I. Introduction

Top Next References

Extracellular purines and pyrimidines have important and diverse effects on many biological processes including smooth muscle contraction, neurotransmission, exocrine and endocrine secretion, the immune response, inflammation, platelet aggregation, pain, and modulation of cardiac function. The concept of purines as extracellular signaling molecules was instigated by Drury and Szent-Györgyi in 1929, in a comprehensive report showing that adenosine and adenosine 5'-monophosphate (AMP), extracted from heart muscle, have pronounced biological effects, including heart block, arterial dilatation, lowering of blood pressure, and inhibition of intestinal contraction. Gillespie, in 1934, drew attention to the structure-activity relationships of adenine compounds, showing that deamination greatly reduces pharmacological activity, and that removal of the phosphates from the molecule influences not only potency, but also the type of response. Removal of phosphates was shown to increase the ability of adenine compounds to cause vasodilatation and hypotension, and ATP caused an increase in rabbit and cat blood pressure that was rarely or never observed with AMP or adenosine. Furthermore, ATP was shown to be more potent than AMP and adenosine in causing contraction of guinea-pig ileum and uterus (Gillespie, 1934). This was the first indication of different actions of adenosine and ATP and, by implication, the first indication of the existence of different purine receptors.

Early investigations into the effects of adenosine and ATP were made on a variety of tissues, but particularly the heart and vasculature (Gaddum and Holtz, 1933; Emmelin and Feldberg, 1948; Folkow, 1949; Green and Stoner, 1950). Initial studies on the effects of extracellular UTP also focused on its cardiovascular effects (Hashimoto et al., 1964; Boyd and Forrester, 1968; Urquilla, 1978; Sakai et al., 1979). Other early lines of purine research concerned the effects of purines on platelet aggregation (Born, 1962) and on mast cells (Cockcroft and Gomperts, 1980). Diverse responses to extracellular purines and pyrimidines have now been documented in a wide range of biological systems, from single cells to whole organisms, and include smooth muscle contraction, neurotransmission in the peripheral and central nervous system, exocrine and endocrine secretion, the immune response, inflammation, platelet aggregation, pain, and modulation of cardiac function (Burnstock and Kennedy, 1986; Gordon, 1986; Seifert and Schultz, 1989; Burnstock, 1990; Olsson and Pearson, 1990; Ralevic and Burnstock, 1991a; Jacobson et al., 1992b; Dubyak and el-Moatassim, 1993; Dalziel and Westfall, 1994; Fredholm, 1995; Burnstock and Wood, 1996; Ongini and Fredholm, 1996; Sebastiâo and Ribeiro, 1996).

Insight into the physiological roles of extracellular purines and pyrimidines comes from studies of their biological sources and the stimuli for their release. In this respect, an important line of research stemmed from studies showing that adenosine is released from the heart during hypoxia to play an important role in reactive hyperemia (Berne, 1963; Gerlach et al., 1963). The general hypothesis of coupling of purine release to metabolic demands via local regulation of blood flow has been applied to other tissues and includes the release of adenine nucleotides, particularly ATP, from skeletal muscle during contraction (Boyd and Forrester, 1968; Forrester and Lind, 1969).

Reports of ATP release from sensory nerves in the rabbit ear (Holton and Holton, 1953; Holton, 1959) were the first in a major line of research concerned with purines as neurotransmitters. ATP was detected in the venous perfusate following antidromic stimulation of the rabbit auricular nerve to elicit vasodilatation of the ear capillary bed, and both antidromic vasodilatation and vasodilatation to arterial infusion of ATP (but not that to other agents) were blocked by cholinesterase inhibitors (Holton and Holton, 1953; Holton, 1959). It is now known that ATP is an important neurotransmitter or cotransmitter and adenosine an important neuromodulator in both the peripheral and central nervous systems (see Stone, 1991; Burnstock, 1990; Edwards and Gibb, 1993; Fredholm, 1995). There are also hints that adenine dinucleotides may play neurotransmitter or neuromodulator roles in the central nervous system (Pintor and Miras-Portugal, 1995b).

Adrenal chromaffin granules (Cena and Rojas, 1990), platelets (Born and Kratzer, 1984; Gordon, 1986), mast cells (Osipchuk and Cahalan, 1992), erythrocytes (Forrester, 1990; Ellsworth et al., 1995), basophilic leukocytes (Osipchuk and Cahalan, 1992), cardiac myocytes (Forrester, 1990), fibroblasts (Grierson and Meldolesi, 1995b), and endothelial (Ralevic et al., 1991a, 1991c, 1995b; Bodin et al., 1992) and epithelial cells (Enomoto et al., 1994; Ferguson et al., 1997) are important sources of purines that can be released under physiological and pathophysiological conditions, which may act on the purine receptors associated with these or neighboring cells. Adenine dinucleotides are released from secretory ganules of thrombocytes, chromaffin cells and neurons, and may represent a novel class of signaling molecules (Hoyle, 1990; Ogilvie, 1992; Ogilvie et al., 1996). Not enough is known about the sources and release of pyrimidines, which limits our understanding of the role played by the widely distributed receptors that are activated by pyrimidines. However, steps toward resolving this are being made with the demonstration that UTP is released by physiologically relevant stimuli from cultured endothelial, epithelial, and astrocytoma cells (Enomoto et al., 1994; Saiag et al., 1995; Lazarowski et al., 1997a).

Purines and pyrimidines mediate their effects by interactions with distinct cell-surface receptors. Early pharmacological evidence for the existence of adenosine receptors has been provided by specific antagonism by methylxanthines of adenosine-mediated accumulation of adenosine 3',5'-cyclic monophosphate (cAMP) in rat brain slices (Sattin and Rall, 1970). "Purinergic" receptors were first formally recognized by Burnstock in 1978, when he proposed that these can be divided into two classes termed "P₁-purinoceptors", at which adenosine is the principal natural ligand, and "P₂-purinoceptors", recognizing ATP and ADP (Burnstock, 1978). This division was based on several criteria, namely the relative potencies of ATP, ADP, AMP, and adenosine, selective antagonism of the effects of adenosine by methylxanthines, activation of adenylate cyclase by adenosine, and stimulation of prostaglandin synthesis by ATP and ADP. This major division remains a fundamental part of purine receptor classification, although adenosine/P1 and P2 receptors are now characterized primarily according to their distinct molecular structures, supported by evidence of distinct effector systems, pharmacological profiles, and tissue distributions. In addition, receptors for pyrimidines are now included within the P2 receptor family (table 1) (Fredholm et al., 1994, 1996). Note that it has been recommended that "P1 receptor" and "P2 receptor" replace the "P₁/P₂-purinoceptor" terminology (Fredholm et al., 1996). The terms "adenosine receptor" and "P1 receptor" are synonymous.

Adenosine/P1 receptors are further divided into four subtypes, A₁, A_2A, A_2B, and A₃, on the basis of their distinct molecular structures and show distinct tissue distributions and pharmacological profiles. All couple to G proteins.

P2 receptors were shown to be ligand-gated cation channels (Benham and Tsien, 1987) or involved G protein activation (Dubyak, 1991), which formed the basis of their subdivision into two main groups termed P2X receptors (ligand-gated cation channels) and P2Y receptors (G protein-coupled receptors) (Abbracchio and Burnstock, 1994; Fredholm et al., 1994). Subtypes are defined according to the molecular structure of cloned mammalian P2 receptors, discriminated by different numerical subscripts (P2X_n or P2Y_n) (Burnstock and King, 1996; Fredholm et al., 1996). This forms the basis of a system of nomenclature that will replace the earlier subtype nomenclature (including P_2X, P_2Y, P_2U, P_2T, and P_2Z receptors) as correlations between cloned and endogenous receptors are established. P3, P4, and P2Y_Ap4A (or P_2D) receptors have been proposed, but evidence for their existence is based solely on the distinct pharmacological profiles exhibited by some biological tissues. As this is no longer tenable for the identification and subclassification of receptors, it remains to be determined, preferably by molecular studies, how these correlate with cloned P2 receptors, and therefore exactly how they will fit within a unifying system of purine and pyrimidine receptor nomenclature.

The main aim of this review is to present the characteristics of receptors for purines and pyrimidines within a framework whereby comparisons can be made between cloned and endogenous receptors. For the P2 receptor family this is in order to promote the conversion from a system of nomenclature that is rapidly losing its value, to a unifying system of classification based on structurally distinct cloned receptors. However, pharmacological characterization of endogenous P2 receptors is often equivocal, largely because of the current lack of selective agonists and antagonists and because of complications introduced by the common and widespread coexpression of different types of P2 receptors. Thus, it will become apparent in the present review that in assigning names to endogenous P2 receptors we have needed to strike a balance between caution (against overinterpretation) and anticipation of the direction in which this field is heading. Signal transduction mechanisms, pharmacological response profiles, receptor desensitization, tissue distribution, and biological effects of receptors for purines and pyrimidines are considered. Because ATP and ADP are rapidly degraded to adenosine, and because most cells and tissues coexpress P1 and P2 receptors, we also consider the interactions that may occur between receptors belonging to these two families. Although modulation of ecto-nucleotidase expression and activity is an important means by which to regulate purine receptor function, this aspect of purinergic signaling is not dealt with in detail in this article; the reader is referred to other reviews on the subject (Ziganshin et al., 1994a; Zimmerman, 1996).

	II. Adenosine/P1 Receptors

Top Previous Next References

The adenosine/P1 receptor family comprises A₁, A_2A, A_2B, and A₃ adenosine receptors, identified by convergent data from molecular, biochemical, and pharmacological studies (table 2). Receptors from each of these four distinct subtypes have been cloned from a variety of species and characterized following functional expression in mammalian cells or Xenopus oocytes (table 3). A₁ and A₂ receptors were described by Van Calker et al. in 1979, in studies showing that activation of these receptors by adenosine and its derivatives inhibited, via A₁, or stimulated, via A₂, adenylate cyclase activity in cultured mouse brain cells (Van Calker et al., 1979). The effects of adenosine could be antagonized by methylxanthines and the order of potency of adenosine analogs was different for the two receptors (Van Calker et al., 1979). Londos et al. (1980) independently drew similar conclusions using membrane preparations from rat adipocytes, hepatocytes, and Leydig tumor cells; the adenosine receptors coupled to inhibition of adenylate cyclase (those in adipocytes) they named R_i (corresponding to the A₁ subtype) and the adenosine receptors coupled to stimulation of adenylate cyclase (those in hepatocytes and Leydig cells) they termed R_a (synonymous with the A₂ subtype). This alternative system of nomenclature was based on "R" to designate the "ribose" moiety of the nucleoside and "i" and "a" to indicate inhibition and activation of adenylate cyclase respectively (Londos et al., 1980). The A₁/A₂ nomenclature is now used, which is fortunate because A₁ receptors act through a variety of transduction mechanisms in addition to adenylate cyclase. A_1a and A_1b receptors have been proposed (Gustafsson et al., 1990), but this subdivision of the A₁ receptor remains equivocal.

A₂ receptors are further subdivided into types A_2A and A_2B. The suggestion that A₂ receptors could be divided into two classes was originally based on the recognition that adenosine-mediated stimulation of adenylate cyclase in rat brain was effected via distinct high affinity binding sites (localized in striatal membranes) and low affinity binding sites (present throughout the brain) (Daly et al., 1983). This subdivision was supported in a study which compared the high affinity striatal A₂ binding site with a low-affinity A₂ binding site characterized in a human fibroblast cell line; the two sites were termed A_2A and A_2B, respectively (Bruns et al., 1986). Definitive evidence for the existence of these two subtypes comes from the cloning and sequencing of distinct A_2A and A_2B receptors which show distinct pharmacological profiles in transfected cells similar to those of the endogenous receptors.

Consistent with the fact that these are distinct receptors, there is a considerable lack of amino acid sequence homology between cloned A₁, A_2A, A_2B, and A₃ receptors. For example, the homology between rat A₁ and A_2B receptors is only 45% (Stehle et al., 1992), and the human A₃ receptor only shows approximately 50%, 43%, and 40% homology with human A₁, A_2A, and A_2B receptors, respectively (Linden, 1994). The homology between A_2A and A_2B receptors is also slight, being approximately 46% when these subtypes are cloned from rat and 61% when cloned from human (Stehle et al., 1992; Pierce et al., 1992).

An adenosine binding site with high affinity for 2-phenylaminoadenosine (CV 1808) (A_2A-selective agonist) in rat striatal membranes has been suggested to be a novel "A₄" adenosine receptor (Cornfield et al., 1992). The very low affinity of 2-[p-(2-carbonyl-ethyl)-phenylethylamino]-5'-N-ethylcarboxamidoadenosine (CGS 21680) and Nethylcarboxamidoadenosine (NECA) at this site were taken to indicate that this is not an A₂ receptor. However, the binding studies were carried out at 4°C (Cornfield et al., 1992), and the existence of a distinct A₄ receptor has been challenged by the demonstration that when similar binding studies are carried out at 21°C, the potency order of compounds at the striatal membrane site is characteristic of the A_2A adenosine receptor (Luthin and Linden, 1995). Furthermore, in COS cells transfected with adenosine A_2A receptors, binding studies carried out at 4°C produce an "A₄" binding profile (Luthin and Linden, 1995). This justifies the more rigorous criteria now demanded for classification of receptors, whose identity must be proved by molecular as well as by biochemical or pharmacological means.

There is a vast and rapidly growing literature on adenosine/P1 receptors; it has not been possible to do justice to this in the present review. Out of necessity, therefore, we focus on the more recent literature.

All adenosine receptors couple to G proteins. In common with other G protein-coupled receptors, they have seven putative transmembrane (TM) domains of hydrophobic amino acids, each believed to constitute an helix of approximately 21 to 28 amino acids. The N-terminal of the protein lies on the extracellular side and the C-terminal on the cytoplasmic side of the membrane. A pocket for the ligand binding site is formed by the three-dimensional arrangement of the -helical TM domains, and the agonist is believed to bind within the upper half of this pore. The transmembrane domains are connected by three extracellular and three cytoplasmic hydrophilic loops of unequal size; typically the extracellular loop between TM4 and TM5 and the cytoplasmic loop between TM5 and TM6 is extended. These features are illustrated in a schematic of the A₁ receptor in figure 1.

View larger version (34K): [in this window] [in a new window]   Fig. 1.   Schematic of the A1 adenosine receptor. In common with other G protein-coupled receptors, the A1 receptor has seven putative transmembrane domains (I-VII) of hydrophobic amino acids, each believed to constitute an -helix, which are connected by three extracellular and three intracellular hydrophilic loops. The number of amino acids comprising the extra- and intracellular loops and the extracellular N-terminal and intracellular C-terminal regions of the bovine A1 receptor are indicated in parentheses (Olah et al., 1992). The transmembrane regions comprise 23 to 25 amino acids in the bovine A1 receptor (Olah et al., 1992). The arrangement of the transmembrane regions forms a pocket for the ligand binding site. The location of histidine residues (H) in transmembrane regions VI (position 254) and VII (position 278) in the bovine A1 receptor, which are believed to be important in ligand binding (Olah et al., 1992), are indicated. Extracellular and transmembrane regions of the protein believed to be important in agonist and antagonist binding are indicated (Olah et al., 1994b,c). S-S denotes the presence of hypothetical disulfide bridges (Jacobson et al., 1993c). Glycosylation occurs on the second extracellular loop.

N-linked glycosylation often occurs on the second extracellular loop; the roles of the carbohydrate moieties of the glycosylated receptor are not clear, although suggested functions include stabilization of protein conformation, protection of proteins from proteases, and modulation of protein function. Current evidence suggests that glycosylation has no obvious influence on ligand binding (Piersen et al., 1994). The intracellular segment of the receptor interacts with the appropriate G protein with subsequent activation of the intracellular signal transduction mechanism. The third intracellular loop of the adenosine A_2A receptor seems to be the main determinant of its G protein selectivity (Olah, 1997). Phosphorylation by protein kinases of amino acid residues on the cytoplasmic domains seems to be involved in desensitization of A_2A and A₃ receptors (Palmer and Stiles, 1997a, 1997b).

The transmembrane regions are generally highly conserved, with particularly long stretches of amino acid homology being found in TM2, TM3, and TM5. Most sequence differences have been observed in a hypervariable region located at the N-terminal half of the second extracellular loop (Tucker and Linden, 1993). It is the residues within the transmembrane regions that are crucial for ligand binding and specificity and, with the exception of the distal (carboxyl) region of the second extracellular loop, the extracellular loops, the C-terminal and the N-terminal do not seem to be involved in ligand recognition (Olah et al., 1994b, 1995). A number of amino acid residues contribute, in different ways, to ligand specificity within the binding pocket. Sitedirected mutagenesis of the bovine A₁ adenosine receptor suggests that conserved histidine residues in TM6 and TM7 are important in ligand binding. Histidine 278 in TM7 seems to be particularly important because mutation of this amino acid abolishes ligand binding (Olah et al., 1992). Mutagenesis of the human A₁ adenosine receptor has shown that threonine 277 in TM7 is important in binding of the non-selective adenosine receptor agonist NECA, but has little effect on the affinity of binding of the A₁ selective agonist (R)-N⁶-(2-phenyl-1-methylethyl)-adenosine (R-PIA), or of antagonists (Townsend-Nicholson and Schofield, 1994). Modification of Glu 16 in TM1 and Asp 55 in TM2 of the human A₁ receptor alters the affinity of binding for [³H]CCPA (2-chloro-N⁶-cyclopentyladenosine) and other agonists, but does not affect antagonist binding (Barbhaiya et al., 1996). Site-directed mutagenesis of the human A_2A adenosine receptor has identified several residues in TM5-7 that are involved in ligand binding (Kim et al., 1995). Glu 13 in TM1 of the human A_2A receptor seems to be critically involved in agonist, but not antagonist recognition (Ijzerman et al., 1996).

A potential problem inherent in the methodology of site-directed mutagenesis is that changes in primary structure may cause changes in tertiary structure of the molecule. This has been addressed by studies with chimeras constructed from structurally similar, but pharmacologically different receptors. The ligand binding properties of A₁/A₃ chimeric receptors support the concept of a crucial role for histidine residues in TM6 and TM7 in ligand binding (Olah et al., 1995). In addition, a critical role in ligand binding of the distal region of the second extracellular loop has been identified, although its specific interactions are not yet clear (Olah et al., 1994b). Possible roles include direct interaction of an amino acid residue(s) within this region with the ligand, an influence on the conformation of the receptor and/or steric hindrance. Construction of chimeric human A₁ and rat A_2A adenosine receptors was used to show that TM1-4 are important in A₁ receptor agonist and antagonist binding and ligand specificity (Rivkees et al., 1995a).

Analogs with greater stability than adenosine are produced by modification of the N⁶ and C2 positions of the adenine ring and the 5'-position of the ribose moiety of adenosine, and have been used extensively in the characterization of adenosine/P1 receptors. NECA (Williams, 1989), N-[2-(4-aminophenyl)ethyl] adenosine (APNEA) (Fozard and Carruthers, 1993), and N⁶-(3-[¹²⁵I]iodo-4-aminobenzyl)-5'-N-methylcarboxamidoadenosine (¹²⁵I-AB-MECA) (Olah et al., 1994a) do not discriminate between adenosine receptor subtypes. Agonists with subtype selectivity are detailed in the sections on individual adenosine receptor subtypes and the chemical structure of some of these are illustrated in figure 2.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   The chemical structure of some agonists at adenosine/P1 receptors.

ATP and metabolically stable ATP derivatives, i.e., adenosine 5'-O-(3-thiotriphosphate)(ATPS) and ,-methylene ATP (,-meATP), can act directly as agonists at adenosine/P1 receptors in some tissues where responses are blocked by methylxanthines, but are not affected by adenosine deaminase or by blockade of 5'-nucleotidase. ,-MeATP is approximately equipotent with adenosine at mediating contraction of smooth muscle adenosine/P1 receptors of rat colon (Bailey and Hourani, 1990), and relaxation via adenosine/P1 receptors of rat duodenum (Hourani et al., 1991), and guinea-pig trachealis muscle (Piper and Hollingsworth, 1996). ATP, ATPS, and ,-meATP inhibit [³H]-NA release in a variety of tissues via receptors that are blocked by the A₁ selective antagonist 1,3-dipropyl-8-cyclopentylxanthine (DPCPX) as well as by the P2 receptor antagonist cibacron blue (Von Kügelgen et al., 1992, 1995b, 1996). ATP (Collis and Pettinger, 1982) and diadenosine polyphosphates (Hoyle et al., 1996; Vahlensieck et al., 1996) have been reported to stimulate directly adenosine/P1 receptors in guinea-pig atria, eliciting negative inotropic and chronotropic effects without prior conversion to adenosine. These effects are not consistent with the pharmacological profile of any of the established subtypes of adenosine/P1 receptor, and in some respects are similar to the profile described for the P3 receptor.

Xanthines and xanthine derivatives, including the natural derivatives theophylline and caffeine, are non-selective adenosine/P1 receptor antagonists. They are not universal adenosine/P1 receptor antagonists; xanthine-resistant relaxations to adenosine and its analogs were observed in guinea-pig aorta (Collis and Brown, 1983; Martin, 1992), rat aorta (Prentice and Hourani, 1996), guinea-pig trachea (Brackett and Daly, 1991), porcine coronary artery (Abebe et al., 1994), and guinea-pig taenia cecum (Prentice et al., 1995). Some A₃ receptors, namely those of rat, rabbit, and gerbil, are characteristically insensitive to methylxanthines, thus it is possible that the xanthine-resistant responses to adenosine described in some tissues occur following actions of adenosine at mast cell A₃ receptors and the subsequent release of vasoactive mediators. This hypothesis would predict that guinea-pig and pig A₃ receptors are also xanthine-insensitive, because xanthine-resistant responses to adenosine have been reported in these species. It would be interesting to see if these responses can be blocked by inhibitors of mast cell degranulation.

8-Phenyltheophylline and the more water soluble 8-(p-sulfophenyl)theophylline (8-SPT) (Daly et al., 1985) are more potent than theophylline at adenosine/P1 receptors, but are not subtype-selective. 8-SPT and its derivative 1,3-dipropyl-8-sulfophenylxanthine (DPSPX) do not cross the blood-brain barrier, being purely peripherally acting adenosine/P1 receptor antagonists (Daly et al., 1985) and thus can be used to discriminate between central and peripheral adenosine receptors. A number of xanthines and non-xanthines identified as adenosine receptor antagonists with reasonable subtype selectivity are described below (see Sections III.F., IV.F., and VI.F.) and their chemical structures illustrated in figure 3.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   The chemical structure of some antagonists at adenosine/P1 receptors.

	III. A1 Receptor

Top Previous Next References

Subdivision of A₁ receptors into high affinity A_1a receptors and low affinity A_1b receptors has been proposed (Gustafsson et al., 1990). This was based on the description of high-affinity binding sites for adenosine agonists and antagonists in rat and guinea-pig brain (A_1a) and low-affinity binding sites in rat vas deferens and guinea-pig ileum (A_1b) (Gustafsson et al., 1990). However, there are no cloned equivalents for these putative subtypes and their existence remains equivocal. It is possible that these reflect high and low affinity states of the same A₁ receptor.

A₁ receptors have been cloned from several species (table 3). The human adenosine A₁ receptor subtype gene (ADORA1) has been localized to chromosome 1q32.1 (Townsend-Nicholson et al., 1995a). The variability in the primary sequence of the A₁ receptor between species is less than 10% for A₁ receptors from dog, rat, and cow, and less than 5% between bovine and human A₁ receptors, but this seems to be sufficient to cause considerable interspecies differences in ligand binding (Tucker and Linden, 1993) and subtle differences in the mechanisms underlying receptor desensitization (Ramkumar et al., 1991; Nie et al., 1997; Palmer and Stiles, 1997b). Species homologs of A₁ receptors have been suggested to differ in their ability to discriminate among the related G_o/G_i protein alpha subunits (Jockers et al., 1994).

The A₁ receptor mediates a broad range of signaling responses, which may be caused by its coupling to different G proteins within the G_i/o family (Freissmuth et al., 1991; Munshi et al., 1991). The G proteins G_i and G_o are substrates for pertussis toxin that ADP-ribosylates the -subunit of G_i/o/t family members, uncoupling them from receptors. Accordingly, effects mediated by A₁ receptors are generally blocked by pertussis toxin. However, presynaptic A₁ receptors seem to be at least partly resistant to pertussis toxin (Fredholm et al., 1989; Hasuo et al., 1992); the reason for this could be the very tight coupling of the presynaptic A₁ receptors to potentially pertussis toxin-sensitive G proteins, rather than coupling to pertussis toxin-insensitive G proteins (Van der Ploeg et al., 1992). A partially-purified protein with selectivity for G protein  subunits has been shown to stabilize the rat brain A₁ receptor-G protein complex, thereby promoting tight coupling of the A₁ receptor with its G protein (Nanoff et al., 1997). Interestingly, this is a feature of the rat brain but not the human brain A₁ receptor; the latter is not under the control of a coupling cofactor, but operates according to the classic ternary complex model of receptor-G protein coupling (Nanoff et al., 1997).

The most widely recognized signaling pathway of A₁ receptors is inhibition of adenylate cyclase causing a decrease in the second-messenger cAMP (Van Calker et al., 1978; Londos et al., 1980). This in turn modulates the activity of cAMP-dependent protein kinase, which phosphorylates diverse protein targets. A₁ coupling to adenylate cyclase has been described in a number of tissues including brain, adipose tissue, and testes. In addition to direct modulation of signaling pathways downstream to cAMP, inhibition of adenylate cyclase via A₁ receptors blocks the effects of other agents which act by stimulating adenylate cyclase activity in cells.

Another signaling mechanism of A₁ receptors is activation of phospholipase C (PLC) leading to membrane phosphoinositide metabolism and increased production of inositol 1,4,5-triphosphate (IP₃) [and diacylglycerol (DAG)] and Ca²⁺ mobilization. This has been described in chinese hamster ovary (CHO)-K1 cells expressing the cloned human A₁ receptor (Iredale et al., 1994; Megson et al., 1995) as well as at endogenous A₁ receptors in a number of tissues including DDT₁ MF-2 smooth muscle cells (Gerwins and Fredholm, 1992a,b; White et al., 1992), heart (Scholz et al., 1993), myometrium (Schiemann et al., 1991a,b), rabbit cortical collecting tubule cells (Arend et al., 1989), renal cells (Weinberg et al., 1989), tracheal epithelial cells (Galietta et al., 1992), cultured mesangial cells (Olivera et al., 1992), and primary astrocytes (Peakman and Hill, 1995). IP₃ stimulates the release of Ca²⁺ from intracellular stores via interactions with specific receptors located on the sarcoplasmic reticulum. Elevation of cytosolic Ca²⁺ by IP₃ can stimulate a variety of signaling pathways, including a family of phosphatidyl serine-dependent serine/threonine-directed kinases collectively called protein kinase C (PKC) (of which there are at least 11 different isoforms), phospholipase A₂ (PLA₂), Ca²⁺-dependent K⁺ channels, and nitric oxide synthase (NOS). Depletion of Ca²⁺ from IP₃-sensitive pools may promote Ca²⁺ influx from extracellular sources.

Activation of phospholipase D (PLD) via A₁ adenosine receptors in DDT₁ MF-2 smooth muscle cells has been described (Gerwins and Fredholm, 1995a, 1995b), although as in the majority of cell systems this may be downstream of phosphoinositide hydrolysis and may require the intermediate activation of PKC or Ca²⁺.

Stimulation of A₁ receptors can activate several types of K⁺ channel, described principally in cardiac muscle and neurons. In supraventricular tissues (sino-atrial and atrioventricular node, and atrium), the A₁ receptor couples directly via pertussis toxin-sensitive G proteins to K⁺ channels (the same K⁺ channels are activated by both adenosine and acetylcholine), and activation causes bradycardia (Belardinelli et al., 1995a; Bünemann and Pott, 1995; Ito et al., 1995). A₁ adenosine receptors also couple to ATP-sensitive K⁺ channels (K_ATP channel); the activity is additionally regulated by metabolic demand (they close when intracellular ATP levels are high). Coupling seems to occur through the G protein in a membrane-delimited manner (Kirsch et al., 1990; Dart and Standen, 1993), although coupling via cytosolic factors is possible given the strong evidence that A₁ receptors, K_ATP channels, and PKC all have a role in ischemic preconditioning. A₁ receptor coupling to K_ATP channels has been described in rat and guinea-pig ventricular myocytes (Kirsch et al., 1990; Ito et al., 1994), porcine coronary arteries (Merkel et al., 1992; Dart and Standen, 1993), rabbit heart (Nakhostine and Lamontagne, 1993), and rat cerebral cells (Heurteaux et al., 1995). Activation of K_ATP channels mediates a reduction in action potential duration, vasodilatation and an increase in blood flow, which is consistent with their having a pivotal role in the coupling of vascular tone to metabolic demand determined both by intracellular purines (ATP/ADP levels) and by the extracellular actions of adenosine (released, for instance, during hypoxia or ischemia).

Neurons express multiple K⁺ channels that A₁ receptors may couple to regulate membrane potential and determine action potential frequency and duration. A₁ receptors reduce neuronal excitability and decrease firing rate by a hyperpolarizing effect mediated mainly by an increase in K⁺ conductance (Trussell and Jackson, 1985; Greene and Haas, 1991; Pan et al., 1995).

A₁ receptors also couple to inhibition of Ca²⁺ currents, which may account for inhibition of neurotransmitter release, although other or multiple mechanisms may be involved in this process (see Fredholm, 1995). Inhibition of Ca²⁺ currents by A₁ receptors has been described in dorsal root ganglion neurons (Dolphin et al., 1986), rat hippocampal pyramidal neurons (Scholz and Miller, 1991), rat sympathetic neurons (N-type Ca²⁺ channels, plus an unidentified Ca²⁺ channel) (Zhu and Ikeda, 1993), rat brainstem (predominantly N-type, but also P-type Ca²⁺ channels) (Umemiya and Berger, 1994), hippocampal CA1 neurons (N-type, plus some unidentified Ca²⁺ channels) (Wu and Saggau, 1994), hippocampal CA3 neurons (N-type Ca²⁺ channel) (Mogul et al., 1993), and mouse motoneurons (N-type Ca²⁺ channel) (Mynlieff and Beam, 1994). In atrial myocytes adenosine has an inhibitory effect on basal L-type Ca²⁺ current, although this is small and may be secondary to a reduction in basal cAMP (Belardinelli et al., 1995a).

Several mechanisms, operational at different levels of the signal transduction cascade, contribute to differential desensitization of G protein-coupled receptors. Rapid desensitization (occuring within a few minutes of agonist exposure) seems to involve phosphorylation of specific residues on the receptor C-terminal or the cytoplasmic loops by G protein-coupled receptor-specific kinases (GRKs) and/or kinases regulated by levels of intracellular second-messengers such as cAMP-dependent protein kinase. The phosphorylated receptor may bind to members of a family of proteins called arrestins, which cause uncoupling of the receptor from its G proteins. Desensitization occuring over a longer time course also involves uncoupling of the receptor-G proteins complex, but phosphorylation does not seem to be a prerequisite. Sequestration of receptors into an intracellular compartment may occur, as described for the increase in A₁ receptors in light vesicle membrane fractions prepared from the hamster vas deferens smooth muscle cell line, DDT₁ MF-2 cells, after chronic exposure to R-PIA (Ramkumar et al., 1991). Prolonged exposure to agonist may additionally lead to down-regulation of receptors and/or of the associated G proteins.

Desensitization of A₁ receptors by exposure to adenosine analogs has consistently been described both in vitro and in vivo, but this usually requires prolonged exposure to agonist (from 15 minutes to several hours or even days) (Parsons and Stiles, 1987; Ramkumar et al., 1991; Abbracchio et al., 1992; Green et al., 1992; Lee et al., 1993; Longabaugh et al., 1989; Casati et al., 1994). This is considerably longer than the time to desensitization of A₃ receptors which typically undergo significant desensitization within several minutes. Interestingly, while an agonist-stimulated increase in phosphorylation has been described for A₁ receptors in hamster DDT₁ MF-2 cells in association with receptor uncoupling from G proteins and desensitization, presumably by GRKs (Ramkumar et al., 1991; Nie et al., 1997), phosphorylation does not occur for the human A₁ receptor expressed in CHO cells at a time when receptor down-regulation is observed (Palmer and Stiles, 1997b). Down-regulation of A₁ receptors and/or of the associated G proteins after prolonged exposure to agonist has been reported in most of the cells and tissues in which this has been studied (Parsons and Stiles, 1987; Longabaugh et al., 1989; Green et al., 1992; Ramkumar et al., 1991, 1993a; Abbracchio et al., 1992).

Down-regulation of G proteins following A₁ receptor activation may lead to heterologous receptor desensitization. Chronic stimulation of A₁ receptors in adipocytes in vivo (Longabaugh et al., 1989) and in isolated adipocytes (Green et al., 1992) with (-)N⁶-phenylisolpropyl adenosine (PIA) for up to 6 and 7 days, respectively, causes down-regulation of A₁ receptors, non-uniform down-regulation of G_i proteins, and heterologous desensitization of other lipolytic hormone responses. In contrast, chronic (7 days) infusion of (R)N⁶-phenylisopropyl adenosine (R-PIA) in guinea-pigs homologously desensitizes the atrioventricular nodal response to adenosine: there is down-regulation of A₁ adenosine receptors, a decrease in high affinity A₁ receptors, and a decrease in G_i and G_o proteins, but no change in responses mediated by muscarinic receptors (Dennis et al., 1995).

Long-term treatment with adenosine/P₁ receptor antagonists generally leads to an increase in the effects of adenosine via a selective increase in the number of A₁ receptors, receptor sensitization and/or altered interaction between the receptor and the associated G proteins (Fredholm, 1982; Murray, 1982; Fredholm et al., 1984; Green and Stiles, 1986; Ramkumar et al., 1991; Fastbom and Fredholm, 1990; Zhang and Wells, 1990; Lupica et al., 1991a, 1991b; Shi et al., 1994). Long-term (12 day) caffeine treatment of rats increases the number of hippocampal A₁ (but not A_2A) receptors, without any changes in A₁ messenger ribonucleic acid (mRNA), suggesting that the adaptive changes are at the posttranslational level (Johansson et al., 1993a). An increase in the density of cortical A₁ receptors has been described after chronic caffeine injestion in mice, but surprisingly, given that striatal adrenergic, cholinergic, GABA, and serotonin receptors and Ca²⁺ channels are also affected by this treatment, there is no change in the density of striatal A_2A receptors (Shi et al., 1993).

Certain N⁶-substituted adenosine derivatives, such as N⁶-cyclopentyladenosine (CPA), N⁶-cyclohexyladenosine (CHA), and R-PIA, are selective agonists at A₁ receptors with K_i values in the range of 0.6 to 1.3 nM (see Jacobson et al., 1992b) (table 2).

Substitutions at both the N⁶- and C2-positions have produced 2-chloro-CPA (CCPA) which is A₁ selective, 1500-fold versus A₂ receptors in binding studies in rat brain, with a K_i of 0.6 nM (Lohse et al., 1988; Thompson et al., 1991; Jacobson et al., 1992b). N-[1S, trans,2hydroxycyclopentyl] adenosine (GR79236) has been reported to be an A₁ selective agonist, which is approximately equipotent with CPA in a variety of isolated tissues and cell types (Reeves et al., 1993; Gurden et al., 1993).

Most of the selective A₁ receptor antagonists described to date are xanthine-based derivatives. The introduction of hydrophobic (particularly phenyl or cycloalkyl) substituents into position 8 of the xanthine ring has yielded potent and A₁-selective antagonists, including 1,3-dipropyl-8-phenyl(2-amino-4-chloro)xanthine (PACPX), DPCPX, and xanthine amine congener (XAC) (Bruns et al., 1987; Martinson et al., 1987; Shimada et al., 1991) (fig. 3). Of these, DPCPX has the greatest affinity (K_i 1.5 nM) for A₁ receptors and the greatest A₁-subtype selectivity (A₂/A₁ affinity ratio 740), as shown in rat brain membranes (Bruns et al., 1987; Lohse et al., 1987). The human A₁ receptor has an approximately lower affinity for DPCPX (Libert et al., 1992; Klotz et al., 1998). A number of other 8-substituted xanthines, including (±)-8-(3-oxocyclopentyl)-1,3-dipropylxanthine (KFM 19) and KW-3902 (8-noradamant-3-yl-1,3-dipropylxanthine), have been shown to be selective antagonists at A₁ receptors (see Williams, 1989; Jacobson et al., 1992b). The alkylxanthine 1,3-dipropyl-8-[2-(5,6-epoxy)norbornyl]xanthine (ENX) is a potent (K_B 3.6 nM) and selective antagonist at A₁ receptors in the guinea- pig heart and brain and in DDT₁ MF-2 cells, with 400-fold greater affinity of binding versus A_2A receptors in guinea-pig brain (Belardinelli et al., 1995b).

Several classes of non-xanthine antagonists have been described, some showing reasonable affinity and selectivity for the A₁ receptor (see Jacobson et al., 1992b; Daly et al., 1993). Some of the more active of these are the tricyclic non-xanthine antagonists, including the triazoloquinazolines (Francis et al., 1988), the triazoloquinoxalines (Trivedi and Bruns, 1988; Sarges et al., 1990), and the imidazoquinolines (Van Galen et al., 1991).

The adenine derivative 1,3-dipropyl-8-[2,(5,6-epoxy)norbornyl]xanthine (N 0861) is reasonably selective (10- to 47-fold versus A_2A receptors) and potent at A₁ receptors in a number of tissues (May et al., 1991; Martin et al., 1993a; Belardinelli et al., 1995b). This compound has been superceded by the S-enantiomer 12 (CVT-124) with nanomolar selectivity and 1800- and 2400-fold selectivity at rat and cloned human A₁ receptors, respectively (Pfister et al., 1997), and by 8-(N-methylisopropyl)amino-N⁶-(5'-endohydroxy-endonorbornyl-)9-methyl adenine (WRC 0571) with 62-fold selectivity versus the A_2A receptor and 4670-selectivity versus the A₃ receptor (Martin et al., 1996).

(+)-(R)-[(E)-3-(2-phenylpyrazolo[1,5-]pyridin-3-yl)acryloyl]-2-piperidine ethanol, FK 453, has been reported to be a potent and selective A₁ receptor antagonist with IC₅₀ values of approximately 17 nM at rat cortical A₁ receptors and 11 µM at striatal A₂ receptors (Terai et al., 1995). Chiral pyrolo[2,3-d]pyrimidine and pyrimido[4,5-b]indole derivatives have been shown to be potent and highly stereoselective A₁ adenosine receptor antagonists (Müller et al., 1996a).

A₁ receptors are particularly ubiquitous within the central nervous system (CNS), with high levels being expressed in the cerebral cortex, hippocampus, cerebellum, thalamus, brain stem, and spinal cord (Reppert et al., 1991; Dixon et al., 1996) (fig. 4). Immunohistochemical analysis using polyclonal antisera generated against rat and human A₁ adenosine receptors has identified different labeling densities of individual cells and their processes in selected regions of the brain (Rivkees et al., 1995b). A₁ receptor mRNA is widely distributed in peripheral tissues having been localized in vas deferens, testis, white adipose tissue, stomach, spleen, pituitary, adrenal, heart, aorta, liver, eye, and bladder (Reppert et al., 1991; Dixon et al., 1996). Only very low levels of A₁ mRNA are present in lung, kidney, and small intestine (Reppert et al., 1991; Stehle et al., 1992; Dixon et al., 1996) (fig. 4).

View larger version (69K): [in this window] [in a new window]   Fig. 4.   Tissue distribution of adenosine receptor mRNA expression as examined by RT-PCR. Sizes of PCR products are given in base pairs. (From Dixon et al., 1996, Br J Pharmacol 118:1461-1468; with permission from McMillan Press Limited.)

It is now well established that adenosine is released from biological tissues during hypoxia and ischemic conditions. One of its effects is to reduce neuronal activity and thereby oxygen consumption; thus it acts as a neuroprotective agent. A significant part of these effects seem to be mediated by the A₁ receptor. A₁ receptors are located pre and postsynaptically on cell bodies, and on axons, where they mediate inhibition of neurotransmission by decreasing transmitter release, hyperpolarizing neuronal membranes, reducing excitability and firing rate, and altering axonal transmission. Adenosine can also exert behavioral effects: adenosine actions at A₁ receptors have been implicated in sedative, anticonvulsant, anxiolytic, and locomotor depressant effects (Nikodijevic et al., 1991; Stone, 1991; Jain et al., 1995; Malhotra and Gupta, 1997). Conversely, xanthine antagonists such as caffeine and theophylline have central stimulatory properties ascribed, at least in part, to inhibition of endogenous adenosine, although inhibition of cyclic nucleotide phosphodiesterases may contribute to this effect.

A₁ receptors mediate cardiac depression through negative chronotropic, dromotropic, and inotropic effects (see Olsson and Pearson, 1990). Slowing of the heart rate occurs via A₁ receptors on sinoatrial and atrioventricular nodes causing bradycardia and heart block, respectively, while the inotropic effects include a decrease in atrial contractility and action potential duration (Olsson and Pearson, 1990). This aspect of A₁ receptor-mediated effects has found application in the clinical use of adenosine to treat supraventricular tachycardia, and in the use of adenosine receptor antagonists in the treatment of bradyarrhythmias.

In the kidney, activation of A₁ receptors mediates diverse effects including vasoconstriction (principally of the afferent arteriole), a decrease in glomerular filtration rate, mesangial cell contraction, inhibition of renin secretion, and inhibition of neurotransmitter release (Olivera et al., 1989; Agmon et al., 1993; Barrett and Droppleman, 1993; Munger and Jackson, 1994). Intravenous and intra-aortic administration of adenosine in rats decrease water and sodium excretion via A₁ receptors, while selective antagonism of A₁ receptors causes diuresis and natriuresis (see Mizumoto et al., 1993; Van Beuren et al., 1993). Intrarenal administration of adenosine, but not of the A_2A selective agonist CGS 21680, in dogs also decreases water and sodium excretion (Levens et al., 1991a,b). Furthermore, A₁ receptors increase transepithelial resistance and reduce Na⁺ uptake in inner medullary collecting duct cells in culture (Yagil et al., 1994). On the other hand, intrarenal administration of adenosine and the A₁-selective agonist CHA in rats has been shown to induce marked diuresis and natriuresis which can be inhibited by the A₁-selective antagonist DPCPX (Yagil, 1994).

Direct effects on blood vessel tone via adenosine actions on A₁ receptors are rare. A more significant role of A₁ receptors with regard to regulation of blood vessel tone appears to be prejunctional modulation of neurotransmitter release. Prejunctional inhibition of neurotransmission via A₁ receptors on perivascular sympathetic (Gonçalves and Queiroz, 1996) and capsaicin-sensitive sensory afferents (Rubino et al., 1993) has been shown. However, A₁ receptors have been observed to mediate relaxation of porcine coronary artery (Merkel et al., 1992), and contraction of guinea-pig aorta (Stoggall and Shaw, 1990) and pulmonary artery (Szentmiklósi et al., 1995). A₁ receptors have also been reported to mediate contraction of rat isolated spleen (Fozard and Milavec-Krizman, 1993) and rat vas deferens (Hourani and Jones, 1994), as well as bronchoconstriction and bronchial hyperresponsiveness (Ali et al., 1994a, 1994b; Pauwels and Joos, 1995; el-Hashim et al., 1996). Diverse A₁-mediated effects in the gut have been described, including inhibition of peristalsis of rat jejunum (Hancock and Coupar, 1995b), relaxation of longitudinal muscle of rat duodenum (Nicholls et al., 1992, 1996), and contraction of rat colonic muscularis mucosa (Bailey et al., 1992; Reeves et al., 1993). Interestingly, adenosine mediates contraction of guinea-pig myometrial smooth muscle via A₁ receptors that in non-pregnant animals are coupled to the formation of IP₃, but in pregnant animals are coupled both to IP₃ and negatively to adenylate cyclase (Schiemann and Buxton, 1991; Schiemann et al., 1991a,b).

Selective inhibition of the synthesis of A₁ receptors with antisense oligonucleotides confirmed that these receptors are involved in an animal model of asthma (Nyce and Metzger, 1997). There was a marked reduction in the number of A₁ receptors in the lung and attenuation of airway constriction to adenosine, histamine, and dust-mite allergen (Nyce and Metzger, 1997). Although the site of action remains to be determined, selective antagonism of A₁ receptors offers a possible new approach in asthma therapy.

A₁ receptors on bovine pulmonary artery endothelial cells have been shown to mediate Cl efflux (Arima et al., 1994). In human airway epithelial cells, A₁ receptors have been reported to mobilize intracellular Ca²⁺ and activate K⁺ and Cl conductance (Rugolo et al., 1993), while selective inhibition of A₁ receptors with DPCPX increases cAMP-activated Cl conductance (McCoy et al., 1995).

A₁ adenosine receptors on rat cochleal membranes (Ramkumar et al., 1994), astrocytes (Peakman and Hill, 1994), and epididymal spermatozoa (Minelli et al., 1995) have been described. Release of Ca²⁺ from internal stores in perisynaptic glial cells of the frog neuromuscular junction via A₁ receptors has been described (Robitaille, 1995).

Adenosine acts via A₁ receptors and inhibition of cAMP to inhibit lipolysis and increase insulin sensitivity in adipose tissue (Londos et al., 1985; Green, 1987). Abnormal A₁ receptor function in genetic obesity has been proposed, showing that lipolysis is less active and A₁ receptor signaling more active, which may be caused by changes in receptor phosphorylation, but also possibly by adenylate cyclase activity (LaNoue and Martin, 1994; Berkich et al., 1995). In contrast, insulin sensitivity is decreased by activation of A₁ receptors in skeletal muscle (Challis et al., 1992). A₁ receptors on pancreatic  cells mediate inhibition of insulin secretion (Hillaire-Buys et al., 1989).

A₁ receptors have been widely reported to mediate the protective effects of adenosine in preconditioning and during ischemia or during reperfusion injury in the heart (Tsuchida et al., 1993, 1994; Yao and Gross, 1993; Lee et al., 1995; Lasley and Mentzer, 1995; Strickler et al., 1996; Grover et al., 1992; van Winkle et al., 1994; Sakamoto et al., 1995; Mizumura et al., 1996; Stambaugh et al., 1997), lung (Neely and Keith, 1995), and brain (Heurteaux et al., 1995). Strong evidence for a protective role of A₁ adenosine receptors comes from studies with transgenic mice over expressing the A₁ receptor. Mice over expressing the A₁ receptor have been shown to have an increased myocardial resistance to ischemia (Matherne et al., 1997). The mechanism involved is not yet clear; it may involve A₁ receptor activation of K_ATP channels as infarct size reduction after activation of A₁ receptors has been reported to be completely abolished by the blockade of K_ATP channels (Grover et al., 1992; van Winkle et al., 1994; Mizumura et al., 1996