ERK1/2 Antagonizes Glycogen Synthase Kinase-3β-induced Apoptosis in Cortical Neurons*

Inhibition of glycogen synthase kinase-3β (GSK3β) is one of the mechanisms by which phosphatidylinositol 3-kinase (PI3K) activation protects neurons from apoptosis. Here, we report that inhibition of ERK1/2 increased the basal activity of GSK3β in cortical neurons and that both ERK1/2 and PI3K were required for brain-derived neurotrophic factor (BDNF) suppression of GSK3β activity. Moreover, cortical neuron apoptosis induced by expression of recombinant GSK3β was inhibited by coexpression of constitutively active MKK1 or PI3K. Activation of both endogenous ERK1/2 and PI3K signaling pathways was required for BDNF to block apoptosis induced by expression of recombinant GSK3β. Furthermore, cortical neuron apoptosis induced by LY294002-mediated activation of endogenous GSK3β was blocked by expression of constitutively active MKK1 or by BDNF via stimulation of the endogenous ERK1/2 pathway. Although both PI3K and ERK1/2 inhibited GSK3β activity, neither had an effect on GSK3β phosphorylation at Tyr-216. Interestingly, PI3K (but not ERK1/2) induced the inhibitory phosphorylation of GSK3β at Ser-9. Significantly, coexpression of constitutively active MKK1 (but not PI3K) still suppressed neuronal apoptosis induced by expression of the GSK3β(S9A) mutant. These data suggest that activation of the ERK1/2 signaling pathway protects neurons from GSK3β-induced apoptosis and that inhibition of GSK3β may be a common target by which ERK1/2 and PI3K protect neurons from apoptosis. Furthermore, ERK1/2 inhibits GSK3β activity via a novel mechanism that is independent of Ser-9 phosphorylation and likely does not involve Tyr-216 phosphorylation.

It has become increasingly evident that there is a complex balance between survival and apoptotic signaling pathways in neurons that determines whether they survive or die. For ex-ample, BDNF 1 activates the ERK1/2 and PI3K/Akt pathways in neurons and protects them from several forms of apoptosis, including those induced by DNA damage, microtubule damage, and trophic deprivation (1)(2)(3)(4)(5)(6). The relative contribution of the ERK1/2 and PI3K/Akt pathways to neuronal survival depends on the specific type of cellular injury (7). In contrast, activation of the stress-activated protein kinases, including the c-Jun NH 2 -terminal protein kinase and the p38 MAPK, contributes to apoptosis in cortical neurons (6,8,9) and other types of neurons (for a review, see Ref. 10).
Recently, GSK3␤ was discovered as another apoptosis-inducing kinase in the nervous system (5,11). Expression of recombinant wild-type GSK3␤ is sufficient to induce apoptosis in PC12 cells (11), primary cortical neurons (5), and sympathetic neurons (12). The basal activity of GSK3␤ in PC12 cells and cortical neurons is relatively high, but can be further activated by inhibition of PI3K/Akt signaling (5,11). Blocking GSK3␤ suppresses apoptosis induced by PI3K inhibition in PC12 cells, cortical neurons, cerebellar granule cells, and sympathetic neurons (5,(11)(12)(13). Although inhibition of GSK3␤ is not sufficient to inhibit sympathetic neuron apoptosis triggered by nerve growth factor withdrawal (12), it protects against trophic deprivation in PC12 cells, cortical neurons, and cerebellar granule cells (5,11,13,14). These data suggest that inhibition of GSK3␤ is one of the mechanisms by which PI3K activation protects neurons from apoptosis.
In this report, we present data supporting the hypothesis that ERK1/2 protects neurons from GSK3␤-induced apoptosis and that inhibition of GSK3␤ may be a common mechanism by which ERK1/2 and PI3K protect neurons from apoptosis. However, ERK1/2 inhibits GSK3␤ activity in cortical neurons via a novel mechanism independent of Ser-9 phosphorylation.
Cell Culture and Transfection-Cortical neurons were prepared from newborn Sprague-Dawley rats and transiently transfected at day 3 or 4 in vitro (DIV3/4) after seeding using a calcium phosphate coprecipitation protocol (2). In Fig. 7, cortical neurons were transfected with LipofectAMINE 2000 (Invitrogen). Briefly, cells were seeded at 500,000/ well in 24-well plates. At DIV3-4, the conditioned medium were removed and saved. Cells were placed in serum-free basal Eagle's medium (Sigma) containing 0.8 g of DNA mixed with 1.5 l of LipofectAMINE 2000/well. After a 2-h incubation at 5% CO 2 and 37°C, the transfection medium were replaced with conditioned medium. Cells were fixed and immunostained 2 days after transfection.
Drug Treatment-PD98059, LY294002, and SL327 were dissolved in Me 2 SO, and Me 2 SO was used as a vehicle control for these drugs. The final concentration of Me 2 SO was 0.2%. When cultures were cotreated with PD98059 and LY294002 or with SL327 and LY294002, the final concentration of Me 2 SO was 0.4%. BDNF was diluted in phosphatebuffered saline containing 0.1% bovine serum albumin before addition to the cells.
Quantitation of Apoptosis by Nuclear Morphological Changes-To visualize nuclear morphology, cells were fixed in 4% paraformaldehyde and stained with the DNA dye Hoechst 33258 (bisbenzimide, Sigma) at 2.5 g/ml (2). Apoptosis was quantitated by scoring the percentage of cells with apoptotic nuclear morphology at the single cell level after Hoechst staining. Uniformly stained nuclei were scored as healthy, viable neurons. Condensed or fragmented nuclei were scored as apoptotic. To obtain unbiased counting, slides were coded, and cells were scored blind without knowledge of their prior treatment.
Western Analysis and Immunostaining-Western blot analysis using anti-phospho-ERK1/2, anti-phospho-Akt, and anti-phospho-GSK3␤ antibodies and immunostaining for ␤-galactosidase were performed as described (2,28,29). Transfected cells were detected by immunostaining with an antibody against ␤-galactosidase and Texas Red-conjugated goat antibody to rabbit IgG. Cells transfected with the hemagglutinin epitope-tagged constructs (MKK1 and PI3K) were also immunostained with an antibody to hemagglutinin followed by fluorescein-conjugated goat antibody to mouse IgG.
Mutagenesis-Site-directed mutagenesis to replace Ser-9 of rat GSK3␤ with alanine was performed using a QuikChange kit (Stratagene) according to the manufacturer's recommendations. The mutagenesis primers were 5Ј-ccgagaaccacggcctttgcggagagctgc-3Ј (sense) and 5Ј-gcagctctccgcaaaggccgtggttctcgg-3Ј (antisense). The primers were purchased from Sigma. The presence of the mutation was confirmed by DNA sequencing.
Statistical Analysis-Statistical analysis of the data was performed using one-or two-way analysis of variance (ANOVA), followed by post hoc tests.

GSK3␤ Activity Is Negatively Regulated by ERK1/2 in Corti-
cal Neurons-To test the hypothesis that ERK1/2 antagonizes GSK3␤-induced apoptosis in cortical neurons, we first determined whether GSK3␤ activity is negatively regulated by ERK1/2 in these cells. We previously showed that LY294002 treatment, which inhibits PI3K activity, activates GSK3␤ (5). Therefore, we used LY294002 treatment as a positive control for the assessment of GSK3␤ activity. To inhibit the ERK1/2 pathway, we applied PD98059, an inhibitor of MKK1 and MKK2 (30), which are upstream kinases that phosphorylate and activate ERK1/2. The effect of these inhibitors on ERK1/2 or PI3K activity was indirectly measured by Western analysis using antibodies that specifically recognize phosphorylated and activated ERK1/2 or Akt, respectively. The effect of these inhibitors on GSK3␤ activity was directly measured by an immune complex kinase assay.
Activation of GSK3␤ induced by trophic deprivation was counteracted by BDNF treatment (Fig. 2) (5). Because BDNF activates both ERK1/2 and PI3K in cortical neurons, we tested whether BDNF-driven ERK1/2 activation contributes to BDNF inhibition of GSK3␤. Cortical neurons were incubated with BDNF under trophic deprivation in the presence of various inhibitors (Fig. 2). Trophic deprivation was achieved by serum withdrawal in the presence of the N-methyl-D-aspartate receptor antagonist MK801 (5). LY294002 completely blocked BDNF stimulation of Akt phosphorylation. PD98059 significantly inhibited BDNF stimulation of ERK1/2 without affecting Akt phosphorylation ( Fig. 2A). However, PD98059 and LY294002 together caused a greater inhibition of ERK1/2 than PD98059 alone. This is consistent with the observation that some isoforms of neuronal PI3K activate the ERK1/2 pathway (31). The effects of these inhibitors on BDNF suppression of GSK3␤ activity were also examined (Fig. 2B). Although PD98059 or LY294002 caused some increase in GSK3␤ activity, the combination of the two caused a significant increase in GSK3␤ activity compared with either drug alone. This suggests that ERK1/2 and PI3K signaling contributes to BDNF suppression of GSK3␤.
Cortical Neuron Apoptosis Induced by Expression of Recombinant GSK3␤ Is Inhibited by Activation of Both ERK1/2 and PI3K Signaling Pathways-Expression of wild-type GSK3␤ is sufficient to induce apoptosis in cortical neurons (5). To elucidate the functional significance of ERK1/2 inhibition of GSK3␤, we determined whether apoptosis induced by expression of wild-type GSK3␤ is blocked by coexpression of constitutively active MKK1 (MKK1CA), which selectively activates ERK1/2 (Fig. 3A). Cortical neurons were cotransfected with a wild-type expression vector for GSK3␤ and MKK1CA. The empty cloning vectors and kinase-dead dominant-negative MKK1 (MKK1KIN) were used as controls. Basal cell death in cells transfected with vectors only was 16%. Expression of GSK3␤ increased apoptosis to 34%. Coexpression of constitutively active MKK1 (but not dominant-negative MKK1) reduced GSK3␤-induced apoptosis to 23% (p Ͻ 0.01). Similarly, GSK3␤induced apoptosis was partially suppressed when the PI3K pathway was selectively activated by transient expression of an active form of the catalytic subunit of PI3K (p110*), but not a kinase-dead mutant of p110 (p110*KIN) (Fig. 3B).
To determine whether stimulation of endogenous ERK1/2 is sufficient to suppress cortical neuron apoptosis due to expression of recombinant GSK3␤, cortical neurons were treated with BDNF to activate endogenous ERK1/2 and PI3K (Fig. 4). BDNF completely inhibited GSK3␤-induced cortical neuron apoptosis. Neuroprotection afforded by BDNF was inhibited by cotreatment with LY294002, PD98059 (Fig. 4A), or SL327 (Fig.  4B). SL327 is a water-soluble structural homolog of the specific MKK1/2 inhibitor U0126 (32,33) and has been widely used to study neuronal signaling mechanisms (34 -40). The data in Fig. 4 5). Cotreatment with LY294002 and PD98059 or SL327 greatly exacerbated GSK3␤ expression-induced apoptosis, which was now seen in 80% of the GSK3␤-expressing cells (Fig. 5). This suggests that under normal culture conditions, apoptosis induced by expression of wild-type GSK3␤ is attenuated by the endogenous basal ERK1/2 and PI3K activities. As shown earlier, addition of BDNF completely suppressed apoptosis induced by expression of GSK3␤. Neuroprotection provided by BDNF was totally reversed by cotreatment with PD98059 and LY294002 (Fig. 5A) or with SL327 and LY294002 (Fig. 5B). These data suggest that A, 20 g of cell lysates were submitted to Western analysis using antibodies recognizing phosphorylated and activated ERK1/2 (p-ERK1, p-ERK2) or Akt (p-Akt; at Ser-473), indicative of activation of the ERK1/2 or PI3K pathway, respectively. To ensure that the loading and transfer were equal, blots were reprobed to visualize the levels of either total ERK2 or Akt (data not shown). B, endogenous GSK3␤ was immunoprecipitated using the anti-GSK3␤ polyclonal antibody from 150 g of cell lysates and assayed for kinase activity using the phosphorylated CREB peptide KRREILSRRPS(P)YR as the substrate. The percentage increase in GSK3␤ activity is relative to vehicle control-treated neurons (C). Data are from four independent experiments of triplicate determinations. Error bars are S.E. **, p Ͻ 0.01; ***, p Ͻ 0.001 (ANOVA) compared with vehicle control-treated cells. C, endogenous GSK3␤ was immunoprecipitated (IP) as described for B and analyzed by Western blotting (IB) using the mouse anti-GSK3␤/␣ monoclonal antibody. A crude cell lysate (Lysate) was included as a positive control for detection of both GSK3␤ and GSK3␣ isoforms in the cortical neuron lysates. As evident from the data, the anti-GSK3␤ polyclonal antibody used for GSK3␤ kinase assay pulls down primarily the GSK3␤ isoform, and an equal amount of endogenous GSK3␤ was immunoprecipitated from each sample. D, 10 g of cell lysates were submitted to Western analysis using the anti-phospho-Ser-9 GSK3␤ antibody (p-S 9 -GSK3␤) (upper panel) or the anti-phospho-Tyr-279 GSK3␣/Tyr-216 GSK3␤ antibody (p-Y 279 GSK3␣, p-Y 216 GSK3␤) (middle panel). To verify equal amounts of total GSK3␤/␣ loading, the blots were stripped and reprobed with the anti-GSK3␤/␣ antibody (lower panel). E, GSK3␤ phosphorylation at Ser-9 was quantitated by densitometric analysis of the anti-phospho-Ser-9 GSK3␤ antibody Western blots. Data are from three independent experiments. Error bars are S.E. ns, not significant; **, p Ͻ 0.01 (ANOVA) compared with vehicle control-treated cells (C). coactivation of ERK1/2 and PI3K provides combinatorial neuroprotection against GSK3␤.
Apoptosis Induced by Activation of Endogenous GSK3␤ Is Blocked by ERK1/2-We previously reported that addition of LY294002 to cortical neuron cultures maintained in the presence of serum inhibits the endogenous PI3K/Akt pathway, activates endogenous GSK3␤, and induces apoptosis in cortical neurons (5). Cortical neuron apoptosis induced by LY294002 was significantly blocked by expression of a dominant-negative mutant form of rat GSK3␤ or inhibitory GBP (p Ͻ 0.001) (Fig.  6A). This confirms that endogenous GSK3␤ plays a significant role in apoptosis induced by LY294002. To determine whether apoptosis induced by activation of endogenous GSK3␤ is blocked by ERK1/2, cortical neurons were treated with LY294002 in the presence of SL327, BDNF, or SL327 plus BDNF (Fig. 6B). LY294002-induced apoptosis was prevented by cotreatment with BDNF (p Ͻ 0.001). Because LY294002 directly inhibited PI3K, and BDNF did not activate PI3K in the presence of LY294002 ( Fig. 2A), BDNF protection against LY294002 is very likely mediated by ERK1/2 signaling. Moreover, BDNF protection against LY294002 was reversed by cotreatment with SL327 (p Ͻ 0.001) (Fig. 6B), further supporting a role for endogenous ERK1/2 in neuroprotection against activation of GSK3␤.
Cortical neurons were also transfected with MKK1CA or its vector control and treated with LY294002. Expression of MKK1CA was sufficient to inhibit cortical neuron apoptosis induced by LY294002 treatment (p Ͻ 0.01) (Fig. 6C). Together, these data suggest that apoptosis induced by activation of endogenous GSK3␤ is blocked by ERK1/2 signaling.
GSK3␤ Phosphorylation at Ser-9 or Tyr-216 Is Not Affected by Inhibition of ERK1/2 in Cortical Neurons-GSK3␤ activity is negatively regulated by phosphorylation at Ser-9. To elucidate mechanisms by which PI3K and ERK1/2 inhibit GSK3␤ activity and GSK3␤-induced apoptosis in cortical neurons, we carried out experiments to determine whether these signaling pathways regulate GSK3␤ phosphorylation at Ser-9. This was accomplished by Western analysis using a phosphopeptide-

FIG. 2. ERK1/2 contributes to BDNF suppression of GSK3␤ activity in cortical neurons via a mechanism that is independent of GSK3␤ phosphorylation at Ser-9 or Tyr-216.
Cortical neurons (DIV5) were stimulated with BDNF (10 ng/ml) in serum-free and MK801 (10 M)-containing medium. The following inhibitors were also included as indicated: vehicle control (C), 40 M PD98059 (PD), 50 M SL327 (SL), or 30 M LY294002 (LY). Cell lysates were prepared after a 3-h treatment with BDNF and drugs. A, 20 g of cell lysates were submitted to Western analysis using antibodies recognizing phosphorylated and activated ERK1/2 (p-ERK1, p-ERK2) or Akt (p-Akt; at Ser-473), indicative of activation of the ERK1/2 or PI3K pathway, respectively. To ensure that the loading and transfer were equal, blots were reprobed to visualize the levels of either total ERK2 or Akt (data not shown). B, both ERK1/2 and PI3K were required for BDNF suppression of GSK3␤ activity. Endogenous GSK3␤ was immunoprecipitated from 150 g of cell lysates and assayed for kinase activity as described in the legend to Fig. 1. The percentage increase in GSK3␤ activity is relative to neurons treated with BDNF and a vehicle control for drugs. Data are from two independent experiments of duplicate determinations. Error bars are S.E. **, p Ͻ 0.01 (ANOVA). C, endogenous GSK3␤ was immunoprecipitated (IP) and analyzed by Western blotting (IB) using the mouse anti-GSK3␤/␣ monoclonal antibody as described in the legend to Fig. 1C to demonstrate the specificity of the anti-GSK3␤ polyclonal antibody used for GSK3␤ kinase assay and that a similar amount of endogenous GSK3␤ was immunoprecipitated from each sample. D, like serum, BDNF induced GSK3␤ phosphorylation at Ser-9. Cortical neurons were cultured in 10% serum (ϩS) or deprived of serum in MK801 (10 M)-containing media in the presence of 0 or 10 ng/ml BDNF as indicated. E, inhibition of ERK1/2 did not affect BDNF-induced GSK3␤ phosphorylation at Ser-9 and had no effect on GSK3␤ phosphorylation at Tyr-216 in cortical neurons. Western analysis using anti-phospho-Ser-9 GSK3␤, anti-phospho-Tyr-279 GSK3␣/phospho-Tyr-216 GSK3␤, or anti-GSK3␤/␣ was performed as described in the legend to Fig. 1. F, GSK3␤ phosphorylation at Ser-9 was quantitated by densitometric analysis of anti-phospho-Ser-9 GSK3␤ antibody Western blots. Data are from three independent experiments. Error bars are S.E. ns, not significant; ***, p Ͻ 0.001 (ANOVA) compared with vehicle control-treated cells (C). There was no statistically significant difference in samples treated with LY294002, LY294002 ϩ SL327, or LY294002 ϩ PD98059. specific antibody that recognizes GSK3␤ phosphorylated at Ser-9 ( Figs. 1 and 2). Under normal culture conditions (i.e. in the presence of serum), LY294002 treatment inhibited Ser-9 phosphorylation by 70% (p Ͻ 0.01) (Fig. 1, D and E). In contrast, PD98059 treatment caused a slight reduction in Ser-9 phosphorylation that was statistically insignificant (p Ͼ 0.05). Furthermore, cotreatment with PD98059 and LY294002 did not inhibit Ser-9 phosphorylation more than LY294002 treatment alone. Because both PD98059 and LY294002 treatment caused an increase in GSK3␤ activity under the same conditions (Fig. 1B), these data suggest that serum-activated PI3K (but not ERK1/2) inhibits GSK3␤ activity in primary cortical neurons via Ser-9 phosphorylation.
Another possible mechanism for ERK1/2 inhibition of GSK3␤ in neurons is through inhibition of GSK3␤ phosphorylation at Tyr-216, a phosphorylation that may activate GSK3␤ (14). However, there was no significant change in Tyr-216 phosphorylation following BDNF treatment, inhibition of ERK1/2, or inhibition of PI3K in cortical neurons (Figs. 1D and  2E). Thus, inhibition of GSK3␤ phosphorylation at Tyr-216 is an unlikely mechanism for ERK1/2 inhibition of GSK3␤ in neurons.

DISCUSSION
The objective of this study was to test the hypothesis that neuronal apoptosis induced by activation of GSK3␤ may be suppressed by stimulation of the ERK1/2 signaling pathway. We report that ERK1/2 negatively regulated GSK3␤ activity in cortical neurons and that this regulation did not involve the inhibitory phosphorylation of GSK3␤ at Ser-9. In addition, this regulation probably does not require the activating phosphorylation of GSK3␤ at Tyr-216. Furthermore, expression of a constitutively active MKK1, which directly activated ERK1/2, suppressed cortical neuron apoptosis induced by expression of recombinant wild-type GSK3␤ or the GSK3␤(S9A) mutant. MKK1 also suppressed cortical neuron apoptosis induced by activation of endogenous GSK3␤ caused by LY294002 inhibition of PI3K. Moreover, ERK1/2 contributed to BDNF neuroprotection against LY294002 or GSK3␤ expression. Maximal BDNF neuroprotection against GSK3␤ expression-induced apoptosis was dependent upon activation of both ERK1/2 and PI3K signaling pathways. Our data provide the first example of a functional consequence of ERK1/2 inhibition of GSK3␤.
Mechanisms for ERK1/2 inhibition of GSK3␤ in neurons are still unclear. Both Akt and protein kinase A have been shown to directly phosphorylate GSK3␤ at Ser-9 (13,15), an inhibitory phosphorylation site. Studies in non-neuronal cells suggest ERK1/2 activation of p90 rsk , which also phosphorylates GSK3␤ at Ser-9 (21,22). However, our data indicate that this mechanism is not operative in primary neurons. This conclusion is based on the observation that the level of GSK3␤ phosphorylation at Ser-9 was not significantly affected by inhibitors of the ERK1/2 pathway. This is in contrast to the profound effect of PI3K inhibitors on phosphorylation of GSK3␤ at Ser-9. In addition, MKK1 (but not PI3K) could still suppress GSK3␤(S9A)-induced apoptosis.
Our data do not support the hypothesis that ERK1/2 inhibition of GSK3␤ in neurons is mediated by inhibition of the activating GSK3␤ phosphorylation at Tyr-216. Alternatively, ERK1/2 may increase the expression or improve the function of the GSK3␤ inhibitory protein GBP. Regardless, it seems that in cortical neurons, ERK1/2 inhibits GSK3␤ activity by an unknown and novel mechanism, which most likely does not involve Ser-9 and Tyr-216 phosphorylation. We conclude that ERK1/2 may inhibit GSK3␤ by a mechanism that is distinct from that utilized by PI3K and protein kinase A. The existence of different regulatory mechanisms for ERK1/2 and PI3K inhi- bition of GSK3␤ may explain why cotreatment of neurons with inhibitors of both signaling pathways activates GSK3␤ more effectively than a single inhibitor alone.
GSK3␤ was first discovered as a pro-apoptotic signaling molecule in PC12 cells (11) and subsequently in primary cortical neurons (5) as well as in sympathetic neurons (12). GSK3␤induced apoptosis is suppressed by activation of the PI3K/Akt and cAMP/protein kinase A signaling pathways (5,(11)(12)(13). Here, we provide evidence that GSK3␤-induced apoptosis is also repressed by the ERK1/2 signaling pathway. Therefore, inhibition of GSK3␤ may be a point of convergence for several survival pathways. In this context, it is also interesting that activation of the PI3K, cAMP, and ERK1/2 signal transduction pathways results in BAD phosphorylation and inhibition of the pro-apoptotic activity of BAD (3,(41)(42)(43)(44)(45)(46). This suggests that seemingly divergent anti-apoptotic signaling pathways may converge on a small group of target molecules (including GSK3␤ and BAD) that are key regulators of neuronal apoptosis.
Since the initial discovery that stimulation of ERK1/2 protects differentiated PC12 cells from apoptosis after nerve growth factor withdrawal (29), additional evidence has accumulated demonstrating the anti-apoptotic function of ERK1/2 in various non-neuronal cells (47)(48)(49)(50)(51) and in neurons (3,52,53). For example, ERK1/2 activation protects central nervous system neurons against DNA (2) and oxidative (1,4) damage. Here, we have discovered that ERK1/2 antagonizes cortical neuron apoptosis induced by GSK3␤, adding another function to its anti-apoptotic repertoire. Furthermore, although trophic promoted neuronal survival is largely mediated by PI3K (2), the data presented here indicate that when PI3K activity is inhibited by LY294002, BDNF stimulation of ERK1/2 is sufficient to compensate for the loss of PI3K activity and to promote cortical neuron survival.
Similar to drug inhibition of PI3K, serum deprivation alone or in combination with the N-methyl-D-aspartate receptor antagonist MK801 also activates GSK3␤ and induces apoptosis in cortical neurons (5). However, ERK1/2 activation protects cortical neurons against LY294002, but not against serum depri-vation (2) or against serum deprivation plus MK801. 2 In addition, ERK1/2 activation does not play a major role in BDNF protection against serum deprivation (2). It seems likely that other pathways, in addition to PI3K and ERK1/2, may also contribute to the neuronal survival provided by serum or BDNF. These may include protein kinase C (54), ERK5 (55,56), and isoforms of p38 MAPKs (57). Although ERK1/2 alone does not promote survival against serum deprivation, which presumably down-regulates all these pathways, it is sufficient to antagonize LY294002-induced apoptosis, which specifically targets PI3K inhibition.
Mechanisms downstream of ERK1/2 activation that lead to neuronal survival are not clearly defined. Activation of p90 rsk and the subsequent phosphorylation of CREB as well as increased CREB-mediated gene expression are potential mechanisms (3,58). Alternatively, ERK1/2 may activate p90 rsk , which phosphorylates and inactivates BAD (3). Our discovery that activation of the ERK1/2 signaling pathway protects against GSK3␤-induced apoptosis suggests another novel downstream target for the ERK1/2 survival pathway.
Why are there multiple downstream effectors mediating the anti-apoptotic activity of ERK1/2? Activation of CREB-mediated transcription and inhibition of BAD or GSK3␤ may function as parallel downstream targets of the anti-apoptotic ERK1/2 pathway. Alternatively, ERK1/2 neuroprotection may employ different downstream targets depending on the nature of the apoptotic insult. For example, camptothecin, a DNAdamaging agent, induces cortical neuron apoptosis, which is blocked by ERK1/2 (2). However, expression of either GBP or dominant-negative GSK3␤ did not protect cortical neurons against camptothecin, 2 excluding GSK3␤ inhibition as a mechanism for ERK1/2 neuroprotection against camptothecin. Finally, there may be cross-talk between the various downstream targets of ERK1/2. For instance, inhibition of GSK3␤ may contribute to ERK1/2 activation of CREB (59).
In summary, our data suggest that ERK1/2 negatively regulates GSK3␤ activity in neurons via a novel mechanism that is independent of phosphorylation of GSK3␤ at Ser-9 and likely does not involve phosphorylation of GSK3␤ at Tyr-216. Furthermore, activation of ERK1/2 suppresses GSK3␤-induced neuronal apoptosis. Our data suggest a new mechanism for the anti-apoptotic action of the ERK1/2 pathway in neurons and implicate GSK3␤ inhibition as a convergence point for several pro-survival signal transduction systems in neurons. FIG. 7. Expression of the GSK3␤(S9A) mutant induces cortical neuron apoptosis that is suppressed by coexpression of constitutively active MKK1, but not PI3K. Cortical neurons (DIV3) were transfected using LipofectAMINE 2000 with a rat GSK3␤ or GSK3␤(S9A) cDNA (0.2 g of plasmid DNA/5 ϫ 10 5 cells/well in a 24-well plate) to induce apoptosis in transfected cells. The cloning vector pEF1␣ was used as a control. Cells were also cotransfected with 0.2 g of plasmid DNA encoding ␤-galactosidase as a marker for transfection and 0.4 g of plasmid DNA of either constitutively active MKK1 (MKK1CA) or constitutively active PI3K (p110*). Two days after transfection, cells were fixed and immunostained with the anti-␤-galactosidase antibody to identify transfected neurons. Apoptosis in the transfected cell population (␤-galactosidase-positive) was scored. Data are representative of four independent experiments. At least 700 -800 transfected cells were counted for each data point. Error bars are S.E. ns, not significant; ***, p Ͻ 0.001 (ANOVA). GSK3␤wt, wild-type GSK3␤.