0306-4522/91 $3.00+ 0.00 Persmuon Pms p~
Nesuoscience Voi. 43, No. 2/3, pp. 585-591, 1991 Printed in Great Britain
© 1991 IBRO
GLUTAMATE
NEUROTOXICITY
IN
SPINAL
CORD
CELL
CULTURE R. F. P . ~ A s and D. W. CHOI* Department of Neurology, Stanford University School of Medicine, Stanford, CA 94305, U.S.A. Ahmm~--Tbe neurotoxicity of glutamate was investigated quantitatively in mixed neuronal and glial spinal cord Hellcultures from fetal mice at 12-13 days of gestation. Five-minute expmure to 10-1000 ~M glutamate produt~d "_wi~__~readacute neuronal swelling, followed by neuronal degeneration over the next 24 h (~s0 for death about 100-200/tM); gila were not injured. Glutamate was neurotoxic in cultures as young as four days in vitro, although greater death was prodLu~,~l in older cultllr~. By 14-20 days in vitro, 80-90% of the neuronal population was destroyed by a 5-rain exposure to 500jaM glutamate. Acute neuronal swelling following glutamate exposure was prevented by r e p o t of e~ttra~l]ular sodium with equimolar choline, with minimal reduction in late Helldeath. Removal of extracellular calcium enhanced acute neuronal swelling but attmuated late neuronal death. Both acute neuronal swelling and late degeneration were efl'ectivelyblocked by the noncompetitive N - m e t h y l - ~ t e receptor antagonist dextrorphan and by the novel Competitive antagonist CGP 37849. Ten micromolar 7-chlorokymat'nate also inhibited glutamate neurotoxicity; protection was reversed by the addition of 1 mM glycine to the bathing medium. These observations suggest that glutamite is a potent and rapidly acting neurotoxin on cultured spinal Cord neurons, and support involvement of excitotoxicity in acute spinal cord injury. Similar to telencephalic neurons, spinal neurom expmed briefly to glutamate desenerate in a manner dependent on extracellulas Ca2+ and the activation of N-methyl-D-aspartate receptors.
Growing evidence suggests that the neurotoxicity of the excitatory transmitter glutamate may contribute to the brain damage associated with several acute insults, including hypoxia, ischemia, hypoglycemia, prolonged seizures, and trauma. 4"z2~ W ~ il~$ults disturb brain cell membrane homeostash, leadin~ to an accumulation of extracellular glutamate l.n which is probably large enough to destroy vulnerable brain neurons, s Glutamate neurotoxicity probably also participates in the p a ~ s of injury in the spinal cord. The spinal cord contaiM large amounts of free glutamate, ~3 consistent with its ubiquitous function as an excitatory neurotransmitter. Extensive electrophysiology9 and ligand binding It studies have indicated the presence of both N-methyl-D-aspartate (NMDA) and non-NMDA-type glutamate receptors, responsive to the same pharmacological agents effective at brain glutamate receptors. Direct evidence supporting involvement of glutamate neurotoxicity in spinal cord injury was recently reported by Faden and colleagues, t° who found that systemic therapy with the noncompetitive N M D A antagonist MK-801
*To whom correspondence should be addreaed. Abbreviations: DIV, days in vitro; GFAP, glial fibfillaryacid
protein; HCSS, HEPES (N-[2-hydroxyethyl]piperazineN'-[2-ethanesulfonic acid])-buffered control salt solution; LDH, lactate dchydrogenase; MEM, minimqln essential medium; NMDA, N-mcthyl-D-aspartate; NSE, neuron-q~cfllc enohtse; TBS, Tris[hydroxymethyl]aminomethane-bufl'ered saline.
improved neurological recovery after spinal cord trauma in rats. In view of this putative disease relevance, it is worthwhile to develop a detailed understanding of the mechanisms by which glutamate can injure spinal cord neurons. However, most existing studies of excitotoxicity have examined the injury of brain neurons. While the susceptibility of spinal cord neurons to excitotoxins has been demonstrated m vivo, 1s23 extension o f / n vitro approaches to spinal cord neurons has been sparse. In the present study, we characterized the neurotoxicity of glutamate on murine spinal cord neurons in ceil culture, using approaches previously developed in work with neocortical cultures. An abstract has already appeared. 25
EXPEltlMgNTAL PitOCEDUilES Cell culture
Mixed Ilpinal cord ~ ctdtUl~, contalnln~ both neuronal and glial ¢dementg were prepared from fetal Swiss-Webster mice at 12-13 days gestation (EI2-EI3) along the lines described by Ranmm et al.~ Embryos were removed from gravid mice within minutes of being killed. Using a dimecting microscope in a laminar flow hood, spinal cords were removed with forceps and dissected free of dorsal root ganglia. Timue was freely mineed with fon:e~ and placed in media containing 0.08"/. acetylated trypsin at 37°C for I h. T'msue collected by Iow-spced centrifugation was resuspended in plating m~dla c o n ~ of Eagle's minimum e~enfial medium (MEM; Em'le's udts, supplied glutaminefree) supplemented with 10% beat-inactivated bon,e serum, 10% fetal bovine ~ glutamine (total 2raM), and glucose (total 21 raM). Dissociation was accomplished by
585
586
R.F. REGANand D. W. CHOI
trituration through narrow-bore (flamed) Pasteur pipettes, and a final cell suspension was plated on Primaria (Falcon) 15-mm multiwell vessels at a density of four spinal cords per 24-well plate (approximately 2.5 x l0 s cells/well). Cultures were maintained at 37°C in a humidified CO2-containing atmosphere. At confluency of the gliai layer [five to seven days in vitro (DIV)] non-neuronal cell division was halted by a two- to three-day exposure to 10-SM cytosine arabinoside; cultures were fed once weekly with MEM supplemented with glucose (25 mM). All key comparisons were made on sister cultures derived from single platings.
to yield the signal specifically associated with glutamate neurotoxicity. The absolute value of the LDH efflux produeed by a given toxic exposure was quite consistent within sister cultures of a single plating, but differed somewhat between platings, largely as a function of neuronal density. Therefore, each observed LDH value was scaled to the mean value obtained by control glutamate exposure in sister cultures (= 100). In experiments with four to eight DIV cultures, injury was assessed by cell counts before and one day after glutamate exposure.
lmmunohistochemL~trv
Measurement of cell area
Cultures to be stained were washed in ice-cold Trisbuffered saline (TBS) and fixed with cold 4% paraformaldehyde for 30 min. After removal of fixative by TBS wash, cultures were exposed to 0.25% Triton X-100 for 10rain. Cultures were then exposed to a blocking solution consisting of 10% normal goat serum in TBS for 15 rain, followed by overnight incubation with rabbit primary antibody at room temperature, with continuous agitation provided by a shaker. After TBS wash, cultures were exposed to biotinylated secondary antibody for 30 rain, then processed for antibody binding with the avidin-biotin--peroxidas¢ system (Vectastain ABC Kit, Rabbit IgG, Vector Laboratories) and reacted with 0.05% diaminobenzidine tetrahydrochloride (Sigma) and 0.01% hydrogen peroxide for 10m in or until stained.
The mean increase in neuronal cell body area induced by 5-15-rain exposure to glutamate was quantitated. Video images of marked microscope fields were obtained under bright-field optics both before and after glutamate exposure. Images were transferred to a microcomputer, and 37-40 neurons selected for having distinct cell borders (e.g. freestanding location, not lost in a clump) were outlined with a cursor. The change in the area of each neuron was expressed as a percentage of its baseline area. Assuming that cell bodies are more or less spherical, the increase in cell area measured by this procedure should correspond roughly to cell swelling and gain in volume. An advantage of the method is the selection of a set of unconstrained neurons. Biochemical methods for determining cell swelling have other advantages, but have the disadvantage of averaging in the behavior of gha or neurons in clumps with possible mechanical constraints on swelling.
Glutamate exposure
Exposure to glutamate, either alone or in the presence of antagonist drugs, was carried out at room temperature in HEPES-buffered control salt solution (HCSS) substituted for culture media by triple exchange. HCSS had the following composition (in raM): NaCI, 120; KCI, 5.4; MgCI2, 0.8; CaCI:, 1.8; HEPES, 20; glucose, 15. Glycine (10pM) was added to HCSS in all experiments except those involving 7-chlorokynurenate. After a 5-rain exposure, this solution was washed out thoroughly (>1000-fold dilution) and replaced with another defined solution, Eagle's MEM with augmented glucose (total 25 mM), before returning cultures to the 37°C incubator. Young cultures (four to eight DIV) did not tolerate this defined replacement solution, so the replacement solution was supplemented with 3.3% horse serum, 3.3% fetal bovine serum, and 667taM glutamine. Control experiments showed that little or no cell damage was produced by this protocol if glutamate was omitted. Assessment of injury
Overall neuronal cell injury was estimated in all experiments by examination of cultures with phase-contrast microscopy at x 100-400. This examination was usually performed one day after exposure to glutamate, at which point the process of cell death was largely complete. In some experiments, this examination was verified by subsequent bright-field examination of Trypan Blue staining (0.4% for 5 min), a dye staining debris and nonviable cells. Neuronal injury was quantitatively assessed by the measurement of lactate dehydrogenase (LDH), released by damaged or destroyed cells, in the extracellular fluid one day after glutamate e x p o s u r e . 19 The small amount of LDH present in the media of sister cultures exposed to sham wash was subtracted from values obtained from treated cultures,
Materials
Dextromethorphan was purchased from Sigma. 7-Chlorokynurenate was obtained from Tocris Neuramin. Dcxtrorphan was kindly provided by Hofl'man-La Roche, and CGP 37849 was kindly provided by Ciba-Geigy. RESULTS Within the first 24-72 h after plating, neurons send out an elaborate network of processes, and over the next 10 days, develop increasingly large, phase-bright cell bodies with prominent nucleoli (Fig. la). Neurons staining for neuron-specific enolase (NSE) are found both singly and in small c l u m p s (Fig. ib), resting on a mat o f astrocytes that do not stain for NSE but do stain for gliai fibriUary acidic protein (GFAP). Mature (12-21 DIV) cultures exposed to 500/~M glutamate for 5 rain developed immediate marked neuronal swelling (Figs lc, Table 2), with cell bodies becoming phase-dark and granular. Over the next few hours, most neurons subsequently degenerated (Fig. id); this late degeneration was largely complete by the next day (20-24 h later) and was associated with a substantial efltux of L D H into the bathing medium. Neuronal degeneration was glutamate concentration dependent between 10 ~ M and ! mM; ECho
Fig. 1. Spinal cord cell culture. (a) Phase-contrast photomicrograph of an untreated spinal cord cell culture at 14 DIV. (b) Fourteen DIV culture stained with rabbit anti-NSE antibody; untreated control for comparison to c-e, which are taken from sister cultures exposed to glutamate. (c) Two hours after exposure to 500taM glutamate, stained with rabbit anti-NSE antibody. There is loss of inmatmostaining from swollen cell bodies, but staining is retained in processes and some nuclei. (d) Twenty-four hours after exposure to 500 taM glutamate, stained with rabbit anti-NSE antibody; most neurons haw degenerated. (e) Sister culture 24 h after exposure to 500 taM glutamate with 100/JM dextrorphan added, stained with rabbit anti-NSE antibody; most neurons have survived. Scale bar = 50 tam.
Glutamate neurotoxicity in spinal cord
Fig. I.
587
588
R. F. REGANand D. W. CHOI
t25
Table 2. Ionic dependence of acute glutamate-induced neuronal swelling
100
Condition Control + Dextrorphan -Sodium - Calcium - S o d i u m / - calcium
0
I
-3
I
=4.s -4 -s.s tog [ a ~ : m t , ]
I
;
-3
Fig. 2. Concentration-toxicity relationship for glutamate. Sister cultures were exposed to indicate conc©ntration of glutamate for 5nfin, and LDH present in the bathing medium was measured one day later (mean ± S.E.M., n = 4 cultures at each point), scaled to the mean value evoked by exposure to 500#M glutamate (= 100). varied somewhat on cultures from different platings, but was generally around 100-200 # M (Fig. 2, representative of nine experiments). At high glutamate concentrations, 80-90% of the original neuronal population was destroyed, as assessed by cell counts after staining with 0.4% Trypan Blue. In contrast, the astrocyte layer showed no gross injury or Trypan Blue staining. The vulnerability of spinal cord cultures to glutamate-induced injury increased with maturation /n vitro. A modest amount of neuronal cell death was noted at four DIV: the youngest at which cultures were sufficiently adherent to the dish surface to permit multiple washes. The extent of damage progresscd to the full 80-90% neuronal cell death by 12 DIV (Table 1).
Ionic dependence of glutamate neurotoxicity Glutamate neurotoxicity in conical cell culture is dependent on extraccllular sodium and calcium? Sodium is required for acute cell swelling, and calcium is required for most late cell degeneration. It was therefore of interest to assess the role of these cations in glutamate-induced injury of spinal cord neurons. The marked acute neuronal swelling induced by glutamate exposure in HCSS was blocked by the substitution of extracellular sodium with equiTable I. Development of glutamate neurotoxicity with age
in vitro Age in vitro (days) 4 8 12 16
Percentage neuronal loss 23.3 + 2.7* 45.8 ± 1.0" 80.3 ± 1.1' 80.4 + 4.3*
At each specifiedage m vitro, sister cultures were exposed to 500#M glutamate for 5min. Neuronal loss was measured by cell counts in three to four microscope fields (75-400 celts/field), expressed as a percentage (mean + S.E.M.) of the baselinecell count, after subtraction of loss due to wash alone. Asterisk indicates significant difference from sham wash control (P < 0.05 by two-tailed t-test).
Perocnlage cell area inctratse 33.75 + 3.58 3.18 + 2.19" -4.24 ± 2.37* 64.11 + 4.68* 1.48 ± 2.90*
Sister cultures were expo~t to 500 # M glutamate in normal HCSS (control), H C ~ with added 100 # M dextrorphan, or HCSS modified by removal of sodium (equimolar choline substitution), removal of calcium, or removal of both sodium and calcium. Percentage change in neuronal cell body area 5 rain after exposure is presented (mean + S.E.M., n = 37-40 cells for each condition). Asterisk indicates signitkant difference compared with control (P < 0.05, Studcnt-Newman-Keuls test). molar choline (Table 2); however, most cells still went on to degenerate (Table 3). Removal of extracellular calcium from the exposure solution actually enhanced acute neuronal swelling (Table 2), but over the next few hours the majority of these neurons returned to normal size, regained phase-bright appearance, and did not release LDH (Table 3). In follow-up observations, these protected neurons were seen to survive for at least the next few days in culture. Removal of both sodium and calcium from the exposure solution effectively prevented both acute swelling and late cell death (Tables 2, 3).
Protection by N-methyl-o-aspartate antagonist~ The selective noncompetitive NMDA receptor antagonist dextrorphan zs attenuated glutamate neurotoxicity in a concentration-dependent fashion over the range 1-100 #M, with ]c~0 approximately 20 #M (Fig. 3), and complete blockade of neuronal degeneration at 100 #M. One-hundred micromolar dextrorphan also produced neax-compicte block of acute cell swelling (Table 2). However, if the duration of glutamate exposure was increased to 15 rain, then 100 # M dextrorphan only partially attenuated nearonal swelling: in one experiment (representative of two), the percentage increase in control area was 51.2 + 3.8 (mean _+ S.E.M., n =40), compared with ! 3.7 + 3.1 with dextrorphan (P < 0.05 by two-tailed t-test). Table 3. Ionic dcpeadcncc of glutamate-induced neuronal degeneration Condition Control -Sodium -Calcium -Sodium/-calcium
LDH release 100 _+.2.4 94.6 _+2.2 20.6 +_2.6* 3.8 ± 5.2*
Sister cultures were exposed to 500 #M glutamate for 5 rain under different ionic conditions as described in the previous table legend. LDH in the bathing medium was measured one day later (mean ± S.E.M., n = 4 for each condition) and is scaled to the control value (= 100): Asterisk indicates significant difl'~'ence from control (P < 0.05 by two-tailed t-test with Bonferroni correction).
Glutamate neurotoxicity in spinal cord
(A) IO0
! 20 0
-8
-S
-3
.-,4
LO¢[ m a m m ~
589
Table 4. Glyc~ne dependence of glutamate neurotoxicity Condition
LDH release
Control 1 ~M 7-C1Kyn 10/~M 7-CIKyn 100pM 7-CIKyn 100~M 7-CIKyn + Gly I mM
100 ± 2.6 87.7 ± 6.6 5.4 -l- 4.1" -0.7 ± 4.5" 82.9 ± 4.8
Sister cultures were exposed to 500/~M glutamate for 5 min without addition of exogenous glycine (except where specifically indicated), but in the presence of the indicated concentrations of 7-chlorokynurcnate (7-CIKyn). LDH in the bathina medium as ~ one day after exposure (mean+S.E.M., n =4). Asterisk indicates significant differenoe from control (P < 0.05, two-tailed t-test with Bonferroni correction).
(B) 100
recognition site but negligible affinity for nonN M D A receptor sitesJ l
Glycine dependence
iJ -6
-S
-.4
-3
,o g'~~
(c)
120 100
!-
40
0 -20
• ' •4
-
.*IS
m -4
a -
'3
'
Fig. 3. Antagonist blockade of glutamate neurotoxicity. Media LDH (mean ± S.E.M., n = 4) in rater cultures one
day after a 5-rain exposure to 500 ~M glutamate in the preg'noe of the indicated concentrations of dextrorplum (A), dextromethorphan (B), or CGP 37849 (C), scaled to the value found in control cultures exposed to glutamate alone (=!00). Approximate ICse: for dextrorphan, 20~M; for
dextromethorphan, 60vM; for CGP 37849, 25~M. LDH value at 10-4 M in (A) was below sham wash value, but was graphed as 0 for ease of display.
The O-methyl derivative of dextrorphan, dextromethorphan, also attenuated glutamate neurotoxicity in a concentration-dependent fashion. It was less potent than dextrorphan; Ics0 was approximately 60/~M (Fig. 3). The novel competitive N M D A antagonist CGP 37849 was neuroprotective over the concentration range 1-100~M, with lcso about 25/~M (Fig. 3). This agent is an orally active phosphonoamino acid with high affinity for the N M D A receptor
Glycine in submicromolar concentrations potentiates N M D A receptor-mediated current in cortical neurons by acting at a strychnine-insensitive receptor site on the NMDA receptor complex. ~7 To eliminate this action as a variable, we routinely included 10/~M glycine in the glutamate exposure solution. However, when exogenous glycine was not added, the ability of 500 ~ M glutamate to destroy spinal cord neurons was grossly unchanged, suggesting that ambient endogenous glycine levels were sufficient to occupy the strychnine-insensitive site. To test for glycine dependence, we used the selective competitive glycine antagonist 7-chlorokynurenate, Is which has been shown to block NMDA receptor-mediated neurotoxicity on bippocampa128 and corticaP 4 neurons. 7-Chlorokynurenate produced a concentration-dependent reduction in glutamate neurotoxicity over the range 1-100~M, with I¢~0 around 2-5/~M, and essentially complete protection at 10-100/~M (Table 4). The specificity of this antagonism for the glycine site was demonstrated by the reversal of neuroprotection upon addition of I mM glycine to the experimental solution (Table 4). DISCUSSION
We present here a quantitative characterization of glutamate neurotoxicity in murine spinal cord cell cultures. Glutamate was a potent and rapidly acting toxin on spinal neurons, a finding which supports the idea that excitotoxicity may contribute substantially to secondary injury after acute spinal cord insults. With 5-min exposure, the toxic ECs0 was 100-200 ~M, a value similar to that previously noted in murine cortical culture (50-100 ~MS). Like cortical neurons, the vulnerability of spinal cord neurons to glutamate-induced injury increased with maturation/n vitro; however, spinal cord neurons exhibited greater vulnerability at younger ages. Although spinal cord neurons were genioved from E12-E13 embryos, modest glutamate neurotoxicity was seen
590
R.F. REG^Nand D. W. Cnol
as early as day 4 in culture, the youngest cultures technically suitable for study. In contrast, murine cortical cell cultures are established from El 5 to El 7 embryos, but show little vulnerability to glutamate until between seven and 10 days in culture, s The early development of vulnerability to glutamate-induced injury in spinal cord neurons may contribute to the vulnerability of the young spinal cord to damage by acute insults. In a newborn rhesus monkey subjected to brief total asphyxia, damage is found in the spinal cord, brainstem, thalamus, and basal ganglia, with relative sparing of cortex. 2
The ionic dependence of injury following brief exposure of spinal cord neurons to high concentrations of glutamate was also similar to that seen on cortical neurons, suggesting two injury components? The first component, immediate neuronal swelling, was prevented by replacement ofextracellular sodium with equimolar choline, with little diminution in subsequent late cell death. In contrast, removal of extracellular calcium increased immediate swelling, but markedly diminished neuronal death. Most likely, the swelling is mediated by the influx of sodium through glutamate receptor-activated channels, accompanied by passive influx of chloride and water. The increased swelling seen after calcium removal may reflect an increase in the mechanical compliance of the cell membrane) In the absence of extracellular calcium, glutamateinduced swelling was largely reversible; most neuronal death was attributable to a second, delayed injury component dependent on extracellular calcium. While further studies will be required to establish the basis of this calcium dependence, a critical role for calcium influx seems a likely possibility. Such a calcium influx could initiate a number of irreversibly lethal processes, including the activation of proteases and phospholipases, and the generation of free radicals. 4 The small amount of neuronal death induced by glutamate in the absence of extraceilular calcium might be a direct result of excessive swelling, or might reflect calcium influx after re-introduction of extracellular calcium.
Despite mediation of glutamate excitation of central neurons by both N M D A and non-NMDA receptors, the late neuronal death induced by brief glutamate exposure on cortical, 7 hippocampal, 27 and cerebella~ neurons has been found to be blocked by selective NMDA antagonists. The present study indicates that glutamate neurotoxicity on spinal neurons is no exception; this injury was blocked by both the noncompetitive NMDA antagonist dextrorphan and the competitive NMDA antagonist CGP 37849. The requirement of NMDA receptor activation for rapidly triggered glutamate-induced neuronal death may be related to high calcium permeability of the NMDA receptor-gated ionophore, 2° and a special linkage of N M D A receptor activation to calcium influx. The ability of N M D A antagonists to reduce glutamate neurotoxicity likely underlies the beneficial effects of these compounds in spinal cord trauma or ischemia in t,ivo. I°~l Another N M D A antagonist found to protect spinal neurons from glutamate neurotoxicity was the glycine site antagonist 7-chlorokynurenate. Since the affinity of 7-chlorokynurenate for the glycine site is similar to that of glycine itself, and since glycine is apparently present in saturating amounts (i.e. added glycine did not increase toxicity), the observed ic~0 for 7-chlorokynurenate, 2-5 # M, provides some approximation of the ambient glycine levels provided by cellular metabolism in the absence of exogenously added glycine. It is interesting to note that this value is 10-fold lower than that seen in cortical cultures, ~4 perhaps reflecting lower effective glycine concentrations in spinal cord cultures. Because glycine is a major inhibitory neurotransmitter in the spinal cord, release of endogenous glycine may be highly regulated, and uptake mechanisms may be more efficient in spinal cord cell cultures than in cortical cell cultures. As expected, the protective action of 7chlorokynurenate was reversed by the addition of 1 mM glycine. Acknowledgements--The study was supported by grant
NS26907 from the NIH, and grants from the American Paralysis Association and the Sunny yon Bulow Coma and Head Trauma Research Foundation.
REFERENCES I. Benveniste H , Drcjer J., Schusboe A. and Demer N. H. (1984) Elevation of the extracellulax concentrations of glutamate and aspar~te in rat hippocampus during transient cerebral ischemia monitored by intrao~ebral microdialysis. J. Neurochem. 43, 1369-1374. 2. Brann A. W. Jr and Schwartz J. F. (1987) Central nervous system disturbances, part 2: birth injury. In NeonatalPerinatal Medicine: Diseases of the Fetus and Infant (eds Fanaroff A. A. and Martin R. J.), pp. 506--520. The C. V.
Mosby Co., St Louis. 3. Choi D. W. (1987) Ionic dependence of glutamate neurotoxicity in cortical cell culture. J. Neurosci. 7, 369-379. 4. Choi D. W. (1988) Cr~talnate neurotoxicity and diseases of the nervous ~ . Neuron 1, 623-643. 5. Choi D. W., M a u l u c c i ~ d e M. A. and Krivgstein A. R. (1987) Glutamate neurotoxicity in cortical ccU vulture. J. Neuroaci. 7, 357-368. 6. Choi D. W., Peters S. and Viseskul V. (1987) Dextrorphan and levorphanol selc~ively block N-m©thyl-v-aspartate rec~tor-mcdiated neurotoxicity on cortical neurons. J. Pharmac. exp. Ther. 242, 713-720. 7. Choi D. W., Koh J. and Petex$ S. (1988) Pharmacology of glutamate neurotoxicity in cortical cell culture: attenuation by NMDA antagonists. J. Neurosci. 8, 185-196.
Glutamate neurotoxicityin spinal cord
591
8. Church J., Lodge D. and Berry S. C. (1985) Differential effects of dextrorphan and levorphanol on the excitation of rat spinal neurom by amino acids. Eur. J. Pharmac. 111, 185-190. 9. Davies J. and Watkim J. C. (1983) Role of excitatory amino acid receptors in mono- and polyJynaptic excitation in the cat spinal cord. Exp/Bra/~ Res. 49, 280-290. 10. Faden A. I. and Simon R. P. (1988) A potential role for excitotoxins in the pathophy~ioiogy of spinal cord injury. Ann. Neurol. 23, 623--626. 11. Fag8 G. E. et al. (1991) CGP 37849 and CGP 39551: novel competitive N-methyl-D-upartate receptor antagonists with potent oral anticonvulsant activity. In Current and Future Trends in Anticonvulsant, Anxiety, and Stroke Therapy (ed$ Meldrum B. ~ Wiiliaml M.), pp. 421--427. Alan R. Liu, New York. 12. Globus M. Y., ~ R., Dietrich W. D., Martinez E., Valdes I. and Ginsberg M. D. (1988) Effect of ischemia on the /n v/vo release of ~'iatal dopamine, glutamate, and a,amma-aminobutyric acid studied by intracerebral microdialysis. J. Neurochem. ~;1, 1455-1464. 13. Graham L. T., Shank R. P., Werman R. and Aprison M. H. (1967) Distribution of some synaptic transmitter suspects in cat spinal cord: glutamic acid, aspartic acid, gamma-aminobutyric acid, glycine, and glutamine. J. Neurochem. 14, 465--472. 14. Hartley D. M., Monyer H., Colamarino S. A. and Choi D. W. (1990) 7-Chlorokynurenate blocks NMDA receptor-mediated neurotoxicity in murine cortical culture. Eur. J. Neurosei. 2, 291-295. 15. Hugon J., Vallet J. M., Spencer P. S., Leboutet M. J. and Barthe D. (1989) Kainic acid induces early and delayed degenerative neuronal changes in rat spinal cord. Neurosci. Lett. 104, 258-262. 16. Janaen K., Faull R., Dragunow M. and Waldvogel H. (1990) Autorndiographic localisation of NMDA, quisqualate, and kaJmc acid receptors in human spinal cord. Neurosci. Lett. 108, 53-57. 17. Johnson J. W. and A~cher P. (1987) Glycine potentiates the NMDA response in cultured mouse brain neurons. Nature 325, 529-531. 18. Kemp J. A., Foster A. C., Leeson P. D., Priestiey T., Tridgett R., Iverson L. L. and Woodruff G. N. (1988) 7-Cholorokynurenic acid is a selective antagonist at the glycine modulatory site of the N-methyl-v-aspartate receptor complex. Proc. hath. Acad. Sci. U.S.A. 85, 6547-6550. 19. Koh J. Y. and Choi D. W. (1987) Quantitative determination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J. Neurosci. Met& 20, 83-90. 20. MacDermott A. B., Mayer M. L., Westbrook G. L., Smith S. J. and Barker J. L. (1986) NMDA-receptor activation incr~___~,~__cytoplasmic calcium concentration in cultured spinal cord neurons. Nature 321, 519-522. 21. Martinez-Ari~la A., Rigamonte D. D., Long J. B., Kraimer J. M. and Holaday J. W. (1990) Effects of NMDA receptor antagonists following spinal ischemia in the rabbit. E,xpl Neurol. 108, 232-240. 22. Meldrum B. (1985) Possible therapeutic applications of excitatory amino acid neurotransmitters. Ciin. Sci. 68, 113-122. 23. Pisharodi M. and Nauta H. J. (1985) An animal model for neuron-specific spinal cord lesions by the microinjection of N-methylag~artate, kainic acid, and quiglualic acid. Appl. Neurophysioi. 4& 226--233. 24. Ransom B. R., Neale E., Henkart M., Bullock P. N. and Nelson P. G. (1977) Mouse spinal cord in cell culture. I. Morphology and intrinsic neuronal electrophysiologic properties. J. Neurophys. 40, 1132-1150. 25. Regan R. F. and Choi D. W. (1990) Glutamate neurotoxicity in spinal cord cell culture. Soc. Neurosci. Abstr. 16, 1339. 26. Rothman S. M. and Olney J. W. (1987) Excitotoxicity and the NMDA receptor. Trends Neurosci. 10, 299-302. 27. Rothman S. M., Thurston J. H. and Hauhart R. E. (1987) Delayed neurotoxicity of excitatory amino acids/n vitro. Neuroscience 22, 471-480. 28. Shalaby I., Chenard B. and Prochniak M. (1989) Glycine reverses 7-C1 kynurenate blockade of glutamate neurotoxicity in cell culture. Fur. J. Pharmac. 160, 309-311. 29. Shramm M., Eimerl S. and Costa E. (1990) Serum and depolarizing agents cause acute neurotoxicity in cultured cerebellar granule cells: role of the glutamate receptor responsive to N-methyl-v-aspartate. Proc. natn. Acad. Sci. U.S.A. 87, 1193-1197. (Accepted 11 January 1991)
Nesuoscience Voi. 43, No. 2/3, pp. 585-591, 1991 Printed in Great Britain
© 1991 IBRO
GLUTAMATE
NEUROTOXICITY
IN
SPINAL
CORD
CELL
CULTURE R. F. P . ~ A s and D. W. CHOI* Department of Neurology, Stanford University School of Medicine, Stanford, CA 94305, U.S.A. Ahmm~--Tbe neurotoxicity of glutamate was investigated quantitatively in mixed neuronal and glial spinal cord Hellcultures from fetal mice at 12-13 days of gestation. Five-minute expmure to 10-1000 ~M glutamate produt~d "_wi~__~readacute neuronal swelling, followed by neuronal degeneration over the next 24 h (~s0 for death about 100-200/tM); gila were not injured. Glutamate was neurotoxic in cultures as young as four days in vitro, although greater death was prodLu~,~l in older cultllr~. By 14-20 days in vitro, 80-90% of the neuronal population was destroyed by a 5-rain exposure to 500jaM glutamate. Acute neuronal swelling following glutamate exposure was prevented by r e p o t of e~ttra~l]ular sodium with equimolar choline, with minimal reduction in late Helldeath. Removal of extracellular calcium enhanced acute neuronal swelling but attmuated late neuronal death. Both acute neuronal swelling and late degeneration were efl'ectivelyblocked by the noncompetitive N - m e t h y l - ~ t e receptor antagonist dextrorphan and by the novel Competitive antagonist CGP 37849. Ten micromolar 7-chlorokymat'nate also inhibited glutamate neurotoxicity; protection was reversed by the addition of 1 mM glycine to the bathing medium. These observations suggest that glutamite is a potent and rapidly acting neurotoxin on cultured spinal Cord neurons, and support involvement of excitotoxicity in acute spinal cord injury. Similar to telencephalic neurons, spinal neurom expmed briefly to glutamate desenerate in a manner dependent on extracellulas Ca2+ and the activation of N-methyl-D-aspartate receptors.
Growing evidence suggests that the neurotoxicity of the excitatory transmitter glutamate may contribute to the brain damage associated with several acute insults, including hypoxia, ischemia, hypoglycemia, prolonged seizures, and trauma. 4"z2~ W ~ il~$ults disturb brain cell membrane homeostash, leadin~ to an accumulation of extracellular glutamate l.n which is probably large enough to destroy vulnerable brain neurons, s Glutamate neurotoxicity probably also participates in the p a ~ s of injury in the spinal cord. The spinal cord contaiM large amounts of free glutamate, ~3 consistent with its ubiquitous function as an excitatory neurotransmitter. Extensive electrophysiology9 and ligand binding It studies have indicated the presence of both N-methyl-D-aspartate (NMDA) and non-NMDA-type glutamate receptors, responsive to the same pharmacological agents effective at brain glutamate receptors. Direct evidence supporting involvement of glutamate neurotoxicity in spinal cord injury was recently reported by Faden and colleagues, t° who found that systemic therapy with the noncompetitive N M D A antagonist MK-801
*To whom correspondence should be addreaed. Abbreviations: DIV, days in vitro; GFAP, glial fibfillaryacid
protein; HCSS, HEPES (N-[2-hydroxyethyl]piperazineN'-[2-ethanesulfonic acid])-buffered control salt solution; LDH, lactate dchydrogenase; MEM, minimqln essential medium; NMDA, N-mcthyl-D-aspartate; NSE, neuron-q~cfllc enohtse; TBS, Tris[hydroxymethyl]aminomethane-bufl'ered saline.
improved neurological recovery after spinal cord trauma in rats. In view of this putative disease relevance, it is worthwhile to develop a detailed understanding of the mechanisms by which glutamate can injure spinal cord neurons. However, most existing studies of excitotoxicity have examined the injury of brain neurons. While the susceptibility of spinal cord neurons to excitotoxins has been demonstrated m vivo, 1s23 extension o f / n vitro approaches to spinal cord neurons has been sparse. In the present study, we characterized the neurotoxicity of glutamate on murine spinal cord neurons in ceil culture, using approaches previously developed in work with neocortical cultures. An abstract has already appeared. 25
EXPEltlMgNTAL PitOCEDUilES Cell culture
Mixed Ilpinal cord ~ ctdtUl~, contalnln~ both neuronal and glial ¢dementg were prepared from fetal Swiss-Webster mice at 12-13 days gestation (EI2-EI3) along the lines described by Ranmm et al.~ Embryos were removed from gravid mice within minutes of being killed. Using a dimecting microscope in a laminar flow hood, spinal cords were removed with forceps and dissected free of dorsal root ganglia. Timue was freely mineed with fon:e~ and placed in media containing 0.08"/. acetylated trypsin at 37°C for I h. T'msue collected by Iow-spced centrifugation was resuspended in plating m~dla c o n ~ of Eagle's minimum e~enfial medium (MEM; Em'le's udts, supplied glutaminefree) supplemented with 10% beat-inactivated bon,e serum, 10% fetal bovine ~ glutamine (total 2raM), and glucose (total 21 raM). Dissociation was accomplished by
585
586
R.F. REGANand D. W. CHOI
trituration through narrow-bore (flamed) Pasteur pipettes, and a final cell suspension was plated on Primaria (Falcon) 15-mm multiwell vessels at a density of four spinal cords per 24-well plate (approximately 2.5 x l0 s cells/well). Cultures were maintained at 37°C in a humidified CO2-containing atmosphere. At confluency of the gliai layer [five to seven days in vitro (DIV)] non-neuronal cell division was halted by a two- to three-day exposure to 10-SM cytosine arabinoside; cultures were fed once weekly with MEM supplemented with glucose (25 mM). All key comparisons were made on sister cultures derived from single platings.
to yield the signal specifically associated with glutamate neurotoxicity. The absolute value of the LDH efflux produeed by a given toxic exposure was quite consistent within sister cultures of a single plating, but differed somewhat between platings, largely as a function of neuronal density. Therefore, each observed LDH value was scaled to the mean value obtained by control glutamate exposure in sister cultures (= 100). In experiments with four to eight DIV cultures, injury was assessed by cell counts before and one day after glutamate exposure.
lmmunohistochemL~trv
Measurement of cell area
Cultures to be stained were washed in ice-cold Trisbuffered saline (TBS) and fixed with cold 4% paraformaldehyde for 30 min. After removal of fixative by TBS wash, cultures were exposed to 0.25% Triton X-100 for 10rain. Cultures were then exposed to a blocking solution consisting of 10% normal goat serum in TBS for 15 rain, followed by overnight incubation with rabbit primary antibody at room temperature, with continuous agitation provided by a shaker. After TBS wash, cultures were exposed to biotinylated secondary antibody for 30 rain, then processed for antibody binding with the avidin-biotin--peroxidas¢ system (Vectastain ABC Kit, Rabbit IgG, Vector Laboratories) and reacted with 0.05% diaminobenzidine tetrahydrochloride (Sigma) and 0.01% hydrogen peroxide for 10m in or until stained.
The mean increase in neuronal cell body area induced by 5-15-rain exposure to glutamate was quantitated. Video images of marked microscope fields were obtained under bright-field optics both before and after glutamate exposure. Images were transferred to a microcomputer, and 37-40 neurons selected for having distinct cell borders (e.g. freestanding location, not lost in a clump) were outlined with a cursor. The change in the area of each neuron was expressed as a percentage of its baseline area. Assuming that cell bodies are more or less spherical, the increase in cell area measured by this procedure should correspond roughly to cell swelling and gain in volume. An advantage of the method is the selection of a set of unconstrained neurons. Biochemical methods for determining cell swelling have other advantages, but have the disadvantage of averaging in the behavior of gha or neurons in clumps with possible mechanical constraints on swelling.
Glutamate exposure
Exposure to glutamate, either alone or in the presence of antagonist drugs, was carried out at room temperature in HEPES-buffered control salt solution (HCSS) substituted for culture media by triple exchange. HCSS had the following composition (in raM): NaCI, 120; KCI, 5.4; MgCI2, 0.8; CaCI:, 1.8; HEPES, 20; glucose, 15. Glycine (10pM) was added to HCSS in all experiments except those involving 7-chlorokynurenate. After a 5-rain exposure, this solution was washed out thoroughly (>1000-fold dilution) and replaced with another defined solution, Eagle's MEM with augmented glucose (total 25 mM), before returning cultures to the 37°C incubator. Young cultures (four to eight DIV) did not tolerate this defined replacement solution, so the replacement solution was supplemented with 3.3% horse serum, 3.3% fetal bovine serum, and 667taM glutamine. Control experiments showed that little or no cell damage was produced by this protocol if glutamate was omitted. Assessment of injury
Overall neuronal cell injury was estimated in all experiments by examination of cultures with phase-contrast microscopy at x 100-400. This examination was usually performed one day after exposure to glutamate, at which point the process of cell death was largely complete. In some experiments, this examination was verified by subsequent bright-field examination of Trypan Blue staining (0.4% for 5 min), a dye staining debris and nonviable cells. Neuronal injury was quantitatively assessed by the measurement of lactate dehydrogenase (LDH), released by damaged or destroyed cells, in the extracellular fluid one day after glutamate e x p o s u r e . 19 The small amount of LDH present in the media of sister cultures exposed to sham wash was subtracted from values obtained from treated cultures,
Materials
Dextromethorphan was purchased from Sigma. 7-Chlorokynurenate was obtained from Tocris Neuramin. Dcxtrorphan was kindly provided by Hofl'man-La Roche, and CGP 37849 was kindly provided by Ciba-Geigy. RESULTS Within the first 24-72 h after plating, neurons send out an elaborate network of processes, and over the next 10 days, develop increasingly large, phase-bright cell bodies with prominent nucleoli (Fig. la). Neurons staining for neuron-specific enolase (NSE) are found both singly and in small c l u m p s (Fig. ib), resting on a mat o f astrocytes that do not stain for NSE but do stain for gliai fibriUary acidic protein (GFAP). Mature (12-21 DIV) cultures exposed to 500/~M glutamate for 5 rain developed immediate marked neuronal swelling (Figs lc, Table 2), with cell bodies becoming phase-dark and granular. Over the next few hours, most neurons subsequently degenerated (Fig. id); this late degeneration was largely complete by the next day (20-24 h later) and was associated with a substantial efltux of L D H into the bathing medium. Neuronal degeneration was glutamate concentration dependent between 10 ~ M and ! mM; ECho
Fig. 1. Spinal cord cell culture. (a) Phase-contrast photomicrograph of an untreated spinal cord cell culture at 14 DIV. (b) Fourteen DIV culture stained with rabbit anti-NSE antibody; untreated control for comparison to c-e, which are taken from sister cultures exposed to glutamate. (c) Two hours after exposure to 500taM glutamate, stained with rabbit anti-NSE antibody. There is loss of inmatmostaining from swollen cell bodies, but staining is retained in processes and some nuclei. (d) Twenty-four hours after exposure to 500 taM glutamate, stained with rabbit anti-NSE antibody; most neurons haw degenerated. (e) Sister culture 24 h after exposure to 500 taM glutamate with 100/JM dextrorphan added, stained with rabbit anti-NSE antibody; most neurons have survived. Scale bar = 50 tam.
Glutamate neurotoxicity in spinal cord
Fig. I.
587
588
R. F. REGANand D. W. CHOI
t25
Table 2. Ionic dependence of acute glutamate-induced neuronal swelling
100
Condition Control + Dextrorphan -Sodium - Calcium - S o d i u m / - calcium
0
I
-3
I
=4.s -4 -s.s tog [ a ~ : m t , ]
I
;
-3
Fig. 2. Concentration-toxicity relationship for glutamate. Sister cultures were exposed to indicate conc©ntration of glutamate for 5nfin, and LDH present in the bathing medium was measured one day later (mean ± S.E.M., n = 4 cultures at each point), scaled to the mean value evoked by exposure to 500#M glutamate (= 100). varied somewhat on cultures from different platings, but was generally around 100-200 # M (Fig. 2, representative of nine experiments). At high glutamate concentrations, 80-90% of the original neuronal population was destroyed, as assessed by cell counts after staining with 0.4% Trypan Blue. In contrast, the astrocyte layer showed no gross injury or Trypan Blue staining. The vulnerability of spinal cord cultures to glutamate-induced injury increased with maturation /n vitro. A modest amount of neuronal cell death was noted at four DIV: the youngest at which cultures were sufficiently adherent to the dish surface to permit multiple washes. The extent of damage progresscd to the full 80-90% neuronal cell death by 12 DIV (Table 1).
Ionic dependence of glutamate neurotoxicity Glutamate neurotoxicity in conical cell culture is dependent on extraccllular sodium and calcium? Sodium is required for acute cell swelling, and calcium is required for most late cell degeneration. It was therefore of interest to assess the role of these cations in glutamate-induced injury of spinal cord neurons. The marked acute neuronal swelling induced by glutamate exposure in HCSS was blocked by the substitution of extracellular sodium with equiTable I. Development of glutamate neurotoxicity with age
in vitro Age in vitro (days) 4 8 12 16
Percentage neuronal loss 23.3 + 2.7* 45.8 ± 1.0" 80.3 ± 1.1' 80.4 + 4.3*
At each specifiedage m vitro, sister cultures were exposed to 500#M glutamate for 5min. Neuronal loss was measured by cell counts in three to four microscope fields (75-400 celts/field), expressed as a percentage (mean + S.E.M.) of the baselinecell count, after subtraction of loss due to wash alone. Asterisk indicates significant difference from sham wash control (P < 0.05 by two-tailed t-test).
Perocnlage cell area inctratse 33.75 + 3.58 3.18 + 2.19" -4.24 ± 2.37* 64.11 + 4.68* 1.48 ± 2.90*
Sister cultures were expo~t to 500 # M glutamate in normal HCSS (control), H C ~ with added 100 # M dextrorphan, or HCSS modified by removal of sodium (equimolar choline substitution), removal of calcium, or removal of both sodium and calcium. Percentage change in neuronal cell body area 5 rain after exposure is presented (mean + S.E.M., n = 37-40 cells for each condition). Asterisk indicates signitkant difference compared with control (P < 0.05, Studcnt-Newman-Keuls test). molar choline (Table 2); however, most cells still went on to degenerate (Table 3). Removal of extracellular calcium from the exposure solution actually enhanced acute neuronal swelling (Table 2), but over the next few hours the majority of these neurons returned to normal size, regained phase-bright appearance, and did not release LDH (Table 3). In follow-up observations, these protected neurons were seen to survive for at least the next few days in culture. Removal of both sodium and calcium from the exposure solution effectively prevented both acute swelling and late cell death (Tables 2, 3).
Protection by N-methyl-o-aspartate antagonist~ The selective noncompetitive NMDA receptor antagonist dextrorphan zs attenuated glutamate neurotoxicity in a concentration-dependent fashion over the range 1-100 #M, with ]c~0 approximately 20 #M (Fig. 3), and complete blockade of neuronal degeneration at 100 #M. One-hundred micromolar dextrorphan also produced neax-compicte block of acute cell swelling (Table 2). However, if the duration of glutamate exposure was increased to 15 rain, then 100 # M dextrorphan only partially attenuated nearonal swelling: in one experiment (representative of two), the percentage increase in control area was 51.2 + 3.8 (mean _+ S.E.M., n =40), compared with ! 3.7 + 3.1 with dextrorphan (P < 0.05 by two-tailed t-test). Table 3. Ionic dcpeadcncc of glutamate-induced neuronal degeneration Condition Control -Sodium -Calcium -Sodium/-calcium
LDH release 100 _+.2.4 94.6 _+2.2 20.6 +_2.6* 3.8 ± 5.2*
Sister cultures were exposed to 500 #M glutamate for 5 rain under different ionic conditions as described in the previous table legend. LDH in the bathing medium was measured one day later (mean ± S.E.M., n = 4 for each condition) and is scaled to the control value (= 100): Asterisk indicates significant difl'~'ence from control (P < 0.05 by two-tailed t-test with Bonferroni correction).
Glutamate neurotoxicity in spinal cord
(A) IO0
! 20 0
-8
-S
-3
.-,4
LO¢[ m a m m ~
589
Table 4. Glyc~ne dependence of glutamate neurotoxicity Condition
LDH release
Control 1 ~M 7-C1Kyn 10/~M 7-CIKyn 100pM 7-CIKyn 100~M 7-CIKyn + Gly I mM
100 ± 2.6 87.7 ± 6.6 5.4 -l- 4.1" -0.7 ± 4.5" 82.9 ± 4.8
Sister cultures were exposed to 500/~M glutamate for 5 min without addition of exogenous glycine (except where specifically indicated), but in the presence of the indicated concentrations of 7-chlorokynurcnate (7-CIKyn). LDH in the bathina medium as ~ one day after exposure (mean+S.E.M., n =4). Asterisk indicates significant differenoe from control (P < 0.05, two-tailed t-test with Bonferroni correction).
(B) 100
recognition site but negligible affinity for nonN M D A receptor sitesJ l
Glycine dependence
iJ -6
-S
-.4
-3
,o g'~~
(c)
120 100
!-
40
0 -20
• ' •4
-
.*IS
m -4
a -
'3
'
Fig. 3. Antagonist blockade of glutamate neurotoxicity. Media LDH (mean ± S.E.M., n = 4) in rater cultures one
day after a 5-rain exposure to 500 ~M glutamate in the preg'noe of the indicated concentrations of dextrorplum (A), dextromethorphan (B), or CGP 37849 (C), scaled to the value found in control cultures exposed to glutamate alone (=!00). Approximate ICse: for dextrorphan, 20~M; for
dextromethorphan, 60vM; for CGP 37849, 25~M. LDH value at 10-4 M in (A) was below sham wash value, but was graphed as 0 for ease of display.
The O-methyl derivative of dextrorphan, dextromethorphan, also attenuated glutamate neurotoxicity in a concentration-dependent fashion. It was less potent than dextrorphan; Ics0 was approximately 60/~M (Fig. 3). The novel competitive N M D A antagonist CGP 37849 was neuroprotective over the concentration range 1-100~M, with lcso about 25/~M (Fig. 3). This agent is an orally active phosphonoamino acid with high affinity for the N M D A receptor
Glycine in submicromolar concentrations potentiates N M D A receptor-mediated current in cortical neurons by acting at a strychnine-insensitive receptor site on the NMDA receptor complex. ~7 To eliminate this action as a variable, we routinely included 10/~M glycine in the glutamate exposure solution. However, when exogenous glycine was not added, the ability of 500 ~ M glutamate to destroy spinal cord neurons was grossly unchanged, suggesting that ambient endogenous glycine levels were sufficient to occupy the strychnine-insensitive site. To test for glycine dependence, we used the selective competitive glycine antagonist 7-chlorokynurenate, Is which has been shown to block NMDA receptor-mediated neurotoxicity on bippocampa128 and corticaP 4 neurons. 7-Chlorokynurenate produced a concentration-dependent reduction in glutamate neurotoxicity over the range 1-100~M, with I¢~0 around 2-5/~M, and essentially complete protection at 10-100/~M (Table 4). The specificity of this antagonism for the glycine site was demonstrated by the reversal of neuroprotection upon addition of I mM glycine to the experimental solution (Table 4). DISCUSSION
We present here a quantitative characterization of glutamate neurotoxicity in murine spinal cord cell cultures. Glutamate was a potent and rapidly acting toxin on spinal neurons, a finding which supports the idea that excitotoxicity may contribute substantially to secondary injury after acute spinal cord insults. With 5-min exposure, the toxic ECs0 was 100-200 ~M, a value similar to that previously noted in murine cortical culture (50-100 ~MS). Like cortical neurons, the vulnerability of spinal cord neurons to glutamate-induced injury increased with maturation/n vitro; however, spinal cord neurons exhibited greater vulnerability at younger ages. Although spinal cord neurons were genioved from E12-E13 embryos, modest glutamate neurotoxicity was seen
590
R.F. REG^Nand D. W. Cnol
as early as day 4 in culture, the youngest cultures technically suitable for study. In contrast, murine cortical cell cultures are established from El 5 to El 7 embryos, but show little vulnerability to glutamate until between seven and 10 days in culture, s The early development of vulnerability to glutamate-induced injury in spinal cord neurons may contribute to the vulnerability of the young spinal cord to damage by acute insults. In a newborn rhesus monkey subjected to brief total asphyxia, damage is found in the spinal cord, brainstem, thalamus, and basal ganglia, with relative sparing of cortex. 2
The ionic dependence of injury following brief exposure of spinal cord neurons to high concentrations of glutamate was also similar to that seen on cortical neurons, suggesting two injury components? The first component, immediate neuronal swelling, was prevented by replacement ofextracellular sodium with equimolar choline, with little diminution in subsequent late cell death. In contrast, removal of extracellular calcium increased immediate swelling, but markedly diminished neuronal death. Most likely, the swelling is mediated by the influx of sodium through glutamate receptor-activated channels, accompanied by passive influx of chloride and water. The increased swelling seen after calcium removal may reflect an increase in the mechanical compliance of the cell membrane) In the absence of extracellular calcium, glutamateinduced swelling was largely reversible; most neuronal death was attributable to a second, delayed injury component dependent on extracellular calcium. While further studies will be required to establish the basis of this calcium dependence, a critical role for calcium influx seems a likely possibility. Such a calcium influx could initiate a number of irreversibly lethal processes, including the activation of proteases and phospholipases, and the generation of free radicals. 4 The small amount of neuronal death induced by glutamate in the absence of extraceilular calcium might be a direct result of excessive swelling, or might reflect calcium influx after re-introduction of extracellular calcium.
Despite mediation of glutamate excitation of central neurons by both N M D A and non-NMDA receptors, the late neuronal death induced by brief glutamate exposure on cortical, 7 hippocampal, 27 and cerebella~ neurons has been found to be blocked by selective NMDA antagonists. The present study indicates that glutamate neurotoxicity on spinal neurons is no exception; this injury was blocked by both the noncompetitive NMDA antagonist dextrorphan and the competitive NMDA antagonist CGP 37849. The requirement of NMDA receptor activation for rapidly triggered glutamate-induced neuronal death may be related to high calcium permeability of the NMDA receptor-gated ionophore, 2° and a special linkage of N M D A receptor activation to calcium influx. The ability of N M D A antagonists to reduce glutamate neurotoxicity likely underlies the beneficial effects of these compounds in spinal cord trauma or ischemia in t,ivo. I°~l Another N M D A antagonist found to protect spinal neurons from glutamate neurotoxicity was the glycine site antagonist 7-chlorokynurenate. Since the affinity of 7-chlorokynurenate for the glycine site is similar to that of glycine itself, and since glycine is apparently present in saturating amounts (i.e. added glycine did not increase toxicity), the observed ic~0 for 7-chlorokynurenate, 2-5 # M, provides some approximation of the ambient glycine levels provided by cellular metabolism in the absence of exogenously added glycine. It is interesting to note that this value is 10-fold lower than that seen in cortical cultures, ~4 perhaps reflecting lower effective glycine concentrations in spinal cord cultures. Because glycine is a major inhibitory neurotransmitter in the spinal cord, release of endogenous glycine may be highly regulated, and uptake mechanisms may be more efficient in spinal cord cell cultures than in cortical cell cultures. As expected, the protective action of 7chlorokynurenate was reversed by the addition of 1 mM glycine. Acknowledgements--The study was supported by grant
NS26907 from the NIH, and grants from the American Paralysis Association and the Sunny yon Bulow Coma and Head Trauma Research Foundation.
REFERENCES I. Benveniste H , Drcjer J., Schusboe A. and Demer N. H. (1984) Elevation of the extracellulax concentrations of glutamate and aspar~te in rat hippocampus during transient cerebral ischemia monitored by intrao~ebral microdialysis. J. Neurochem. 43, 1369-1374. 2. Brann A. W. Jr and Schwartz J. F. (1987) Central nervous system disturbances, part 2: birth injury. In NeonatalPerinatal Medicine: Diseases of the Fetus and Infant (eds Fanaroff A. A. and Martin R. J.), pp. 506--520. The C. V.
Mosby Co., St Louis. 3. Choi D. W. (1987) Ionic dependence of glutamate neurotoxicity in cortical cell culture. J. Neurosci. 7, 369-379. 4. Choi D. W. (1988) Cr~talnate neurotoxicity and diseases of the nervous ~ . Neuron 1, 623-643. 5. Choi D. W., M a u l u c c i ~ d e M. A. and Krivgstein A. R. (1987) Glutamate neurotoxicity in cortical ccU vulture. J. Neuroaci. 7, 357-368. 6. Choi D. W., Peters S. and Viseskul V. (1987) Dextrorphan and levorphanol selc~ively block N-m©thyl-v-aspartate rec~tor-mcdiated neurotoxicity on cortical neurons. J. Pharmac. exp. Ther. 242, 713-720. 7. Choi D. W., Koh J. and Petex$ S. (1988) Pharmacology of glutamate neurotoxicity in cortical cell culture: attenuation by NMDA antagonists. J. Neurosci. 8, 185-196.
Glutamate neurotoxicityin spinal cord
591
8. Church J., Lodge D. and Berry S. C. (1985) Differential effects of dextrorphan and levorphanol on the excitation of rat spinal neurom by amino acids. Eur. J. Pharmac. 111, 185-190. 9. Davies J. and Watkim J. C. (1983) Role of excitatory amino acid receptors in mono- and polyJynaptic excitation in the cat spinal cord. Exp/Bra/~ Res. 49, 280-290. 10. Faden A. I. and Simon R. P. (1988) A potential role for excitotoxins in the pathophy~ioiogy of spinal cord injury. Ann. Neurol. 23, 623--626. 11. Fag8 G. E. et al. (1991) CGP 37849 and CGP 39551: novel competitive N-methyl-D-upartate receptor antagonists with potent oral anticonvulsant activity. In Current and Future Trends in Anticonvulsant, Anxiety, and Stroke Therapy (ed$ Meldrum B. ~ Wiiliaml M.), pp. 421--427. Alan R. Liu, New York. 12. Globus M. Y., ~ R., Dietrich W. D., Martinez E., Valdes I. and Ginsberg M. D. (1988) Effect of ischemia on the /n v/vo release of ~'iatal dopamine, glutamate, and a,amma-aminobutyric acid studied by intracerebral microdialysis. J. Neurochem. ~;1, 1455-1464. 13. Graham L. T., Shank R. P., Werman R. and Aprison M. H. (1967) Distribution of some synaptic transmitter suspects in cat spinal cord: glutamic acid, aspartic acid, gamma-aminobutyric acid, glycine, and glutamine. J. Neurochem. 14, 465--472. 14. Hartley D. M., Monyer H., Colamarino S. A. and Choi D. W. (1990) 7-Chlorokynurenate blocks NMDA receptor-mediated neurotoxicity in murine cortical culture. Eur. J. Neurosei. 2, 291-295. 15. Hugon J., Vallet J. M., Spencer P. S., Leboutet M. J. and Barthe D. (1989) Kainic acid induces early and delayed degenerative neuronal changes in rat spinal cord. Neurosci. Lett. 104, 258-262. 16. Janaen K., Faull R., Dragunow M. and Waldvogel H. (1990) Autorndiographic localisation of NMDA, quisqualate, and kaJmc acid receptors in human spinal cord. Neurosci. Lett. 108, 53-57. 17. Johnson J. W. and A~cher P. (1987) Glycine potentiates the NMDA response in cultured mouse brain neurons. Nature 325, 529-531. 18. Kemp J. A., Foster A. C., Leeson P. D., Priestiey T., Tridgett R., Iverson L. L. and Woodruff G. N. (1988) 7-Cholorokynurenic acid is a selective antagonist at the glycine modulatory site of the N-methyl-v-aspartate receptor complex. Proc. hath. Acad. Sci. U.S.A. 85, 6547-6550. 19. Koh J. Y. and Choi D. W. (1987) Quantitative determination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J. Neurosci. Met& 20, 83-90. 20. MacDermott A. B., Mayer M. L., Westbrook G. L., Smith S. J. and Barker J. L. (1986) NMDA-receptor activation incr~___~,~__cytoplasmic calcium concentration in cultured spinal cord neurons. Nature 321, 519-522. 21. Martinez-Ari~la A., Rigamonte D. D., Long J. B., Kraimer J. M. and Holaday J. W. (1990) Effects of NMDA receptor antagonists following spinal ischemia in the rabbit. E,xpl Neurol. 108, 232-240. 22. Meldrum B. (1985) Possible therapeutic applications of excitatory amino acid neurotransmitters. Ciin. Sci. 68, 113-122. 23. Pisharodi M. and Nauta H. J. (1985) An animal model for neuron-specific spinal cord lesions by the microinjection of N-methylag~artate, kainic acid, and quiglualic acid. Appl. Neurophysioi. 4& 226--233. 24. Ransom B. R., Neale E., Henkart M., Bullock P. N. and Nelson P. G. (1977) Mouse spinal cord in cell culture. I. Morphology and intrinsic neuronal electrophysiologic properties. J. Neurophys. 40, 1132-1150. 25. Regan R. F. and Choi D. W. (1990) Glutamate neurotoxicity in spinal cord cell culture. Soc. Neurosci. Abstr. 16, 1339. 26. Rothman S. M. and Olney J. W. (1987) Excitotoxicity and the NMDA receptor. Trends Neurosci. 10, 299-302. 27. Rothman S. M., Thurston J. H. and Hauhart R. E. (1987) Delayed neurotoxicity of excitatory amino acids/n vitro. Neuroscience 22, 471-480. 28. Shalaby I., Chenard B. and Prochniak M. (1989) Glycine reverses 7-C1 kynurenate blockade of glutamate neurotoxicity in cell culture. Fur. J. Pharmac. 160, 309-311. 29. Shramm M., Eimerl S. and Costa E. (1990) Serum and depolarizing agents cause acute neurotoxicity in cultured cerebellar granule cells: role of the glutamate receptor responsive to N-methyl-v-aspartate. Proc. natn. Acad. Sci. U.S.A. 87, 1193-1197. (Accepted 11 January 1991)
