Thiamine Deficiency Results inDownregulation of the GLAST

Glutamate Transporter inCultured Astrocytes

ALAN S. HAZELL,1* PIERRE PANNUNZIO,1 K.V. RAMA RAO,1 DAVID V. POW,2AND ANDREA RAMBALDI1

1Department of Medicine, Hopital Saint-Luc, University of Montreal, Montreal, Quebec, Canada2Department of Physiology and Pharmacology, University of Queensland, Brisbane, Australia

KEY WORDS cell culture; pyrithiamine; glutamate transporter; excitotoxicity;metabotropic receptor

ABSTRACT Pyrithiamine-induced thiamine deficiency (TD) is a well-establishedmodel of Wernicke’s encephalopathy in which a glutamate-mediated excitotoxic mech-anism may play an important role in determining selective vulnerability. In order toexamine this possibility, cultured astrocytes were exposed to TD and effects on gluta-mate transport and metabolic function were studied. TD led to decreases in cellularlevels of thiamine and thiamine diphosphate (TDP) after 24 h of treatment and de-creased activities of the TDP-dependent enzymes �-ketoglutarate dehydrogenase andtransketolase after 4 and 7 days, respectively. TD treatment for 10 days led to areversible decrease in the uptake of [3H]-D-aspartate, a nonmetabolizable analogue ofglutamate. Kinetic analysis revealed that the uptake inhibition was caused by a 47%decrease in the Vmax for uptake of [3H]-D-aspartate, with no change in the Km value.Immunoblotting showed that this decrease in uptake was due to an 81% downregulationof the astrocyte-specific GLAST glutamate transporter. Loss of uptake activity andGLAST protein were blocked by treatment with the protein kinase C inhibitor H7, whileexposure to DCG IV, a group II metabotropic glutamate receptor (mGluR) agonist,resulted in improvement of [3H]-D-aspartate uptake and a partial reversal of trans-porter downregulation. These results are consistent with our recent in vivo findings of aloss of astrocytic glutamate transporters in TD and provide evidence that TD conditionsmay increase phosphorylation of GLAST, contributing to its downregulation. In addi-tion, manipulation of group II mGluR activity may provide an important strategy in thetreatment of this disorder. © 2003 Wiley-Liss, Inc.

INTRODUCTION

The importance of thiamine in the effective function-ing of the CNS is well established. A deficiency of thisvitamin in association with the administration of pyri-thiamine, a thiamine antagonist that inhibits both thi-amine transport and the enzyme thiamine pyrophos-phokinase, leads to profound structural lesions in focalregions of the rat brain, which closely recapitulate theneuropathology of Wernicke’s encephalopathy (WE)(Troncoso et al., 1981). Despite investigations thathave focussed on such parameters as the thiamine-

dependent enzymes pyruvate dehydrogenase complex(PDH), �-ketoglutarate dehydrogenase complex (KGDH),and transketolase, energy metabolism, receptor bind-

Grant sponsor: the CHUM Foundation (University of Montreal); Grant spon-sor: the Canadian Institutes of Health Research.

*Correspondence to: Dr. Alan S. Hazell, Department of Medicine, HopitalSaint-Luc (CHUM), 1058 St-Denis Street, Montreal, Quebec H2X 3J4, Canada.E-mail: alan.stewart.hazell@umontreal.ca

Received 6 December 2001; Accepted 13 March 2003

DOI 10.1002/glia.10241

GLIA 43:175–184 (2003)

© 2003 Wiley-Liss, Inc.

ing sites, the blood-brain barrier, and oxidative stress(Hazell et al., 1998a), the basis of the selective histo-logical damage due to pyrithiamine-induced thiaminedeficiency (TD) remains elusive.

On the other hand, previous studies have revealedthat extracellular glutamate concentrations are in-creased in vulnerable regions of the brain such as thethalamus at the symptomatic stage in TD (Hazell et al.,1993; Langlais and Zhang, 1993), suggesting an in-volvement of the glutamatergic system in the neuropa-thology of this disorder. In the brain, regulation ofextracellular glutamate concentration is a well-estab-lished feature of astrocytes (Schousboe et al., 1997).Astrocytic glutamate transport may be altered as aresult of metabolic changes and thus may play an im-portant role in the pathophysiology of TD-inducedbrain damage. Decreased activity of KGDH (Butter-worth et al., 1986; Sheu et al., 1998), a key rate-limit-ing enzyme of the tricarboxylic acid cycle, is an impor-tant contributor to the disruption of energy metabolismcharacteristic of TD (Gibson et al., 2000). Indeed, con-sequences of TD including decreased ATP levels(Aikawa et al., 1984) and depolarization (Hazell et al.,1998b) in brain have been shown to lead to impairmentof glutamate transport and decreased glutamate up-take in other systems (Kanner and Marva, 1982; Sza-tkowski et al., 1990) and may be attributable to thelowered KGDH activity.

Thus far, five different glutamate transporters havebeen cloned: GLT-1 (Pines et al., 1992), GLAST (Storcket al., 1992), EAAC1 (Kanai and Hediger, 1992),EAAT4 (Fairman et al., 1995), and EAAT5 (Arriza etal., 1997). Of these, GLT-1 and GLAST are known to belocalized primarily to astrocytes while EAAC1, EAAT4,and EAAT5 are expressed in neurons, predominantlyat extrasynaptic locations. Previous neuropathologicreports indicate that astrocytes are among the firstcells to be affected in thiamine deficiency (Collins,1967; Robertson et al., 1968) in advance of neuronalcell death, suggesting that the supporting frameworkfor neuronal activity may be compromised as a result ofastrocytic dysfunction.

Recently, we have demonstrated an in vivo down-regulation of GLT-1 and GLAST transporters in TD(Hazell et al., 2001). This finding provides a likelyexplanation for the increased interstitial glutamatelevels, and, along with evidence of NMDA receptorinvolvement (Langlais and Zhang, 1993), and regionalactivation of L-type voltage-sensitive calcium channels(Hazell et al., 1998b), strongly reinforce the likelihoodof a glutamate-mediated excitotoxic process being re-sponsible for the neuronal cell death in TD.

Previous cell culture studies on thiamine deficiencyin neural-derived cells have focussed mainly on theeffects of amprolium, an inhibitor of thiamine trans-port only (Bettendorff et al., 1995a, 1995b, 1997; Parket al., 2000). In another report, pyrithiamine was usedto produce TD in glioma and neuroblastoma cell lines(Schwartz et al., 1975). Since pyrithiamine-induced TDmost closely reproduces the characteristics of WE, and

in order to study the effects of this form of TD onastrocytes and glutamate transport, primary culturesof these cells were exposed to this thiamine antagonistand the effects on metabolism and glutamate transportwere investigated for up to 10 days. This study repre-sents the first report of the effects of pyrithiamine oncultured astrocytes.

MATERIALS AND METHODS

Pyrithiamine hydrobromide, protease inhibitor cock-tail, ATP, ouabain, ammonium molybdate, and anti-serum against actin were purchased from SigmaChemical (St. Louis, MO). Polyvinylidene difluoride(PVDF) membranes and broad-range protein markerswere purchased from Bio-Rad Laboratories (Hercules,CA), enhanced chemiluminescence (ECL) kits werepurchased from New England Nuclear (Boston, MA),and X-OMAT autoradiography film was purchasedfrom Kodak (Ile des Soeurs, Quebec, Canada). Purifiedrabbit polyclonal antisera to GLT-1 and glial fibrillaryacidic protein (GFAP) were purchased from Calbio-chem-Novabiochem (San Diego, CA) and Santa CruzBiotechnology (Santa Cruz, CA), respectively. Horse-radish peroxidase (HRP)-coupled antirabbit IgG waspurchased from Promega (Madison, WI). Other chemi-cals were purchased from Amersham Canada(Oakville, Ontario, Canada).

Cell Culture Preparation

Primary astrocyte cultures from newborn rats wereprepared using a modification of the method of Booherand Sensenbrenner (1972). Briefly, cerebral corticeswere removed and the tissue was dissociated, passedthrough sterile nylon sieves, and then suspended inDulbecco’s modified Eagle medium (DMEM; Life Tech-nologies, Burlington, Ontario, Canada) containing 10%fetal calf serum. Approximately 0.25 � 106 cells per mlwere seeded in 35 and 100 mm culture dishes, whichwere maintained in an incubator at 37°C provided witha mixture of 5% CO2 and 95% air. After 2 weeks, cellsattained confluency, at which point the fetal calf serumwas replaced by horse serum. Cultures were grown fora total of 3–5 weeks, during which the medium waschanged twice a week. At least 95% of cells were deter-mined to be astrocytes based on GFAP immunocyto-chemistry.

To induce TD, cells were exposed to a custom-de-signed DMEM media lacking in thiamine (Life Tech-nologies) and containing 5% horse serum, in the pres-ence of pyrithiamine. Initial studies indicated that thisconcentration of horse serum in TD media was suffi-cient to provide levels of thiamine well below the Kmvalue for high-affinity thiamine uptake. Control astro-cytes were treated with TD media in which normallevels of thiamine (11 �M) had been added. Dependingon the treatment time, growth media were changed

176 HAZELL ET AL.

every 3 or 4 days. Cultures were treated in a staggeredmanner, with all plates being harvested on the sameday for each experiment.

Measurement of Thiamine andThiamine Diphosphate

Thiamine was measured essentially as described byBettendorf et al. (1986) with minor modifications. As-trocyte cultures were harvested in 400 �l ice-cold po-tassium phosphate buffer 50 mM, pH 7.4, and soni-cated. Samples were then deproteinized with 75 �L ofcold TCA (50%) and centrifuged at 12,500 rpm for 20min at 4°C. The supernatant was extracted twice withfive volumes of water-saturated diethyl ether. The sol-vent delivery system for HPLC (model 501; Waters)was equipped with an automatic sampler (Varian9090) and fluorescence detection. Eighty microliters ofthe supernatant was derivatized by mixing 30 �L ofalkaline ferricyanide solution (4.3 mM potassium fer-ricyanide in 15% sodium hydroxide) prior to injectionon the column PRP-1 (150 mm � 4.1 mm; internaldiameter, 5 �m; Hamilton, Reno, NV). The system wasrun in an isocratic mode with 1.4% tetrahydrofuran in15 mM sodium phosphate buffer (pH 8.5). Flow rateswere 0.7 ml/min for thiamine determinations and 1.0ml/min for thiamine diphosphate (TDP). Standardcurves for thiamine and TDP were generated usingexternal standards prepared in the same extractionsolutions as the samples. Peak area measurementswere computed using the Baseline 810 program. Pro-tein content was determined by the method of Lowry etal. (1951).

Measurement of PDH Activity

The activity of pyruvate dehydrogenase (PDH) wasassayed by measuring the rate of reduction of iodo-nitrotetrazolium violet (INT) by NADH (Elnageh andGaitonde, 1988). Cells were harvested in homogenizingbuffer: 20 mM MOPS containing 0.2 mM mercapto-ethanol, 5 mM MgCl2, 0.1 mM CaCl2, 1 g/L TritonX-100, 1 g/L Lubrol-Px, and the pH was adjusted to 7.2with Tris. The reaction mixture contained 0.2 mMTDP, 2.5 mM NAD, 0.1 mM CoA, 1 mM MgCl2, 0.1 mMoxalate, 1 mg/ml bovine serum albumin (BSA), 0.6 mMINT, 5 mM pyruvate, 7 U of lipoamide dehydrogenase,0.2% Triton X-100, 0.5 mM EDTA, and 50 mM Tris-HCl, pH 7.4. The reaction was initiated by the additionof pyruvate and PDH activity was measured spectro-photometrically (500 nm) as the NADH-linked increasein absorbance of INT at 25°C.

Measurement of KGDH Activity

To measure the activity of KGDH, cells were firstharvested in homogenizing buffer: 20 mM MOPS con-

taining 0.2 mM mercaptoethanol, 5 mM MgCl2, 0.1 mMCaCl2, 1 g/L Triton X-100, 1 g/L Lubrol-Px, and the pHwas adjusted to 7.2. The harvestate was then assayedfor KGDH in a reaction mixture containing 0.15 mMTDP, 1 mM NAD, 0.12 mM CoA, 0.5 mM MgCl2, 0.5mM DTT, 1 g/L Triton X-100, 1 mM �-ketoglutarate,0.1 mM CaCl2, 50 �M EDTA, 0.5 mM dithiothreitol,and 40 �M rotenone in 50 mM MOPS, pH 7.6. Thereaction was initiated by the addition of CoA and theincrease of absorbance was measured spectrophoto-metrically (340 nm) as the �-ketoglutarate-dependentformation of NADH at 25°C as previously described(Lai and Cooper, 1986).

Measurement of Transketolase Activity

Transketolase activity was measured using a modi-fication of the technique of Dreyfus and Moniz (1962)as previously described (Giguere and Butterworth,1987). Briefly, cells were harvested in 50 mM potas-sium phosphate buffer (pH 7.6) and sonicated. Sampleswere added to ribose-5-phosphate (9 mM) and the mix-ture was incubated for 30 min at 37°C. The reactionwas stopped by the addition of 20% TCA and the sam-ples were centrifuged at 2,500 rpm for 15 min. Aliquotsof the supernatant were added to concentrated H2SO4

and boiled for 4 min. Addition of cysteine (3%) wasfollowed by measurement of sedoheptulose-7-phos-phate values 15 h later at 25°C from the difference inabsorbance at 510 and 540 nm using a standard curve.

Measurement of Na�,K�-ATPase Activity

Determination of the activity of Na�,K�-ATPase wasperformed according to Ratnakumari et al. (1995) andbased on Bonting et al. (1961). Cultures were harvestedin 30 mM Tris-HCl buffer (pH 7.4) containing 100 mMNaCl, 20 mM KCl, 3 mM MgCl2, and 3 mM ATP assubstrate. For measurement of Mg2�-ATPase activity,assays were performed in the presence of 1 mMouabain and NaCl and KCl were replaced with equalvolumes of distilled water. Experiments were carriedout at 37°C with a preincubation time of 10 min fol-lowed by the addition of ATP and a 15-min incubationperiod. The reaction was subsequently stopped by ad-dition of an equal volume of 10% trichloroacetic acid(w/v). Samples were centrifuged and equal volumes ofthe supernatant and coloring reagent (400 mg ferroussulfate per 10 ml of 1% ammonium molybdate in 1.15M H2SO4) were mixed and left to stand at room tem-perature for 15 min followed by estimation of liberatedinorganic phosphate by spectrophotometry at 700 nm.Na�,K�-ATPase activity was calculated as the differ-ence between total ATPase and Mg2�-ATPase activi-ties.

177THIAMINE DEFICIENCY IN CULTURED ASTROCYTES

Uptake of [3H]-D-Aspartate

[3H]-D-aspartate uptake studies were carried out ac-cording to Hazell et al. (1997) with modifications.Briefly, cells were preincubated for 30 min to allowequilibration in a serum-free solution containing thefollowing (in mmol/L): 120 NaCl, 2.7 KCl, 0.9 CaCl2, 0.5MgCl2, 6 glucose, 10 phosphate buffer, pH 7.4. Cultureswere then incubated in similar media containing thenonmetabolized glutamate analogue D-aspartate and0.2 �Ci/ml of [3H]-D-aspartate in 5% CO2/95% air at37°C. Uptake was stopped by aspiration of the mediaand rapid washing of cells three times with ice-coldDMEM. Cells were then solubilized in 0.5 ml of 1 MNaOH. Sample aliquots were then measured for pro-tein content (Lowry et al., 1951) and incorporated ra-dioactivity by liquid scintillation counting for calcula-tion of uptake rates. Studies of the effect of TDconditions on [3H]-D-aspartate uptake were carried outin which the incubation time for all experiments was 2min as linearity of uptake was maintained for at least5 min at all concentrations of D-aspartate studied (datanot shown). Cells were incubated with D-aspartate atconcentrations of 1–1,000 �M with correction of theresulting uptake rates for zero-time uptake at 0–4°C.Estimates of the kinetic parameters (Vmax and Km) for[3H]-D-aspartate uptake were obtained from the result-ing uptake curves, which were assumed to consist of asaturable component that follows Michaelis-Menten ki-netics using Eadie-Hofstee analysis. Also, for compar-ison, kinetic parameters were calculated by nonlinearregression analysis using curve-fitting algorithms onKaleidagraph software (Synergy Software, Reading,PA) for the Macintosh.

Immunoblotting

Astrocytes were harvested in buffer containing 50mM Tris, 150 mM NaCl, 0.1% sodium dodecyl sulfate(SDS), 1% NP-40, 0.5% sodium deoxycholate (pH 8.0),and a protease inhibitor cocktail (Sigma Chemical) andcentrifuged at 10,000 g for 10 min at 4°C. Aliquots ofthe resulting supernatant (50 �g) were subjected toSDS-polyacrylamide gel electrophoresis (8% polyacryl-amide) and the proteins were subsequently transferredto polyvinylidene difluoride membranes by wet trans-fer at 20 V over 24 h. The transfer buffer consisted of 48mM Tris (pH 8.3), 39 mM glycine, 0.037% SDS, and20% methanol. Membranes were subsequently incu-bated in blocking buffer (10 mM Tris, 100 mM NaCl,5% nonfat dried milk, and 0.1% Tween-20) followed byincubations with antisera against GLT-1 (1:1,000) orGLAST (1:5,000), corresponding to proteins of molecu-lar weight 75 and 65 kDa, respectively. Details of char-acterization of the rabbit polyclonal antiserum to theC-terminal domain of GLAST have previously beendescribed (Pow and Barnett, 1999). Membranes werealso incubated with an antiserum against the (40 kDa)actin (1:40,000) as housekeeping gene. Reblocking was

followed by incubation with diluted horseradish perox-idase-coupled antirabbit IgG (1:40,000) secondary an-tiserum. Each incubation step was of 1-h duration,following which blots were washed several times withbuffer (10 mM Tris, 100 mM NaCl, and 0.1% Tween-20). For the detection of specific antibody binding, themembranes were treated in accordance with the ECLkit instructions and apposed to photosensitiveX-OMAT film for 30–60 s. Signal intensities were sub-sequently measured by densitometry using a micro-computer-based image display system (Imaging Re-search, St. Catherines, Ontario, Canada). Linearity ofthe relationship between optical density and proteinconcentration was verified using appropriate standardcurves. Blots were reversibly stained with Ponceau-Sto determine uniformity of protein loading and evalu-ate protein transfer efficiency. Negative controls com-prised experiments in which transferred proteins wereincubated with blocking buffer in which the primaryantiserum was replaced with an appropriate amount ofBSA protein.

ATP Assay

Astrocytes were prepared for ATP measurements aspreviously described (Hazell et al., 1997). Briefly, cellswere initially frozen in liquid nitrogen and stored at�80°C. Following harvesting in ice-cold 0.1 M NaOHcontaining 1 mM EDTA, cells were immediately trans-ferred to tubes containing a final concentration of 7%perchloric acid. Samples were centrifuged at 5,000 g at4°C for 20 min. The supernatants were neutralizedwith 3 M K2CO3 and centrifuged again. Resulting su-pernatant extracts were assayed for ATP content byspectrofluorometry as described by Lowry and Passon-neau (1972).

Cytotoxicity Assay

Lactate dehydrogenase (LDH) was measured atroom temperature using the method of Wroblewski andLaDue (1955) with modifications as previously de-scribed (Hazell et al., 1997).

Morphology

Cells were stained with Giemsa and May-Grunwaldand examined for morphology at the light microscopelevel.

RESULTSGeneral Observations

Cultured cells showed characteristic morphologicalfeatures of type 1 astrocytes based on morphology andpositive GFAP immunstaining. Treatment of cells with

178 HAZELL ET AL.

TD for up to 10 days produced no obvious changes incell morphology.

Effect of TD on Levels of Thiamine and TDP

Exposure of cultured astrocytes to TD led to a 55%lowering of intracellular thiamine levels after 1 day(P � 0.01; Fig. 1A). More lengthy treatment with TDfor up to 10 days decreased thiamine levels by up to69%. Treatment of astrocytes with TD resulted in a58% decrease in cellular TDP levels after 1 day (P �

0.01; Fig. 1B). This effect was maintained with moreprolonged TD treatment for up to 10 days.

Effect of TD on Thiamine-DependentEnzyme Activities

Exposure of astrocytes to TD conditions for up to 4days resulted in no significant change in TK activity(Fig. 2). However, 7 and 10 days of treatment produceda reduction in activity of this enzyme of up to 91%compared to controls (P � 0.01). Treatment of astro-cytes with TD led to a 55% decrease in KGDH activityafter 4 days (P � 0.05), and after 7 and 10 days by 38%and 50%, respectively (P � 0.05; Fig. 2). In contrast, nochange in the activity of PDH was observed followingTD treatment for up to 10 days (Fig. 2).

Effects of TD on ATP Levels, Na�,K�-ATPaseActivity, and Cell Viability

When treated with up to 10 days of TD, astrocytesexhibited no change in ATP levels (controls 21.5 � 1.3vs. TD 17.9 � 2.6 nmol/mg protein; n � 9) or evidenceof loss of viability based on the amount of LDH releasedinto the media following TD treatment (controls 6.3 �0.5 vs. TD 7.0% � 0.6%; n � 9). In addition, treatmentof astrocytes for 10 days with TD produced no changein Na�,K�-ATPase activity (controls 11.5 � 1.3 vs. TD10.6 � 1.6 �mol/h/mg protein; n � 8).

Fig. 1. Levels of thiamine (A) and thiamine diphosphate (B) incultured astrocytes under conditions of thiamine deficiency. Cellswere exposed to thiamine-deficient media in the presence of 10 �Mpyrithiamine and treated for a period of up to 10 days. Controlastrocytes (Ctrl) were treated with thiamine-deficient media supple-mented with thiamine (11 �M) for the appropriate period of time andaveraged. For each experiment, all cultures were harvested on thesame day. Results show the mean � SEM of two separate experimentseach performed in triplicate. Asterisk, P � 0.01 compared with con-trols (Student’s t-test with Bonferroni correction for multiple compar-isons).

Fig. 2. Effect of thiamine deficiency on activities of the thiamine-dependent enzymes �-ketoglutarate dehydrogenase (KGDH), trans-ketolase (TK), and pyruvate dehydrogenase (PDH) in cultured astro-cytes. Cells were exposed to thiamine-deficient media in the presenceof 10 �M pyrithiamine and treated for a period of up to 10 days.Control astrocytes (Ctrl) were treated with TD media supplementedwith thiamine (11 �M) for the appropriate period of time and aver-aged. For each experiment, all cultures were harvested on the sameday. Results show the mean � SEM of two separate experiments eachperformed in triplicate. Asterisk, P � 0.05 compared with controls(one-way ANOVA with Dunnett’s test for multiple comparisons).

179THIAMINE DEFICIENCY IN CULTURED ASTROCYTES

Effect of TD on [3H]-D-Aspartate Uptake andGlutamate Transporter Protein Levels

TD treatment of cultured astrocytes led to a 22%decrease in [3H]-D-aspartate uptake after 7 days (P �0.05) and a 51% decrease after 10 days (P � 0.01)compared to controls (Fig. 3A). In order to examine theinfluence of TD on the kinetics of glutamate transport,concentration dependence studies were performed andshowed a decrease in [3H]-D-aspartate uptake follow-ing 10 days of TD treatment (Fig. 3B). Eadie-Hofsteeanalysis of kinetic parameters showed that TD led to a47% decrease in Vmax with no change in the Km value(Fig. 3C, Table 1).

To examine the effects of TD on the levels of astro-cyte glutamate transporter protein, GLT-1 and GLASTimmunoblots were prepared and the resulting bandswere quantitated by normalization to actin, used as ahousekeeping gene in all cases. While constitutiveGLAST protein content was high (Fig. 4A), levels ofGLT-1 protein were almost undetectable in the pres-ence of TD (� 10% of GLAST; data not shown). Expo-sure of astrocytes to TD for 1, 4, 7, and 10 days led to a

81% downregulation of GLAST protein after 10 days(P � 0.01; Fig. 4B).

Influence of Thiamine and Antioxidants on theTD-Induced Inhibition of D-Aspartate Uptake

TD cultures treated for 10 days and subsequentlyreplenished with thiamine for 3 days showed a 90%reversal of the inhibitory effect on [3H]-D-aspartateuptake, which was also reflected in a complete recoveryof GLAST protein content (Fig. 5). In order to examinethe mechanism by which TD may inhibit transporter

Fig. 3. Effect of thiamine deficiency on D-aspartate uptake in culturedastrocytes. Cells were exposed to thiamine-deficient media in the pres-ence of 10 �M pyrithiamine. Control astrocytes (Ctrl) were treated withthiamine-deficient media supplemented with thiamine (11 �M) for theappropriate period of time and averaged. A: Time course. Uptake wasstudied at a D-aspartate concentration of 100 �M. B: Concentrationdependence of D-aspartate uptake. Cells were treated for 10 days withTD and exposed to varying concentrations of D-aspartate ranging from 1to 1,000 �M with [3H]-D-aspartate (0.2 �Ci/ml) and [3H]-D-aspartateuptake was measured. C: Eadie-Hofstee analysis from a representativeexperiment. For each experiment, all cultures were harvested on thesame day. Error bars represent SEM of either two or three separateexperiments each performed in triplicate. Asterisk, P � 0.05; doubleasterisk, P � 0.01 compared with control cells (Student’s t-test withBonferroni correction for multiple comparisons).

TABLE 1. Effect of Thiamine Deficiency on the Kinetic Parametersof D-Aspartate Uptake in Cultured Astrocytes*

Kinetic parameters Control TD

Km (�M) 100.4 � 8.2 126.3 � 9.2Vmax (nmol/min/mg protein) 36.1 � 2.1 18.9 � 1.3a

*Cells were treated with thiamine-deficient media in the presence of 10 �Mpyrithiamine for 10 days and then incubated with [3H]-D-aspartate for 2 min at37°C as described in text. Data are mean � SEM from three separate experi-ments each performed in triplicate. aP � 0.01 compared with control values(Mann-Whitney U-test).

180 HAZELL ET AL.

uptake of [3H]-D-aspartate, cells were exposed for 10days in the absence or presence of the antioxidantsglutathione (145 �M), �-tocopherol (55 �M), or buty-lated hydroxy anisole (BHA; 100 �M). All three anti-oxidants were ineffective at reversing the inhibition of[3H]-D-aspartate uptake (Fig. 5).

Influence of Protein Kinase C Inhibition andGroup II mGluR Agonists on the TD-Induced

Decrease in D-Aspartate Uptake

Astrocytes exposed to 10 days of TD in the presenceof the protein kinase C (PKC) inhibitor H7 [1-(5-iso-quinolinesulfonyl)-2-methylpiperazine dihydrochoride](500 �M) from days 7 to 10 showed a 95% recovery of[3H]-D-aspartate uptake (Fig. 6A). In addition, expo-sure of TD-treated cells for 3 days to the group IImGluR agonist DCG IV (100 �M) resulted in a 89%

recovery of uptake (Fig. 6A). Immunoblot studies re-vealed that these uptake recoveries were associatedwith a reversal of GLAST downregulation in the case ofH7 (55% recovery) and DCG IV (27% recovery). BothH7 and DCG IV on their own increased basal uptake of[3H]-D-aspartate by 56% and 29%, respectively, whichwas reflected in a 64% and 36% upregulation of GLAST(Fig. 6).

DISCUSSION

The findings of this study indicate that 10 days of TDtreatment produces a metabolic impairment associatedwith a decrease in D-aspartate uptake in cultured as-trocytes. Kinetic analysis showed that this decrease inuptake was due to a decrease in the Vmax associatedwith a downregulation of the GLAST glutamate trans-porter. These results are consistent with our recentfindings of a downregulation of GLAST and GLT-1astrocyte glutamate transporters in the TD rat (Hazellet al., 2001) and provide further evidence of an involve-ment of the GLAST transporter in the pathophysiologyof thiamine deficiency. While GLAST protein contentwas initially high in these astrocyte cultures, levels ofGLT-1 protein were almost undetectable. This is con-sistent with the well-known property of low GLT-1levels in astrocyte cultures of high purity (Gegelashviliet al., 1997; Swanson et al., 1997).

The present results indicate that downregulation ofGLAST and the consequent loss of glutamate transport

Fig. 4. Effect of thiamine deficiency on GLAST transporter proteinlevels in cultured astrocytes. Blots were probed with an anti-GLASTantibody. For each experiment, all lanes were loaded with equalamounts of protein in each case (30 �g). A: Representative blot ofGLAST protein in cultured astrocytes (lanes 1–3, control (Ctrl); lanes4–6, 1 day; lanes 7–9, 4 days; lanes 10–12, 7 days; and lanes 13–15,10 days of treatment with thiamine-deficient media and 10 �M pyri-thiamine). B: Quantitative immunoblot analysis of GLAST proteinlevels. Data are the mean � SEM of three separate experiments eachperformed in triplicate. Asterisk, P � 0.01 compared with control cells(one-way ANOVA with Dunnett’s test for multiple comparisons).

Fig. 5. Influence of thiamine replenishment and antioxidants onthiamine deficiency-mediated inhibition of D-aspartate uptake. Astro-cytes were exposed to thiamine-deficient media in the presence of 10�M pyrithiamine for 10 days followed by reversal with thiamine (11�M) for 3 days (TD�Th). Inset: Representative immunoblot of GLASTprotein for these conditions. C, Ctrl; T, TD�Th. In other experiments,cells were exposed to thiamine-deficient media in the presence of 10�M pyrithiamine for 10 days and either glutathione (Glut; 145 �M), or�-tocopherol (Toco; 55 �M), or BHA (100 �M). Uptake was studied ata D-aspartate concentration of 100 �M. Results are expressed asmean � SEM of 2–3 separate experiments each performed in tripli-cate. Asterisk, P � 0.01 compared with control cells (one-way ANOVAwith Bonferroni correction for multiple comparisons).

181THIAMINE DEFICIENCY IN CULTURED ASTROCYTES

function following TD treatment was not related to alowered energy status or loss of cell viability since ATPlevels were unchanged and LDH levels were unaf-fected. Furthermore, TD astrocytes, subsequentlywashed out and exposed to thiamine-replete medium,showed evidence of reversibility of the glutamatetransport deficit due to reversal of GLAST downregu-lation. Interestingly, 7 days of TD treatment resultedin a decrease in glutamate transport in the absence ofany change in the level of GLAST protein. This findingindicates that GLAST transporter dysfunction pre-cedes its downregulation.

Treatment of astrocytes with the protein kinase in-hibitor H7 resulted in a recovery of glutamate uptakefunction in TD-exposed cells in association with block-age of the GLAST protein downregulation. Recentstudies have demonstrated that phorbol esters (whichactivate PKC) decrease GLAST expression and gluta-mate transport activity (Conradt and Stoffel, 1997;Gonzalez et al., 1999). Since H7 was able to prevent theuptake inhibition produced by TD, this indicates thatTD may increase the phosphorylation of GLAST. Howthis may occur is unclear at the present time. When TDastrocytes were treated with the group II mGluR ago-nist DCG IV, glutamate uptake was also increased,concomitant with improvement of GLAST levels. Eightsubtypes of mGluRs are presently known, and they aresubdivided into three groups. Of these, group I mGluRagonists have been reported to cause a downregulationof the GLAST transporter, while the group II agonistDCG IV upregulates the expression of this protein(Gegelashvili et al., 2000). Thus, mGluRs may also beimplicated in GLAST function in TD. Group II mGluRsare negatively coupled to cyclic AMP and found on glialcells as well as both pre- and postsynaptic membranes(Petralia et al., 1996). It is possible, therefore, thatDCG IV treatment antagonizes the decrease of GLASTprotein in TD by interfering with cyclic AMP levels inthe astrocyte. DCG IV has been shown to have neuro-protective efficacy in cultured cortical neurons (Brunoet al., 1995; Miyamoto et al., 1997) and in a model oftraumatic brain injury (Zwienenberg et al., 2001). Useof this ligand may in future offer a potential therapeu-tic strategy for treatment of excitotoxicity in thiaminedeficiency as well as in other brain disorders, includingneurodegenerative diseases, in which evidence of a dys-function of glutamate transport has been reported.

In the present study, selectivity of the glutamatetransport deficit was evaluated by measuring Na�,K�-ATPase activity in TD-treated astrocytes. Glutamatetransport is strongly coupled to the maintenance ofionic gradients across the cell membrane, which in turnis dependent on the Na�,K�-ATPase pump. Since thefunction of many transporters is also dependent on thisenzyme, a change in its activity would suggest thatdecreased glutamate transport is not a selective pro-cess in TD. Our findings indicate that Na�,K�-ATPaseactivity is unchanged, providing evidence that TDtreatment selectively affects the GLAST transporter.

In the TD rat, previous studies have reported anincrease in reactive oxygen species (ROS) (Langlais etal., 1997). Since ROS can lead to decreased activity ofthe Na�,K�-ATPase (Hexum and Fried, 1979) and areversal of the glutamate transporter, astrocytes weretreated with the antioxidants glutathione or �-tocoph-erol to examine whether such a process might contrib-ute to the glutamate transport dysfunction indepen-dent of GLAST downregulation. However, neitheragent had an effect on glutamate uptake inhibition,suggesting that oxidative stress is unlikely to contrib-ute to the transport defect. In addition, TD treatmentdid not lead to any change in Na�,K�-ATPase activity,

Fig. 6. Influence of H7 and DCG IV on thiamine deficiency-medi-ated inhibition of D-aspartate uptake and GLAST downregulation.Astrocytes were exposed to thiamine-deficient media in the presenceof 10 �M pyrithiamine for 10 days in the absence or presence of thePKC inhibitor H7 (500 �M) or the group II mGluR agonist DCG IV(100 �M). Cells were exposed to H7 or DCG IV during days 7–10.A: Uptake studied at a D-aspartate concentration of 100 �M.B: Representative immunoblot and quantitative analysis of GLASTprotein levels. D � DCG IV; H � H7. Results are expressed as mean �SEM of two separate experiments each performed in quadruplicate.Asterisk, P � 0.05 compared with control cells (one-way ANOVA withDunnett’s test for multiple comparisons).

182 HAZELL ET AL.

supporting this concept. Furthermore, exposure toBHA (another antioxidant and inhibitor of lipid peroxi-dation) was not beneficial, suggesting that lipid peroxi-dation is also unlikely to be a contributor to the de-crease in glutamate uptake.

Treatment of cultured astrocytes with TD led to arapid and sustained decrease in the intracellular con-tent of thiamine and TDP levels, thus indicating thedevelopment of TD. These findings reflect the impor-tance of a constant external supply of the vitamin formaintenance of the levels of TDP levels in these cells.Rapid depletion of intracellular stores of thiamine andTDP, the biologically active (enzyme cofactor) form ofthis vitamin, after the first 24 h in astrocytes exposedto TD was followed by decreases in activity of theTDP-dependent enzymes KGDH and transketolase. Inaddition, activity of PDH in the different treatmentgroups remained unchanged throughout the timecourse, consistent with previous reports (Butterworth,1986; Elnageh and Gaitonde, 1988). These findingsindicate that during the early stages of TD in astro-cytes, all three TDP-dependent enzyme systems areable to maintain normal activity despite large de-creases in thiamine and TDP levels. This may be asso-ciated with a high degree of apoenzyme/cofactor (TDP)binding and/or a low enzyme turnover rate. However,TD lowers mRNA levels of transketolase (and the E1subunit of PDH), but not mRNA levels of the TDP-binding E1 subunit of KGDH, with the eventual de-crease in activity of both enzymes likely to be due to theloss of TDP in the case of transketolase and loss of theE1 subunit in the case of KGDH (Gibson et al., 2000).In addition, Park et al. (1999) have reported that ROSinhibit KGDH activity in cultured microglia. The sus-tained decrease in activity of transketolase by approx-imately 90% between day 7 and 10 in the absence of celldeath suggests that this enzyme is unlikely to play amajor role in the determination of astrocyte survivalunder conditions of TD.

Ten days of TD treatment did not affect cell viabilityas determined by LDH release status, and astrocytesmaintained normal morphological integrity during thistime, consistent with the findings of unchanged ATPlevels. The ability of these cells to maintain a highenergy status despite a large decrease in the activity ofKGDH, a major rate-limiting enzyme of the tricarbox-ylic acid cycle (presumably due to increased anerobicglycolysis), may be an important factor in their abilityto survive the TD insult.

In conclusion, TD treatment of primary cultures ofastrocytes led to a dramatic decline in intracellularlevels of thiamine, TDP, activities of thiamine-depen-dent enzymes, and a decrease in glutamate uptake thatwas associated with a loss of GLAST. This effect ap-pears to be at least partly due to an involvement ofPKC and phosphorylation in the regulation of thistransporter. In addition, activation of group II mGluRsreduces the extent of GLAST transporter dysfunction,suggesting a potential therapeutic role for agonists ofthis glutamate receptor subtype in the treatment or

prevention of thiamine-deficient lesions. Use of thisastrocyte preparation may thus afford an importantmeans towards further understanding the loss of glu-tamate transporter-related functionality in TD and inpatients with WE.

ACKNOWLEDGMENT

The authors thank Dr. Roger F. Butterworth foruseful discussions.

REFERENCES

Aikawa H, Watanabe IS, Furuse T, Iwasaki Y, Satoyoshi E, Sumi T,Moroji T. 1984. Low energy levels in thiamine-deficient encepha-lopathy. J Neuropathol Exp Neurol 43:276–287.

Arriza JL, Eliasof S, Kavanaugh MP, Amara SG. 1997. Excitatoryamino acid transporter 5, a retinal glutamate transporter coupledto a chloride conductance. Proc Natl Acad Sci USA 94:4155–4160.

Bettendorf L, Grandfils C, de Rycker C, Schoffeniels E. 1986. Deter-mination of thiamine and its phosphate esters in human bloodserum at femtomole levels. J Chromatogr 382:297–302.

Bettendorff L, Sluse F, Goessens G, Wins P, Grisar T. 1995a. Thia-mine deficiency-induced partial necrosis and mitochondrial uncou-pling in neuroblastoma cells are rapidly reversed by addition ofthiamine. J Neurochem 65:2178–2184.

Bettendorff L, Goessens G, Sluse F, Wins P, Bureau M, Laschet J,Grisar T. 1995b. Thiamine deficiency in cultured neuroblastomacells: effect on mitochondrial function and peripheral benzodiaz-epine receptors. J Neurochem 64:2013–2021.

Bettendorff L, Goessens G, Sluse FE. 1997. Reversibility of thiaminedeficiency-induced partial necrosis and mitochondrial uncouplingby addition of thiamine to neuroblastoma cell suspensions. Mol CellBiol 174:121–124.

Bonting SL, Simon KA, Hawkins NM. 1961. Studies on sodium-potassium-activated adenosinetriphosphatase: I, quantitative dis-tribution in several tissues of the cat. Arch Biochem Biophys 95:416–423.

Booher J, Sensenbrenner M. 1972. Growth and cultivation of dissoci-ated neurons and glial cells from embryonic chick, rat and humanbrain in flask cultures. Neurobiology 2:97–105.

Bruno V, Battaglia G, Copani A, Giffard RG, Raciti G, Raffaele R,Shinozaki H, Nicoletti F. 1995. Activation of class II or III metabo-tropic glutamate receptors protects cultured cortical neuronsagainst excitotoxic degeneration. Eur J Neurosci 7:1906–1913.

Butterworth RF. 1986. Cerebral thiamine-dependent enzyme changesin experimental Wernicke’s encephalopathy. Metab Brain Dis1:165–175.

Butterworth RF, Giguere JF, Bernard AM. 1986. Activities of thia-mine-dependent enzymes in two experimental models of thiamine-deficiency encephalopathy. 2. alpha-Ketoglutarate dehydrogenase.Neurochem Res 11:567–577.

Collins GH. 1967. Glial cell changes in the brain stem of thiamine-deficient rats. Am J Pathol 50:791–814.

Conradt M, Stoffel W. 1997. Inhibition of the high-affinity brainglutamate transporter GLAST-1 via direct phosphorylation. J Neu-rochem 68:1244–1251.

Dreyfus P, Moniz R. 1962. The quantitative histochemical estimationof transketolase in the nervous system of the rat. Biochem BiophysActa 65:181–189.

Elnageh KM, Gaitonde MK. 1988. Effect of a deficiency of thiamine onbrain pyruvate dehydrogenase: enzyme assay by three differentmethods. J Neurochem 51:1482–1489.

Fairman WA, Vandenberg RJ, Arriza JL, Kavanaugh MP, Amara SG.1995. An excitatory amino-acid transporter with properties of aligand-gated chloride channel. Nature 375:599–603.

Gegelashvili G, Danbolt NC, Schousboe A. 1997. Neuronal solublefactors differentially regulate the expression of the GLT1 andGLAST glutamate transporters in cultured astroglia. J Neurochem69:2612–2615.

Gegelashvili G, Dehnes Y, Danbolt NC, Schousboe A. 2000. The high-affinity glutamate transporters GLT1, GLAST, and EAAT4 areregulated via different signalling mechanisms. Neurochem Int 37:163–170.

183THIAMINE DEFICIENCY IN CULTURED ASTROCYTES

Gibson GE, Park LC, Sheu KF, Blass JP, Calingasan NY. 2000. The�-ketoglutarate dehydrogenase complex in neurodegeneration.Neurochem Int 36:97–112.

Giguere J-F, Butterworth RF. 1987. Activities of thiamine-dependentenzymes in two experimental models of thiamine deficiency enceph-alopathy: III, transketolase. Neurochem Res 12:305–310.

Gonzalez MI, Lopez-Colom AM, Ortega A. 1999. Sodium-dependentglutamate transport in Muller glial cells: regulation by phorbolesters. Brain Res 831:140–145.

Hazell AS, Butterworth RF, Hakim AM. 1993. Cerebral vulnerabilityis associated with selective increase in extracellular glutamate con-centration in experimental thiamine deficiency. J Neurochem 61:1155–1158.

Hazell AS, Itzhak Y, Liu H, Norenberg MD. 1997. 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) decreases glutamate uptake incultured astrocytes. J Neurochem 68:2216–2219.

Hazell AS, Todd KG, Butterworth RF. 1998a. Mechanisms of neuro-nal cell death in Wernicke’s encephalopathy. Metab Brain Dis 12:97–122.

Hazell AS, Hakim AM, Senterman MK, Hogan MJ. 1998b. Regionalactivation of L-type voltage-sensitive calcium channels in experi-mental thiamine deficiency. J Neurosci Res 52:742–749.

Hazell AS, Rama Rao KV, Danbolt NC, Pow DV, Butterworth RF.2001. Selective down-regulation of the astrocyte glutamate trans-porters GLT-1 and GLAST within the medial thalamus in experi-mental Wernicke’s encephalopathy. J Neurochem 78:560–568.

Hexum TD, Fried R. 1979. Effects of superoxide radicals on transport(Na � K) adenosine triphosphatase and protection by superoxidedismutase. Neurochem Res 4:73–82.

Kanai Y, Hediger MA. 1992. Primary structure and functional char-acterization of a high-affinity glutamate transporter. Nature 360:467–471.

Kanner BI, Marva E. 1982. Efflux of L-glutamate by synaptic plasmamembrane vesicles isolated from rat brain. Biochemistry 21:3143–3147.

Lai JC, Cooper AJ. 1986. Brain alpha-ketoglutarate dehydrogenasecomplex: kinetic properties, regional distribution, and effects ofinhibitors. J Neurochem 47:1376–1386.

Langlais PJ, Zhang SX. 1993. Extracellular glutamate is increased inthalamus during thiamine deficiency-induced lesions and is blockedby MK-801. J Neurochem 61:2175–2182.

Langlais PJ, Anderson G, Guo SX, Bondy SC. 1997. Increased cere-bral free radical production during thiamine deficiency. MetabBrain Dis 12:137–143.

Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. 1951. Protein mea-surement with the folin phenol reagent. J Biol Chem 193:263–275.

Lowry OH, Passonneau JV. 1972. A flexible system of enzymaticanalysis: New York: Academic Press.

Miyamoto M, Ishida M, Shinozaki H. 1997. Anticonvulsive and neu-roprotective actions of a potent agonist (DCG-IV) for group IImetabotropic glutamate receptors against intraventricular kainatein the rat. Neuroscience 77:131–140.

Park LC, Zhang H, Sheu KF, Calingasan NY, Kristal BS, Lindsay JG,Gibson GE. 1999. Metabolic impairment induces oxidative stress,

compromises inflammatory responses, and inactivates a key mito-chondrial enzyme in microglia. J Neurochem 72:1948–1958.

Park LC, Calingasan NY, Uchida K, Zhang H, Gibson GE. 2000.Metabolic impairment elicits brain cell type-selective changes inoxidative stress and cell death in culture. J Neurochem 74:114–124.

Petralia RS, Wang YX, Niedzielski AS, Wenthold RJ. 1996. Themetabotropic glutamate receptors, mGluR2 and mGluR3, showunique postsynaptic, presynaptic and glial localizations. Neuro-science 71:949–976.

Pines G, Danbolt NC, Bjoras M, Zhang Y, Bendahan A, Eide L,Koepsell H, Storm-Mathisen J, Seeberg E, Kanner BI. 1992. Clon-ing and expression of a rat brain L-glutamate transporter. Nature360:464–467.

Pow DV, Barnett NL. 1999. Changing patterns of spatial buffering ofglutamate in developing rat retinae are mediated by the Muller cellglutamate transporter GLAST. Cell Tissue Res 297:57–66.

Ratnakumari L, Audet R, Qureshi IA, Butterworth RF. 1995.Na�,K�-ATPase activities are increased in brain in both congenitaland acquired hyperammonemic syndromes. Neurosci Lett 197:89–92.

Robertson DM, Wasan SM, Skinner DB. 1968. Ultrastructural fea-tures of early brain stem lesions of thiamine-deficient rats. Am JPathol 52:1081–1097.

Schousboe A, Sonnewald U, Civenni G, Gegelashvili G. 1997. Role ofastrocytes in glutamate homeostasis: implications for excitotoxicity.Adv Exp Med Biol 429:195–206.

Schwartz JP, Lust WD, Shirazawa R, Passonneau JV. 1975. Glyco-lytic metabolism in cultured cells of the nervous system: III, theeffects of thiamine deficiency and pyrithiamine on the C-6 gliomaand C-1300 neuroblastoma cell lines. Mol Cell Biochem 9:73–78.

Sheu KF, Calingasan NY, Lindsay JG, Gibson GE. 1998. Immuno-chemical characterization of the deficiency of the alpha-ketoglu-tarate dehydrogenase complex in thiamine-deficient rat brain.J Neurochem 70:1143–1150.

Storck T, Schulte S, Hofmann K, Stoffel W. 1992. Structure, expres-sion, and functional analysis of a Na(�)-dependent glutamate/as-partate transporter from rat brain. Proc Natl Acad Sci USA 89:10955–10959.

Swanson RA, Liu J, Miller JW, Rothstein JD, Farrell K, Stein BA.1997. Longuemare MC Neuronal regulation of glutamate trans-porter subtype expression in astrocytes. J Neurosci 17:932–940.

Szatkowski M, Barbour B, Attwell D. 1990. Non-vesicular release ofglutamate from glial cells by reversed electrogenic glutamate up-take. Nature 348:443–446.

Troncoso JC, Johnston MV, Hess KM, Griffin JW, Price DL. 1981.Model of Wernicke’s encephalopathy. Arch Neurol 38:350–354.

Wroblewski F, LaDue JS. 1955. Lactic dehydrogenase activity inblood. Proc Soc Exp Biol Med 90:210–213.

Zwienenberg M, Gong QZ, Berman RF, Muizelaar JP, Lyeth BG.2001. The effect of groups II and III metabotropic glutamate recep-tor activation on neuronal injury in a rodent model of traumaticbrain injury. Neurosurgery 48:1119–1126.

184 HAZELL ET AL.