Glutamate Neurotoxicity in Cortical Cell Culture

The Journal of Neuroscience, February 1987, 7(2): 357-388

Dennis W. Choi, Margaret Maulucci-Gedde, and Arnold R. Kriegstein Department of Neurology, Stanford University Medical Center, Stanford, California 94305

The central neurotoxicity of the excitatory amino acid neu- 1982), epilepsy (Nadler et al., 1978; Sloviter, 1983), stroke rotransmitter glutamate has been postulated to participate (Rothman, 1984; Simon et al., 1984), and hypoglycemic en- in the pathogenesis of the neuronal cell loss associated cephalopathy (Wieloch, 1985). Despite this potential clinical with several neurological disease states, but the complexity importance, little is presently known about the phenomenonof of the intact nervous system has impeded detailed analysis GNT at the cellular level. The major obstacle to the study of of the phenomenon. In the present study, glutamate neu- GNT has been the complexity of the intact nervous system; rotoxicity was studied with novel precision in dissociated indeed, it is somewhat problematic to demonstrate the phe- cell cultures prepared from the fetal mouse neocortex. nomenon at all in viva Systemic administration of glutamate Brief exposure to glutamate was found to produce mor- to rodents generally produces brain lesionsonly in immature phological changes in mature cortical neurons beginning as animalswithout a fully developed blood-brain barrier, and then quickly as 90 set after exposure, followed by widespread predominantly only in the circumventricular regions (Olney, neuronal degeneration over the next hours. Quantitative 1978). Although it is possibleto produce brain lesionsin adult dose-toxicity study suggested an ED, of 50-l 00 MM for a 5 animals by direct intraparenchymal administration (McBean min exposure to glutamate. Immature cortical neurons and and Roberts, 1984), extremely high dosesare needed,possibly glia were not injured by such exposures to glutamate. Up- reflecting the protective efficacy of glutamate uptake systems. take processes probably do not limit GNT in culture, as the Much of what we presently know about GNT is therefore de- uptake inhibitor dihydrokainate did not potentiate GNT. Pos- rived indirectly, from studies of the structural analog kainate, sibly reflecting the lack of uptake limitation, glutamate was a plant compound not endogenousto the mammalian CNS. found to be actually more potent than kainate as a neuro- The purpose of this study, and of the companion study re- toxin in these cultures, a dramatic reversal of the in vivo ported in the following paper, was to investigate the phenom- potency rank order. Some neurons regularly survived brief enon of GNT directly, utilizing the experimental leveragegained glutamate exposure; these possibly glutamate-resistant in a primary dissociatedcell culture system, derived from fetal neurons had electrophysiologic properties, including che- mouseneocortex. In cell culture, definedconcentrations of glu- mosensitivity to glutamate, that were grossly similar to those tamate can be delivered directly to neuronsin a controlled en- of the original population. vironment, and the resultant morphologicalchanges can be pre- cisely monitoredin a serialfashion. We report herethat glutamate The amino acid glutamate is present at several millimolar con- is a remarkably potent and rapidly acting neurotoxin in cortical centration in mammalian central gray matter (Waelsch, 1951) cell culture. and has a ubiquitous excitatory effect on central neurons(Curtis et al., 1960; Crawford and Curtis, 1964; Krnjevic, 1974), prop- Materials and Methods erties that support the current belief that glutamate is likely a Cell culture. Dissociated murine cortical cell cultures were prepared major mammalian central excitatory neurotransmitter (Di- using minor modification of an established technique (Dichter, 1978). Whole cerebral neocortices were removed from fetal mice (14-l 7 d Chiara and Gessa, 1981). It is therefore somewhatsurprising gestation),taking care to discardthe hippocampalformation, basal gan- that glutamate is alsoa neurotoxin (Lucasand Newhouse,1957; glia, andmost of the meninges.The tissue was then minced,incubated Olney, 1969; Coyle et al., 1981). Under normal circumstances, in 0.08% acetylated trypsin for 45-75 min at 37”C, dissociated by trit- protective mechanismspresumably limit neuronal glutamate uration, and plated as a single-cell suspension on Primaria (Falcon) 35 exposureand prevent toxicity, but a potential for pathogenesis mm dishes (lo6 cells/dish) in a plating medium of Eagle’s minimal essential media (MEM, Earle’s salts) supplemented with 10% heat-in- is surely ever present. Glutamate neurotoxicity (GNT) has now activated horse serum, 101 fetal bovine serum, glutamine (total 2 mM, beenproposed to participate specificallyin the neuronal cell loss added the week of use), glucose (total 21 mM), and bicarbonate (total associatedwith a number of neurologic diseases,including Hun- 38 mM). Cultures were maintained at 37°C in a humidified 9% CO, tington’s disease(Coyle and Schwartz, 1976; McGeer and atmosphere. After 5-7 d in vitro, non-neuronal cell division was halted by l-3 d of exposure to 1O-5 M cytosine arabinoside, and the cells were McGeer, 1976), olivopontocerebellar atrophy (Plaitakis et al., shifted into a maintenance medium similar to the plating media but lacking fetal serum. Subsequent media replacement was carried out on Received Feb. 25, 1986; revised June 12, 1986; accepted July 23, 1986. a biweekly schedule. The culture media used did not contain any glu- We thank M. Yokoyama for expert technical assistance.This work was sup- tamate except for that present in the serum (Hyclone defined sera); based ported by Grants BRSG RR5353 and NS21628 from the National Institute of on the supplier’s measurements of serum glutamate levels, the initial Health, grants from the Wills and Hereditary DiseaseFoundations, and a Hartford glutamate concentration in maintenance media was less than 20 PM. Fellowship (to D.W.C.) from the John A. Hartford Foundation. Cortical glial cell cultures were prepared using the same protocol as Correspondence should be addressed to Dennis W. Choi at the above address. above but using cortices removed from early postnatal mice (postnatal Copyright 0 1987 Society for Neuroscience 0270-6474/87/020357-12$02.00/O days l-3) instead of fetal mice, since neurons removed from such older 82

83 84

Figure I. Cortical cell culture. A, Field of cortical cells photographed just prior to electrophysiologic study. The cells labeled with arrows were all penetrated for intracellular recording and found to be neurons (all exhibiting spontaneous synaptic potentials; all but 2, probably damaged by the recording electrode, exhibiting evoked or spontaneous action potentials). Bl- B3, Oscilloscope traces obtained from cells l-3 in A, respectively. B4, Penwriter record obtained in cell 4; the downward lines were produced by passing constant hyperpolarizing current pulses. Delivery of a 2 set pulse of 0.5 mM glutamate from a nearby puffer pipette occurs at the arrowand is followed by cell depolarization, reduction in input resistance, and a flurry of IPSPs. Vertical calibration, 10 mV (BI, B2), 5 mV (B3), 24 mV (B4); horizontal calibration, 10 msec (BI-B3), 7 set (B4). Cl, Fast cresyl violet Nissl staining shows darkly stained neuronal cell bodies against a lightly stained non-neuronal cell layer. C2, Heavy staining of this background cell layer with anti-GFAP antibody; staining is essentially absent in the absence of specific anti-GFAP antibody (C3), except for slight darkening of neuronal cell bodies. Distance bar in A and Cl corresponds to 50 pm; the Cl bar applies to C2 and C3 as well. The Journal of Neuroscience, February 1997, 7(2) 359

Figure 2. Acute time course of GNT. Serial phase-contrast micrographs showing a field of cortical cells before (A), immediately after (B), 10 min after (C), and 25 min after (D) exposure to a single 90 set pulse of 1 mM glutamate. The dish was maintained on the stage of the physiology microscope, and the field was continuously observed after glutamate was delivered by pressure ejection from a puffer pipette placed in the middle of the field. Bar, 50 pm. animals do not survive the plating period (Booher and Sensenbrenner, either absent or labeled with trypan blue on the follow-up study, were 1972; McCarthy and devellis, 1980). counted as damaged by glutamate; the counts were expressed as a per- GIutamate exposure. Prior to study, culture dishes were examined centage of the number of cells originally present. In control experiments, under phase-contrast microscopy, and (in most experiments) l-5 rep- dishes exposed to the procedure described above, but without addition resentative 200x microscope fields per dish were photographed and of glutamate, showed damage to no more than a few percent of neurons. marked on a coordinate system. Exposure to glutamate was generally Immunohistochemical stainingfor glial~rillary acidic protein (GFAP). carried out at room temperature in a Tris-buffered “control salt solu- Cultures to be stained were washed in CSS and fixed by immersion in tion” (CSS) substituted for culture medium by triple exchange; CSS had Perhx (Fisher Scientific) for 30 min at room temperature. Fixative was the following composition (in mM): NaCl, 120; KCl, 5.4; MgCl,, 0.8; removed by rinsing 3 times with 20 mM PBS (pH 7.5). Cultures were CaCl,, 1.8; Tris-Cl @H 7.4 at 25°C); glucose, 15. After 5 min, the exposed first to a blocking solution consisting of 1.8% normal goat serum glutamate was washed out thoroughly (effective dilution >600) and in PBS for 1 hr, followed by overnight incubation with polyclonal rabbit reolaced with Eaale’s MEM with augmented alucose (21 mM) and Na anti-GFAP antiserum (kindly supplied by Dr. Larry Eng) at 5°C (1:250 bicarbonate (38 &M), prior to returmng the dishes to‘the 9Oi CO, in- dilution of antiserum in blocking solution plus 0.3% Triton X-100). cubator. Continuous monitoring of pH with phenol red indicator was After a 60 min PBS wash (3 exchanges), cultures were processed for done in some experiments and did not show evidence of any pH dis- antibody binding with the avidin-biotin-peroxidase system (Vectastain turbance with the procedure. Cultures were later removed from the ABC Kit, Rabbit IgG, Vector Laboratories) and reacted with 0.05 mgl incubator and reexamined at specified intervals, with follow-up photos ml diaminobenzidine tetrahydrochloride (Sigma) for 10 min or until taken of previously identified fields. stained. Controls were processed in an identical manner except that For quantitative assessment of GNT, cell counts were made from anti-GFAP was omitted from the primary antibody solution. photomicrographs of representative identified fields taken before and Electrophysiology, Dishes were maintained on the stage ofan inverted after glutamate exposure; in most cases, the follow-up photomicrograph Nikon Diaphot microscope, in a warmed aluminum block (bath tem- was taken both with uhase-contrast and with bright-field following 5 perature, 33”c), and perfused continuously with a recording medium min of incubation in-0.4% trypan blue, a dye normally excluded-by consisting of 3 parts Earle’s balanced salt solution and 1 part MEM, healthy cells. Specific neurons present on the baseline photographs, but augmented with extra CaCl, (total 3.8 mM). Visualized cells (phase- 360 Choi et al. l Glutamate Neurotoxicity in Cell Culture

Figure 3. Long-term course of GNT. Serial phase-contrast micrographs of an identified field of cortical cells taken before (A), 1 hr after (B), 2 hr after (C), 4 hr after (D), 9 hr after (I?), and 24 hr after (F) pulse exposure to 0.5 mM glutamate for 5 min. Between photographs the dish was kept in the incubator. Bar, 100 pm.

contrast, 400 X) were impaled for intracellular recording with glass microelectrodes filled with 4 M potassium acetate (50-80 MQ), and the Results signal was displayed on a storage oscilloscope and a chart recorder. A In murine cortical cell cultures prepared as described, pre- bridge circuit enabled current to be injected through the recording elec- sumptive neurons send out an extensive network of processes, trode for the purpose of stimulation or measurement of cell input resistance. Defined concentrations of glutamate were delivered to cells by and, over the first 10 d in culture, develop increasingly large, pressure ejection from a nearby “puffer” pipette, using the recording phase-bright cell bodies; such cells stain darkly with a Nissl medium as a drug carrier (Choi and Fischbach, 198 1). stain (fast cresyl violet) (Fig. 1 Cl). The background mat of cells The Journal of Neuroscience, February 1987, 7(2) 361

Figure 4. Lack of gliotoxicity of glutamate. A and B, Phase-contrast micrographs of a representative field of a cortical non-neuronal cell culture taken before (A) and 1 d after (B) exposure to 0.5 mM glutamate for 5 min. C, bright-field micrograph of a sister non-neuronal culture stained for GFAP. D, Phase-contrast appearance of another sister culture 1 d after exposure to 80 PM A23 187 in the presence of high (6.8 mM) Ga for 20 min; only debris is apparent. A23187 was first dissolved in dimethyl sulfoxide. Bar, 50 pm. for the most part showedthe typical flat, polygonal morphology serialphase-contrast photomicrographs of a field of cortical neu- of astrocytes in culture (Booher and Sensenbrenner, 1972; rons taken before and after pressureejection of glutamate; 90 McCarthy and devellis, 1980)and stainedpositively with rabbit set after exposure onset (Fig. 28), 1 neuronal cell body (in the anti-GFAP (Fig. 1, C2, C3). center of the photo, at the top of the cluster of 4 neurons)had The neuronal identity of these phase-bright cells was con- already swollen by about 20% in diameter. firmed by intracellular recordings(Fig. 1, A, B). In a seriesof In order to study the late effects of glutamate exposure, the 45 cells with stablemembrane potentials equal to or more neg- bath application proceduredescribed in Materials and Methods ative than -50 mV, all had spontaneousor stimulus-evoked was used. Pulseexposure of mature (1424 d in vitro) cultures neuronal action potentials, and most (42/45) exhibited spon- to 0.5 mM glutamate for 5 min in CSS was invariably followed taneousexcitatory and/or inhibitory synaptic activity. Resting by widespreadprogressive neuronal degenerationover subse- potential was 60.6 + 6.7 mV and input resistancewas 73.9 f quent hours; by the following morning, most of the neurons 30.7 MB (mean + SD). Brief (l-2 set) pressureejection of OS- werereplaced by debris (> 50 experiments; Fig. 3). Stainingwith 1.O mM glutamate from a puffer pipette 20-50 Km distant from dyes (fast cresyl violet, 0.4% trypan blue), as well as probing a neuronalcell body wasinvariably followed by reversible mem- with microelectrodes,confirmed the cell-free nature of much of brane depolarization and reduction in input resistance(27127 this debris visualized with phase-contrast.Neuronal cell counts cells) (Fig. lB4); in most casesa brief burst of synaptic poten- performed on 14 representative 200 x microscopefields in sev- tials/action potentials was also seen. eral separateexperiments showed that 88.1 + 2.6% (SE) of If the pressureejection of glutamate wasnot terminated after specifically identified neurons were damagedor destroyed by a few seconds,but maintained for a minute or longer, neurons this glutamate exposure. Interestingly, neuronsin clumps(reag- near the puffer electrodewere seento swell,darken, and become gregatedfrom the original dispersedsingle-cell plating) tended more granular. These changeswere often detectableas early as to be more rapidly and more thoroughly destroyedby glutamate l-2 min after onset of glutamate exposure and continued to exposure than more isolated counterparts. evolve after the ejection pulsewas terminated. Figure 2 shows The morphology of the underlying astrocyte bed appeared Figure 5. Developmentof glutamateneurotoxicity with age in vitro.Phase-contrast micrographs of identified fieldsof cortical cells taken before(top row) and 1 dafter (bottom row) a 5 min exposureto 0.5 mM glutamate,at selectedages in vitro:A, day 5; B, day 13; and C, day 17. The exposureswere done sequentiallyon sistercultures derived from a singleplating. Bar, 50 firn. The Journal of Neuroscience, February 1987, 7(2) 363 completely unaffected by such glutamate exposure and remained stable for days afterwards; virtually no staining of the glia was ever seen (> 30 experiments) following incubation in 0.4% trypan blue. A similar lack of glutamate effect was seen with the nearly pure astrocyte cultures prepared from postnatal cortices (Fig. 4, A-C, n = 3), in contrast to the complete glial disintegration that followed sufficient exposure to the Ca ionophore A23 187 (Fig. 40; n = 3). The rate of glucose disappearance from the culture medium of a glial culture exposed to glutamate did not differ from that of a control glial culture (the media glucose concentration in both dropped from 7 mM to undetectable over 4 d), AGE IN VITRO suggesting that glutamate exposure does not grossly alter glial Figure 6. Quantitationof the developmentof glutamateneurotoxicity glucose utilization. with agein vitro. At eachspecified age in vitro, a sisterculture from a The reliable neurotoxicity produced by exposure of cortical singleplating was exposed to 0.5 mM glutamatefor 5 min,and the mean cultures to 0.5 mM glutamate for 5 min was only seen with and SEM of the injury or lossto corticalneurons in 3 representative mature cultures. Immature (5-7 d in vitro) cultures, character- microscopefields, expressed as a percentageof the originalcell count, ized by small, phase-dark neurons and incomplete spreading of arepresented. At DIV 13 and DIV 17, countswere performed on specificallyidentified neurons as described in Materialsand Methods; how- the glial mat, were unaffected by such glutamate exposure, and ever, at DIV 5, the cultureswere changing too rapidly to permit un- partial effects were seen with cultures of intermediate maturity ambiguousrecounting of previouslyidentified cells, so absolute counts (3 experiments; Fig. 5). This development of vulnerability to at selectedlocations in the dish werecompared. (In fact, in 1 field, glutamate with increasing age in vitro was repeatedtwice with substantiallymore neuronswere found at the selectedlocation 24 hr after glutamateexposure than wereoriginally present, accounting for sister cultures from a single plating; neuronal cell counts ob- the presentationof a negativemean cell loss.) tained from 1 of theserepeat experimentsare presentedin Fig- ure 6. The neurotoxicity producedby a 5 min exposureto glutamate increasedwith increasingconcentrations of glutamate (Fig. 7). 0.5 mM glutamate, in almost all casesa considerablenumber The ED,, wasconsistent among sistercultures of a given plating of apparently intact neuronscould be found a day after exposure but varied somewhat between about 50 and 100 PM among (nearly 12% of the identified neuronsin the cell countspresented different platings. A quantitative determination of neuronal in- above; Figs. 9, lOA). Thesesurviving cells excludedtrypan blue jury or lossversus glutamate concentration, measuredon sister (Fig. 1OB) and remainedmorphologically stablefor at leastsev- culturesof a single plating and representative of 2 other exper- eral more days. Electrophysiologic study of 8 such surviving iments, is presentedin Figure 8. cells (Fig. 10, C, D) showedno grossdifferences in resting po- The effective concentration of an acute pulseof bath-applied tential (63.3 + 6.5 mV, mean _+SD), input resistance(74.8 + glutamate at neuronal cell membranesis probably not much 39.7 Mn), excitability (8/8 cells), or responseto 0.5-l mM glu- reduced by the presenceof uptake mechanismsin the open, tamate (7/7 cells depolarized) compared to control neurons(see 2-dimensionalenvironment of dissociatedcell culture. One might above). Interestingly, 7/8 of the cells still exhibited spontaneous therefore expect that the neurotoxic potency of glutamatein cell synaptic activity, a surprisingly high percentagegiven the recent culture might be increasedcompared to that found in vivo. On massiveneuronal loss; this may suggestthe existence of a high the other hand, kainate, a structural analog of glutamatethat is density of synaptic interconnectionsin thesecultures. not taken up by the mammalian CNS (Johnston et al., 1979), a characteristic that perhapscontributes to the fact that it is 2 Discussion ordersof magnitudemore potent than glutamateas a neurotoxin The present experiments establishthat glutamate is a remark- in vivo (Schwartz et al., 1978), would not be expected to be ably potent, rapidly acting neurotoxin in cortical cell culture, more potent in culture than in vivo. To test the prediction that and they exploit the accessibility of culture to define the first the glutamate/kainateneurotoxicity potency ratio would be in- quantitative concentration-effect relationship for GNT on cor- creasedin culture, the drugs were directly compared on sister tical neurons.The observed ED,, of 50-100 PM is near the ED,, cultures.In 4 suchdirect comparisons,1 mM kainate wasfound of 100-300 PM reported by Brookes and Burt (1980) for GNT to be actually lesstoxic than 0.5 mM glutamate (Fig. 9A). Neu- on spinal cord neurons.The concentration of glutamaterequired ronal cell counts corroborated this finding: 1 mM kainate for 5 to kill a large percentageof the in vitro cortical neuronal pop- min wasassociated with the lossof only 13.0 + 8.5% (SE, n = ulation after 5 min of exposure (100 PM) is only 11100thof the 6) of specificallyidentified neurons,compared with the 88.1 + 10 mM concentration normally presentin wholecortex (Waelsch, 2.6% neuronal lossobserved with 0.5 mM glutamate for 5 min 195l), and presumablya smallerfraction of the actual glutamate (seeabove); different at p < 0.001, 2-tail t test). concentrationspresent in intracellular compartments.With ex- Furthermore, additional evidence that GNT in culture is not posureslonger than 5 min, even lower glutamateconcentrations affectedby uptake wasthat the toxicity of an intermediate con- could possibly be toxic. centration of glutamate (60 PM) was not detectably altered by The present experimentsalso provide a novel degreeof tem- the addition of the glutamate uptake inhibitor dihydrokainate poral resolution into the serialmorphological changes in central (DHK) (Johnstonet al., 1979)at 1 mM to the exposuresolution neuronsthat follow in the wake of excessexposure to glutamate. (Fig. 9, B-D). In control experiments, a 5 min exposure to 1 The neuronal swelling seenafter only 90 set of glutamate ap- mM DHK alone did not injure neurons. plication is comparableto the rapid swellingof locustleg muscle Although most neuronswere destroyed by brief exposure to reported within 5 min of exposure to glutamate (Duce et al., Figure 7. Relationshipof glutamatedose to neuronal loss.Phase-contrast micrographs of identified fieldsof cortical cells taken before (top row) and 1 dafter (bottom row) a 5 min exposureto various concentrationsof glutamate:A, 10 ELM;B, 50 PM; C, 100 PM;and D, 1 mM. The exposureswere done during a singleexperiment on sister cultures.The height of the letter A correspondsto 28 pm in all parts. The Journal of Neuroscience, February 1987, 7(2) 365

80 G 0 -I 50 Figure 8. Dose-response relationship i! for glutamate neurotoxicity. At each 0 40 tested concentration, the mean and SEM a of cortical neuronal injury or loss in 3 ii identified fields are expressed as a per- 2 30 centage of the original cell count. The z determinations were made in a single 20 experiment on sister cultures and are typical of 2 other experiments. The small 10 percentage of neuronal damage seen after exposure to 10-30 MM glutamate is similar to that seen in control experi- ”n I, ~-I ments without glutamate addition and, -4 -‘4 -3 thus, may represent “natural” cell death LOG GLUTAMATE CONCENTRATION in the cultures.

1983); previous in vivo observations have documented neuronal study would therefore support the hypothesis (Nadler et al., dendritic swelling only after 15-30 min (Coyle et al., 198 1). 1980)that in vivo glutamate gliotoxicity is primarily an indirect The potency ratio of glutamate relative to kainate as a neu- effect secondaryto neuronal damage,perhaps resulting from the rotoxin in the present experiments was dramatically increased accumulation of potassium or other substancesin the closed compared with that customarily found in vivo. This finding is extracellular spacesof intact tissue. similar to that reported by Brookes and Burt (1980) for spinal Besidesthe glia, immature neuronswere also apparently record cells in culture, and by Garthwaite (1985) for cerebellar sistantto GNT. The brain injury producedby parenteralkainate cells in culture but not with cerebellar cells in slice. As dis- similarly is minimal in immature rats younger than age 18 d cussed above, a likely explanation for the finding is a heightened and then increaseswith maturation until the adult pattern is neurotoxic effect of glutamate in cell culture, due to the free attained at age 35 d (Nitecka et al., 1984). The basis for the access of bath glutamate to neuronal membranes in culture. The resistanceof immature neurons to excitotoxins is not known; high vulnerability of cultured neurons to excess glutamate ex- one possibility is that immature neuronshave fewer membrane posure may thus reflect the true intrinsic vulnerability of central receptors for glutamate than mature neurons. The number of neurons to GNT. Neurons in vivo are likely protected against Na-independent glutamate binding sitesin rat cerebral cortex prolonged glutamate exposure by both the blood-brain barrier increasesdramatically betweenbirth and age 50 d (Sanderson and powerful neuronal and glial (McLennan, 1976; Schousboe, and Murphy, 1982). 198 1) glutamate uptake systems. GNT in vivo may therefore It is interesting that someneurons in culture do survive brief only occur when these protective mechanisms are either im- glutamate exposure.A subject for future study is whether these paired (e.g., a low-energy state reducing uptake function) or surviving neuronsare a random samplingof the original pop- overwhelmed (Mangano and Schwartz, 1983). ulation, surviving becauseof chance,or whether they represent It is also possible that a lower neurotoxic potency of kainate a relatively resistant subpopulationof cortical neurons.Differ- in vitro may contribute to the observed reversal ofthe glutamate/ ential regional vulnerability to kainate is seenin several loca- kainate potency ratio. The neurotoxic effect of kainate in vivo tions in vivo, including cochlear nucleus(Bird et al., 1978) and appears to depend in part on the presence of local glutamatergic hippocampus(Nadler et al., 1978); even within a single brain terminals (Biziere and Coyle, 1978), perhaps because some of region, some neurons do appear to be more vulnerable than the neurotoxicity of kainate is produced by release of endoge- others to injury by excitatory amino acids(Nadler et al., 1980; nous glutamate. Cell culture may lack the concentrated, highly Kohler and Schwartz, 1983). organized glutamatergic projections found in those in vivo re- No gross differencesin resting potential or input resistance gions (e.g., striatum, hippocampal CAl) that are highly vulner- were seenbetween the surviving neurons studiedand the orig- able to exogenously applied kainate. inal population. The simplest hypothesis, that the surviving No evidence of direct glutamate gliotoxicity was seen with neuronslack membraneglutamate receptors, is excludedby our the exposures to glutamate used here, either on glial elements direct demonstration of glutamate chemosensitivity on those in the mixed cortical cultures or in the essentially pure glial cells. Additional study will be required to searchfor other cel- cultures prepared from postnatal animals. Swelling of glial and lular characteristicsthat are correlatedwith vulnerability to glu- ependymal elements is seen with GNT in vivo (Olney, 1971), tamate. Ferrante and colleagues(1985) have recently reported but with low doses of glutamate, neurotoxicity without gliotox- that a distinct subpopulation of striatal neuronscontaining the icity has been observed (Olney et al., 197 1). Evidence of glu- enzyme nicotinamide adeninedinucleotide phosphatedispho- tamate gliotoxicity has been observed in pure glial cultures pre- rase are preserved in Huntington’s disease;a searchfor such pared from chick retinal cells (Hyndman, 1984) but only with cells in culturesexposed to glutamateseems warranted (seeNote immature, proliferating glial cultures and with more intense addedin proof). In the following paper, we report evidencethat glutamate exposure than used here (5 mM for days). The present GNT may be mediated largely by a calcium influx; thus, a .- I

GLU GLUbDHK

Figure 9. Importanceof uptake in glutamate neurotoxicityin culture. AI-3, Micrographs of an identified field of cortical cells taken before (Al, phase-contrast)and 1 dalter (A2 phase-contrast;A3 bright-fieldafter trypan blue incubation) a 5 min exposureto 1 mM kainate.A4, For comparison,a representativefield in a sisterculture exposedto 0.5 mM glutamate for 5 min (bright-field after trypan blue incubation). B, Representativefield of cortical cells before (Bl, phase-contrast)and 1 dafter (B2, bright-field after trypan blue incubation) exposureto 60 pM glutamate.C, Sameas B, in a sisterculture, exceptthat the exposureto 60 NM glutamateoccurred in the presenceof 1 mM dihydrokainate(DHK). D, Comparative cell losssustained in 5 identified fieldsdue to 60 PM glutamate,with and without 1 mM DHK (mean + SEM). Bar in AI represents100 pm and appliesto A-C. 03 L

Figure 10. Someneurons surviveglutamate exposure. A, Phase-contrastmicrograph of cortical cellstaken 1 d afterexposure to 1 mM glutamate.Some intact neuronscan be seen,which exclude0.4% trypan blue dyeafter 5 min incubation (B, bright-field).C, Higher magnification(400 x) phase-contrastview of a singlesurviving neuron (white arrow) 2 d afterexposure to 0.5 mM glutamate.Inset, Chart record of intracellular recording from that neuron; at the black arrow, a 1set pulse of 0.5 mM glutamatewas ejectedfrom a nearbypuffer pipette, resulting in depolarization, shunting of the hyperpolarizing constantcurrent pulses,and abarrage of action potentials and synaptic potentials.DI and 02, Oscilloscopetraces obtained during the samerecording from the cell in C, showing spontaneousaction potentials and anEPSP, respectively.03, Another representativeoscilloscope trace showing IPSPs recordedin another neuron that survivedglutamate exposure. Space bar in A, 100 pm (A, B) and 50gm (C). Vertical cahbration,24 mV (chart record in C), 10 mV (DZ), and 5mV (02, 03).Horizontal calibration, 3 set (C’), 10 msec(01-03). 368 Choi et al. * Glutamate Neurotoxicity in Cell Culture

reduced availability of membrane calcium conductances could Kmjevic, K. (1974) Chemical nature of synaptic transmission in ver- be another parameter that is correlated with GNT resistance. tebrates. Physiol. Res. 54: 4 18-540. Lucas, D. R., and J. P. Newhouse (1957) The toxic effect of sodium Note added in proof We have now searched for these L-glutamate on the inner layers of the retina. Arch. Opthalmol. 58: NADPH-d( +) in our cortical cultures,and have found thesecells 193-201. to be selectively resistant to the toxicity of N-methyl-n-aspar- Mangano, R. M., and R. Schwartz (1983) Chronic infusion of endog- tate, but not to the toxicity of glutamate (Koh et al., 1986). enous excitatory amino acids into rat striatum and hippocampus. Brain Res. Bull. 10: 47-5 1. McBean, G. J., and P. J. Roberts (1984) Chronic infusion of L&I- tamate causes neurotoxicity in rat striatum. Brain Res. 290: 372-375. References McCarthy, K. D., and J. deVellis (1980) Preparation of separate as- Bird, S. J., R. L. Gulley, R. J. Wenthold, and J. Fex (1978) Kainic troglial and oligodendroglial cell cultures from rat cerebral tissue. J. acid injections result in degeneration of cochlear nucleus cells inner- Cell Biol. 8.5: 890-902. vated by the auditory nerve. Science 202: 1087-1089. McGeer, E. G., and P. L. McGeer (1976) Duplication of biochemical Biziere, K., and J. T. Coyle (1978) Influence of cortio-striatal alIerents changes of Huntington’s chorea by intrastriatal injections of glutamic on striatal kainic acid neurotoxicity. Neurosci. Lett. 8: 303-3 10. and kainic acids. Nature 263: 5 17-5 19. Booher, J., and M. Sensenbrenner (1972) Growth and cultivation of McLennan, H. (1976) The autoradiographic localization of dissociated neurons and glial cells from embryonic chick, rat and t#Hldutamate in rat brain tissue. Brain Res. 115: 139-144. human brain in flask cultures. Neurobiology 2: 97-105. Nadier,J. V., B. W. Perry, and C. W. Cotman (1978) Intraventricular Brookes, N., and D. R. Burt (1980) Development of muscarinic re- kainic acid preferentially destroys hippocampal pyramidal cells. Na- ceptor binding in spinal cord cell cultures and its reduction by glutamic ture 271: 676-677. and kainic acids. Dev. Neurosci. 3: 118-127. Nadler, J. V., B. W. Perry, C. Gentry, and C. W. Cotman (1980) Choi, D. W., and G. D. Fischbach (198 1) GABA conductance of chick Degeneration of hippocampal CA3 pyramidal cells induced by in- spinal cord and dorsal root ganglion neurons in cell culture. J. Neu- traventricular kainic acid. J. Comn. Neurol. 192: 333-359. rophysiol. 45: 605-620. Nitecka, L., E. Tremblay, G. Charton, J. P. Bouillot, M. L. Berger, and Coyle, J. T., and R. Schwartz (1976) Lesion of striatal neurones with Y. Ben&i (1984) Maturation of kainic acid seizure-brain damage kainic acid provides a model for Huntington’s chorea. Nature 263: syndrome in the rat. II. Histopathological sequelae. Neuroscience 13: 244-246. 1073-1094. Coyle, J. T., S. J. Bird, R. H. Evans, R. L. Gulley, J. V. Nadler, W. J. Olney, J. W. (1969) Brain lesion, obesity and other disturbances in Nicklas, and J. W. Olney (198 1) Excitatory amino acid neurotoxins: mice treated with monosodium glutamate. Science 164: 7 19-72 1. Selectivity, specificity, and mechanisms of action. Neurosci. Res. Prog. Olney, J. W. (197 1) Glutamate-induced neuronal necrosis in the infant Bull. 19: 331-427. mouse hypothalamus. An electron microscopic study. J. Neuropathol. Crawford, J. M., and D. R. Curtis (1964) The excitation and depression Exp. Neurol. 30: 75-90. of mammalian cortical neurons by amino acids. Br. J. Pharmacol. Olney, J. W. (1978) Neurotoxicity of excitatory amino acids. In Kainic 23: 323-329. Acid as a Tool in Neurobiology, E.G. McGeer, J.W. Olney, and P.L. Curtis, D. R., J. W. Phillis, and J. C. Watkins (1960) The chemical McGeer, eds., pp. 95-121, Raven, New York. excitation of spinal neurons by certain acidic amino acids. J. Physiol. Olney, J. W., 0. L. Ho, and V. Rhee (197 1) Cytotoxic effects of acidic (Lond.) 150: 656-682. and sulphur containing amino acids on the infant mouse central ner- DiChiara, G. D., and G. L. Gessa (198 1) Glutamate as a neurotrans- vous system. Exp. Brain Res. 14: 61-76. mitter. Adv. Biochem. Psychopharmacol., Vol. 27. Plaitakis, A., S. Berl, and M. D. Yahr (1982) Abnormal glutamate Dichter, M. A. (1978) Rat cortical neurons in cell culture: Culture metabolism in an adult-onset degenerative neurological disorder. Sci- methods, cell morphology, electrophysiology, and synapse formation. ence 216: 193-196. Brain Res. 149: 279-293. Rothman, S. (1984) Synaptic release of excitatory amino acid neu- Duce, I. R., P. L. Donaldson, and P. N. R. Usherwood (1983) Inves- rotransmitter mediates anoxic neuronal death. J. Neurosci. 4: 1884- t&&ions into the mechanism ofexcitant amino acid cytotoxicity using 1891. a well-characterized dutamateraic svstem. Brain Res. 263: 77-87. Sanderson, C., and S. Murphy (1982) Glutamate binding in the rat Ferrante, R. J., N. W. Kowall, M.F. Bkal, E. P. Richardson, E. D. Bird, cerebral cortex during ontogeny. Dev. Brain Res. 2: 329-339. and J. B. Martin (1985) Selective sparing of a class of striatal neurons Schousboe, A. (198 1) Transport and metabolism of glutamate and in Huntington’s disease. Science 230: 561-563. GABA in neuronsand glial cells. Int. Rev. Neurobiol. 22: l-45. Garthwaite, J. (1985) Cellular uptake disguises action of L-glutamate Schwartz, R., D. Scholz, and J. T. Coyle (1978) Structure-activity on N-methyl-D-aspartate receptors. Br. J. Pharmacol. 85: 297-307. relations for the neurotoxicity of kainic acid derivatives and glutamate Hyndman, A: G. (1984) The effects of glutamate and kainate on cell analogues. Neuropharmacology 17: 145-l 5 1. nroliferation in retinal cultures. Invest. Gnhthalmol. Vis. Sci. 25: 558- Simon, R. P., J. H. Swan, T. Griffiths, and B. S. Meldrum (1984) 563. Blockade of N-methyl-D-aspartate receptors may protect against isch- Johnston, G. A. R., S. M. E. Kennedy, and B. Twitchin (1979) Action emit damage in the-brain. Science 226: 850-852. - of the neurotoxin kainic acid on hiah affinitv uptake of L-alutamic Sloviter. R. S. (1983) “EDikDtiC" brain damage in rats induced bv acid in rat brain slices. J. Neurochem. 32: 121-127. - sustained electrical stimulation of the perfo& path. I. Acute elecl Koh, J., S. Peters, and D. W. Choi (1986) Neurons containing NADPH- trophysiological and light microscopic studies. Brain Res. Bull. 10: diaphorase are selectively resistant to quinolinate toxicity. Science 675-697. 234: 73-76. Waelsch, H. (195 1) Glutamic acid and cerebral function. Adv. Protein Kohler, C., and R. Schwartz (1983) Comparison of ibotenate and Chem. 6: 299-341. kainate neurotoxicity in rat brain: A histological study. Neuroscience Wieloch, T. (1985) Hypoglycemia-induced neuronal damage pre- 8: 819-835. vented by an N-methyl-D-aspartate antagonist. Science 230: 68 l-683.