|Home | About | Journals | Submit | Contact Us | Français|
Adenosine A2A receptors are highly enriched in the basal ganglia system. They are predominantly expressed in enkephalin-expressing GABAergic striatopallidal neurons and therefore are highly relevant to the function of the indirect efferent pathway of the basal ganglia system. In these GABAergic enkephalinergic neurons, the A2A receptor tightly interacts structurally and functionally with the dopamine D2 receptor. Both by forming receptor heteromers and by targeting common intracellular signaling cascades, A2A and D2 receptors exhibit reciprocal antagonistic interactions that are central to the function of the indirect pathway and hence to basal ganglia control of movement, motor learning, motivation and reward. Consequently, this A2A / D2 receptors antagonistic interaction is also central to basal ganglia dysfunction in Parkinson's disease. However, recent evidence demonstrates that, in addition to this postsynaptic site of action, striatal A2A receptors are also expressed and have physiological relevance on presynaptic glutamatergic terminals of the cortico-limbic-striatal and thalamo-striatal pathways, where they form heteromeric receptor complexes with adenosine A1 receptors. Therefore, A2A receptors play an important fine-tuning role, boosting the efficiency of glutamatergic information flow in the indirect pathway by exerting control, either pre- and/or post-synaptically, over other key modulators of glutamatergic synapses, including D2 receptors, group I metabotropic mGlu5 glutamate receptors and cannabinoid CB1 receptors, and by triggering the cAMP-protein kinase A signaling cascade.
The basal ganglia are a richly interconnected neural network involved in adaptive control of behavior through interactions with sensorimotor, motivational and cognitive brain areas (Graybiel et al., 1994). The striatum is the major input structure of the basal ganglia (Fig. 1). 90% of striatal neurons are medium-size spiny neurons, named for their high density of dendritic spines. This population of GABAergic striatal efferent neurons directly receives most striatal inputs from various intrinsic and extrinsic sources. Of these, the two predominant inputs are glutamatergic afferents from cortical, limbic and thalamic areas and dopaminergic afferents from the mesencephalon, either the substantia nigra pars compacta or the ventral tegmental area. Both inputs converge in the dendritic spine but have different roles: whereas the glutamatergic input serves as a trigger of striatal circuits, the dopaminergic input subserves a crucial modulatory role, since it is unable to trigger electrical responses in medium spiny neurons in the absence of glutamatergic inputs. The morphological organization of these inputs is consistent with these functional roles: the glutamatergic terminal makes synaptic contact with the head of the dendritic spine, while the dopaminergic terminal makes synaptic contact with the neck of the dendritic spine. In this way, dopaminergic neurotransmission is in a position to regulate glutamatergic neurotransmission (Gerfen, 2004).
The striatum, and hence GABAergic striatal efferent neurons, is critical for the control of movement. Indeed, it is essential both for the selection and initiation of actions (Graybiel et al., 1994) and for the learning of habits and skills (Graybiel 1995; White, 1997), a task mainly controlled by the dorsal striatum. GABAergic striatal efferent neurons also have a prominent role in motivation and reward, in which the ventral striatum plays a central role. Both motor control and motivation/reward are highly dependent on modulation by dopamine. A leading hypothesis for striatal function is that it contributes to the formation of stimulus–response associations through reinforcement learning (Schultz et al. 2003). It is therefore proposed that the striatum processes reward signals that are encoded by the association of dopaminergic input with sensory cues from the cortico-limbic-thalamic striatal projections, to generate appropriate behavioral responses to given contextual situations. In addition, circuits in the striatum generate learning and memory about behavioral responses and reward attainment through use-dependent long-term changes in synaptic efficacy (Wickens et al., 2003).
Adenosine is another very important modulator of striatal glutamatergic neurotransmission through its actions on adenosine receptors (Fredholm et al., 2001), which are highly abundant in the striatum (see below). However, the aim and mechanisms of adenosinergic modulation are fundamentally different from dopaminergic modulation. In fact, whereas dopamine is an extrinsic signal (depending on the firing of neurons whose cell bodies are localized in the substantia nigra pars compacta), adenosine is an intrinsic signal since it is locally produced as a function of the activity of striatal circuits. There seem to be two main sources of extracellular adenosine. First, there is the hormonal-like formation of adenosine whereby extracellular levels of adenosine increase as a function of the general workload of the circuit. Thus, with increasing workload (i.e. increased firing per unit of time), greater consumption of ATP in neurons and astrocytes is required to maintain ionic homeostasis and membrane potential. Dephosphorylation of ATP (present intracellularly in the millimolar range) generates adenosine, which levels increase substantially over basal intracellular levels, which are in the nanomolar range. This build-up of intracellular adenosine is translated into increased levels of extracellular adenosine since virtually all cell types are equipped with bidirectional and non-concentrative nucleoside transporters (Geiger and Fyda, 1991). This mechanism ensures that there is a local fluctuation of extracellular adenosine levels as a function of local activity in the striatum.
In parallel, there is a second mechanism that generates extracellular adenosine, which might be particularly related to the control of synaptic activity since it specifically generates a synaptic pool of adenosine. Several studies have provided evidence for the formation of extracellular adenosine as a result of the action of ecto-nucleotidases (see Zimmermann, 2000) on ATP released upon nerve stimulation (reviewed in Cunha, 2001). ATP can be released with most neurotransmitters since it is present in synaptic vesicles together with different neurotransmitters, including glutamate (reviewed in Cunha and Ribeiro, 2000). In this manner, the synaptic pool of adenosine is replenished as a function of neuronal firing (Cunha, 2001). Extracellular ATP can also be originated from glia (Zhang et al., 2003), since ATP is stored and released from synaptic vesicles in astrocytes (Koizumi et al., 2003; Newman, 2003). The relative contribution of the different cellular (astrocytes or neurons, presynaptically or non-synaptically) and metabolic sources of extracellular adenosine (released as such or formed from released ATP) is still not well-defined. However, irrespective of its source, under physiological conditions extracellular levels of adenosine are always expected to build up locally as a function of the workload of the system.
GABAergic striatal efferent neurons display particular passive and active membrane properties that shape their intrinsic excitability and their responsiveness to synaptic inputs. Electrophysiological recordings performed in vivo showed that these neurons present a unique type of spontaneous electrical behavior, consisting of fluctuations of the membrane potential between two preferred potentials, the “up-state” close to firing threshold and a “down-state” near the potassium equilibrium potential (Wilson and Kawaguchi, 1996; Stern et al., 1997, 1998). It has been proposed that the down-state is due to an inwardly rectifying potassium conductance, the major membrane conductance of these neurons at rest, and that transitions from down- to up-state are mainly triggered by the powerful excitatory glutamatergic synaptic input. This means that these neurons are not intrinsically bistable. However, even if these transitions are triggered by excitatory input, they are also strongly influenced by intrinsic conductances and their regulation. Although such transitions do not spontaneously occur in brain slice preparations, mimicking the synaptic input either by stimulation of the cortico-limbic-striatal fibers or by application of NMDA in the bath can restore this specific behavior in vitro (Vergara et al., 2003; Olson et al., 2005)
GABAergic striatal efferent neurons can be classified into two major classes according to their peptide expression: GABAergic enkephalinergic and GABAergic dynorphinergic neurons (Gerfen, 2004) (Fig. 1). In addition, there are different types of GABAergic interneurons (parvalbumin, calretinin or somatostatin internerons) and large cholinergic interneurons (Tepper and Bolam, 2004). In the dorsal striatum the GABAergic dynorphinergic and enkephalinergic neurons give rise to two striatal efferent pathways, which connect the dorsal striatum with the output structures of the basal ganglia, the substantia nigra pars reticulata and the internal segment of the globus pallidus (GPi; entopeduncular nucleus in rodents) (Gerfen, 2004) (see Fig. 1 for a schematic representation of basal ganglia circuitry). These are called “direct” and “indirect” pathways. In the direct pathway, GABAergic dynorphinergic neurons directly connect the striatum with the output structures. The indirect pathway consists of GABAergic enkephalinergic neurons, which connect the striatum with GABAergic neurons in the external segment of the globus pallidus (GPe; globus pallidus in rodents), which project to glutamatergic neurons in the subthalamic nucleus (STN), which in turn connect the STN with the output structures. GPe GABAergic neurons also project directly to the output structures without using the STN relay (Gerfen, 2004). Because of these differences in connectivity, stimulation of the direct pathway results in motor activation and stimulation of the indirect pathway produces motor inhibition. Dopamine, or dopamine agonists, will induce motor activation by activating the direct pathway (via D1 receptors localized in GABAergic dynorphinergic neurons) and by depressing the indirect pathway (acting on inhibitory D2 receptors localized in GABAergic enkephalinergic neurons) (Gerfen, 2004) (Fig. 1).
Different from the dorsal striatum, the ventral striatum (mostly represented by the nucleus accumbens) receives glutamatergic input from limbic and paralimbic cortices, the amygdala and hipoccampus, and dopaminergic input from the ventral tegmental area (Parent and Hazrati, 1995; Gerfen, 2004). The same two classes of GABAergic efferent neurons found in the dorsal striatum exist in the ventral striatum: here, GABAergic enkephalinergic cells project to the ventral pallidum, which is the equivalent of the GPe in the dorsal striatopallidal complex. However, unlike the dorsal striatum, a high proportion of ventral GABAergic enkephalinergic neurons also express substance P and ventral GABAergic dynorphinergic neurons also project to the ventral pallidum (Robertson and Jian, 1995; Lu et al., 1998; Zhou et al., 2003). Furthermore, the ventral pallidum has characteristics of both the GPe and the GPi in its afferent and efferent systems and, therefore, it can also be considered an output structure of the basal ganglia (reviewed in Ferré, 1997).
One of the most notable features of these two subtypes of GABAergic striatal efferent neurons is the segregation of dopamine and adenosine receptors subtypes (Fig. 1). Thus, GABAergic enkephalinergic neurons express predominantly facilitatory A2A adenosine receptors and inhibitory D2 dopamine receptors, while GABAergic dynorphinergic neurons express predominantly inhibitory A1 adenosine and facilitatory D1 dopamine receptors (Schiffmann et al., 1991a, 1993; Fink et al., 1992; Ferré et al., 1991, 1997, 2005; Svenningsson et al., 1999; Gerfen, 2004). Furthermore, there is a tight interplay between the adenosine and dopamine receptors in each of the two striatal pathways. Previous studies have shown that antagonistic interactions between A1 and D1 receptors modulate the function of the GABAergic dynorphinergic neuron and antagonistic interactions between A2A and D2 receptors modulate the function of the GABAergic enkephalinergic neuron (Ferré et al., 1997, 2005). It has been hypothesized that these interactions provide a mechanism of action for the depressant motor effects of adenosine agonists and for the motor stimulant effects of adenosine antagonists, like caffeine. Indeed, the adenosine system is the major target for methylxanthines, a class of substances that includes popular psychostimulants such as caffeine and theophylline. Caffeine is a competitive antagonist at both A1 and A2A receptors, which under normal conditions are activated by endogenous adenosine (Fredholm et al., 1999).
It was first suggested in the early nineties that the A2A-D2 receptor interactions could provide a new therapeutic strategy for Parkinson's disease, mostly based on the positive results seen following co-administration of A2A receptor antagonists with L-DOPA or preferential D2 receptor agonists (Ferré et al., 1997). More recently, it has been demonstrated that, when co-transfected into mammalian cell lines, A2A receptors form heteromeric complexes with D2 receptors and A1 receptors form heteromeric complexes with D1 receptors (Ginès et al., 2000; Hillion et al., 2002; Canals et al., 2003; Ciruela et al., 2004). It is believed that these heteromeric receptor complexes are mainly localized in the perisynaptic ring adjacent to the “postsynaptic density” (PSD) of the glutamatergic synapses on dendritic spines of GABAergic dynorphinergic and enkephalinergic neurons (Ferré et al., 2005).
Since the early 1990's, imaging techniques including in situ hybridization, immunocytochemistry, and binding autoradiography have all demonstrated high enrichment of A2A receptors in the different parts of the striatum (caudate-putamen, nucleus accumbens and olfactory tubercle) in all mammalian species studied (from mouse to human) (Fig. 2) (Schiffmann et al., 1991a; 1991b, 1993; Svenningsson et al., 1999; Rosin et al., 1998, 2003), strongly suggesting their involvement in basal ganglia functions. As stated above, in situ hybridization analyzed at the cellular level revealed the specific expression of A2A receptor in GABAergic enkephalinergic striatopallidal neurons (Schiffmann et al., 1991a, 1993, Svenningsson et al., 1999).
The development of positron emission tomography (PET) techniques and selected radiopharmaceuticals has allowed imaging and quantification of defined molecular targets located in the brains of living subjects. These in vivo PET methods allow correlation of molecular data with defined clinical or behavioral presentations of human and animal subjects. In this context, the basal ganglia have received great attention in PET research. The development of analytical techniques based on the combined use of morphological and functional imaging (magnetic resonance imaging or MRI, and PET, respectively) has allowed improved analysis of both the dorsal and ventral aspects of basal ganglia. For instance, striatal dopaminergic afferents may be successfully visualized in normal or pathological conditions using tracers that selectively interact with biological targets expressed on dopaminergic nerve endings, such as the amino acid decarboxylase, the dopamine transporter, or the vesicular dopamine transporter. Dopaminergic receptors present on GABAergic efferent neurons have also been visualized in pathological conditions using a number of PET ligands (Fig. 3A). The existence of tracers selective for D1- and D2-like receptors has permitted the separate investigation of the direct and indirect pathways. Finally, the sensitivity of tracers such as [11C]raclopride (Fig. 3A) to extracellular levels of dopamine has extended the use of PET to the in vivo measurement of basal and stimulated dopamine levels.
Both adenosine A1 and A2A receptors are present at high density in the basal ganglia and have been visualized in vivo. A1 receptors may be efficiently visualized using selective PET ligands like [11C]FR194921 and [18F]CPFPX (Matsuya et al., 2005; Meyer et al., 2005). The first in vivo imaging of A2A receptors in animals and humans was obtained using the xanthine derivative [11C]KF18446 (Ishiwata et al., 2000; 2005). However, its binding profile did not completely overlap with the distribution of A2A receptors, showing a similar accumulation in thalamus and striatum. The kinetic behaviour in rodents and monkeys of the non-xanthine antagonist and [4,3-e]1,2,4-triazolo[1,5-c]pyrimidine derivative [11C]SCH442416 indicated its potential use for in vivo imaging of A2A adenosine receptors (Todde et al., 2000). Indeed, ex vivo experiments in rats (Fig. 3B, 3C) and in vivo experiments in monkeys (Fig. 3E) showed a high abundance of [11C]SCH442416 labelling in the striatum, both in its ventral and dorsal parts (Moresco et al., 2005). The sensitivity of [11C]SCH442416 to changes in A2A receptor density is demonstrated by the significant reduction of striatal [11C]SCH442416 binding observed in rats pre-treated with quinolinic acid (QA, Fig. 3D), a neurotoxin that selectively destroys striatal neurons (Ishiwata et al., 2002).
The D1 receptor is a Gs-Golf-coupled receptor, whose main signaling pathway is the cAMP-protein kinase A (PKA) cascade. D1 receptor stimulation activates adenylyl-cyclase, which increases the formation of cAMP, leading to the activation of PKA. In the GABAergic dynorphinergic neuron, the A1 receptor, which is a Gi-coupled receptor, antagonistically modulates D1 receptor function by inhibiting adenylyl cyclase activation and by means of direct interaction in the A1-D1 receptor heteromer, by which stimulation of A1 receptors decreases the binding of dopamine to the D1 receptor (Ferré et al., 1998; Ginès et al., 2000). The D1 receptor modulates neuronal excitability and glutamatergic neurotransmission by inducing PKA-mediated phosphorylation of different substrates such as ion channels, the dopamine and cAMP-regulated phophoprotein, 32 kDa (DARPP-32), NMDA and AMPA glutamate receptors, and the transcription factor cAMP response element binding protein (CREB).
D1 receptor activation also results in PKA-mediated phosphorylation of voltage-dependent Na+ channels (Surmeier et al., 1992; Schiffmann et al., 1995), leading to decreased Na+ current and, hence, to reduced neuronal excitability (Schiffmann et al., 1995). On the other hand, D1 receptor-induced phosphorylation of L-type voltage-dependent Ca2+ channels increases their conductance (Surmeier et al., 1995) resulting in facilitation of neuronal excitability (Hernandez-Lopez et al., 1997). These modulations may explain the functional consequences of D1 receptor activation on the transitions between down- and up-state. At the down-state level, D1 receptor activation will mostly decrease the probability of shifting to the up-state and of evoking spikes whereas when the striatal neuron succeeds in making a transition to the up-state, D1 receptor activation will promote increased spiking.
PKA-mediated phosphorylation of NMDA receptors (in the C terminus of the NR1 subunit) potentiates NMDA receptor-mediated currents (Greengard et al., 1999) and there is also evidence for a tight cross-talk between the second messenger pathways of D1 and NMDA receptors (Dudman et al., 2003). These functional interactions depend on heteromerization between the D1 receptor and specific subunits of the NMDA receptor (Lee et al., 2002).
Another important D1 receptor-dependent PKA substrate is DARPP-32 (Greengard et al., 1999). DARPP-32 is highly expressed in both striatonigral dynorphinergic and striatopallidal enkephalinergic neurons (Ouimet et al., 1998), where it acts as a modulator of the cAMP-PKA pathway (Fienberg et al., 1998). Phosphorylation catalyzed by PKA at Thr34 converts DARPP-32 into an inhibitor of protein phosphatase-1 (PP-1) (Hemmings et al., 1984), whereas phosphorylation catalyzed by cyclin dependent kinase-5 (Cdk-5) at Thr75 converts DARPP-32 into an inhibitor of PKA (Bibb et al., 1999). In this way, regulation of DARPP-32 produces opposite biochemical effects depending on the site of phosphorylation. Activation of D1 receptors results in PKA-catalyzed phosphorylation of DARPP-32 at Thr34 (Nishi et al., 1997; Svenningsson et al., 1998). Phospho[Thr34]DARPP-32 amplifies the effects of PKA by inhibiting PP-1 and reducing dephosphorylation of downstream target proteins, such as voltage-dependent Ca2+ and Na+ channels, and NMDA, AMPA, and GABAA receptors (Nairn et al., 2004) with a series of functional consequences (Schiffmann et al., 1998). This positive feedback on protein phosphorylation provided by DARPP-32 appears to be necessary to elicit full behavioral responses produced by activation of the cAMP-PKA cascade in striatal efferent neurons. For example, the hyperlocomotor effect of cocaine, a drug which increases Thr34 and decreases Thr75 phosphorylation via stimulation of D1 receptors (Bibb et al., 2001; Svenningsson et al., 2000), is strongly attenuated in DARPP-32 knockout mice (Fienberg et al., 1998).
In addition, D1 receptor-mediated PKA activation leads to changes that have implications for synaptic plasticity, such as phosphorylation of AMPA receptors and increase in gene transcription. Phosphorylation of AMPA receptors plays a crucial role in the initial plastic changes in glutamatergic synapses seen in long term potentiation (LTP) and depression (LTD). Phosphorylation of AMPA receptors increases their channel conductance and is associated with their recruitment to the postsynaptic density (PSD) (Song and Huganir, 2002) as demonstrated following D1 receptor-mediated phosphorylation of striatal AMPA receptors (Wolf et al., 2003). These early changes involved in synaptic plasticity are transitory and quickly reversible, but could be followed by gene transcription and protein synthesis, inducing more permanent phenotypic changes, as formation of new spines or spine pruning (Segal, 2005). PKA activated by D1 receptor stimulation can translocate to the nucleus and phosphorylate the nuclear constitutive transcription factor CREB (Greengard et al., 1999). Activation of CREB is involved in the D1 receptor-mediated increase in expression of immediate-early genes, such as c-fos, NGFI-A and jun-B and in the increased expression of the preprodynorphin gene, which encodes the precursor of dynorphin (Xu et al., 1994).
The adenosine A2A receptor is functionally very similar to the D1 receptor. We can say that the A2A receptor is the equivalent of the D1 receptor in the enkephalinergic neuron. Thus, the main signaling pathway of the A2A receptor, which is also a Gs-Golf-coupled receptor, is the stimulation of the cAMP-PKA cascade (Fig. 4). A2A receptors form heteromers with the D2 receptor by means of an electrostatic epitope-epitope interaction between an Arg-rich domain of the D2 receptor (localized in its long third intracellular loop) and a pSer localized in the C-terminus of the A2A receptor (Ciruela et al., 2004). This interaction is very similar to that involved in D1-NMDA receptor heteromerization. The D2 receptor is coupled to Gi proteins and antagonizes A2A receptor function by inhibiting adenylyl cyclase activation (Kull et al., 1999; Hillion et al., 2002). In addition, there is a reciprocal intramembrane A2A-D2 receptor interaction, by which stimulation of the A2A receptor decreases binding of dopamine to the D2 receptor (Ferré et al., 1991).
The A2A receptor can also modulate neuronal excitability and synaptic transmission. By modulating intrinsic properties as well as the balance between excitatory and inhibitory inputs, it can control the activity of these striatopallidal enkephalinergic neurons over both short and long term. As mentioned before, the membrane potential of striatal neuron undergoes state transitions in vivo (Wilson and Kagawashi, 1996) that can be mimicked in brain slice preparations by application of NMDA (Vergara et al., 2003; Olson et al., 2005). These transitions require ion channels that may be regulated by striatal transmitters acting through G protein-coupled receptors such as D2 and A2A receptors. Preliminary experiments indicate that D2 receptor activation inhibits the down- to up-state transition in a subpopulation of striatal neurons and abolishes the firing of these neurons in the up-state. A2A receptor activation fully counteracts these effects of D2 receptor, whereas application of an A2A receptor agonist alone has no effect (Azdad et al., 2006). These observations suggest that, unlike the D1 receptor, A2A receptor activation triggers a PKA-independent mechanism which remains to be fully determined, but which could involve the antagonistic A2A-D2 receptor intramembrane interaction. A similar mechanism was proposed for the interaction of these receptors in a human neuroblastoma cell line, in which D2 receptor stimulation counteracted depolarization-induced Ca2+ influx through voltage-dependent Ca2+ channels and A2A receptor stimulation antagonized the D2 effect (Salim et al., 2000).
A2A receptor activation also regulates striatal and pallidal inhibitory transmission. A2A receptor stimulation enhances the GABAA-mediated inhibitory postsynaptic currents (IPSCs) evoked in principal neurons of the globus pallidus (Shindou et al., 2001, 2002, 2003). This effect involved the cAMP-PKA pathway since it could be mimicked by cAMP analogs and occurred presynaptically in terminals of GABAergic enkephalinergic efferent neurons. This effect confirms that the A2A receptor positively regulates GABA release in the globus pallidus (Ferré et al., 1993; Mayfield et al., 1996). On the other hand, A2A receptor agonists decreased GABAA-mediated IPSCs evoked in the striatal efferent neurons by intra-striatal stimulation (Mori et al., 1996). This effect was also mimicked by cAMP analogs and was proposed to be mediated at a presynaptic site either on recurrent terminals of the GABAergic enkephalinergic efferent neurons, on terminals of GABAergic interneurons or via a disynaptic pathway involving both terminals. Reports showing that A2A receptor antagonists blocked the depressant effects of kainate receptors on large-sized striatal IPSCs favour the latter hypothesis (Chergui et al. 2000).
Modulation of excitatory glutamatergic neurotransmission at cortico-limbic-striatal and thalamo-striatal synapses is a main target for A2A receptor regulation of striatal function. This regulation seems to occur both at the post-and pre-synaptic levels (see below). At the post-synaptic side, A2A receptor activation reduced the amplitude of NMDA-mediated inward currents (Norenberg et al., 1997, 1998) and of the NMDA component of the excitatory postsynaptic currents (EPSCs) (Gerevich et al., 2002; Wirkner et al., 2004) in a subpopulation of striatal efferent neurons. One report suggests that the postsynaptic effect is independent of the cAMP-PKA pathway and appears to involve phospholipase C-induced stimulation of the Ca2+-calmodulin kinase II pathway (Wirkner et al., 2000). The same group showed that A2A receptor activation did not affect the AMPA-mediated inward current or the AMPA component of excitatory postsynaptic potentials (EPSCs; Norenberg et al., 1997; 1998; Gerevich et al., 2002) despite its ability to lead to PKA-induced phosphorylation of GluR1 subunit of the AMPA receptor (see below). Reports of stimulation of the Ca2+-calmodulin kinase II pathway by A2A receptors are rare. Instead, other groups have shown that striatal NMDA currents are increased upon activation of the cAMP-PKA pathway (Colwell and Lewine, 1995), the prototypal cascade induced by A2A receptors. Clearly, additional research is required to clarify which pathways are activated by A2A receptor stimulation.
Considerable evidence indicates the importance of DARPP-32 in the interaction between A2A and D2 receptors in GABAergic enkephalinergic neurons. Activation of A2A receptors results in increased PKA-dependent phosphorylation of DARPP-32 at Thr34 (Svenningsson et al., 2000) and decreased phosphorylation at Thr75 (Lindskog et al., 2002). The opposite regulation of the cAMP-PKA cascade exerted by A2A and D2 receptors is reflected by the observation that the increase in phosphorylation of DARPP-32 at Thr34 produced by the A2A receptor agonist CGS21680 is counteracted by quinpirole, a D2-like receptor agonist (Lindskog et al., 1999). The negative regulation exerted by D2 receptors on the phosphorylation of DARPP-32 at Thr34 is further demonstrated by studies showing that, in intact animals, administration of D2 receptor antagonists, such as eticlopride and haloperidol, increases phosphoThr34-DARPP-32 levels (Svenningsson et al., 2000; Håkansson et al., 2006). This effect depends on tonic dopamine D2 receptor activation and some maintenance of basal cAMP production and PKA activity by endogenous adenosine via A2A receptors. In line with this interpretation it has been shown that the increase in phosphoThr34-DARPP-32 produced by eticlopride is prevented by administration of KW-6002, an A2A receptor antagonist (Svenningsson et al., 2000), and is strongly reduced in A2A receptor knock out mice (Svenningsson et al., 2000). In summary, the state of Thr34 phosphorylation of DARPP-32 in GABAergic enkephalinergic neurons is in large part controlled by the opposing actions of adenosine at A2A receptors, and dopamine at D2 receptors. These competing effects on the cAMP-PKA-DARPP-32 cascade account for the opposite regulation of protein phosphorylation (Håkansson et al. 2006) and gene expression by A2A and D2 receptors (for review, see Fisone et al., 2004).
The functional significance of changes in the state of phosphorylation of DARPP-32 is demonstrated by studies using adenosine receptor antagonists such as caffeine. Blockade of A2A receptors reduces basal cAMP production, suppressing phosphorylation of DARPP-32 at Thr34 (Andersson et al., 2005) and increasing phosphorylation at Thr75 (Lindskog et al., 2002). In addition, the motor stimulant effect typically exerted by caffeine, or by selective A2A receptor antagonists, is attenuated in DARPP-32-deficient mice (Lindskog et al., 2002). Therefore, DARPP-32 not only promotes responses to drugs that activate cAMP signalling in GABAergic efferent neurons, but also amplifies the behavioural effects produced by inhibition of the cAMP-PKA cascade. This latter action is most likely exerted via a parallel pathway involving increased phosphorylation of Thr75 and further inhibition of PKA activity.
Recent work has led to the identification of downstream target proteins regulated by the cAMP-PKA-DARPP-32 pathway and involved in the control of GABAergic enkephalinergic neuron excitability. It is well established that activation of D2 receptors negatively modulates striatal glutamatergic function via inhibition of AMPA receptor transmission. Thus, AMPA current amplitude is reduced by quinpirole (Hernández-Echeagaray et al., 2004), and glutamatergic transmission is potentiated in D2 receptor knockout mice (Cepeda et al., 2001). Recently, it has been demonstrated that, in the dorsal striatum, activation of D2 receptors decreases phosphorylation of GluR1 at the PKA site, Ser845 (Håkansson et al., 2006). This finding suggests that the inhibition exerted by D2 receptors on glutamate transmission is mediated, at least in part, via suppression of PKA-catalyzed phosphorylation of AMPA receptors. Blockade of D2 receptors results in increased phosphorylation of GluR1 at Ser845 (Håkansson et al., 2006). This effect is dependent on tonic PKA activation in GABAergic enkephalinergic neurons, as it is blocked by an adenosine A2A receptor antagonist, but not by a dopamine D1 receptor antagonist. Furthermore, eticlopride does not induce phosphorylation of GluR1 in DARPP-32-deficient mice, or in mice in which the phosphorylation site for PKA in GluR1 is replaced by a non-phosphorylable Ala (Håkansson et al., 2006). It therefore appears that, in GABAergic enkephalinergic neurons, D2 receptor antagonist-dependent phosphorylation of AMPA receptors occurs via disinhibition of the cAMP-PKA-DARPP-32 cascade. Hypolocomotion induced by D2 receptor blockade is thought to occur following activation of GABAergic enkephalinergic neurons, inhibition of thalamocortical neurons and reduction of motor cortex activity (Parr-Brownlie and Hyland, 2005). In this regard, it is conceivable that the enhancement in glutamate transmission produced in GABAergic enkephalinergic neurons by eticlopride and haloperidol, via DARPP-32-dependent GluR1 phosphorylation, is involved in the motor depressant effect produced by these drugs. This possibility is supported by the observation that catalepsy produced by raclopride, a potent D2 receptor antagonist, is attenuated in DARPP-32 deficient mice (Fienberg et al., 1998).
As with the D1 receptor, A2A receptor-mediated PKA activation can potentially phosphorylate several substrates involved in synaptic plasticity. However, to our knowledge, A2A receptor-mediated phosphorylation of L-type voltage dependent Ca2+ channels or NMDA receptors has not been reported. Furthermore, A2A receptor blockade does not produce a decrease in the basal PKA-mediated phosphorylation of GluR1, although it counteracts GluR1 phosphorylation induced by D2 receptor blockade (Hakansson et al., 2006). This seems to be related to the fact that, under basal conditions, there is a strong tonic activation of D2 receptors in the striatum that impairs the ability of A2A receptor to signal through the cAMP-PKA cascade. Given these observations, what are the conditions that allow A2A receptors to fully activate PKA? It has been shown that co-stimulation of the group I metabotropic glutamate receptor mGlu5 allows A2A receptor stimulation to override the inhibitory tone imposed by endogenous dopamine via D2 receptors (Ferré et al., 2002).
The mGlu5 receptor is a Gq-coupled receptor located primarily in the perisynaptic ring (Kennedy, 2000), although some of our recent studies suggest that they can also be located at the PSD together with A2A receptors (Rodrigues et al., 2005) (Fig. 4). mGlu5 functionally interacts with the NMDA receptor and is anatomically linked to the NMDA receptor by means of intermediate scaffolding proteins of the PSD. A2A and mGlu5 receptors form heteromeric receptor complexes in transfected cells and in the rat striatum. mGlu5 receptor stimulation potentiates the effects of A2A receptors both at the intramembrane level by increasing its ability to inhibit dopamine D2 receptor binding (Popoli et al., 2001; Domenici et al., 2004), and at the mitogen-activated protein kinase (MAPK) level (Ferré et al., 2002; Nishi et al., 2003). In fact, central co-administration (in the lateral ventricle) of a selective A2A receptor agonist and a selective mGlu5 agonist induces an increase in the striatal expression of c-fos, while no significant effect is obtained when each is administered alone (Ferré et al., 2002). Furthermore, the synergistic interaction between A2A and mGlu5 receptors is in notable agreement with the recently reported synergistic effects of A2A and mGlu5 receptor antagonists in animal models of Parkinson's disease (Coccurello et al., 2004; Kachroo et al., 2005).
Changes in the efficiency of glutamatergic synapses (i.e. plasticity) do not only depend on postsynaptic changes, but also on presynaptic changes. Specifically, there is clear evidence for increases in the probability of vesicular neurotransmitter release associated with LTP and decreases associated with LTD (Schulz, 1997; Sola et al., 2004; Ronesi and Lovinger, 2005). Vesicular neurotransmitter release follows a cycle, with filling, docking, fusion and recycling. The protein machinery involved in vesicular fusion seems to play a key role in presynaptic plasticity. Stimulation of G protein-coupled receptors localized in the glutamatergic terminals can significantly modify the probability of neurotransmitter release by acting on the mechanisms involved in vesicular fusion. There are two major reported mechanisms involved in the modulation of neurotransmitter release by G protein-coupled receptors. Stimulation of Gi protein-coupled receptors, such as A1 receptors (Fig. 4), induces a decrease in the probability of neurotransmitter release (Lovinger and Choi, 1995; Flagmeyer et al., 1997) by direct inhibition of N- and P-Q-type voltage dependent Ca2+ channels by beta-gamma subunits of Gi proteins (Wu and Saggau, 1997; Jarvis and Zamponi, 2001) as well as direct actions on the release machinery (reviewed in Silinsky et al., 1999). Stimulation of Gs protein coupled receptors, such as A2A receptors (Fig. 4), on the other hand, increases the probability of neurotransmitter release by a cAMP-PKA-dependent mechanism (Evans and Morgan, 2003; Leenders and Sheng, 2005) or alternatively, through PKC-mediated facilitation of neurotransmitter release (reviewed in Fredholm et al., 2005). In the striatum, A2A receptors are located in glutamatergic nerve terminals and their activation enhances the evoked release of glutamate (Rodrigues et al., 2005; Ciruela et al., 2006). This raises the question of whether the modulatory effects operated by A2A receptors in the striatum are only due to post-synaptic A2A receptors or if the presynaptic A2A receptors located in glutamatergic terminals may also play a relevant role.
Moreover, it is also worth mentioning that activation of A2A receptors has been repeatedly suggested to result in a stimulation of striatal dopamine as well as acetylcholine release (Brown et al., 1990; Zetterstrom and Fillenz, 1990; Kurokawa et al., 1994; Okada et al., 1996; 1997; but see Jin and Fredholm, 1997). Although still a subject of debate, this was supported by a decreased basal dopamine level in the striatum of A2A receptor-deficient mice (Dassesse et al., 2001). A2A receptor activation also leads to the stimulation of GABA release by striatopallidal neurons as detected in the globus pallidus (Ferré et al., 1993; Shindou et al., 2001; 2002; 2003).
Most of the receptor-receptor interactions first described post-synaptically in GABAergic enkephalinergic neurons apparently also exist presynaptically to control glutamate release from cortico-limbic-thalamic nerve terminals. In fact, there is also a presynaptic co-localization of A2A and mGlu5 receptors in striatal glutamatergic terminals (Fig. 4) and there is a synergistic effect between A2A and mGlu5 receptors in the facilitation of glutamate release (Rodrigues et al., 2005). Furthermore, dopamine D2 receptors also control glutamate release in the striatum (Bamford et al., 2004) and preliminary experiments also indicate a tight inter-dependence between presynaptic A2A and D2 receptors (Tscherter et al., 2006). Likewise, other receptor subtypes located presynaptically in glutamatergic terminals, such as cannabinoid CB1 receptors (Kofalvi et al., 2005), may control striatal function by modulating neurotransmitter release and may do so by interacting with adenosine A2A receptors (Carriba et al., 2007). Particularly surprising is the observation that there are functional antagonistic interactions between A1 and A2A (Fig. 4) receptors that modulate glutamate release in the striatum (Ciruela et al., 2006a; 2006b; Quarta et al., 2004). The use of synaptosomes allowed us to conclude a robust facilitation of glutamate release by activation of presynaptic A2A receptors (Rodrigues et al., 2005) that co-exists with an ability of presynaptic A1 receptors to inhibit striatal glutamate release (Ciruela et al., 2006a; 2006b). It was found, using the in vivo microdialysis technique in freely moving rats, that intrastriatal perfusion with either an A2A receptor agonist or an A1 receptor antagonist induces glutamate outflow (Quarta et al., 2004).
Two questions arise: are A1 and A2A receptors co-localized in the same glutamatergic terminals? If that is the case, what is the functional significance of a co-localization of two receptors operated by the same ligand that have opposite effects? Electron microscopy studies have demonstrated the co-localization of A1 and A2A receptors in the same striatal glutamatergic terminal (Ciruela et al., 2006). Furthermore, studies with isolated striatal nerve terminal preparations indicated that most striatal glutamatergic terminals contain both A1 and A2A receptors (Ciruela et al., 2006a). Experiments in co-transfected cells using BRET (Bioluminescence Resonance Energy Transfer) techniques demonstrated the existence A1-A2A receptor heteromers. Furthermore, radioligand-binding experiments showed the existence of a strong intramembrane A1-A2A receptor interaction, by which stimulation of A2A receptors decreases the affinity of A1 receptors for their agonists (Ciruela et al., 2006a). The intramembrane A1-A2A receptor interaction was used as a biochemical fingerprint of the A1-A2A receptor heteromer, which allowed demonstrating the existence of A1-A2A receptor heteromers in the striatum (Ciruela et al., 2006a). Finally, functional experiments in striatal glutamatergic terminals demonstrated that the A1-A2A receptor heteromer provides a “switch mechanism” by which low and high concentrations of adenosine produce opposite effects on glutamate release (Ciruela et al., 2006a; 2006b).
However, in electrophysiological experiments, in most conditions of patch clamp or extracellular field recordings, activation of A2A receptors does not modify striatal evoked EPSCs or EPSPs which are fully dependent on AMPA receptors (Lovinger and Choi, 1995; Flagmeyer et al., 1997; D'Alcantara et al., 2001; Gerevich et al., 2002; Schiffmann et al., 2003). This lack of A2A receptor effect was demonstrated not only by using pharmacological tools in wild type animals (Lovinger and Choi, 1995; Flagmeyer et al., 1997; D'Alcantara et al., 2001; Gerevich et al., 2002) but also by the observation that there was no difference between wild-type and A2A knockout mice (D'Alcantara et al., 2001; Schiffmann et al., 2003). In contrast with these negative results, it was recently shown that presynaptic A2A receptors are able to modulate excitatory synaptic cortico-striatal transmission but only in the presence of 4-aminopyridine (Tebano et al., 2004) or by controlling miniature events upon activation of D2 receptors (Tscherter et al., 2006), demonstrating that in certain conditions of presynaptic activity, A2A receptor activation may exhibit a facilitatory effect.
This brief description of the known effects mediated by adenosine A2A receptors in the striatum makes it evident that this receptor is engaged in tight interactions with other metabotropic receptors to fine-tune glutamatergic transmission at cortico-limbic-striatal and thalamo-striatal synapses. Interactions between postsynaptic A2A, D2 and mGlu5 receptors (and also CB1 receptors) are well-documented in the control of medium size efferent neuron responsiveness. However, presynaptic interactions also appear to control the release of glutamate, which actually triggers the functioning of striatal circuits. For most of these interactions between metabotropic receptors and A2A receptors there is strong evidence for the formation of heterodimers (Ferré et al., 2005 for review). However, from the functional point of view, the most relevant question is to understand how presynaptic and postsynaptic heteromers containing A2A receptors modulate glutamatergic neurotransmission in the striatal spines of enkephalinergic neurons.
At the presynaptic side, low concentrations of adenosine, probably obtained during weak cortico-limbic-thalamic input (Fig. 4A), bind preferentially to A1 receptors, which decreases the probabilility of glutamate release. On the other hand, high concentrations of synaptic adenosine, probably obtained by strong cortico-limbic-thalamic input (Fig. 4B) with high co-release of glutamate and ATP, will also induce occupation of A2A receptors, counteracting the effects of A1 receptor stimulation and increasing the probability of neurotransmitter release. This shift of function of adenosine from inhibition into facilitation is probably aided by mGlu5 receptors. Thus, the frequency-dependent increase in extracellular levels of adenosine as well as glutamate will increase activation of mGlu5 receptors in parallel with increased activation of A2A receptors and these two receptor systems synergize to facilitate glutamate release (Rodrigues et al., 2005). Finally, at the presynaptic site, it remains to be seen whether increased firing frequencies in the cortico-limbic-thalamic-striatal projection bring presynaptic A2A receptors into play and if this is able to overcome the tonic D2 receptor inhibition of glutamate release (Bamford et al., 2004; Tscherter et al., 2006).
At the postsynaptic side, under weak cortico-limbic-thalamic input (Fig. 4A) there is a preferential D2 receptor-mediated modulation of the A2A-D2-mGlu5 heteromeric receptor complex, which is associated with low glutamatergic neurotransmission, low neuronal excitability and low gene expression. Under strong cortico-limbic-thalamic input (fig. 4B), there is a strong release of glutamate and formation of synaptic adenosine, which activates postsynaptic A2A receptors, probably concentrated in the perisynaptic ring adjacent to the PSD. Furthermore, glutamate is able to overflow from the synaptic cleft and stimulate the mGlu5 receptors of the perisynaptic ring. Thus, under these conditions there is strong activation of the A2A and mGlu5 receptors in the A2A-D2-mGlu5 receptor heteromer, which allows gene expression, protein synthesis and plastic changes. In conclusion, the A2A receptor, which is localized pre- and postsnaptically forming part of different heteromeric receptor complexes, plays a key role in the functional changes of glutamatergic synapses of the enkephalinergic striatal neurons during conditions of strong cortico-limbic-thalamic input. In agreement with this hypothesis, we found that administration of a selective A2A receptor antagonist or caffeine counteracted both MAPK activation and AMPA receptor phosphorylation in GABAergic enkephalinergic neurons following cortical electrical stimulation (Quiroz et al., 2006).
A key question that remains to be clarified is the relative importance of presynaptic and postsynaptic A2A receptors in the control of glutamatergic synapses between cortico- and thalamo-striatal projections and medium size efferent neurons. Electron microscopy, immunohistochemical studies, and in situ hybridization studies suggest that striatal A2A receptors are mainly localized postsynaptically (Schiffmann et al., 1993; Rosin et al., 2003). In fact, the available evidence makes it clear that postsynaptic A2A receptors have a strong impact on several key biochemical features that are accepted as determinants of striatal efferent neuron function, such as cAMP levels, DARPP-32 phosphorylation and regulation of L-type voltage-dependent Ca2+ channel and AMPA receptor activities. Less evidence is available on the contribution of presynaptic A2A receptors. From a teleological point of view, the strong impact of A2A receptors on glutamate release would be expected to play a crucial role. In fact, it is the release of glutamate that triggers postsynaptic activity and minor changes in the probability of release of glutamate are expected to have a significant impact on the function of the striatal efferent neurons. For instance, this should affect the down- to up-state transitions of striatal medium size spiny neurons with a higher proportion of time in the up-state. However, there are still no electrophysiological studies evaluating the relative importance of presynaptic and postsynaptic A2A receptors in glutamatergic synapses of the dorsal striatum. It should be pointed out that in order for presynaptic A2A receptors to play a relevant role in the control of the indirect pathway, A2A receptors would need to be selectively localized in the glutamatergic nerve terminals that impinge on the GABAergic enkephalinergic neurons of the indirect pathway. This possibility awaits experimental support, although it is known that there is a selective tagging of cortico-striatal synapses to different medium size spiny neuron populations (Parthasarathy and Gaybriel, 1997; Lei et al., 2004) and previous studies in different brain areas have documented presynaptic modulatory systems that are confined to particular synapses contacting particular targets within the same neuron (Khakh et al., 2003).
Finally, it is essential to keep in mind the fact that A2A receptors may come into play only under particular physiological conditions. There is evidence indicating that the activation of A2A receptors might be achieved by a particular pool of adenosine, i.e., the adenosine derived from the extracellular catabolism of released ATP (for reviews see Cunha, 2001; Ferré et al., 2005). Thus, one could anticipate that A2A receptors might play a prominent role in the control of glutamatergic striatal synapses under conditions of enhanced release of ATP, which occurs upon increased firing rates of glutamatergic neurons (Cunha et al., 1996; Wieraszko et al., 1989). This leads to long-term modifications of synaptic strength such as LTP and LTD at the cortico-striatal or cortico-accumbal synapses. In accordance with the anticipated particular role of A2A receptors at higher firing frequencies, LTP of the AMPA receptor-mediated EPSP could be elicited in the nucleus accumbens of both wild-type and A2A receptor-deficient mice but it was quantitatively reduced in the mutant animals as it was in slices from wild-type mice treated with an A2A receptor antagonist (D'Alcantara et al., 2001; Schiffmann et al., 2003). The involvement of PKA was supported by a reduced level of LTP in wild-type slices treated with an inhibitor of this enzyme. As previously discussed, it is unclear whether this A2A receptor-mediated regulation of cortico-striatal synaptic plasticity occurs pre- or postsynaptically or both. Whatever the mechanism, the regulation of LTP induction by A2A receptors should have important consequences for motor learning and reward processes as suggested during chronic cocaine adaptation (Baldo et al., 1999) or behavioral sensitization to repeated amphetamine (Bastia et al., 2005) or L-DOPA treatment (Freduzzi et al., 2002). It should also be pointed out that this global (pre- and post-synaptic) facilitation of glutamatergic transmission by adenosine A2A receptors may not only contribute to implement changes in synaptic efficiency but may also be responsible for excitotoxic effects once over-stimulated, in a manner analogous to the double-edge sword role played by NMDA receptors. This is re-enforced by the observation that the pharmacological or genetic blockade of A2A receptors indeed confers robust neuroprotection against different chronic noxious stimuli (Popoli et al., 2002; Blum et al., 2003; Cunha, 2005).
In summary, the body of available evidence supports the conclusion that A2A receptors play an important fine-tuning role in striatal function by boosting the efficiency of information flow in glutamatergic synapses of the indirect pathway. This is mostly achieved by the organization of A2A receptors in different heteromers allowing control of other key modulators of glutamatergic synapses, such as dopamine (acting through D2 receptors), glutamate (acting through group I metabotropic receptors) and endocannabinoids (acting through CB1 receptors) as well as by triggering the cAMP-PKA signaling cascade. However, further neurochemical and electrophysiological studies are clearly required to better understand the relative participation of presynaptic and postsynaptic A2A receptors in modulation of striatal pathways.
These works were supported by Fondation Médicale Reine Elisabeth (FMRE 2005−2007), Fonds National de la Recherche Scientifique (FNRS grants 3.4507.02 F and 3.4509.06F), Van Buuren Foundation, Action de Recherche Concertée Communauté Française Wallonie-Bruxelles (ARC N° 02/07−290) to SNS, the National Institute on Drug Abuse Intramural Research Funds to SF and Fundação para a Ciência e Tecnologia (grant POCI/SAU-FCF/59215/2004) to RAC. The authors thank N.Rebola and K. Azdad for helpful discussion and contribution to these studies.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.