Overview of dinucleotides

Dinucleotides or dinucleoside polyphosphates, commonly abbreviated as NpnNs, comprise a group of compounds formed by two nucleoside moieties linked by their ribose 5′-ends to a variable number of phosphates, which can change from two to seven. The most abundant and widely studied of these dinucleotides contain either one or two adenine bases in their structure. They are naturally occurring substances ubiquitously present in the cytoplasm of prokaryotic and eukaryotic cells where they are synthesized by some aminoacyl-tRNA synthetases and other enzymes [60]. The cytosolic levels of the dinucleoside polyphosphates are modified under a variety of physiological and pathological conditions and these compounds have emerged as putative intracellular signaling implicated in the regulation of essential cellular process [40, 45]. Some of the intracellular functions of adenine dinucleotides may be related to environmental stress [3, 4, 70–72]. Apart from their well-known intracellular roles, diadenosine polyphosphates also exert important extracellular actions, which are of special relevance in the vascular system. These compounds are stored in the secretory dense granules of platelets and are released into the circulation when platelets become activated. Circulating diadenosine polyphosphates have deep effects on platelet aggregation and are involved in the control of vascular tone and myocardial function [21, 64, 69]. Our group has been largely working on the presence and functional significance of diadenosine polyphosphates in the nervous system. We have demonstrated that these compounds fulfil with the requirements to be considered as signaling molecules in the CNS [46]. Diadenosine polyphosphates are co-stored with ATP and classical neurotransmitters in the secretory granules of chromaffin cells, the cholinergic synaptic vesicles from Torpedo marmorata electric organ, and in rat brain synaptic terminals [50, 57, 62]. They are co-released to the extracellular media in a calcium-dependent manner and can interact extracellularly with membrane receptors. Finally, their effects are terminated by the action of ecto-ApnA-hydrolases, rendering adenosine as final product. This nucleoside is re-uptaked into the cells by means of the nucleoside transporter [46].

Due to their close structural relationship, some of the extracellular actions of diadenosine polyphosphates can be attributed to the interaction of these compounds with ATP receptors. The ATP-sensitive receptors are divided into two families, ionotropic P2X (multimeric ligand-gated cationic channels) and metabotropic P2Y receptors (G-protein-coupled 7TM receptors) [61]. Both P2X and P2Y receptors are present in the nervous system and their expression change during development [9, 75]. Seven P2X subunits have been cloned to date (P2X1, P2X2, P2X3, P2X4, P2X5, P2X6, and P2X7), which can assemble in different combinations to form homomeric or heteromeric P2X receptors. Similarly to the great variety on ionotropic receptors, eight metabotropic P2Y receptors have been currently characterized (P2Y1, P2Y2, P2Y4, P2Y6, P2Y11, P2Y12, P2Y13, and P2Y14). Among the great diversity of nucleotidic receptors, P2X1, P2X3, and also P2Y1 receptors have been described as possible targets for dinucleotides. This opens a broad spectrum of actions for dinucleotides in the CNS considering the wide distribution of multiple types of P2X and P2Y receptors at this level. The question regarding the role of these dinucleotides in the CNS that needs to be addressed is: are there specific receptors for diadenosine polyphosphates different from ATP receptors? This issue presented some difficulties to overcome, first due to the poor available tools, specific agonists and antagonists, which could allow to clearly distinguish the actions mediated through the different types of nucleotidic receptors.

Specific targets for dinucleotides at the presynaptic level in the nervous system

In spite of these initial problems, the presence of a specific receptor for dinucleotides of ionotropic characteristic was firstly described in the pioneering work of Pintor and Miras-Portugal (in 1995), who found specific responses for dinucleotides in midbrain synaptic terminals. The increase in the intrasynaptosomal calcium concentration mediated by diadenosine polyphosphates was unaffected after cross-desensitization with ATP, when ApnA and ATP were consecutively applied, though each agonist was able to desensitize itself. Additional proof for the presence of independent presynaptic receptors for ATP and ApnA was provided by the use of the classical P2 antagonists PPADS and suramin. These compounds did not affect the Ca2+-increase evoked by diadenosine polyphosphates but blocked Ca2+ entry induced by ATP and its synthetic analogues [54]. The synthesis of a new series of dinucleotides denoted diinosine polyphosphates, which resulted by enzymatic deamination of the corresponding diadenosine polyphosphate, supposed a novel contribution in the characterization of dinucleotide receptors. These substances did not show any agonistic effect on the rat synaptosomal receptors but effectively antagonized the responses induced by ApnA. Diinosine pentaphosphate (Ip5I) was the most potent among these substances, being able to block ApnA responses at concentrations in the nanomolar range (IC50=4 nM). In contrast, micromolar concentrations of the antagonist were required to abolish ATP-elicited responses in the same preparation [52]. In addition, a series of experiments using structural analogs of the diadenosine polyphosphates demonstrated that the ApnA receptor is also sensitive to guanosine polyphosphates (GpnG) as agonists [51]. This is noteworthy, as the different P2 receptors described so far (either P2X or P2Y receptors) only show sensitivity to adenine or uridine nucleotides (ATP, ADP, UTP, and UDP) but none of them respond to guanine nucleotides. Taken together, these differences between the actions of ApnA compounds and ATP and the inability of methylxanthines to block dinucleotide actions (excluding the participation of adenosine receptors) strongly suggested that there are specific receptors for diadenosine polyphosphates in the rat midbrain synaptic terminals.

In the following years, there was some discrepancy over the nomenclature to be adopted for this new type of receptor and several terms have been used in the literature. According to its pharmacological properties and in the absence of any structural data, we propose it is not appropriate to define this receptor as a mere variant of the ATP receptors. We consequently prefer to use the term dinucleotide receptor over any other cited in the literature [55].

The presence of an independent dinucleotide receptor at the presynaptic level has important implications in neural transmission. In fact, the ionotropic character of the dinucleotide receptor implies Ca2+ entry through the receptor, which depolarizes the terminal and activates Na+- and Ca2+-voltage-dependent channels [16, 54, 58, 59]. It seems reasonable to propose that dinucleotides acting through this mechanism of action may modulate the release of the neurotransmitters present in the synaptic terminal. We explored the interaction between dinucleotide receptor activation and the exocytotic release of other neurotransmitters. Using microfluorimetric Ca2+ measurement techniques combined with immunocytochemical studies, we showed the presence of the dinucleotide receptor in aminergic terminals isolated from basal ganglia [23]. A specific receptor for diadenosine polyphosphates is also present in cholinergic terminals from rat midbrain as well as in GABAergic or glutamatergic terminals obtained from the same source. It is interesting that the stimulation of the midbrain synaptosomes with Ap5A leads to the exocytotic release of the stored neurotransmitter: glutamate, GABA, or acetylcholine [17, 24, 27].

Several reviews on extracellular and neurotransmitter roles of dinucleotides acting through presynaptic ionotropic receptors, their specific dinucleotide receptor, and/or ionotropic nucleotide P2X receptors have been previously published [46, 55]. We refer the reader to these works to better understand some aspects, which are only briefly mentioned here. We will specifically address this review on the new interactions of the presynaptic ionotropic dinucleotide receptor with the nucleotide and other neurotransmitter signaling systems; we will also describe the new findings on the actions of dinucleotides in the cerebellum. In addition, we will pay special attention on dinucleotide signaling in glial cells where a new specific metabotropic receptor for these substances has been described.

Modulation of the presynaptic dinucleotide receptor by other neurotransmitter systems

Previous work in rat midbrain synaptic terminals has demonstrated that the dinucleotide receptor was submitted to acute regulation at presynaptic level. In this sense, the dinucleotide receptor activity is modulated by effectors of protein kinases and phosphatases [53]. While activation of protein kinases A and C led to inhibition of Ap5A-induced responses, protein kinase inhibitors as well as protein phosphatase modulators clearly potentiated these responses. To continue with these regulatory studies, modulation of the dinucleotide receptor by presynaptic purinergic receptors of adenosine and ATP was investigated. ATP, acting on a not fully characterized P2 receptor, induces a decrease in the maximal response elicited by Ap5A in these terminals. Adenosine, on its hand, shows a dramatic effect on the affinity of the dinucleotide receptor for its ligand. This nucleoside, acting presumably on A1 receptors, allows the dinucleotide receptor to attain a high-affinity state and, thus, to be stimulated by lower concentrations of diadenosine polyphosphates, even in the picomolar/low nanomolar range [15]. In the absence of A1 receptor activation, the dinucleotide receptor in rat midbrain synaptosomes is only sensitive to ApnA concentrations in the micromolar range [51, 54]. Thus, the increased affinity induced by adenosine could confer the dinucleotide receptor the capacity to respond to more physiologically relevant ApnA concentrations. The ability of adenosine to induce a high-affinity state of the dinucleotide receptor has also been described in rat hippocampal nerve terminals [14].

Changes in dinucleotide receptor affinity have also been observed after the activation of presynaptic GABAB receptors [25]. The sigmoidal concentration response for Ap5A (EC50=44 μM) changes into a biphasic curve in the presence of the GABAB agonist baclofen. Thus, when GABAB receptors are activated, the curve shows a high-affinity component in the picomolar range (EC50=77 pM) and a low-affinity component in the micromolar range (EC50=17 μM). The effect of baclofen was blocked by the selective antagonist saclofen and by preincubation with pertussis toxin, clearly demonstrating the implication of GABAB receptors in these effects. Both A1 and GABAB are Gi/Go-coupled receptors that decrease the activity of adenylate cyclase and, as a consequence, reduce cAMP levels. It seems to be reasonable that a decreased cAMP concentration reduces PKA activity leading to a lower degree of phosphorylation of the dinucleotide receptor that could modify its affinity to Ap5A. Experiments performed with modulators of the adenylate cyclase and PKA activities have demonstrated the involvement of this signaling pathway in the potentiatory effect of baclofen [25]. Thus, it could be a general mechanism of regulation of the dinucleotide receptor activity (Fig. 1).

Fig. 1
figure 1

Modulation of the presynaptic dinucleotide receptor. Activation of the dinucleotide receptor induces calcium entry into the terminals with the subsequent activation of CaMKII, finally leading to the inhibition of the calcium signaling through the dinucleotide receptor. A similar inhibition of the functionality of ApnA receptors can be achieved after activation of calcium-permeable ionotropic receptors that could be present at the same terminal, as is the case of the acetylcholine nicotinic receptors [18]. On the other hand, stimulation of second messenger systems that are negatively coupled to adenylate cyclase leads to a reduction of PKA activity and this allows the dinucleotide receptor to reach a higher-affinity state, being sensitive to lower diadenosine polyphosphate concentrations, even in the nanomolar range [15, 25]

Full size image

The functionality of the dinucleotide receptor cannot only be modulated by stimulation of metabotropic receptors present at the same terminal, but a fine-tuning interaction between the dinucleotide receptor and epibatidine-responding nicotinic receptors has also been described in rat midbrain cholinergic terminals [18]. Dinucleotide receptors and epibatidine-sensitive nicotinic receptors are coexpressed in 19% of the total of rat midbrain cholinergic terminals, as determined by a combination of immunological and microfluorimetric techniques. Activation of each independent receptor induces intrasynaptosomal [Ca2+]i increases and acetylcholine release in a dose-dependent way. However, stimulation of the cholinergic terminals with both agonists induces a response that is significantly reduced with respect to the expected signal (the summatory of the individual responses for each agonist). The mutual negative control of these two ionotropic presynaptic receptors is perhaps the physiological way to turn off the synaptic excitability and prevent an excessive release of neurotransmitters (Fig. 1). This inhibitory cross-talk is reverted in the presence of specific inhibitors of calcium/calmodulin-dependent protein kinase II (CaMKII), such as KN-62 and KN-93. These results demonstrate the participation of this kinase in the mechanism of reciprocal regulation between the presynaptic dinucleotide and nicotinic receptors [18].

Dinucleotide actions in cerebellum

Although, as mentioned before, the pharmacological and molecular characterization of native nucleotide receptors, both ionotropic P2X and metabotropic P2Y receptors, have been greatly delayed by the lack of specific agonists and antagonists for the different receptor subtypes, several binding studies carried out with the available tools, such as a nonhydrolyzable ATP analog, [3H]αβ-meATP, and the dinucleotide [3H]Ap4A, revealed that the cerebellum, particularly the cerebellar cortex, was one of the structures exhibiting highest levels of specific binding in the brain [63]. Based on these findings, we decided to study the functional responses to nucleotides in this area.

We developed enriched cultures in the two major neuronal types of cerebellum, Purkinje neurons, at the beginning, and granule neurons, more recently. Astrocyte cultures have been carried out in parallel to separately study the nucleotide receptors present in neurons and glia, the two components of the brain multifunctional unit. We found that all of them expressed multiple functional nucleotide receptors. Purkinje and granule neurons co-expressed ionotropic P2X and metabotropic P2Y receptors, whereas astrocytes only expressed P2Y receptors [29, 34, 44]. This affirmation made for cerebellar astrocytes several years ago can now be ruled out. The presence of functional P2X7 receptors has been recently detected, which are currently being investigated [11]. However, neurons and glial cells differed in their sensitivity to dinucleotides. In our experimental conditions, Purkinje neurons were completely insensitive to dinucleotide stimulations (Ap5A and Ap4A stimulations) and granule neurons displayed some intracellular responses to the dinucleotide Ap5A; but, there can be no doubt that astrocytes have resulted to be one of the most active “sensors” of extracellular dinucleotide concentrations.

Cerebellar granule neurons

Beginning with granule neurons, among the great diversity of nucleotidic receptors functionally expressed in these cells, there are possible targets for dinucleotides, such as P2Y1, P2X1, and P2X3 [29]. Among metabotropic receptors, the major population of nucleotidic responding cells presented a pharmacological profile typical of P2Y1 receptors, being mainly found at the cell bodies. Concerning ionotropic receptors, although P2X2 and P2X3 were also present at the somas, the preferential distribution pattern of P2X subunits in granule cells were along the cell fibers. Significant responses at this level were found for αβ-meATP and BzATP, which could activate P2X1/3 and P2X7 receptors, respectively (unpublished observations). Taking into account that along the cell fibers of granule neurons in culture it is not possible to differentiate between dendritic and axonal compartments, the responses found for nucleotidic agonists could be both of presynaptic and/or postsynaptic nature. However, the characteristic punctuated immunostaining pattern observed for P2X3 and P2X7 receptors, which exhibited a high degree of colocalization with the synaptic marker synaptophysin, suggested a presynaptic function for these receptors in granule neurons. This was further confirmed in synaptosomal preparations of rat cerebellum. In fact, the only ionotropic nucleotidic receptors significantly expressed in glutamatergic cerebellar synaptic terminals were P2X3 and P2X7 receptors. The immunostaining pattern was corroborated by calcium recordings in single cerebellar synaptosomes, in which the responses found for the specific agonists, αβ-meATP and BzATP, fitted well with the distinctive pharmacology of P2X3 and P2X7 receptors, respectively [30].

Assuming the high percentage of P2X3 receptor expression in cerebellar synaptosomes, experiments were carried out to challenge for dinucleotide responses. In this model, a significant population of synaptosomes was able to respond to the dinucleotide Ap5A, either in the presence or absence of previous ATP responses. The fact that around 44% of these responses remained in cross-desensitization studies with ATP gives evidence in favor of the presence of a specific dinucleotide receptor at cerebellar synaptic terminals (unpublished results). Considering the high proportion of cerebellar synaptosomes exhibiting P2X3 immunostaining and the lack of P2X1 receptor expression with the antibodies commercially available, P2X3 receptors are good candidates for assuming the remaining 56% Ap5A responses [30].

Coming back to the granule neuron model, recent studies have shown that, beyond the great diversity and compartmentalization of nucleotidic and/or dinucleotidic receptors between cell bodies and fibers, the calcium signals primarily triggered for all these nucleotidic agonists become confluent in a specific calcium-derived signal, that is CaMKII. The phosphorylation and subsequent activation of CaMKII by different nucleotidic agonists spatially correlates with the distribution of the nucleotide receptors that they activate. Although the highest levels of phosphorylation were found for 2MeSADP and BzATP, P2Y1 and P2X7 agonists, respectively, a significant response was observed for the dinucleotide Ap5A, and the P2X1,3 agonist αβ-meATP, reaching values of about 70–80%, increased for both agonists (see Fig. 2) [42]. It is interesting to mention the similarity found for αβ-meATP and Ap5A in the phosphorylation pattern of CaMKII at the granule cell fibers, which appeared in some but not all fibers and in a dotted appearance. This pattern fitted well with the immunostaining distribution of P2X3 subunits along granule cell fibers [29]. In addition, the effect of both agonists on CaMKII was partially sensitive to Ip5I inhibition at 100-nM concentration, with the inhibition being significantly higher, although not complete, in the case of Ap5A. These results are consistent with αβ-meATP and Ap5A acting through P2X3 receptors to activate CaMKII and also imply that another receptor, possibly the previously cited specific dinucleotide receptor, could be accounted for the effect seen with Ap5A, as demonstrated at the presynaptic level in synaptosomes.

Fig. 2
figure 2

CaMKII activation by Ap5A and αβ-meATP in cerebellar granule neurons. CaMKII phosphorylation at Thr286/287 was detected by immunocytochemistry and quantified as described in [42], after the treatment of granule cells with Ap5A and αβ-meATP at 100 μM concentrations for 5 min. Although the phosphorylation pattern for both agonists was similarly distributed along the cell bodies and fibers of granule cells, Ip5I, used at concentrations (100 nM) that inhibited more specifically the dinucleotide receptor over the P2X3 receptors, exhibited a higher inhibitory effect on Ap5A responses in comparison to that observed for αβ-meATP

Full size image

The above data are consistent with the dinucleotide Ap5A acting mainly through P2X3 receptors and mediating a broad spectrum of actions in granule cell neurons. The putative modulatory functions will depend on the specific cellular compartment or microdomain in which Ap5A could act. In addition, the possible actions of Ap5A acting through the specific dinucleotide receptor described in cerebellar synaptic terminals cannot be excluded, which enriched even more the versatility of this dinucleotide as a wide modulator of cerebellar neurotransmission.

Considering the convergent localization of Ap5A responses in cerebellar synaptic terminals of glutamatergic nature, it is tempting to speculate a possible interaction between glutamate and nucleotides, at the level of glutamate release. Taking into account that Ap5A is coupled to the activation of CaMKII, it is well documented that this enzyme actively participates in synaptic vesicle mobilization for neurotransmitter release. In fact, between the targets of CaMKII at the presynaptic level are synaptic vesicle proteins, such as synapsin I, which, upon phosphorylation by CaMKII, stops interaction with actin filaments, releasing the synaptic vesicle and, thus, facilitating the vesicle movement towards the active synaptic sites. In this context, preliminary results have shown CaMKII-dependent phosphorylation of synapsin I by several nucleotides, 2MeSADP, αβ-meATP, and BzATP, in granule neurons. These nucleotidic agonists are also involved in the modulation of glutamate release in this neuronal model (León et al., manuscript in preparation). Whether Ap5A acting through a specific dinucleotide receptor or through nucleotidic receptors, P2Y1 and P2X3, can be coupled to glutamate release modulation will be the aim of future studies.

In addition to these promising perspectives for Ap5A, it has to be considered that its main action localizes at the postsynaptic level and partially acting through P2X3 receptors. At this level, CaMKII is only the first step in the nucleotide-derived signaling following calcium influx. This enzyme is a link for different intracellular signaling pathways, such as MAPK kinases, NFκB, CREB, etc., and leads its actions even at the transcriptional level. Due to its effects on CaMKII activation, the nucleotides effectively become potential modulators of a great range of actions, as this enzyme has been involved as a critical mediator of synaptic plasticity, survival, and neuritogenesis [6, 20]. A recent study has demonstrated a role for CaMKII as a regulator of the increased expression of plasma membrane P2X3 receptors, which leads to potentiation of ATP-mediated responses in dorsal root ganglion neurons [77]. This resembles the effect of CaMKII on the increase in synaptic AMPA receptor transmission, by facilitating the delivery of additional AMPA receptors to the synapse, an early event occurring in long-term potentiation [65]. This allows us to hypothesize that, similarly to the well-known actions of AMPA receptors as modulators of synapse formation and spine morphogenesis, nucleotide receptors could be playing a relevant role in these processes. In this sense, CaMKII is directly involved in promoting dendritic growth and maintenance on granule neurons, and this effect has been attributed to its regulatory effect on the transcription factor NeuroD involved in axonal and dendritic morphogenesis [22].

Cerebellar astrocytes

As mentioned previously, we select purified astrocyte cultures as a model of glial cells. Astrocyte cultures have been widely used for the characterization of receptors present in astroglia, such as glutamate, GABA, histamine, ATP, etc. [73]. Over the past decade, it has been concluded that astrocytes are the “new stars” in the nervous system. Neuron-to-astrocyte signaling in the brain represents a multifunctional unit, modulating synaptic activity and exerting a dynamic control of brain microcirculation [19]. Cerebellar astrocytes could perhaps be good candidates to study nucleotide signaling. In fact, in our first studies, we detected not only the presence of functional metabotropic ATP receptors in all tested cells but also the existence of an interaction with dinucleotides in a subpopulation of astrocytes, which accounted for 65% of cells [32].

Dinucleotides potentiate the P2Y-mediated calcium responses in single astrocytes

Preincubation of astrocytes with 100 nM Ap5A for 3 min and co-stimulation with previous ineffective ATP concentrations (0.1–1 μM, depending on the cell) induced remarkable [Ca2+]i increases, which ranged from 40–60% of the maximal ATP response obtained at 100 μM ATP. The potentiation was independent of extracellular calcium and, once it occurred, persisted for a prolonged period (several hours), during which the cells continued to be sensitive to low ATP concentrations that were previously ineffective (Fig. 3, the upper panel shows an example of the described effect). The potentiation was exclusive for Ap5A and was not reproduced by any other dinucleotide. These findings prompted us to characterize the receptors mediating ATP calcium responses and the “target” and mechanism triggered by Ap5A. Using different approaches, we have found that all cerebellar astrocytes co-express several subtypes of metabotropic nucleotide receptors, ADP-sensitive receptors, the P2Y1 and/or the P2Y13-like receptors, which are heterogeneously distributed among astrocyte population, and the ATP/UTP-sensitive receptors, the P2Y2/P2Y4 receptors, which are homogeneous distributed in all the cells [8, 34].

Fig. 3
figure 3

Cross-talk among Ap5A, EGF, and ATP nucleotide receptors in cerebellar astrocytes. The upper panels show the potentiation of ATP calcium responses by Ap5A and EGF in individual cerebellar astrocytes. Cells were stimulated with different ATP concentrations: the maximal effective concentrations (100 μM, solid bars) and an ineffective concentration (100 nM, open bars) for 30 s. Where indicated, cells were preincubated with 100 nM Ap5A or 50 ng/ml EGF for 3 min before co-stimulation with ineffective ATP challenges. The basal [Ca2+]i levels were around 80 nM and the magnitude of the potentiated ATP responses amounted to 304±18 nM. These cells are representative of the astrocyte subpopulation in which the potentiation was observed. The lower panel shows a model representing the intracellular signaling triggered by Ap5A in cerebellar astrocytes, adapted from [12]. The specific metabotropic diadenosine pentaphosphate receptor can adopt several active conformations, which can be differently stabilized in a ligand concentration manner to discriminate among various intracellular signaling pathways. At nanomolar concentration, Ap5A engaged to ERK signaling via src and ras proteins, whereas at micromolar concentrations, it activated the typical intracellular calcium mobilization dependent on PLC activation. The inset shows the phosphorylated ERK1/2 forms detected by immunoblotting from a representative experiment, in which cells were stimulated with 100 nM Ap5A or 50 ng/ml EGF for 3 min, lysed, and with p-ERK1/2 forms detected as described in [35]

Full size image

Searching into the Ap5A “target”, at first, the fact that the potentiation was triggered by nanomolar concentrations of the dinucleotide induced us to think that Ap5A could be interacting with A1 adenosine receptors, which exhibited high affinity for the nucleoside. At that time, same potentiatory actions of adenosine had been reported. However, neither nanomolar adenosine concentrations nor specific A1 receptor agonists or antagonists had any effect on ATP calcium responses. ATP calcium responses were surprisingly potentiated by simple co-stimulation with micromolar adenosine concentrations or other signals coupled to Gs proteins. Although, in contrast to that found for Ap5A, this potentiation was totally reversible and was not exclusive for the ATP receptors, P2Y2/P2Y4 receptors, as other G-protein-coupled receptors mediating PLC stimulation were also potentiated, such as ADP and bradykinin receptors, which are also present in cerebellar astrocytes [13, 33].

Another candidate for Ap5A target could be the ADP-sensitive P2Y1 receptors, at which diadenosine polyphosphates can act as agonists or even as antagonists, as is the case of the P2Y1 receptor present in rat brain endothelial cells [56, 74]. However, the possible Ap5A interaction with this receptor could also be discarded. As mentioned before, all astrocytes were sensitive to ADP stimulations, whereas only around 65% were sensitive to Ap5A, indicating that they are acting through different receptors. In addition, specific agonists of P2Y1 receptor, such as ADP and 2MeSATP, were unable to reproduce the dinucleotide potentiatory effect. These results suggested the existence of a specific metabotropic receptor for Ap5A in this astrocyte subpopulation.

Metabotropic Ap5A signaling cascades: analogies and divergences with EGF

Trying to find out the intracellular signaling triggered after the stimulation of the “putative” Ap5A receptor in astrocytes, considering that the potentiation was maintained for hours and the numerous examples of cross-talk between GPCR and tyrosine receptor kinases resulting in ERK activation (as examples, see [43, 78]), we examined ERKs as possible mediators of this event. The stimulation of cells with 100 nM Ap5A for 3 min induced ERK activation, in a similar extent to that produced by EGF under the same experimental conditions (Fig. 3, insert in lower panel). ERK activation induced by Ap5A was concentration-dependent (EC50 value of 9.8±2.4 nM) but exhibited a bell-shaped curve, reaching the maximal effect at 100 nM Ap5A, just the concentration at which the potentiaton was displayed. At concentrations higher than 100 nM, ERK activation was considerably reduced. Additional studies using specific inhibitors for several steps of MAPK cascade activation showed that Ap5A engaged to MAPK cascade by a mechanism that was insensitive to pertussis toxin, independent of EGFR transactivation, and required the involvement of src and ras proteins [12, 35]. This signaling is quite distinct from those reported for some members of the P2Y nucleotide receptor family [26, 41, 66, 67]. The “putative” Ap5A receptor also exhibited another peculiar characteristic, when it was stimulated with micromolar concentrations of the dinucleotide also coupled to PLC, and evoked intracellular calcium mobilization (Fig. 3, lower panel). This switch was not observed for any other nucleotide receptor. Although P2Y1 and P2Y2/P2Y4 receptor stimulation also induced ERK activation in cerebellar astrocytes, it ran parallel to PLC activation (unpublished results).

Ap5A and EGF mediate similar potentiation on ATP calcium responses

Continuing with the potentiation phenomenon, whether Ap5A induced ERK activation in a similar extent to those induced by EGF, it could be plausible that EGF or other growth factors reproduced the potentiation. As expected, it was exactly reproduced by EGF and NGF [12]. When cells were preincubated with 50 ng/ml EGF or NGF and co-stimulated with previous ineffective ATP concentrations, intracellular calcium responses ranging from 55–60% of the maximal ATP response (obtained at 100 μM ATP) were elicited. The potentiation was also permanent, as is the case of Ap5A (Fig. 3, upper panel). In all the cases, an acute (3 min) stimulation of the “priming” receptor (dinucleotide or growth factor receptor) was required before co-stimulation with ATP. The potentiation of ATP responses by Ap5A and EGF was completely abolished by the MAP kinase (MEK) inhibitor U-0126, proving that ERK activation is a required step for the potentiation. It was also sensitive to src-like kinases inhibitors, such as herbimycin, p21ras farnesyltransferase inhibitor peptide, and some PKC inhibitors. Taken together, these data revealed that Ap5A triggered the potentiation of ATP calcium responses through an insensitive pertussis toxin G protein and required src protein-mediated ERK activation and the participation of atypical protein kinase C isoforms activated down-stream from ERK. The question that still remains open is: which are the substrate or the target of the activated PKC? We speculated that PKC could phosphorylate the RGS2 protein, a member of the regulators of G protein signaling (RGS). This phosphorylation decreased its capacity to reduce GTP-stimulated PLC activation, with the corresponding enhancement in inositol phosphate accumulation, as have been proposed by Cunningham et al. [10]. This could be feasible based on the fact that the potentiation required co-stimulation with the agonist. Independently of the intracellular mechanism that took place, these results clearly suggest that Ap5A at nanomolar concentrations could act as a growth factor in these glial cells. In this sense, dinucleotides have been also proposed as growth-promoting extracellular mediators [31]. Besides, the interaction with growth factors represents a novel regulation of nucleotide signaling that could be relevant in both physiological and pathological conditions. Although several interactions between growth factors, cytokines, and nucleotide signaling have been previously reported [36, 37, 41], all of them described long-term effect, which required long incubation periods, and are very far from the acute effect reported here.

Conclusions and perspectives

The aim of this review has been to present an overview of the dinucleotide actions in the nervous system and their numerous cross-talks resulting in the majority of the cases in a considerable alteration of calcium signaling. Considering that dinucleotides are more resistant to hydrolysis than nucleotides, these interactions would contribute to maintain the functionality of nucleotide receptors, when the extracellular ATP levels fall because of hydrolysis and/or diffusion. The interactions revised here are isolated examples, which would represent only a small part among the great diversity of signals and mechanisms that must be integrated in vivo. However, this fact does not diminish the possible physiological implications, which unfortunately remain to be elucidated. Nucleotide responses and functional receptors have been described in intact preparations, but there was a controversy concerning the receptors implied [39, 48]. As noticed previously, the nucleotide receptor characterization is even more complex than those for any neurotransmitter, first by the lack of specific agonist and antagonist and, also, by other factors [1, 38, 47]. Nucleotide and dinucleotide actions depend on the local environment at which are released the receptors coexisting in this particular “microdomain”, on the presence of ectoenzymes responsible for their hydrolysis and even also on the sources, as have been described for glutamatergic neurotransmission. Excellent works from Araque and colleagues have revealed that astrocyte Ca2+ signal can be spatially localized and compartmentalized in restricted regions of the cells—the subcellular microdomains—that constitute the elementary units of the Ca2+ elevations. Furthermore, astrocytes can also discriminate between synaptically released glutamate from different afferent axons [2, 49]. In this sense, the nucleotide sources are even more diverse than those of glutamate. Nucleotides or dinucleotides can be released from healthy cells in a calcium-dependent or calcium-independent manner from neurons, glial and endothelial cells, and also from damaged or dead cells, in all the cases reaching sufficient extracellular concentrations to activate both ionotropic and metabotropic receptors and, after being metabolized in the extracellular media, also give products that continue being “active”, such as ADP and, finally, adenosine, which can also act on specific receptors, modulating nucleotide responses at presynaptic level in neurons and also in glial cells [14, 33]. Thus, it is quite difficult to determine the set-point of nucleotide signaling and to determine the “basal” activity of any cellular responses to nucleotides. This complexity of purinergic system itself together with the multiple interactions that could be produced in a local and temporal localization set us into thinking that any cellular response, the different types of neurotransmission, secretion, migration, etc., results from the integration of the multiple extracellular signals.

To finish this review, we would like to highlight another aspect of nucleotide signaling, the existence of scarce data concerning the subcellular distribution of the receptors, their intracellular trafficking, etc. In this way, the interactions revised here would bring out some light about this. The interactions described with GABA, acethylcholine, glutamate, and EGF receptors suggest that they must be colocalized within the same membrane domains facilitating the functional interactions. Considering that several of them have been localized in lipid rafts, one would speculate that the nucleotide receptor subtypes co-interacting must be also there. The first candidates would be the P2Y2/P2Y4 receptor, which is potentiated by EGF and Ap5A, as well as the metabotropic Ap5A receptor present in cerebellar astrocytes. EGF receptor and src family kinases were ones of proteins primarily identified in the lipid rafts [76]. These findings have opened new perspective in our research. Lipid rafts have also been identified in neurons, which appear to be crucial for some functions, including neurotrophic factors, adhesion, axon guidance, vesicular trafficking, and protein sorting. It has been described that lipid rafts are essential for the maintenance of α7 nicotinic acethylcholine receptor clusters in somatic spines in ciliary ganglion neurons and for the synaptic stability of the AMPA-type glutamate receptors in hippocampal neurons [7, 28]. In addition, the GABAB receptors have also been associated with lipid rafts in cerebellum [5]. The evidence of functional cross-talk between the presynaptic dinucleotide receptor and GABAB receptors described before could be based on the co-localization of both receptors in specific microdomains. This assumption is feasible as P2X3 and GABAB receptors have been found to be associated with lipid rafts in cerebellar granule neurons [68].

All these data suggest that the models under study, synaptosomal preparations, astrocytes, and granule neuron cultures from cerebellum, provide good models in which to explore the open new perspectives for different aspects of nucleotide and dinucleotide receptors.