AMPA receptor-mediated toxicity in oligodendrocyte progenitors involves free radical generation and activation of JNK, calpain and caspase 3


Address correspondence and reprint requests to Guillermina Almazan, McGill University, Department of Pharmacology and Therapeutics, 3655 Sir William Osler Promenade, Room 1321, Montreal, Quebec H3G-1Y6, Canada. E-mail:


The molecular mechanisms underlying AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionate) receptor-mediated excitotoxicity were characterized in rat oligodendrocyte progenitor cultures. Activation of AMPA receptors, in the presence of cyclothiazide to selectively block desensitization, produced a massive Ca2+ influx and cytotoxicity which were blocked by the antagonists CNQX and GYKI 52466. A role for free radical generation in oligodendrocyte progenitor cell death was deduced from three observations: (i) treatment with AMPA agonists decreased intracellular glutathione; (ii) depletion of intracellular glutathione with buthionine sulfoximine potentiated cell death; and (iii) the antioxidant N-acetylcysteine replenished intracellular glutathione and protected cultures from AMPA receptor-mediated toxicity. Cell death displayed some characteristics of apoptosis, including DNA fragmentation, chromatin condensation and activation of caspase-3 and c-Jun N-terminal kinase (JNK). A substrate of calpain and caspase-3, α-spectrin, was cleaved into characteristic products following treatment with AMPA agonists. In contrast, inhibition of either caspase-3 by DEVD-CHO or calpain by PD 150606 protected cells from excitotoxicity. Our results indicate that overactivation of AMPA receptors causes apoptosis in oligodendrocyte progenitors through mechanisms involving Ca2+ influx, depletion of glutathione, and activation of JNK, calpain, and caspase-3.

Abbreviations used



buthionine sulfoximine




Dulbecco's modified Eagle's medium


experimental autoimmune encephalomyelitis




Hank's balanced salt solution


inositol trisphosphate


c-Jun N-terminal kinase




mitochondrial membrane potential


multiple sclerosis


3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide




polyacrylamide gel electrophoresis


phosphate-buffered saline


reactive oxygen species


sodium dodecyl sulfate


terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling.

Sustained activation of AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionate) receptors is toxic to oligodendroglial cell lines and to differentiated oligodendrocytes in culture (Yoshioka et al. 1996; Matute et al. 1997; McDonald et al. 1998a; Sanchez-Gomez and Matute 1999; Itoh et al. 2000; Kavanaugh et al. 2000; Kelland and Toms 2001; Alberdi et al. 2002). Consistent with the in vitro observations, injection of AMPA agonists into white matter, gray matter or optic nerve causes oligodendrocyte death and demyelination resembling multiple sclerosis (MS) lesions (Matute et al. 1997; Matute 1998; McDonald et al. 1998a; Hopkins et al. 2000; Jamin et al. 2001). Myelin is also damaged during acute brain insults causing glutamate release (Stys 1998; Li and Stys 2000). Loss of oligodendrocytes under hypoxic/ischemic conditions and in spinal cord injury is prevented by non-NMDA receptor antagonists, suggesting that AMPA/kainate (KA) receptor-mediated death of oligodendroglia may contribute to white matter damage (McDonald et al. 1998a; Rosenberg et al. 1999; Follett et al. 2000; Yoshioka et al. 2000; Tekkok and Goldberg 2001).

The rapid ischemic death of oligodendrocytes was shown to be mediated by Ca2+ influx via AMPA receptors (Fern and Moller 2000; Yoshioka et al. 2000). Furthermore, blockage of AMPA receptors increases oligodendrocyte survival and ameliorates the neurological score of animals with experimental autoimmune encephalomyelitis (EAE), an animal model of MS (Fern and Moller 2000; Pitt et al. 2000; Smith et al. 2000; Yoshioka et al. 2000; Matute et al. 2001). A recent report on MS lesions indicates that down-regulation of glutamate transport and glutamate-metabolizing enzymes may underlie oligodendrocyte pathology (Werner et al. 2001).

A few studies of excitotoxicity in mature oligodendrocytes suggested that either apoptotic or necrotic cell death can occur depending on culture conditions and the methods used to induce toxicity (Yoshioka et al. 1996; McDonald et al. 1998a; Yoshioka et al. 2000). However, the mode of excitotoxic death in oligodendrocyte progenitors and relevant signaling pathways remain unknown. It was postulated that glutamate toxicity is an important factor in periventricular leukomalacia, the most common ischemic lesion of white matter in premature infants (Back et al. 2001). Preferential damage to the white matter is related to the presence of late oligodendrocyte progenitors (Back et al. 2001), which are more vulnerable than mature oligodendrocytes to ischemic (Fern and Moller 2000; Follett et al. 2000) and as well as other insults (Almazan et al. 2000).

In this study, we examined the course of excitotoxicity in oligodendrocyte progenitors and characterized the molecular mechanisms involved in overactivation of AMPA receptors. Our results demonstrate that prolonged activation of AMPA receptors achieved by blocking receptor desensitization causes apoptosis in rat brain oligodendrocyte progenitors. Cytotoxicity was prevented by the antioxidant N-acetylcysteine (NAC) through a mechanism involving an increase in intracellular glutathione (GSH) levels. We also show for the first time that signaling events implicated in AMPA receptor-mediated excitotoxicity include a massive influx of Ca2+, depletion of GSH, and activation of JNK, calpain, and caspase-3.

Materials and methods


Kainate, AMPA, CNQX, GYKI 52466, MK-801, L-AP3 and cyclothiazide were from RBI (Natick, MA, USA). Anti-phospho-specific JNK (Thr183 and Tyr185) and anticleaved caspase-3 (17 kDa) antibodies were from New England Biolabs (Mississauga, ON, Canada). Anti-α-spectrin antibody was from Chemicon (Temecula, CA, USA). Dulbecco's modified Eagle medium (DMEM), Ham's F12 medium, Hank's balanced salt solution (HBSS) and 7.5% bovine serum albumin fraction V were from Gibco BRL (Burlington, ON, Canada). A cell death detection kit was from Roche (Laval, QC, Canada). Glutathione reductase wasfrom Boehringer Mannheim (Laval, QC, Canada). 45Ca2+ (21 mCi/mg), chemiluminescence reagents from Mandel (Guelph, ON, Canada) and PD 150606 and DEVD-CHO from Calbiochem (La Jolla, CA, USA). Immobilon-P membrane was from Millipore (Mississauga, ON, Canada), Protein A-Sepharose CL-4B was from Pharmacia Canada (Baie d'Urfe, QC, Canada) and platelet-derived growth factor AA and basic fibroblast growth factor were from Peprotech (Rocky Hill, NJ, USA). Poly d-lysine, 5,5′-dithiobis-2-nitrobenzoic acid, NADPH, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT), N-acetylcysteine and buthionine sulfoximine were from Sigma-Aldrich (Oakville, ON, Canada). Anti-JNK, anti-β-actin and glutathione-S-transferase-cJun were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Anti-GluR1, -GluR2, -GluR2/3 and -GluR4 antibodies were generous gifts from Dr R. Wenthold. All other reagents were obtained from Fisher (Montreal, QC, Canada) or VWR (Mississauga, ON, Canada).

Cell culture

Primary cultures were prepared from newborn rat brains as described by Almazan et al. (1993) based on the technique of McCarthy and de Vellis (1980). Oligodendrocyte progenitors were plated on poly d-lysine-coated culture dishes and grown in serum-free medium consisting of a DMEM-F12 mixture (1 : 1), 10 mm HEPES, 0.1% bovine serum albumin, 25 µg/mL human transferrin, 30 nm triiodothyronine, 20 nm hydrocortisone, 20 nm progesterone, 10 nm biotin, 5 µg/mL insulin, 16 µg/mL putrescine, 30 nm selenium and 2.5 ng/mL each of platelet-derived growth factor AA and basic fibroblast growth factor. Cultures were characterized immunocytochemically with cell type-specific antibodies (Cohen and Almazan 1994; Radhakrishna and Almazan 1994). More than 95% of the cells were positive for monoclonal antibody A2B5, a marker for oligodendrocyte progenitors in culture while less than 5% were galactocerebroside-positive oligodendrocytes (see Figs 1a and b), glial fibrillary acidic protein-positive astrocytes or complement type-3-positive microglia.

Figure 1.

Immunofluorescence staining of oligodendrocyte progenitor cultures and AMPA receptor expression. Cells were labeled with A2B5 (a) or with anti-galactocerebroside antibodies (b). (c) Samples containing 30 µg of total cell lysate protein were subjected to western blotting using anti-GluR1, 2, 2/3, and 4 or β-actin antibodies. A representative immunoblot of duplicate samples is shown.

MTT survival assay

Mitochondrial dehydrogenase activity assessed by cleavage of MTT provided an index of cell viability (Mosmann 1983). Cultures were incubated with 500 µg/mL MTT at 37°C for 3 h, washed and solubilized in an acidic-isopropanol mixture. Absorbance was measured at 600 nm with a micro-ELISA spectrophotometer.

Measurement of 45Ca2+ influx

Oligodendrocyte progenitors were incubated in Mg2+-free Locke's solution containing 1 µCi/mL 45Ca2+ for 5 min at 37°C (Liu et al. 1997). Cells were washed three times with ice-cold buffer (154 mm choline chloride, 2 mm EGTA, and 10 mm HEPES, pH 7.4) and lysed with 0.1 m NaOH/0.1% Triton X-100. Radioactivity was measured by liquid scintillation counting.

Western blot analysis and immunoprecipitation kinase assay

Cells were lysed in ice-cold buffer (20 mm Tris-HCl, pH 8, 1% Nonidet P-40, 10% glycerol, 137 mm NaCl, 1 mm phenylmethylsulfonyl fluoride, 1 mm aprotinin, 1 mm sodium vanadate and 20 mm NaF). Protein content was determined with a Bio-Rad kit, and 30 µg samples were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and transferred to Immobilon-P membranes. Blots were blocked with 5% milk in Tris-buffered saline containing 0.1% Tween 20, incubated with appropriate primary and secondary antibodies and immunoreactive bands were visualized by enhanced chemiluminescence and quantified by densitometry. To normalize for equal loading and protein transfer, the membranes were stripped and incubated with an antibody for β-actin.

JNK activity was determined by immunoprecipitation kinase assay (Giasson and Mushynski 1997). Cells were harvested in 200 µL of JNK lysis buffer (20 mm Tris, pH 7.4, 1% Triton X-100, 10% glycerol, 137 mm NaCl, 2 mm EDTA, 25 mmβ-glycerophosphate, 1 mm sodium orthovanadate, 2 mm pyrophosphate, 1 mm phenylmethylsulfonyl fluoride, 10 µg/mL leupeptin). JNK was immunoprecipitated from 100 µg of cellular protein using 1 µg of anti-JNK antibody and 20 µL of protein A-Sepharose suspension. The immunocomplexes were washed and assayed for JNK kinase activity using GST-cJun and [γ-32P]ATP as described. Phosphorylation of the substrate protein was visualized after SDS–PAGE by autoradiography and quantified by densitometry.

Measurement of glutathione

Total glutathione was determined in oligodendrocyte progenitors bya kinetic assay (Tietze 1969). Upon addition of 2.4 mm 5,5′-dithiobis-2-nitrobenzoic acid, 40 µg/mL glutathione reductase, and 0.8 mm NADPH to the cellular lysates, total glutathione content was determined by measuring the change in absorbance at 414 nm with a micro-ELISA spectrophotometer and compared with a glutathione standard curve. To determine the amount of oxidized glutathione, samples or standards were mixed with 2-vinylpyridine for 1 h to derivatize reduced glutathione (GSH) (Griffith 1980) before the assay. Under all experimental conditions, the amount of oxidized glutathione was always < 1% of the total glutathione. Therefore, the results are presented as GSH.

Terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) and 4,6-diamidino-2-phenylindole (DAPI) staining

Cells were stained using an ‘in situ cell death detection kit’ (Roche) to reveal DNA fragmentation. Cultures were fixed for 20 min in fresh 4% paraformaldehyde in phosphate-buffered saline (PBS). Endogenous peroxidase was inactivated by incubation in 0.3% H2O2/methanol for 30 min. Cells were permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate at 4°C, washed and incubated with TUNEL reaction mixture for 60 min at 37°C, followed by incubation with peroxidase-conjugated antibody for 30 min. Staining was visualized with a 3,3′-diaminobenzidine substrate kit (Vector Laboratories) and analyzed by light microscopy.

To visualize DNA condensation, fixed cells were stained with 1 µg/mL DAPI in PBS for 10 min and viewed under ultraviolet illumination using a fluorescence microscope.

Data analysis

One-way analysis of variance followed by the Tukey–Kramer test was used to determine statistical significance: p-values less than 0.05 were considered significant.


Expression of AMPA receptor subunits in oligodendrocyte progenitors

AMPA receptor subunit expression in oligodendrocyte progenitors was assessed by western blot analysis. Antibodies specific for AMPA receptor subunits GluR1, GluR2, GluR2/3 or GluR4 detected distinct proteins with apparent molecular masses of around 108 kDa as reported (Wenthold et al. 1992). All of the AMPA receptor subunits are expressed in progenitor cultures. The different intensities of the various bands compared to the β-actin standard suggest high expression levels for GluR2, GluR3 and GluR4 and a relatively lower level of GluR1 (Fig. 1c).

Prolonged activation of AMPA receptors kills oligodendrocyte progenitors

As we previously reported (Liu and Almazan 1995), exposure of oligodendrocyte progenitors to 100 µm AMPA or KA from 1 h−24 h is not cytotoxic as determined by MTT reduction. To assess whether this lack of AMPA or KA toxicity was due to rapid receptor desensitization, we examined the viability of progenitors treated with agonists as well as cyclothiazide, which selectively blocks desensitization of the AMPA receptor. After 1 h of KA plus cyclothiazide (100 and 25 µm, respectively) exposure, 15% of the cells had died, increasing to 30% at 6 h, and reaching 40% at 24 h (Fig. 2). A similar but more pronounced time-dependent cytotoxic effect was observed with AMPA treatment, reaching 50% at 24 h. The effect of cyclothiazide on KA or AMPA-mediated cell death was also concentration-dependent as 50 µm cyclothiazide caused even greater loss of progenitors.

Figure 2.

Prolonged activation of AMPA receptors causes time- and concentration-dependent cytotoxicity. Progenitors were treated with 100 µm KA (a) or AMPA (b) in the presence or absence of cyclothiazide (C) (25 or 50 µm) for 0.5–24 h. Toxicity was assessed by MTT reduction as described in Materials and methods. Data are expressed as mean ± SEM of three experiments performed in quadruplicate and are represented as percentages of the value for untreated control (100%). Statistical differences compared with control were as follows: AMPA or KA plus cyclothiazide for 30 min, 1 h, 2 h, 4 h, 18 h and 24 h (p < 0.001).

KA toxicity in the presence of cyclothiazide was prevented by the competitive AMPA/KA receptor antagonist CNQX (50 µm) and by the non-competitive AMPA receptor antagonist GYKI 52466 (50 µm). In contrast, MK-801 (100 µm) or L-AP3 (100 µm), respective NMDA and metabotropic receptor antagonists, had no effect (Table 1).

Table 1.  Pharmacological characterization of kainate plus cyclothiazide-mediated cytotoxicity and 45Ca2+ uptake in oligodendrocyte progenitors
TreatmentMTT assay
(relative OD)
45Ca2+ uptake
(nmol/mg protein)
  1. Cultures were preincubated with antagonists for 30 min prior to kainate (KA) plus cyclothiazide (CYZ) stimulation for 24 h for MTT assay or 5 min for 45Ca2+ uptake. Cell viability was determined by MTT assay and 45Ca2+ uptake was measured by liquid scintillation counting. Data represent the mean ± SEM from three experiments performed in quadruplicate. Statistical differences compared with control: *p < 0.001; compared with KA plus cyclothiazide: p < 0.001.

Control1.301 ± 0.033 5.38 ± 0.15
KA + CYZ (100 + 50 µm)0.426 ± 0.006*57.67 ± 0.84*
+ CNQX (50 µm)1.292 ± 0.010 6.28 ± 0.48
+ GYKI 52466 (50 µm)1.218 ± 0.013 7.58 ± 0.18
+ MK801 (100 µm)0.460 ± 0.03255.11 ± 0.68
+ LAP3 (100 µm)0.450 ± 0.01654.76 ± 1.64

Overactivation of AMPA receptors induces apoptosis in oligodendrocyte progenitors

We then investigated whether the cytotoxic effect of KA plus cyclothiazide involved cellular changes characteristic of apoptosis. DNA fragmentation was assessed by TUNEL assay and chromatin condensation by DAPI staining. Progenitors exposed to KA (100 µm) plus cyclothiazide (50 µm) for 24 h showed intense TUNEL labeling (Fig. 3b), as well as fragmented or condensed nuclei (Fig. 3d), indicative of an apoptotic mode of cell death.

Figure 3.

Overactivation of AMPA receptors causes apoptosis. Cultures were treated with KA plus cyclothiazide (100 µm + 50 µm) for 24 h followed by TUNEL assay (a and b) or DAPI staining (c and d). Controls (a and c) show intact nuclear staining with DAPI and few TUNEL-positive cells. In KA plus cyclothiazide-treated cultures, massive DNA fragmentation and chromatin condensation were visualized by TUNEL labeling (b) and DAPI staining (d), respectively.

Role of caspase-3 in KA plus cyclothiazide-mediated cell death

Activation of one or more members of the caspase family of cysteine proteases mediates apoptosis in many systems (Cohen 1997). To assess whether caspase-3 was involved in excitotoxic cell death, progenitors were pretreated with the selective inhibitor DEVD-CHO (100 µm) 1 h before addition of KA plus cyclothiazide for 3 h and 24 h later the MTT assay revealed that protection was complete (Table 2). Similar results were obtained with TUNEL staining (data not shown). Caspase-3 exists as a 32-kDa proenzyme that is proteolytically cleaved during activation into 17–20 kDa and 12-kDa subunits. We determined caspase-3 activation by western blotting with an antibody recognizing the 17 kDa fragment. A significant increase (1.7-fold) in the 17 kDa caspase-3 fragment was detected after 3 h of KA plus cyclothiazide treatment with a maximum at 9 h (4.3-fold) (Fig. 4).

Table 2.  Effects of the caspase-3 inhibitor DEVD-CHO and calpain inhibitor PD 150606 on kainate plus cyclothiazide-induced cytotoxicity
TreatmentMTT assay (relative OD units)
  1. Cells were preincubated with DEVD-CHO or PD 150606 for 1 h. After exposure to KA and cyclothiazide for 3 h, cells were washed and maintained in serum-free medium. MTT assay was carried out 24 h later to determine cell viability. Data represent the mean ± SEM from three experiments performed in quadruplicate. Statistical differences compared with control: *p < 0.01; compared with kainate (KA) plus cyclothiazide (CYZ): p < 0.001, p < 0.05.

Control1.308 ± 0.020
DEVD-CHO (100 µm)1.440 ± 0.032
PD 150606 (2 µm)1.294 ± 0.050
KA + CYZ (100 + 25 µm)1.086 ± 0.022*
KA + CYZ + DEVD-CHO1.364 ± 0.028
KA + CYZ + PD 1506061.268 ± 0.034
Figure 4.

KA plus cyclothiazide promote caspase-3 activation. Progenitors were treated with 100 µm KA and 25 µm cyclothiazide for 1–9 h. Caspase-3 activation was assessed by western blotting with an antibody against cleaved caspase-3 (17 kDa). (a) Representative blot with duplicate samples. (b) Signals were analyzed by densitometry and were expressed in arbitrary OD units as mean ± SEM of three independent experiments performed in triplicate. Statistical differences compared with basal values were as follows: 3 h (p < 0.01), 6 h and 9 h (p < 0.001).

Roles of Ca2+ and calpain in AMPA receptor-triggered excitotoxicity

AMPA receptors in oligodendrocyte progenitors are permeable to Ca2+ (Holzwarth et al. 1994; Pende et al. 1994; Liu et al. 1997), and Ca2+ overload plays a key role in the early phase of excitotoxicity (Choi 1988). We therefore investigated the role of Ca2+ in AMPA receptor-mediated cell death by measuring 45Ca2+ uptake elicited by AMPA receptor agonists in the presence of cyclothiazide. KA alone, at 100 µm, caused a small increase (1.8-fold from basal level) in 45Ca2+ uptake and its effect was greatly potentiated by cyclothiazide. A significant effect was detected with 5 µm cyclothiazide (1.8-fold) reaching a maximum at 50 µm (six-fold) (Fig. 5). 45Ca2+ influx in response to AMPA plus cyclothiazide increased with a similar profile but was of a greater magnitude (4.9-fold at 10 µm and 12.4-fold at 50 µm cyclothiazide) (Fig. 5). 45Ca2+ uptake was blocked by CNQX or GYKI 52466, but not by MK-801 or L-AP3 (Table 1).

Figure 5.

Cyclothiazide potentiates AMPA and KA-stimulated 45Ca2+ uptake. Cells were stimulated with AMPA or KA, both at 100 µm, with cyclothiazide (1–100 µm). 45Ca2+ uptake after a 5-min incubation was analyzed by liquid scintillation counting. The basal value was 5.17 nmol/min/mg protein. Values are mean ± SEM of three different experiments in triplicate. Statistical differences compared with the basal value were as follows: KA + 5 µm cyclothiazide (p < 0.05); KA or AMPA + 10–100 µm cyclothiazide (p < 0.001).

Since overactivation of AMPA receptors in oligodendrocyte progenitors causes a large increase in intracellular Ca2+ levels, we next examined the role of the Ca2+-activated protease, calpain, which has also been implicated in excitotoxic cell death (Rami et al. 1997; Zhao et al. 2000). Pretreatment of progenitors with the calpain inhibitor, PD 150606 (2 µm), prior to KA plus cyclothiazide reduced cell death, suggesting the involvement of calpain in excitotoxicity (Table 2).

Proteolysis of α-spectrin in KA plus cyclothiazide-treated cells

Having shown that both calpain and caspase-3 were involved in cell death induced by overactivation of AMPA receptors, we next examined the fragmentation pattern of α-spectrin, a cytoskeletal protein that is cleaved by calpain and caspase-3 during apoptosis (Nath et al. 1996). Both proteases cleave α-spectrin at specific sites to form two non-identical 150 kDa fragments. In addition, calpain subsequently cleaves this fragment to produce a 145-kDa product, while caspase-3 produces the apoptotic 120 kDa fragment. Exposure to KA plus cyclothiazide caused a time-dependent proteolysis of α-spectrin (Fig. 6a). The presence of calpain-mediated α-spectrin breakdown to a 150 and 145 kDa doublet increased dramatically following KA plus cyclothiazide treatment and a maximal increase was observed at 6 h (6.9-fold). The caspase-3 specific 120 kDa fragment also increased significantly with time and reached a 5.8-fold increase at 6 h exposure (Fig. 6b). These results indicate that calpain and caspase-3 are activated in oligodendrocyte progenitors during AMPA receptor-induced cell death.

Figure 6.

Calpain and caspase-3 cleave α-spectrin in KA and cyclothiazide-induced cell death. Cells were treated with KA and cyclothiazide (100 + 25 µm) for 1–9 h. Proteins were extracted and analyzed by western blotting with an antibody against α-spectrin. (a) Representative blot with duplicate samples. Arrows indicate a calpain-specific 150 and 145 kDa doublet and caspase-3-specific 120 kDa fragment. (b) Signals were analyzed by densitometry and expressed in arbitrary OD units as mean ± SEM of three different experiments in triplicate. Statistical differences compared with the basal value were as follows: for 145/150 kDa fragments: 1 h−6 h (p < 0.001); 9 h (p < 0.01); for 120 kDa fragment: 3 h (p < 0.05); 6 h (p < 0.001); 9 h (p < 0.01).

KA plus cyclothiazide activate c-Jun amino-terminal kinase (JNK)

JNKs are activated by different stimuli, including inflammatory cytokines and environmental stresses, some of which may lead to apoptosis (for a review, see Ip and Davis 1998). We monitored JNK activation by KA plus cyclothiazide with an immunoprecipitation kinase assay using GST-cJun as substrate (Fig. 7). KA (100 µm) activated low levels of JNK at time points ranging from 15 min to 6 h. A more significant increase in JNK activity (3.7-fold) was observed 6 h after KA plus cyclothiazide treatment. JNK activation was further confirmed by western blot analysis using a phospho-specific-JNK antibody recognizing the activated forms of p46 and p54 JNK (Fig. 7a). The phosphorylation levels for both JNK isoforms were significantly increased 6 h after KA plus cyclothiazide exposure, a time when there is a significant level of cell death.

Figure 7.

KA plus cyclothiazide activate JNK. Progenitors were exposed to KA (100 µm) or KA + cyclothiazide (25 µm). JNK activity was analyzed at the times indicated. (a) Upper panel: representative autoradiogram for in vitro kinase assay. Bottom panel: representative western blot using a phospho-specific JNK antibody. The upper and lower bands represent the 54 kDa and 46 kDa JNK isoforms. (b) Autoradiograms from in vitro kinase assay were analyzed by densitometry. Values are expressed in arbitrary OD units as mean ± SEM ofthree independent experiments performed in triplicate. Statistical differences compared with basal values were as follows: KA + cyclothiazide 6 h (p < 0.01).

N-Acetylcysteine prevents KA plus cyclothiazide-induced oligodendrocyte progenitor cell death: role of glutathione

To determine whether reactive oxygen species (ROS) are involved in cell death, cells were exposed to the antioxidant NAC, prior to treatment with KA plus cyclothiazide. NAC (5 mm) prevented cytotoxicity induced by overactivation of AMPA receptors, as assessed by MTT reduction (Table 3) or by TUNEL labeling (Table 4). As NAC is a precursor of GSH, it could exert its protective effect by increasing intracellular GSH levels. We tested this notion by measuring GSH content in cultures exposed to KA plus cyclothiazide in the presence or absence of NAC. KA plus cyclothiazide reduced intracellular GSH levels by 35% (Table 3). At 5 mm, NAC alone caused an increase in GSH content and it reduced the GSH depletion caused by KA plus cyclothiazide (Table 3). These results suggest that the protective effect of NAC is mediated, in part, by preventing the intracellular depletion of GSH.

Table 3.  Effects of N-acetylcysteine and buthionine sulfoximine on kainate plus cyclothiazide-induced cell death and intracellular glutathione content
TreatmentMTT assay
(relative OD units)
Glutathione content
(nmol/mg protein)
  1. Progenitors were pretreated with N-acetylcysteine (NAC) or buthionine sulfoximine (BSO) for 1 h followed by exposure to kainate (KA) plus cyclothiazide (CZY) for 24 h. MTT assay and measurement of intracellular glutathione (GSH) content were carried out as described in Materials and methods. Data represent the mean ± SEM from three experiments performed in quadruplicate. Statistical differences compared with control: *p < 0.001; compared with KA plus cyclothiazide: p < 0.001, p < 0.05.

Control1.387 ± 0.0474.37 ± 0.03
NAC (5 mm)1.598 ± 0.0725.69 ± 0.06*
BSO (10 µm)1.376 ± 0.0360.88 ± 0.03*
KA + CYZ (100 + 25 µm)0.792 ± 0.023*2.79 ± 0.02*
KA + CYZ + NAC1.272 ± 0.0994.24 ± 0.01
KA + CYZ + BSO0.521 ± 0.0180.77 ± 0.02
Table 4.  Effect of N-acetylcysteine on kainate plus cyclothiazide-induced DNA fragmentation
TreatmentTUNEL-positive progenitors (%)
  1. Progenitors were pretreated with NAC (5 mm) for 30 min followed by kainate (KA) plus cyclothiazide (CZY) (100 + 25 µm) stimulation for 3 h. DNA fragmentation was assessed by TUNEL assay and positive cells were quantified by light microscopy. Values are expressed as percentage of TUNEL-positive cells from total cells counted and represented as mean ± SEM. Statistical differences compared with control: *p < 0.001; compared with KA and cyclothiazide: p < 0.001.

Control5.91 ± 0.14
NAC (5 mm)5.27 ± 0.13
KA + CYZ (100 + 25 µm)26.69 ± 0.63*
KA + CYZ + NAC9.84 ± 0.32

Further support for the involvement of GSH in moderating KA toxicity was obtained by testing the effect of buthionine sulfoximine (BSO), a specific γ-glutamylcysteine synthetase inhibitor which blocks the rate-limiting step of GSH synthesis, on intracellular GSH levels and cell viability. At 10 µm, BSO alone dramatically decreased GSH content but had no effect on cell viability (Table 3). GSH depletion and cell death caused by KA plus cyclothiazide were further enhanced when cultures were pretreated withBSO (Table 3). These data suggest that GSH is a crucial component of NAC-mediated protection against excitotoxicity.


In this study we demonstrate that prolonged activation of AMPA receptors achieved by concomitant blockage of receptor desensitization was cytotoxic to oligodendrocyte progenitors. Dying cells displayed several characteristics typical of apoptosis, including DNA fragmentation, chromatin condensation and caspase-3 activation. We also show for the first time that the underlying molecular mechanisms implicated in this type of cell death involve massive Ca2+ influx, GSH depletion and JNK and calpain activation. Furthermore, the antioxidant, NAC, prevents cell death by maintaining intracellular GSH levels.

Cells of the oligodendrocyte lineage express functional AMPA and KA receptors (for a review, see Gallo and Ghiani 2000). In addition, glutamatergic synapses were demonstrated between CA3 pyramidal neurons and CA1 oligodendrocyte progenitors in rat hippocampus (Bergles et al. 2000). Electrophysiological studies showed rapidly desensitizing responses upon AMPA or KA stimulation in oligodendrocytes (Gallo et al. 1994; Patneau et al. 1994). This type of agonist-induced receptor desensitization was proposed as a neuronal mechanism for protection against excitotoxicity. Indeed, a lack of AMPA toxicity was observed in cerebellar granule (Cebers et al. 1997; Impagnatiello et al. 1997), neocortical (Jensen et al. 1998) and hippocampal neurons (May and Robison 1993). Here we confirm our previous findings and those of others showing that AMPA or KA alone in the micromolar concentrations and under standard culture conditions are not toxic to rat oligodendrocyte progenitors (Pende et al. 1994; Liu and Almazan 1995; Gallo et al. 1996; Yuan et al. 1998) and further demonstrate that AMPA receptor-mediated excitotoxicity is highly dependent on receptor desensitization. In contrast, mouse progenitors derived from forebrain and cocultured with type-1 astrocytes are highly vulnerable to excitotoxicity (McDonald et al. 1998b). The differences in vulnerability may reflect species differences (mouse versus rat), culture origins (forebrain versus whole brain), cell preparations (coculture with type-1 astrocytes versus pure cultures) and the molecular composition of the AMPA receptor channel as discussed below.

The rate of desensitization of AMPA receptors can be pharmacologically manipulated with the diuretic compound, cyclothiazide, to enhance glutamatergic synaptic currents (Yamada and Tang 1993). In oligodendrocyte progenitors, cyclothiazide blocks desensitizing responses to AMPA and potentiates steady-state responses to KA (Patneau et al. 1994; Berger 1995). We observed a time- and concentration-dependent cell death induced by AMPA or KA when cyclothiazide was present. These results together with other studies (Yoshioka et al. 1995; Matute et al. 1997; Sanchez-Gomez and Matute 1999; Itoh et al. 2000; Kavanaugh et al. 2000) further support the notion that rapid desensitization of AMPA receptors protects oligodendrocyte progenitors against excitotoxicity under normal physiological conditions. A recent report showed that progenitors and immature oligodendrocytes are more vulnerable to AMPA receptor-mediated injury than mature oligodendrocytes, suggesting that excitotoxicity in cells of the oligodendrocyte lineage is developmentally regulated (Kavanaugh et al. 2000).

Our studies of oligodendrocyte progenitors showed that both AMPA receptor-mediated cell death and 45Ca2+ uptake were blocked by an AMPA/KA receptor antagonist, suggesting that entry of excess Ca2+ is the primary trigger of consequent injury. A similar effect was observed in progenitors from the JS3/16 oligodendrocyte-like cell line (Itoh et al. 2000). In addition, AMPA receptor-mediated cell death in mature oligodendrocytes was prevented in Ca2+-free medium indicating a Ca2+-dependent excitotoxicity (Yoshioka et al. 1995; Matute et al. 1997; Sanchez-Gomez and Matute 1999; Alberdi et al. 2002). As we previously reported, Ca2+ influx triggers AMPA receptor-mediated phospholipase C activation and resultant inositol trisphosphate (IP3) accumulation in oligodendrocyte progenitors (Liu et al. 1997). The IP3-mediated release of Ca2+ from intracellular stores may serve as a positive feedback loop resulting in Ca2+ overload. One important determinant of the Ca2+ permeability of AMPA receptors is the presence of an edited version of the GluR2 subunit, GluR2(R), which makes the channel impermeable to Ca2+. Analysis of AMPA receptor in our cultures revealed the expression of all subunits, GluR1-4, consistent with the presence of GluR2 in progenitors from the CG-4 and JS3/16 cell lines (Patneau et al. 1994; Yoshioka et al. 1995; Meucci et al. 1996; Yoshioka et al. 1996; Chew et al. 1997; Itoh et al. 2000) and their low permeability to Ca2+. In contrast, the pronounced vulnerability to excitotoxicity of differentiated oligodendrocytes isolated from optic nerve may be due to activation of AMPA receptors that do not contain GluR2 and are thus highly permeable to Ca2+(Matute et al. 1997; Sanchez-Gomez and Matute 1999).

Oligodendrocyte progenitors exposed to KA and cyclothiazide showed marked TUNEL-positive labeling and condensed or fragmented nuclei suggesting that a subpopulation of cells died via apoptosis. TUNEL labeling and DNA laddering were also observed in CG-4 mature oligodendrocytes treated with millimolar concentrations of KA (Yoshioka et al. 1996). A recent study in JS3/16 progenitors demonstrated that mitochondria play a crucial role in buffering the large Ca2+ load evoked by AMPA plus cyclothiazide and Ca2+ sequestration within mitochondria may account for a fall in mitochondrial membrane potential (MMP) (Itoh et al. 2000), which is one of the earliest events in apoptosis (Wadia et al. 1998). In neurons the rapid influx of Ca2+ through Ca2+-permeable AMPA/KA receptors resulted in mitochondrial Ca2+ overload and consequent production of ROS (Carriedo et al. 1998; Carriedo et al. 2000). Consistent with previous studies (Mayer and Noble 1994; Yoshioka et al. 1996), we found that the antioxidant, NAC, prevented excitotoxicity suggesting that ROS were being generated. In addition to functioning directly as an ROS-scavenging antioxidant, NAC by increasing intracellular cysteine levels, provides a vital substrate for GSH synthesis. Indeed, we found that KA plus cyclothiazide caused a decrease in the level of intracellular GSH, and NAC was able to prevent this decrease. The finding that GSH depletion by BSO rendered progenitors more vulnerable to KA plus cyclothiazide suggests a role for GSH in protecting cells from excitotoxicity. This notion is supported by the reported increase in ROS production and apoptotic features in GSH-depleted progenitors (Back et al. 1998).

It has been shown that the concurrent buildup of intramitochondrial Ca2+ and attenuation of MMP induce the opening of the mitochondrial inner membrane permeability transition pore (Schinder et al. 1996). Subsequent rupture of the outer mitochondrial membrane releases pro-apoptotic factors from the mitochondrial intermembrane space, including cytochrome c (Scarlett and Murphy 1997), which in turn initiates the downstream caspase cascades that characterize apoptosis. Recent reports have shown that oligodendrocytes express several initiator and effector caspases, including caspase-3, and caspase activation was observed during apoptosis (Craighead et al. 1999; Gu et al. 1999). Our findings that the caspase-3 inhibitor DEVD-CHO blocks cell death and that caspase-3 is activated under excitotoxic conditions suggest that this protease plays a pivotal role in the execution of apoptosis induced by KA plus cyclothiazide. Increased expression of the p17 fragment, representing activated caspase-3, was also evident in apoptotic oligodendrocytes following experimental trauma and EAE (Beer et al. 2000; Han et al. 2000; Hisahara et al. 2000; Shibata et al. 2000; Hisahara et al. 2001). Similarly, hypoxic-ischemic brain injury caused apoptotic death of oligodendrocytes at various stages of development (Fern and Moller 2000; Skoff et al. 2001; Back et al. 2002) and involved excitotoxicity by AMPA/KA receptors (Follett et al. 2000; Tekkok and Goldberg 2001).

A rapidly increasing number of caspase substrates has been identified, including structural proteins, proteins involved in signal transduction pathways and nuclear proteins (for a review, see Chan and Mattson 1999). We showed that treatment with KA plus cyclothiazide caused the proteolysis of α-spectrin, a cytoskeletal protein and known substrate for caspase-3. Accumulation of the caspase-3-cleaved α-spectrin breakdown product, a 120-kDa fragment, provided further evidence for caspase-3 activation in apoptosis induced by excitotoxicity. Our findings are in line with those seen in apoptotic oligodendrocytes where the active from of caspase-3 and the caspase-3-specific cleavage product of α-spectrin were detected (Craighead et al. 2000; Beesley et al. 2001). However, we cannot exclude the possibility that necrosis may also occur in AMPA receptor-mediated injury. Indeed, necrotic cell death has been reported in differentiated oligodendrocytes cocultured with astrocytes and in CG-4 oligodendrocytes exposed to high concentrations of KA or to oxygen-glucose deprivation (Yoshioka et al. 1996; McDonald et al. 1998a; Yoshioka et al. 2000).

Sustained elevation of intracellular Ca2+ levels can trigger multiple Ca2+-dependent cascades that ultimately lead to excitotoxicity. One of the important mediators is the Ca2+-activated protease, calpain. Early studies localized calpain in myelin and in myelin-forming oligodendrocytes (Banik 1992; Li and Banik 1995). Our results showing that the calpain inhibitor, PD 150606, decreased cell death induced by KA plus cyclothiazide suggest that high intracellular Ca2+ levels increase calpain activation which contributes to cytoskeletal degradation and cell death. Calpain activation during excitotoxicity was further confirmed by the observed cleavage of α-spectrin to calpain-specific 145 kDa fragment.

We demonstrate for the first time that overstimulation of AMPA receptors results in the activation of JNK in oligodendrocyte progenitors. JNK activation in differentiated oligodendrocytes has also been reported in response to oxidative stress leading to cell death (Casaccia-Bonnefil et al. 1996; Zhang et al. 1996; Zhang et al. 1998; Bhat and Zhang 1999). We found a significant increase in JNK activity in cells exposed to KA plus cyclothiazide for 6 h, a condition where 30% cell loss was observed, suggesting the involvement of JNK signaling in apoptosis induced by excitotoxicity. Recent studies have demonstrated JNK activation in neuronal cultures undergoing glutamate-induced apoptosis (Kawasaki et al. 1997; Schwarzschild et al. 1997). A role for JNK in excitotoxicity is further supported by the report of Yang et al. (Yang et al. 1997) showing that mice lacking the JNK3 gene, which encodes a JNK species selectively expressed in the brain, exhibited a remarkable resistance to KA-induced apoptosis. Although the molecular mechanisms relating to JNK-dependent apoptosis remain unclear, an influence on the mitochondrial death signaling pathway has been suggested (Tournier et al. 2000).

In summary, we have demonstrated that overactivation of AMPA receptors by blocking receptor desensitization with cyclothiazide causes apoptosis in oligodendrocyte progenitors. The events implicated in the signaling cascades leading to cell death include increased influx of Ca2+, depletion of GSH, and activation of JNK, calpain, and caspase-3. Excitotoxicity can be prevented by the antioxidant NAC through mechanisms involving replenishment of intracellular GSH. During inflammation in both EAE and MS, the excessive amounts of glutamate released from lymphocytes, brain microglia and macrophages can activate AMPA receptors in oligodendrocytes, leading to increased Ca2+ influx and cell death (Steinman 2000; Werner et al. 2001). Understanding the signaling pathways linking AMPA receptors with cell death may therefore aid in the development of new strategies for treatment of demyelinating diseases.


This work was funded by operating grants from the Multiple Sclerosis Society of Canada and Canadian Institutes of Health and Research to GA. HNL was supported by a studentship from the Multiple Sclerosis Society of Canada.