosine uptake by erythrocytes [66,110]. Together with adenosine degraded from ATP in this pathway, ticagrelor, by targeting erythrocytes, leads to increases in circulating adenosine levels [101]. Considering the beneficial effects of adenosine on cardiovascular function [110], ticagrelor could have pleiotropic effects beyond its platelet inhibitory effects, as treatment with ticagrelor reduced major cardiovascular adverse events (MACE) compared to clopidogrel, another P2Y 12 R antagonist that does not have impact on erythrocytes for purinergic activation, in patients with acute coronary syndrome [13]. Indeed, increased adenosine concentrations by ticagrelor reduced anti-inflammatory responses, improved vascular function and attenuated ischemia-reperfusion injury [13]. Moreover, ticagrelor significantly enhanced adenosine-increased coronary blood flow in human and adenosine-mediated hyperemia in dogs [101,112]. Of note, a clinical study evaluating the effects of ticagrelor in stable multivessel ischemic heart disease is ongoing [17]. However, how much adenosine-mediated secondary effect of ticagrelor contributes to overall cardiovascular outcomes remains unclear.
In addition to ticagrelor, magnesium has been applied in patients for possible treatment of acute myocardial infarction [113]. Evidence has shown that the beneficial effect of magnesium in an animal model of myocardial infarction on the infarct size is by adenosine through enhancement of 5′-nucleosidase activity [55]. It is of interest to monitor the effect of magnesium on the coronary microvascular function. Further studies are required to better elucidate the extent to which enhanced adenosine responses contribute to the clinical profile of those compounds. More studies aiming at pinpointing ARs and manipulating receptor sensitivity in coronary microvasculature are also needed to evaluate the possible therapeutic potential.

Conclusions and perspectives
Adenosine is an endogenous purine nucleoside that functions as an extracellular signaling molecule via activation of ARs. Adenosine and adenosine-mediated ARs play a significant role in the regulation of coronary microcirculation in certain conditions in physiology and pathophysiology. Adenosine mediates coronary microvascular tone and reactive hyperemia mainly through A 2A R activation on both endothelial and smooth muscle cells and also via interaction with other ARs. ARs further activate the downstream effectors including H 2 O 2 , K ATP , K V and K Ca2+ channels leading to coronary vasodilation.
Adenosine-mediated AR activation also plays a role in several cardiovascular diseases. Downregulation of A 2A R, A 3 R and potential downstream potassium channels play a crucial role in coronary dysfunction in hypertension. A 1 R, A 2B R and A 3 R are thought to be involved in the angiogenesis and microvascular growth in coronary atherosclerosis. The coronary microvascular responses to adenosine are not consistent in atherosclerosis, which may underestimate diagnosis of coronary artery disease using adenosine as a stimulator. There is a decreased A 2 R sensitivity in coronary microcirculation after ischemia-reperfusion and myocardial infarction. The adenosine effect on coronary flow regulation in diabetes is not consistent and may depend on the etiology of diabetes. More studies are needed to evaluate the adenosine and AR modulation for the treatment. Indirect modulation of adenosine by a compound like ticagrelor may be of potential for the improvement of coronary microvascular function in certain cardiovascular disorders.
Collectively, there is a complexity of adenosine and ARmediated effects in coronary microcirculation. Many aspects are still not fully understood due to a number of discrepant observations. The discrepancy arises from (1) endogenous adenosine vs. exogenous adenosine effects and adenosine concentration vs. AR sensitivity, (2) different conditions/ stimuli (basal condition, ischemia, hypoxia, exercise/pacing and diseases) and (3) differences in AR expression and distribution in different microvascular segments of different species. Better understanding of these aspects will help with elucidation of the role of adenosine and AR in the regulation of coronary microcirculation and development of novel therapeutic strategies. and the Science and Technology Program of Sichuan Province, China (20GJHZ0036) (to XC). We acknowledge many of those who have made significant contributions to adenosine and coronary microcirculation field that are not included in the present study.
Funding Open access funding provided by Karolinska Institute.

Conflict of interest None.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.