Downstream Pathways of Adenosine

Abstract

Adenosine belongs to the class of neuromodulators rather than neurotransmitters, since it is not stored in vesicles, nor released by exocytosis as a classical neurotransmitter. Moreover, it does not induce synaptic potentials but influences the release and the action of neurotransmitters. This mostly occurs through interactions with other G protein-coupled receptors as well as of receptors for neurotrophic factors, ion channels, ionotropic receptors, and neurotransmitter transporters. The actions of adenosine are operated by four different G protein-coupled membrane receptors (A₁, A_2A, A_2B, A₃), which activate several downstream signaling pathways, the main focus of the present review. Cross talk between adenosine receptors and receptors for neurotransmitters or other neuromodulators may result from interactions between common signaling cascades, as well as through receptor–receptor interactions, including receptor heteromerization. The key receptor in this synaptic interplay appears to be the A_2A receptor, whereas A₁ receptors mainly act as modulators of neurotransmitter release or by counteracting A_2A receptor-mediated actions. We herein review some of the most recent data on the regulation of adenosine availability, as well as on the consequences of adenosine actions in synapses and the corresponding downstream signaling pathways. Moreover, we discuss how activation of adenosine receptors and regulation of extracellular adenosine levels is operated by combined mechanisms. It is highlighted that modulation of neuronal activity by adenosine involves a diversity of enzymes, receptors and signaling cascades that act in a concerted way to fine tune the activity of neurons and glia, including astrocyte-to-neuron signaling.

1 Introduction

Adenosine, due to the way it operates in the nervous system, belongs to the class of neuromodulators rather than neurotransmitters (for a review on these concepts see, for instance, Ribeiro and Sebastião 2010). So, intracellular adenosine is not stored in vesicles, nor released by exocytosis as a classical neurotransmitter. Instead, adenosine is released to the extracellular space through equilibrative nucleoside transporters that function bidirectionally according to the gradient across the cell membrane. Adenosine is also formed in the extracellular space through degradation of released ATP. Once in the extracellular space, adenosine activates membrane located G protein-coupled receptors (GPCR) and through these receptors affects neuronal functioning at different levels, including changes in the ability to release or respond to neurotransmitters or even gliotransmitters, but so far, no neurotransmitter-like actions for adenosine have been identified.

Adenosine also behaves as a retaliatory metabolite, influencing and reflecting cell energy state, as well as metabolic demand and nutrient supply. More than 25 years have elapsed since it was first proposed that adenosine efficiently connects synaptic activity, energy expenditure, and nucleic acid metabolism by acting as a sensor of the bioenergetic state of the cell (Newby et al. 1985). Moreover, intracellular adenosine directly regulates transmethylation reactions, including DNA methylation (Boison et al. 2002), which can lead to long-lasting epigenetic modifications. Other than at the nerve tissue level, the influence of adenosine upon cerebral blood flow enables it to further act as an energy balancing metabolite. Thus, when metabolism is increased, the elevated ATP catabolism will produce higher amounts of adenosine, which through the activation of A_2AR, (Phillis 1989) will induce vasodilation, allowing an improvement of oxygen and nutrient delivery via the cerebral vasculature.

In the present work we discuss some of the most recent data on the regulation of adenosine availability and its effects through adenosine receptors and how their activation is regulated by extracellular adenosine levels. Downstream mechanisms of receptor activation and receptor cross talk are then reviewed on the basis of recently published data, highlighting the functional outcomes of the subtle ways adenosine fine-tunes neuronal activity.

2 Adenosine Formation

The main source of adenosine, in the central nervous system, is the dephosphorylation of 5′-AMP by 5′-nucleotidases (5′-NTs) (Meghji 1993). These enzymes dephosphorylate noncyclic nucleoside monophosphate to nucleosides and inorganic phosphate. So far, seven distinct nucleoside monophosphate phosphohydrolases or 5′-nucleotidases (EC 3.1.3.5 and EC 3.1.3.6) have been cloned. Five are localized in the cytosol; one is attached to the outer side of the plasma membrane and one in the mitochondrial matrix. Nucleotidases are responsible for both intracellular and extracellular synthesis of adenosine from the dephosphorylation of AMP. An alternative source of adenosine synthesis results from the hydrolysis of S-adenylhomocysteine (SAH), which is catalyzed by SAH hydrolase (SAHH) (Palmer and Abeles 1979) (see Fig. 7.1).

Fig. 7.1

Adenosine metabolism. Adenosine can be synthesized intra- and extracellularly. Inside the cell adenosine is formed from AMP metabolism through endo-5′-nucleotidase (5′NT) or by the transmethylation reaction catalyzed by SAHH, which converts SAH into adenosine and homocysteine. At the extracellular space, adenosine derives from the metabolism of ATP/ADP/AMP, being the last reaction catalyzed by the ectonucleotidase (Ecto 5′NT). The release of adenosine through equilibrative nucleoside transporters is an alternative source of adenosine. Regarding clearance of extracellular adenosine, in some cases it can be converted into inosine by ecto-adenosine deaminase (ecto-ADA) in the extracellular space, but in most cases adenosine is taken up by the equilibrative nucleoside transporter into cells where adenosine can be phosphorylated to AMP by adenosine kinase (ADK) or deaminated to inosine by adenosine deaminase (ADA)

2.1 Extracellular Formation of Adenosine

Adenosine found in the extracellular space can be released via equilibrative nucleotide transporters or be synthesized locally via ATP catabolism, which involves several enzymes, including the enzymes of ectonucleoside triphosphate diphosphohydrolase (E-NTPDase) family, ectonucleotide pyrophosphatase/phosphodiesterase (E-NPP) family, ecto-5′-nucleotidase/CD73, and alkaline phosphatases (Yegutkin 2008). Through this cascade, adenine nucleotides are dephosphorylated into 5′-AMP, which is then dephosphorylated by ecto-5′-nucleotidase into adenosine. The entire catalytic pathway is complete in a few hundred milliseconds, and the rate-limiting step being the dephosphorylation of AMP into adenosine by ecto-5′-nucleotidase (Dunwiddie et al. 1997).

2.1.1 Ecto-5′-Nucleotidase

After the vesicular release of ATP (which is cosecreted with neurotransmitters or even released as a neurotransmitter), ATP is metabolized by a cascade of ectonucleotidases, including ecto-ATPase, ecto-ADPase (E-NTPDase family) and apyrase (E-NPP family) and finally ecto-5′-nucleotidase, producing adenosine (Ribeiro and Sebastião 1987; Richardson and Brown 1987; Zimmermann et al. 1986). Additionally cAMP can also be released into the extracellular space, by a probenecid-sensitive transporter (Rosenberg and Li 1995) in sufficient amounts to increase extracellular adenosine concentrations (Brundege et al. 1997; Dunwiddie et al. 1992).

In neuronal cells, ectonucleotidases are able to convert most adenine nucleotides (except cAMP) into adenosine in less than a second (Dunwiddie et al. 1997). Indeed, even stable ATP analogues can be converted into adenosine by ectonucleotidases (e.g., Cascalheira and Sebastião 1992; Cunha et al. 1998). Several ectonucleotidases, including alkaline phosphatase and nucleoside triphosphate diphosphohydrolase 2, are associated with subsets of progenitor cell populations in the mouse embryonic, postnatal, and adult neurogenic zones (Langer et al. 2007). Knockdown of tissue nonspecific alkaline phosphatase impairs neural stem cell proliferation and differentiation (Kermer et al. 2010), highlighting their relevance in neurogenesis, including adult neurogenesis (see Zimmermann 2011).

The ecto-5′-nucleotidase is a cell surface protein attached to the plasma membrane by a glycosyl phosphatidylinositol (GPI) anchor at its C terminal (Misumi et al. 1990). The hydrolysis of extracellular AMP is considered the main function of this enzyme, but 5′-nucleotidase is also involved in cell adhesion, as it also binds laminin and fibronectin (Mrhul et al. 1993; Olmo et al. 1992). Ecto-5′-nucleotidase acts also as a coreceptor in T cell activation (see Resta and Yamashita 1998).

Ecto-5′-nucleotidase is highly expressed in the brain, where it is mainly associated with glial cell membranes, namely, astrocytes, oligodendrocytes, and also microglia (Kreutzberg et al. 1978; Naidoo 1962; Schoen and Kreutzberg 1995). In fact, the predominant glial expression of ecto-5′-nucleotidase is related to an enhancement in the contribution of extracellular conversion of AMP into adenosine when astrocytes are cocultured with neurons (Zamzow et al. 2008a).

Regarding localization in neurons, some initial cytochemical studies associated ecto-5′-nucleotidase with the surface of migrating and immature nerve cells and with subsets of synapses during part of their regeneration period as well as during synapse remodeling and regeneration (Schoen et al. 1991, 1993). Later on, the ecto-5′-nucleotidase expression by mature neurons was also demonstrated in the cerebellum (Maienshein and Zimmermann 1996) and in hippocampal nerve terminals (Cunha et al. 2000). More recently, it has been demonstrated that ecto-5′-nucleotidase is expressed by nociceptive neurons in dorsal root ganglia and on terminals in substantia gelatinosa of spinal cord, where the conversion of AMP into adenosine promotes antinociception (Sowa et al. 2010).

Functionally, there is some evidence that ectonucleotidases are in close physical proximity with presynaptic adenosine receptors (Cunha et al. 1996; Dunwiddie and Masino 2001), so that recently formed adenosine becomes immediately available to the presynaptic receptors involved in modulation of neurotransmitter release. Topographical arrangement of membrane bound molecules involved in purinergic signaling may determine the type of receptor activated by adenosine, since there is evidence that adenosine formed from released ATP preferentially activates facilitatory receptors (Cunha et al. 1996).

2.2 Intracellular Synthesis of Adenosine

For the net extracellular adenosine levels, intracellular synthesis of adenosine is at least as important as adenosine formation from breakdown of extracellular ATP (Lloyd and Fredholm 1995). Intracellular synthesis of adenosine occurs mainly by AMP dephosphorylation, which is catalyzed by cytosolic nucleotidases. The presence of cytosolic 5′-nucleotidase in the brain was firstly demonstrated in 1982 (Montero and Fes 1982). Although differential expression of 5′-nucleotidase among different brain areas has not been established so far, its ubiquitous role in the intracellular synthesis of adenosine is well known. Additionally, the cytosolic nucleotidases participate in substrate cycles that regulate the cellular levels of ribo- and deoxyribonucleoside monophosphates, regulating the intracellular pools of ribo- and deoxyribonucleotides (Reichard 1988) which are crucial for DNA/RNA synthesis.

Furthermore, adenosine produced inside the cell contributes to restoring ATP levels by decreasing ATP utilization and increasing oxygen and nutrients supply via blood flow (Newby 1984). Thus, adenosine is commonly considered to be a retaliatory metabolite, since adenosine produced during cytosolic ATP degradation behaves as a metabolic stress sign promoting retaliatory effects against the stress-causing conditions (Newby 1984).

Another source of adenosine is the transmethylation pathway, where adenosine results from the hydrolysis of S-adenosylhomocysteine (SAH) catalyzed by SAH hydrolase (SAHH, EC 3.3.1.1), which also produces l-homocysteine (Palmer and Abeles 1979; Schrader et al. 1981). This enzyme was firstly described in 1959, in rat liver. SAHH catalyzes a reversible reaction, that preferentially evolves towards S-adenosylhomocysteine synthesis (de la Haba and Cantoni 1959). In the heart, this pathway provides one-third of the total cardiac adenosine at normoxic conditions but generates undetectable levels under hypoxic conditions (Deussen et al. 1989). SAHH expression is widespread in the brain, with higher expression levels present in cortex and cerebellum. Inside the cell, SAHH displays a nuclear expression where it is involved in transmethylation mechanisms. In detail, different methyltransferases convert S-adenosylmethionine (SAM) into SAH, which is then metabolized into adenosine and l-homocysteine by SAHH. So, adenosine is an obligatory end product of SAM-dependent transmethylation reactions and because of that is able to inhibit methylation reactions. To avoid this inhibition, adenosine is phosphorylated into AMP by a long isoform of adenosine kinase (ADK), which was described as a nuclear ADK (Cui et al. 2009).

Under normal conditions SAHH has low impact upon neuronal excitability (Pak et al. 1994), suggestive of a minor role in the control of cytoplasmic levels of adenosine in neurons.

3 Nucleoside Transporters

The relevance of adenosine uptake by nucleoside transporters in terminating adenosine effects was first supported by different studies showing that the nucleoside transporter blockade produces vasodilation, potentiates the ability of adenosine to decrease locomotor activity (Crawley et al. 1983), depresses neuronal activity (Motley and Collins 1983), increases nociceptive thresholds (Yarbrough and McGuffin-Clineschmidt 1981), and exerts anticonvulsive effects (Dragunow and Goddard 1984). Therefore, nucleoside transporter inhibitors exacerbate the effects mediated by adenosine.

Nucleoside transporters can be divided in two main classes: the equilibrative (Na⁺-independent) nucleoside transporters (ENTs) and the concentrative (Na⁺-dependent) nucleoside transporters (CNTs) (Baldwin et al. 1999). Six isoforms of CNTs (CNT1–CNT6) and four isoforms of ENTs (ENT1–ENT4) have been cloned, to date. Equilibrative transporters mediate nucleoside transport in both directions, depending on the nucleoside concentration gradient across the membrane. The four transporters are widely distributed, and all of them are able to transport adenosine but they have different abilities to transport other nucleosides (Baldwin et al. 2004). The transport mediated by concentrative transporters is independent of nucleoside gradient and is coupled to sodium gradient. As intracellular concentrations of adenosine are kept low due to its conversion into AMP, and as catabolism of released nucleotides constitutes an additional and transporter-independent source of extracellular adenosine, the extracellular concentrations of adenosine are usually higher than the intracellular ones. Therefore, the usual direction of equilibrative adenosine transport is uptake into cells, rather than release. Indeed, adenosine uptake in the brain occurs primarily by facilitated diffusion via equilibrative transporters, although some of it (10–20 %) can be mediated by concentrative transporters (Geiger and Fyda 1991; Parkinson et al. 1994).

Equilibrative nucleoside transporters are crucial to regulate the levels of extracellular adenosine, being the main entity responsible for removing adenosine from the extracellular space. They are thus responsible for restraining, both spatially and temporally, adenosinergic modulation. Due to the equilibrative nature of the adenosine transporters in neuronal cells, changes in the activity of enzymes involved in ADO metabolism will modify the transporters’ activity. Accordingly, transporter inhibitors can either increase (Dunwiddie and Diao 1994; Phillis et al. 1989; Sanderson and Scholfield 1986) or decrease (Gu et al. 1995) extracellular adenosine levels, depending on the transmembrane adenosine gradient and consequently depending on transport direction, into or out of the cell. However, because the extracellular synthesis of adenosine from catabolism of nucleotides constitutes an alternative source of adenosine, which is not affected by transport blockade, the transporters inhibitors usually lead to an increase in the extracellular levels of the nucleoside.

The amount of adenosine released by nucleoside transporters is enhanced under some circumstances such as hypoxia or ischemia, when a massive increase in extracellular adenosine levels is observed, a process prevented by transporter blockade (Parkinson et al. 2002). At the synaptic level, however, the rise in extracellular levels of adenosine during hypoxia may increase rather than decrease upon inhibition of equilibrative nucleoside transporters (Frenguelli et al. 2007). Furthermore, recent evidence (Zhang et al. 2011) showed that neuronal nucleoside transporters contribute to the removal of extracellular adenosine from the synaptic space even during hypoxic or ischemic insults. Therefore, evidence now available allows suggesting that the control of extracellular adenosine levels may differ in different microdomains. As highlighted recently (Sebastião 2011), a deeper understanding of those microdomains as well as of the relative contribution of the different cell types (i.e., neurons vs. astrocytes) for the net production of adenosine is required to better predict the direction of the changes in adenosine levels after pharmacologic or genetic manipulation of adenosine transporters in pathological conditions.

Interestingly, adenosine release by nucleoside transporters is promoted by neurotransmitters. For example, glutamatergic agonists such as NMDA and kainate increase, in a dose-dependent manner, adenosine release (Carswell et al. 1997; Delaney et al. 1998). In fact, activation of NMDA receptors seems to promote release of adenosine itself instead of its precursor, ATP (Harvey and Lacey 1997; Manzoni et al. 1994). This may be part of a protective feedback loop since adenosine released through the transporters seems to preferentially activate adenosine A₁ receptors (A₁R; Cunha et al. 1996) and these are neuroprotective, namely, through inhibition of NMDA currents not only under normoxic (de Mendonça et al. 1995) but also under hypoxic (Sebastião et al. 2001) conditions.

4 Intracellular Adenosine Clearance

After being taken up through nucleoside transporters, adenosine is inactivated either by deamination through adenosine deaminase or by phosphorylation through adenosine kinase (Fig. 7.1). It is accepted that the pathway responsible for intracellular adenosine clearance is dependent on its concentration. As such, at low concentrations, adenosine is mainly inactivated by phosphorylation while at higher concentrations adenosine is predominantly deaminated by adenosine deaminase (Meghji and Newby 1990), in accordance with the affinity for adenosine and enzymatic capacity of those two enzymes.

4.1 Adenosine Kinase

ADK (EC 2.7.1.20) phosphorylates intracellular adenosine into AMP. Due to its low K _m for adenosine, it is the main enzyme responsible for intracellular adenosine catabolism, at least, for low adenosine concentration. ADK is therefore a key target whenever manipulation of the neuromodulatory actions of adenosine is desirable. By phosphorylating adenosine into AMP, ADK has a double role for maintaining a homeostatic energy flux: (1) a direct ability to influence the cellular energy pool (AMP, ADP, and ATP) and (2) an influence upon intra- and extracellular levels of the homeostatic regulator, adenosine. The relevance of ADK for the homeostatic control (Boison et al. 2011) is supported by several lines of evidence, namely, (1) the release of higher amounts of adenosine by ADK-deficient fibroblasts in cultures, when compared to that released by ADA-deficient fibroblasts (Huber et al. 2001), (2) the ability of ADK inhibition to depress neuronal activity in hippocampal slices, in a way sensitive to A₁R antagonists (Diógenes et al. 2004; Pak et al. 1994), and (3) the suppression of seizure activity caused by ADK inhibition in various models for epilepsy (Kowaluk and Jarvis 2000).

The immature brain is more vulnerable to seizure activity than the adult brain (Moshe 2000), an action probably related to the higher expression of ADK at early developmental stages (Studer et al. 2006). Interestingly, during maturation, there is a shift from neuronal to glial expression of ADK, suggestive of distinct functions of ADK and adenosine in immature and adult brain; thus, during neuronal development expression of ADK in neurons may provide a salvage pathway to utilize adenosine in nucleic acid synthesis, whereas in the mature brain predominant ADK expression in astrocytes contributes to maintenance of tonic adenosinergic inhibition in the central nervous system (Studer et al. 2006). Overexpression of adenosine kinase in epileptic hippocampus contributes to epileptogenesis (Gouder et al. 2004). Furthermore, there is a prominent upregulation of ADK in astrocytes after induction of status epilepticus (SE) in animals as well as in humans with temporal lobe epilepsy (Aronica et al. 2011). Selective ADK downregulation in astrocytes almost completely abolishes spontaneous recurrent seizures in epileptic mice (Theofilas et al. 2011). Thus, ADK emerges as a key link in astrocyte-to-neuron communication, and its dysregulation after intense neuronal activity may contribute to epileptogenesis. Permanent changes in ADK expression in astrocytes will be reflected in decreases in ambient adenosine, leading to a further enhancement of neuronal activity and in such a way being part of a positive feedback loop to promote epileptogenesis. Accordingly, focal adenosine augmentation therapeutic strategies, mainly based in local manipulation of ADK activity, have been proposed as a useful strategy to control pharmacoresistant seizures (Boison et al. 2011).

The regulation of ambient adenosine levels by ADK might also have a key role in the susceptibility of brain tissue to ischemic injury (Lynch et al. 1998; Pignataro et al. 2007; Shen et al. 2011). Indeed, pharmacological inhibition of ADK in animal models is also an effective strategy to protect from stroke (Boison 2006; Kowaluk and Jarvis 2000).

4.2 Adenosine Deaminase

Adenosine deaminase (ADA, EC 3.5.4.4) catalyzes the hydrolytic deamination of adenosine into inosine. It has been known for several years that inhibition of ADA causes adenosine-like effects such as sedation (Major et al. 1981; Radulovacki et al. 1983), reduction of the infarct area in the hippocampus and decrease in neuronal degeneration in animal models of global forebrain ischemia or focal ischemia (Lin and Phillis 1992; Phillis and O’Regan 1989). Although ADA is expressed by both neurons and astrocytes (Haun et al. 1996; Nagy et al. 1984) it seems that it is in glial cells that this enzyme assumes a major role in the control of adenosine levels. This role is more relevant during stress conditions (like trauma or ischemia), when adenosine levels rise and astrocytes became reactive, probably playing an important role in adenosine conversion to inosine (Zamzow et al. 2008b). Inosine by itself can have a protective effect in stroke models (Shen et al. 2005). However, neuroprotection conferred by inhibitors of ADA during hypoxia or ischemia (Lin and Phillis 1992) mostly results from potentiation of the stress-induced increase in intracellular adenosine, which leads to enhanced adenosine release through transport reversal (Phillis and O’Regan 1989).

Although the enzyme localization is mainly cytosolic, there is evidence of the existence of an ectoenzyme, bound to the extracellular side of the membrane (Franco et al. 1998). The A₁R may act as an anchoring protein for ecto-ADA, which through a nonenzymatic but allosteric interaction facilitates agonist and antagonist binding to A₁R (Ciruela et al. 1996; Gracia et al. 2008; Ruíz et al. 2000; Saura et al. 1998). Like A₁R, A_2BR was also found to be anchored to ADA in lymphocytes and cultured cells. Similarly, binding of enzymatically active or inactive ADA to this receptor increases its affinity and signaling by a protein–protein interaction (Herrera et al. 2001).

5 Control of Extracellular Adenosine Levels by Astrocytes

In the brain, extracellular adenosine concentrations are normally kept in the range of 25–250 nM, therefore at concentrations that can tonically activate a substantial proportion of the high affinity A₁R and A_2AR (Dunwiddie and Masino 2001). A major player in the steady-state levels of adenosine is ADK, which has high affinity for adenosine and is mostly expressed in astrocytes (see above and Boison et al. 2010). ADA also predominates in astrocytes. Equilibrative nucleoside transporters are also expressed in astrocytes. Therefore, under physiological conditions astrocytes probably function as a major sink for adenosine, since its uptake is driven by the intracellular activity of ADK. It is also likely that under conditions that prompt increases in intracellular adenosine, such as hypoxia or ischemia, astrocytes provide a major source of adenosine, which is released by reversal of transport direction, but direct evidence for a predominant astrocytic origin of extracellular adenosine during hypoxic/ischemic insults is still lacking.

The levels of extracellular adenosine can be regulated by both A₁R and A₂R activity since A₁R blockade increases extracellular levels of adenosine in cardiac fibroblasts (Andresen et al. 1999) and activation of A₂R promotes adenosine transport in chromaffin cells (Delicado et al. 1990). Furthermore, it has been shown that A_2AR activation at nerve endings enhances the activity of nucleoside transporters, leading to a decrease in the availability of adenosine to activate A₁R under high frequency neuronal firing (Pinto-Duarte et al. 2005). Again, information is still lacking regarding the role of astrocytes in this process. In astrocytes, A₁R and A_2AR form tetramers constituted by two A₁R and two A_2AR molecules bound to G_i/0 and G_s proteins to regulate GABA transport in a deeply interactive and concerted way (Cristóvão-Ferreira et al. 2011). Whether these A₁R-A_2AR tetramers control adenosine transporters is also unknown.

6 Purines and Intracellular Signaling

ATP acts upon different classes of membrane receptors, the ionotropic P2X and the metabotropic P2Y (for reviews see, for instance, Illes and Ribeiro 2004; Ralevic and Burnstock 1998). Adenosine operates through activation of four distinct metabotropic receptors: A₁R, A_2AR, A_2BR, and A₃R. All these receptors are expressed in the brain, where adenosine is involved in a variety of physiological and pathological processes, namely, regulation of sleep–arousal cycle, neuroprotection, epilepsy, pain, fine control of movement, fine-tuning of neurotransmission (see Boison 2006; Dunwiddie and Masino 2001; Ribeiro 2005; Sebastião and Ribeiro 2009a).

7 Adenosine Receptors and Signaling Pathways

All four adenosine receptors have been cloned. Being GPCRs, adenosine receptors are formed by a single peptide chain, with seven alpha-helical transmembrane domains, an intracellular C-terminal, and an N-terminal facing the extracellular space. The N-terminal usually contains one or more glycosylation sites. The C-terminal contains phosphorylation and palmitoylation sites, which are involved in regulation of receptor desensitization and internalization (Perez and Karnik 2005).

Adenosine receptors are widely distributed throughout the body. In the brain they can be found pre-, post-, or nonsynaptically, in neurons as well as in glia. Their expression is not homogenous in the central nervous system (Fig. 7.2). Higher A₁R expression levels are found in the cortex, hippocampus, cerebellum, thalamus, brain stem, and spinal cord (see Ribeiro et al. 2002 and references therein). Though at low density, A₁Rs are also present in basal ganglia, both on dopaminergic nigrostriatal and glutamatergic corticostriatal terminals. Adenosine A_2ARs are mostly expressed in the basal ganglia and olfactory bulb. However, it is possible to find mRNA encoding A_2ARs or the A_2AR protein in other brain regions where they are weakly expressed, namely, in the hippocampus and the cortex (see Sebastião and Ribeiro 1996). The A_2BRs are mainly expressed in peripheral organs, being weakly expressed in the whole brain (Dixon et al. 1996). Finally, A₃Rs have an intermediate expression level in the human cerebellum, and as A_2BRs, they display a low expression in the entire brain (see Fredholm et al. 2001). The A₁R and A_2AR are high affinity receptors, with adenosine K _d values of 70 and 150 nM, for A₁R and A_2AR respectively, which allow their tonic activation by basal adenosine levels. The A_2BR and A₃R are considered low affinity receptors, with adenosine affinity constant values around 5,100 and 6,500 nM, respectively (see Dunwiddie and Masino 2001).

Fig. 7.2

Differential distribution of adenosine receptors (A₁, A_2A, A₃) among the brain (adapted from Ribeiro et al. 2002). Higher expression corresponds to larger text size

Classically, A₁R and A₃R inhibit adenylate cyclase (AC) through coupling to G_i/0. A_2AR and A_2BR are coupled to G_s or G_olf, promoting AC activity. The A_2BR subtype is also coupled to G_q/11, through which it can activate phospholipase C (Ryzhov et al. 2006). The A₃R can also couple to G_q/11, also activating phospholipase C (Fredholm et al. 2001). The increase of cAMP mediated by AC leads to activation of cAMP dependent protein kinase (PKA), which then phosphorylates different targets such as ionotropic receptors or neurotransmitter transporters. On the other hand, activation of phospholipase C converts phosphatidylinositol 4,5-bisphosphate (PIP₂) into diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP₃). Then DAG activates protein kinase C (PKC), which phosphorylates different substrates, while IP₃ triggers calcium release from intracellular stores. Then, elevation of cytosolic Ca²⁺ can stimulate a variety of signaling pathways, including a family of phosphatidyl serine-dependent serine/threonine-directed kinases collectively called protein kinase C (PKC), phospholipase A₂ (PLA₂), as well as Ca²⁺-dependent K⁺ channels, and nitric oxide synthase (NOS). IP₃ can also promote calcium influx from extracellular sources if Ca²⁺ intracellular stores are depleted due to previous activation of IP₃ receptors (see Ralevic and Burnstock 1998).

The activation of G proteins can also directly modify the activity of several enzymes and ion channels that directly or indirectly influence intracellular calcium levels. For example, A₁R, via G_i/0 activation, leads to activation of several types of K⁺ channels and to blockade of N-, P-, and Q-type Ca²⁺ channels (see Fredholm et al. 2001). Direct evidence that A₁R can inhibit calcium channels in nerve terminals under hypoxic conditions has been reported (Coelho et al. 2002). A₃R are also able to modulate Ca²⁺ levels, through the inhibition of AC. Both A_2AR and A_2BR can also modify the levels of intracellular calcium (see Fredholm et al. 2001).

Aside from the involvement of AC/cAMP/PKA and PLC/IP₃-DAG/PKC, other transduction pathways are associated with adenosine receptor activation, namely, the mitogen-activated protein kinase (MAPKs) (Schulte and Fredholm 2003). The MAPK family is constituted by three main groups: extracellular regulated kinases (ERK) such as ERK1 and ERK2, stress-activated protein kinases (SAPK) such as p38, and jun-N-terminal kinase (JNK). These kinases are usually activated by receptors with tyrosine-kinase activity (Seger and Krebs 1995), but GPCR can also signal through them (Gutkind 1998; Liebmann 2001; Marinissen and Gutkind 2001). In fact, all adenosine receptors can affect the MAPK pathway. This was first shown in COS-7 cells transiently transfected with A₁R, leading to activation of ERK1/2 via G_i/0 (Faure et al. 1994). It was also early recognized that activation of A_2AR can increase MAPK activity (Sexl et al. 1997). Interestingly, the signal pathway used by A_2AR to activate MAPK can vary, depending on the cellular machinery available. Thus, in CHO cells, A_2AR-mediated ERK1/ERK2 activation involves G_s-AC-cAMP-PKA-MEK1, while in HEK-293 cells, MAPK activation by A_2AR involves PKC but not PKA, even though cAMP levels are found to be enhanced by G_s activity (Seidel et al. 1999). A_2AR can also inhibit ERK phosphorylation (Hirano et al. 1996). The activation of A_2BR can trigger the three main branches of MAPK family (ERK1/2, p38, and JNK) (see Fredholm et al. 2001). Finally, the A₃R activate ERK1/2 in human fetal astrocytes (Neary et al. 1998). Also, the phosphorylation of ERK1/2 was also clearly demonstrated in CHO cells transfected with A₃R (see Schulte and Fredholm 2000).

To conclude, MAPK activation by adenosine receptors is quite similar to that prompted by other GPCR (Gutkind 1998; Luttrell et al. 1999; Sugden and Clerk 1998). Interestingly, ERK1/2 phosphorylation is promoted either by receptors coupled to G_s (A_2A/A_2B) or to G_i/0 (A₁/A₃) proteins. The MAPK-mediated effects of adenosine receptors mainly impact modulation of DNA synthesis, cellular differentiation, proliferation, and apoptosis (see Schulte and Fredholm 2003).

Adenosine receptor activity is regulated by its expression at the membrane level which results from a balance between endocytosis and exocytosis rates. An enhancement of the endocytosis rate, which restrains the intensity and duration of the signal, is often preceded by receptor phosphorylation and uncoupling from G proteins, a well-known process of receptor desensitization. Adenosine receptor subtypes desensitize differently. A₁R are slowly phosphorylated and internalized, a time frame of several hours being needed to complete the process. A_2AR and A_2BR desensitize in a faster way, with downregulation kinetics less than 1 h. The A₃R have the fastest desensitization profile, a process often occurring within minutes (Klaasse et al. 2008).

8 Implications for Modulation of Neuronal Function

The variety of downstream pathways operated by adenosine receptors highly supports the pluripotential of this nucleoside to interfere with a multiplicity of intracellular functions essential to regulate neuronal activity either directly or indirectly via interaction with several neurotransmitters and/or neuromodulators. These interactions can occur within the same cell, in some cases involving receptor heteromerization, or be a result of transcellular communication. Being a small and easily diffusible molecule, adenosine easily acts in a paracrine way, affecting cells away from the release point. Its role as a trans-synaptic modulator, involving neuron-to-astrocyte communication at tripartite synapses is now well accepted (Fields and Burnstock 2006; Hamilton and Attwell 2010; Perea et al. 2009).

Within the same cell there are many possibilities of cross talk between transduction pathways that have several kinases and other key molecules in common (see Fig. 7.3). GPCRs can interact at the G protein level, by sharing βγ-subunits or common α-subunits, affecting the activation kinetics of other GPCRs. This also applies to A₁R and A_2AR, and related mechanisms are most probably involved in the ability of adenosine receptors to interact with other receptors for neurotransmitters or neuromodulators (Sebastião and Ribeiro 2000).

Fig. 7.3

Schematic representation of the different signaling pathways associated with adenosine receptors. Adenosine receptors are GPCRs. A₁R and A₃R couple to G_i/0, inhibiting AC, which will reduce cAMP levels and consequently decrease PKA activity. A_2AR and A_2BR are coupled to G_s, promoting AC activity and consequently PKA activity. A₃R and A_2BR can also couple to G_q/11, enhancing PLC activity. PLC catalyzes PIP₂ into DAG and IP₃. DAG will directly activate PKC, while IP₃ will increase intracellular Ca²⁺ levels. Furthermore, all adenosine receptors can activate the MAPK pathway

A₁R and A_2AR can form heteromeric complexes (Ciruela et al. 2006). As clearly shown in astrocytes, A₁R-A_2AR heteromers appear as heteromers of homomers with a minimal structure consisting of an A₁R-A₁R-A_2AR-A_2AR complex (Cristóvão-Ferreira et al. 2011). The heterotetramer makes it possible to accommodate the two different G proteins, and the A₁R-A_2AR heteromer in astrocytes is clearly coupled to G_i/0 and G_s proteins (Cristóvão-Ferreira et al. 2011). Importantly, the blockade of a single partner in the A₁R/A_2AR/G_i/0/G_s complex leads to adjustments in the functioning of the whole unit (Cristóvão-Ferreira et al. 2011).

Heteromerization between adenosine receptors and receptors of other neurotransmitters/neuromodulators also occurs, being the A_2AR-D₂R heteromer the first to be recognized (Hillion et al. 2002). Through this heteromer, adenosine restrains D₂R-mediated effects. The relevance of dopaminergic signaling and dysfunction in several pathologies turns A_2AR-D₂R heteromers into promising therapeutic targets (Altamura et al. 2005; Ferré et al. 1997). A close interaction between A₁R and D₁R was also described (Ginés et al. 2000). Once again, adenosine, through the activation of A₁R, inhibits D₁R-mediated effects. In detail, A₁R activation leads to uncoupling of D₁R from AC (Ginés et al. 2000), reinforcing the A₁R-induced inhibition of D₁R, which is mediated by G_i/0 activation (Ferré et al. 1994, 1998).

Adenosine A_2AR can also heteromerize with cannabinoid CB₁R in the striatum (Carriba et al. 2007), and this has putative implications for pharmacotherapy drug addiction (Ferré et al. 2010). Psychomotor stimulation by A_2AR antagonists also depends upon A_2AR-CB₁R cross talk (Lerner et al. 2010). A₁R also interact with CB₁R receptor-mediated actions in the hippocampus, a process with implications for CB₁R induced memory impairment, which is exacerbated by chronic caffeine consumption (Sousa et al. 2011).

The predominant neuromodulatory action of adenosine, inhibition of neurotransmitter release, is controlled by A₁R and relates to presynaptic inhibition of calcium responses (Fredholm et al. 1990; Fossier et al. 1999; Ribeiro 1978). However, A_2AR can also modulate extracellular transmitter levels and they do so either by enhancing release and/or by influencing uptake. Indeed, A_2AR activation in the hippocampus facilitates GABA release (Cunha and Ribeiro 2000) and GABA uptake into presynaptic terminals (Cristóvão-Ferreira et al. 2009) and astrocytes (Cristóvão-Ferreira et al. 2011). In the striatum, where A_2AR expression is higher, the activation of these receptors leads to an inhibition of GABA uptake (Gonzalez et al. 2006). Glutamate release from hippocampal (Lopes et al. 2002) and cortical (Marchi et al. 2002) synaptosomes is also enhanced by A_2AR activation. The same occurs with acetylcholine release from hippocampal nerve terminals (Cunha et al. 1995).

During excitotoxic conditions, such as hypoxia, the A₁R-mediated presynaptic inhibition of calcium responses (Coelho et al. 2002), together with inhibition of NMDA responses (Sebastião et al. 2001) confers protection against synaptic damage. In contrast, A_2AR may facilitate ionotropic receptor activation, as it is the case of their ability to enhance AMPA receptor-mediated responses at the postsynaptic and extrasynaptic level, affecting the reserve of the GluR₁-containing AMPA receptors at the extrasynaptic pool, priming them for synaptic insertion and for reinforcement of synaptic strength (Dias et al. 2012). This action is sustained even after brief A_2AR activation, involves cyclic AMP and PKA activation and leads to enhancement of long-term potentiation (LTP) in Schaffer collateral-CA1 synapses of the hippocampus (Dias et al. 2012). LTP facilitation by A_2AR is also evident in aged animals (Costenla et al. 2011). Interestingly, A_2AR blockade in vivo impairs conditional learning as well as potentiation of CA1 hippocampal potentials recorded concomitantly in freely moving animals (Fontinha et al. 2009). A_2AR localized postsynaptically at synapses between mossy fibers and CA3 pyramidal cells are essential for a form of long-term potentiation (LTP) induced by short bursts of mossy fiber stimulation, which requires the activation of NMDA and metabotropic glutamate receptors (mGluR₅) to increase cytoplasmic Ca²⁺ levels (Rebola et al. 2008).

A_2AR seem to be devoted to interacting with other metabotropic receptors, not only of the GPCR family (Sebastião and Ribeiro 2000) but also with tropomyosin-related kinase (Trk) receptors (Sebastião and Ribeiro 2009b). A_2AR are able to transactivate TrkB receptors in the absence of the neurotrophin (Lee and Chao 2001). This transactivation requires long-term incubation with A_2AR agonist and requires receptor internalization (Rajagopal et al. 2004). Furthermore, adenosine A_2AR activation is also a crucial step for the functioning of neurotrophic receptors at synapses, through a mechanism most probably different from TrkB transactivation, and which involves translocation of TrkB molecules to lipid rafts (Assaife-Lopes et al. 2010). The A_2AR-mediated gating and/or facilitation of the actions of neurotrophins, has been shown for the facilitatory actions of brain derived neurotrophic factor (BDNF) on synaptic transmission (Diógenes et al. 2004; Tebano et al. 2008) and on plasticity (Fontinha et al. 2008) at the CA1 area of the hippocampus. The actions of BDNF are blocked by either A_2AR blockade or inhibition of Trk phosphorylation, but a Trk phosphorylation inhibitor does not prevent A_2AR-mediated facilitation of synaptic transmission (Pousinha et al. 2006, indicating that A_2AR operate upstream of TrkB activation. Synaptic actions of other neurotrophic factors, such as glial derived neurotrophic factor (GDNF) are also under influence of A_2AR in the striatum (Gomes et al. 2006, 2009). Adenosine A_2AR and BDNF TrkB receptors can coexist in the same nerve ending since the facilitatory action of A_2AR upon TrkB-mediated BDNF action is also visible at the neuromuscular junction (Pousinha et al. 2006), a single nerve ending synapse model. Colocalization of A_2AR and Ret, a component of the GDNF receptor complex, has also been shown in single axon terminals in the striatum (Gomes et al. 2009).

The ability of BDNF to facilitate synaptic transmission and synaptic plasticity is dependent on the age of the animals (Diógenes et al. 2007, 2011) and this may be related to the degree of activation of adenosine A_2AR by endogenous adenosine at different ages. Thus, to trigger a BDNF facilitatory action at synapses of infant animals it is necessary to increase the extracellular levels of adenosine, either by inhibiting ADK or by a brief depolarization (Diógenes et al. 2004; Pousinha et al. 2006) or even by high frequency neuronal firing (Fontinha et al. 2008). These adenosine-triggered BDNF actions are lost by blocking adenosine A_2AR with selective antagonists. In adult animals, BDNF per se can facilitate synaptic transmission, but this effect is also fully lost with blockade of adenosine A_2AR (Diógenes et al. 2007) or in A_2AR knockout mice (Tebano et al. 2008). Interestingly, the enhanced hippocampal synaptic plasticity in aged animals can be related not only to a higher influence of adenosine A_2AR (Costenla et al. 2011) but also to an enhanced BDNF TrkB-mediated facilitatory tonus also dependent from cross talk with A_2AR (Diógenes et al. 2011). Nicotinic alpha7 cholinergic currents in GABAergic hippocampal neurons are inhibited by BDNF, and this also requires coactivation of adenosine A_2AR (Fernandes et al. 2008). Inhibition of GABA transporters (GAT) by BDNF at nerve terminals does not fully depend upon coactivation of A_2AR, since it is not abolished by A_2AR blockade, though this inhibitory BDNF action can be exacerbated by A_2AR coactivation (Vaz et al. 2008). Interestingly, in astrocytes BDNF facilitates GABA transport, and this facilitation is fully dependent upon A_2AR activation (Vaz et al. 2011). Contrasting with A_2AR which promote the actions of neurotrophic factors, A₁R inhibit neurite outgrowth of cultured dorsal root ganglion neurons, both in the absence and in the presence of NGF (Thevananther et al. 2001).

A_2AR, due to their ability to enhance excitotoxic phenomena, including glutamate release (Lopes et al. 2002; Marchi et al. 2002), are mostly regarded as promoters of neuronal death. However, in some cases, such as cultured retinal neurones, A_2AR have been shown to protect neurones against glutamate-induced excitotoxicity (Ferreira and Paes-de-Carvalho 2001). Whether this is due to the ability of A_2AR to facilitate actions of neurotrophic factors, as it has been shown to occur in relation to A_2AR-mediated neuroprotection of motor neurones (Wiese et al. 2007), is not yet known. The pathophysiological implications of the cross talk between A_2AR and receptors for neurotrophic factors have been discussed in detail elsewhere (Sebastião and Ribeiro 2009c).

Adenosine receptor activation may also induce release of neurotrophic factors. Thus, the expression and/or release of NGF are enhanced by activation of A_2AR in microglia (Heese et al. 1997) and by activation of A₁R in astrocytes (Ciccarelli et al. 1999). Adenosine A_2BR in astrocytes are also able to enhance GDNF expression (Yamagata et al. 2007). In the whole hippocampus, A_2AR are required for normal BDNF levels (Tebano et al. 2008). Interestingly, in a mouse model of Huntington’s disease, A_2AR are also required to keep striatal BDNF levels close to those obtained in wild-type mice (Potenza et al. 2007).

Finally, interactions among purinergic, growth factor, and cytokine signaling regulate neuronal and glial maturation as well as development. In neuronal-dependent glial maturation both ATP and adenosine purinoceptors are involved (see, for instance, Fields and Burnstock 2006). The extracellular adenosine levels during high frequency neuronal firing are sufficient to stimulate adenosine receptors in oligodendrocyte ancestor cells inhibiting their proliferation and stimulating their differentiation into myelinating oligodendrocytes (Stevens et al. 2002). In premyelinating Schwann cells, A_2AR activate phosphorylation of extracellular signal-regulated kinases (ERKs), namely, ERK1/2, and inhibit Schwann cell proliferation without arresting differentiation (Stevens et al. 2004). Contrasting with A_2AR, which usually promote the actions of neurotrophic factors, adenosine A₁R inhibit neurite outgrowth of cultured dorsal root ganglion neurons, both in the absence and in the presence of NGF (Thevananther et al. 2001).

Besides influencing the activity of other neuromodulatory receptors, adenosine A_2AR also affect the activity of equilibrative transporters, as in the case of adenosine transporters (Pinto-Duarte et al. 2005). Thus, activation of A_2AR with the selective agonist CGS 21680 facilitates adenosine uptake and enhances release of adenosine, which points to a direct effect of A_2AR on nucleoside transporters, rather than an indirect action resulting from a modification of the adenosine gradient of concentrations across the plasma membrane (i.e., a metabolic effect). Furthermore, high frequency neuronal firing activates A_2AR and concomitantly enhances nucleoside transporters. The main consequence of this A_2AR-mediated enhancement of nucleoside transporters is a marked reduction of the tonic activation of inhibitory A₁R upon high-frequency firing. This action of A_2AR on the activity of the adenosine transporters constitutes a clear demonstration that a neuromodulatory receptor is able to control the extracellular levels of its endogenous ligand and, hence, to influence its ability to control neurotransmitter release (Pinto-Duarte et al. 2005).

A_2ARs facilitate GAT-1 mediated GABA transport into nerve terminals, an effect that is mediated by AC activation, which restrains the inhibition of GAT-1 by PKC (Cristóvão-Ferreira et al. 2009). Transport facilitation by A_2ARs is due to an increase in the membrane expression of GAT-1 molecules, reflected in increased maximum transport velocity (Cristóvão-Ferreira et al. 2009). This A_2AR-mediated facilitation of GABA transport into nerve endings, if coupled to an increase in the release of GABA (Cunha and Ribeiro 2000), may contribute to faster neurotransmitter recycling, leading to an enhancement of phasic GABAergic signaling.

A_2ARs also facilitate GABA transport into astrocytes, by enhancing GAT-1 and GAT-3 mediated transport, an action under tight control of A₁R, due to formation of A₁R-A_2AR heteromers (Cristóvão-Ferreira et al. 2011). While the A₁ protomer of the heteromer inhibits GABA transport, the A_2A protomer enhances it, the shift from inhibition to enhancement of GABA uptake probably occurring at low micromolar concentrations of extracellular adenosine (Cristóvão-Ferreira et al. 2011). These concentrations are easily attained at a tripartite synapse, where astrocytes and neurons release considerable amounts of ATP. The higher the release of ATP (as may occur at high neuronal firing rates in reciprocal neuron-to-astrocyte communication) the higher the expected concentration of extracellular adenosine. It is therefore likely that sustained neuronal firing promotes activation of the A_2AR protomer of the A₁R-A_2AR heteromer leading to facilitation of GABA uptake. Activation of GABA uptake by astrocytes will lead to a decrease in ambient GABA and a subsequent depression of tonic GABAergic inhibition resulting in enhanced excitatory tonus. Conversely, at submicromolar adenosine concentrations, there is a preferential activation of the A₁ protomer of the heteromer and so, GABA uptake by astrocytes would be inhibited. Consequently, tonic inhibition by GABA would be enhanced. Thus, through an adenosine action upon A₁R-A_2AR heteromers, astrocytes might behave as dual amplifiers, facilitating excitation of intense astrocytic-to-neuronal signaling and increasing inhibition at low neuronal firing rates. This switch in neural activity requires a highly efficient control to avoid sudden state transitions, and this seems to be the main advantage of heteromerization of A₁R and A_2AR in astrocytes. Indeed, overstimulation of just one of the receptor protomers leads to internalization of the whole functional unit (Cristóvão-Ferreira et al. 2011), therefore allowing a double brake in the system and avoiding an abrupt inhibitory signaling and a sudden switch from excitation to inhibition as a consequence of desensitization of only the excitatory protomer.

9 Concluding Remarks

Operating on multiple downstream signaling pathways, adenosine receptors influence the activity of other GPCRs as well as of receptors for neurotrophic factors, ion channels, ionotropic receptors, and neurotransmitter transporters. Modifications of extracellular adenosine levels, due to changes in its metabolic pathways, lead to alterations in the degree of activation of adenosine receptors, which will impact their ability to enhance or restrain the action of other neurotransmitters or neuromodulators. The cross talk between adenosine receptors and other membrane receptors results in part from intracellular cascade processes occurring between common transducing systems and through protein phosphorylation processes that involve PKC or PKA (Fig. 7.3). The key receptor in this synaptic interplay appears to be the A_2AR, whereas the A₁R can counteract A_2AR mediated actions.

In summary, synaptic transmission is under tight control of endogenous extracellular adenosine, which through pre- and postsynaptic actions interplays with other synaptic molecules involved in neurotransmission as well as with membrane proteins (receptors, and transporters) essential for transmission to harmonically influence neuronal activity.