I. INTRODUCTION

Top Previous Next References

Interest in ATP as a molecule was for many years devoted to the concept of the "high-energy phosphate bond" introduced by Lippman (292); this despite the fact that roles for extracellular purines had been described by Drury and Szent-Györgyi in 1929 (137) soon after the discovery of ATP (144, 295). In fact, the crude tissue extracts used by Drury and Szent-Györgyi contained mostly AMP and induced a negative chronotropic effect and a decrease in blood pressure. ATP was later shown to produce transient tachycardia at low doses or to slow heart and to induce atrioventricular block at higher doses (207, 454). These effects were associated with vasodilatation of coronary vessels and led Berne (36) to postulate that adenosine was the mediator. Adenosine then became the topic of most studies concerning purines and the cardiovascular system.

ATP is an amphiphilic compound showing both hydrophilic and strong hydrophobic interaction. Adenine, its 6-aminopurine base, has pK_a <1, 4.1, and 9.8. Like free fatty acids, ATP binds to BSA, which is generally used to mimic protein solutions and establish an oncotic pressure. Up to 94%, but not all, ATP could be liberated from BSA by the addition of palmitate. Interaction of BSA with AMP is considerably weaker (21). ATP strongly binds divalent cations such as to form ~70% MgATP², 22% CaATP², and 2.4% free ATP⁴ in a physiological solution (441). The diffusion coefficient of Na₂ATP in water at 25°C is 7.01 × 10⁶ cm²/s not far from 7.54 × 10⁶ cm²/s for CaCl₂ (196).

Autonomic nonadrenergic and noncholinergic nerves (71) were suggested to contain ATP (72), and a tentative model of storage and release was proposed (68). Nerves utilizing ATP as their principal transmitter were named "purinergic" (67). Based on the actions of purine nucleotides and nucleosides in a wide variety of tissues, Burnstock (69) proposed the P1- and P2-receptor classification, which was then extended to the heart (74). P1 purinoceptors are much more sensitive to adenosine and AMP than to ADP and ATP. The reverse is true for P2 purinoceptors. P2 receptors are unaffected by methylxanthines such as caffeine and theophylline that selectively and completely inhibited the P1 purinoceptors. Subclasses of P1 adenosine receptors were introduced when their stimulation was shown to inhibit (A₁ subtype) or activate (A₂ subtype) adenylyl cyclase activity. A similar lack of homogeneity of P2 receptors was indicated by the use of phosphate-modified ATP and ADP (166, 302); this was reinforced by the observation that apamin, a K⁺-channel blocker, antagonizes the inhibitory action of ATP in guinea pig cecum and stomach but not its excitatory action in guinea pig bladder and uterus (439). The first clear subdivision of P2 receptors proposed P_2X purinoceptors that mediate vasoconstriction with ,-methylene adenosine 5'-triphosphate (,-met-ATP) as a potent agonist, and P_2Y purinoceptors that mediate vasodilatation with 2-methylthioadenosine 5'-triphosphate (2-MeSATP) as a potent agonist (73).

P2 purinoceptors were initially classified according to pharmacological studies. This pharmacological nomenclature has now been replaced by molecular-based nomenclature. The first defined subtypes according to pharmacological criteria include the following: P_2X for excitatory and P_2Y for inhibitory, then P_2U for those also activated by UTP and P_2T at which ADP induces platelet aggregation, finally P_2Z for a large molecular pore activated by ATP⁴ and P_2D which is activated by dinucleotides. The second system is based on the cloning of P2 receptors where, the IUPHAR, Committee on Receptor Nomenclature (164, 165), has defined as P2X_1-n the ligand-gated ion channel and as P2Y_1-n the G protein-coupled receptors. Very slow progress has been made to establish the correspondence of the cloned to the pharmacologically defined P2 subtypes. This has proven difficult to realize in part because some P2 purinoceptors are functional heteroligomers and in part because the molecule binding to a given receptor is often a hydrolysis product of that which was added to the bath. Moreover, this field is characterized by the paucity of receptor subtype-selective agonists and antagonists that have their own propensity to undergo rapid hydrolysis and interconversion just like the natural agonists. It is also a problem that commercially available compounds might already contain breakdown products that act as effective agonists (42, 198, 342, 513), a point already considered in 1934 by Gillepsie (180).

After a characterization of the sources and extracellular metabolism pathways of ATP and the various purinergic receptors known today, the present review concentrates on the mechanical, chronotropic, and arrhythmogenic effects of ATP at the cellular level and under the whole heart as well as in pathological conditions. These pieces of information are then interpreted on the basis of our present knowledge concerning the electrophysiological consequences and signal transduction pathways activated by applying ATP to isolated cardiomyocytes. Inasmuch as it is possible, similar pieces of information are given for UTP and diadenosine polyphosphastes. Much more is known about the effects of ATP in wider aspects of cardiovascular system, and this has been extensively reviewed elsewhere (59, 143, 191, 272, 355, 402, 488).

	II. EXTRACELLULAR ADENOSINE 5'-TRIPHOSPHATE: SOURCES, METABOLISM, AND RECEPTORS

Top Previous Next References

Adenine nucleotides are continually present in quite variable amounts in the extracellular space of the heart. More precise knowledge of ATP exocytosis could be expected with the development of new bioluminescence technologies (464) involving cell surface-attached firefly luciferase (28) or atomic force microscopy in combination with myosin functional cantilevers (431). Basal ATP found in the coronary effluent from isolated, saline-perfused hearts had been shown to range below 1 nM (54, 491). This rather low value reflects the rapid degradation of ATP, which is over 95% during a single coronary passage. With the use of microdialysis, ATP in the interstitial space has been estimated to be 40 nM (276). These levels increase markedly during electrical stimulation as already noted in 1962 (1): hypoxia or ischemia (36, 163, 276, 359, 491, 519), challenge by cardiotonic agents (54, 117, 244, 491), increased blood flow (117, 492), mechanical stretch (483), and even work load in frog heart (136).

There is strong evidence supporting the notion that ATP is a cotransmitter in perivascular sympathetic nerves (68). Other likely sources of the nucleotide include the following: 1) ischemic myocytes (36, 163, 359, 519), 2) activated platelets (118, 209, 328), 3) nerve terminals (70, 210, 408), 4) inflammatory cells (133), 5) erythrocytes (33, 141, 449), 6) endothelial cells (46, 401, 526), 7) smooth muscle cells (45, 245, 369), and 8) exercising muscle cells (162, 364).

ATP can be released by exocytosis from platelets and nerves like other neurotransmitters. It can also leak out during cell lysis (139, 182). In recent years, the cystic fibrosis transmembrane conductance regulator (CFTR) has been suggested to act as an ATP channel and enables intracellular ATP to cross the cell membrane (2, 3, 6, 366, 432, 509). This is strongly disputed (see Ref. 123 for review), and it has been shown that mechanical stimulation of the cell surface, rather than CFTR activation, is sufficient to release ATP (510). A recent work (457) suggests that the permeation pathways for Cl and ATP are distinct, in which case CFTR would be a regulator of the ATP channel. Identification of the molecular basis of the CFTR-associated ATP channel is of critical interest.

Supporting the idea that an anion channel can allow for ATP to cross through membranes, it has been recently reported that the mitochondrial voltage-dependent anion channel (412) as well as the 116-pS Cl channel in cardiac sarcoplasmic reticulum (SR) conduct ATP and adenine nucleotides (247).

The breakdown of ATP and ADP to form adenosine was first reported nearly 50 years ago (235), and it is long recognized that this susceptibility to degradation does limit the potency of ATP and its degradable analogs (515). In considering the biological responses to extracellular ATP in multicellular tissues and even isolated cells, it should be noted that such effects are not only complicated by the presence of multiple P2-purinergic receptors but also by the rapid catabolism of ATP to produce adenosine. This cascade of surface-located enzymes converts P2 into P1 signaling. The extracellularly formed adenosine can itself modulate cardiac functions and also serve purine salvage after reuptake via plasma membrane-located adenosine transporters (471). The majority of the ATP perfused into the heart is dephosphorylated during a single passage through the coronary vasculature (29, 158, 359). The half-life of ATP is 0.2 s when perfused in blood through the lung vasculature (413) instead of ~10 min in whole blood ex vivo (479). It is likely that this catabolism is mostly due to endothelial cell ectonucleotidases. This hydrolysis of ATP, ADP, and AMP was first demonstrated in isolated rat myocytes by Bowditch et al. (56), but the nature and characteristics of the enzymes were not studied. Kinetic properties of the extracellular reaction sequences ATP  ADP  AMP  adenosine catalyzed by ectonucleotidases has been investigated at the surface of adult rat cardiomyocytes by following the catabolism of ³H-labeled nucleotides (158, 324) and by NMR (35) or in the ventricular muscle by microdialysis (276). It is worth noting that a significant ATP catabolism occurs even on isolated cardiomyocytes. Thus ATP is rapidly hydrolyzed to 65 and 33% after 3 and 15 min, respectively, by plasma membrane at 20°C (42).

A variety of extracellular enzymes utilize ATP to induce biological responses (256, 324, 325, 368, 373, 541, 542) (Table 1). In many tissues including heart, extracellular ATP as well as other nucleotide tri- and diphosphates appear to be hydrolyzed by E-type enzymes. They include ecto-ATPases and ecto-ATP diphosphohydrolases or ecto-ATPDases. All (but the -sarcoglycan or Adhalin) lack the Walker consensus ATP binding motif; instead, they have several apyrase conserved domains, a putative ATP -phosphate 1 binding motif, found in proteins as diverse as actins, 70-kDa heat shock protein, and sugar kinases. Ecto-ATPases are insensitive to known inhibitors of various intracellular ATPases such as ouabain and vanadate (P-type) or N-ethylmaleimide (V type) and F for the alkaline phosphatase. The turnover number of the ecto-ATPases is estimated to be 500,000 molecules/min, a value 50 times higher than the value of the Na⁺-K⁺-ATPase and two orders of magnitude higher than the one of the SR Ca²⁺-ATPase (380). These enzymes are activated by either Ca²⁺ or Mg²⁺ and inhibited however at near millimolar concentration by Cu²⁺, Zn²⁺, or La³⁺ (539). Aluminium is present at 1-10 µM in the extracellular fluid and is bound by ATP with a significantly higher affinity than either Ca²⁺ or Mg²⁺; it has, however, no effect on the hydrolysis of ATP by the enzymes (266). Ecto-nucleotidases have been proposed to be coreleased with ATP from nerve terminals (472).

Ecto-ATPDase is a highly glycosylated protein with six potential N-linked glycosylation, five apyrase-conserved regions, and two potential transmembrane domains. Both NH₂ and COOH terminals are short with COOH terminals demonstrating phosphorylation consensus sites (92, 304). This 82-kDa protein is identical to a lymphoid cell activation antigen CD39, an ecto-apyrase (227, 238, 507). It hydrolyzes nearly equally ATP and ADP to produce PP_i and P_i, respectively. It is inhibited by azide (265). Ecto-ATPDase has been found in heart tissues by Northern hybridization (238, 249). It has also been purified from the bovine heart and immunocytochemically localized in Purkinje and ventricular cells (25). Using an ATPDase purified from the bovine aorta, the same group demonstrated that the ATP analogs modified on the phosphate chain were not hydrolyzed at variance with those modified on the purine base. Thus 2-chloro-ATP, 2-MeSATP, and 8-bromo-ATP demonstrated similar Michaelis constant (K_m) and maximal hydrolysis rate (16-20 µmol·min¹·mg protein¹) (377).

Ecto-ATPases are millimolar cation-dependent, low-specificity enzymes that hydrolyze all nucleotide triphosphates (NTPases). A microsomal Mg²⁺-ATPase has originally been found in low-density vesicles that contained plasma membrane of various rat tissues with the following properties: 1) K_m for ATP of 0.2 mM; 2) ATP-induced inactivation of the enzyme prevented by concanavalin A; 3) hydrolysis of all NTP tested but no nonnucleotide phosphocompounds; and 4) Ca²⁺, Mn²⁺, Zn²⁺, or Co²⁺ could substitute for Mg²⁺, but Ca²⁺, Sr²⁺, Ba²⁺, or Be²⁺ could not fulfill the bivalent cation requirement (26). They can be distinguished from ecto-ATPDases by their inability to hydrolyze ADP and other nucleotide diphosphates at a rate >1-2% that of their ATP hydrolysis rate. Also, they are insensitive to azide. However, 8-azido-ATP is a good substrate of ecto-ATPase activity while it is an irreversible partial inhibitor after photoactivation (410). Another difference with ecto-ATPDases is fast inactivation of their ATP hydrolysis activity, an effect prevented by concanavalin A, a known inhibitor of 5'-nucleotidase (368). The first vertebrate ecto-ATPase was cloned and sequenced from the chicken muscle (257). It has a molecular mass of 54.4 kDa. Ecto-ATPases share some homologies with CD39 ecto-ATPDases in that they have two transmembrane domains, four external putative glycosylation sites, and four apyrase-conserved regions. Like ecto-ATPDases, they do not show Walker consensus sequence. The COOH terminal contains a single putative cAMP/cGMP-dependent protein kinase phosphorylation site as well as a single putative tyrosine kinase phosphorylation site (257).

A Ca²⁺/Mg²⁺ ecto-ATPase activity has long been reported in the plasma membrane of cardiac myocytes (128, 414, 480, 531). Rat cardiac sarcolemmal Ca²⁺/Mg²⁺ ecto-ATPase (Myoglein) requires millimolar concentrations of Ca²⁺ or Mg²⁺ for maximal ATP hydrolysis. It has been purified to apparent homogeneity, and a partial cDNA clone has been produced that revealed 100% homology to human platelet CD36 (241). In its native form, this enzyme has a molecular mass of 180 kDa and two subunits of ~90 kDa each (240). The limited sequence information indicates that there is no sequence similarity with any of the previously reported ecto-ATPases, chicken gizzard smooth muscle ecto-ATPase (257), or CD39 (242, 507), but partial homology with SR and sarcolemmal pump Ca²⁺-ATPases.

-Sarcoglycan (Adhalin) has recently been shown to have ecto-ATPase activity in skeletal muscle (38). It is different from the t-tubule ecto-ATPase of 56 kDa (119). Its deduced amino acid sequence for ATP binding reveals two Walter consensus sequences present in several intracellular ATPases. It binds ATP in a Mg²⁺-dependent, Ca²⁺-independent manner. ATP binding is inhibited by 3'-O-(4-benzoyl)benzoyl ATP (BzATP). The authors thus suggest that -sarcoglycan also modulates the activity of P2X (P2X₇) receptor by buffering the extracellular ATP concentration (38).

An ecto-nucleoside diphosphokinase (NDPK) has also been described but is, as yet, not reported in cardiac muscle. It controls the interconversion of P2-receptor agonists with a rate of extracellular transphophorylation up to 20-fold the rate of nucleotide hydrolysis (279). The latter can just couple ATP- and UTP-dependent stimulations. Other ecto-enzymes that also catalyze the hydrolysis of ATP such as alkaline phosphatase (518) or autotaxin, a nucleotide phosphodiesterase/ATP pyrophosphatase (93, 181), have not yet been studied in cardiac tissues.

Furthermore, the ATP degradation product AMP will be hydrolyzed to adenosine by 5'-nucleotidase, a phosphatase active mostly at alkaline pH (25, 146, 326, 377, 380, 433).

The kinetic properties of the ATP hydrolysis catalyzed by ectonucleotidases at the surface of cardiomyocytes were characterized (324). It is worthy of noting that the activity of these ecto-nucleotidases is modulated by Mg²⁺ (305), protein kinase C (PKC), and ₁-adrenergic stimulation (260, 262, 347, 420) as well as during preconditioning or postmyocardial infarction (263, 276, 329) and also differs in isolated perfused versus in situ hearts (276). The activity of the ecto-ATPase is not affected by treadmill exercise training or by exhaustive exercise (120). In isolated cardiomyocytes, exogenous adenosine, via A₁-receptor activation and a G_i protein, decreases both ectosolic and cytosolic 5'-nucleotidase activity (261).

Intracellular protein kinases are important in the regulation of cellular functions. Protein kinases that use extracellular ATP to phosphorylate proteins localized at the external surface of the plasma membrane (ecto-protein kinases) have now been demonstrated in a variety of tissues. Several surface proteins are reported to be phosphorylated in nerve cells and in aortic endothelial cells (140, 142; see Ref. 139 for review). In addition to an ecto-protein kinase with catalytic properties of atypical PKC- implicated in hippocampal long-term potentiation (85), a cAMP-dependent kinase has been shown to phosphorylate the atrial natriuretic peptides (270, 271). In view of the fact that ATP and cAMP are released in the extracellular space, one should expect a role for these ecto-kinases in the cardiac tissues.

Most P2Y receptors (P2Y_{2, 4, 6}) and, maybe, some P2X receptors are also activated by uracil nucleotides. This suggests the presence of extracellular UTP and derivatives. Recently, it has been demonstrated that mechanical stimulation induces a 15-fold increase in UTP release from 1321 N1 human astrocytoma (278). A recent review has appeared (10). Presently, little is known in heart. UTP stimulates cardiomyocytes. Most studies deal with the mechanisms of nucleoside plasma salvage leading to uracil nucleotide synthesis (175, 297). Total uracil nucleotides is 1.7 µmol/g protein in myocardium and in freshly isolated cardiomyocytes much less than the ATP content (27 µmol/g protein) in myocardium (176, 354).

Adenosine polyphosphates are a group of adenosine dinucleotides that consist of two adenosine molecules bridged by a variable number of phosphates thus abbreviated Ap_nA (n = 2-6). They are ubiquitous molecules found in prokaryotes and eukaryotes. Intracellular Ap_nAs appear to alert the cell under stress conditions, thus their name "alarmones." The potency and efficacy of Ap_nAs to inhibit ATP-activated K⁺ channel activity were described in cardiac cells and appeared similar to those of intracellular ATP (236). Also, Ap_nAs stimulate a Ca²⁺-induced Ca²⁺ release channel from skeletal muscle SR as well as increasing the binding of ryanodine to the calcium release channel by, respectively, nine- and threefold in skeletal and cardiac muscles (206). Another alarmone receptor is hemoglobin. Ap_nAs bind preferentially with high affinity to the deoxy conformation of hemoglobin in a ratio of one per tetramer. This binding markedly enhances the Bohr effect (52). Recently, it has been reported that the mammalian myocardium contains abundant amounts of Ap₅A (237) or of Ap₄A in humans (485). In mammalian tissues, Ap_nA are produced and released from platelets or chromaffin cells where it is costored with ATP and catecholamines. Ischemia induces a 10-fold decrease in the Ap_nA myocardial concentration (237, 264, 338). Ap_nAs are degraded in rat plasma to ATP (from Ap₅A) and AMP, and these nucleotides are further degraded to adenosine. However, compared with ATP, Ap₃A and Ap₄A have a relatively long half-life (300). The rate of degradation is dose dependent: the biological half-life was ~3 s at 1 mg/kg after intravenous infusion (253). The endothelial ecto-diadenosine polyphosphate hydrolase of cultured adrenomedullary vascular endothelial cells is activated by Mg²⁺ and Mn²⁺ and inhibited by Ca²⁺, adenosine 5'-O-(3-thiotriphosphate) (ATPS), and suramin (308). These compounds are considered to act as extra- and intracellular signaling molecules, and many of their properties are close to those of ATP (22, 259, 353).

Purified heart sarcolemma membranes were found to bind a slowly hydrolyzable analog of ATP, [³⁵S]ATPS, in a specific manner and exhibited two apparent affinity sites (532). The high-affinity site, which may represent the ATP receptor according to the authors, had a dissociation constant (K_d) of 4.7-8.3 nM and a maximal binding (B_max) of 9.5-18.4 pmol/mg protein, whereas the low-affinity site had a K_d of 655-1.227 nM and a B_max of 812-2.955 pmol/mg protein. [³⁵S]ATPS binding was displaced by GTP, UTP, CTP, and ITP. The number of high-affinity binding sites decreases during oxidative stress (334). Similar high- and low-affinity binding sites for ,-[³H]met-ATP, then considered to be a P2X agonist, were also reported in rat heart membranes (327). In another study, 8-azido-ATP (8-Az-ATP) was used for labeling intact rat ventricular myocytes. After minimizing background labeling by large concentrations of UTP, two bands, one near 48 kDa and a less pronounced at 90 kDa, are labeled by radioactive 8-Az-ATP (177). 2-MeSATP and ATPS both partially and specifically inhibited labeling of the 48-kDa band. This was related to activation of the fast Ca²⁺ transient by 2-MeSATP but not by ATPS and of the Na-P_i cotransport by ATPS only. However, much caution should be taken during these binding studies, not only because it would be difficult to distinguish sites with relatively similar binding affinities but mostly because secured experimental conditions are difficult to settle. Optimal binding conditions to obtain reliable equilibrium parameters require a 15-min incubation, a time during which >75% of the radioligand must remain intact, a situation not easily reached in view of the high ecto-ATPase activies (42).

Several reviews directly aimed at P2X receptors have appeared (17, 59, 64, 112, 149, 178, 349-351, 403, 448, 459). In this section I outline the characteristics of the P2X receptors in as far as they help understanding the following focus of attention on cardiac cells and tissue. Particular attention is paid to recent publications not covered by previous reviews. It is a general feature of all P2 purinoceptors to bind ATP⁴, uncomplexed to cations, whereas ATPMg² is the substrate of ATPases and kinases.

P2X receptors are ion channels opened by micromolar extracellular ATP. The seven subunits currently known (Table 2) are encoded by different genes. Each subunit has a topology that is fundamentally different from that of other known ligand-operated channels. P2X receptors are made of proteins, with 379-472 amino acids, inserted in the membrane to form a pore comprising two hydrophobic transmembrane domains with a large extracellular hydrophyllic loop. The putative extracellular loop of cloned P2X_1-7 has 10 conserved cysteine and 14 conserved glycine residues as well as two to six potential N-linked glycosylation sites. Glycosylation of at least two of these asparagine residues that sit on the extracellular loop is needed for P2X₂ receptor expression at the cell surface and its function (339, 474, 475).

A) P2X₁ receptor. In 1994, two joined papers in Nature described a new family of ligand-gated ion channels defined by the P2X receptor (60, 487). Valera et al. (487) first generated a unidirectional cDNA library from the poly(A)⁺ RNA of vas deferens that was screened in oocytes after injection with RNA transcribed in vitro. The cation-selective channel generated by expression of this P2X₁ receptor activates and desensitizes rapidly and shows a relative high permeability for Ca²⁺. The order of agonist potency is 2-MeSATP  ATP > ,-met-ATP > ADP, UTP, GTP; ACh and 5-hydroxytryptamine were ineffective. A recent report on P2X₁ receptors in acutely dissociated smooth muscle cells of the rat tail artery proposes that UTP is an effective agonist but 100-fold less potent than ATP (318). Both suramin and pyridoxal-phosphate-6-azophenyl-2',4'-disulfonic acid (PPADS), but not amiloride, block this current. The conductance, 19 pS at negative potential, shows anomalous inward rectification. These electrophysiological properties are close to those observed in smooth muscle (24). The cloned 399-amino acid-long protein is mostly extracellular and contains two transmembrane domains plus a pore-forming motif that bears a striking similarity with the P domain of the inward rectifying K⁺ channel, Kv 2.1 (205). A similar molecular architecture was reported for the mechanosensitive channels of Caenorhabditis elegans (211) and the related amiloride-sensitive epithelial sodium channel (77). However, there is no primary sequence homology between these channels and the P2X purinoceptor.

B) P2X₂ receptor. Essentially the same structure and properties were reported by Brake et al. (60) for a P2X₂ channel cloned from a complementary DNA library constructed from PC12 mRNA screened in oocytes. The homomeric P2X₂ channel is most readily characterized by its insensitivity to ,-met-ATP and its lack of desensitization during ATP applications of up to 10 s. Its conductance is 21 pS at 100 mV in 150 mM NaCl. The behavior of this cloned P2X₂ receptor resembles native P2X receptors on PC12 cells but differs from those on vascular smooth muscle and vas deferens where ,-met-ATP acts as a potent agonist. ATP, ATPS, and 2-MeSATP were roughly equipotent agonists. CTP and 2'-deoxy-ATP (dATP) elicited small currents, whereas UTP, GTP, ADP, and AMP were inactive. As in neurons (96, 286), extracellular Zn²⁺ potentiates the ATP effect on the cloned receptor by shifting the EC₅₀ for ATP to 15 µM. The intracellular COOH terminus of P2X₂ receptor contains several consensus sequences for phosphorylation by protein kinase A (PKA) and by PKC. After its reexpression in HEK-293 cells, dialyzing with phorbol 12-myristate 13-acetate (PMA) fails to alter the ATP-induced cationic current. However, 8-bromo-cAMP or the purified catalytic subunit of PKA causes a reduction in the magnitude of the ATP-activated current without altering its kinetics of inactivation or its reversal potential (90).

Both groups (60, 487) noticed the similarity of the P2X₁ and P2X₂ receptors with a partial cDNA called RP-2 that was isolated from cells induced to die (358). This led them to suggest that extracellular ATP might initiate programmed cell death as is known from the application of ATP to various cells.

C) P2X₃ receptor. A single cDNA (P2X₃ receptor) was obtained from a rat dorsal root ganglion library (84, 285). Both ATP and ,-met-ATP evoked fast-activating and rapidly desensitizing current. However, only the coexpression of P2X₃ and P2X₂ yielded ATP-activated currents that closely resembled the ,-met-ATP-sensitive, nondesensitizing current in sensory neurons. Direct evidence for heteromeric assembly of P2X₂ and P2X₃ receptors was provided by Baculovirus expression (400). That led to the new idea that these channels are formed by subunit heteropolymerization at variance with the homomeric channels formed by the cloned P2X₁ and P2X₂ receptors (60, 487). It is notable that after expression of these cDNA, the irreversible block by PPADS requires lysine at position 249 of P2X₁. P2X₃ lacks this lysine, and the inhibition by PPADS is fast and shows rapid recovery (285).

D) P2X₄ receptors. P2X₄ was initially cloned from rat brain (43, 447). It is distantly related to P2X₁, P2X₂, and P2X₃. Its expression gives an ATP-activated cation-selective channel that is highly sensitive to Ca²⁺ and whose agonist sensitivity is increased by Zn²⁺ without altering maximal response (447). Its conductance is 9 pS. This ligand-gated channel is activated by ATP > ATPS > 2-MeSATP > ADP ~ ,-metATP. However, a distinct characteristic is its insensitivity to the currently used P2X-purinoceptor antagonists, suramin, PPADS, and reactive blue 2 (RB2). Surprinsingly, however, the agonist response of the mouse ortholog of the P2X₄ receptor, mP2X₄, is potentiated by suramin, RB2, and in part by PPADS (478).

E) P2X₅ and P2X₆ receptors. The ATP-induced current after expression of the P2X₅ receptor in HEK-293 cells showed rapid activation with minimal desensitization and a lack of effect of ,-met-ATP (99). Like with P2X₁ and P2X₂, this current was inhibited by suramin and PPADS. Together with P2X₁, P2X₅ forms an heteromeric receptor (477). The same group (99) reported the cloning of the P2X₆ purinoceptor that, like P2X₄, was not sensitive to the antagonists suramin and PPADS.

F) P2X₇ receptor. The P2X₇ presently cloned from rat and human macrophages and brain (406, 460) shows a unique feature with its long intracellular COOH terminal, 240 amino acids of which at least 177 are involved in the induction of the nonselective cation channel and do not contain any known signaling motifs. P2X₇ receptors share many similarities with previously named P_2Z receptors through which ATP permeabilizes macrophages and other cells and leads to cell death (132). Thus P2X₇ and P2Z show both high sensitivity to BzATP relative to ATP and marked potentiation of the responses with reduced external divalent cation concentration so that ATP⁴ could be considered as the agonist (98). Indeed, the binding of ATP⁴ to P2X₇ receptors induces within milliseconds the opening of a channel selective for small cations and within seconds a larger pore whose single-channel conductance is 409 pS in macrophages and which is permeable to molecules up to 900 Da (108, 460). These effects of ATP are antagonized by PPADS not by suramin (87).

Several experimental situations suggest a role for connexin43 (Cx43) in the ATP-induced cell leakage of cations and small molecules. First, mouse macrophage J774 cells that do not express Cx43 are ATP resistant. It was, thus, suggested that Cx43 forms "half-gap junctions" in response to extracellular ATP (40). Second, both ATP and low-Ca solutions similarly evoked dye leakage from Novikoff hepatoma cells (293). The latter solution is known to induce hemigap-junction channels with single-channel conductance of 145 pS in catfish retina (124). Third, in HEK-293 cells overexpressing Cx43, as well as in ventricular cardiomyocytes, low-Ca solutions and metabolic inhibition open hemichannels (234). The latter case represents a physiopathological situation during which, not only cations, but also ATP could leak out of the cardiac cells and might induce autocrine stimulation.

P2X₇ receptors are insensitive to UTP and inhibited by isoquinolene derivatives 1-(N,O-bis[5-isoquinolinesulphonyl]-N-methyl-L-tyrosyl)-4-phenylpiperazine (KN62), oxidized ATP (460), and calmidazolium (88, 224, 495). P2X₇ receptors in macrophages and lymphocytes cause activation of phospholipase D (143) and modulate lipopolysaccharide signaling and inducible nitric oxide synthase expression in macrophages (220).

A) HOMO- AND HETEROMERIC RECEPTORS. Like P2X₂ and P2X₃ that combine to form a unique heteromeric channel (285, 400), with specific properties (496, 497), others assemble. In cells expressing the heteromeric P2X_1/5, ,-met-ATP evoked biphasic currents with a pronounced nondesensitizing plateau phase (477). The heteromeric assembly was suggested by in situ hybridization studies (280) and is confirmed by coimmunoprecipitation (477). Similarly, P2X₄ and P2X₆ are major subunits with highly overlapping mRNA distribution in the mammalian central nervous system (99).

A systematic study of subunit P2X coassembly has been tested by protein-protein interactions using a coimmunoprecipitation assay (476). P2X₄ and P2X₇ do not form heteromers, whereas P2X₆ needs to be associated in heteromers with P2X_1,2,4,5 to form an active channel. It has been proposed that a novel structural motif of quaternary structure of P2X receptors together (as with voltage-dependent Na⁺, Ca²⁺, or K⁺ channels) with a bundle of -helices contributed by the putative transmembrane segment M₂ near the COOH terminal (405) might lead to a tetrameric (as with the nicotinic channel) or pentameric organization that forms a pore. In another elegant study, it is suggested that P2X₁, P2X₄, or P2X₃ receptors form stable trimers (344). The dose-response curves for ATP activation of the 30-pS P2X₂ channel showing a Hill coefficient of 2.3 also suggests the channel to be a trimer (131).

B) PERMEABILITY. The substituted cysteine accessibility method was used to identify parts of the molecule that form the ionic pore of the P2X₂ receptor (405). L338 and D349 are on either side of the channel gate with D349 located near the middle of P2X₂ channel. It is a negatively charged amino acid conserved among all the seven P2X receptors. It is thus possible that D349 is the site of permeant cation binding and be responsible for ionic selectivity (131). The asparagine residue Asn-333 was found to regulate the conductance such that the unitary conductance of 80 pS in 100 Na⁺ was roughly halved when it was replaced by isoleucine (N333I) (336).

The mechanism of inward rectification has been extensively studied in several types of channels. Small peptides that formed the channel might aggregate to lead to an intrinsic inward rectification (252). More often inward rectification results from open channel block by Mg²⁺ (310) and polyamines (296, 343). The ATP-receptor subunit P2X₂ expressed in Xenopus oocytes and in HEK-293 cells shows profound inward rectification in all patch-clamp recording conditions. That suggests single-channel conductance may decrease when membrane potential becomes more positive, and furthermore, there is a substantial contribution of voltage-dependent gating to inward rectification of the steady-state current-voltage relation (537). P2X activation by external ATP (see sect. IVA1) was suggested.

Externally applied ATP activates a voltage-independent conductance on single smooth muscle cells dispersed from rabbit ear artery (31). This is also true in amphibian atrial cells (167). However, weak inward rectifying properties are shown for the ATP-induced current in rat ventricular, guinea pig atrial (204, 311, 424), and rabbit sinoatrial node (SAN) cells (436).

With the use of a combination of whole cell patch-clamp and fura 2 fluorescence measurements in HEK-293 cells transfected with hP2X₄, it was determined that the recombinant channel allows a substantial amount of Ca²⁺ to permeate. The 8% current carried by Ca²⁺ is very close to the value previously reported for native ATP-induced currents in sympathetic neurons (411). Such a high Ca²⁺ permeation through hP2X₄ receptors suggests that activation of these proteins may directly carry Ca²⁺ that will contribute to synaptic transmission (94).

The Ca²⁺ permeability of P2X₁ receptors is greater than that of P2X₂ receptors (P_Ca/P_Na 3.9 and 2.2, respectively) despite no difference between the two receptors with respect to their permeability to a series of monovalent organic cations (148). In later work, advantage was taken of the clearly different properties of P2X₂ and P2X₃ (,-met-ATP sensitivity and desensitization) to compare Ca²⁺ permeability in homomeric and heteromeric expressed receptors (496). P2X₃ has a relatively low Ca²⁺ permeability (P_Ca/P_Na 1.2-1.5) that dominates the heteromer P2X_2/3 Ca²⁺ permeability while the heteromeric receptor is nearly as sensitive to external Ca²⁺ inhibition as the P2X₂ receptor. These properties reinforced the view that the native ,-met-ATP-sensitive receptor in nodose ganglion neurons is a P2X_2/3 heteromultimer (496). P2X_2/3 desensitization is much slower than that of P2X₃ subtype alone; thus the heterogeneous expressed P2X_2/3 acquires more effective Ca²⁺ dynamics than P2X₂ or P2X₃ receptor alone (481).

C) DESENSITIZATION. P2X receptors can be divided into two broad groups according to whether they show fast desensitization within 100-300 ms or slowly if at all (Table 1). Desensitizing currents and the mechanisms of P2X-receptor desensitization have been recently reviewed (403). Briefly, rapidly desensitizing P2X receptors are activated by ATP, ,-met-ATP, and 2-MeSATP. They include recombinant P2X₁ and P2X₃. The nondesensitizing ,-met-ATP-insensitive P2X receptors are the expressed cloned P2X₂, P2X₄, P2X₅, P2X₆, and P2X₇ receptors. Several mechanisms have been proposed to account for desensitization. Desensitization implies the first or the second hydrophobic domain since substitution of these domains in chimeric P2X₁ or P2X₂ receptors confer the characteristic (516).

The rate of desensitization of the heterologously expressed P2X₁ receptor channel in HEK-293 cells has been reported to be 50-150 ms, while it is 5-10 s when recorded from Xenopus oocytes (147, 516). A slower rate of desensitization of the P2X₁ receptor from rat vas deferens stably expressed into HEK-293 cells was observed by the second day of culture. This effect was reversed by cytochalasins B and D (362). Mutations, M332I and T333S, identified in the porelike domain near or within the second putative transmembrane domain also prevented changes in kinetics.

Desensitization of P2X₂ receptors expressed in either Xenopus oocytes or HEK cells is markedly accelerated by increasing ATP concentrations (538). However, this concentration-dependent effect could in part be attributable to the concomitant acidification that occurs when high concentrations of ATP are dissolved in weakly buffered solutions (456). Desensitization of P2X₂ receptor was proposed to be controlled by alternative splicing (61). The splice isoform P2X_2b or P2X_2-2R, which lacks a stretch of COOH-terminal amino acids (Val370-Gln438), exhibits rapid and complete desensitization, whereas the wild-type channel desensitizes slowly (268, 442). The Pro373-Pro376 sequence of P2X_2R represents a functional motif that is critical for the development of the slow desensitization profile (267).

The truncated P2X₃ clone lacking the NH₂-terminal intracellular region expresses functional channels that do not desensitize in oocytes (254). Desensitization of the ATP-gated cation channel P2X₃ is abolished by removal of external Ca²⁺ or by cyclosporin pretreatment (254). The rate of desensitization is also decreased by injection of the autoinhibitory peptide CaNA457-481 in the oocyte. Thus it is thought that P2X₃ desensitizes through a Ca²⁺-dependent calcineurin-mediated phosphorylation on NH₂-terminal residues (254).

Most efforts in cloning of P2X receptors and their tissue distribution has been devoted to the brain. Much less is known for the heart and cardiomyocytes in particular. Immunoreactivity has been first reported with antisera against P2X₁ receptors on rat cardiac tissues that showed positive staining at the intercalated disk (501).

One of the most remarkable characteristics of the rat P2X₃ receptor was its apparent tissue expression pattern with transcripts restricted to nociceptive neurons (84, 285). However, major species difference seems to exist, since P2X₃ was found in human heart (173) despite the fact that this gene is highly conserved from rat to human, coding for channel proteins that share 93% identity. The majority of sequence variations are located within the putative extracellular loop. P2X₃ is also found in human fetal heart where it appears to be the most abundant P2X-receptor subtype (47). Electrophysiological characterization showed however minor differences with an inhibitory potency of suramin five times lower in human than in rat (IC₅₀ 15 vs. 3 µM, respectively) (173). It is also remarkable that hP2X₃ had a high affinity for CTP (EC₅₀ ~20 µM). The authors could not rule out the possibility that the detection of human P2X₃ RNA in heart and spinal cord reflects neuronal transcripts located in primary sensory afferents densely innervating these structures.

In situ hybridization histochemistry was recently performed together with RT-PCR on the rat cardiovascular system (348). In heart sections, the authors noticed that mRNA transcripts for P2X₁, P2X₂, and P2X₄ all colocalized in smooth muscle cells of coronary vessels with no specific apparent positivity in myocardium. However, RT-PCR from microdissected tissues from various areas of the heart confirm the presence of P2X₁, P2X₂, and P2X₄ receptor mRNAs, with strong signals in the atria and only the typical band for P2X₄ in ventricles (348). Furthermore, two splice variants of the P2X₂ were identified. The discrepancy between the PCR and in situ hybridization data might result from a too low level of mRNA expression for detection or, as well, tiny blood vessels and nerve endings may also contribute.

Northern blot and RT-PCR analysis demonstrate P2X₄ transcripts in many tissues of the rat including blood vessels and heart (447). In a subsequent study, Garcia-Guzman et al. (171), characterization of recombinant human P2X₄ receptor reveals pharmacological differences in the two species. The homology to the rat P2X₄ receptor shows 87% identity. Zn²⁺ increases the apparent gating efficiency at low concentration (5-10 µM) but inhibits the ATP-evoked current at 100 µM and higher. hP2X₄ and rP2X₄ receptors display similar agonist potency profiles, but the human receptor has a notably higher one.

The P2X₅ purinoceptor was cloned in rat heart (172) simultaneously to its cloning in brain (99). Like in brain, functional expression of the recombinant rat P2X₅ receptor shows a current that resembles mostly the P2X₂ phenotype: slow desensitization, inhibition by PPADS and suramin, and no activation by ,-met-ATP. The EC₅₀ for ATP is 8 µM. The rank order of potency is ATP  2-MeSATP > AMP ~ ADP > dATP ~ ,-met-ATP. No current is recorded after application of ,-met-ATP, CTP, GTP, UTP, or adenosine, each at 500 µM (172).

P2X₁ receptor mRNA level is increased 2.7-fold in rats under congestive heart failure (216).

To date a total of 13 P2Y receptor-like DNA sequences have been cloned (P2Y_1-11, tp2Y, and fb1). However, the P2Y₇ receptor is definitely a leukotriene B₄ receptor, and there is no functional evidence that P2Y₅, P2Y₉, and P2Y₁₀ receptors are nucleotide receptors. Therefore, there are currently five genuine human P2Y receptors (P2Y_1,2,4,6,11). It is most likely that the chicken P2Y₃ receptor is the avian ortholog of the P2Y₆ receptor, whereas Xenopus P2Y₈ and tp2y are probably orthologs of the P2Y₄ receptor. Members of the P2Y-receptor family couple to heteromeric G proteins which, in turn, activate intracellular second messenger systems to modulate the physiological function of the cells. All have a wide tissue distribution (Table 3).

A number of specific reviews have appeared (18, 44, 59, 100, 191, 255, 403).

A) P2Y₁ receptor. The first cloning of a P2Y receptor from chick brain was reported in 1993 (514). Subsequently, P2Y₁ was isolated from a range of species including rat, mouse, bovine, and human with a wide tissue distribution including heart (12, 47, 199, 473, 511). Pharmacological characterization of the P2Y₁ receptor indicates that only purines are active. UTP and derivatives are not active at this receptor subtype (443). More precisely, the P2Y₁ receptor had been initially characterized by an agonist potency order of 2-MeSADP > 2-MeSATP > ADP > ATP (57, 58, 427, 514). However, this has been questioned (198, 283). Whereas ADP and 2-MeSADP are potent full agonists, purified ATP and 2-MeSATP rather act as weak antagonists on the heterologously expressed hP2Y₁. This observed adenine nucleoside diphosphate specificity of the P2Y₁ receptor is very similar to the pharmacological selectivity of the P_2T receptor in platelets (217). It is now well established that part of the ADP effect on platelets (calcium release, shape change) is mediated by the P2Y₁ receptor. Other effects (cyclase inhibition) are mediated by a still uncharacterized distinct receptor, whereas aggregation requires both receptors (283). The apparent discrepancy with ATP being a weak but full agonist on P2Y₁ expressed in HEK and an antagonist on the same receptor expressed in Jurkat T cells was analyzed in a recent study and attributed to differences in the degree of P2Y₁-receptor reserve (361). In another recent work, adenosine 2'-phosphate-5'-phosphate (A2P5P) and adenosine 3'-phosphate-5'-phosphate (A3P5P) inhibited ADP-induced platelet shape change and aggregation and competitively antagonized Ca²⁺ movements in response to ADP in fura 2-loaded platelets, a B10 clone of brain capillary endothelial cells, and Jurkat cells expressing the human P2Y₁. In contrast, these compounds had no effects on ADP-induced inhibition of adenylyl cyclase in platelets or B10 cells (197).

B) P2Y₂ receptor. Concurrently in 1993, a receptor was isolated from mouse neuroblastoma cells and found to be activated by UTP and by ATP with equal potency and efficacy. ATPS and ITP are less potent while 2-MeSATP and ,-met-ATP are weak partial agonists (299). Orthologs of P2U, now named P2Y₂, have subsequently been isolated from human and rat (365, 511). P2Y₂ is expressed in a wide variety of tissue including heart (81, 407, 511). After expression in Xenopus oocytes, P2Y₂ coupled to two classes of G proteins to increase an endogenous Ca²⁺-dependent Cl channels and mediate a pertussis toxin-sensitive increase in the inward-rectifier K⁺ channels of the Kir3.0 subfamily (330).

C) P2Y₃ receptor. The P2Y₃ was cloned from chick brain and exhibits the following rank order of agonist potency: UDP > UTP > ADP > ATP (512). It is most likely the avian ortholog of the mammalian P2Y₆ receptor.

D) P2Y₄ receptor. Another clone originally thought to be the pyrimidinoceptor, P2Y₄, was isolated from human genomic DNA libraries (104, 340). The hP2Y₄ receptor is highly selective for UTP, whereas ATP, ADP, and ITP appear to be weak partial agonists (104, 340, 342). This rules out P2Y₄ as a true pyrimidinoceptor (440). It is not antagonized by suramin, nor by PPADS (but see for rP2Y₄), and is blocked by RB2. Pyrimidinoceptors were mainly found in rat sympathetic neurons. Recently, a rP2Y₄ has been cloned (48, 513). It shares 83% overall identity with the hP2Y₄ and is less related to the rP2Y₂ with 51% identity. Its agonist profile is close to the native P_2U receptor: ATP ~ UTP > Ap₄A > ATPS > 2-MeSATP. It is also strongly activated by ITP and UDP, although the latter could be a partial agonist. These observations, including suramin efficiency, help differentiating the rP2Y₄ from the orthologs of P2Y₂ (48, 154). ATP and UTP act on the same receptor, since they show cross-desensitization. Besides, the hP2Y₄ is among the few G protein-coupled receptors to show no N-glycosylation consensus sequence (104), but the rat ortholog does (513).

E) P2Y₆ receptor. The P2Y₆ was cloned from a rat aortic smooth muscle library and from a human placenta cDNA library; the two orthologs show 88% amino acid identity (80, 103). As for P2Y₄ receptors, the selectivity of P2Y₆ receptors for UDP had to be reexamined because of the use of UDP preparations contaminated by UTP and the degradation of UTP into UDP during incubation of the cells (342). With these precautions the P2Y₆ receptor is activated most potently by UDP and weakly by UTP, ATP, and ADP.

F) P2Y₁₁ receptor. The newly cloned human P2Y receptor provisionally called hP2Y₁₁ is characterized by considerably larger second and third extracellular loops (101). It exhibits only 33% homology with hP2Y₁, its closest homolog, and 28% homology with hP2Y₂. It shows one potential site for N-linked glycosylation and two potential sites for phosphorylation by PKC or calmodulin-dependent protein kinases. hP2Y₁₁ couples positively to both phosphoinositide and adenylyl cyclase, a unique feature among the P2Y family. Following stable expression in 1321 N1 astrocytoma and CHO-K₁ cells, the rank order of agonist potency for the two pathways is ATPS ~ BzATP > dATP > ATP > adenosine 5'-O-(2-thiodiphosphate) (ADPS) > 2-MeSATP > ADP with UTP and UDP being inactive, thus showing hP2Y₁₁ is presently the most adenine nucleotide selective P2Y receptor (106).

G) DETERMINANTS OF ATP BINDING ON P2Y PURINOCEPTORS. Site-directed mutagenesis of P2Y₂ led to the suggestion that the charged amino acids His-262, Arg-265 in TM6 and Arg-292 in TM7 interact with the phosphate groups of the nucleotides (145). The amino acid sequence of hP2Y₄ shows some similarities with the P2Y₂ receptor at sites directly involved in the binding of negatively charged phosphate groups. Three residues are conserved: His-262, Arg-265, and Arg-292, whereas Arg-265 is replaced by a Lys-259 in P2Y₁ as in the P2Y₆ (105, 340). Moreover, Lys-289 residue also plays a major role, since its substitution by Arg shifts the preference of the P2Y₂ receptor for triphosphate nucleotides to diphosphate nucleosides (145). Note that this Lys residue is conserved in both P2Y₄ and in P2Y₆; the latter which, however, has a clear preference for UDP.

Activation of the recombinant P2Y₁ receptor mediates IP₃ formation and increases intracellular Ca²⁺ concentration ([Ca²⁺]_i), but it does not change cAMP (443). However, in a recent study BzATP, an antagonist of rat and human P2Y₁ expressed in Jurkat cells, prevents the ADP-induced inhibition of the cAMP pathway (493). IP₃ and [Ca²⁺]_i increases can stimulate a variety of pathways including PKC, phospholipase A₂ (PLA₂), and nitric oxide synthase which subsequently can activate other pathways such as phospholipase D (PLD) and mitogen-activated protein kinase (MAPK).

Activation of P2Y₄ and P2Y₆ receptors also leads to the formation of inositol phosphates (103, 104) and a rise in [Ca²⁺]_i (80, 340). However, the two receptors seem to involve distinct G proteins. The P2Y₄ but not the P2Y₆ response was inhibited by pertussis toxin (80, 102, 409). hP2Y₁₁ has the unique property of simultaneously activating the adenylyl cyclase with the same rank order of potency of agonists (101, 106). Surprisingly, the ATP derivative AR-C67085, a potent inhibitor of ADP-induced platelet aggregation, was the most potent agonist at the recombinant hP2Y₁₁ receptor (106).

The expression of P2Y_1,2,4,6 receptor transcripts was reported by RT-PCR in rat heart (511). All receptor sequences could be amplified from neonatal whole heart, with P2Y₆ appearing the most abundant transcript of the four. However, in neonatal cardiomyocytes, P2Y₁ is expressed at higher levels than the others. In adult myocytes, P2Y₁, P2Y₂, and P2Y₆ could be amplified while P2Y₄ could not be detected (511). Two major conclusions can be made from this work: 1) the need to specifically work on isolated cardiomyocytes, and 2) P2Y expression varies during development with the arrest of P2Y₄ expression in adult cardiomyocytes. P2Y₂, P2Y₄, and P2Y₆ receptors were recently cloned in fetal human heart using degenerated oligonucleotides (47). Other works reporting the presence of one or the other of the P2Y receptors are quoted in Table 2. P2Y₁₁ was not found in human heart using Northern blot analysis (101). In a recent study however, using a new quantifying RT-PCR protocol, P2-receptor mRNA expression was compared in control and rats under congestive heart failure. In the sham-operated rats, P2Y receptors are expressed at a higher level than P2X₁ receptors, with P2Y₆ being the most abundant. In failing hearts, a prominent change was seen: P2X₁ and P2Y₂ receptor mRNA levels were increased 2.7 and 4.7, respectively (216). Extending their study to adult human myocardium, the authors could detect P2X₁, P2Y₁, P2Y₂, P2Y₆, and P2Y₁₁ receptors in both right and left atria and ventricles, while no P2Y₄ receptor was seen.

The selectivity and activity of adenine dinucleotides for neuronally derived recombinant P2 purinoceptors were studied using P2X₂ and P2Y₁ subtypes expressed in Xenopus oocytes (378). Ap₄A is as active as ATP but less potent (EC₅₀ ~15 and 4 µM, respectively) on P2X₂ receptors. Other adenine dinucleotides are inactive. In a previous work it had been shown that Ap₅A is a partial agonist at the human ortholog hP2X₁ (147). However, Ap₅A potentiates the ATP responses but not the Ap₄A response at the P2X₂ subtype with an EC₅₀ of ~3 nM, and Ap₅A also enhances the efficacy of suramin (378). In a recent work, Ap₅A also potentiated the ATP response in rat cerebellar astrocytes (233). Such an effect appears to be mediated by the metabotropic P2Y receptors. At the P2Y₁ subtype, Ap₃A is equipotent and active as ATP. Ap₄A is a weak partial agonist and other dinucleotides are inactive. In another study, Ap₄A is reported to be a potent agonist at P2Y₂ but not P2Y₁, P2Y₄, and P2Y₆ expressed in 1321 NI human astrocytoma cells (341). Thus some dinucleotides have the capacity to potentiate ATP responses at both P2X and P2Y receptors.

To date, in cardiac tissues, only the presence of Ap₄A receptor has been reported (202, 203, 504), with at least 77% of the active receptors on the plasma membranes (502). Recent work demonstrates photocross-linking of an Ap₄A derivative with a 50-kDa polypeptide, suggesting a homogeneous population of receptors (42). At the high-affinity binding site, the apparent K_d values for Ap₄A and ATPS are 0.08 and 0.04 µM, respectively; there is also a low-affinity site for ATPS with a K_d at 1.0 µM. The P2X agonist ,-met-ATP did not compete effectively with these two agonists and indicates that the Ap₄A receptor is a P2Y receptor in cardiac plasma membranes. The same group had previously suggested Ap₄A binds to a specific 30-kDa polypeptide dinucleotide receptor. This observation resulted from receptor proteolysis. Ap₄A receptor was shown to undergo at least two proteolytic processing steps, one of which is carried out by a serine protease, and this serine protease is required for receptor activation (202, 503, 504). Specific dinucleotide receptors activated by Ap_nAs, but not ATP or UTP, have been suggested on synaptic terminals in guinea pig diencephalon and cerebellum (379). This follows the identification of binding sites that are highly selective, or even specific, for Ap_nA as opposed to mononucleotides. This possibly represents P2D receptors, P2Y_ApnA receptors, or also P4 receptor that are distinct from those activated by ATP to induce an elevation in intrasynaptosomal Ca²⁺.

Despite much effort there is a still strong need of potent and subtype-specific P2 receptor agonists and antagonists. Several substances that display selectivity for P2X or P2Y subtypes have been recently carefully reviewed (403) and are summarized in Tables 1 and 2. In the following I would just like to draw attention to side effects of most of the presently available compounds that make their use difficult even in isolated cells.

It had been already recognized that the ecto-nucleotidase activity accounts, in most part, for the difference in the concentration curves of P2X agonists in vascular smooth muscle contraction and electrophysiological response, since ATP degradation products might act besides ATP itself (250). Inhibitors of the ecto-ATPases include NaN₃ or AP₅A as well as many of the so-called purinergic antagonists, suramin, PPADS, RB2, FLP 66301, and FLP 67156 (39, 66, 82, 110, 111, 223, 524, 540). All the nonhydrolyzable ATP analogs also do (83, 377). Consequently, such an inhibition of ecto-ATPases might, in part, account for the observed facilitation effects of ATP in tissues or in cell incubation (82). Suramin is also an active inhibitor of protein-tyrosine phosphatases (530), a side effect of potential importance knowing the involvement of tyrosine kinase pathways after ATP stimulation (394). PPADS also appears to inhibit UTP- and ATP-induced Ca²⁺ mobilization by a nonspecific mechanism independent of P2Y-receptor recognition that might involve the inhibition of intracellular IP₃ channels (494). Moreover, there might be even species selectivity. Thus rat recombinant P2X₄ and P2X₆ receptors are not blocked by PPADS, but the human homolog of the P2X₄ receptor is (171).

Recently, 2',3'-O-(2,4,6-trinitrophenyl)adenosine 5'-triphosphate (TNT-ATP) has been reported to inhibit specifically P2X₁ and P2X₃ and heteromeric P2X_2/3 with an IC₅₀ of 1 nM. However, its use in tissue might be of limited interest due to its fast breakdown (497).

BzATP, an antagonist of P2X₇ receptor, is also a photoaffinity probe that binds covalently to the nucleoside sites of ATPases (20, 360); it prevents ATP binding to -sarcoglycan, a skeletal muscle ecto-ATPase (38), and to P2Y₁ receptors (493).

Quinidine was first used by Burnstock (69) as a P2-receptor inhibitor. It is interesting to note that quinidine antagonizes the ATP-induced nonselective cationic current in rat ventricular cells (91) and the late phase of positive inotropy in guinea pig atria (135) in line with the view that a part of contractile activity might result from P2X-mediated Ca entry (169).

DIDS is a commonly used anion transport inhibitor. It has also been reported to block a number of presumably purinoceptor-mediated responses. Indeed, DIDS causes a long-lasting blockade of P2X receptor activity in rat vas deferens (65, 112, 319). This effect is in line with the fact that DIDS decreases the binding of [³²P]ATP in rat parotid acinar cells (319). DIDS is now known to be a P2-receptor antagonist (65, 147, 445).

Ion selectivity of the P2X-receptor channels resembles that of nicotinic ACh, glutamate-activated, or serotonin-activated channels. It might be of interest to test blockers of these channels (local anesthetics QX314, QX222, D-tubocurarine, dizocilpine, and phencyclidine). In fact, D-tubocurarine is a blocker of P2X₂ receptors (60).

	III. PHYSIOLOGICAL AND PHYSIOPATHOLOGICAL EFFECTS OF EXTRACELLULAR ADENOSINE 5'-TRIPHOSPHATE

Top Previous Next References

Most initial reports describe negative inotropic effects after applying ATP in mammalian hearts, particularly in the atrium (75, 137, 159, 207, 212). These effects should be attributable to A₁-adenosine receptor activation after ATP breakdown in the tissue. Later studies on guinea pig and rat atria revealed positive inotropic effects of ATP that developed after a transient rapid decrease (135, 168, 169, 306). In these studies, ATP, ADP, AMP, adenosine, ,-met-ATP, ,-met-ATP, as well as UTP were shown to have this dual effect. Desensibilization by long exposure to ,-met-ATP (135), suramin, but not RB2 could antagonize the positive inotropic effect (168), whereas 8-phenyltheophylline (8-PT) or 8-cyclopentyl-1,3-dipropylxanthine (DPCPX) prevented the negative effects (168, 306). It is noteworthy that 2-MeSATP induces only a negative inotropic effect (168). A similar dual inotropic effect of ATP is observed in human atrial strips (J. Alvarez and G. Vassort, unpublished results). Pyridoxal 5'-phosphate, an active form of vitamin B₆, shows antagonism toward ATP-induced positive inotropic effect in perfused rat heart, with both pyridoxal and phosphate moities being essential for the action (508).

Fifty years ago, Green and Stoner (185) reported that after a brief period of cardiac arrest, adenine nucleotides had a strong positive inotropic effect in isolated perfused mammalian hearts. They reported that increasing states of phosphorylation of the adenosine molecule correlated well with the increasing inotropic effects. They suggested that the adenine nucleotides exerted directly their effect on the myocardium rather than through vasodilatation. Several reports established ATP as a full positive inotropic agent in frog ventricles (11, 161, 183, 489 and see references therein). ATP also induces positive inotropic effects in the ventricle of axolotl (321) but has no effect in turtle heart, dogfish, and trout ventricle (320, 322, 323). Positive inotropy in frog heart is also induced by UTP, ITP, CTP, GTP, and their derivatives (489, 490, 527). The fast initial positive phase is followed, after a transient decrease, by a secondary increase of variable amplitude according to the agonists. ATP and its derivatives exert clear positive inotropy in rat papillary muscle (282, 421) and enhance cell shortening in various mammalian ventricular cardiomyocytes (Fig. 1) (91, 116, 381, 421).

View larger version (22K): [in this window] [in a new window]   Fig. 1. Inotropic effects of ATP. A: in the frog heart, the application of ATP (20 mg%) is able to compensate for the partial removal of extracellular Ca2+. [From Antoni et al. (11). Copyright 1960 Springer-Verlag.] B: ATP at 1 µM in the presence of 1 mM Ca2+ induces parallel changes in transient Ca2+ and cell shortening in an indo 1-loaded rat ventricular myocyte. [From Danziger et al. (116).] C: negative and positive inotropic effects of ATP in a guinea pig left atria, respectively, in absence and presence of 0.1 µM 1,3-dipropyl-8-cyclopentylxanthine (DPCPX). [From Mantelli et al. (306).] D: positive inotropic effects of 100 µM ATP on an auricular strip isolated from a pertussis toxin-treated rat before and after 0.5 µM isoproterenol (ISO) stimulation. [From Scamps et al. (421).]

A beneficial effect of adenine nucleotides on contractility of postischemic papillary muscles is long known; a situation where ATP effect was tentatively attributed to its high-energy supplier properties such as it would contribute to high-energy phosphate replenishment (434). However, with the present knowledge, positive inotropy in control and diseased cells could be attributable to an increase in L-type long-lasting Ca²⁺ current (I_CaL) (204, 421-423) that mediates an increase in Ca²⁺ transient. Moreover, ATP could enhance SR Ca²⁺ release, although this mechanism is unclear (41, 91, 204, 389). Other mechanisms have been suggested as well. They include an interaction of ATP an