- Top of page
- Experimental procedures
We investigated the role of mitogen-activated protein kinase (MAPK) pathways in hypoxic neuronal injury using primary cultures from murine cerebral cortex. Hypoxia caused the death of ∼50% of neurons at 16 h and ∼65% of neurons at 24 h. This was associated with phospho-activation of the MAPK/extracellular signal-regulated kinase (ERK) kinase MEK1/2 and its downstream target ERK1/2, but not p38 MAPK or c-Jun N-terminal kinase (JNK), as detected by western blotting. The MEK1/2 inhibitor, PD98059, increased neuronal death in hypoxic cultures, suggesting that MEK1/2 promotes neuronal survival, whereas the p38 inhibitors, SB202190 and SB203580, had no effect. To identify downstream effects of ERK1/2 that might regulate hypoxic neuronal death, we measured hypoxia-induced phosphorylation of three ERK1/2 targets: the 90-kDa ribosomal protein S6 kinase (RSK), the transcription factor ELK1, and the pro-apoptotic Bcl-2 family protein Bad. We observed increased abundance of inactivated (phospho-)Bad, but no change in phospho-RSK or phospho-ELK1. Moreover, the MEK inhibitor PD98059 reduced phospho-inactivation of Bad in hypoxic cultures. These findings suggest that a cell-survival program involving phospho-activation of MEK1/2 and ERK1/2 and inactivation of Bad is mobilized in hypoxic neurons, and may help to regulate neuronal fate following hypoxic-ischemic injury.
Mitogen-activated protein kinases (MAPKs) transduce extracellular signals from tyrosine-kinase receptors and G protein-coupled receptors to cytoplasmic and nuclear effectors (Schaeffer and Weber 1999; Chang and Karin 2001). MAPK signaling pathways consist of sequentially acting serine-threonine protein kinases designated MAPK kinase kinases, MAPK kinases and MAPKs (Davis 2000). Multiple MAPK signaling modules have been identified in mammalian cells, including neurons (Sweatt 2001). These include the extracellular signal-regulated kinase (ERK) cascade, which is activated by growth factors and mitogens and stimulates cell growth and differentiation, and the c-Jun N-terminal kinase (JNK) and p38 cascades, which respond to inflammatory cytokines and cellular stress to promote inflammation and cell death (Schaeffer and Weber 1999).
Signaling through ERK1/2 involves the receptor-mediated activation of MAPK kinase kinases such as Raf or Mos, which activate MAPK/ERK kinase (MEK1/2); MEK1/2, in␣turn, activates ERK1/2. Downstream targets of ERK1/2 include the 90-kDa ribosomal protein S6 kinase (RSK), Ets-like transcription factor-1 (ELK1), and the pro-apoptotic Bcl-2 family protein Bad (Schaeffer and Weber 1999). Phosphorylation of Bad on Ser112 dissociates Bcl-2/Bad heterodimers and unmasks the anti-apoptotic effect of Bcl-2 (Bonni et al. 1999; Scheid et al. 1999).
Within the nervous system, MAPK pathways have been implicated in diverse physiological functions, including synaptic plasticity, gene expression and ion-channel activation (Sweatt 2001). In addition, MAPK signaling regulates excitotoxic neuronal cell death in vitro. Stimulation of N-methyl-d-aspartate receptors on cultured rat cerebellar granule neurons activates cell-death mechanisms that appear to involve p38, since cell death is attenuated by the p38 inhibitor, SB205380 (Kawasaki et al. 1997). In cultured rat hippocampal neurons, N-methyl-d-aspartate activates ERK, JNK and p38, and blocking activation reduces cell death (Mukherjee et al. 1999). In contrast, in cultured cerebral cortical neurons, MEK/ERK signaling has been implicated in the neuroprotective effect of ischemic tolerance (Gonzalez-Zulueta et al. 2000), in which brief ischemia protects cells against subsequent prolonged ischemia (Chen and Simon 1997). The capacity of MAPK pathways to exert opposite effects on cell survival is well documented in myocardial ischemia, where transient activation of p38 is associated with ischemic tolerance, while its sustained activation contributes to cell death (Mackay and Mochly-Rosen 1999).
Activation of MAPK pathways has also been studied in in␣vivo rodent models of global and focal cerebral ischemia. Global ischemia activated p38 in microglial cells adjacent to selectively vulnerable CA1 hippocampal neurons in one study (Walton et al. 1998), while another report showed transient activation of p38 and JNK in CA1 and of ERK2 in CA3, followed by more persistent activation of p38 and JNK in CA3 (Sugino et al. 2000). The p38 inhibitor SB203580 reduced cell death in CA1, suggesting that p38 promoted ischemic neuronal death in this model.
Other studies have addressed the role of MAPK signaling in focal cerebral ischemia. In the rat, middle cerebral artery occlusion increased activation of p38 in astrocytes within the ischemic core and penumbra, whereas ERK1/2 activation was increased in penumbral neurons and oligodendrocytes (Irving et al. 2000). Since these latter cells survived the ischemic insult, this was interpreted as evidence for a neuroprotective effect of ERK1/2. A similar study showed activation of JNK in cortical and striatal neurons within the ischemic core, favoring a death-promoting role (Hayashi et al. 2000). Finally, focal cerebral ischemia in the mouse led to increased expression of phospho-ERK1/2, but no change in phosphorylation of p38 or of the JNK substrate c-Jun (Alessandrini et al. 1999). Moreover, the MEK1/2 inhibitor PD98059, which blocks activation of ERK1/2, reduced infarct size by as much as 55%. Thus, in this system, signaling through MEK1/2 and ERK1/2 appeared to potentiate rather than reduce ischemic injury.
The complexity of MAPK signal transduction pathways, the fact that activation can elicit different responses in different cells (Schaeffer and Weber 1999) and under different conditions (Mackay and Mochly-Rosen 1999), and the presence in brain of numerous cell types with differential sensitivity to hypoxic-ischemic insults, suggest that studies in an isolated neuronal system might help to clarify the role of MAPK signaling in neuronal responses to hypoxia or ischemia. Therefore, we examined the effect of hypoxia on the activation of MAPK signaling intermediates in primary cultures of murine cerebral cortical neurons. The results suggest that hypoxic activation of a signaling module involving MEK1/2 and ERK1/2 leads to inactivation of the pro-apoptotic Bcl-2 family member Bad, and thereby promotes cell survival.
- Top of page
- Experimental procedures
ERK, which plays a critical role in the regulation of cell growth and differentiation and is activated by a wide variety of extracellular signals, has also been implicated in regulating cell survival and death after ischemia. To investigate whether MAPK pathways are activated by hypoxia in primary cultures of cerebral cortical neurons, cultures were maintained without oxygen for up to 24 h and western blotting was performed with an antibody against activated (phospho-)ERK. As shown in the third panel of Fig. 1, phosphorylated ERK1 (Mr∼44 kDa) and ERK2 (Mr∼42 kDa) were present at low levels in normoxic neuronal cultures, and exposure to hypoxia for 4–24 h caused a progressive increase in phospho-ERK1/2 immunoreactivity.
Figure 1. Western blot analysis of MAPK pathway phospho-proteins after hypoxia. Cell lysates (100 µg of protein) from cultured mouse cortical neurons, maintained under normoxic conditions (0 h) or under hypoxic conditions for 4, 8, 16 or 24 h, were loaded on 12% SDS–PAGE and analyzed by immunoblotting using antibodies against phospho-Raf, phospho-MEK1/2, phospho-ERK1/2 (p44/p42 MAPK), phospho-ELK1 and phospho-RSK. Blots shown are each representative of at least three independent experiments.
Download figure to PowerPoint
Activation of ERK1/2 occurs through phosphorylation of threonine and tyrosine by MEK1/2. To investigate whether phosphorylation of MEK1/2 is also increased in hypoxia, which could account for the enhanced activation of its downstream target, ERK 1/2, western blotting was performed using an antibody against activated (phospho-)MEK 1/2. As shown in Fig. 1, phospho-MEK1/2 was barely detectable in normoxic neurons, but its abundance increased beginning 4 h after hypoxia, with maximal activation at about 16 h. This is consistent with a role for MEK1/2 in the hypoxia-induced activation of neuronal ERK1/2.
MAPK signaling modules diverge to activate multiple downstream effectors, but also converge upon shared targets. One important target of various MAPKs, including ERKs, is␣the ternary complex transcription factor ELK1. ELK1 is phosphorylated by ERK or JNK, acting on the S/T(P) motif in the transcription domain, which permits ELK1 to bind to a serum response element and activate transcription (Li et al. 2000). However, we observed no hypoxia-induced increase in the activation of ELK1 by western blotting (Fig. 1). The 90-kDa ribosomal protein S6 kinase (RSK), a cytosolic protein, is also a substrate for ERK. Activated RSK regulates gene expression and protein synthesis by phosphorylating transcription factors and polyribosomal proteins, and also provides feedback inhibition of the ERK pathway (Frodin and Gammeltoft 1999). Nevertheless, as in the case of ELK1, we did not detect activation of RSK after hypoxia. Therefore, neither of these substrates appears to mediate the downstream effects of hypoxic activation of ERK1/2 in our cultures. Instead, these effects may involve one or more of the other transcription factors that are regulated by ERK1/2, such as Ets1, Sap1, c-Myc, Ta1, STAT, Myb or c-Jun (Treisman 1996).
Raf-1 is a serine-threonine protein kinase that activates MEK1/2 (Kyriakis et al. 1992). Several such MEK1/2 activators exist (Schaeffer and Weber 1999), including A-Raf, B-Raf, Mos and tumor progression locus (TPL) 2, but we examined the possible involvement of Raf-1 because it is one of the MEK kinases that is prominently expressed in neurons (Morice et al. 1999). Activation of Raf-1 was assessed by western blotting using an antibody against activated (phospho-)Raf. However, as shown in Fig. 1, in contrast to the activation observed for MEK1/2, phospho-Raf immunoreactivity decreased following ischemia.
In addition to ERK1/2, at least two other MAPKs – JNK and p38 – have been implicated in cellular stress responses. To examine the specificity of induction of the MEK/ERK signaling module in our neuronal hypoxia model, we also performed western blotting with antibodies against activated (phospho)-JNK and phospho-p38 (Fig. 2). Neither was induced in cultured cortical neurons after hypoxia, although hypoxic activation of JNK could be demonstrated in HN33 (not shown), a mouse hippocampal neuron x neuroblastoma hybrid cell line (Lee et al. 1990). Thus, the spectrum of MAPK signaling pathways that are induced by hypoxia appears to vary across cell types.
Figure 2. Western blot analysis of phospho-JNK and phospho-p38 MAPK after hypoxia. Cell lysates (100 µg of protein) from cultured mouse cortical neurons, maintained under normoxic conditions (0 h) or under hypoxic conditions for 4, 8, 16 or 24 h, were loaded on 12% SDS–PAGE and analyzed by immunoblotting using antibodies against (a) phospho-JNK and (b) phospho-p38. In contrast to the negative result shown in (b), the phospho-p38 antibody was capable of detecting phospho-activation of p38 in PC12 cell cultures treated with 25 ng/mL of nerve growth factor (NGF) for 9 days (c). Blots shown are each representative of three independent experiments.
Download figure to PowerPoint
To test the hypothesis that activation of MEK1/2 helps to attenuate hypoxic neuronal injury, we examined the effects of MAPK inhibitors on neuronal survival following hypoxia. The compounds used were the MEK1/2 inhibitor PD98059, the p38 inhibitors SB202190 and SB203580, and the inactive SB202190/SB203580 analog, SB202474. As shown in Fig. 3, hypoxia reduced cell viability to ∼50% of control levels after 16 h and to ∼35% after 24 h, as measured by MTT absorbance. Of the inhibitors tested, only PD98059 (50 µm) altered this outcome, further reducing viability to ∼35% at 16 h and to ∼25% at 24 h. The effective concentration of PD98059 is comparable to IC50 values of 10–60 µm for inhibition of MEK activity and of downstream effects of MEK in other systems (Dudley et al. 1995). A␣qualitatively similar result was observed when cell␣viability was measured using calcein-AM, although the magnitude of the incremental effect of PD98059 appeared to be greater in this case. This is probably because different assays of cell viability measure different things (such as reducing activity for MTT and esterase activity for calcein-AM), and may vary in sensitivity depending on the nature of the insult being studied.
Figure 3. Neuronal viability after 16 h or 24 h of hypoxia, in the absence and presence of the MEK1/2 inhibitor, PD98059. Viability of cultured mouse cortical neurons (% control viability in normoxic cultures) was measured by (a) MTT assay or (b) staining of viable cells (green) with calcein-AM and nonviable cells (red) with ethidium homodimer (see Experimental procedures). Data shown in (a) are mean ± SE from three independent experiments; *signifies that viability at 50 µm PD98059 is different from that in the absence of PD98059 (p < 0.05 by anova followed by post hoc t-test), whereas there was no statistically significant effect of any concentration of PD98059 in the 0 h hypoxia condition (p = 0.150 by anova). Images in (b) are each representative of at least three separate experiments. The p38 inhibitors SB202190 and SB203580 and the inactive SB202190/SB203580 analog, SB202474, had no significant effect on cell viability at either 16 h or 24 h (not shown).
Download figure to PowerPoint
The Bcl-2 family protein Bad is considered a cell-death promoter because it can bind to Bcl-2 and Bcl-xL and inhibit their cytoprotective effects (Zha et al. 1996). MEK1/2-dependent phosphorylation of Bad on Ser112 reduces its ability to form heterodimers with Bcl-2 or Bcl-xL, thereby promoting cell survival (Bonni et al. 1999; Scheid et al. 1999). To investigate the possible role of Bad phosphorylation in the protective effect of MEK1/2, cultures were exposed to hypoxia for 0–24 h and western blotting was performed using an antibody specific for Ser112 phospho-Bad. Hypoxia increased the abundance of Ser112 phospho-Bad (Fig. 4) without altering the overall expression of Bad (not shown), consistent with hypoxic induction of Bad phosphorylation. Since phosphorylation of Bad on Ser112 is thought to require MEK (Scheid et al. 1999), there might be a cause-and-effect relationship between hypoxic activation of MEK1/2 (Fig. 1) and Ser112 phosphorylation of Bad (Fig. 4). If so, and considering that Ser112 phosphorylation of Bad nullifies its pro-apoptotic effect, this could account for the ability of the MEK1/2 inhibitor PD98059 to potentiate ischemic neuronal death. To test this directly, cultures were exposed to hypoxia for 8 h with inhibitors of MEK1/2 or p38, and western blots were probed with the Ser112 phospho-Bad antibody. Figure 4 shows that the MEK1/2 inhibitor PD98059 reduced phosphorylation of Bad on Ser112, compared with phosphorylation in the presence of the p38 inhibitors SB202190 and SB203580 and the inactive analog SB202474.
Figure 4. Western blot analysis of Bad and phospho-Bad expression after hypoxia. (a) Cell lysates (100 µg of protein) from cultured mouse cortical neurons, maintained under normoxic conditions (0) or under hypoxic conditions for 4, 8, 16 or 24 h, were loaded on 12% SDS–polyacrylamide gels and analyzed by immunoblotting using an antibody against Bad or Ser112 phospho-Bad (pBad). For pBad experiments, cell lysates were immunoprecipitated with the same anti-pBad antibody used for western blotting. The blots shown are each representative of 3–4 independent experiments. (b) Phospho-Bad was quantified by computer densitometry and plotted as a percentage increase in optical density over that in normoxic cultures (0). Data shown are mean ± SE from four independent experiments.␣(c)␣Cultures were exposed to hypoxia for 8 h with the MEK1/2␣inhibitor PD98059, the p38 inhibitors SB203580 or SB202190, or the inactive SB202190/SB203580 analog SB202474, and western blots were probed with the same Ser112 phospho-Bad antibody used in (a). The blot shown is representative of three independent experiments.
Download figure to PowerPoint
- Top of page
- Experimental procedures
Hypoxic or ischemic neuronal injury triggers transcriptional and post-transcriptional processes that regulate neuronal fate. These include the activation of MAPK signaling pathways by neurotrophins and neurotransmitters (Han and Holtzman 2000) and the induction or activation of pro- and anti-apoptotic proteins, including those of the Bcl-2 family (Chen et al. 1995). The manner in which these two phases of hypoxia signaling are coupled is of considerable interest, since molecular mediators of hypoxic-ischemic injury might serve as targets for therapy. In addition, there may be an advantage to targeting early steps of cell-death cascades, since injury may be most readily reversible at this stage.
Among the MAPKs, ERK has been widely associated with cell survival, while JNK and p38 are often implicated in cell death (Chang and Karin 2001). In particular, most studies on the role of ERKs in neuronal hypoxia or ischemia suggest a neuroprotective role. ERKs are preferentially activated in the resistant CA3 compared with the vulnerable CA1 hippocampal sector during global cerebral ischemia (Sugino et al. 2000) and in the ischemic penumbra compared with the ischemic core after focal cerebral ischemia (Irving et al. 2000); these differences in expression could contribute to the differential survival of the cell populations in question. In␣addition, ERK1/2 mediates the protective effect of agents like brain-derived neurotrophic factor in hypoxic-ischemic brain injury (Han and Holtzman 2000).
The major finding of this study is that in cortical neuron cultures, hypoxia induces the activation of a MEK/ERK signaling pathway that appears to limit the extent of hypoxic injury and to involve downstream phospho-inactivation of the pro-apoptotic protein Bad. Consistent with this conclusion, the MEK1/2 inhibitor PD98059 increased hypoxic neuronal death and reduced the phospho-inactivation of Bad. What initiates the MEK/ERK signaling cascade in hypoxic neurons is unclear, but possible triggering factors include neurotransmitters like glutamate and growth factors like brain-derived neurotrophic factor (Han and Holtzman 2000), which are known to stimulate the MEK/ERK pathway. Although several upstream activators (MAPK kinase kinases or MEK kinases) of MEK/ERK signaling have also been identified, we did not observe evidence for the involvement of Raf-1. This implies that another neuronal MAPK kinase kinase, such as B-Raf (Olah et al. 1991; Dugan et al. 1999; Morice et al. 1999; Grewal et al. 2000), may be the principal activator of MEK1/2 in our hypoxic neurons.
If ERK1/2 (Fig. 1) and Bad (Figs 4a and b) are maximally phosphorylated 8 h after hypoxia and are protective, why does cell death continue to increase between 16 h and 24 h (Fig. 3a)? First, the protective effects of ERK1/2 and Bad are probably exerted downstream of these signaling molecules, so protection is likely to be seen at later times (e.g. 16–24 h, Fig. 3a). Second, competing cell-death pathways may oppose the protective effects of ERK1/2 and Bad. As suggested by results shown in Fig. 3(a), the result is that hypoxia still causes cell death, but causes less cell death if the MEK/ERK pathway is intact.
ERK1/2 is incapable of directly phospho-inactivating Bad, because the site surrounding Ser112 of Bad lacks the consensus sequence required for phosphorylation by ERKs. Among known substrates of ERK1/2 that might be responsible for phospho-inactivating Bad, we saw no evidence that RSK was activated by hypoxia in our cultures. Additional studies will be required to determine which of several candidate protein kinase substrates of ERK1/2 – such as MAPK signal-integrating kinase 1, mitogen- and stress-activated protein kinase 1 or MAPK-activated protein kinase␣3 (Schaeffer and Weber 1999) – leads to phospho-inactivation of Bad in our system.
The ability of hypoxia-induced MAPK signaling to modify cell fate by altering the phosphorylation state of Bad, a Bcl-2 family protein implicated in apoptosis, suggests that hypoxia initiates a form of programmed cell death in our neuronal cultures. Although it is unclear how best to categorize hypoxic-ischemic neuronal death in in vivo animal models or human stroke (Colbourne et al. 1999; Love et al. 2000), Bcl-2 family proteins are involved both in␣vivo (Chen et al. 1995) and in vitro (Tamatani et al. 2000). Our results suggest that Bad may have an important role in this process.
In contrast to the finding that ERK1/2 is preferentially activated in neurons that survive global (Sugino et al. 2000) or focal (Irving et al. 2000) cerebral ischemia in vivo, as well as our own results, one study (involving 1–2 h of middle cerebral artery occlusion followed by 3 min of reperfusion in the mouse) reported a detrimental effect of ERK activation in murine focal ischemia (Alessandrini et al. 1999). This was based on the observation that infarct size was reduced by inhibiting the action of MEK with PD98059. As in our system, phosphorylation of ERK1/2 was induced by ischemia, whereas neither JNK nor p38 was activated. Hemodynamic and other systemic factors, such as cerebral blood flow and temperature, were unaffected by PD98059. The reason for the discrepancy between the cell death-promoting effect of MEK/ERK activation in this study and the protective effect that we and others (Irving et al. 2000; Sugino et al. 2000) have observed is unclear. However, it may relate to differences in the coupling of MAPK signaling pathways to downstream mechanisms regulating cell death in different brain regions, in different neurons within a given region, or in neurons of different ages, or to the presence of non-neuronal cell types or unidentified systemic factors in the in vivo ischemia model.