Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Purinergic Mechanisms in Nervous System Function and Disease States

Michael F. Jarvis and Michael Williams


The role of the purine nucleoside, adenosine and its nucleotides, AMP, ADP and ATP, in intercellular signaling processes originated with the report of Drury and Szent-Gyorgi in 1929 (25) that adenosine (ADO) (Fig. 1) and AMP could reduce cardiac contractility, increase coronary vasodilatation, inhibit intestinal contraction and elicit CNS sedation in guinea pigs. These findings were rapidly followed by the evaluation of the therapeutic potential of ADO as an antihypertensive  agent. However, the short half life of ADO (2- 10 s) precluded further interest in the utility of the purine as a potential drug although this shortcoming proved to be an advantage in the subsequently approved use of ADO for the treatment supraventricular tachycardia some 60 years later (19).

It was not until the late 1940s that interest in the role of purines as extracellular messenger molecules re-emerged, culminating in the 1970s with Burnstock’s now seminal hypothesis of purinergic transmission that proposed the existence of distinct P1 (adenosine) and P2 (ATP) receptors (73). The P2 receptor family was then further subdivided on the basis of pharmacological data into P2X, P2Y  and P2T and P2Z receptors. With the exception of the P2T receptor, molecular clones of all the receptors proposed from these pharmacological studies have been identified.

Extracellular levels of ATP (Fig. 2)and ADO are increased as the result of tissue trauma, particularly ischemia and hypoxia (4,92) ADO thus acts as an autocrine homeostatic agent to conserve tissue function under adverse conditions (91). The widely consumed psychoactive drug, caffeine produces its CNS stimulatory actions by antagonizing the sedative, sleep inducing actions of endogenous ADO acting via the A2A receptor (55) indicating that the purine nucleoside is normally present in the extracellular environment.

ATP has neuromodulatory actions in amoeba, annelids, molluscs, coelenterates, crustacea and various insects (8) and are thought to precede neuropeptides as an intercellular messenger in evolutionary terms

Like the neuropeptides, the purines, ADO and ATP are also primary cellular constituents that are involved in nearly all aspects of cell function acting as metabolic cofactors, the building blocks for nucleic acids and proteins and as key molecules involved in the storage and production of cellular energy. The concept that ADO and ATP could function as neuromodulator agents via their presence in the synaptic cleft has been the subject of considerable debate focusing on why such a “uniquely valuable small molecule to the cell” (10) would be released to the extracellular milieu. However, the body can synthesize its own weight in ATP per day (66), the nucleotide being formed as fast as it is required (10). Thus the amounts of ATP, ADO and their various intermediates that are  involved in neurotransmission/neuromodulation processes appear inconsequential  compared to the total amounts available. In addition to their role as signalling molecules, ATP and ADO may also have the potential to alter the cellular energy charge (2) distal to their site of release. ATP may also act as a substrate for synaptic membrane phosphorylation events (10).

 Abundant evidence now exists to show that ADO has both pre and postsynaptic effects on neurotransmission processes (6) while ATP has excitatory actions both centrally and peripherally (90).

Four distinct ADO receptors and 11 different receptors sensitive to purine and pyrimidine nucleotides, (Tables 1, 2, and 3; 36, 51, 67, 73) have been cloned and characterized providing a diversity of cellular targets through which purines can elicit effects on tissue function.

Purinergic Receptor Dynamics, Interactions, and the Purinergic Cascade

The metabolic pathways that link ATP, ADP, AMP and ADO and the potential for each of these purines to elicit receptor-mediated effects on cell function forms the basis of a potentially complex physiological cascade that is comparable to those involved blood clotting and complement activation. Thus ATP released synaptically as a co-transmitter leads to the sequential formation of ADP, AMP and ADO (Fig. 3). The  ATP in the extracellular space can activate the various classes of P2 receptors and its actions are terminated by receptor desensitization or dephosphorylation. The latter occurs via group of  some 11 enzymes that metabolize ATP, diadenosine polyphosphates like Ap4A (Fig. 2), and NAD (99). Ecto-ATPases preferentially hydrolyze ATP to ADP and ecto-apyrases convert both ATP and ADP to AMP. Ecto-5’-nucleotidase converts AMP to ADO. The activity of this enzyme class is dynamic.  In myeloid leukocytes, ecto-apyrase and ecto-5’-nucleotidase show stage specific transient expression (16, 40) while in guinea-pig vas deferens, soluble nucleotidase are released from neurons together with ATP and norepinephrine (87) suggesting that the ATP released by nerve activity can undergo increased inactivation as a result of the same nerve activity that results in its release.

The term inactivation is however a misnomer since the products of ATP breakdown their own functional activities, some of which are mutually antagonistic to ATP, and thus comprise a purinergic cascade (Figure 2). ATP can antagonize the actions of ADP on platelet aggregation and ADO-elicited sedation in the CNS activity contrasts with the excitatory actions of ATP on nerve cells (73). Furthermore, in the broader framework of ATP-modulated proteins, ATP-sensitive potassium channels (KATP) can also be activated when intracellular ATP levels are reduced (32). Thus as P2 receptor mediated responses are attenuated as a result of ATP hydrolysis to adenosine, P1 mediated responses and KATP -mediated responses become enhanced. ADP activates platelet P2T receptors and can also enhances its own availability while ADO activates members of the P1 receptor family. Activation of A1 or A2A receptors can inhibit ATP availability (49) while activation of hippocampal A3 receptors can desensitize A1 receptor-mediated inhibition of excitatory neurotransmission (6). For UTP, while high concentrations of uracil have been reported to modulate CNS dopaminergic systems in animals, there is currently no pharmacological evidence for a uracil equivalent of the P1 receptor (91).

The purinergic cascade is an elegant and complex system for the regulation of cell to cell communication that in physiological terms will be dependent on  the dynamics of the local milieu in which ATP is made available thus reflecting the purinoceptor phenotype of the tissue, ectonucleotidase activities and ADA, AK and nucleoside transporter activity.

More recently, electrophysiological studies on P2X and nicotinic receptor-mediated responses have led to the suggestion (80) that these two ligand gated ion channels interact with one another with each receptor containing a inhibitory binding site for agonists active at the other receptor. If this hypothesis proves to be correct, then nicotinic agonist may function as P2X receptor antagonists. This is particularly interesting inasmuch as ATP, acting via the P2X receptor is a nociceptive agent while neuronal nicotinic receptor agonists like ABT-594 are potent analgesic agents (3)

Purine Availability and Modulation

The factors governing the extracellular availability of ADO and ATP in nervous tissue have been and remain a controversial issue. Basal levels of ADO in the extracellular space in the CNS are thought to be in the 30 -300 nM range. Under conditions of tissue hypoxia or ischemia or in seizure activity, extracellular ADO levels can approach micromolar levels (28,96)  Similarly, extracellular ATP levels can also reach millimolar concentrations in the local environment either through release as a co-transmitter or following cellular perturbation (73). ATP is released together with other neurotransmitters including acetylcholine, norepinephrine, glutamate, GABA and neuropeptide Y (83) depending on the transmitter repertoire of the neuron.

Under basal conditions, ADO levels in the extracellular milieu are tightly regulated by ongoing metabolic activity. Bi-directional nucleoside transporters and the enzymes, ADO deaminase (ADA) and ADO kinase (AK), regulate the removal of ADO from the extracellular space (Fig. 4; 35, 42, 52). ADO can also be formed intracellularly from ATP and transported to the extracellular milieu via nucleoside transporters thus representing a major source of extracellular ADO (35).

Under conditions of hypoxia or ischemia, ADO levels in the extracellular space are markedly increased in response to increased metabolic demand with the purine acting to regulate the energy supply / demand balance in a given tissue in response to changes in blood flow and energy availability (4, 92). Reductions in oxygen or glucose availability due to tissue trauma such as that that occurs during stroke, epileptogenic activity and reduced cerebral blood flow lead to the breakdown of ATP with the sequential formation of ADP, AMP and ADO. Thus the normal homeostatic role of extracellular ADO can be locally amplified several fold resulting in an enhanced protective role to prevent further traumatic insult to affected tissues (76).

Studies with a limited number of ADA and AK inhibitors have shown that inhibition of AK is physiologically more relevant in increasing extracellular ADO availability than ADA inhibition (47, 97). AK inhibitors are also more effective in enhancing the neuroprotective actions of endogenous ADO as compared to inhibitors of ADA or ADO transport (47,97).  Compounds that act to potentiate the actions of endogenous ADO have effects that are limited to those areas where tissue insult results in increased production of extracellular ADO, e.g in stroke, reperfusion injury and epilepsy (Fig. 3). Such compounds have been termed "site and event specific" agents (29).

P1 and P2 Receptors

The adenosine or P1 receptor family is activated by adenosine and its many analogs and, with the exception of the rat A3 receptor, is selectively blocked by methyl and arylxanthines derived from caffeine and theophylline e.g. CPX (Fig. 5) and by a number of other novel heterocyclic molecules that include CGS 15943A, SCH 58261, ZM 241385, MRS 1067, MRS 1191, MRS 1220 and MRS 1222 (43, 44, 73;Table 1; Fig. 5).

ATP and related purine and pyrimidine nucleotides activate the P2 receptor family (Tables 2 and 3). However, the selectivity of the various agonists is extremely dependent on the tissue preparation(s) used, the species and the experimental protocol. As a result, the functional characterization of P2 receptors has been limited by a paucity of potent, selective and bioavailable ligands, both agonists and antagonists (43, 44).           All P2 agonist ligands known are closely related to ATP, UTP, ADP etc. (Fig. 2) and, irrespective of their degree of chemical modification, show varying degrees of susceptibility to degradation and intrinsic activity (48). There are also very few studies where a systematic evaluation of the selectivity of P2 receptor agonists has been carried out. As a result, their use as probes for the functional characterization of the various P2 receptor subtypes has been limited. For example, BzATP is widely used as an agonist for the P2X7 receptor active where it is active in the micromolar range being 13-times more potent than ATP in activating this receptor P2X7 receptor (EC50 = 18 mM). It is however, far more potent at transfected rat and human P2X1 (EC50 = 1.9 nM) and P2X3 (EC50 = 98 nM) receptors (5)

Evolution of Receptor Nomenclature

While the nomenclature for P1 receptors is straightforward from both a pharmacological and a molecular perspective, that for P2 receptors has evolved in a somewhat haphazard manner reflecting both the complexity of this superfamily and the limited pharmacological tools available for receptor characterization. Thus the P2X, P2Y , P2T and P2Z nomenclature was followed by the identification of various pharmacologically defined receptors designated P2D, P2U, P3, P4, P2YAp4A etc. (73)  Since ATP was known to produce its receptor -mediated effects via either ion channels or G protein -coupled receptors, P2 receptors were divided into two main classes, P2X that are ligand-gated ion channels and P2Y which are GPCRs. With the cloning of the members of the P2X and P2Y families, the previous nomenclature systems have been replaced with P2Xn and P2Yn designations (73). For  the P2X receptor family, these receptors are sequentially numbered 1 through 7 (P2X1 - P2X7) For the P2Y family, receptors designated  P2Y1, P2Y2, P2Y4, P2Y6, and P2Y11 have been cloned and shown to have functional activity (51,73; Table 3)

This unusual numbering reflects the fact that at least six other putative P2Y receptors have been identified based on putative sequence homology which are either non-mammalian homologues or receptors for which nucleotides are not the preferred agonists (51, 73).  For example, the putative P2Y7 receptor (1) proved to be a leukotriene B4 receptor (94). For those receptors that are not valid members of the P2Y receptor family but have been cloned, lower case is used to define the receptor. Thus what would be the P2Y3 receptor is designated as p2y3. For those receptors that have been pharmacologically defined but not cloned, italics and subscripts are used, e.g. P2T and P2D.


Four P1 receptors designated A1, A2A, A2B and A3 (Table 1) have been cloned and pharmacologically characterized (67). All four are members of the G protein -coupled receptor (GPCR) superfamily and are heterogeneously distributed in a variety of mammalian tissues including heart, smooth muscle,  kidney, testis, platelets, leukocytes, adipocytes in addition to the central nervous system (CNS) 

The A1 receptor has a wide distribution in the CNS and is functionally coupled to inhibition of cAMP formation, stimulation of potassium conductance, inhibition of N-channel-mediated calcium conductance, stimulation of phospholipase C production and modulation of nitric oxide production (73). Selective agonists for A1 receptors are ADO analogs substituted in the N6-position and include cyclohexyl (CHA), cyclopentyl (CPA; A1 Ki = 0.6 nM)and 2-chlorocyclopentyl (CCPA; Ki = 0.6 nM) ADO (Fig. 1) that are 780- and 1500- fold selective for the A1 receptor as compared to other P1 receptors (44). Agonist effects at the A1 receptor are selectively blocked by 8 - phenyl substituted xanthines including BIIP-20 (Ki = 15 nM) and cyclopentylxanthine (CPX; Ki = 0.46 nM) that are 180- and 740-fold selective for the A1 receptor, respectively (Fig. 5). Pharmacologically, the A1 receptor shows distinct species differences at the receptor level (91). Like other GPCRs, the A1 receptor can be allosterically modulated by compounds that do not directly interact with the agonist binding site of the receptor. The thiophenes, PD 81,723 and RS 74513  (Fig. 5) can selectively enhance A1 receptor binding and function (7, 58) by stabilizing an agonist preferring conformation of the A1 receptor independent of an interaction of G proteins (60).

Two molecular and pharmacologically distinct subtypes of the A2 receptor exist that are linked to activation of adenylate cyclase (67). The A2A high affinity receptor may also utilize N- and P-type Ca2+ channels as signal transduction mechanisms. This receptor is localized in the striatum, nucleus accumbens and olfactory tubercule regions of mammalian brain. The lower affinity A2B receptor is more ubiquitously distributed throughout the CNS and periphery. CGS 21680 (Ki A2A = 15 nM; Fig. 1) is an 2-substituted ADO analog that is 39- and 173-fold selective for the A2A receptor versus A3 and A1 receptors in vitro. The xanthine antagonists, KF 17837 (Ki A2A = 24 nM; Fig. 5) and CSC (8 -(3-chlorylstyryl) caffeine (Ki A2A = 9 nM) are 108- and greater than 3000-fold selective for A2A receptors versus other members of the P1 receptor family (43). SCH 58261 (Ki A2A =  nM) and ZM241385 (Ki A2A = x nM; Fig. 5) are novel and potent non-xanthine antagonists that are 60 - 1000- and 6800-fold selective, respectively for the A2A receptor (43). The A2B receptor has proven considerably more difficult to characterize due to a paucity of selective agonist and antagonist ligands. Responses mediated by the non -selective ADO agonist, 5’N6-ethylcarboxamido adenosine (NECA) and not by other A1, A2A or A3 receptor selective agonists can however, be attributed to A2B receptor activation.

The A3 receptor is a relatively new member of the P1 receptor family that is linked to inhibition of adenylate cyclase and elevation of cellular IP3 levels and intracellular Ca2+ and also shows distinct species-dependent pharmacology. The human A3 receptor is sensitive to xanthine blockade while the rat receptor is not (67, 91). The A3 receptor shows wide spread distribution with low levels in brain. IB-MECA (Ki = 1nM) and its 2-chloro analog (Ki = 0.3 - 0.7 nM; Fig. 1) are potent and selective A3 receptor agonists, the latter being 2500- and 1400-fold selective for A3 versus A1 and A2A receptors, respectively. Efforts to identify non-xanthine A3 receptor antagonists have been unusually successful with  the flavonoid MRS 1067, the dihydropyridine, MRS 1191, the triazolonaphthyridine, L-249313 and the thiazolopyrimidine, L-268605 having been recently identified (43). A3 receptors are involved in mast cell function, eosinophil apoptosis and the phenomenon known as preconditioning in ischemic reperfusion (61).


The P2 receptor family is divided into two major subclasses; P2X (Table 2) receptors that are ligand-gated ion channel (LGIC) receptors specific for ATP and P2Y (Table 3) receptors that are members of the GCPR superfamily (26, 73). While the initial classification of P2 receptors was based on the rank order potency in vitro of a limited series of agonists related to ATP in native tissues, the majority of these receptors have subsequently been cloned and functionally characterized in various heterologous expression systems (51, 73).

P2X Receptors.  P2X receptor subunits share a common motif of two transmembrane spanning regions, a large extracellular domain with both the N and C termini being located intracellularly in a manner similar to that  of the amiloride-sensitive epithelial Na+ channel  (73). As noted by Ralevic and Burnstock (73), the subunits by themselves are not functional entities. The functional receptor channel, a non-selective pore permeable to calcium, potassium and sodium and mediates rapid (~ 10 ms) neurotransmission processes (Fig. 6), is formed by multimeric combinations of the various P2 receptor subunits. When a given receptor is discussed below, unless stated otherwise, it will refer to a homomeric functional LGIC. The homomeric and heteromeric combinations and their proportions in the functional receptor(s) is unclear with evidence for three, five (65) and four (74) subunit combinations. Seven functional members of the P2X receptor family, P2X1 - P2X7 have been cloned and characterized (Table 2). These have been grouped into three major classes based on agonist efficacy (27). Group 1 includes P2X1 and P2X3 receptors that have high affinity for ATP (EC50 = 1 mM) and are rapidly activated and desensitized; Group 2 includes P2X2, P2X4, P2X5 and P2X6 receptors that have lower affinity for ATP (EC50 = 10 mM) and show a slow desensitization and sustained depolarizing currents; and Group 3 that is represented by the P2X7 LGIC that has very low affinity for ATP (EC50 = 300 -400 mM), shows little or no desensitization and in addition to functioning as an ATP-gated ion channel, also functions as a non-selective ion pore.  The physiological significance of P2X receptor desensitization remains to be determined but clearly reflects one mechanism by which to terminate the actions of ATP, another being degradation of the nucleotide to ADO.

Known P2 receptor antagonists include PPADS, DIDS, various blue dyes that include Evans, trypan and reactive blue-2 and suramin, the usefulness of which is limited by their lack of selectivity, potency and bioavailability, not only for the different subtypes of the P2 receptor but also for other classes of neurotransmitter receptors, G proteins and various enzymes (43,44,91). In addition, these compounds may also inhibit the enzymes responsible for the breakdown of ATP and related nucleotides thus further confounding receptor characterization (48). The lack of any reliable binding assays for P2 receptor subtypes together with species nuances in receptor pharmacology (13,14,41) has tended to limit the discovery of improved compounds.

                More recent P2X selective antagonists have been described that have reduced ectonucleotidase activity:  Examples a DIDS analog NH01 and the suramin analogs, NF023 and NF 279 (Fig. 7; 43).  A series of trinitrophenyl (TNP) substituted nucleotides have been reported as noncompetitive, reversible antagonists at P2X1 and P2X3 receptors (88). TNP-ATP(Fig. 7) which has antagonist activity in the 1 nM range has being characterized as an allosteric modulator has weak activity at P2X4 and P2X7 receptors. The ATP analogs, A3P5PS and A3P5P (Fig.1) are partial agonists/ competitive antagonists at the turkey erythrocyte P2Y1 receptor (43) and a derivative, MRS 2179 (Fig. 7) is a full P2Y1 receptor antagonist (IC50 =330 nM; 11). ARL 67085 (57) is 2-alkylthio substituted bioisostere of ATP that is a selective antagonist at the ADP-sensitive P2T/P2YAC receptor involved in platelet aggregation (21).

P2X1 receptors are activated by 2meSATP, ATP and ab-meATP, exhibit rapid desensitization kinetics and are present in the dorsal root, trigeminal and celiac ganglia and in spinal cord and brain. The P2X1/5 receptor heteromeric polymers display slow desensitization kinetics and reduced affinity for ab meATP.

The P2X2 receptor is activated by 2MeSATP and ATPgS but is insensitive to ab-meATP and bg-meATP. This receptor desensitizes slowly to agonist activation and is present in brain, spinal cord, superior cervical ganglia and adrenal medulla. Splice variants of the P2X2 receptor have been localized to the cochlea of the ear (85). P2X2-1 and P2X2-3R receptors have been identified in rat tissues and P2X 2-1, P2X 2-2 and P2X 2-3 receptors in guinea - pig tissues. P2X2-1  and P2X2-3R splice variants are present on the endolymphatic surface of the cochlear endothelium, an area associated with sound transduction (37).

The P2X3 receptor has a rank order of activation where 2meSATP >> ATP > ab-meATP. This receptor is localized to a subset of sensory neurons that includes the dorsal root, trigeminal and nodose ganglia. The P2X3 receptor has similar properties to the P2X1 subtype including ab-meATP sensitivity and rapid desensitization kinetics. P2X2 and P2X3 subunits can form functional heteromeric receptors in vivo Functionally, the P2X2/3 heteromeric receptor appears to combine the pharmacological properties of P2X3 (ab-meATP sensitivity) with the kinetic properties of P2X2 (slow desensitization) thereby facilitating its detection in situ or in heterologous expression systems (73).

The P2X4 receptor is activated by 2MeSATP and is only weakly activated by ab meATP. The human P2X4 receptor is weakly sensitive, and the rat P2X4 receptor insensitive to putative P2X receptor antagonists. This receptor is present in rat hippocampus, superior cervical ganglion, spinal cord, bronchial epithelium, adrenal gland, testis and human brain and can form functional  heteromers with P2X6 subunits in vitro (54).

The P2X5 receptor shows an activation profile of ATP > 2MeSATP > ADP with ab meATP being inactive (17). The P2X5 receptor does not exhibit rapid desensitization kinetics but is blocked by suramin and PPADS. Message for the P2X5 receptor is present in the central horn of the cervical spinal cord and in trigeminal and dorsal root ganglia neurons. The P2X5 receptor is found in the brain only in the mesencephalic nucleus of the  trigeminal nerve (17,73) and can form heteromers with P2X1 subunits.

The P2X6 is present in the superior cervical ganglion, cerebellar Purkinje cells, spinal motoneurons of lamina IX of the spinal cord and the superficial dorsal horn neurons of lamina II. It is also present in trigeminal, dorsal root and celiac ganglia (17).  In vivo, neither the P2X5 nor the P2X6 receptor appear to exist as homomers but form functional heteromers with other P2X receptor subunits.

The P2X7 receptor, known as the P2Z receptor before it was cloned (84), is an atypical member of the P2X receptor family that has been extensively studied in mast cells and macrophages (22, 26). This receptor has a long (240 amino acid) intracellular C-terminal region that confers a unique phenotype of forming a large non-selective cytolytic pore upon prolonged or repeated agonist stimulation (22, 26). Application of agonists to the P2X7 receptor for brief periods (1-2 sec) results in transient pore opening that is thought to be involved in intercellular signaling processes. Prolonged P2X7 receptor activation can trigger apoptosis (programmed cell death), a complex intracellular process that is important in both embryogenesis and in removing cancerous, infected and dying cells from tissues (15). The P2X7 receptor can be partially activated by saturating concentrations of ATP, but it is fully activated by the synthetic ATP analog, BzATP (34). The physiological function of the P2X7 receptor remains unclear since its involvement in apoptotic events would suggest that its presence within a cell is designed, under conditions where there are high local concentrations of ATP (300 - 400 mM; 27) leads to the elimination of the cell.

In the immune system, hemopoetic cell differentiation and activation are modulated by P2 receptors (27) and IFN-g  and LPS can upregulate P2X7 receptor expression (39,40) an accompanying decrease in ecto-ATPase activity which makes them more susceptible to the cytolytic actions of extracellular ATP.  ATP can induce cytolysis in macrophages that are infected with mycobacterium via P2X7 receptor - mediated apoptotic and necrotic events (53). This novel antimicrobial activity of ATP while having potential utility in the treatment of tuberculosis may also provide a more basic understanding of P2X7 receptor- mediated apoptotic events than that can be derived from more complex mammalian cell systems. The P2X7 receptor is found in the SCG and spinal cord (84) and cerebral artery occlusion results in an increase in P2X7 immunoreactive cells in the penumbral region around the stroke (17).The release and maturation of IL-1b from macrophages can be stimulated by ATP acting via P2X7 receptor-mediated mechanisms (22, 68) that involve activation of the cysteine protease/ caspase, interleukin -1b convertase (ICE) that is involved in the initiation of apoptosis (86).

KN-62 is an isoquinoline inhibitor of calcium-calmodulin dependent protein kinase-II (CamK-II) with an IC50 value of 900 nM. The compound is also a potent, non-competitive antagonist of the human P2X7 receptor (IC50 = 9 -13 nM; 34). KN-62 was inactive at the transfected rat P2X7 receptor (41) highlighting species related responses. The novel anti-inflammatory, tenidap, a putative ICE inhibitor can enhance ATP activation of P2X7 receptors in mouse macrophages (77).

                P2Y Receptors.   The P2Y receptors are all members of the GPCR superfamily and are activated by both  purine and pyrimidine nucleotides (18, 36, 51).  While 13 P2Y-like receptors have been cloned (51) only five mammalian subtypes designated P2Y1, P2Y2, P2Y4, P2Y6, and P2Y11 have been cloned and shown to have functional activity (Table 3; 51). All five are coupled to Gq11 and receptor activation results in stimulation of PLC and IP3 activation and subsequent release of calcium from intracellular stores (36). 

                The  P2t receptor in platelet is distinguished from other P2 receptors in that ADP is the preferred agonist and ATP can function as a competitive antagonist  P2t and P2Y1 receptors share a similar agonist pharmacology leading to the proposal that they are the same molecular entity although it has yet to be cloned. This is however, controversial and there is now a body of evidence showing that purine responses in platelets are mediated via multiple P2 receptor subtypes including a putative P2t receptor (21).

                The P2Y1 receptor is preferentially activated by adenine nucleotides with 2MeSATP being the most potent agonist. Uridine nucleotides (e.g. UTP, UDP) are inactive at this receptor.  Suramin, PPADS and cibacron blue can antagonize P2Y1 receptor activation. The ADO-bisphosphate analogs, A3P5PS, A3P5P and MRS 2179 (Fig. 1 & Fig. 7) are competitive antagonists at this receptor (11,43)

The P2Y2 receptor is activated by both ATP and UTP, with diphosphate nucleotides are inactive. The efficacy of UTP at this receptor provided molecular evidence for the concept of pyrimidinergic transmission.  P2 receptor antagonists like suramin and are less efficacious at the P2Y2 receptor.  UTP is the preferred agonist at the P2Y4 receptor with ATP and the nucleotide diphosphates being inactive. The latter are more active at the P2Y6 receptor as compared to nucleoside triphosphates. The P2Y6 receptor can thus be classified as a UDP-preferring receptor. The P2Y11 receptor (18) is unique in regard to other P2Y receptors in that only ATP serves as a agonist for this receptor.

                The diadenosine polyphosphates represent an additional group of purine based signalling molecules that can modulate cell function via activation of cell surface receptors (38). A pharmacologically defined Ap4A receptor in nervous tissue can modulate neurotransmitter release. However, it is unclear whether the actions of the diadenosine polyphosphates involve unique receptor subtypes since these purines non- selectively interact with other cloned P2 receptors (73). No distinct receptors for the diadenosine polyphosphates have yet been cloned.

Therapeutic Potential of Purinoceptor Ligands in Nervous Tissue

                ADO is a potent inhibitor of dopamine, GABA, glutamate, acetylcholine, serotonin and norepinephrine release acting via presynaptic A1 receptors (6). These effects of ADO occur preferentially on excitatory as opposed to inhibitory neurotransmitter release data (71, 95) implying a degree of specificity in regard to its effects on brain function. Postsynaptically, ADO can modulate excitability acting via both A1 and A2A receptors causing hyperpolarization of the postsynaptic membrane.

In the past 25 years there have been many studies suggesting that ADO systems in both the CNS and PNS are involved in the actions of a wide variety of CNS active drugs that include analgesics, antipsychotics, antidepressants, anxiolytics, nootropics / cognition enhancers and the various agents effective in stroke related CNS damage. These studies have involved test paradigms in which either the effects of CNS drugs on ADO responses have been evaluated or, alternatively, the effects of ADO agonists or antagonists on the effects of prototypic CNS agents were assessed. In many instances single, somewhat high, concentrations of a limited number of compounds were used to draw conclusions related to a complete class of psychotherapeutic agent, often with no negative control data thus representing a somewhat reductionistic approach to delineating the role of the purine in drug actions.  Thus, while ADO has been implicated in the actions of a wide variety of CNS active agents, much of this data must be viewed as interesting in the absence of more rigorous evaluation. For P2 receptors where the absence of ligands, either agonists and antagonists, limit the functional characterization of the various receptor subtypes,  the delineation of a role(s) for P2 receptors in CNS pathology has been postulated largely on the basis of in situ localization of the mRNAs encoding the different P2 receptor subtypes.

One way to circumvent these limitations is the use of mice made deficient in a targeted receptor by gene disruption, the phenotype of which will provide information on the role of the receptor. In A2A receptor knockout mice, caffeine no longer acts as a stimulant but depresses exploratory activity. These mice which were insensitive to the effects of the A2A agonist, CGS 21680, were more anxious, more aggressive, had higher nociceptive thresholds and increases in blood pressure and heart rate (55) 

Compounds that produce their effects via purinoceptor systems, either P1 or P2 receptors, comprise three distinct classes - i) conventional agonists, partial agonists or antagonists; ii) allosteric modulators of receptor function and ii) modulators of the endogenous systems that regulate the extracellular availability of ATP, ADO, UTP and their respective nucleotides. This latter group includes the various ecto-ATPases, adenosine deaminase, adenosine kinase and the bi-directional member transporter systems (35, 42, 52) that remove ADO from the extracellular environment (Fig. 4).

Efforts over the last 25 years to develop directly acting P1 receptor agonists and antagonists as therapeutic agents (91) have been unsuccessful due to a combination of the choice of disease states in which other therapeutic modalities have been successful and the side-effects associated with global receptor modulation. The identification of partial agonists that may have enhanced tissue specificity (42), allosteric modulators (7, 34)  or of novel modulators of ADO and ATP metabolism may result in the discovery of clinically useful agents with greater therapeutic indices (35, 52).


                In hypoxia (92)and focal ischemia (76) there is rapid extracellular accumulation of ADO supporting the role of the purine as a homeostatic neuroprotective function of endogenous ADO (52, 91). In support of this concept, directly acting ADO receptor agonists can reduce cerebral ischemic damage while, in contrast, ADO receptor antagonists exacerbate ischemic brain damage in animal models of focal and global cerebral ischemia (98). The neuroprotective effects of ADO are mediated by several distinct but complementary mechanisms (76). ADO A1 receptor activation stabilizes neuronal membrane potential, inhibiting neuronal excitability and excitatory amino acid (EAA) release (76). Blockade of EAA release thus prevents the neurotoxic sequalae associated with activation on NMDA receptor.  ADO also hyperpolarizes astrocyte membranes, limiting extracellular glutamate and potassium accumulation and modulates local cerebral blood flow and local inflammatory responses such as platelet aggregation and neutrophil recruitment and adhesion acting via the A2A receptor (31)

A1 receptor agonists like CHA can reduce stroke related cell death and hippocampal neurodegeneration (76) while ADO antagonists increase ischemic damage by enhancing glutamate release. The prototypic AK inhibitor, 5’d-5IT is also neuroprotective (52, 63) Expression of P2X7 receptor mRNA is upregulated on microglial cells in the ischemic penumbral region 24 hr following middle cerebral artery occlusion in the rat  (17) suggesting that cytolytic pore formation and inflammatory cytokine release (22) are events associated with neural trauma and neurodegeneration


                 Seizure activity is associated with rapid and marked increases in brain ADO concentrations in both animals (50) and in epileptic patients with spontaneous onset seizures (28). ADO agonists reduce seizure activity induced by a variety of chemical and electrical stimuli in animal models (23) acting via A1 receptors (50)  In electrically kindled seizure models, ADO agonists reduce seizure severity and duration without significantly altering seizure threshold (50). The anticonvulsant effects of ADO agonists are blocked by doses of methylxanthines that, when given alone, have no observable effect on seizure activity (50).

                 The direct administration of the AK inhibitors, 2’-deoxycoformycin, NH2dADO and 5-IT (Fig. 1) into rat brain reduced bicuculline methiodide-induced seizures (97). AK inhibitors were more effective than either ADA inhibitors(2’-deoxycoformycin) or nucleoside transport inhibitors (dilazep) in this model.  5’d-5IT given systemically was more potent in reducing PTZ-induced seizures in mice as compared to NH2dADO and 5-IT in agreement with the rank order of potency of these AK inhibitors in inhibiting ADO phosphorylation in intact cells. (52)  These effects appear to be centrally mediated as the peripherally active ADO receptor antagonist, 8-(p-sulfophenyl)theophylline (8-PST) was unable to antagonize the anticonvulsant actions of 5’d-5IT while the A1-selective antagonist, 8-cyclopentyl-1,3-dimethylxanthine (CPT), completely block PTZ-induced seizures in mice (50). Other AK inhibitors like GP515, GP683, and GP3269 also have enhanced in vivo potency as compared to NH2dADO in reducing PTZ and maximal electroshock seizures in mice (52, 63).

                The role of ATP as an excitatory neurotransmitter coupled with the finding of reduced ecto-ATP activity in situations of increased seizure sensitivity suggests that ATP may participate in pathological neuronal hyperexcitability. Recent studies have shown that ATP agonists potent induced generalized seizures following direct infusion into the prepiriform cortex (50). Interestingly, direct administration of 8-PST was shown to potentiate the proconvulsant effects of ATP in this brain region further supporting a specific and discrete role for ATP in the excitatory neurotransmission events associated with seizure generation.


                It is widely accepted that the neuronal hyperexcitability associated with ischemia, hypoxia and epilepsy also underlies the neurodegenerative processes associated with aging. Thus the toxicity  following excessive glutamate release and the resultant changes in calcium homeostasis that lead to nerve cell death may reflect an acute manifestation of more subtle, long term changes that are associated with Alzheimer's and Parkinson's Disease.  ADO antagonists like caffeine and theophylline are potent CNS stimulants and enhance cognition in animal models acting by blocking the actions of endogenous ADO. For instance, the  8 - substituted xanthine, BIIP 20 ((+)-8 - (3-oxocyclopentyl)- 1, 3- dipropylxanthine; Fig. 1) has been clinically evaluated as a cognition enhancing agent with potential utility in the treatment of Alzheimer's and other age-related dementias (91).

                The utility of ADO ligands in addressing the processes underlying neurodegeneration is bifunctional. Agonists reduce EAA-induced neurotoxicity while antagonists enhance cognition by disinhibition of the inhibitory effects of ADO on excitatory neurotransmission. If CNS - selective purinoceptor ligands are to be effective agents for the amelioration of the cognitive decline associated with aging, a functional equilibrium will need to be established between the neuroprotective and CNS stimulatory actions of the purine. In this context, P1 receptor partial agonists may represent useful entities (42).

                Amyloid-b-protein (Ab) may elicit nerve cell death in Alzheimer's Disease via activation of the classical complement cascade via an immunoglobulin-independent mechanism and epidemiological data that indicates that aged patients with rheumatoid arthritis who consume large quantities of traditional anti-inflammatory agents have a reduced incidence of Alzheimer's disease has underlined a role for inflammation in he onset of this disease. ADO agonists are potent anti-inflammatory agents (31) acting to inhibit free radical production and may thus  provide additional benefit in AD over and above direct effects on neurotransmitter - mediated neuronal events.

In nervous tissue, trophic factors ensure neuronal viability and regeneration (64). Increases in the polypeptide growth factors e.g. fibroblast growth factors, epidermal growth factor and platelet-derived growth factor are increased following neural injury (64). ATP can act in combination with various growth factors to stimulate astrocyte proliferation ,contributing to the process of reactive astrogliosis, a hypertrophic/hyperplastic response that is frequently associated with brain trauma, stroke/ischemia, seizures and various neurodegenerative disorders.In reactive astrogliosis, astrocytes undergo process elongation and express GFAP (glial fibrillary acidic protein), an astrocyte specific intermediate filament protein. ATP increases GFAP and AP-1 complex formation in astrocytes (64) in a manner quantitatively similar to that seen with bFGF. In addition, ATP as well as GTP can induce trophic factor (NGF, NT-3, FGF) synthesis in astrocytes and neurons.

ADO DOPAMINE INTERACTIONS: Parkinson's Disease and Psychosis

                A detailed body of behavioral and biochemical data supports a functional interaction between central dopaminergic and purinergic systems (30). Methylxanthines like caffeine stimulate rotational behavior and potentiate the effects of dopamine (DA) agonists in rats with unilateral striatal lesions. Conversely, ADO agonists can blocked the behavioral effects of DA via A2A receptor activation (75). Interestingly, in mouse A2A receptor knockouts, exploratory motor activity was reduced relative to controls and caffeine was able to further reduced this activity, an effect opposite to its well characterized psychomotor stimulant effects (55).

ADO A2A receptors are highly localized in striatum, nucleus accumbens and olfactory tubercule, brain regions that also have high densities of DA D1 and D2 receptors. Furthermore, mRNAs  for ADO A2A receptors and DA D2 receptors are co-localized in GABAergic- enkephalin striatopallidal neurons in the basal ganglia (Fig. 8) that represent the so-called “indirect' pathway from the striatum to the globus pallidus. Dysfunction of this pathway may be involved in the etiology of Huntingtons' chorea as well as the movement disorders associated with Parkinson's disease (PD; 30). The indirect pathway originates from striatal GABA - enkephalinergic neurons and via GABAergic relays interacts with a glutaminergic pathway arising in the subthalamic nucleus (Fig. 8). The latter can activate the internal segement of the  pars reticulata which relays through a pars reticulata thalamic GABAergic pathway to inhibit the thalmic/ cortical glutaminergic pathway (Fig. 9a).

                A direct pathway also exists that originates in the striatum from GABA - Substance P - dynorphinergic neurons which,acting via a GABAergic pathway also inhibits the internal segment of the pars reticulata to disinhibit the ascending thalamic glutaminergic pathway (Fig. 8 & Fig. 9b). The balance between these direct (cortical activating) and indirect (cortical inhibiting) striatal dopaminergic pathways  tonically regulates normal motor activity. Dopaminergic inputs arising from the pars compacta can facilitate motor activity, inhibiting the indirect pathway by activation of D2 receptors and activating the direct pathway via D1 receptor activation.

                These findings have led to the hypothesis that striatal ADO systems, specifically those involving A2A receptors, may play a pivotal role in neurological disorders involving basal ganglia malfunction like PD. Intrastriatal administration of the A2A agonist, CGS 21680 attenuates rotational behavior produced by both direct and indirect DA agonists in unilaterally lesioned rats. Radioligand binding studies have shown an increased efficacy of CGS 21680 in reducing the binding affinity of supersensitive D2 receptors, a mechanistic finding that supports the increased sensitivity of animals with supersensitive DA receptors to CGS 21680. Repeated administration of the DA antagonist, haloperidol can upregulate the density of both DA D2 and ADO A2A receptors in rat striatum. The modulatory influence of ADO on dopaminergic neurotransmission is thus enhanced in situations of increased DA receptor sensitivity (30, 75). 

                The modulatory relationship between ADO and DA extends to the ADO A1 receptor, activation of which can reduce the high affinity state of striatal DA D1 receptors {30).  Functionally, the A1 selective agonist, CPA blocks DA D1 receptor-mediated locomotor activation in reserpinized mice. Similarly, the non- selective ADO agonist, NECA can attenuate perioral dyskinesias induced by selective DA D1 activation in rabbits.. 

                ADO acting at both striatal A2a and A1 receptors directly modulates DA receptor -mediated effects on striatal GABA-enkephalinergic neurons and DA D1 receptors on striatal GABA-Substance P neurons (Fig. 5).  While the exact mechanisms by which ADO agonists modulate the binding of dopaminergic agonists at the D1 and D2 receptors is currently unknown, these interactions appear to be independent of G protein coupling and an intramembrane modulatory mechanism has been proposed (30).

This dynamic inter-relationship between dopaminergic and purinergic systems in the neurochemistry of psychomotor function offers new possibilities for the amelioration of dopaminergic dysfunction via ADO receptor modulation. Selective adenosine A2A receptor antagonists like KW 17837 have been evaluated as novel treatments for PD in MPTP lesioned marmosets(75). Aa related analog, KW-6002 potentiates the antiparkinsonian effects of L-dopa without producing dyskinesia in this model (46)

                As would be predicted, ADO agonists mimic the biochemical and behavioral actions of DA antagonists in animal models, an effect mediated via A2A receptors.  ADO agonists inhibit DA release and synthesis and attenuate DA transductional processes effects that contribute to a diminution in dopaminergic neurotransmission. ADO receptor agonists thus act as functional DA antagonists (30).

                The lack of a clear dose-dependent effect for the A2A selective agonist CGS 21680 to induce catalepsy illustrates the possibility that activation of  A2A receptors can produce favorable antipsychotic activity without untoward motor impairment (30,62).  One ADO agonist, CI -936 was advanced to the clinic over a decade ago as a novel antipsychotic but was terminated in Phase II trials due to undisclosed side effect liabilities.


                The hypnotic and sedative effects of ADO are well known as are the central stimulant activities of the various xanthine ADO antagonists that include caffeine (45). Direct administration of ADO into the brain elicits an EEG profile similar to that seen in deep sleep, manifested as an increase in REM sleep with a reduction in REM sleep latency that results in an increase in total sleep (12) Caffeine, on the other hand, suppresses REM sleep and decreases total sleep time. Sleep deprivation can increase ADO A1 receptor density in the cortex and corpus striatum (12).  In vivo microdialysis studies have shown that extracellular ADO concentrations increase in basal forebrain in proportion to periods of sustained wakefulness and declined during sleep (70) suggesting that ADO functions as a endogenous sleep regulator. Infusion of the A2A agonist, CGS 21680 into the subarachnoid space associated with the ventral surface of the rostral basal forebrain, an area defined as the prostaglandin D2-sensitive sleep promoting zone increased slow wave (SWS) and paradoxical (PS) sleep, effects blocked by the A2A antagonist, KW 17837 (78). The A1 selective agonist, CHA, suppressed SWS and PS prior to eliciting an increase in SWS.


                Application of ATP to sensory afferents results in hyperexcitability and the perception of intense pain. The nucleotide can also induce nociceptive responses at local sites of administration and can facilitate nociceptive responses to other noxious stimuli (79). P2 receptor antagonists like suramin and PPADS reduce nociceptive responses in animal models of acute and persistent pain (24).

                The pronociceptive actions of ATP are mediated via P2X receptors present on sensory afferents and in the spinal cord. Homomeric P2X3 and heteromeric P2X2/3 receptors are highly localized the  sensory nerves that specifically transmit nociceptive signals (20). ATP is released from a number of cell types (e.g. sympathetic nerves, endothelial cells, visceral smooth muscle) in response to trauma (9) and there is a substantive body of evidence that activation of P2X3 receptors may initiate and contribute to the peripheral and central sensitization associated with visceral nociception (9). P2X3 receptor expression is upregulated in sensory afferents and spinal cord following damage to peripheral sensory fibers (90).Thus the  development of selective, bioavailable P2X3 receptor antagonists may be anticipated to provide novel compounds for the treatment of pain.

        While  ATP acts to facilitate nociceptive sensory information processing, ADO has opposite effects, inhibiting nociceptive processes in the brain and spinal cord. Intrathecal  ADO, ADO receptor agonists and AK inhibitors provide pain relief in a broad spectrum of animal models (e.g. mouse hot plate test, mouse tail flick assay, rat formalin test, mouse abdominal constriction assay (47, 52, 69, 79). These effects are blocked by systemic or intrathecal ADO receptor antagonist administration. ADO agonists are also effective in relieving neuropathic pain in rat models (56).

                ADO receptor agonists also inhibit pain behaviors elicited by spinal injection of substance P and the glutamate agonist, NMDA .Glutamate is a key mediator of the abnormal hyperexcitability of spinal cord dorsal horn neurons (central sensitization) that is associated with clinical pain states (93) A1 agonists inhibit the spinal cord glutamate release and also reduce cerebrospinal fluid levels of substance P in rat, another key mediator of nociceptive responses (79, 81,82). Electrophysiological studies have shown both pre- and post-synaptic actions of ADO on synaptic transmission from primary afferent fibers to neurons of the substantia gelatinosa of the spinal dorsal horn (59). Thus a combination of peripheral and supraspinal mechanisms contribute to ADO modulation of nociception. The ADO agonists, CHA, R-PIA and NECA, were 10 to 1000-fold more potent in inhibiting acetylcholine-induced writhing in mice when administered i.c.v than orally, suggesting a supraspinal site of action (52). ATP can also modulate nociceptive transmission processes in the spinal cord (59). The ability of ADO to inhibit neurotransmitter release (6) and inflammatory processes (31) may contribute to the blockade of peripheral sensitization which is a feature of the pain resulting from tissue injury and inflammation (92).  ADO agonists have also shown utility in relieving human pain (82). Spinal administration of the A1 agonist, R-PIA relieved allodynia in a neuropathic pain patient without affecting normal sensory perception and ADO infusion (at doses without effect on the cardiovascular responses) improved pain symptoms in clinical pain models reducing spontaneous pain, ongoing hyperalgesia and allodynia in patients with neuropathic pain. In addition, low dose infusions of  perioperative ADO during surgery reduced the requirement for volatile anesthetic and for postoperative opioid analgesia (33, 81).

                AK inhibitors are also active in a variety of animal pain models (52). As noted in animal seizure models, NH2dADO was a more effective antinociceptive agent than the ADO deaminase inhibitor, deoxycoformycin (47). The effects of NH2dADO were blocked by intrathecal co- administration of theophylline. NH2dADO,  and two other AK inhibitors, 5'd-5IT and 5IT were also active when given systemically in the mouse hotplate test (52).

Challenges in the Development of CNS Selective Therapeutic Agents

In the past decade, the field of purinergic pharmacology has undergone significant growth as the complexity of the receptor families and the various enzymes involved in purine metabolism have  been cloned. Progress in delineating bona fide targets for therapeutic intervention and the development of selective compounds that may represent drugs has continued to be slow. One theme that has emerged however, is the complex interactions of P1 and P2 receptor systems and their ligands. A wide spectrum of ligands have been developed for P1 receptors over the past 30 years. Interestingly, the majority of these were developed in the compete absence of information regarding the molecular structure of these receptors and without the advantages of high throughput screening to identify novel structures. In contrast, the search for P2 receptor ligands with improved potency, selectivity and bioavailability is occurring in a very different era of drug discovery with the molecular structure of the targets firmly established coupled with the ability to functionally characterize these receptors in transfected cell systems.. 

Efforts to develop therapeutics based on the modulation of purine receptor responses for CNS disorders have not been very successful. CI-936 has already been mentioned while the potential use of ADO antagonists as cognition enhancers/CNS stimulants has been investigated with compounds like BIIP 20 with no results are available on any clinical utility.  The extensive research efforts at the Karolinska Institute on the interrelationships between A2A receptors and  D2 receptors in the basal ganglia offer a novel approach to the treatment of PD and KF 17837 and KW 6002 have already shown positive results in predictive animal models of this disease (46, 75). Direct acting ADO agonists have shown problems with tolerance and a lack of tissue selectivity (44, 91) However, novel AK inhibitors have shown potential to enhance the endogenous neuroprotective actions of ADO in epilepsy, stoke, and chronic pain (52). 

While much less is currently known regarding the P2 receptor families in CNS function, the discrete localization of the P2X3 receptor on sensory nociceptive neurons (20)  together with the nociceptive actions of P2 agonists (9, 24) suggests that antagonists for this receptor may  have potential as novel analgesic agents. Finally, the emerging role of the P2X7 receptor in apoptotic events and its increase in ischemic tissues (17) suggests that this receptor may be as a key mediator of the molecular events related to neurodegeneration as extracellular ATP levels are increased under conditions of neuronal trauma including hypoxia/ischemia, neuropathic pain and viral load. The development on novel ligands for these receptors coupled with genetic analysis of tissues from various disease states may be anticipated to provide a wealth of novel targets to aid in the understanding on CNS disease states and their amelioration.

Acknowledgements: The authors would like to thank Ed Burgard and Wende Niforatos for providing the information in Fig. 6.  They would also like to acknowledge the extensive work of their colleagues in the purine research area whose work, for reasons of space limitations, could not be cited.

published 2000