The mitogen-activated protein (MAP) kinases extracellular signal-regulated kinases 1 and 2 (ERK1/2) regulate multiple processes in neural cells, such as the balance between survival and cell death, proliferation and differentiation, and they also play a crucial role during nervous system development and in neuroregeneration. Numerous factors modulate ERK1/2 signaling, alone or in combination, including extracellular signals, scaffolding and signaling proteins recruited to specific cascades, and other intrinsic tissue or cell factors. Moreover, the spatio-temporal regulation of the ERK1/2 pathway shapes the final biological response to its activation. In neural models, transient or sustained early activation of ERK1/2 signaling is involved in either proliferation or differentiation, respectively. By contrast, prolonged and delayed ERK1/2 activation is more closely related to cell death following certain toxic stimuli (Marshall, 1995; Chu et al., 2004). Hence, fine tuning of ERK1/2 signaling is necessary to obtain adequate responses.

Different upstream and downstream mechanisms influence the activation kinetics of ERK1/2 signaling (Junttila et al., 2008). Among those involved in their negative regulation, MAPK protein phosphatases (MKPs) are considered crucial elements that determine the final duration and magnitude of MAPK signaling (Dickinson and Keyse, 2006; Patterson et al., 2009). There are 10 dual specificity protein phosphatases (DUSPs) in the MKP family that specifically dephosphorylate Ser/Thr and Tyr residues within the activation loop of MAPKs. DUSPs can be classified according to their substrate selectivity (ERK1/2, p38 and JNK MAPKs) and their subcellular localization. Understandably, the regulation of these phosphatases is complex, particularly since the same MAPKs that represent their substrates can also regulate their expression, implicating them in a negative feedback loop that offers some control over MAPK signaling (Dickinson and Keyse, 2006; Patterson et al., 2009).

DUSP expression and activity are tightly regulated at different levels, and finally depend on cell type and specific cellular context. Regulatory mechanisms involve several processes at the transcriptional level, but also cover post-transcriptional and post-translational events. Actually, DUSP expression is rapidly increased upon growth factor receptor stimulation. Several transcription factors activated by one or more MAPK pathways are responsible for the rapid expression of Dusp genes. This implies that MAPKs acting as DUSP substrates are responsible for the transcriptional induction of Dusp genes, functioning in negative feedback loops, and in crosstalk between distinct MAPK modules (Ekerot et al., 2008; Zeliadt et al., 2008). In addition, for some inducible DUSPS, stress conditions/factors that induce stress can also favor and activate its transcriptional expression (Staples et al., 2010; Bermudez et al., 2011). At the second level, post-transcriptional regulation, involve the participation of several RNA binding proteins that have the ability to stabilize Dusp mRNA increasing its half-life, which also accounts for increases in the DUSP6 protein levels (Kuwano et al., 2008; Bermudez et al., 2011). On the other hand, specific micro-RNAs can also promote Dusp gene silencing (Zhu et al., 2010). In some aspects, MAPK themselves also have the ability to mediate both DUSP transcription and mRNA stabilization through its 3′-UTR (Bermudez et al., 2011). The third mode of control takes place by post-translational modifications that rapidly affects DUSP turnover, thereby providing a dynamic control over MAPK signaling. DUSP1 and DUSP6 provide numerous examples of phosphorylation processes in different key residues that account for either protein stabilization or increase in the rate of protein degradation. Not only MAPKs, but also different types of signaling kinases participate in this regulatory process, such as mTOR, Protein kinase C (PKC) and CK2 (Marchetti et al., 2005; Choi et al., 2006; Lin and Yang, 2006; Bermudez et al., 2008; Cagnol and Chambard, 2010).

One of the most representative members of the DUSP family that selectively dephosphorylates ERK1/2 is DUSP6, also termed MKP-3 (Muda et al., 1996; Camps et al., 1998). This is a constitutive protein phosphatase that regulates the basal levels of phosphorylated ERKs in the cytosolic compartment, and that is a negative regulator of mitogen signaling (Ekerot et al., 2008). Indeed, although it does not behave as an immediate early gene, Dusp6 can be induced transcriptionally by the Ras/MEK/ERK cascade triggered by growth factors like FGF, albeit with delayed kinetics. ERK activation also regulates DUSP6 protein expression and stability through mechanisms that involve direct phosphorylation and protein complex assembly (Marchetti et al., 2005). However, it should be noted that most studies into DUSP6 regulation have been carried out in exogenous overexpression systems, cell lines overexpressing the phosphatase or tumor cells. Regarding the latter, it remains unclear how this protein phosphatase may be implicated in cancer. While DUSP6 expression is enhanced in some kinds of cancer (melanoma and glioma cells), suggesting it may be a tumor promotor (Messina et al., 2011; Li et al., 2012), its expression is dampened in other tumors where it may exert a role as a tumor suppressor, such as pancreatic invasive cancer, primary lung cancer and ovarian cancer (Furukawa et al., 2003; Chan et al., 2008; Okudela et al., 2009). Significantly, very little is known about the intrinsic regulation of DUSP6 in native tissues and its modulation by endogenous extracellular signals that activate receptor tyrosine kinases (RTKs) e.g., mitogens (Ham et al., 2015; Donaubauer et al., 2016).

In an attempt to investigate the role of endogenous DUSP6 in nervous system, we took advantage of earlier studies carried out on primary cultures of granule neurons and astrocytes obtained from the rat cerebellum. ERK1/2 signaling cascades are well characterized in both these cell types and they are activated by stimulating different kinds of growth factor receptors or G protein-coupled receptors (GPCRs) (Delicado et al., 2005; Subramaniam et al., 2005). Interestingly, an ionotropic receptor, the P2X7 nucleotide receptor that is activated by high extracellular concentrations of ATP, is attracting growing interest in both these cell models. Indeed, ERK1/2 activation elicited by stimulation of the P2X7 receptor underlies the neuroprotection against glutamate excitotoxicity driven by this receptor (Ortega et al., 2011). ERK activation by P2X7 receptors in granule neurons is dependent on the entry of extracellular calcium and calcium/calmodulin kinase II (CaMKII) activation. Conversely, ERK activation in cerebellar astrocytes was not dependent on extracellular calcium nor does it involve nPKC isoforms (Carrasquero et al., 2010; Ortega et al., 2011). Here, we investigated the role of endogenous DUSP6 phosphatase in regulating ERK1/2 signaling, both in basal conditions and when P2X7 receptors are stimulated. Moreover, the regulation of DUSP6 by epidermal growth factor (EGF) receptors was taken as a positive control of growth factor regulation of this phosphatase. This study set out to determine if DUSP6 is specifically involved in regulating cytosolic ERK1/2 signaling in these two cellular types, and whether it is regulated by nucleotide receptors in a similar fashion as it is by growth factors.

In the present study we demonstrate the presence of the dual specificity protein phosphatase DUSP6 in granule neurons and astrocytes from the rat cerebellum, and its regulation by nucleotide P2X7 and EGF. Stimulation with a P2X7 specific nucleotide and of EGF receptors regulates the levels of DUSP6 according to biphasic kinetics. This regulation involves an early degradation phase that takes place following a short stimulation, and that represents a feedforward mechanism to amplify and prolong ERK1/2 signaling. This is followed by a second phase in which there is a recovery in DUSP6 protein levels that works as a negative feedback mechanism to restore basal ERK1/2 activity. Interestingly, ERK1/2 proteins are involved in the biphasic modulation of DUSP6 as ERK activity is required for both degradation and for the recovery phase. The data reported here indicate that this biphasic regulation seems to work as a general and universal mechanism shared by different types of extracellular signals, and that it operates similarly in both native primary cells and established cell lines (Jurek et al., 2009; Ndong et al., 2014; Ham et al., 2015). Thus, we conclude that both P2X7 and EGF receptors are key elements that shape the final biological outcome of ERK1/2 signaling through the regulation of the ERK-selective phosphatase DUSP6 in neurons and glial cells (Figure 11).

In both cell models, granule neurons and astrocytes, the DUSP6 protein undergoes rapid degradation after short stimulations with the P2X7 agonist or EGF. The half-life of DUSP6 degradation after short stimulations of 15–30 min is similar in both cell types, in accordance with previous data (Jurek et al., 2009). The proteasome inhibitor MG-132 prevents DUSP6 degradation, indicating the ubiquitin-proteasomal pathway is involved in this process. The degradation of DUSP6 protein is inversely correlated with the phosphorylation of ERK1/2 and indeed, as the DUSP6 inhibitor BCI enhances ERK1/2 phosphorylation, this protein phosphatase appears to be involved in regulating basal cytosolic ERK activation (Donaubauer et al., 2016). The specificity of BCI as an inhibitor of DUSP6 was first reported by studying the FGF-mediated increase in ERK signaling in zebrafish embryos (Molina et al., 2009). Yet BCI is also known to inhibit the nuclear and inducible phosphatase DUSP1, which exhibits broader substrate selectivity and interacts with ERK, p38 and c-Jun N-terminal kinases (JNK). However, given that enhanced ERK phosphorylation was mainly observed in the cytoplasmic compartment, we conclude that BCI is likely to have inhibited DUSP6 in both granule neurons and astrocytes.

Another interesting aspect of DUSP6 degradation in response to BzATP and EGF stimulation is the need for ERK activity, with DUSP6 degradation attenuated in the presence of the inhibitor of MEK1, U0126. Hence, ERK1/2 proteins participate in a self-regulatory loop acting via direct targeting of DUSP6, probably through the DUSP6 phosphorylation on Ser¹⁹⁷, that was dampened by the MEK inhibitor. However, U0126 only partially prevented the loss of the DUSP6 protein, as noted in BzATP stimulated granule neurons. Moreover, no DUSP6 phosphorylation was evident in these neurons after BzATP stimulation. Therefore, we cannot exclude the influence of other kinases and intracellular pathways in the regulation of DUSP6, or even the involvement of other residues susceptible to phosphorylation. Additional work is necessary to further confirm that DUSP6 phosphorylated in Ser¹⁹⁷ is subsequently ubiquitinated in astrocytes. In this regard, DUSP6 undergoes similar regulation following growth factor stimulation in systems overexpressing Ras mitogenic signaling and systems with inducible DUSP6 expression (Zeliadt et al., 2008). Accordingly, Ser¹⁵⁹ and Ser¹⁹⁷ residues in the N-terminal domain of DUSP6 were first recognized as targets for ERK1/2 and mTOR-mediated phosphorylation, promoting DUSP6 degradation in response to serum in fibroblasts ectopically expressing DUSP6 (Marchetti et al., 2005; Bermudez et al., 2008). In endothelial cells overexpressing PDGF receptors, DUSP6 underwent rapid degradation similar to that observed here in neurons and astrocytes when stimulated with BzATP or EGF. The Ser¹⁷⁴ residue was identified as a target for ERK1/2 phosphorylation and it was responsible for ubiquitin-ligase recruitment (Jurek et al., 2009), similar to the regulation of other DUSP family members. ERK1/2-mediated phosphorylation of the DUSP1 inducible phosphatase at different residues may promote either degradation or stabilization (Brondello et al., 1999; Lin and Yang, 2006). In addition, signaling kinases other than ERK1/2 can regulate DUSP1, such as PKCδ, driving its degradation via the ubiquitin-proteasome pathway in a hippocampal cell line and in cortical neurons during glutamate-induced excitotoxicity (Choi et al., 2006). These data indicate that DUSP proteins are tightly regulated and that they are submitted to multiple levels of post-translational regulation. Indeed, perhaps little is still known about the specific regulatory mechanisms that operate for particular stimuli or in specific cell types.

In terms of DUSP6 recovery, it must be taken into account that although Dusp6 is not an immediate early gene, its expression is transient and it peaks rapidly after 1–2 h of P2X7 agonist or EGF stimulation, occurring in the same time window to that found for other growth factors (Zeliadt et al., 2008; Jurek et al., 2009). Studies carried out with the transcriptional inhibitor, Actinomycin D, clearly supported that Dusp6 transcriptional induction appears to be the mechanism responsible for the increases in DUSP6 expression mediated by BzATP and EGF in both cellular models. Indeed, Dusp6 expression is influenced by the Ets-1 and Ets-2 transcriptional regulators, known targets of ERK1/2-mediated signaling (Ekerot et al., 2008; Jurek et al., 2009). Nevertheless, the recovery phase occurs much more slowly in BzATP stimulated granule neurons, as this agonist prolongs DUSP6 degradation for 1–2 h. Indeed, BzATP stimulation of granule neurons seems to be less effective in supporting DUSP6 synthesis and in promoting the wave of expression that restores basal DUSP6 protein levels after 2–4 h.

The delay in DUSP6 recovery helps maintain ERK signaling active for longer, suggesting that neural P2X7 receptors amplify cytosolic ERK signaling and redirect ERK signaling to their cytosolic targets for longer periods. The physiological meaning of transient vs. sustained ERK1/2 activation may depend on the model under study and the subcellular compartment in which ERK is activated. Prolonged ERK activation favors the survival of macrophages subjected to hypoxia, although in this case the down-regulation of DUSP6 activity was essential for survival, modulating the balance between pro- and anti-apoptotic Bcl-2 proteins (Nyunoya et al., 2005). However, in other models prolonged ERK signaling are correlated to cell death in different models, as occurs in granule neurons submitted to apoptotic stimuli (Subramaniam et al., 2005). Here, DUSP6 inhibition by BCI clearly compromises cell survival of both granule neurons and astrocytes when such inhibition exceeds 1–2 h or when the concentration used exceeds 10 μM. Therefore, DUSP6 expression is crucial to avoid the apoptotic effect of sustained ERK1/2 signaling, as described in astrocytes exposed to glutamate excitotoxicity in the ischemic brain (Szydlowska et al., 2010). Along similar lines, the detrimental effect of elevated ERK signaling in cortical neurons induced by the DNA alkylating agent cisplatin is due to the loss of DUSP6 phosphatase activity (Gozdz et al., 2008). The protective role of DUSP6 in mechanical allodynia has been also reported. Induction of DUSP6 expression by cannabinoid receptors seems to be essential to end elevated ERK signaling during mechanical allodynia in spinal neurons and glial cells (Landry et al., 2012). As well, dysregulation of DUSP6 prevents spontaneous resolution of acute postoperative pain and drives its transition to persistent pain via prolonged neuronal and microglial MAPK activation on spinal cord (Saha et al., 2013).

Regulation of cytosolic ERK activity by DUSP6 is also critical in differentiation. NGF suppresses ERK signaling in PC12 cells by inducing Dusp6 and this directly drives terminal differentiation into a neuronal phenotype (Camps et al., 1998). Interestingly, constitutive DUSP6 levels are elevated in granule neurons during their differentiation in culture, whereas DUSP6 declines after 1 week of culture, coincident with the onset of P2X7 receptor expression (unpublished results). This is consistent with the P2X7 receptor behaving as a key signal to terminate differentiation. In the presence of P2X7 antagonists, neurite outgrowth is enhanced in hippocampal neurons and neuroblastoma cells (Díaz-Hernandez et al., 2008; Gómez-Villafuertes et al., 2009). Moreover, P2X7 receptor stimulation has a trophic effect on cerebellar astrocytes, which involves PKC/PKD activation and that could be involved in astrogliosis (Carrasquero et al., 2010).

In conclusion, nucleotide P2X7 receptors behave like growth factors in terms of the fine-tuning of ERK1/2 MAPK signaling, acting through the regulation of the DUSP6 protein phosphatase. The biphasic kinetics of DUSP6 regulation involves an early degradation step that prolongs ERK1/2 activation, followed by the synthesis of new DUSP6 protein to terminate ERK1/2 signaling. Both processes occur in an ERK1,2–dependent manner and therefore, ERK1/2 regulates its own activation via DUSP6 activity through feed-forward and negative feed-back mechanisms (Figure 11). This regulation ensures the reduction in ERK phosphorylation necessary to allow the cell to respond to a subsequent stimulation, apparently functioning as a common universal mechanism.

MJQ is the recipient of a FPU fellowship and conducted most of the experiments performed in granule neurons. JCG-R is the recipient of a FPI fellowship and conducted most of the experiments performed in astrocytes. VM contributed to the experimental part of the study. FO and MTM-P contributed to the preparation of the figures and discussion of the results. EGD and RP-S contributed to the design and writing of the manuscript. All the authors analyzed the data, revised the manuscript and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.