Inhibition of the ATP-gated P2X7 receptor promotes axonal growth and branching in cultured hippocampal neurons

The formation of neuronal circuits initially depends on establishing specialized domains in neurons – the axon and the dendrites. The first important morphological change in neuronal differentiation involves the formation and growth of an axon, which permits the transfer of information between neurons. Axon growth requires the coordinated reorganization of the microtubule and microfilament cytoskeleton through specific signalling cascades.

Axon formation and growth is positively regulated by different extracellular signals, such as neurotrophic factors, neurotransmitters and other signalling molecules. A variety of proteins in different pathways have been implicated in axon formation and growth, including PI3-kinase (PI3K), Akt, GSK3, JNK, ζPKC, mPar3/6, rap1B and LKB1 (Garrido et al., 2007; Schwamborn and Puschel, 2004; Shelly et al., 2007; Shi et al., 2004; Shi et al., 2003). The coordination of these pathways controls cytoskeleton dynamics, regulating the polymerization and depolymerization of microtubules and microfilaments (Bradke and Dotti, 1999; da Silva and Dotti, 2002). Significantly, these processes not only facilitate axon growth but are also important for the localization of membrane receptors at specific sites along the axon (Hirokawa and Takemura, 2005). However, although the role of receptors of neurotrophic factors in axon development has been studied extensively, little attention has been paid to the influence of purinergic receptors on axon development and growth.

Purines act as neurotransmitters and modulators in the central and peripheral nervous system, but extracellular purines can exert a direct trophic influence on the development and maintenance of the nervous system, and in its response to disease and injury (Rathbone et al., 1999). One such purine, ATP, behaves as a fast neurotransmitter in the central nervous system by acting through ligand-gated cationic channels known as P2X receptors (Burnstock, 2007; Edwards et al., 1992; Evans et al., 1992). These receptors can transiently increase intracellular Ca²⁺ concentrations by permitting the influx of this ion independently of voltage-dependent Ca²⁺-channel activation (Khakh, 2001). Ca²⁺ influx and the transient elevation of the Ca²⁺ concentration in growth cones regulates the rate of axon outgrowth and, whereas suppressing Ca²⁺ transients accelerates axon extension, Ca²⁺ influx slows otherwise rapid axonal growth (Gomez and Spitzer, 1999). P2X receptors are formed by the homomeric or heteromeric combinations of seven different subunits, P2X1-P2X7 (Torres et al., 1999), which are widely distributed throughout the mammalian central nervous system (North, 2002; Rubio and Soto, 2001; Yu et al., 2008). In the hippocampus, the P2X receptors modulate neurotransmitter release (Fellin et al., 2006; Rodrigues et al., 2005) and facilitate the induction of long-term potentiation (LTP) (Pankratov et al., 2002), as well as interacting with other membrane receptors (Khakh et al., 2005). Moreover, P2X1, P2X2 and P2X4 might participate in the formation of neuronal networks during hippocampal development, indicating a trophic role of purinergic signalling (Heine et al., 2006; Rathbone et al., 1999). However, little is known about the participation of P2X receptors in the regulation of axonal growth or about the signalling cascades regulated by these receptors in neurons during their differentiation.

We investigated the role of ATP and its P2X receptors in the regulation of axonal growth in cultured hippocampal neurons, and demonstrated that ATP exerts a negative influence on the elongation and branching of the axons produced by hippocampal neurons. Functional P2X7 receptors were identified on the growth cones of hippocampal neurons in culture, a location that would support their participation in axon growth. These purinergic P2X7 receptors mediate Ca²⁺ influx in the distal domain of the axons and the effects of ATP on axonal growth. Indeed, the inhibition or suppression of P2X7-receptor activity with interference RNA promoted increased axonal growth and branching. With regards to the signalling pathway that promotes this increase in axonal growth, we show that P2X7-receptor inhibition decreases Ca²⁺-calmodulin dependent protein kinase II (CaMKII) phosphorylation whereas it increases the activity of focal adhesion kinase (FAK), the activation of PI3K and the modification of its downstream targets, Akt and GSK3.

To study the role of ATP in the regulation of axonal growth, we used a well-established model of cultured hippocampal neurons. As ATP binds to ionotropic and metabotropic purinergic receptors, we first tested whether ATP could mobilize Ca²⁺ in axons. Neurons were plated on coverslips and cultured to stage 3, when the axon has started its elongation (Dotti et al., 1988), before they were loaded with Fura-2 dye. When the Ca²⁺ response of these neurons was then analyzed in the presence of 1 mM ATP, an increase in the concentration of Ca²⁺ was evident in their distal axon and in the axonal growth cone (Fig. 1A, trace 3 and image b). Following the ATP-induced increase in Ca²⁺, neuronal viability was confirmed by stimulation with 60 mM KCl. The changes in Ca²⁺ that were brought about by ATP suggested that it might play a role in axon growth. We examined this possibility by exposing neurons that had been cultured for 24 hours to 1 mM ATP every 12 hours for the following 72 hours in culture. These neurons were then fixed and stained with an anti-tyrosinated-tubulin antibody and phalloidin–Alexa-Fluor-594 to observe neuronal morphology. When axonal length and branching was analyzed in 100 neurons from three independent experiments, the axons of neurons that were exposed to ATP were significantly shorter (153.03±13.17 μm) than those from control neurons (199.04±14.94 μm), and they also had fewer axonal ramifications (0.1±0.05 versus 1.39±0.19, respectively; Fig. 1B,C).

In view of the effect of ATP on axonal growth and its site of action, we examined which ATP receptor(s) might influence axonal growth, the P2X or the P2Y purinergic receptors. The local Ca²⁺ response provoked by ATP suggests that P2X purinergic receptors, known to act as Ca²⁺ channels, are likely to be involved in the reduction of axon length. P2Y metabotropic receptors mobilize Ca²⁺ from the endoplasmic reticulum (Burnstock, 2007). Thus, we treated cultured hippocampal neurons with pharmacological antagonists specific for the different P2X receptors. Whereas Ip5I (1 μM) blocks P2X1 and P2X3 receptors in the micromolar range (King et al., 1999), PPADS (30 μM) is an antagonist of P2X2 and P2X4 receptors (North, 2002), and Brilliant Blue G (BBG, from 0.1 to 5 μM), KN-62 (50 nM) and A-438079 (200 nM) are specific antagonists of P2X7 (Humphreys et al., 1998; Jiang et al., 2000; McGaraughty et al., 2007). Neurons were cultured for 72 hours in the presence of these antagonists before they were stained with different markers to observe their morphology (Fig. 2). The number of secondary axonal ramifications was defined as the number of branching points from the primary or longest axon, and tertiary ramifications were those growing from secondary ramifications. The neurons exposed to BBG (5 μM) had longer axons (531.65±150.45 μm) than control neurons (295.9±12.66 μm) and, on average, these longer axons also produced more secondary (4.35±0.08) and tertiary (0.72±0.13) axonal ramifications than control neurons (0.92±0.1 and 0.10±0.05, respectively: Fig. 2A,B). Indeed, similar results were obtained with lower concentrations of BBG (100 and 500 nM) and, not only was the total axonal length greater, but also the length of the primary axon increased without considering the length of the axon ramifications (supplementary material Fig. S1). This effect on axonal growth and ramification was not produced by the P2X1 and P2X3 antagonist Ip5I (1 μM). In addition, neurons developed longer axons (488.66±118.71 μm) and produced more secondary ramifications (2.71±0.54) than control neurons in the presence of PPADS (30 μM), although this effect was more modest than that produced by BBG (Fig. 2B). When the other specific P2X7 antagonists were assessed, both KN-62 (50 nM) and A-438079 (200 nM) produced a similar effect on axonal elongation and branching as BBG (Fig. 2C,D and supplementary material Fig. S1). To rule out the possibility that the increment in axonal branching could be linked to longer axons, we analyzed the ratio between the number of ramifications and axonal length. This ratio was only significantly different, when compared to control neurons, in neurons treated with BBG, KN-62 or A-438079 (Fig. 2B,D and supplementary material Fig. S1). Hence, it appeared that inhibition of P2X7 receptors did indeed promote axonal growth and branching.

In order to validate the modifications to axon growth that are provoked by P2X7 antagonists, we adopted a second approach based on the expression of P2X7-receptor short-hairpin RNA (shRNA). We designed a pSUPER-neo-GFP-vector-derived shRNA strategy to target the P2X7 receptor (see Materials and Methods). The effectiveness of this shRNA-P2X7 approach was first confirmed in the HEK 293T cell line, which does not express native P2X7. This receptor was first expressed in HEK 293T cells using a pcDNA3-P2X7 vector, and these cells were then co-transfected with either the vector expressing the interference RNA for P2X7 (hereafter termed shRNA-P2X7) or with an unspecific shRNA (shRNA-Luc). Accordingly, the expression of exogenous P2X7 was specifically reduced by 65% in cells that were co-transfected with shRNA P2X7 (Fig. 3A). Having confirmed the efficiency and specificity of shRNA-P2X7, hippocampal neurons were transfected with the shRNA-P2X7-containing vector at 1 DIV (day in vitro), and these neurons were fixed at 3 DIV to examine the length and ramifications of their axons. Similar to those exposed to P2X7 antagonists (Fig. 2), neurons transfected with shRNA-P2X7 had significantly longer axons (908±221.7 μm) than neurons transfected with a plasmid expressing GFP alone (395.9±93.39 μm). Moreover, the axonal length of neurons transfected with a plasmid expressing a non-specific shRNA was similar to those expressing GFP alone (517.8±105.9 μm; Fig. 3B,C). Furthermore, knockdown of P2X7 expression also produced a significant increment in the total number of axon ramifications (4.62±0.53) when compared with control GFP-transfected neurons (1.16±0.33) at 3 DIV.

Neurons were also nucleofected before plating with plasmids expressing shRNA-P2X7 or GFP (Fig. 3D). As previously shown, the reduction in P2X7 expression by shRNA-P2X7 induced longer and more-branched axons; thus, we analyzed the effect of expressing an exogenous P2X7-GFP receptor (Fig. 3D). Interestingly, P2X7-GFP-nucleofected neurons had shorter axons than GFP-nucleofected neurons, as previously observed when ATP was added to neurons (Fig. 1B). Moreover, in the neurons expressing P2X7-GFP, GFP fluorescence was preferentially localized in the distal region of axons rather than in their proximal domain, although P2X7-GFP was also detected in the soma and dendrites (Fig. 3D). These results confirm the involvement of P2X7 in axonal development, as previously shown using a pharmacological approach.

In light of our data, we analyzed the distribution of the P2X7 receptor in cultured hippocampal neurons at 3 DIV. Neurons were stained with two different anti-tubulin antibodies (anti-α-tubulin and anti-tyrosinated-α-tubulin) to define their morphology and with an antibody previously shown to specifically recognize P2X7 receptors (Miras-Portugal et al., 2003) (Fig. 3A). The P2X7 receptor was expressed in the soma and was also polarized to the distal region of axons (Fig. 4A,B), where we have shown that ATP induces an increment in Ca²⁺ concentration (Fig. 1A). We confirmed this differential expression by measuring the fluorescence intensity of P2X7 and anti-tyrosinated-α-tubulin (Fig. 4A,C) or anti-α-tubulin (Fig. 4B,D) along the axons. The intensity of the fluorescence that was associated with P2X7 displayed a gradient that was most intense at the distal extreme of the axon (Fig. 4C,D). In addition, P2X7 was present in the actin-rich region of the growth cone (Fig. 4E). This localization suggests that P2X7 regulates microtubule and microfilament dynamics in the axonal growth cone.

To confirm that the P2X7 at the growth cone is functional, we performed microfluorimetric Ca²⁺ studies to demonstrate that the Ca²⁺ influx mediated by ATP in axonal growth cones of hippocampal neurons (Fig. 1A) was blocked by specific antagonists of P2X7. Axon growth cones (Fig. 4F,G) were identified in Fura-2-dye-loaded hippocampal neurons and their response to 1 mM ATP in the absence or presence of physiological Mg²⁺ concentrations was analyzed in growth cones, as well as at different points along the axon (Fig. 4H). In accordance with the pharmacological properties of P2X7, the Ca²⁺ response induced by ATP was higher in the absence of Mg²⁺. Moreover, the response that was evoked by ATP in the absence of Mg²⁺ was abolished when neurons were pre-treated with BBG (1 μM, Fig. 4I), and identical results were obtained when Bz-ATP (100 μM) and o-ATP (100 μM) were used as a P2X7 agonist and antagonist, respectively (data not shown). Hence, we concluded that P2X7 is active at the axon growth cone and that it is involved in the regulation of Ca²⁺ concentrations in the growth cone as well as on the effect of ATP on axonal growth.

We examined the intracellular signalling pathway that might be regulated by the P2X7 receptor. Among the Ca²⁺-sensitive Ca²⁺ kinases, the specific properties of CaMKII make it a suitable candidate to participate in P2X7-receptor signalling. CaMKII is one of the most abundant proteins in the brain and is phosphorylated by activation of P2X7 receptors in cerebellar granule cells (Leon et al., 2006). We analyzed the phosphorylation status of CaMKII in hippocampal neurons cultured in the presence or absence of BBG in western blots. Exposure to BBG for 30 or 60 minutes did not appear to produce any change in CaMKII phosphorylation, probably due to the intense expression of CaMKII in the somatodendritic compartment, which could mask the possible effect of BBG on CaMKII phosphorylation at axonal growth cones (supplementary material Fig. S2). Thus, we analyzed phosphorylated CaMKII (CaMKII-P) expression at axon growth cones in control and BBG-treated neurons (Fig. 5A-C). Interestingly, when we quantified the expression of CaMKII-P at the growth cone, we detected a significant reduction in CaMKII-P fluorescence intensity in BBG-treated neurons (85.93±2.34%) when compared with control neurons (100±3.9%; Fig. 5C). Moreover, when we analyzed the phosphorylation of synapsin I, a well-established presynaptic CaMKII substrate, the reduced levels of CaMKII phosphorylation at growth cones were associated with a decrease in synapsin-I phosphorylation on exposure to BBG (Fig. 5D). Furthermore, the morphology of neurons exposed for 3 days in culture to a specific inhibitor of CaMKII activity, KN-93 (1 μM), was analogous to that of the neurons treated with P2X7 antagonists (Fig. 5E). Indeed, neurons exposed to KN-93 had longer axons (570.37±21.16 μm) and more ramifications (2.71±0.17) than control neurons (182.46±9.74 μm and 0.21±0.05, respectively). These results suggest that CaMKII mediates the P2X7 signals that control axonal elongation. Moreover, alterations to growth-cone morphology were detected in neurons treated with BBG, suggesting that alterations in the actin cytoskeleton might influence the increase in axon elongation (Fig. 5A,B).

As mentioned, the changes in axon length and ramification induced by P2X7 antagonists in hippocampal neurons (Figs 2 and 3) might be due to alterations in the actin cytoskeleton. Thus, we analyzed the morphology of the axonal growth cones in neurons maintained in the presence or absence of P2X7 antagonists. The growth cones of neurons exposed to P2X7 antagonists displayed a filopodia-like morphology, without an expanded lamellipodia (Fig. 6B,C), unlike the more-extended growth cones observed in control neurons (Fig. 6A). Whereas control neurons presented a microfilament lattice that dominates the growth cone margin and its protrusions (Fig. 6A), neurons treated with P2X7-receptor antagonists presented growth cones characterized by a more intense actin staining and the presence of bundled microtubules terminated proximally to the axonal tip (Fig. 6B,C). It has been previously described that local Ca²⁺ increments can depolymerize microtubules (Keith et al., 1983). Thus, a lower concentration of Ca²⁺ at axon terminals due to P2X7 inhibition can induce a higher rate of microtubule polymerization, which invades the actin domain of the growth cone.

To study how these morphological modifications were produced, we examined FAK, a protein that interacts with actin and that controls focal-adhesion-contact formation and lamellipodia stability (Robles and Gomez, 2006). It has been shown that CaMKIIα phosphorylated on threonine 286 phosphorylates FAK at serine 843 and that, when this residue is not phosphorylated, activated FAK autophosphorylates its own tyrosine 397 residue (Fan et al., 2005). We examined the expression of FAK in growth cones of neurons treated with BBG (Fig. 7A) and found that they had fewer focal adhesions than those that develop in control neurons. This difference might be due to the different morphologies between control and treated neurons; thus, we examined whether the amount or activity of FAK was modified by exposing cultured hippocampal neurons to P2X7 antagonists. In agreement with the decrease in CaMKII activity (Fig. 5), serine-843 phosphorylation of FAK was lower in neurons treated with BBG for 60 minutes (Fig. 7B) compared with control neurons. The reduced levels of phosphorylated FAK^S843 were coupled with increased tyrosine-397 phosphorylation of FAK (FAK^Y397-P). FAK^Y397-P is considered as an active form of FAK (Fig. 7C).

The aforementioned results demonstrate that the inhibition of functional P2X7 receptors at the axon growth cone promotes axon growth and branching. In addition, axon growth was accompanied by morphological changes of the growth cone that were associated with an increment in FAK activity. Active FAK (FAK^Y397-P) can activate PI3K (Chen et al., 1996; Xia et al., 2004), which in turn regulates axonal growth (Shi et al., 2003). Thus, we analyzed whether inhibition of PI3K abolished the axon growth that was induced by P2X7 inhibition and, thus, whether PI3K was a component of the signalling cascade regulated by P2X7. Neurons were cultured for 24 hours and were then exposed to BBG (5 μM) in the presence or absence of a PI3K inhibitor (LY-294002, 50 μM). When fixed and stained with anti-α-tubulin antibody at 3 DIV, neurons exposed to BBG had longer axons and more ramifications than control neurons (as shown previously, Fig. 2). By contrast, the axons of neurons exposed to BBG and the PI3K inhibitor had a similar morphology to control neurons (Fig. 8A). Whereas the axon length of BBG-treated neurons was 551.2±30.69 μm, in the presence of LY-294002 the mean axon length was 104.19±5.15 μm. Also, the increase in the number of axon ramifications was abrogated by the presence of the PI3K inhibitor (Fig. 8B) and, hence, PI3K appeared to mediate the inhibition of P2X7. We evaluated whether the activity of proteins regulated by PI3K activity were modulated by P2X7 inhibition. Indeed, when 2-DIV hippocampal neurons were exposed to 5 μM BBG for 30 or 60 minutes, Akt phosphorylation was augmented in extracts of these cells (Fig. 8C). GSK3 phosphorylation also lies downstream of PI3K and Akt, and has been related to axon growth and ramification (Garrido et al., 2007; Kim et al., 2006). GSK3 phosphorylation increased significantly following exposure to BBG, indicating that PI3K activity was augmented when P2X7 was inhibited (Fig. 8D). Because the increment in Akt and GSK3 phosphorylation was small but significant, we analyzed the phosphorylation status of a well-known GSK3 substrate, tau. The ratio between tau-1 (dephosphorylated epitope) and PHF-1 (the hyperphosphorylated epitope) increased almost twofold after exposure to BBG, demonstrating that BBG inhibits the activity of GSK3 (Fig. 8E). Moreover, a specific GSK3 inhibitor, AR-A014418, impeded the inhibitory influence of ATP on axon growth (Fig. 8G), as also occurred when neurons were cultured in the presence of BBG and ATP (supplementary material Fig. S3). Finally, when hippocampal neurons were maintained in the presence of BBG (5 μM), Ip5I (1 μM) or PPADS (30 μM) for 24 hours to 6 DIV (see Materials and Methods), only BBG treatment increased GSK3 phosphorylation, whereas the other P2X antagonists did not significantly modify GSK3 phosphorylation (Fig. 8F).

Several factors and receptors have been shown to induce axon growth and its development, as well as the establishment of synaptic connections. However, the correct establishment of neuronal circuits also needs mechanisms that negatively control the growth of axons, as occurs during axon guidance. Different secreted proteins and extracellular matrix proteins have been shown to negatively regulate axon growth, such as semaphorins and ephrins (Yamamoto et al., 2002). Nevertheless, it remains unclear whether other membrane receptors, factors or molecules might also mediate a similar negative effect on growth. Here, we show that the ionotropic P2X7 purinergic receptor plays an important role in the negative control of axon growth in cultured hippocampal neurons when activated by ATP. The negative effect of ATP in neuritogenesis has been previously demonstrated in neural-tube explants (Cheung et al., 2005) and in spiral ganglion neurons, acting through P2X2 and P2X3 receptors (Greenwood et al., 2007). Here we show that the P2X7 receptor is present in hippocampal neurons and that it is located in axon growth cones, in accordance with a recent study that identified P2X7 mRNA in hippocampal neurons (Yu et al., 2008). P2X7 can also be found in presynaptic terminals in the central nervous system (Deuchars et al., 2001), mainly in the hippocampal neurons (Anderson and Nedergaard, 2006; Armstrong et al., 2002; Kukley et al., 2004), as well as in presynaptic terminals of cultured cerebellar granule cells (Miras-Portugal et al., 2003). Despite the localization of the P2X7 receptor in axon growth cones (presynaptic terminals) and its relation to the modulation of neurotransmitter release, the function of this receptor during axon development and regeneration has yet to be studied in neurons.

Our results demonstrate that ATP diminishes the length of the axon and also increases Ca²⁺ levels, which can be prevented by the addition of a specific P2X7 antagonist, BBG. A reduction in intracellular Ca²⁺ at the axon growth cone has been linked with increased axon growth (Gomez and Spitzer, 1999). Hence, these results indicate that P2X7 exerts a negative influence on axonal growth. Interestingly, our results show, for the first time, that the inhibition of P2X7 receptors, or their knockdown with shRNAs, permits hippocampal neurons to grow longer axons with multiple ramifications, suggesting a relationship between P2X7 and cytoskeletal dynamics. This is supported by the localization of P2X7 and the exogenous P2X7-GFP protein at the most distal region of the axon. P2X7 inhibition has also been related to the maintenance of neuromuscular synapses (Bettini et al., 2007) and, in fact, we found modifications to axon growth-cone morphology when P2X7 receptors were inhibited, changing from a lamellipodial to a filopodial morphology. In agreement with these morphological changes in our model of cultured hippocampal neurons, Ca²⁺ influx only occurred at the axon growth cone and the distal region of the axon, where P2X7 is expressed. These regions are the most dynamic in terms of axon outgrowth and actin depolymerization, which permits microtubule growth. Previous studies have also shown that increasing the Ca²⁺ concentration can depolymerize microtubules (Keith et al., 1983). Thus, the expression of the P2X7 receptor and its activation at the axon growth cone could mediate local actin polymerization and repress axon growth. Indeed, P2X7 co-precipitates with proteins that are associated with the actin cytoskeleton (Kim et al., 2001) and, recently, P2X-receptor activation by ATP was shown to induce the formation of cofillin rods and the retraction of neurites in PC12 cells (Homma et al., 2008). Accordingly, our results imply that the P2X7 receptor participates in the negative control of axon growth, as demonstrated by the use of P2X7 antagonists and P2X7 shRNAs. The inhibition or suppression of these receptors could lead to actin depolymerization and permit axon growth, as well as the formation of multiple axon terminals.

How does the inhibition of P2X7 increase axon growth? In response to ATP, P2X7 functions as a Ca²⁺ channel and CaMKII might therefore be involved in the signalling cascade activated by P2X7. Indeed, CaMKII is phosphorylated by activation of P2X7 receptors in cerebellar granule cells (Leon et al., 2006). CaMKIIβ can bundle and cross-link F-actin filaments (O'Leary et al., 2006) and it dissociates from F-actin upon Ca²⁺/calmodulin stimulation (Fink et al., 2003). In postnatal day 19 (P19) neurons, inhibition of CaMKII induces the reorganization of F-actin and the formation of growth cones with a filopodial structure (Easley et al., 2006). Our data show that the inhibition of P2X7 reduces the amount of phosphorylated CaMKII at the axonal growth cones, supported by a reduction in the levels of phosphorylated synapsin I. Furthermore, inhibition of CaMKII activity induced the same morphological changes as P2X7 inhibition.

Our results therefore suggest that P2X7 could regulate actin-associated proteins through CaMKII. One of these, FAK, is thought to play a pivotal role in the cellular dynamics controlling axon branching and growth in hippocampal neurons (Rico et al., 2004). FAK can be phosphorylated at serine 843 by CaMKII and, in the absence of this serine phosphorylation, FAK activity is augmented by the phosphorylation of tyrosine 397 (Fan et al., 2005). We show that BBG increases the tyrosine-397 phosphorylation of FAK, thereby activating this kinase. This autophosphorylation of FAK at tyrosine 397 was coupled with less serine-843 phosphorylation of FAK, suggesting that these changes in FAK phosphorylation were due to P2X7 inhibition, effected through CaMKII. In addition, local increases in Ca²⁺ levels contribute to the disappearance of FAK^Y397-P and the loss of focal contacts. By contrast, the elimination of extracellular Ca²⁺ increases the FAK^Y397-P staining at growth cones (Conklin et al., 2005). Thus, the changes in axon growth-cone morphology provoked by BBG that we observed are probably due to a decrease in the Ca²⁺ influx through P2X7 receptors.

FAK acts upstream of PI3K (Xia et al., 2004) and FAK^Y397-P binds and activates PI3K (Chen et al., 1996). Our results show that the inhibition of PI3K activity abrogates the effects of P2X7 antagonists on axon growth and branching. Thus, the inhibition of P2X7 receptors by specific antagonists induces FAK activation and regulates PI3K activity, promoting axon growth and branching. Moreover, P2X7 antagonists induce the phosphorylation of downstream PI3K targets such as Akt and GSK3, activating Akt and inhibiting GSK3. Phosphorylated Akt is present in the axon growth cone and has been associated with axon formation (Schwamborn and Puschel, 2004). Increased Akt phosphorylation can also induce the inhibition of GSK3 (GSK3^S9/21-P), which has been shown to promote axon growth (Zhou et al., 2004) and branching (Garrido et al., 2007). The inhibition of GSK3 induced by P2X7 antagonists is also confirmed by the increase in dephosphorylated tau. Moreover, a GSK3 inhibitor can overcome the negative effect of ATP in axonal growth.

In conclusion, P2X7 receptors could act as sensors of regions of cell death (Jun et al., 2007; Khakh and North, 2006), avoiding the growth and establishment of synaptic connexions in the regions where there are higher concentrations of extracellular ATP produced by cell rupture. This idea is supported by the demonstration that inhibiting P2X7 induces axon growth and branching in cultured hippocampal neurons. These morphological alterations were also observed when the endogenous expression of P2X7 was abolished in hippocampal neurons, confirming the specific effect of the P2X7 antagonist, BBG. Together, these results strongly suggest that the inhibition of P2X7 could be considered as a therapeutic approach to recover dysfunctional neuronal circuits. In fact, the use of P2X7 antagonists in mice improves recovery after spinal-cord injury (Wang et al., 2004). This improvement was mediated by the decrease in the inflammatory response produced by microglia and macrophages, which also express P2X7 (Wang et al., 2004). Our results provide new evidence suggesting that P2X7 inhibition improves the growth and possibly the regeneration of injured axons.

The following reagents were used in this study: ATP (A5394), Bz-ATP (B6396), o-ATP (A6779), PPADS (027K4601) and BBG (B5133), all purchased from Sigma; KN-62 (1277) and A-438079 (2972) were obtained from Tocris; Ip5I was synthesized as previously described (Pintor et al., 1997); and LY-2940002 was purchased from Calbiochem.

Hippocampal neurons were prepared as described previously (Banker and Goslin, 1988). Briefly, the hippocampus was obtained from E17 mouse embryos and, after dissection and washing three times in Ca²⁺/Mg²⁺-free HBSS, the tissue was digested in the same solution with 0.25% trypsin for 15 minutes at 37°C. The hippocampi were then washed three times in Ca²⁺/Mg²⁺-free HBSS and dissociated with a fire-polished Pasteur pipette. The cells were counted, resuspended in plating medium (MEM, 10% horse serum, 0.6% glucose) and plated at a density of 5000/cm² on polylysine-coated coverslips (1 mg/ml). After plating, neurons were cultured for 3 days in neuronal culture medium (Neurobasal, B-27, glutamax-I). For biochemical experiments, hippocampal neurons were plated on polylysine-coated (1 mg/ml) 60-mm plates at a density of 200,000 cells/cm². They were then cultured for 72 hours or 6 DIV in neuronal culture medium, and treated or not. AraC was added at a final concentration of 5 μM after 2 days in culture. To analyze the effect of P2X-receptor agonists and antagonists, the compounds were added to the cultured neurons 4 hours or 1 day after plating. For biochemical experiments, hippocampal neurons were plated on 60-mm plates coated with polylysine (1 mg/ml) at a density of 200,000/cm², and they were cultured for 72 hours or 6 DIV with or without treatment.

HEK 293T cells were maintained in DMEM (Gibco) supplemented with 10% (v/v) FCS. Cells were reseeded at 10⁵ cells/cm² 1 day before transfection, after which FCS was reduced to 0.5% (v/v).

The human P2X7 full-length cDNA was purchased and sequenced by Geneservice (cDNA clone number MGC: 20089, IMAGE: 4298811; Cambridge, UK). The P2X7 cDNA was isolated from the original plasmid (pOTB7) by digestion with EcoRI and XhoI, and then subcloned into the corresponding sites of pcDNA3 for expression in mammalian cells. To construct the P2X7-GFP plasmid, P2X7 was cloned into the pd2EGFP-N1 vector (Clontech) and the ligation product was confirmed by sequencing. P2X7-receptor knockdown was achieved by RNA interference (RNAi) using a vector-based shRNA approach. The shRNA target sequence 5′-GTTTTGACATCCTGGTTTT-3′ was selected for the P2X7 receptor according to a previously reported rational-design protocol (Reynolds et al., 2004). As a control, we used the firefly-luciferase-targeted oligonucleotide 5′-CTGACGCGGAATACTTCGA-3′. The specificity of the sequence was confirmed by a BLAST analysis for human, mouse and rat P2X7. Synthetic forward and reverse 64-nucleotide oligonucleotides (Sigma Genosys) were designed, annealed and inserted into the BglII-HindIII sites of the pSUPER-neo-GFP vector (OligoEngine, Seattle, WA) following the manufacturer's instructions. These constructs express 19-bp nine-nucleotide stem-loop shRNAs targeted against P2X7 or luciferase (control shRNA) mRNAs. The concomitant expression of GFP from this vector allowed transfected cells to be identified by fluorescence.

HEK 293T cell transfections were performed with the pSUPER-neo-GFP-derived plasmid constructs using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen). After 6 hours, the medium was removed and the cells were further incubated for the periods indicated in the figure legends in the presence of culture medium. Neuronal transfection was carried out 24 hours after plating using Lipofectamine 2000 (9 μl, Invitrogen) and 3 μg of control shRNA-Luc or shRNA-P2X7R vectors. The transfection mix was removed after 2 hours and the neurons were washed and maintained for 3 DIV. In another set of experiments, neurons were nucleofected before plating using the Amaxa nucleofection kit for hippocampal neurons.

Hippocampal neurons were cultured on coverslips treated with polylysine as described above. The day after the cells were plated, neurons were washed with HBM buffer (140 mM NaCl, 5 mM KCl, 1.2 mM NaH₂PO₄, 1.2 mM NaHCO₃, 1 mM MgCl₂, 10 mM glucose and 10 mM HEPES, pH 7.4) and loaded with Fura-2AM solution (5 μM) for 30 minutes at 37°C. This period facilitated the intracellular hydrolysis of Fura-2AM. Subsequently, the coverslips were washed again with HBM medium and mounted in a superfusion chamber on a NIKON Eclipse TE-2000 microscope. Neurons were continuously superfused at 1.2 ml/minute with HBM perfusion media during functional assays. Control pulses of 30 seconds with ATP (1 mM) or Bz-ATP (100 μM) were applied to neurons. In some cases, neurons were pre-incubated for 5 minutes with the specific P2X7 antagonists o-ATP (100 μM) or BBG (1 μM) before ATP or BzATP was applied in the presence of both P2X7 antagonists. A pulse of 60 mM K⁺ was applied at the end of each experiment to confirm the viability of the neurons under study. In some cases, the perfusion medium and drugs were dissolved in HBM Mg²⁺-free medium, in which the MgCl₂ was substituted by glucose at a concentration that preserved the osmolarity. Neurons were visualized using a Nikon microscope containing a ×40 S Fluor 0.5-1.3 oil lens. The wavelength of the incoming light was filtered to 340 nm and 380 nm with the aid of a monochromator (10-nm bandwidth, Optoscan monocromator, Cairin). The 12-bit images were acquired with an ORCA-ER C 47 42-98 CCD camera from Hamamatsu (Hamamatsu City, Japan) controlled by Metafluor 6.3r6 PC software (Universal Imaging, Cambridge, UK). The exposure time was 250 ms for each wavelength and the changing time was <5 ms. The images were acquired continuously and buffered in a fast SCSI disk. The time-course data represent the average light intensity in a small elliptical region within each cell. The background and autofluorescence components were subtracted at each wavelength.

The commercial antibodies used here were raised against: P2X7, obtained from Alomone (Jerusalem, Israel), Chemicon (Temecula, CA) and GE Health Care-Pharmacia (Buckinghamshire, UK); Akt, obtained from Santa Cruz Biotechnology (Santa Cruz, CA); tyrosinated-α-tubulin, α-tubulin and β-actin from Sigma (St Louis, MI); phospho-GSK3 (pS9/21) and phospho-Akt (pS473), obtained from Cell Signalling (Beverly, MA); GSK3 α/β,, CaMKII α/β, phospho-FAK (pY397), phospho-FAK (pS843) and FAK, purchased from Invitrogen-Biosource (San Francisco, CA); phospho-CaMKII α/β (pT286/287) or phospho-synapsin-I (pS603), from Upstate cell signalling solutions (Lake placid, NY, USA); tau-1, from Chemicon; and PHF-I, a gift from Jesús Avila (CBM, Madrid).

Immunocytochemistry was performed on neurons cultured for 3 DIV following fixation in 4% paraformaldehyde for 20 minutes. Non-specific binding was blocked with 0.22% gelatin and 0.1% Triton X-100 in 0.1 M phosphate buffer (PB). The cells were then incubated with primary antibodies for 1 hour at room temperature, washed and incubated with Alexa-Fluor-488 or Alexa-Fluor-594-conjugated secondary antibodies (1:1000) and Alexa-Fluor-594-conjugated phalloidin (1:100). The coverslips were finally mounted using fluoromount G (Southern Biotech), and images were acquired using a Leica DMI 6000 B coupled to a Leica DFC 350 FX camera and a Leica TCSSP5 confocal microscope. Analysis of axon length and ramifications were carried out using the Neuron J program. Fluorescence intensity was evaluated using the RGB colour profiler tool of the ImageJ program. Images were processed and presented using Adobe Photoshop and Illustrator CS3.

Cultured neurons were lysed and homogenized in a buffer containing 20 mM HEPES, pH 7.4, 100 mM NaCl, 10 mM NaF, 1% Triton X-100, 1 mM sodium orthovanadate, 10 mM EDTA and protease inhibitors (2 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin and 10 μg/ml pepstatin). Proteins were separated on 10% SDS-PAGE gels and transferred to nitrocellulose membranes (Schleicher and Schuell). The experiments were performed using primary polyclonal antisera (and dilutions) against: P2X7 (1:1000); CaMKIIα/β^T286/287-P (1:1000); Akt (1:1000); Akt^S473-P (1:1000); GSK3^S9/21-P (1:1000); FAK^Y397-P (1:1000); FAK^S843-P (1:1000) and FAK (1:1000), or monoclonal antibodies against: α-tubulin (1:10,000); CaMKIIα/β (1:1000); synapsin-I^S603-P (1:1000); GSK3α/β (1:1000); tau-1 (1:5000); tau-PHF1 (1:100); and β-actin (1:1000). Membranes were incubated with the selected antibodies overnight at 4°C in 5% non-fat dried milk. A secondary goat anti-mouse monoclonal antibody or a goat anti-rabbit polyclonal antiserum (both at 1:5000; Amersham) was used to detect the primary antibodies, which were visualized by ECL (Amersham).

All experiments were repeated at least three times and the results are presented as the mean and standard deviation. Statistical differences were analyzed with the aid of the Origin 7.0 software using a paired t-test or ANOVA test as indicated in figure legends.

The authors thank Francisco Wandosell for critical reading of the manuscript. This work was supported by the Centro de Investigación Biomédica en Red de Enfermedades Neurodegenerativas (CIBERNED), and by grants from the Ministerio de Educación y Ciencia (SAF2006-00906 to J.J.G.; BFU2005-02079 to M.T.M.-P.), the Comunidad Autónoma de Madrid (SAL/0551/2004 to M.T.M.-P.), the Fundación `La Caixa' (BM05-114-0 to M.T.M.-P.) and the Fundación Marcelino Botín. M.D.-H. was supported by a `Juan de la Cierva' contract and is currently the holder of a `Ramón y Cajal' contract. J.I.D.-H. was supported by a `Juan de la Cierva' contract. M.D.-Z. is the recipient of a FPU grant from the Ministerio de Educación y Ciencia.