Opposing Roles for ERK1/2 in Neuronal Oxidative Toxicity

Glutamate-induced oxidative toxicity is mediated by glutathione depletion in the HT22 mouse hippocampal cell line. Previous results with pharmacological agents implicated the extracellular signal-regulated kinases-1/2 (ERK1/2) in glutamate toxicity in HT22 cells and immature embryonic rat cortical neurons. In this report, we definitively establish a role for ERK1/2 in oxidative toxicity using dominant negative MEK1 expression in transiently transfected HT22 cells to block glutamate-induced cell death. In contrast, chronic activation of ERK (i.e. brought about by transfection of constitutively active ERK2 chimera) is not sufficient to trigger HT22 cell death demonstrating that ERK1/2 activation is not sufficient for toxicity. Activation of ERK1/2 in HT22 cells has a distinct kinetic profile with an initial peak occurring between 30 min and 1 h of glutamate treatment and a second peak typically emerging after 6 h. We demonstrate here that the initial phase of ERK1/2 induction is because of activation of metabotropic glutamate receptor type I (mGluRI). ERK1/2 activation by mGluRI contributes to an HT22 cell adaptive response to oxidative stress as glutamate-induced toxicity is enhanced upon pharmacological inhibition of mGluRI. The protective effect of ERK1/2 activation at early times after glutamate treatment is mediated by a restoration of glutathione (GSH) levels that are reduced because of depletion of intracellular cysteine pools. Thus, ERK1/2 appears to play dual roles in HT22 cells acting as part of a cellular adaptive response during the initial phases of glutamate-induced oxidative stress and contributing to toxicity during later stages of stress.

Glutamate-induced oxidative toxicity is mediated by glutathione depletion in the HT22 mouse hippocampal cell line. Previous results with pharmacological agents implicated the extracellular signal-regulated kinases-1/2 (ERK1/2) in glutamate toxicity in HT22 cells and immature embryonic rat cortical neurons. In this report, we definitively establish a role for ERK1/2 in oxidative toxicity using dominant negative MEK1 expression in transiently transfected HT22 cells to block glutamate-induced cell death. In contrast, chronic activation of ERK (i.e. brought about by transfection of constitutively active ERK2 chimera) is not sufficient to trigger HT22 cell death demonstrating that ERK1/2 activation is not sufficient for toxicity. Activation of ERK1/2 in HT22 cells has a distinct kinetic profile with an initial peak occurring between 30 min and 1 h of glutamate treatment and a second peak typically emerging after 6 h. We demonstrate here that the initial phase of ERK1/2 induction is because of activation of metabotropic glutamate receptor type I (mGluRI). ERK1/2 activation by mGluRI contributes to an HT22 cell adaptive response to oxidative stress as glutamate-induced toxicity is enhanced upon pharmacological inhibition of mGluRI. The protective effect of ERK1/2 activation at early times after glutamate treatment is mediated by a restoration of glutathione (GSH) levels that are reduced because of depletion of intracellular cysteine pools. Thus, ERK1/2 appears to play dual roles in HT22 cells acting as part of a cellular adaptive response during the initial phases of glutamate-induced oxidative stress and contributing to toxicity during later stages of stress.
Oxidative stress can contribute to neuronal toxicity and has been implicated in both acute injury and chronic neuropathological conditions (1,2). Many in vitro models have been used to examine the mechanistic basis for neuronal cell death induced by oxidative stress. For example, oxidative toxicity can be induced by glutamate treatment in the HT22 mouse hippocampal cell line (3)(4)(5) and immature primary embryonic rat cortical neuron cultures (4,6,7). In these models, glutamate treatment leads to glutathione (GSH) 2 depletion and subsequent accumulation of reactive oxygen species (ROS) (8). Many intracellular second messenger pathways are required for oxidative toxicity in HT22 cells including arachidonic acid metabolites, cyclic GMP (cGMP), and calcium (8). In addition, signaling kinases, such as mitogen-activated protein kinases (MAPK), are activated upon glutamate-induced oxidative stress in HT22 cells and primary neurons and are likely to affect targets that either limit or promote oxidative toxicity.
Extracellular signal-regulated kinases-1/2 (ERK1/2) has been implicated in glutamate-induced neuronal oxidative toxicity based upon the neuroprotective effects of U0126, a specific inhibitor of the ERK1/2activating kinase, MEK1/2 (5). U0126 is also effective at reducing brain injury following focal ischemia in rodents suggesting that ERK1/2 may also promote neuronal cell death resulting from acute injury in vivo (9). However, the role of ERK1/2 in neuronal cell death remains controversial (10 -12). In a number of studies, ERK1/2 has been found to promote neuronal survival and reduce cell death induced by various insults (13)(14)(15). Furthermore, even in HT22 cells, ERK1/2 may limit toxicity in response to specific insults such as serum withdrawal (16).
Distinct kinetic profiles of ERK1/2 activation are observed in response to different extracellular signals and also can be associated with differential compartmentalization of active ERK1/2 within the cell. For example, in PC12 cells, prolonged, nerve growth factor-induced activation of ERK1/2 leads to its persistent accumulation within nuclei; however, rapid and transient epidermal growth factor-induced activation of ERK1/2 is unable to trigger its efficient nuclear translocation (15,17). Interestingly, increasing evidence suggests that the kinetics and duration of ERK1/2 activation, as well as its subcellular localization, may direct ERK1/2 toward downstream targets that will either promote or limit neuronal survival (10,15,18).
In HT22 cells and primary neurons, glutamate induces a biphasic activation of ERK1/2 (5). Persistent activation of ERK1/2 in these cells is mediated primarily through the oxidative inhibition of select ERK phosphatases (19). However, the mechanisms responsible for the initial rapid activation of ERK1/2 by glutamate in HT22 cells and primary neurons are not known. Furthermore, the impact of the first wave of ERK1/2 activation on the response of HT22 cells and primary neurons to oxidative stress has not been established.
In this report, we show that rapid activation of ERK1/2 by glutamate in HT22 cells is driven by type I metabotropic glutamate receptors (mGluRIs). Furthermore, mGluRI activation of ERK1/2 represents a cellular defense response that attempts to limit glutathione depletion resulting from glutamate-induced cysteine depletion. This protective response mediated by ERK1/2 is however unable to overcome an overwhelming and chronic oxidative stress that utilizes ERK1/2 in its final stages to promote neuronal cell death. Thus, ERK1/2 activation may serve opposing roles in neuronal oxidative toxicity acting initially through effects on glutathione metabolism to limit oxidative stress but serving as a necessary signal to trigger cell death when cellular defense against oxidative stress is exhausted.

MATERIALS AND METHODS
Plasmids-The expression plasmids for LAERK2-MEK1 and ERK2-MEK1 were kind gifts from Dr. Melanie Cobb (20). pMCL-MEK1 Lys-97 3 Met (K97M), an HA-tagged dominant negative (DN) mutant of MEK1 was provided by Dr. Jane Cavanaugh (12). The mitochondrial-targeted enhanced yellow fluorescent protein (eYFP) expression plasmid was a gift from Dr. Ian Reynolds, and the expression plasmid for the ELK-1-GAL4 fusion protein, the luciferase reporter, and the constitutive Renilla reporter plasmids were obtained from Stratagene (La Jolla, CA).
Cell Culture-HT22 cells, a mouse hippocampal cell line, were maintained in Dulbecco's minimal essential medium supplemented with 10% fetal calf serum (Atlanta Biologicals, Norcross, GA), 100 units of penicillin, and 100 g/ml streptomycin at 37°C and 5% CO 2 .
Cell Viability Assay-HT22 cells grown in 24-well plates were incubated for 10 min with 1 l (1:1000 dilution) of a 6.25 mg/ml solution of propidium iodide (PI) to visualize dead or dying PI-positive cells (7). Cells were observed under an inverted fluorescence microscope equipped with phase-contrast optics (Nikon Eclipse TE200). Three random fields were counted for each condition in at least three separate cultures. For experiments that did not involve transfection, the fraction of PI-positive cells was scored relative to total cells/field visualized by phase contrast microscopy. In transfection experiments, the extent of PI-positive staining was scored in eYFP-positive transfected cells.
Transfection and Western Blotting-Cells were transfected with Lipofectamine 2000 (Invitrogen) using conditions recommended by the supplier. During transfection, cells were maintained in serum-and antibiotic-free medium. Following 4 h of exposure to DNA-Lipofectamine mixture, cells were refed with medium containing 10% fetal bovine serum. On the following day, cells were washed, scraped, collected into ice-cold phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 , 1.4 mM KH 2 PO 4 , pH 7.4), pelleted at 2-3 ϫ 10 3 ϫ g for 5 min, and disrupted in Lysis Buffer (50 mM Tris-Cl, pH 7.5, 2 mM EDTA, 100 mM NaCl, 1% Nonidet P-40, 100 M NaVO4, 100 M NaF, 2 mM dithiothreitol) supplemented with 5 l/ml of a protease inhibitor mixture (Sigma). The lysates were then centrifuged at 14,000 ϫ g for 10 min at 4°C. Equivalent amounts of total protein, 30 g, were separated by SDS-PAGE on 10% polyacrylamide gels and then transferred to polyvinylidene fluoride membranes (Millipore, Bedford, MA). Membranes were blocked with 5% dry milk in phosphate-buffered saline, 0.1% Tween 20. Membranes were then incubated with primary antibodies (anti-phospho-ERK, anti-total ERK, all from Cell Signaling, Beverly, MA) overnight at 4°C with 3% dry milk, washed 3 ϫ 10 min with phosphate-buffered saline, 0.1% v/v Tween 20, and then exposed to the appropriate horseradish-peroxidase-conjugated secondary antibody for 1 h at room temperature. Membranes were again washed 3 ϫ 10 min with phosphate-buffered saline, 0.1% Tween 20, and immunoreactive bands were detected by enhanced chemiluminescence (ECL, Amersham Biosciences) using standard x-ray film (Kodak). Several different exposure times were used for each blot to ensure linearity of band intensities. Densitometry was performed using a Personal Densitometer SI (Molecular Dynamics, Amersham Biosciences) linked to the Image-Quant 5.2 software (Molecular Dynamics).
Luciferase Assay-Components of the Path Detect In Vivo Signal Transduction Pathway trans-Reporting System (Stratagene) were used to monitor ELK-1-dependent transcription in HT22 cells (19). Cells were transfected with an expression plasmid coding for an ELK-1-GAL4 fusion protein, a reporter plasmid containing the luciferase gene linked to a synthetic promoter containing five tandem GAL4 binding sites, a Renilla expression plasmid, and various MEK1 expression plasmids (i.e. DN-MEK1, LA-MEK1ERK2). In this system, luciferase activity is a measure of the extent of ELK-1-activated transcription. Both luciferase activity and Renilla activity were examined with the Dual-Glo Luciferase Assay System (Promega, Madison, WI) using a luminometer (Wallac Victor3, PerkinElmer Biosciences) according to the manufacturer's specifications. Results shown were from at least three separate experiments. All luciferase values were normalized to the internal control, Renilla, within each sample.
Glutathione Concentration Measurement-A glutathione assay kit from Cayman Chemical Company (Ann Arbor, MI) was used to measure glutathione concentrations. Cells from two 60-mm plates were collected and sonicated in 50 l of ice-cold lysis buffer (i.e. 50 mM MES buffer, pH 6 -7, containing 1 mM EDTA). After centrifuging at 10,000 ϫ g for 15 min at 4°C, the supernatant was collected for deproteination with the exception of a small amount of cell lysate that was used to determine lysate protein concentration. Deproteination was necessary to avoid interference in the assay because of particulates and protein sulfhydryl groups. An MPA reagent (10% metaphosphoric acid in water; Sigma) and 4 M TEAM reagent (triethanolamine from Sigma) were used to deproteinize lysates. An equal volume of ice-cold MPA was added to ice-cold supernatant and left on ice for 10 min. Samples were then centrifuged at top speed in a microfuge for 5 min at 4°C. 50 l of TEAM reagent was added per ml of supernatant. 50 l of sample or the standard provided in the kit was applied to each sample well. An assay mixture mix was prepared with the following reagents: MES buffer (11.25 ml), reconstituted cofactor mixture (0.45 ml), reconstituted enzyme mixture (2.1 ml), water (2.3 ml), and reconstituted DTNB (5,5Ј-dithiobis(nitrobenzoic acid), 0.45 ml). 150 l of the freshly prepared assay mixture was added to each of the wells containing standards and samples. The plate was incubated in the dark on an orbital shaker at room temperature. Absorbance was measured at 405 or 414 nm using a plate reader 25 min later. An end point method was used to calculate and determine the sample glutathione concentration according to the instructions provided by the supplier. All glutathione concentration values were normalized to the protein concentration within each sample.
Statistical Analysis-Comparisons of multiple mean values were accomplished by analysis of variance with Bonferroni's post-hoc tests for significance. Comparisons of two means were performed using a paired t test. p values less than 0.05 were taken to be significant, and all data were analyzed using Graph Pad Prism version 3.0 for Windows (GraphPad Software, San Diego, CA).

ERK1/2 Activation Is Necessary for Glutamate-induced Oxidative
Toxicity in HT22 Cells-ERK1/2 has been implicated in both neuronal cell survival and cell death although the basis for its disparate effects is not known. In fact, even in HT22 cells, opposing roles for ERK1/2 have been proposed (5,16,22). Because part of the discrepancy for the role of ERK1/2 in neuronal cell survival may be because of secondary effects of pharmacological agents used to block ERK1/2 activation, we used a molecular approach to definitively assess the role of ERK1/2 in HT22 cell response to oxidative stress.
HT22 cells were therefore cotransfected with dominant negative MEK1 (DN-MEK1) and mitochondrial-targeted enhanced yellow fluorescent protein (mt-eYFP) expression plasmids. Because the cotransfected plasmids enter cells with equal probability, eYFP-positive cells were likely to contain DN-MEK1 DNA. DN-MEK1 contains a point mutation at position 97 (i.e. lysine to methionine) that inactivates its kinase activity (21). As a result, the DN-MEK1 interferes with activation of wild type MEK1. Cell death was measured in transfected cells 7 h after 5 mM glutamate treatment using a PI staining assay (7). As shown in Fig.  1A, DN-MEK1 overexpression blocked glutamate-induced oxidative toxicity in HT22 cells. These data confirm our previous results with a pharmacological inhibitor of MEK1 (i.e. U0126) and demonstrate that ERK1/2 activation is necessary for glutamate toxicity.
To confirm the inhibitory effect of DN-MEK1 on ERK1/2 activation, we used a sensitive assay to detect functional ERK1/2 in nuclei. Specifically, HT22 cells were transfected with the DN-MEK1 expression plasmid along with an expression vector for an ELK-1-GAL4 fusion protein and a luciferase reporter gene under the control of a promoter containing five tandem GAL4 DNA binding sites i.e. GAL4UAS-luciferase (UAS, upstream activator sequences). Given that ELK-1 is an established nuclear target of ERK1/2, enhanced transactivation activity resulting from its phosphorylation by nuclearly localized ERK1/2 is easily monitored through the activity of the luciferase reporter (19).
As shown in Fig. 1B, 10 M U0126 treatment decreased luciferase activity from GAL4UAS-reporter significantly in both untreated and glutamate treatment HT22 cells. Luciferase activity from the GAL4UAS-luciferase reporter was induced nearly 2-fold following a 7-h treatment with 5 mM glutamate (Fig. 1B) consistent with previous results of ERK1/2 activation using this assay (19). The expression of DN-MEK1 significantly reduced luciferase activity from the GAL4UAS-reporter in glutamate treated cells (Fig. 1C). The reduced effectiveness of DN-MEK1 to block ELK-1-activated transcription relative to U0126 treatment most likely results from variable expression of the DN-MEK1 within individual transfected cells. Nonetheless, these results confirm the expected inhibition of ERK1/2 activity in the nucleus of HT22 cells upon overexpression of DN-MEK1.
Prolonged ERK Activation Is Not Sufficient to Induce Toxicity in HT22 Cells-To assess whether chronic activation of ERK was sufficient to induce toxicity in HT22 cells, we used constitutively active ERK2 chimeric proteins. MEK1ERK2 is a fusion protein with constitutive ERK2 activity that localizes in the cytoplasm (20). In the constitutively active ERK2 fusion LAMEK1ERK2, four leucines in MEK1 that are crucial for its nuclear export have been mutated to alanines (23). As a result, the LAMEK1ERK2 fusion protein is localized in the nucleus (20). Furthermore, by virtue of its retention within nucleus, LAMEK1ERK2 exhibits a 5-10-fold higher ERK2 activity than MEK1ERK2 (20). To verify the activation of ERK signaling by LAMEK1ERK2, the Path Detect in vivo signal transduction pathway trans-reporting system was used to access ELK-1-dependent gene expression. MEK1ERK2/LAMEK1ERK2 expression plasmids along with an expression vector for an ELK-1-GAL4 fusion protein and the GAL4UAS-luciferase reporter gene were transfected into HT22 cells. Cell lysates were collected, and luciferase activity was measured 18 h after transfection. As shown in Fig. 2A, LAMEK1ERK2 overexpression can induce robust ERK1/2 activation compared with that of MEK1ERK2. The enhanced activation of the GAL4UAS luciferase reporter in LAMEK1ERK2-transfected cells likely results from its more efficient localization within nuclei.
We then accessed the effects of LAMEK1ERK2 overexpression on HT22 cell viability using the PI staining method. As shown in Fig. 2B, both LAMEK1ERK2-and MEK1ERK2-transfected cells showed no enhanced toxicity as compared with empty vector-transfected HT22 cells. Robust ERK activation upon overexpression of LAMEK1ERK2 chimera also did not unleash toxicity in HT22 cells exposed to a low dose of glutamate (1.5 mM). As will be shown below, 1.5 mM glutamate can trigger cell death under conditions of altered mGluRIs activity. Thus, although necessary, prolonged ERK activation is not sufficient to induce toxicity in HT22 cells nor does it predispose cells to toxicity under conditions of minimal glutamate exposure.
mGluRIs Are Involved in Glutamate-induced Oxidative Toxicity-HT22 cells express both type 1 and type 5 metabotropic glutamate receptors (mGluR1 and mGluR5) (24). These receptors are likely to be activated during glutamate exposure that triggers oxidative stress. Because mGluRIs have been suggested to play a protective role in glutamate toxicity (24), we sought to examine their role in promoting or attenuating ERK1/2-dependent toxicity in HT22 cells. HT22 cells were therefore pretreated with a mGluRI antagonist (R,S)-1-aminoindan-1,5-dicarboxylic acid (AIDA) or mGluRI agonist (R,S)-3,5-dihydroxyphenylglycine (DHPG) at 500 or 1000 M, respectively. Glutamate was added 30 min later at 5 mM (maximally toxic dose) or at 1.5 mM, a dose that exhibits minimal cytotoxic effects. 7 h after 5 mM glutamate and 22 h after 1.5 mM glutamate treatment, cell viability was measured by PI staining. 1.5 mM glutamate treatment of HT22 cells for 22 h only results The following day, cell death was induced by a 7-h treatment with 5 mM glutamate. Cell death in transfected cells was measured using a PI staining assay and was expressed as the percentage of PI and mt-eYFP-double-positive cells relative to the total mt-eYFP-positive cells. The percentage of cell death was decreased from 48 to 5% by expression of DN-MEK1. Data shown are from three separate experiments (**, p Ͻ 0.01). B, HT22 cells were transfected with an empty plasmid vector and then treated with Me 2 SO, 10 M U0126, 5 mM glutamate, or glutamate plus U0126. Luciferase activity was measured from a cotransfected ELK-1 reporter plasmid. Luciferase activity from the GAL4UAS-luciferase reporter was induced nearly 2-fold upon a 7-h 5 mM glutamate treatment. However, U0126 inhibited luciferase activity in both Me 2 SOand glutamate-treated groups. C, DN-MEK1 expression decreased ELK-1-driven luciferase activity in cells treated for 7 h with 5 mM glutamate (*, p Ͻ 0.05; **, p Ͻ 0.01).
in a 17% reduction in cell viability as assessed by PI staining (Fig. 3A). However, a 30-min pretreatment with AIDA potentiated toxicity in the presence of 1.5 mM glutamate resulting in 66% cell death after 22 h (Fig.  3A). In contrast, a 30-min pretreatment of HT22 cells with DHPG significantly reduced the minimal toxicity induced by 1.5 mM glutamate treatment to 5% (Fig. 3A). The toxicity observed in HT22 cells treated with glutamate and AIDA or DHPG was reduced by U0126 establishing that the toxicity regulated by the mGluRIs remains ERK1/2-dependent. In HT22 cells treated with 5 mM glutamate, AIDA and DHPG have similar effects as those observed in 1.5 mM glutamate-treated cell (Fig.  3B). These results confirm that mGluRIs play a role in limiting oxidative toxicity induced by glutamate treatment (24) and establish that limiting mGluRI activation promotes ERK1/2-dependent toxicity at subthreshold levels of glutamate.
Effects of mGluRIs Activation on the ERK1/2 Activation in Glutamate-induced Oxidative Toxicity-Given the protective effects of mGluRI activation on ERK1/2-dependent oxidative toxicity in HT22 cells, we set out to investigate whether mGluRI impacts ERK1/2 activation. Western blots were therefore performed with lysates prepared from HT22 cells treated with 1.5 mM glutamate and AIDA or DHPG and probed to detect phosphorylated ERK1/2. To assess the kinetics of ERK1/2 activation under the various conditions, protein samples were then collected at different time points (i.e. 30 min, 1, 6, 10, and 22 h) following glutamate addition. In all cases, the extent of ERK1/2 activation was measured relative to total ERK1/2 quantified from stripped blots (e.g. see Fig. 5).
As shown in Fig. 4A, treatment of HT22 cells with 1.5 mM glutamate alone induced a 3-fold activation of ERK1/2 that peaked within 30 min. ERK1/2 activation did not persist and eventually was reduced to levels ϳ40% below baseline following 10 h of treatment with 1.5 mM glutamate (Fig. 4A). Continued exposure to 1.5 mM glutamate alone (i.e. 22 h total) generated a second peak of ERK1/2 activation (Figs. 4A and 5B).
A role for mGluRIs in rapid transient activation of ERK1/2 by 1.5 mM glutamate was supported by the observed prevention of this activation by a 30-min pretreatment with AIDA (Fig. 4B). Specifically, the early FIGURE 2. Chronic activation of ERK1/2 is not sufficient to induce HT22 cell death. A, a constitutively active LAMEK1ERK2 fusion plasmid was transfected into HT22 cells, and luciferase activity was measured from an ELK-1-driven luciferase reporter plasmid. Expression of the LAMEK1ERK2 plasmid induced a robust increase in luciferase activity compared with transfections with a MEK1ERK2 plasmid (**, p Ͻ 0.01). B, HT22 cells were transfected with LAMEK1ERK2 (LA), MEK1ERK2 fusion plasmid, or empty vector PSG5 (PSG5). Cell toxicity was detected using the PI staining assay in the presence or absence of a 22-h 1.5 mM glutamate treatment (glu). Activation of ERK1/2 in LA-transfected cells did not trigger cell death (*, p Ͻ 0.05).

FIGURE 3. Effects of mGluRI activation or inactivation on glutamate toxicity in HT22 cells.
The percentage of cell death was measured using a PI staining assay as described above. The mGluRI antagonist AIDA enhanced glutamate (glu) toxicity while the mGluRI agonist DHPG decreased glutamate toxicity at both 1.5 mM (A) and 5 mM glutamate (B). In both groups, U0126 prevented cell death (*, p Ͻ 0.05; ***, p Ͻ 0.001). Relative phospho-ERK1/2 levels following 1, 4, and 6 h of glutamate treatment remained ϳ2.5-fold elevated but were eventually reduced to ϳ40% below baseline following 10 h of glutamate treatment. An additional 12-h exposure to 1.5 mM glutamate led to a 2.7-fold activation of ERK1/2 (n Ͼ ϭ 3). B, the level of ERK1/2 activation within a 30-min glutamate treatment (i.e. "early") was reduced by pretreatment with AIDA (see Fig. 5 for statistical analysis). Between 1 and 10 h of glutamate treatment, phospho-ERK1/2 levels were gradually reduced to a nadir of 24% of base line levels. However, AIDA enhanced a second phase of ERK1/2 activation that appears 22 h after glutamate treatment (see Fig. 5 for statistical analysis).

ERK1/2 and Glutathione Metabolism in Neurons
phase of ERK1/2 activation (i.e. within 30 min of glutamate treatment) was reduced 34% by AIDA (Fig. 5A). The addition of DHPG along with 1.5 mM glutamate did not lead to a significant effect on the initial phase of ERK1/2 activation (Fig. 5A). Baseline levels of activated ERK1/2 during the nadir between early and late phases of activation were often lower than that observed in unstimulated cells. Therefore, we compared the extent of ERK1/2 activation at 22 h relative to 10 h of glutamate treatment to more clearly illustrate the late phase of activation (Fig. 5B).
Using this approach, we observed that AIDA enhanced the second phase of ERK/2 activation nearly 4-fold (Fig. 5B). However, continuous exposure of HT22 cells to 1.5 mM glutamate and DHPG did not generate a second peak of ERK1/2 activation (Fig. 5B). These results identify a dual role for mGluRIs in regulating ERK1/2 activation. Activation of mGluRIs by glutamate triggers activation of ERK1/2 that may function to limit toxicity. However, in the absence of mGluRI activation, a delayed activation of ERK1/2 is enhanced that functions to promote cell death.
The Role of ERK1/2 in Glutathione Depletion-Glutamate-induced oxidative toxicity is triggered by glutathione depletion (8). A role for ERK1/2 in this component of glutamate toxicity has not been established. We therefore measured intracellular glutathione concentrations at various times after glutamate treatment in the presence or absence of U0126. Consistent with previous studies, a 5 mM glutamate treatment induced prolonged glutathione depletion in HT22 cells (Fig. 6A). GSH levels dropped to 84% in 2 h and 33% in 6 -8 h relative to baseline levels. Although U0126 protects HT22 cells from glutamate toxicity, it does not trigger the restoration of GSH levels in glutamate-treated HT22 cells (Fig. 6B). In this experiment, U0126 was added coincident with glutamate and therefore blocked both early and late phases of ERK1/2 activation. However, when U0126 was added 2 h following 5 mM glutamate treatment to allow for the early but not late phase of ERK1/2 activation, GSH levels returned to ϳ82% of basal line levels within 22 h (Fig. 6C). This delay in U0126 treatment protects HT22 cells from glutamate toxicity (5). Thus, ERK1/2 activation via mGluRIs signaling within the first few hours of glutamate treatment is required for HT22 cells to restore GSH levels that were reduced due to cysteine depletion.
As illustrated above, a suboptimal dose (i.e. 1.5 mM) of glutamate generates minimal toxicity in HT22 cells that is only apparent following prolonged exposure (22 h). Therefore, we investigated the kinetics of GSH depletion in response to a low dose of glutamate. Under these conditions, ERK1/2 activation is observed only at early times after glutamate addition. 1.5 mM glutamate induced an early depletion of GSH in 4-8 h to ϳ27% of basal line levels, which is comparable to the extent of GSH depletion in 5 mM glutamate-treated cells. However, a rebound in GSH levels occurred between 16 -22 h, which restores GSH to basal line levels (Fig. 6D). Interestingly, inhibition of mGluRI activation by a 30-min pretreatment with AIDA abolished the rebound in GSH levels that occurs in response to a 1.5 mM glutamate treatment. GSH levels remained low (i.e. 18 -46% of basal line levels) from 4 to 22 h (Fig. 6E). Under these conditions, only the late phase of ERK1/2 activation is observed. Thus, in the absence of mGluRI activation, a compensatory mechanism that acts to restore GSH levels is not operating, and cell death can result in HT22 cells from even a minimal exposure to glutamate.

ERK1/2 and Neuronal Cell
Death-It is now well recognized that ERK1/2 plays disparate roles in neurons, acting in some cases to promote cell survival (14,25) while also participating in neuronal cell death and the pathogenesis of neurodegeneration (10). Although a mechanistic basis for diverse effects of ERK1/2 in neurons is beginning to emerge, it is often difficult to make meaningful comparisons of results obtained in different cell lines or neuronal cell types. In this study, we have uncovered both the prosurvival and cell-death-promoting activity of ERK1/2 in one neuronal cell line exposed to a single toxic stimulus. We (5) and others (26) had previously shown that ERK1/2 is required for glutamateinduced oxidative toxicity relying principally on results obtained with pharmacological inhibitors of ERK1/2 activation such as the MEK1 inhibitor U0126. In the present study, we provide more definite support for this conclusion by showing a protective effect of DN-MEK1 on glutamateinduced oxidative toxicity in HT22 cells. Interestingly, we also showed that chronic ERK1/2 activation itself is not sufficient to induce cellular toxicity. Thus, ERK1/2 must cooperate with other pathways or cellular components affected by oxidative stress to contribute to toxicity.
Our initial studies of the kinetics of ERK1/2 activation in response to glutamate treatment revealed a biphasic pattern with an early peak of activation occurring within 30 -60 min and a second peak emerging within ϳ6 -8 h (5). Although recent results from our laboratory established a role for oxidative inhibition of ERK1/2 phosphatases in the second peak of ERK1/2 activation in glutamate-treated HT22 cells and primary cortical neurons (7), the signaling pathway that triggers the initial phase of ERK1/2 activation in these models remained undefined. In this report, we show that the initial phase of ERK1/2 activation is because of the activation of mGluRI. The transient nature of this rapid activation of ERK1/2 is likely because of desensitization of mGluRI, which occurs upon continuous exposure of HT22 cells to glutamate (24). Importantly, our results demonstrate that ERK1/2 can indeed function, as it does in many other neuronal and non-neuronal cells, as a prosurvival kinase. Specifically, in early phases of glutamate-induced FIGURE 5. Quantitative analysis of ERK1/2 activation in glutamate-treated HT22 cells. Autoradiograms from Western blots shown in Fig. 4 and replicates were subjected to densitometric analysis and relative phospho-ERK1/2 levels (normalized to total ERK1/2) were quantified using NIH Image (n Ͼ ϭ 3). As shown in A, AIDA inhibited by 4-fold the activation of ERK1/2 that occurs within 30 min of 1.5 mM glutamate (glu) treatment (*, p Ͻ 0.05). DHPG did not significantly affect ERK1/2 activation within 30 min of glutamate treatment. B, the second phase of ERK1/2 activation is illustrated by comparisons of ERK1/2 phosphorylation at 22 versus 10 h of 1.5 mM glutamate treatment. An AIDA pretreatment enhanced this late phase of ERK1/2 activation ϳ3-fold, whereas DHPG inhibited the late phase of ERK1/2 activation 5-fold (*, p Ͻ 0.05).
oxidative stress, ERK1/2 is a component of a cellular defense pathway that seeks to overcome oxidative stress by restoring depleted glutathione levels. However, as oxidative stress continues to develop, this cellular defense pathway is overwhelmed, perhaps due in part to desensitization of mGluRI, and ERK1/2 becomes a necessary player as oxidatively stressed cells enter the final stages of a unique cell death pathway (8). If the initial phase of ERK1/2 activation is blocked (i.e. by a mGluRI antagonist), HT22 cells become hypersensitive to a subthreshold dose of glutamate. The toxicity that results in HT22 cells from low doses of glutamate in the absence of mGluRI action remains ERK1/2-dependent.
Mechanism of ERK1/2 Activation by mGluRIs-mGluRs are G protein-coupled receptors with seven transmembrane domains. These receptors have been implicated in a variety of physiological functions, including neurotransmission, long term potentiation, and reciprocal interactions with the ionotropic glutamate receptors (27). mGluR-mediated neuroprotection against a variety of insults is well established, and each of the subgroups has been shown to provide neuroprotection in various contexts. Furthermore, the activation of mGluRs have been shown to be protective against a variety of neurotoxic insults in vivo including spinal cord injury, ischemia, epilepsy, multiple sclerosis, amyotrophic lateral sclerosis, and more recently, oxidative stress and diabetes (28).
HT22 cells and primary embryonic rat cortical neurons lack ionotropic glutamate receptors (29) and do not express mGluRII (24). Therefore, HT22 cells serve as a good model to illustrate the role of mGluRI in neuronal cell death. In fact, specific agonists and antagonists of mGluRIs have been shown to have anti-and proapoptotic roles, respectively, in HT22 cells and primary cortical neurons (24). Our results not only establish ERK1/2 as a mediator of mGluRI effect on oxidative toxicity but identify the downstream target of ERK1/2 (i.e. glutathione metabolism, see below) that is responsible for its initial neuroprotective effects.
The prosurvival effects of mGluR have previously been shown to be because of activation of the phosphatidylinositol 3-kinase and MAPK signaling pathways, which in turn may lead to decreased ROS accumulation (30,31). Additional evidence suggests that mGluRs regulate ROS production and oxidative stress albeit by distinct subtypes (e.g. group I, II, and III) depending upon the neurotoxic insult and neuronal type. For example, activation of mGluRII prevents ROS generation in the mitochondria of dorsal root ganglia neurons in response to elevated glucose levels (28). The mechanism of the neuroprotective action of mGluRI in FIGURE 6. Effects of ERK1/2 on glutathione depletion. A, HT22 cells were treated with 5 mM glutamate, and GSH levels were measured at 2, 4, 6, and 8 h afterward. A significant reduction of GSH levels to 30% of basal line levels was detected by 6 -8 h. B, HT22 cells were treated with 5 mM glutamate and 10 M U0126 simultaneously, and GSH levels were examined 4, 8, 16, and 22 h afterward. U0126 did not affect GSH depletion during the time course, although it blocks glutamate toxicity. C, HT22 cells were exposed to 10 M U0126 2 h after the initiation of a 5 mM glutamate treatment, and GSH levels were examined 4, 8, 16, and 22 h after glutamate addition. GSH levels decreased to 25 and 11% after 4 and 8 h, respectively. A rebound of GSH levels to 82% was observed within 22 h of glutamate treatment. D, HT22 cells were treated with 1.5 mM glutamate, and GSH levels were assayed at 4, 8, 16, and 22 h afterward. 1.5 mM glutamate induced a significant decrease of GSH in 4 -8 h, which rebounded to basal line GSH levels within 16 -22 h. E, HT22 cells were pretreated with AIDA for 30 min then exposed to 1.5 mM glutamate. GSH levels were assayed at 4, 8, 16, and 22 h afterward. AIDA treatment induced a prolonged and significant GSH depletion. Data shown above are representative of four separate experiments. The GSH reading was normalized to protein concentration. The GSH level of HT22 cells was set as 100% (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001).
these cells was not clear, although it was suggested to involve activation of phosphatidylinositol 3-kinase and ERK1/2 signaling pathways (30,31). Our results confirm the neuroprotective effects of mGluRI in HT22 cells and establish a role for ERK1/2 activation in mGluRI action to limit toxicity. Unfortunately, this protective response of HT22 cells driven by active ERK1/2 is not sufficient to overcome the overwhelming oxidative stress that ensues following prolonged exposure to a high concentration of glutamate.
ERK1/2 Activation and Glutathione Depletion-Glutathione is the most predominant non-protein thiol antioxidant in mammalian and plant cells (32). Glutamate, cysteine, and glycine are the three precursors of GSH, which plays an important role in many cellular processes such as metabolism of xenobiotic (33) and endogenous oxidants, cell proliferation (34), and regulation of gene transcription (35). Among those, the most well known function of GSH is antioxidant defense by acting as a coenzyme in glutathione peroxidase or GST-catalyzed reactions (36 -38). GSH depletion has been implicated in the pathogenesis of many neurological diseases such as Parkinson disease (39). Although some cells can take up GSH directly from their surroundings, adequate GSH levels are maintained by de novo synthesis. GSH concentrations in cells are determined by the balance of its synthesis and consumption. As discussed and evaluated by Hansen et al. (40), GSH synthesis is controlled by the first enzyme (rate-limiting enzyme) in the synthetic pathway, ␥-glutamylcysteine synthetase (GCS), which converts glutamate to cysteine (40). Some environmental factors can alter intracellular GSH levels by influencing this reaction (34). For example, H 2 O 2 (41) and some reactive species generated during exposure to toxins could upregulate GSH production (42). Unfortunately, this increase is limited and short lived (36).
GSH depletion and elevations in ROS levels do not represent the terminal, irreversible phase of cell death in oxidatively stressed HT22 cells. The protective effects of U0126 are not associated with a restoration of GSH levels (this work) or reduction in elevated ROS (5). Therefore, active ERK1/2 is more proximal to the irreversible step in HT22 cell oxidative toxicity and is downstream of ROS activation. Although there is not a link between GSH metabolism and the death-promoting action of the late phase of ERK1/2 activation, the initial prosurvival effects of active ERK1/2 (i.e. driven by mGluRI activation) appear to be due to its impact on GSH metabolism. Specifically, in HT22 cells that are triggered to activate ERK1/2 only at early times after glutamate addition (i.e. delayed U0126 addition), a rebound occurs that restores GSH to basal line levels. Without the initial activation of ERK1/2, this restoration of GSH levels in glutamate-treated HT22 cells does not occur.
Previous studies in HNE treated cells have shown that inhibition of ERK1/2 by PD98059 blocked GCSc (catalytic subunit of GCS) mRNA but not GCSm (modulator subunit of GCS) mRNA (36). Furthermore, ERK1/2 was found to up-regulate GCSc/GCSm mRNA through two possible pathways. One is through the increase in AP-1 DNA binding and the other through an increase in binding of the Nrf2 transcription factor to ARE binding sites (43). Future studies will reveal whether any of these mechanisms apply to the effect of ERK1/2 on GSH metabolism in HT22 cells.
In summary, the dual roles of ERK1/2 as a prosurvival and deathpromoting kinase can be observed within a single neuronal cell type (i.e. HT22 cells) exposed to a single toxin (i.e. glutamate). Specifically, at early stages of glutathione depletion-induced oxidative stress, ERK1/2 influences GSH metabolism and facilitates the recovery of GSH levels, although at later stages ERK1/2 becomes a necessary component of oxidative toxicity. Thus, any therapeutic intervention directed at the ERK1/2 pathway must take into account the divergent effects of this signaling pathway particularly within cells and tissues where oxidative stress is a contributing factor in cell death.