Keywords

1 Introduction

In 1972 Ginsborg and Hirst (1972) described for the first time that the nucleoside adenosine inhibited acetylcholine release at the rat-diaphragm neuromuscular junction and that this effect was abolished by theophylline. This observation was soon supported by Ribeiro and Walker (1975) using the frog neuromuscular junction preparation. Authors discussed these results in relation to the capability of adenosine to increase cyclic AMP (cAMP), an effect that was abolished by theophylline as previously demonstrated by Sattin and Rall (1970). In 1979 van Calker et al. (1979) proposed that adenosine regulates the accumulation of cAMP in cultured brain cells, via two different types of adenosine receptors and in 1980, Londos et al. (1980) proposed subclasses of membrane adenosine receptors: the A1/Ri (that inhibited adenylate cyclase) and the A2/Ra (that activated adenylate cyclase). In 1984 Ribeiro and Sebastião (1984) in an attempt to characterize the type of adenosine receptors involved in the inhibitory action of adenosine at the frog neuromuscular junction, suggested that the adenosine receptors at the frog neuromuscular junction should not be classified as A1/A2 because the potency profile of adenosine agonists did not fit the pharmacological profile proposed for either A1 or A2 receptors. As a consequence, Ribeiro and his Ph.D. student, Ana M. Sebastião, were requested by the editors of Progress in Neurobiology to write a review on adenosine receptors. In reviewing the available pharmacological information, three different adenosine receptor entities emerged based on the affinity of different adenosine agonists: A1 with a pharmacological profile with L-PIA, CHA>CADO>D-PIA, NECA (high stereoselectivity for the PIA isomers) and negatively coupled to adenylate cyclase; the A2 with NECA>CADO>L-PIA, CHA, D-PIA with low stereoselectivity for the PIA isomers (L-PIA∼D-PIA) and positively coupled to adenylate cyclase. A third entity was then proposed, an A3 adenosine receptor (A3AR) with an agonist profile with L-PIA, CHA, NECA>CADO, and D-PIA usually less potent than CADO, possibly linked to calcium (Ribeiro and Sebastião 1986).

After cloning, it was clarified that rat A3AR is homologous with the adenosine A1 and A2A receptors and belongs to the G-protein-coupled receptor family (Meyerhof et al. 1991a, b). After cellular expression of cloned A3AR and pharmacological characterization, rat A3AR emerged as having a very low affinity for xanthine-based adenosine receptor antagonists such as theophylline. Various specie homologues of this receptor have been cloned, including the human A3AR (Salvatore et al. 1993). It emerged that A3AR cloned from different species show different pharmacological properties. Cloning and expression of the human A3AR stably expressed in Chinese hamster ovary cells pointed out that, contrary to rat A3AR, human A3AR is xanthine sensitive.

2 Distribution of A3 AR in the Central Nervous System (CNS)

The first binding studies on solubilized membranes from rat brain demonstrated the presence of a low affinity adenosine receptor with characteristics of the A3 subtype (Oliveira et al. 1991). However, in situ hybridization studies in the rat indicated the presence of A3AR mRNA only in the testis (Meyerhof et al. 1991a; Zhou et al. 1992; Rivkees 1994) and not in the CNS (Rivkees et al. 2000). Similarly, no expression of A3AR in the brain of mice or in the hippocampi of humans was detected (Rivkees et al. 2000). However, by reverse transcription–polymerase chain reaction (RT-PCR) A3AR expression was found distributed in the rat heart, lung and widespread in the rat and mouse brain (Zhou et al. 1992; Dixon et al. 1996; von Arnim et al. 2000). Low levels were detected by A3AR binding in various regions of the mouse brain (Jacobson et al. 1993), more than five times below that of the A1 receptor (Cunha et al. 1995) or of the A2A receptor (Cunha et al. 1996a).

There are significant brain regional differences in the levels of A3AR mRNA. In the mice, there is evidence for the expression of the A3AR in the hippocampus, thalamus and hypothalamus (Yaar et al. 2002). It is generally accepted that A3AR has species-specific tissue distribution. In humans and sheep, A3AR is significantly expressed in many peripheral tissues with lower levels in the CNS and testis (Linden et al. 1993; Salvatore et al. 1993).

As to A3AR localization in the different cell type of the CNS, the presence of A3AR in neurons, primarily at presynaptic sites, was demonstrated by PCR of laser dissected hippocampal neurons and by western blotting in rat hippocampal nerve terminal membranes (Lopes et al. 2003). Moreover, A3AR mRNA is expressed in microglia (Fiebich et al. 1996) and has been identified by Northern blot analysis in mouse astrocytes (Zhao et al. 1999).

3 The Roles of A3AR in the CNS

The endogenous neuromodulator, adenosine, controls and integrates a wide range of brain functions; its extracellular levels vary according to behavioral state and pathophysiological condition. Dysfunction of the adenosine system is involved in pathologies ranging from epilepsy to neurodegenerative disorders and psychiatric conditions. Less is known about the contribution of the low-affinity A3AR to the regulation of brain function and neuropathological conditions if compared to high-affinity A1 and A2A adenosine receptors. Its role in several pathophysiological conditions is often enigmatic and controversial.

3.1 Role of A3 AR in Memory and Cognition

Phenomena of synaptic plasticity, long-term potentiation (LTP) and long-term depression (LTD) are the likely cellular substrates for learning and memory. It is possible to induce either LTP or LTD according to the magnitude of the transient calcium levels attained at restricted synaptic spine domains. Smaller calcium increases predominantly activate protein phosphatases, leading to LTD, whereas higher calcium levels activate protein kinases, causing LTP (Dudek and Bear 1993).

Adenosine affects synaptic plasticity phenomena acting on both A1 and A2A receptors (de Mendonca and Ribeiro 2000; Rebola et al. 2008). The estimated affinity of rat A3AR for the endogenous ligand, adenosine (Ki = 1 μM), is considerably lower compared to that of A1 receptors (Ki = 10 nM) or A2A receptors (Ki = 30 nM) (Jacobson et al. 1995). Since the extracellular concentration of endogenous adenosine does not exceed 300 nM (Latini and Pedata 2001) it is difficult to detect a physiological role of A3AR in the CNS. However, conditions of stimulation that elicit synaptic plasticity can also transiently raise the extracellular adenosine concentration three to tenfold over basal levels (Cunha et al. 1996b). A role for A3AR on LTP and LTD has also been reported in the hippocampus (Costenla et al. 2001). Activation of A3AR essentially attenuates LTD (Costenla et al. 2001; Huang et al. 2007) and allows induction of LTP. The facilitating effects of the A3AR agonist, (1-[2-chloro-6-[[(3-iodophenyl)methyl]amino]-9H-purin-p-yl]-1-deoxy-N-methyl-beta-D-ribofuranuronamide (Cl-IB-MECA) on LTP were observed with weak subliminal θ-burst induction conditions but not with high frequency stimulation, which would elicit marked postsynaptic depolarization and Ca2+ increase, thus overcoming the facilitory effect of A3AR activation. This effect is consistent with a modification in the threshold for the induction of long-term synaptic changes, and might be due to coupling of A3AR to the phospholipase C transducing pathway (Abbracchio et al. 1995), thus promoting a shift to increased Ca2+ levels and protein kinase activation. The Cl-IB-MECA effect is prevented by the selective A3AR antagonist, MRS1191, indicating a genuine A3AR mediated response. However, the A3AR antagonist, MRS1191, does not by itself modify θ-burst-induced LTP, suggesting that tonically released adenosine is not able to activate A3AR in order to modulate LTP.

Behavioral studies have confirmed the involvement of A3AR in spatial learning and memory. In female Swiss mice, 1-deoxy-1-[6-[[(3-iodophenyl)-methyl]amino]-9H-purin-9-yl]-N-methyl-beta-D-ribofuranuronamide (IB-MECA), administered i.p. 20 min before tests, diminished scopolamine- and MK-801-induced impairment of spontaneous alternation in Y-maze and learning abilities in a passive avoidance task indicating that A3AR stimulation may ameliorate spatial memory and long term memory impairments in terms of cholinergic and glutamatergic deficits (Rubaj et al. 2003). It must be mentioned, however, that in the same animal species, Borowicz et al. (1997), using the passive avoidance task, demonstrated that a non selective A3AR agonist, N(6)-2-(4 aminophenyl)ethyladenosine (APNEA), administered i.p. 30 min before behavioral tests, impaired long-term memory.

When looking at A3AR-mediated behavioral effects, it should be remembered that A3AR agonists have depressant effects on locomotor activity (Jacobson et al. 1993).

3.2 Role of A3AR in Locomotion

By studying locomotor activity in an open field, Jacobson et al. (1993) demonstrated that intraperitoneal administration of 3-IB-MECA in mice induces a depression of motor activity. These results are consistent with the most recent experiments performed in A3AR-deleted mice. Significant increases in some aspects of motor function were observed in A3AR-deleted mice by using three different tests: activity in the open field; number of arm entries in the elevated-plus maze; and number of transitions in the light/dark box (Fedorova et al. 2003). The change in motor activity appears selective without evidence of ataxia. The increase in motor activity was attributed to disinhibition of cortical neurons because selective A3AR stimulation inhibits excitatory neurotransmission in rat cortical neurons (Brand et al. 2001). Most recently it was confirmed that both adolescent (21-day old) and adult A3AR knockout (KO) mice showed an increase of motor activity in the open field (Bjorklund et al. 2008a). Moreover, a reduced response to the motor-stimulating effect of caffeine or amphetamine was found in A3AR KO mice. These data are surprising in view of the poor affinity of A3AR for caffeine (Bjorklund et al. 2008a).

Work reviewed above indicates that even though A3AR are expressed at low level in the central nervous system, they play a tonic role in modulating motor activity.

3.3 Role of A3AR Receptors in Convulsions

Adenosine was identified as an endogenous anticonvulsant in the brain more than 20 years ago (Dunwiddie 1980; Lee et al. 1984; Dragunow et al. 1985; Dragunow 1991) and it was suggested that dysfunction of the adenosine-based neuromodulatory system might contribute to epileptogenesis (Boison 2007, 2008). Many experimental convulsive procedures led to a considerable rise in extracellular levels of adenosine (for a review see: Boison 2008). Clear-cut evidence that this really occurs in epileptic patients was provided by During and Spencer (1992). Their studies, with microdialysis probes implanted in the hippocampi of epileptic patients with intractable complex partial epilepsy, revealed that extracellular levels of adenosine in the dialysate were elevated by six- to 31-fold during seizures.

Recently, the therapeutic potential of stem cells engineered to release adenosine as a local source to augment endogenous adenosinergic functions was assayed in two cell transplantation experiments (Li et al. 2007, 2008; Boison 2008). Most of the studies about the role of adenosine as an anticonvulsant emphasize the preeminent involvement of A1 adenosine receptors. However, several studies using different experimental models of epilepsy have investigated the role of adenosine A2A and A3 receptors in this condition.

A first report conducted in an in vivo seizure model in mice indicates that stimulation of A3AR protects from seizures (von Lubitz et al. 1995). Acute systemic administration of IB-MECA protects against chemically-induced (NMDA injection) but not electrically-induced seizures while a protective effect of chronically administered IB-MECA is evident in both chemically- and electrically-evoked seizures (von Lubitz et al. 1995). The protective effect of acute administration of IB-MECA is attributed to both arteriolar constriction and severe hypotension (von Lubitz et al. 1994), which can reduce the final intracerebral concentration of the chemoconvulsant NMDA (von Lubitz et al. 1995). In a study conducted on seizure-sensitive DBA/2 mice, an animal model of generalized reflex epilepsy, the intraperitoneal administration of A3AR agonist IB-MECA is without effect (De Sarro et al. 1999). More recently, Vianna et al. (2005) demonstrated in adult rats in status epilepticus (SE) induced by pilocarpine, that pretreatment with the A3AR antagonist MRS1220 does not alter the incidence of SE but reduces the latency to develop it.

Although the above reviewed evidence would support the concept that A3AR exerts a protective role against seizures, in some cases data indicates that A3AR stimulation by endogenous adenosine may aggravate epileptic activity, as determined in CA1 area of rat immature hippocampal slices in which seizure activity was induced by Mg2+ deprivation together with brief high frequency stimulation of Shaffer collaterals (Etherington and Frenguelli 2004). Such an A3AR-mediated excitatory effect is in agreement with the notion that activation of A3AR in the adult hippocampus exerts excitatory effects, that is increases high-threshold calcium currents (Fleming and Mogul 1997), desensitizes the inhibitory A1 receptors (Dunwiddie et al. 1997) and inhibits presynaptic inhibitory metabotropic glutamate receptors (Macek et al. 1998). In addition, 2-Cl-IB-MECA increases the amplitude of electrically evoked and the frequency of spontaneous epileptiform field potentials recorded during GABAA receptor blockade in CA3 area of rat immature hippocampal slices (Laudadio and Psarropoulou 2004). An enhanced amplitude of evoked responses associated with the increase in frequency of spontaneous discharges may facilitate seizure-induced neuronal damage. The frequency of spontaneous discharges recorded during GABAA and A1 or A2A receptor blockade is not affected by the A3AR antagonist, MRS1220 (Laudadio and Psarropoulou 2004). This indicates that adenosine, tonically released from immature slices, is unlikely to increase the rate of spontaneous discharges. However, the high affinity uptake blocker NBTI (Deckert et al. 1988), which increases extracellular adenosine, induced spontaneous discharges in a subset of rat immature slices (Laudadio and Psarropoulou 2004). Two lines of evidence suggest that this effect is mediated by A3AR. First, 2-Cl-IB-MECA has no additional excitatory effect following an NBTI-induced excitation, and secondly, the NBTI-induced increase in the frequency of spontaneous discharges (in the presence of the A1 antagonist DPCPX) was reversed by the A3AR antagonist MRS1220. The NBTI-induced excitation in immature slices is a novel finding which may be peculiar to the developing hippocampus. In fact in the adult hippocampus NBTI has an inhibitory effect on evoked or spontaneous discharges; this is in line with the consistent A1-mediated inhibitory effects of adenosine (Sanderson and Scholfield 1986).

The variable effects of NBTI in the immature and mature hippocampus might be ascribed to changes in the levels of endogenous adenosine (Park et al. 1987) and/or in the proportion of A1 and A3AR.

In conclusion, the excitatory effects of A3AR activation on synchronous epileptiform discharges in vitro suggest that an increase of endogenous adenosine in conditions of stress, that is seizures or hypoxia, may enhance synaptic activity in the immature brain. In addition, they raise the possibility that the A3AR subtype may play a role in the establishment of activity-dependent plastic changes.

The blockade of A3AR with the selective antagonist, MRS1334, improves the stability of GABAAergic neurotransmission, as assessed in a patch clamp study on GABAA receptor isolated from neurosurgically resected epileptic human nervous tissues and microtransplanted into Xenopus oocytes, and on human epileptic slices obtained from neurosurgical resection (Roseti et al. 2008).

Therefore, antagonism of A3AR may increase the inhibitory efficacy of GABAA receptor in some forms of human epilepsy, pointing towards new therapeutic targets to fight epilepsy.

3.4 Role of A3AR in Nociception

Adenosine exerts complex influences on pain transmission by different mechanisms in the brain and spinal cord, as demonstrated in a broad spectrum of animal pain models (for a review see: Sawynok 1998). Several therapeutic approaches to pain and inflammation based on mimicking or modulating the effects of endogenous adenosine are currently under preclinical and clinical investigation. These include the use of adenosine itself, the use of direct-acting adenosine receptor agonists and the use of agents designed to modulate the levels and therefore the actions of adenosine in the extracellular space (adenosine-kinase (AK) inhibitors) (see: Gao and Jacobson 2007). Much evidence indicates that adenosine receptor activation in the spinal cord produces antinociception due to stimulation of A1 receptors. This effect is attributed to presynaptic inhibition of excitatory neurotransmitter release with subsequent reduction of substance P concentration, as detected in cerebrospinal fluid (Sjolund et al. 1997) and therefore due to postsynaptic inhibition of the glutamate effects (DeLander and Wahl 1988).

A3AR ko have decreased nociception, as assessed by the hot-plate test (Fedorova et al. 2003), probably due to a decrease in the supraspinal processing and “recognition” of painful stimuli. This is consistent with the localization of A3AR in thalamic nuclei (Yaar et al. 2002) where they may play a role in processing nociceptive information.

Following carrageenan-induced inflammation in the hind paw, heat hyperalgesia, plasma extravasation and edema were significantly reduced in A3AR-deleted mice compared to wild type mice (Wu et al. 2002) suggesting that A3AR plays a role in generating the localized inflammatory response which is in agreement with previous evidence that adenosine A3AR activation produces pain behaviors secondary to mast cell degranulation and release of histamine and 5-hydroxytryptamine (5-HT) that exert nociceptive actions at sensory nerve terminal (Sawynok et al. 1999). Subcutaneous administration of adenosine A3AR agonists produces nociceptive behavior (Sawynok et al. 1997).

The above data support the pro-nociceptive role of A3AR involved in both central nervous system effects and pro-inflammatory effects on peripheral tissues.

It must be mentioned, however, that intrathechal administration of IB-MECA does not exhibit an antinociceptive profile in acute nociception as assessed in the early phase pain response of the formalin test, but it does depress the late phase of prolonged pain that measures hyperalgesia related to formalin-induced inflammation (Yoon et al. 2005, 2006). This suggests an involvement of spinal adenosine A3AR in protection from nociception and is in agreement with the observation that A3AR agonists control the in vitro release of pain-related neuropeptides from the rat spinal cord (Mauborgne et al. 2002).

3.5 Role of A3AR in Mood and Affects

Miller and Hoffman (1994) demonstrated that activation of A3AR results in increased 5-HT-uptake in rat basophilic leukemia cells. Likewise, but in rat central nervous system, the activation of adenosine A3AR by the agonist APNEA during adenosine A1 receptor blockade, decreases hippocampal extracellular 5-HT levels in freely moving rats (Okada et al. 1997). The inhibitory effect of the A3AR agonist on extracellular 5-HT levels is abolished by inhibition of 5-HT reuptake activity with DU24565 and fluoxetine (Okada et al. 1999). It was later established using rat basophilic leukemia 2H3 cells, that activation of A3AR, via both PKG and p38 MAPK, stimulates the activity of 5-HT transporter (SERT) (Zhu et al. 2004) that is the main responsible for inactivation of synaptic 5-HT and a main target of antidepressant drugs which inhibit SERT activity.

More recently, it has been reported that the A3AR agonist IB-MECA rapidly (10 min) and selectively stimulates 5-HT transport in mouse midbrain, hippocampal, and cortical synaptosomes (Zhu et al. 2007). IB-MECA-induced stimulation of 5-HT uptake is blocked by the selective A3AR antagonist MRS1191 and is absent in synaptosomes prepared from A3AR-knockout mice (Zhu et al. 2007). In view of these results, a possible antidepressant effect of A3AR antagonists could be envisaged. However such extrapolation is not supported by behavioral depression tests evaluated in A3AR deleted mice. A3AR ko mice show an increase in the amount of time spent immobile in two tests of behavioral depression, the forced-swim test and the tail-suspension test, respectively (Fedorova et al. 2003). This response is probably not attributable to a decrease in motor activity, mainly because of the increased locomotion expressed by the A3AR deleted genotype (Fedorova et al. 2003).

Although there is evidence that A3AR ko mice are more prone to depressive behavior, it is also reported that they have increased performance in the elevated-plus maze and light/dark box suggestive of reduced anxiety but this is most probably a consequence of the increase in exploratory activity due to increased motor activity (Fedorova et al. 2003).

3.6 A3AR and Cerebral Blood Flow Regulation

One important issue not always sufficiently clarified is whether modification of blood pressure contributes to some effects of systemically administered A3AR agonists. Von Lubitz et al. (1994) have shown that preischemic administration of IB-MECA results in a significant delay in the return of postischemic blood flow, and this may affect neuronal survival (see Section 9.4.1). The hypotensive response observed following adenosine A3AR activation in the anaesthetized rat may involve mediator release from mast cells (Fozard and Carruthers 1993; Fozard et al. 1996). Indeed A3AR activation results in rapid mast cell degranulation in the anaesthetized rat. Moreover, a direct central regulation of arterial blood pressure is suggested by the study of Stella et al (1998).

4 Role of A3AR in Neurodegeneration

There is major evidence regarding the role of A3AR in neurodegenerative phenomena from in vivo and in vitro studies in hypoxia/ischemia models.

Although at relatively low levels in comparison to A1 and A2A receptors, A3AR mRNA was detected in the rat and mouse brain by RT-PCR (Dixon et al. 1996) and radioligand binding (Jacobson et al. 1993). This receptor is widespread in the rat and mouse brain (see: Gessi et al. 2008). In comparison to A1 and A2A receptors, A3AR has less affinity for adenosine (10–30 nM versus 1 µM). However, since extracellular adenosine concentrations increase dramatically during ischemia (Hagberg et al. 1987), evidence now indicates that all three adenosine receptors are potential targets for therapeutic treatment of stroke.

It is well accepted that by stimulation of A1 receptors, adenosine exerts a protective role in ischemia by reducing Ca2+ influx by counteracting the presynaptic release of excitatory neurotransmitters (Corradetti et al. 1984). Moreover, by directly increasing the K+ and Cl− ion conductance, adenosine stabilizes neuronal membrane potentials, thus reducing neuronal excitability (Greene and Haas 1991). Furthermore, adenosine, through A1 receptor activation, inhibits NMDA receptor-mediated currents (de Mendonca et al. 1995) and the NMDA receptor component of synaptic potentials during hypoxia (Sebastião et al. 2001). Reductions in synaptic transmission, in cellular metabolism and in energy consumption as well as moderate lowering of the body/brain temperature are protective in ischemia. Although data converge in demonstrating a neuroprotective effect of adenosine through A1 receptors during ischemia, the use of selective A1 agonists is hampered by unwanted peripheral effects, i.e. sedation, bradycardia, hypotension (Kafka and Corbett 1996). More recently, the role of A2A receptors in ischemic neuroprotection has been studied. Most data report a beneficial effect evoked by A2A antagonists that is attributed to reduced excitotoxicity and lower production of intracellular mediators involved in transcription mechanisms that may be relevant to neurodegeneration (Chen et al. 2007; Chen and Pedata 2008).

4.1 Role of A3AR in Hypoxia/Ischemia

The studies currently in the literature concerning the role of A3AR in the pathophysiology of cerebral ischemia are rather contradictory and have been matter of discussion in several review papers (Jacobson 1998; Jacobson et al. 1999; Baraldi et al. 2000; von Lubitz 1999; von Lubitz et al. 1999, 2001).

An early in vivo study in the model of global forebrain ischemia in the gerbil showed that a selective agonist of A3AR, IB-MECA, acutely administered 15 min prior to ischemia, impaired post-ischemic blood flow, increased mortality and exacerbated the loss of hippocampal neurons (von Lubitz et al. 1994). IB-MECA administration 20 min prior to transient middle cerebral ischemia also resulted in a significant increase in infarct size(von Lubitz et al. 2001).

In agreement with a possible noxious role of A3AR in ischemia, it has been more recently shown in a model of in vitro ischemic preconditioning, that the selective A3AR antagonist, 5-propyl-2-ethyl-4-propyl-3-(ethylsulfanylcarbonyl)-6-phenylpyridine-5-carboxylate (MRS1523) applied before and during oxygen–glucose deprivation (OGD), facilitated the full recovery of CA1 hippocampal neurotransmission after a severe (7 min), irreversible OGD period (Pugliese et al. 2003). The harmful role of A3AR during in vitro OGD is confirmed by the observation that the A3AR selective antagonists MRS1523, the new antagonists, LJ1251 ((2R,3R,4S)-2-(2-chloro-6-(3-iodobenzylamino)-9H-purin-9-yl)tetrahydrothiophene-3,4-diol), the 2-arylpyrazolo[3,4-c]quinoline and 4-modified-2-aryl-1,2,4-triazolo[4,3-a]quinoxalin-1-one derivatives and the 4-bismethanesulfonylamino-2-phenyl-1,2,4-triazolo[4,3-a]quinoxalin-1-one compound, consistently abolish or delay the occurrence of anoxic depolarization (AD) and significantly prevent the irreversible disruption of excitatory neurotransmission caused by a severe (7 min) ischemic episode (Pugliese et al. 2006, 2007; Colotta et al. 2007, 2008). The appearance of AD (Pugliese et al. 2006; Tanaka et al. 1997) is strictly correlated with the extent of brain damage during ischemia both in vivo and in vitro (Somjen 2001) and alterations in AD characteristics caused by A3AR antagonists may be attributable to their actions on glutamate-mediated cellular responses (see: Pugliese et al. 2006, 2007). NMDA receptors are essential to AD initiation and propagation (Somjen 2001). The block of A3AR, by removing A3AR-mediated impairment of the feedback inhibition of glutamate release exerted by specific metabotropic glutamate receptor subtypes (Macek et al. 1998) may reduce the participation of glutamate in triggering the AD. Interestingly, Cl-IB-MECA facilitates epileptiform discharges in the CA3 area of immature rat hippocampal slices (Laudadio and Psarropoulou 2004), suggesting that following a rise of endogenous adenosine, as occurs during convulsions associated with hypoxia, A3AR facilitates excitation, thus limiting the known protective inhibitory effect of adenosine in the brain. An A3AR direct regulation of glutamate efflux appears unlikely on the basis of in vitro results showing that neither A3AR activation with Cl-IB-MECA nor its blockade with MRS1191 modify neurotoxicity caused by kainate and cyclothiazide in cultured neurons (Rebola et al. 2005).

Contrary to the above information, Hentschel et al. (2003) demonstrated that under 5 min hypoxic conditions (95% N2–5% CO2) in vitro, selective activation of A3AR by a brief (5 min) application of IB-MECA, inhibits excitatory neurotransmission on cortical neurons. Such effect is blocked by the selective A3AR antagonist MRS1220. These data indicate that A3AR may sustain inhibition of synaptic activity during hypoxia and therefore mediate neuroprotection. Furthermore, IB-MECA acutely administered 20 min after transient (30 min) focal cerebral ischemia decreases the infarct volume (von Lubitz et al. 2001). A possible protective role of A3AR is supported by the observation that mice deleted for A3AR showed more pronounced hippocampal pyramidal neuron damage following repeated episodes of moderate hypoxia (Fedorova et al. 2003) and an increase in cerebral infarction after transient ligation of the middle cerebral artery (Chen et al. 2006). In addition, intracerebroventricular or repeated intravenous administration (i.e., at 165 and 15 min before transient ligation of the middle cerebral artery) of Cl-IB-MECA decrease cerebral infarction assessed 2 days later. Cl-IB-MECA decreased the size of infarction in the wild-type controls, but not in the A3AR knockout animals, confirming that Cl-IB-MECA-induced protection was mediated through the A3AR (Chen et al. 2006). We must notice that the selective A3AR antagonist, MRS1191, administered intracerebroventricularly 30 min before transient (60 min) middle cerebral artery occlusion (MCAo) did not modify the extent of infarction (Shen et al. 2005). This observation is in agreement with the poor role of A3AR in normal physiological transmission. MRS1523 and LJ1251 do not in fact modify synaptic transmission in the CA1 area of the hippocampus under normoxic conditions (Pugliese et al. 2006, 2007).

These conflicting results on the excitatory or inhibitory role of A3AR on synaptic activity under hypoxia/ischemia may be reconciled by recent data reported by Pugliese et al. (2007). The A3AR antagonist, MRS1523, applied before a brief (2 min) OGD reduces the depression of CA1 hippocampal neurotransmission. This result indicates an inhibitory role of A3 AR on synaptic transmission during brief OGD. Indeed a depression of synaptic activity such as that brought about by adenosine A1 receptors during ischemia is considered neuroprotective. In fact, antagonists of A1 receptors reduce inhibition of synaptic transmission, impair the recovery of synaptic potentials (Sebastião et al. 2001) and shorten the onset of AD induced by hypoxia in the CA1 region of hippocampal slices (Lee and Lowenkopf 1993). Results by Brand et al. (2001), Lopes et al (2003) and Pugliese et al (2007) indicate no interplay between A3AR and A1 receptors on the inhibition of excitatory transmission. The decrease in synaptic depression brought about by A3AR antagonists after hypoxia or a brief OGD period suggests that A3AR as A1 receptor inhibit fEPSP amplitude during the first few minutes of OGD, therefore sharing a neuroprotective role with A1 receptors.

Results suggest that in the first phase of ischemia A3AR, by decreasing synaptic transmission, play a protective synergistic role with A1 receptors. However, recovery of transmission after ischaemia may be influenced by several processes, some eventually unrelated to the level of synaptic transmission during hypoxia. Prolonged ischemic conditions could play a pivotal role in switching the effects of A3AR stimulation from A1-like inhibition to potentiation of an excitotoxic glutamate effect. The activation of phospholipase C (PLC) by A3AR has been reported in striatal and hippocampal slices (Abbracchio et al. 1995). Rat cortical neurons exposed to hypoxia in vitro show an increase in activation of protein kinase C (PKC) after selective A3AR stimulation (Nieber and Hentschel 2006). Similarly to what is described in the heart, PKC-dependent activation of KATP channels may enhance adenosine protection (Liang and Jacobson 1998), but if OGD is applied long enough to be considered severe, PKC activation induced by A3AR could account for an increase in intracellular calcium, which may participate in increasing tissue excitability and thus lead to irreversible synaptic failure.

Taking into account that during ischemic conditions, adenosine is released from hippocampal slices, reaching concentrations up to 30 µM after 5-min OGD (Latini et al. 1998; Pearson et al. 2006) and that A3AR are stimulated by µM concentrations of adenosine (Fredholm et al. 2001) we may speculate that A3AR mediated effects would become particularly deleterious during ischemia, when high levels of adenosine are reached extracellularly and detrimental effects of A3AR activation may be due, at least in part, to increased excitoxicity.

Altogether these data suggest that the outcome of A3AR stimulation on synaptic transmission during hypoxic/ischemic phenomena depends on the duration and severity of the ischemic episode. Although A3AR may play a protective role in the first phase of ischemia, prolonged A3AR stimulation by high adenosine concentrations could be pivotal in transforming the A3AR-mediated effects from protective to injurious.

Interestingly, in the in vitro OGD model in hippocampal slices, it was found that a long application (before and during OGD) of 5′-N-methylcarboxamidoadenosine derivatives Cl-IB-MECA and of new selective A3 agonists (Volpini et al. 2002, 2007): VT72 (N6-methoxy-2-phenylethynyl), VT158 (N6-methoxy-2-phenylethynyl), VT160 (N6-methoxy-2-(2-pyridinyl)-ethynyl), VT163 (N6-methoxy-2-p-acetylphenylethynyl) and AR132 (N6-methyl-2-phenylethynyladenosine) have effects similar to those of antagonists upon the OGD-induced depression of synaptic transmission and on the appearance of AD after the severe (7 min) OGD period (Pugliese et al. 2007). These effects may be attributed to desensitization of A3AR. In fact, both human and rat A3AR are desensitized within a few minutes after agonist exposure (Palmer et al. 1995; Trincavelli et al. 2002). A stimulation as massive as that reached in the presence of endogenous adenosine plus exogenous A3AR agonists might induce substantial A3AR plastic adjustments such as desensitization. These in vitro results probably concord with the observation in the model of global forebrain ischemia in the gerbil (von Lubitz et al. 1994), that chronic administration (10-day pre-ischemic) of IB-MECA improves post-ischemic cerebral flow circulation, survival and neuronal preservation (von Lubitz et al. 1994, 1999) and that repeated intravenous administration of Cl-IB MECA before MCAo ligation increases locomotor activity and decreases cerebral infarction (Chen et al. 2006). Chronic preischemic administration of IB-MECA also results in significant preservation of ischemia-sensitive microtubule-associated protein 2 (Map-2), enhancement of the expression of glial fibrillary acidic protein (GFAP) and depression of nitric oxide synthase in ischemic brain tissue (von Lubitz 1999).

Summarizing the above discussed evidence, the protective/injurious effects of A3AR during ischemia appear to depend on time after the onset of the ischemic insult which is consistent with the cascade of events described after ischemia (Dirnagl et al. 1999), some of which may also relate to inflamation. Initially, massive excitoxicity may be controlled by A3AR and later, an inflammatory cascade could be potentiated by prolonged A3AR stimulation. Timing of treatment with respect to the onset of the ischemic insult may therefore account for the different effects of pre- versus post-ischemia administration of A3AR agonists. Moreover, A3AR desensitization may account for the different effects of acute versus chronic agonist treatments reported in different studies.

These observations raise the question of the time-related utility of A3AR antagonists/agonists for treatment of ischemia. It may be speculated that after ischemia, prolonged treatment with A3AR agonists first protects by reducing glutamate-mediated excitotoxicity thus supporting a depression of neuronal activity and energy save; later after ischemia because it desensitizes A3AR, avoiding late onset deleterious A3AR influences. This last-described protective effect could also be ascribed to antagonists administered late with respect to the onset of ischemia. However, the fact that preischemic stimulation of A3AR in the in vivo model of global forebrain ischemia in the gerbil aggravates cerebral damage (von Lubitz et al. 1994) dictates caution in using A3AR agonists to protection from ischemia-induced brain damage. Certainly further studies aimed at verifying the effect of agonists versus antagonists at different times after in vivo ischemia will help clarify the utility of this class of drugs in ischemia.

4.2 A3AR and Neuroinflammation

The evidence summarized in the previous section suggests that A3AR may control ischemic brain injury by controlling excitoxicity. Although excitotoxicity is invoked in the pathophysiology of most neurodegenerative central diseases, how A3AR directly contribute to modulation of brain injury is largely unknown. Under neurodegenerative conditions involving ischemia, trauma, excitoxicity and bioenergetic dysfunctions, the interplay of resident glial cells with infiltrating peripheral bone marrow-derived cells produces neuroinflammation. An important role for A3AR in modifying the inflammatory response was pointed out by first studies in mast cells where activation of these receptors appears to be responsible for release of allergic mediators contributing to inflammatory expansion (Ramkumar et al. 1993;Fozard et al. 1996; Gao et al. 2001) and mast cell degranulation (Reeves et al. 1997; Salvatore et al. 2000; Tilley et al. 2000; Zhong et al. 2003). On the other hand, exposure of blood peripheral cell lines to selective A3AR agonists resulted in both anti- and proinflammatory effects (see: Gessi et al. 2008).

Microglia, astrocytes and oligodendrocytes are cell type sensors responding to neurodegenerative phenomena in the CNS. Evidence suggests that central A3AR exert an important role in brain injury by affecting not only neurons but also glial function controlling important intracellular signaling pathways that are involved in neuroinflammation (Fig. 9.1). Below is a review of A3AR influence on glial functions.

Fig. 9.1
figure 1_9

Schematic diagram illustrating A3 adenosine receptor localization in the brain. ADO: adenosine; ADA: adenosine deaminase; ATP: adenosine triphosphate, AMP: adenosine monophospate; AKA: adenosine kinase; T: bidirezional nucleoside transporter; NPTDase: family of ecto-nucleotidases, including NPTDase 1,2,3. During cerebral ischemia, extracellular ADO concentration increases acting on A3 adenosine receptors located on different cell type

Full size image

4.2.1 Effects of A3AR in Astrocytes

A3AR mRNA has been identified by Northern blot analysis in mouse astrocytes (Zhao et al. 1999). Early evidence indicates that A3AR on astrocytes mediate both protection and cell death, depending on A3AR agonist concentration (Abbracchio et al. 1997; Yao et al. 1997; Jacobson et al. 1999; Di Iorio et al. 2002).

On astroglial cell lines (human astrocytoma ADF cells) low (nM) concentrations of the selective A3AR agonist Cl-IB-MECA induced a marked reorganization of cell cytoskeleton accompanied by induction of expression of small GTP-binding protein of the Rho family that is involved in control of actin cytoskeleton and by changes of intracellular distribution of the antiapoptotic protein Bcl-XL (Abbracchio et al. 1997, 2001).

In addition, stimulation of cultured murine astrocytes with Cl-IB-MECA induces the release of CCL-2, a chemokine which may exert neuroprotective effects (Wittendorp et al. 2004). Recently, in human D384 astrocytoma cells, Cl-IB-MECA at relatively low concentration (0.8 μM), reduced ATP depletion and apoptosis caused by hypoxic conditions. Furthermore, primary astrocytes prepared from A3AR KO mice were more affected by hypoxia than those prepared from WT mice (Bjorklund et al. 2008b). In vivo, in the ischemia model of MCAo (transient, 30 min occlusions), IB-MECA administered after ischemia proved to decrease the intensity of reactive gliosis involving microglia and astrocytes as evaluated 7 days after ischemia. (von Lubitz et al. 2001). However some data have indicated no effect of A3AR selective stimulation. In rat primary cultures, IB-MECA (1 μM) failed to modulate intracellular calcium signaling ([Ca2+]i) elicited by ATP (Alloisio et al. 2004) and no evidence was found that A3AR affects intracellular calcium levels in acutely isolated rat astrocytes (Pilitsis and Kimelberg 1998).

Conversely, a high concentration (μM) of IB-MECA induced apoptosis of various cell lines including astrocytes (Yao et al. 1997; Jacobson et al. 1999). In agreement with this evidence, in primary cultures of rat astrocytes and in C6 glial cells, it was shown that treatment with the A3AR agonist Cl-IB-MECA (10 μM) induced apoptosis and reduced the expression of endogenous Bcl-2, whereas it did not affect the expression of Bax. this suggests that intense activation of A3AR is pro-apoptotic in glial cells via bcl2 and caspase-3 dependent pathways. (Appel et al. 2001). In primary cultures of mouse astrocytes, adenosine caused an increase in [Ca(2+)i] most probably by acting on A3AR (Chen et al. 2001). In rat cultured astrocytes, apoptosis caused by adenosine was significantly reduced by the selective A3AR antagonist MRS1523 (Di Iorio et al. 2002).

On the whole, data are supportive that A3AR may exert a cytoprotective or noxious effect in astrocytes, depending on the intensity of receptor stimulation.

4.2.2 Effects of A3AR in Microglia

Several possible antiinflammatory effects are directly mediated by A3AR on microglial cells. A3AR mRNA is expressed in microglia (Fiebich et al. 1996) where it mediates several effects. Lee et al. (2006) have demonstrated that Cl-IB-MECA suppresses LPS-induced NF-kappaB activation and TNF-alpha production in mouse BV2 microglial cells. In primary mouse microglial cells and N13 microglia cell line, A3AR stimulation increases both ERK1/2 and p38 MAPK phosphorylation via phosphatidylinositol-3′-kinase (Hammarberg et al. 2003, 2004). Interestingly, a concentration-dependent effect was noticed, that is high ERK1/2 phosphorylation occurred at low A3AR agonist concentration, decreasing with increasing agonist concentration (Hammarberg et al. 2003).

Knowledge of the intracellular networks activated by adenosine A3AR may help to elucidate the pathophysiological role of this receptor. Reactive gliosis, in response to central trauma, hypoxia/ischemia and neurodegeneration, includes phenotypical alterations of microglia and astrocytes and increased astrocyte number, occurs. Oligodendrocytes first encounter damage and death. After a central trauma, reactive gliosis is generally regarded as beneficial at first but, if prolonged, it may enhance tissue damage by production of deleterious factors (Neary and Snowden 1996). Glial A3AR stimulated by high adenosine levels caused by a prolonged central trauma may well be implicated in neuroinflammatory tissue responses.

Plastic adjustments of A3AR induced by brain injury might also be relevant to the modulation of intracellular pathways and cell safety. A3AR appear to be very sensitive to prolonged stress in vitro. An up-regulation of A3AR mRNA was observed 1 h after 3-nitropropionate exposure in hippocampal slices, normalization ensued 24 h later (von Arnim et al. 2000). A3AR up-regulation was also reported in the hippocampus of a transgenic mouse model of Alzheimer’s disease (APP23tg) where impaired oxidative phosphorylation was detected prior to amyloid deposition (von Arnim et al. 2006).

5 Conclusions and Perspectives

Adenosine A3 receptors are widely distributed in the CNS but are expressed at a low level and have lower affinity for adenosine in comparison to the A1 and A2A receptors. Nevertheless, they appear to tonically modulate motor activity as pointed out in A3AR-deleted mice. The role of A3AR in several pathophysiological conditions is often controversial. In an attempt to synthesize the major evidence in the literature, it emerges that in conditions that create an extracellular increase of adenosine such as seizures or ischemia, A3AR may contribute to neurotransmission and cell damage. Recently a pro-convulsant effect of A3AR stimulation emerged, especially in the immature brain, thus raising the possibility that A3AR might facilitate seizure-induced neuronal damage. This also raises the possibility that the A3AR subtype plays a role in the establishment of activity-dependent plastic changes. Moreover, most data support a pro-nociceptive role of A3AR involving both central nervous system and pro-inflammatory effects at peripheral tissues.

Major evidence for A3AR in neurodegenerative phenomena emerges from studies performed in in vivo and in vitro models of hypoxia/ischemia. Data from the current literature suggest that the outcome of A3AR stimulation on synaptic transmission during hypoxic/ischemic phenomena depends on the duration and intensity of the ischemic episode. It has been hypothesized that while A3AR play a protective role in the first phase of ischemia by decreasing synaptic transmission, prolonged A3AR stimulation by high adenosine concentrations could be pivotal in transforming the A3AR-mediated effects from protective to injurious. Detrimental effects of A3AR activation may be due, at least in part, to increased excitoxicity. Glial A3AR stimulated by high adenosine levels caused by a prolonged central trauma may well be implicated in neuroinflammatory tissue responses. Moreover, A3AR appear very sensitive to prolonged stress in vitro and plastic adjustments of A3AR induced by brain injury and pharmacological treatment with agonists might be relevant to their final role. All these observations raise the question of the time-related utility of A3AR antagonists/agonists for treatment of ischemia. Certainly further studies aimed at verifying the effect of agonists versus antagonists at different times after in vivo ischemia will help clarify the utility of this potent class of drugs in ischemia and in different neurodegenerative diseases.