Differential Oxidative Stress in Oligodendrocytes andNeurons After Excitotoxic Insults and Protection byNatural PolyphenolsGASKON IBARRETXE, MAR�IA VICTORIA S�ANCHEZ-G �OMEZ, MAR�IA ROSARIO CAMPOS-ESPARZA,ELENA ALBERDI, AND CARLOS MATUTE*

Departamento de Neurociencias, Facultad de Medicina y Odontologia, Universidad del Pa�ıs Vasco, Leioa, Spain

KEY WORDSAMPA receptor; kainate receptor; calcium; mitochondrion;reactive oxygen species; free radicals; excitotoxicity; poly-phenol; mangiferin; morin

ABSTRACTOligodendrocytes are vulnerable to overactivation of boththeir AMPA receptors and their high- and low-affinity kai-nate receptors. Depending on the intensity of the insultand the type of receptor activated, excitotoxic oligodendro-cyte death mediated by these receptors has different char-acteristics. One important consequence at a cellular level isthe ensuing oxidative stress, related to Ca21-dependentalterations in mitochondrial functioning. We observed thatoxidative stress associated with selective AMPA receptoractivation is much higher than that associated with theselective activation of high- and low-affinity kainate recep-tors. Moreover, excitotoxic insults generate more intenseoxidative stress in oligodendrocytes than in cortical neu-rons, though similar alterations in [Ca21]i and mitochon-drial potential were observed in both cell types. Nanomolarconcentrations of mangiferin and morin, two natural poly-phenols with antioxidant properties, partially protect oligo-dendrocytes as well as cortical neurons from mild, but notintense, insults mediated by AMPA receptors. In additionto presenting oxygen radical scavenging activity, mangiferinand morin attenuate the intracellular Ca21 overload subse-quent to the activation of AMPA receptors, a mechanismthat may contribute to their protective properties. Theinclusion of these antioxidant agents in therapeutic strate-gies for the treatment of diseases in which oligodendrocyteas well as neuron loss occurs may prove to be beneficial.VVC 2005 Wiley-Liss, Inc.

INTRODUCTION

Oligodendrocytes are sensitive to excitotoxic insultsmediated by overactivation of their AMPA/kainate iono-tropic glutamate receptors (Yoshioka et al., 1996; Matuteet al., 1997; McDonald et al., 1998; S�anchez-G�omez andMatute, 1999). Excitotoxicity in oligodendrocytes, as wellas in neurons, is completely dependent on Ca21 entrythrough the ionotropic glutamate receptor (Choi, 1995;Alberdi et al., 2002; Krieger and Duchen, 2002). Overac-tivation of AMPA and kainate receptors in neurons andoligodendrocytes causes abrupt rises in the concentrationof cytoplasmic Ca21 in both cell types (Alberdi et al.,2002; Krieger and Duchen, 2002).

One important intracellular target for Ca21-mediatedtoxicity is the mitochondrion, which can take up highloads of this cation in an electric potential-dependentfashion (Duchen, 2000; Hajn�oczky et al., 2003). Isolatedmitochondria exposed to high Ca21 concentrations, likethe ones present in extracellular medium (�2 mM), gen-erate reactive oxygen species, an effect that is dependenton their uptake of this ion (Ermark and Davies, 2001;Sousa et al., 2003). A similar process occurs in wholecells, both in neurons and in oligodendrocytes, after Ca21

uptake into mitochondria as a consequence of excitotoxicinsults (Carriedo et al., 2000; S�anchez-G�omez et al.,2003).

Exposure of oligodendrocytes to AMPA and kainatecauses dose-dependent toxicity (S�anchez-G�omez andMatute, 1999) and caspase-dependent and -independentcell death (S�anchez-G�omez et al., 2003). Thus, maximalactivation of high- and low-affinity kainate receptors, aswell as submaximal activation of AMPA receptors, inducemoderate oligodendrocyte death, which can be preventedby caspase-3 inhibitors. In contrast, maximal activationof AMPA receptors causes severe oligodendroglial deaththat is caspase-independent. In all instances, excitotoxicinsults alter Ca21 homeostasis and mitochondrial func-tion and oxidative stress ensues (S�anchez-G�omez et al.,2003). The present results corroborate these observationsand provide evidence that oxidative stress in optic nerveoligodendrocytes is much higher in excitotoxic insultsdue to AMPA in comparison with kainate receptor activa-tion. Interestingly, excitotoxic insults to neurons produceless oxidative stress than that measured in oligodendro-cytes. In addition, we show that two natural antioxidantpolyphenols protect against mild excitotoxic insultsmediated by AMPA receptors and that the protectivemechanisms involve free radical scavenging and Ca21

handling in the cytosol.

Grant sponsor: Spanish Ministry of Health and Consumption; Grant sponsor: LaCaixa; Grant sponsor: Basque Government; Grant sponsor: Basque Country Uni-versity; Grant sponsor: CONYTEA; Grant sponsor: Carolina Foundation; Grantsponsor: Spanish Ministry of Education and Science.

*Correspondence to: Carlos Matute, Departamento de Neurociencias, Facultadde Medicina y Odontolog�ıa, Universidad del Pa�ıs Vasco, 48940-Leioa, Spain.E-mail: onpmaalc@lg.ehu.es

Received 21 April 2005; Accepted 30 June 2005

DOI 10.1002/glia.20267

Published online 3 October 2005 in Wiley InterScience (www.interscience.wiley.com).

VVC 2005 Wiley-Liss, Inc.

GLIA 53:201–211 (2006)

MATERIALS AND METHODSDrug Application

AMPA and cyclothiazide (CTZ) (Tocris Cookson, Bris-tol, UK), kainate (Sigma, St. Louis, MO) and GYKI53655,kindly supplied by D. Leander (Eli Lilly and Company,Indianapolis, IN), were first dissolved in an equimolarsolution of NaOH (AMPA and kainate), ethanol (CTZ), orDMSO (GYKI53655) and were then added to culturemedium to achieve the desired final concentration. Man-giferin and morin (Sigma, St. Louis, MO) were dissolvedin DMSO at a concentration of 100 mM.

Optic Nerve Cultures

Primary cultures of oligodendrocytes derived from theoptic nerves of 12 day old Sprague-Dawley rats wereobtained as described previously (Barres et al., 1992),with minor modifications (Alberdi et al., 2002). Cellswere seeded into 24-well plates bearing 12–14-mm-dia-meter coverslips coated with poly-D-lysine (10 lg/ml) at adensity of 10,000 cells/well and maintained at 37�C and5% CO2 in a chemically defined medium (Barres et al.,1992). At 1–3 days in vitro, cultures were composed of atleast 98% O4/GalC1 cells, as determined by immuno-fluorescence labeling with a mouse IgG3 anti-galactocer-ebroside antibody (3 lg/ml; Boehringer-Mannheim, Ger-many); the majority of the very few remaining cellsimmunolabeled for GFAP (Z0334; Dakocitomation). Cul-tures were employed for experiments at this stage. NoA2B51 or microglial cells were detected. (Alberdi et al.,2002).

Cortical Neuron Cultures

Primary neuron cultures were obtained from E17 Spra-gue-Dawley rat cortices as described elsewhere (Dawsonet al., 1993). Cells were seeded into 24-well plates bear-ing 12–14-mm-diameter coverslips coated with poly-L-ornithine (10 lg/ml) and maintained at 37�C and 5% CO2

in Neurobasal medium, supplemented with B27 ‘‘minusantioxidants’’ (Gibco). Cultures were used at 1 week invitro. At this time point, at least 98% of the cells wereMAP21, as determined by immunolabeling with a mono-clonal anti-microtubule associated protein antibody(M1406; Sigma), and the majority of remaining cells wereGFAP1.

Exposure of Cells to Agonists

All the agonists were applied for 15 min, or alternativelyfor 5 min when indicated. We employed four differenttypes of excitotoxic stimulation, as described in otherreports (S�anchez-G�omez et al., 2003). Selective sustainedactivation of oligodendroglial and neuronal AMPA recep-tors was achieved with the agonist in the presence of 100lM CTZ. Moderate and maximal activation of these recep-tors was achieved using 10 lM and 100 lM AMPA, respec-tively. Kainate activates both AMPA and kainate receptors.To selectively stimulate the latter subtypes of receptors inoligodendrocytes, we used kainate in combination withthe AMPA-selective antagonist GYKI 53655 at 100 lM. Wedistinguished between activation of high- and low-affinitykainate receptors by incubating cells with this agonist at3 lM and 3 mM, respectively. Cells were exposed to100 lM CTZ or 100 lM GYKI 53655 for 10 min beforeincubation with AMPA or kainate, respectively. Morin andmangiferin were diluted in culture medium to the desiredconcentration, added 24 h before exposure to agonists andleft there until the end of the experiments.

In Vitro Estimation of Antioxidant Activity

The antioxidant power of polyphenolic test agents wasevaluated by the 1,1-diphenyl-2-picrylhydrazyl (DPPH·,Sigma) assay (Aruoma, 2003). We monitored the reductionof optical absorbance at 517 nm of the DPPH· stable freeradical upon reaction with test compounds in ethanol solu-tion, every 10 min after the beginning of the reaction. Thiswas initiated by adding 10 ll of test compounds at 1 mM to300 ll of 100 lM DPPH·. The endpoint decrease in absor-bance represents the antioxidant power of the test species.

Mitotracker Orange Staining

Reduced-Mitotracker Orange (CM-H2TMRos; Molecu-lar Probes, Eugene, OR; M7511) is a mitochondrial poten-tial-sensitive probe that can also be used to measure oxi-dative stress (Bernardi et al., 1999; Kweon et al., 2001).As in the case of dichlorofluorescein-related probes, thefluorescence of this species is dramatically enhancedwhen its reduced moieties are oxidized by free radicals.Oligodendrocytes were exposed to agonists and were sub-sequently incubated with Reduced-Mitotracker Orange(200 nM) for 15 min. Thereafter, cells were fixed with 4%paraformaldehyde and images were taken using anOlympus EC600 confocal microscope with a Plan-Neo-fluar 1003 immersion oil objective. Excitation with aHelium–Neon laser at a wavelength of 546 nM was car-ried out, using a pinhole of 50 lM. Fluorescence emissionwas detected using a conventional rhodamine filter.

Intracellular Reactive Oxygen SpeciesMeasurement

Oligodendrocyte and neuron cultures (10,000 cells/well) were exposed to AMPA or kainate receptor agonists

Abbreviations used

AMPA 3-amino-5-hydroxy-4-methylisoxazole propionic acidkainate (2S, 3S, 4R)-carboxy-4-(1-methylethenyl)-3-pyrrolidinea-

cetic acidROS reactive oxygen speciesCTZ cyclothiazideGYKI methyl-dioxol-2,3-benzodiazepin-5-benzenamideMAP 2 microtubule-associated protein 2GFAP glial fibrillary acidic protein[Ca21]i intracellular calcium concentration

202 IBARRETXE ET AL.

as described. Then, accumulation of reactive oxygen spe-cies within cells was measured by loading cells with10 lM 5,6-chloromethyl-20,70-dichlorodihydrofluoresceindiacetate (CM-DCFDA; Molecular Probes C6827) for20 min at 37�C, 5% CO2, using 5 lg/ml Hoechst 33258(Molecular Probes H1398) as a control dye. In our hands,CM-DCFA was better retained in living cells than itsrelated 20,70-dichlorodihydrofluorescein diacetate (DCFDA),from which it is derived. Fluorescence was measured usinga Fluoroskan Ascent plate fluorimeter (Thermo LabSystems, Altrincham, UK), and data were expressed as anormalized percentage of CM-DCFDA/Hoechst fluorescencein controls. Excitation and emission wavelengths for CM-DCFDA and Hoechst were as suggested by the supplier. Allexperiments were performed in duplicate and data wereplotted as the mean of at least three independent experi-ments6 SEM.

Measurement of Mitochondrial Potential

Oligodendrocyte and neuron cultures were exposed toAMPA or kainate receptor agonists as above. Immediatelythereafter, cells were loaded with 100 nM tetramethylrho-damine ethyl ester (TMRE; Molecular Probes T669) and 1lM calcein AM (Molecular Probes C3100) for 10 min at37�C, 5% CO2. TMRE is a highly membrane-permeantcationic fluorophore that accumulates in negativelycharged subcellular compartments, notably mitochondria.Under nonquenching conditions, quantifying the retentionof the dye by whole cells provides an estimation of theiraverage mitochondrial potential (Bernardi et al., 1999).Calcein fluorescence, a common method used to test cellviability, was used to quantify the number of cells presentwithin the reading field. We chose a plating density of10,000 cells/well in order to minimize TMRE backgroundfluorescence levels (Scaduto and Grotyohann, 1999).Fluorescence was measured using a Fluoroskan Ascentplate fluorimeter (Thermo Lab Systems, Altrincham, UK),and the data were expressed as a normalized percentageof TMRE/calcein fluorescence in controls. Excitation andemission wavelengths for TMRE and calcein were as sug-gested by the supplier. All assays were performed in dupli-cate and the values are the average of at least three inde-pendent experiments (mean 6 SEM).

Measurement of [Ca21]i

The concentration of cytoplasmic calcium ([Ca21]i) wasdetermined by microfluorimetry. Cells were incubatedwith fura-2 AM (Molecular Probes, Eugene, OR) at 5 lMin culture medium for 30–45 min at 37�C, and werewashed of excess fura-2 AM by incubating in HBSS con-taining 20 mM HEPES [pH 7.4], 10 mM glucose, and2 mM CaCl2 (incubation buffer) for 5 min at room tem-perature. Experiments were carried out in a coverslipchamber, continuously perfused with incubation buffer at2 ml/min. The perfusion chamber was mounted on thestage of a Zeiss (Oberkochen, Germany) inverted epi-

fluorescence microscope (Axiovert 35), equipped with a150 W xenon Polychrome IV lamp (T.I.L.L. Photonics,Martinsried, Germany) and a Plan Neofluar 40X oilimmersion objective (Zeiss). Cells were visualized with ahigh-resolution digital B/W CCD camera (ORCA), andimage acquisition and data analysis were performedusing the AquaCosmos software program (Hamamatsu,Iberica, Spain). At the end of the assay, in situ calibra-tion was performed with the successive addition of10 mM ionomycin and 2 M Tris-50 mM EGTA, pH 8.5.The [Ca21]i concentration was estimated by the 340/380ratio method, using a Kd value of 224 nM.

Detection of Active Caspase 3

Oligodendrocytes were exposed to AMPA 10 lM for5 min in the presence or absence of polyphenols and fixed 1h thereafter with 4% paraformaldehyde. Activation of cas-pase 3 was evaluated by immunofluorescence, using a pri-mary polyclonal antibody that specifically recognizes theactive form of this protease (Cell Signaling, D175; 1:100from stock) followed by Alexa Fluor 488 goat-anti rabbitsecondary antibody (Molecular Probes, A-11008; 1:200).Immunoreactive cells were counted and data were plottedas a percentage with respect to controls, nontreated cells.Negative controls included omission of the primary anti-body and no staining was observed under these conditions.

Cell Viability and Toxicity Assays

Cells were exposed to agonists as described and thenwere incubated for 24 h in fresh medium. To quantify oli-godendrocyte viability, these were loaded with 1 lM cal-cein-AM (Molecular Probes C3100) and fluorescence wasmeasured using a Fluoroskan Ascent plate fluorimeter(Thermo Lab Systems, Altrincham, UK). Excitation andemission wavelengths were as suggested by the supplier.To quantify neuron viability, the MTT assay (Mossnan,1983) was employed, using neuronal cultures at highdensity (300.000 cells/well). Neurons were incubatedwith tetrazolium salt, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (1 mg/ml; MTT; Sigma) for1 h in culture medium, the formazan precipitate wassolubilized thereafter with DMSO and absorbance of thecolored reaction quantified at 570 nm. All assays wereperformed in duplicate and data were plotted as mean 6SEM of at least three independent experiments.

Data Analysis

All data are expressed as mean 6 SEM. The student ttest was performed to ascertain whether the differencesbetween two experimental conditions were statisticallysignificant, considering them so when P < 0.05. For ana-lysis of correlation between two variables, Pearson’sdetermination coefficient (r square) and P-values werecalculated for n 5 4 as the number of different exposure

203EXCITOTOXICITY AND OXIDATIVE STRESS IN OLIGODENDROGLIA

paradigms tested. The coefficient r2, taking values from 0to 1, shows the extent of variation of one variable whichis explained by the variation in the other variable. Thecorrelation was considered significant when P < 0.05.The data were analyzed with Excel (Microsoft, Seattle,WA) and Prism (Lake Forest, CA) software.

RESULTSActivation of AMPA and Kainate ReceptorsInduces Differential Oxidative Stress in

Oligodendrocytes

Immunocytochemical characterization of optic nerve oli-godendrocyte cultures used in these assays showed thatthe vast majority of cells (>98%) were GalC1 (Fig. 1A,B).We monitored oxidative stress in these cultures by meansof the free radical sensitive probes CM-H2TMRos and CM-DCFDA (Fig. 1). We observed that simultaneous activa-tion of both AMPA and kainate receptors, using kainatealone (10 lM) induced intense CM-H2TMRos fluorescencewhich was not observed in control, nonstimulated cultures(cf. Fig. 1C,D). Likewise, selective sustained activation ofAMPA receptors elicited a marked increase in the numberof cells strongly labeled with CM-DCFDA in comparisonwith control, nonstimulated cultures (Fig. 1E–H).

To quantify the levels of ROS generated by activationof these receptors, we employed experimental paradigmsrepresenting maximal and submaximal activation ofAMPA receptors and of the high- and low-affinity kainatereceptors known to mediate oligodendrocyte excitotoxi-city (S�anchez-G�omez and Matute, 1999). Toxicity assaysconfirmed that the levels of oligodendrocyte death trig-gered by AMPA and kainate receptors under those excito-toxic paradigms (Fig. 2A) were identical to those pre-viously reported (S�anchez-G�omez and Matute, 1999). Inaddition, quantitative assessment in oligodendrocytes ofCM-DCFDA fluorescence showed that the selective acti-vation of AMPA receptors, both at submaximal (10 lM)and maximal (100 lM) concentration of the agonist,caused a three- to fourfold increase in ROS levels. In con-trast, activation of high-affinity (KA 3 lM) and low-affi-nity (KA 3 mM) kainate receptors induced a 15–20% risein those levels (Fig. 2C). Interestingly, we observed ahigh correlation (r2 5 92.01%; P 5 0.04) between theROS levels measured in the present study and the peakincrease in [Ca21]i after AMPA or kainate receptor acti-vation reported previously (S�anchez-G�omez et al., 2003),which suggests that free radical production by oligoden-drocytes after excitotoxic insults is related to increases inthe intracellular concentration of [Ca21]i.

Since the generation of ROS in excitotoxic insults isassociated with alterations in mitochondria, we nextmeasured mitochondrial membrane potential usingTMRE as a fluorescent probe. Oligodendrocytes exposedto AMPA for 15 min underwent a 60–65% reduction ofTMRE uptake, indicating a severe reduction in the mem-brane potential of mitochondria (Fig. 2D). In contrast,this reduction was much smaller (about 10%) in magni-tude after activation of high- and low-affinity kainate

receptors. These changes inversely correlate with theincrease in ROS levels (r2 5 97.35%; P 5 0.01), whichindicates that free radical production is related to depo-larization of their mitochondria in oligodendrocytes.

In contrast, mitochondrial potential and ROS levelswere found to be less affected by selective activation ofhigh- and low-affinity kainate receptors than by submax-imal activation of AMPA receptors, though these threeexcitotoxic insults cause similar levels of oligodendroglial

Fig. 1. Activation of AMPA/kainate receptors causes oxidative stressin optic nerve oligodendrocytes. A: Immunocytochemical characteriza-tion of optic nerve oligodendrocyte cultures shows that the vast major-ity of cells stain for the myelin lipid galactocerebroside. B: Chromatinstaining with Hoechst 33258 of the field shown in A. C: Control oligo-dendrocytes labeled with the ROS-sensitive probe CM-H2TMRos (200nM; 15 min), showing the typical accumulation of the probe withinmitochondria. D: Oligodendrocytes exposed to 10 lM kainate, and sub-sequently labeled as in C. Note dramatic changes in staining includingthe appearance of bright spots corresponding to free radical generation.E: Control oligodendrocytes labeled with the ROS-sensitive probe CM-DCFDA. F: Note the overall increase in CM-DCFDA staining when cul-tures were exposed to 10 lM AMPA for 15 min. G: Higher magnifica-tion of AMPA-exposed and subsequently CM-DCFDA labeled oligoden-drocytes. H: Hoechst 33258 of the field shown in G indicates chromatincondensation in cells with elevated ROS. The arrow in G,H points to acell with a low level of ROS and with apparently normal chromatin.(This figure can be viewed in color online at www.interscience.wiley.com.)

204 IBARRETXE ET AL.

death. This indicates that different death mechanismsare triggered by AMPA and kainate receptors in oligoden-drocytes.

Oligodendrocytes Accumulate Higher ROSLevels Than Do Cortical Neurons After

Activation of AMPA Receptors

Neurons, like oligodendrocytes, are also vulnerable toexcitotoxic insults mediated by AMPA receptors (Car-riedo et al., 2000). Thus, we next examined to whatextent these insults induced oxidative stress and depolar-ization of the mitochondrial membrane in pure culturesof cortical neurons. Surprisingly, we found that both sub-maximal and maximal activation of AMPA receptors for

15 min elicited only a modest increase (about 20%) inROS above control levels (Fig. 3A). In contrast, thereduction of the mitochondrial membrane potential inneurons after exposure to AMPAwas similar to that mea-sured in oligodendrocytes (Fig. 3B). Together, theseresults indicate that the consequences of excitotoxicinsults on the mitochondrial potential in neurons and oli-godendrocytes are very similar, but that the levels ofROS accumulated by oligodendrocytes are much higherthan that found in neurons.

Moderate But Not Severe AMPA-ReceptorMediated Oligodendrocyte Death Is PartiallyPrevented by Two Polyphenolic Antioxidants

Mangiferin is the main active ingredient of Vimang, aMangifera indica extract of Cuban origin, which has beenreported to have strong antioxidant and anti-inflamma-tory properties (Mart�ınez et al., 2000; Garrido et al.,2001). Morin is ubiquitous in vegetables, berries andfruits (Ross and Kasum, 2002) and its antioxidant proper-ties have also been described elsewhere (Kok et al., 2000).

Initially, we assayed the antioxidant properties of man-giferin and morin (Fig. 4A) by measuring their ability toscavenge the free-radical 2,2-diphenyl-1-picrylhydrazyl(DPPH·) in an ethanol solution (Aruoma, 2003; Li et al.,2003). Both mangiferin and morin (1 mM) scavengedDPPH· in a significant way compared with control sam-ples tested in their absence (Fig. 4B). The antioxidantpower in this assay was found to be higher for morin thanfor mangiferin, which decreased absorbance at 517 nm byabout 15% and 8%, respectively.

We then tested whether mangiferin and morin attenu-ated oligodendrocyte death caused by excitotoxic insults.

Fig. 2. Excitotoxicity and oxidative stress in oligodendrocytes. Cellswere cultured from the optic nerve and exposed for 15 min to agonists.A: Cell viability assessed by calcein fluorimetry 24 h after selective acti-vation of AMPA and kainate receptors. B: Increased [Ca21]i elicited bythese excitotoxic insults, as measured by Fura-2 AM microfluorimetry.Histogram illustrates the peak response (mean 6 SEM) of 10–15 cellsfrom 3–5 different cultures. C,D: AMPA receptor activation causes highlevels of ROS and large mitochondrial depolarization as assessed CM-DCFDA (10 lM) and TMRE (100 nM) fluorescence, respectively. In con-trast, these parameters were much less altered after kainate receptoractivation. Values represent the mean 6 SEM of duplicates-triplicatesfrom three to five different experiments and are normalized to Hoechst33258 fluorescence (C) or to calcein fluorescence (D), both used as acontrol of cells the cell number measured in each field. 100% representscontrol values in the absence of agonists. E,F: Mitochondrial potentialand oxidative stress show a negative linear correlation whereas thecorrelation between [Ca21]i and oxidative stress is positive. *P < 0.05;**P < 0.01; ***P < 0.001. Student’s t-test.

Fig. 3. Oxidative stress and mitochondrial potential in neurons.Cells were cultured from the cerebral cortex and exposed for 15 minto agonists. A: AMPA receptor activation in neurons only induced amild increase in CM-DCFDA (10 lM) fluorescence indicating that lowlevels of ROS are generated after the insult. B: In contrast, mito-chondrial potential, as measured with TMRE (100 nM) fluorescence,was greatly diminished under the conditions assayed. Values repre-sent the mean 6 SEM of duplicates/triplicates from three differentexperiments and were normalized to Hoechst 33258 fluorescence (A)or to calcein fluorescence (B). *P < 0.05; **P < 0.01; ***P < 0.001.Student’s t-test.

205EXCITOTOXICITY AND OXIDATIVE STRESS IN OLIGODENDROGLIA

Dose-response curves showed that both polyphenols, at 1nM to 100 lM, did not improve oligodendrocyte survivalafter 15 min activation of AMPA or kainate receptors.However, submicromolar concentrations of mangiferinand morin displayed a marked protective effect againstshorter (5 min) activation of AMPA receptors with AMPA10 lM (Fig. 5A), but not at AMPA 100 lM (Fig. 5B). Bothantioxidants were most effective in preventing oligoden-drocyte death at 1–10 nM. In contrast, no protection wasobserved after 5-min activation of kainate receptors withhigh and low affinity (Fig. 5C).

Fig. 4. Mangiferin and morin are natural polyphenols that displayantioxidant activity. A: Chemical structure of mangiferin and morin.B: Curves showing the relative antioxidant potency of these two spe-cies, assayed by in vitro scavenging of the DPPH radical. Under theseconditions, morin displayed higher antioxidant power than mangiferin.Absorbance at 517 nm was measured every 10 min after the beginningof the reaction. *P < 0.05; **P < 0.01; *** P < 0.001. Student’s t-test(n 5 3).

Fig. 5. Morin and mangiferin substantially reduce oligodendrocytedeath caused by submaximal activation of AMPA receptors. Cultures wereincubated with agonists for 5 min and cell viability assessed by calceinfluorimetric 24 h later. Polyphenols were added 24 h before agonist expo-sure and until the end of the experiments. A,B: Dose-response curves ofthe protective effects of mangiferin and morin after a 5 min incubationwith AMPA 10 lM and 100 lM respectively. Both polyphenols at submi-cromolar concentrations protected against moderate excitotoxic insults. C:Mangiferin and morin did not improve cell viability after insults mediatedby high- and low-affinity kainate receptor activated by 3 lM and 3 mMkainate, respectively, in the presence of GYKI53655 (100 lM). Valuesrepresent the mean 6 SEM of duplicates-triplicates from three differentexperiments. *P < 0.05; **P < 0.01; ***P < 0.001. Student’s t-test.

206 IBARRETXE ET AL.

Both Mangiferin And Morin ReduceOligodendrocyte Apoptosis by Different

Mechanisms

Oligodendroglial death mediated by AMPA at 10 lMhas apoptotic features and is dependent on caspase 3

activation (S�anchez-G�omez et al, 2003). Because ofthat, we evaluated whether incubating oligodendrocyteswith these polyphenols diminished the amount of cellspresenting activation of caspase 3, as a consequence ofan exposure to AMPA 10 lM for 5 min. For this pur-pose, we used immunocytochemical staining with anantibody that specifically recognizes the active form ofcaspase 3. The result showed that both mangiferin andmorin significantly diminished the amount of cells pre-senting active caspase 3 in comparison with sister cul-tures treated with AMPA in the absence of polyphenols(Fig. 6). Active caspase 3 was present in oligodendro-cytes displaying chromatin condensation, and thusfurther indicating that they were undergoing apoptosis(Fig. 6A).

We then studied whether mangiferin and morin atte-nuated oligodendrocyte apoptosis by reducing ROSlevels, as expected from their antioxidant activity, o byinterfering with other events in the apoptotic cascadeinitiated by excitotoxicity in oligodendrocytes, such asdepolarization of the mitochondrial membrane potentialand [Ca21]i overload (S�anchez-G�omez et al., 2003). Inter-estingly, these experiments revealed that mangiferin andmorin have a differential effect over each of these para-meters. Thus, only morin diminished ROS levels andsupported a substantial recovery of the mitochondrialpotential in oligodendrocytes after exposure to AMPA10 lM for 5 min (Fig. 7A and B, respectively). In con-trast, mangiferin, but not morin, consistently diminished(about 35%) the increase in [Ca21]i elicited by AMPA 10 lM(Fig. 7C,D). Morin was also effective in reducing ROSlevels and recovering mitochondrial potential after expo-sure to 100 lM AMPA (Fig. 7A,B), which causes caspase-independent cell death (S�anchez-G�omez et al., 2003).However, this polyphenol does not attenuate cell deathcaused by this experimental condition, as illustratedabove (Fig. 5B).

Mangiferin and Morin Also Protect CorticalNeurons Against Moderate AMPA Receptor-

Induced Excitotoxicity

Excitotoxicity induce similar alterations in neuronsand oligodendrocytes relating cell Ca21 overload, mito-chondrial depolarization and free radical production,though oxidative stress is less intense in neurons under

Fig. 6. Mangiferin and morin partially block oligodendroglial apop-tosis mediated by exposure to AMPA 10 lM. Oligodendrocyte cultureswere exposed to AMPA for 5 min in the presence and absence of poly-phenols, and cells were fixed 1 h later. Activation of caspase 3 was eval-uated by counting of cells stained with an antibody that recognizes theactive form of this protease. Mangiferin and morin were added 24 hbefore agonist exposure and until the end of the experiments. A: Repre-sentative fields showing Hoechst labeled nuclei (left) and caspase 31

cells (right). Notice that cells showing chromatin condensation (arrows)are also caspase 31. B: Histogram representing the amount of caspase31 cells in the conditions indicated. Both antioxidants reduce the numberof apoptotic cells observed after incubation with AMPA 10 lM. Valuesrepresent the mean 6 SEM of duplicates-triplicates from three differentexperiments. *P < 0.05; **P < 0.01; ***P < 0.001. Student’s t-test. (Thisfigure can be viewed in color online at www.interscience.wiley.com.)

207EXCITOTOXICITY AND OXIDATIVE STRESS IN OLIGODENDROGLIA

the conditions assayed in this study. Because of that, westudied whether mangiferin and morin would attenuateexcitotoxic neuronal death under the conditions usedwith optic nerve oligodendrocytes. We observed that bothmangiferin and morin were also highly protective in cor-tical neurons following a 5 min exposure to AMPA 10 lM(Fig. 8A). As in oligodendrocytes, the maximal death-res-cuing effect was at around 10 nM of antioxidant, andneuronal excitotoxic death reduced to a 30% of thatoccurring in controls. Both mangiferin and morin werealso protective after longer (15 min) incubation with theagonist, a condition that was not effective in oligodendro-cytes (see above). In contrast, neither morin nor mangi-ferin attenuated neuronal death at any of the tested con-centrations when cultures were exposed to AMPA100 lM (data not shown).

In addition, since these polyphenolic species can affect[Ca21]i load in oligodendrocytes after excitotoxic insults,we evaluated whether a similar mechanism might occurin cortical neurons. We found that [Ca21]i in these cellsfollowing an exposure to AMPA 10 lM was similarlyreduced around 20% by application of both mangiferinand morin at a concentration of 10 nM (Fig. 8B,C).

DISCUSSION

We have demonstrated that excitotoxic insults media-ted by AMPA and kainate receptors in oligodendrocytesinduce differential oxidative stress, which correlates withincreases in the intracellular calcium concentration andwith the degree of depolarization of the mitochondrial mem-

brane. Strikingly, oligodendrocytes accumulate higher ROSlevels than cortical neurons upon sustained activation ofAMPA receptors. In addition, we observed that low concen-trations of mangiferin and morin, two natural polyphenols,improve oligodendrocyte and neuronal viability againstmoderate excitotoxic insults that, in oligodendrocytes, areassociated with apoptotic cell death. The mechanismsaccounting for these effects include reduction of oxidativestress and attenuation of Ca21 overload.

Differential oxidative stress is induced by excitotoxicinsults in oligodendrocytes Our results show that opticnerve oligodendrocytes exposed to AMPA undergo a dra-matic increase in their levels of ROS and an intense depo-larization of their mitochondrial membrane. Thesechanges correlate with the raise of [Ca21]i subsequent toreceptor activation. However, activation of both high- andlow-affinity kainate receptors induces only minor changesin these parameters. These differences are striking sinceprolonged activation of either AMPA or kainate receptorsis toxic to oligodendrocytes, indicating that oligodendro-cyte death mediated by these receptors occurs via distinctmechanisms. Thus, moderate activation of AMPA recep-tors or high- and low-affinity kainate receptors causesimilar levels of oligodendrocyte toxicity as shown in thisstudy and in a previous report (S�anchez-G�omez andMatute, 1999). However, toxicity after maximal activationof AMPA receptors is higher than under sub-maximalAMPA stimuli and yet depolarization of the mitochondrialmembrane and the generation of ROS are similar. Conse-quently, AMPA and kainate receptor-mediated oxidativestress and mitochondrial depolarization do not correlateequivalently with oligodendrocyte death.

Fig. 7. Mangiferin and morindifferentially affect ROS produc-tion, loss of mitochondrial potentialand [Ca21]i overload induced byactivation of AMPA receptors in oli-godendrocytes. A: Levels of ROSafter exposure to AMPA 10 lM for5 min were reduced by morin but notby mangiferin. B: Likewise, morinbut not mangiferin attenuated theloss of mitochondrial membrane po-tential. Values in A and B representthe mean 6 SEM of duplicates-tri-plicates from three different experi-ments. In all instances, polyphenolswere added 24 h before agonistexposure and until the end of theexperiments. C,D: [Ca21]i increasesinduced by AMPA 10 lM are reducedby mangiferin, but not by morin.Curves in C illustrate the timecourse of the [Ca21]i increase (mean 6SEM) with respect to basal valuesbefore incubation with the agonist.Histogram in D illustrates the peakresponse (mean 6 SEM) of 38–53cells from 3–5 different cultures.Mangiferin, but not morin, diminishessignificantly AMPA-mediated [Ca21]ioverload. *P < 0.05; **P < 0.01.Student’s t-test.

208 IBARRETXE ET AL.

Excitotoxic death is dependent primarily on Ca21 entryinto the cell by ionotropic glutamate receptors (Alberdiet al., 2002; Krieger and Duchen, 2002). An increase incytosolic Ca21 concentration depolarizes mitochondriawhich possess a high-affinity Ca21 uniporter on theirinner membrane (Kirichok et al., 2004). Accordingly, acti-vation of AMPA and kainate receptors in oligodendro-cytes induces a large increase in the concentration ofcytosolic Ca21, the bulk of which is rapidly sequestered bymitochondria (S�anchez-G�omez et al., 2003). Notably, cyto-solic Ca21 overload in oligodendrocytes is much higherafter activation of AMPA receptors in comparison withkainate receptors, a feature which determines the magni-tude of the depolarization of the mitochondrial membraneand the ensuing levels of ROS generated, as observed inthis study. Taken together, these results suggest that oxi-dative stress after excitotoxic insults in oligodendrocytesis dependent on Ca21 uptake by mitochondria.

Differential Oxidative Stress Is Induced by AMPAReceptor Activation in Oligodendrocytes and

Cortical Neurons

An important finding of this work is that optic nerveoligodendrocytes generate many more ROS than corticalneurons upon sustained activation of AMPA receptors.These differences are unlikely to be due to differentlevels of AMPA receptor expression, since we found simi-lar levels of depolarization in the mitochondrial mem-brane and of [Ca21]i increases in both cell types. Thus,an intrinsic factor in oligodendrocytes renders these cellsespecially sensitive to oxidative stress caused by activa-tion of AMPA receptors. This factor may be the higherlevels of iron, which, in oligodendrocytes, acts as anessential cofactor for the synthesis of myelin (Connor andMenzies, 1996). Indeed, high levels of free iron are pre-sent in the mitochondrion, since the final assembly of fer-rous iron to protein leading to synthesis of heme groupsnaturally takes place in this organelle (Dailey, 2002).However, high concentrations of free ferrous iron are alsolinked to oxidative stress by its capability to catalyze theFenton reaction (Liochev and Fridovich., 1999), whichgenerates hydroxyl radicals and cause lipid peroxidation.Thus, oxidative stress due to excessive calcium uptakeinto mitochondria could be further amplified in oligoden-drocytes by the iron-catalyzed generation of free radicalspecies occurring in the Fenton reaction. This mayaccount for the enhanced accumulation of ROS weobserved in oligodendrocytes as compared with neuronssubjected to the same excitotoxic insults.

Mangiferin and Morin Protect Against ModerateInsults Mediated by AMPA Receptors and Inhibit

Oligodendroglial Apoptosis

Since we found that oligodendrocytes generate a greatamount of free radicals when exposed to AMPA, weexamined the protective potential of antioxidants. Wefocused on the use of two polyphenols, mangiferin and

Fig. 8. Mangiferin and morin protect cortical neurons against mod-erate excitotoxic insults mediated by AMPA receptors and attenuate[Ca21]i overload induced by activation of these receptors. A: Mangiferinand morin protect against AMPA (10 lM) toxicity. B,C: Mangiferin andmorin reduced the [Ca21]i overload provoked by exposure to this ago-nist. Values in A represent the mean 6 SEM of triplicates from six dif-ferent experiments and in B and C, the mean 6 SEM from 35–51 cellsfrom at least three different cultures. Histogram in C depicts thechanges in the peak amplitude of [Ca21]i. In all cases, polyphenols(10 nM) were present 24 h before agonist exposure and until the end ofthe experiments. *P < 0.05; **P < 0.01; ***P < 0.001. Student’s t-test.

209EXCITOTOXICITY AND OXIDATIVE STRESS IN OLIGODENDROGLIA

morin, which are natural constituents of several plantspecies. These antioxidants were tested in oligodendro-cyte and in neuron cultures, which undergo differentialoxidative stress under the excitotoxic insults assayed.Both mangiferin and morin protected oligodendrocyteand neuron cultures from moderate but not stronginsults mediated by AMPA receptors. Interestingly, theseagents are most effective at very low concentrations,within the nanomolar range, a concentration that couldbe achieved in the CNS in potential future clinical trials,due to the limited permeability of the blood-brain barrier.

In oligodendrocytes, mild excitotoxic insults mediatedby AMPA receptors induce apoptotic cell death which isdependent on caspase 3 activation, whereas deathmediated by intense insults is caspase-independent andlikely of necrotic nature (S�anchez-G�omez et al., 2003).We wondered whether the effect of these polyphenols inimproving oligodendrocyte survival could be attributed totheir ability to prevent apoptosis. Thus, we evaluatedcaspase 3 activation by immunocytochemistry in oligo-dendrocytes exposed to AMPA 10 lM and found that bothmangiferin and morin inhibit the activation of this celldeath effector caspase.

The Mechanisms Underlying Neuroprotection byMorin and Mangiferin Are Different

The neuroprotective properties of mangiferin andmorin are diverse. Initially, we postulated that theseantioxidants would attenuate cell death by reducing oxi-dative stress ensuing excitotoxic insults. Indeed, weobserved in this study that both polyphenols have antiox-idant activity, which is higher for morin. This finding isconsistent with the idea that glycosylated polyphenols,such as mangiferin, are less active than their relatedaglycones in terms of antioxidant potency (Kim et al.,2002; Hou et al., 2004). Accordingly, morin reduced oxi-dative stress as well as the loss of mitochondrial poten-tial ensuing submaximal activation of AMPA receptor inoligodendrocytes, which suggests that protection in thesecells is mainly due to its antioxidant properties. Unex-pectedly, mangiferin did not reduce ROS levels or the lossof mitochondrial potential. However, this polyphenolreduces the increase of [Ca21]i subsequent to sustainedAMPA receptor activation both in oligodendrocytes andneurons, a feature that may underlie neuroprotection bymangiferin in both cell types. In addition, morin alsoattenuates [Ca21]i overload following mild insults chan-neled by AMPA receptors, a property that may contributeto reduced cell death in neurons after excitotoxic condi-tions that do not generate intense oxidative stress. Thefinding that these polyphenols can affect the homeostasisof Ca21 in oligodendrocytes and neurons constitutes anovel and strong argument that supports their oligoden-droprotective and neuroprotective potential.

Oxidative Stress and Demyelinating Diseases

Oligodendrocyte and myelin loss are two of the princi-pal pathological hallmarks of demyelinating diseases,

including multiple sclerosis. Interestingly, oxidativestress and in vivo oxidation markers have been identifiedin white matter tissue of patients affected by this disease(Smith et al., 1999), along with disturbances in gluta-mate homeostasis, which correlate with the severity ofrelapses (Matute et al., 2001; Groom et al., 2003; Sarch-ielli et al., 2003). In MS, this oxidative stress is asso-ciated with inflammatory incidents, such as the releaseof free radicals from activated immune cells (Gilgun-Sherki et al., 2004). Free radicals are also generated as aconsequence of sustained activation of oligodendrocyteAMPA/kainate receptors due to subtle increases in theconcentration of extracellular glutamate (Li et al., 2003).These findings, together with those reported in the pre-sent study, indicate that oxidative stress may representan important component in the pathophysiology of oligo-dendrocytes in demyelinating diseases such as MS.

In summary, we provide evidence that activation ofAMPA and kainate receptors triggers differential levels ofoxidative stress in oligodendrocytes, and that excitotoxicinsults to these cells results in the production of higherlevels of ROS than in neurons. The polyphenolic antioxi-dants mangiferin and morin partially prevent AMPA-induced cell death in both cell types by mechanisms thatinclude attenuation of Ca21 overload. Overall, theseresults suggest that these polyphenols may be useful com-ponents in therapies for neurodegenerative diseases,including demyelinating diseases, in which disturbancesin glutamate homeostasis and oxidative stress occur.

ACKNOWLEDGMENTS

The authors sincerely thank Dr. Fernando Vidal-Vana-clocha and Rogelio Arellano-Ostoa for providing access toequipment in their laboratories. We are also grateful toDr. Leander (Elli Lilly and Company, Indianapolis) forproviding GYKI 53655 and to David Fogarty for havingaided to improve this manuscript. This work was sup-ported by grants from the Spanish Ministry of Healthand Consumption, La Caixa, the Basque Government,and the Basque Country University. G.I. was supportedby the Basque Government, M.R.C-E. by CONYTEA andthe Carolina Foundation, and E.A. by a Ram�on y Cajalcontract from the Spanish Ministry of Education andScience.

REFERENCES

Alberdi E, S�anchez-G�omez MV, Marino A, Matute C. 2002. Ca21 influxthrough AMPA or kainate receptors alone is sufficient to initiate exci-totoxicity in cultured oligodendrocytes. Neurobiol Dis 9:234–243.

Aruoma OI. 2003. Methodological considerations for characterizingpotential antioxidant actions of bioactive components in plant foods.Mutat Res 523/524:9–20.

Barres BA, Hart IK, Coles HS, Burne JF, Voyvodic JT, Richardson WD,Raff MC. 1992. Cell death and control of cell survival in the oligoden-drocyte lineage. Cell 70:31–46.

Bernardi P, Scorrano L, Colonna R, Petronilli V, Di Lisa F. 1999. Mito-chondria and cell death. Mechanistical aspects and methodologicalissues. Eur J Biochem 264:687–701.

Carriedo SG, Sensi SL, Yin HZ, Weiss JH. 2000. AMPA exposuresinduce mitochondrial Ca21 overload and ROS generation in spinalmotor neurons in vitro. J Neurosci 20:240–250.

210 IBARRETXE ET AL.

Choi DW. 1995. Calcium: still center-stage in hypoxic-ischemic neuronaldeath. Trends Neurosci 18:58–60.

Connor JR, Menzies SL. 1996. Relationship of iron to oligodendrocytesand myelination. Glia 17:83–93.

Dailey HA. 2002. Terminal steps of haem biosynthesis. Biochem SocTrans 30:590–595.

Dawson VL, Dawson TM, Bartley DA, Uhl GR, Snyder SH. 1993.Mechanisms of nitric oxide-mediated neurotoxicity in primary braincultures. J Neurosci 13:2651–2661.

Duchen MR. 2000. Mitochondria and Ca21: from cell signalling to celldeath. J Physiol 529:57–68.

Ermark G, Davies K. 2001. Calcium and oxidative stress: from cell sig-naling to cell death. Mol Immun 38:713–721.

Garrido G, Gonz�alez D, Delporte C, Backhouse N, Quintero G, Nu~nez-Selles AJ, Morales MA. 2001. Analgesic and anti-inflammatory effectsof Mangifera indica L. extract (Vimang) Phytother Res 15:18–21.

Groom AJ, Smith T, Turski L. 2003. Multiple sclerosis and glutamate.Ann NY Acad Sci 993:229–275.

Hajn�oczky G, Davies E, Madesh M. 2003. Calcium signaling and apop-tosis. Biochem Biophys Res Commun 304:445–454.

Hou L, Zhou B, Yang L, Liu ZL. 2004. Inhibition of human low densitylipoprotein oxidation by flavonoids and their glycosides. Chem PhysLipids 129:209–219.

Kim SR, Park MJ, Lee MK, Sung SH, Park EJ, Kim J, Kim SY, OhTH, Markelonis GJ, Kim YC. 2002. Flavonoids from Inula britannicaprotect cultured cortical cells from necrotic cell death induced by glu-tamate. Free Radic Biol Med 32:596–604.

Kirichok Y, Krapivinsky G, Clapham DE. 2004. The mitochondrial cal-cium uniporter is a highly selective ion channel. Nature 427:360–364.

Kok LD, Wong YP, Wu TW, Chon HC, Kwok TT, Fung KP. 2000. Morinhydrate: a potential antioxidant in minimizing the free-radicals-mediated damage to cardiovascular cells by anti-tumor drugs. LifeSci 67:94–99.

Krieger C, Duchen MR. 2002. Mitochondria, Ca21 and neurodegenera-tive disease. Eur J Pharmacol 447 :177–188.

Kweon SM, Kim HJ, Lee ZW, Kim SJ, Kim SI, Paik SG, Ha KS. 2001.Real-time measurement of intracellular reactive oxygen species usingMitotracker orange (CMH2TMRos) Biosci Rep 21:341–352.

Li J, Lin JC, Wang H, Peterson JW, Furie BC, Furie B, Booth SL, VolpeJJ, Rosenberg PA. 2003. Novel role of vitamin K in preventing oxida-

tive injury to developing oligodendrocytes and neurons. J Neurosci23:5816–5826.

Liochev SI, Fridovich I. 1999. Superoxide and iron: partners in crimeIUBMBLife 48:157–161.

Mart�ınez G, Delgado R, P�erez G, Garrido G, Nu~nez-Selles AJ, Le�on-OS.2000. Evaluation of the in vitro antioxidant activity of Mangiferaindica L. extract (Vimang). Phytother Res 14:424–427.

Matute C, S�anchez-G�omez MV, Mart�ınez-Mill�an L, Miledi R. 1997. Glu-tamate receptor-mediated toxicity in optic nerve oligodendrocytes.Proc Natl Acad Sci 94:8830–8835.

Matute C, Alberdi E, Domercq M, P�erez-Cerd�a F, P�erez-Samart�ın A,S�anchez-G�omez MV. 2001. The link between excitotoxic oligodendro-glial death and demyelinating diseases. Trends Neurosci 24:224–230.

McDonald JW, Althomsons SP, Hyrc KL, Choi DW, Goldberg MP. 1998.Oligodendrocytes from forebrain are highly vulnerable to AMPA/kai-nate receptor-mediated excitotoxicity. Nat Med 4:291–297.

Mosmann T 1983. Rapid colorimetric assay for cellular growth and sur-vival: application to proliferation and cytotoxicity assays. J ImmunolMethods 65:55–63.

Ross JA, Kasum CM. 2002. Dietary flavonoids: bioavailability, meta-bolic effects, and safety. Annu Rev Nutr 22:19–34.

S�anchez-G�omez MV, Matute C. 1999. AMPA and kainate receptors eachmediate excitotoxicity in oligodendroglial cultures. Neurobiol Dis 6:475–485.

S�anchez-G�omez MV, Alberdi E, Ibarretxe G, Torre I, Matute C. 2003.Caspase-dependent and caspase-independent oligodendrocyte deathmediated by AMPA and kainate receptors. J Neurosci 23:9519–9528.

Sarchielli P, Greco L, Floridi A, Floridi A, Gallai V. 2003. Excitatoryamino acids and multiple sclerosis: evidence from cerebrospinal fluid.Arch Neurol 60:1082–1088.

Scaduto RC, Grotyohann LW. 1999. Measurement of mitochondrialmembrane potential using fluorescent rhodamine derivatives. Bio-phys J 76:469–77.

Smith KJ, Kapoor R, Felts PA. 1999. Demyelination: the role of reactiveoxygen and nitrogen species. Brain Pathol 9:69–92.

Sousa SC, Maciel EN, Vercesi AE, Castilho RF. 2003. Ca21 induced oxi-dative stress in brain mitochondria treated with the respiratory chaininhibitor rotenone. FEBS Lett 543:179–183.

Yoshioka A, Bacskai B, Pleasure D. 1996. Pathophysiology of oligoden-droglial excitotoxicity. J Neurosci Res 46:427–437.

211EXCITOTOXICITY AND OXIDATIVE STRESS IN OLIGODENDROGLIA