Apoptosis has been implicated in the progression of acute and chronic neurodegenerative conditions such as stroke, spinal cord injury, Alzheimer’s disease, Parkinson’s disease and Huntington’s disease. Several kinases have been implicated in the regulation of neuronal apoptosis including AKT, GSK3 and JNK family kinases. The AKT pathway has been found to promote cell survival in many neuronal cell types 
while inhibition of AKT signaling has been shown to promote neuronal cell death 
. In contrast, the GSK3β and JNK family kinases are known to function to promote cell death in several types of neurons and inhibition or knockdown of these kinases protects neurons from a variety of apoptotic stimuli 
. While it is well established that these kinases play a key role in determining neuronal survival the mechanisms by which they regulate the apoptotic machinery remains unclear. Importantly, in the present study we have demonstrated that the AKT, GSK3β and JNK signaling pathways converge to regulate the transcriptional induction of the pro-apoptotic Bcl-2 family member Puma. Furthermore we demonstrate that induction of Puma by these kinase pathways is a critical determinant of apoptosis in cerebellar granule neurons both in vitro
and in vivo
The Bcl-2 family proteins are critical mediators of apoptosis and several studies have demonstrated that the multi-domain pro-apoptotic member Bax is essential for the execution of apoptosis in diverse neuronal death paradigms 
. It is now recognized that the BH3-only subfamily of Bcl-2 proteins play a key role in activating Bax in response to apoptotic stimuli making them likely candidates for kinase mediated regulation 
. The BH3-only family consists of multiple members and indeed several of these have been shown to be affected by AKT and JNK signaling. For example, AKT has been reported to phosphorylate Bad resulting in its sequestration by protein 14-3-3 and inhibiting its ability to induce apoptosis 
. Similar to our results with Puma, it has been reported that AKT upregulation by IGF-1 can suppress the transcriptional induction of Bim in potassium deprived CGNs 
. Furthermore, it has been shown that JNK inhibition can block transcriptional induction of the BH3-only members Bim and Hrk/DP5 in trophic factor deprived neurons 
. The role of Hrk/DP5 in trophic factor deprivation induced neuronal apoptosis appears to be neuronal subtype dependent as apoptosis is not reduced in Hrk/DP5-deficient CGNs subjected to potassium deprivation, but is partially reduced in superior cervical ganglia cells following nerve growth factor withdrawal 
. Similarly, it has previously been reported that trophic factor deprivation induced apoptotic cell death is significantly reduced in Bim-deficient neurons 
. However, we have found that potassium deprivation induced apoptosis is only modestly reduced in Bim-deficient CGNs. On the other hand we have determined that Puma plays a major role in regulating trophic factor deprivation induced apoptosis in CGNS both in vitro
and in vivo
. Furthermore, Puma-deficient neurons have been shown to be remarkably resistant to the induction of apoptosis by diverse stimuli including DNA damage, oxidative stress, ER stress/dysfunction, and proteasome inhibition 
. In addition, Puma-deletion has been shown to be neuroprotective in mouse models of severe status epilepticus
and Amyotrophic Lateral Sclerosis 
We have determined that inhibition of either JNK or GSK3β markedly reduces Puma induction and cell death suggesting that simultaneous activation of both pathways is required for Puma induction. Furthermore, our results suggest that these pathways are functioning independently and converge to regulate Puma transcription. Specifically we have determined that suppression of the AKT/GSK3β pathway by either IGF-1 mediated AKT activation or pharmacological inhibition of GSK3β does not affect the induction of JNK targets including P-c-Jun, P-ATF2 or ATF3. Similarly, we find that inhibition of JNK does not affect AKT activity as it does not appear to influence AKT mediated GSK3β (ser-9) phosphorylation. However, we cannot rule out the possibility that JNK could indirectly modulate GSK3β activity independently of AKT. Interestingly, we found that prolonged inactivation of the PI3K-AKT pathway by LY294002 was sufficient to induce Puma expression and neuronal cell death. However, we found that cell death induced by LY294002 was inhibited by the JNK inhibitor SP600125 (data not shown) suggesting that basal levels of JNK activity may be contributing to Puma induction in this context. This would be consistent with the lower levels of Puma induction and cell death observed following LY294002 mediated PI3K/AKT inactivation as compared with potassium withdrawal. Our finding that activation of both the AKT/GSK3β and JNK pathways is required to control Puma induction suggests a signaling cascade which has a built-in safety mechanism to prevent spontaneous neuronal apoptosis.
The activation of Puma mRNA induction provides the convergence point of these kinase signaling pathways, however, the exact mechanism by which they converge on Puma induction remains to be determined. As Puma is regulated at the transcriptional level it seems logical that these kinases alter the activity of transcriptional repressors or activators which in turn control Puma expression. Puma was originally identified as a target gene of the transcription factor p53, and indeed our laboratory, as well as others have demonstrated that Puma is an essential pro-apoptotic factor in p53-mediated neuronal apoptosis 
. However, Puma has been shown in many cases to be induced independently of p53 (reviewed in 
), and it is unlikely that p53 contributes to Puma induction in this model as it has previously been demonstrated that p53 is not required for potassium withdrawal induced apoptosis in CGNs 
. As such, we expected that other transcription factors, downstream of the AKT/GSK3β and JNK pathways, would be responsible for Puma upregulation following potassium deprivation in CGNs. Previous studies have implicated the transcription factor FoxO3a in trophic factor deprivation induced neuronal cell death 
. Importantly, we demonstrate that FoxO3a promotes neuronal apoptosis through the transcriptional induction of Puma. Similar to our results it has previously been reported that FoxO3a can activate Puma transcription and apoptosis in cytokine-deprived lymphoid cells 
. The nuclear localization and transcriptional activity of FoxO3a is negatively regulated by AKT-mediated phosphorylation. Consistent with this we found that IGF-1 prevented the potassium deprivation induced decrease in AKT activity, FoxO3a dephosphorylation and attenuated Puma induction. Interestingly, we found that inhibition of either JNK or GSK3β also inhibited FoxO3a dephosphorylation/activation. These results were surprising given that GSK3β is activated downstream of AKT and that JNK signaling does not appear to affect AKT activity in this context (). This suggests that JNK and GSK3β can regulate FoxO3a phosphorylation by an indirect mechanism or via an AKT-independent mechanism perhaps by regulating the activity of a phosphatase involved in FoxO3a dephosphorylation.
Although JNK and GSK3β were found to affect FoxO3a activation we cannot rule out the possibility that they may also regulate other transcription factors involved in Puma induction. A candidate factor downstream of GSK3β is nuclear factor of activated T cells (NFAT) which has been shown to be phosphorylated by GSK3β resulting in its export from the nucleus and promotion of survival in CGNs 
. In this case NFAT may act as a repressor of Puma transcription which is removed upon GSK3β activation. Similarly, beta-catenin may be acting to suppress Puma induction until inactivated by GSK3β. Phosphorylation of beta-catenin by GSK3β causes its translocation out of the nucleus and targets it for degradation and inhibition of this phosphorylation event has been associated with neuronal survival 
. Finally, there are several downstream targets of the JNK pathway which could control Puma expression following JNK activation, these include c-Jun, activating transcription factor 2 (ATF2) and activating transcription factor 3 (ATF3). A main downstream target of JNK, c-Jun has been found to be upregulated in trophic factor deprived neurons and ectopic expression of dominant negative c-Jun was found to protect against cell death 
. The JNK regulated transcription factors ATF2 and ATF3 are also induced in response to potassium deprivation and it has been reported that knockdown or inhibition of these factors can protect neurons against apoptosis 
. It is noteworthy that the Puma promoter contains putative AP1 binding sites which are the known target sequence for all three of these transcription factors, suggesting a potential role for these factors in Puma induction. Interestingly, a recent study implicated c-Jun in the regulation of Puma expression in fatty acid induced apoptosis of hepatocytes 
, although the AP-1 binding site identified in this study does not appear to be conserved. While these transcription factors have been implicated in neuronal apoptosis it is unclear whether or not they play a role in Puma upregulation in this context and is currently under investigation.
In summary, we have delineated a key pathway involved in the regulation of apoptosis induced by potassium deprivation in CGNs. Cell death in this paradigm results from the loss of activity dependent survival signals which is believed to mimic aspects of synaptic dysfunction common to many neuronal injury and neurodegenerative conditions. Therefore, in future studies it will be important to investigate the role of this pathway in in vivo models of neuronal injury and neurodegenerative diseases and to explore the therapeutic potential of targeting this pathway.