Brain Research, 555 (1991) 99-106 © 1991 Elsevier Science Publishers B.V. All rights reserved. 0006-8993/91/$03.50 ADONIS 0006899391168302
99
BRES 16830
Enhancement of NMDA receptor-mediated neurotoxicity in the hippocampal slice by depolarization and ischemia James J. Vornov I and Joseph T. Coyle 2 1Department of Neurology and 2Departments of Psychiatry and Neuroscience, The Johns Hopkins School of Medicine, Baltimore, MD 21205 (U.S.A.) (Accepted 5 March 1991) Key words: Ischemia; Hippocampus; Glutamate; Protein synthesis; N-Methyl-r,-aspartate; Neurotoxicity; In vitro
Evidence from animal stroke models suggests that the proximate cause of neuronal degeneration after ischemia is massive release of glutamate and activation of NMDA receptors. However, in the physiologic presence of oxygen and glucose in the rat hippoeampal slice preparation, the neurotoxicity of glutamate, as measured by inhibition of protein synthesis, requires high concentrations and is not prevented by glutamate receptor antagonists. Thus, the NMDA receptor-mediated neurotoxic effects of extraceUular glutamate accumulation during isehemia might depend on additional factors, such as neuronal depolarization. In the experiments reported here, slices were exposed to glutamate in a medium intended to mimic the ionic conditions found during ischemia, high potassium (128 mM) and low sodium (26 raM). This depolarizing medium itself inhibited protein synthesis in a manner which was partially mediated by NMDA receptor activation, since it was significantly reversed by the noncompetitive NMDA antagonist, MK-801. Furthermore, the effect of glutamate under depolarizing conditions was also significantly decreased by MK-801, suggesting that glutamate was acting at NMDA receptors. Thus, depolarization appears to enhance the sensitivity of neurons to toxic NMDA receptor activation by glutamate. Under conditions that mimic ischemia, hypoxia plus hypoglycemia, a similar protective effect of NMDA receptor antagonists was observed. Depolarization and ischemia both appeared to attenuate the neurotoxicity of non-NMDA receptor agonists. It appears that under conditions of normal glucose and oxygen, high concentrations of bath applied glutamate inhibit protein synthesis at sites other than the NMDA receptor. However, when the Na + gradient is decreased, as occurs during ischemia, glutamate's NMDA effects predominate. These findings suggest that ionic shifts may play a central role in permitting NMDA receptor-mediated ischemic neuronal damage. INTRODUCTION Large amounts of glutamate and other excitatory amino acid neurotransmitters are released into the extraceUular space during ischemia and appear to contribute to neuronal injury by activation of N M D A receptors. The best evidence for the role of N M D A receptor activation in producing ischemic damage is that neuronal degeneration can be prevented by N M D A receptor antagonists like 2-amino-5-phosphonovalerate (APV) and MK-8019,25. In dissociated neuronal tissue culture, the N M D A receptor-mediated injury caused by hypoxia 22 can be reproduced by a brief exposure to glutamate 4. The similarity between the effects of hypoxia and exogenous glutamate on these neuronal cultures has been taken as further evidence for the role of glutamate release and subsequent N M D A receptor activation in producing neuronal injury after ischemia. This evidence supports the hypothesis that glutamate released during ischemia injures neurons by neurotoxic activation of N M D A receptors. Because of the many possible differences between
dissociated neuronal culture and the intact brain, we have chosen the acute hippocampal slice preparation as a model system in which to test this hypothesis in vitro 26. The hippocampal slice has proved to be a useful model of ischemic and excitotoxic damage 5`7,SASA6,2°,27, allowing the study of well characterized synaptic connections in vitro 6. While brain slice preparations have the advantage of preserving much of the normal anatomical and structural state of fully differentiated adult neurons in vitro, the limited survival of the preparation favors acute metabolic or functional changes as measures of injury. Since the histologic signs of injury may take days to develop after an ischemic insult in vivo, only severe, acute morphologic changes are amenable to study. For example, after a 5 min period of carotid artery occlusion in the gerbil, standard histological examination of the hippocampus does not reveal the full extent of neuronal degeneration for at least 3 days 12. We have previously demonstrated that the inhibition of protein synthesis is a useful functional measure of neuronal distress in the hippocampal slice preparation 26. We initially chose inhibition of protein synthesis as a
Correspondence: J.J. Vornov, Department of Neurology, Meyer 1-130, The Johns Hopkins Hospital, Baltimore, MD 21205, U.S.A.
I00 measure of injury because an inhibition of protein synthesis is one of the few acute signs of injury in the hippocampus in vivo after a brief ischemic insult, and so might serve as a sensitive, acute indicator of neuronal injury in the slice preparation. As measured by incorporation of [3H]leucine, the rate of protein synthesis can be determined conveniently in whole slices by scintillation counting or regionally by autoradiography. Effects of many treatments on large numbers of slices can be measured simultaneously, an advantage compared to electrophysiological observation of neuronal function. As we and others have shown, protein synthesis in the slice is predominantly neuronal, so that the inhibition is a specific, quantitative measure of neuronal metabolism 15, t8,e6. However, since the limited survival of the slice does not allow reliable long-term observation of neuronal degeneration, inhibition of protein synthesis must be considered a measure of toxicity or injury and is not necessarily a correlate of neuronal death. Brief bath application of glutamate to the acute hippocampal slice preparation acutely inhibits n e u r o n a l protein synthesis and causes histological signs of neurotoxicity about 5 h after exposure 26. However, these effects could not be shown to be mediated by N M D A receptor activation, since the N M D A receptor antagonists ketamine and MK-801, alone and in combination with the less selective antagonists kynurenic acid and D N Q X , had no effect on the inhibition of protein synthesis caused by glutamate. These results did not support the hypothesis that, during ischemia, selective N M D A receptor-mediated toxicity occurs following glutamate release into the extracellular space. Because we could not induce N M D A receptor-mediated toxicity with bath applied glutamate, it seemed that other events during ischemia might enhance the N M D A receptor-mediated toxicity of glutamate. This could occur by a combination of a suppression of toxicity mediated at sites other than the N M D A receptor and/or an enhancement of N M D A receptor mediated toxicity. During ischemia or hypoxia, neurons are known to undergo a rapid loss of m e m b r a n e potential, 'ischemic depolarization '5'1°'19'24 associated with a large shift of potassium ions. Since high extracellular potassium could conceivably result in both glutamate release and an e n h a n c e m e n t of n e u r o n a l sensitivity to N M D A receptor-mediated neurotoxicity, we examined the effect of incubating slices in m e d i u m that might mimic the ionic conditions of ischemia.
MATERIALS AND METHODS Slices were prepared as described previously26. Briefly, 450 #m thick transverse rat hippocampal slices were prepared with a
Macllwain tissue chopper using standard techniques. The slices were incubated in 6 glass scintillation vials, 2 slices per vial, each containing 5 ml of Krebs-bicarbonate buffer (NaCI 122.6 mM, NaHCO 3 26.2 mM, KCI 5.4 mM, MgSO4 1.2 mM, NaH2PO4 1.2 mM, CaCI2 2.0, D-glucose 10 raM) individually bubbled with 95%O2/5%CO2 in a 37 °C water bath. Bubbling was adjusted to keep the slices moving gently on the bottom of the vial. Each experiment consisted of 2 sets of 6 vials, each set prepared independently from one hippocampus. One vial of slices from each animal was always untreated to serve as a control for that set of slices. As in previously reported experiments 26 the protocol consisted of a 30 min preincubation period, followed by a 30 rain drug treatment. For treatments of less than 30 min, drugs were added to the slices during the final part of the 30 min to ensure that incubation and recovery times were identical. In experiments involving 'ischemia'. after the usual preincubation period, the slices were placed in a glucose-free Krebs buffer pre-equilibrated with 95%N2/5%CO2 and the vials were bubbled with this oxygen-free gas. In all experiments, slices recovered for 30 min and then were labeled for 30 min with tracer concentrations of [3H]leucine (spec. act. 60 Ci/mmol; 3 /~Ci/vial, 10 nM leucine). Slices were rinsed for 5 min in ice-cold Krebs buffer, which depletes tracer from the most rapidly exchanging compartment, considered to be the extracellular space. The slices were then blotted and individually placed in tared microcentrifuge tubes and weighed. The slices were then sonicated in 6% TCA, centrifuged for 10 min at 14,000 g, and the radioactivity of an aliquot of the supernatant was determined by liquid scintillation counting as a measure of TCA-soluble [3H]leucine. In certain experiments, the TCA-soluble material was analyzed by HPLC 11 to determine total slice leucine and its specific radioactivity. The radioactivity was recovered as a single peak corresponding to leucine. In all conditions tested, the total amount of leucine per/tg protein was unchanged by experimental manipulation. The specific radioactivity of leucine in the slice was therefore proportional to the intracellular TCA-soluble leucine, the precursor pool available for protein synthesis. The TCA-precipitable pellet, containing the labeled, newly synthesized protein, was washed by resuspending in TCA, recentrifuging and dissolving in 0.1 M NaOH. The radioactivity in this NaOH solute was measured and the protein was determined by the method of Lowry~7. As described previously, glutamate increases the uptake of [3H]teucine from the extracellular space, thus increasing the specific activity of the precursor pool for protein synthesis. To correct for precursor specific activity, protein synthesis was expressed as the ratio of TCA-precipitable to TCA-soluble radioactivity. In each experiment, the soluble and precipitable [3H]leucine/~g protein and [3H]leucine/mg wet weight were calculated and compared for consistency with the calculated ratio reported here. The two slices in each vial were treated as duplicates of the same experimental condition, averaged and expressed as a percentage of the control slices in the same experiment. Each independent experiment was treated as a single observation. All statistics were performed on the multiple conditions in each experiment with multi-way factorial ANOVAs with the Sheff6 F-test as a supplementary post hoc test for differences between individual means. Statistical significance was considered to be at the P < 0.05 level. RESULTS
Glutamate exposure under depolarizing conditions We first tested the hypothesis that ionic conditions like those found during ischemia can induce N M D A receptormediated toxicity by examining the effects of large depolarizations on slices. As detailed in the first column of Table I, a 30 rain incubation of slices in a depolarizing
101 TABLE I Effects of glutamate on protein synthesis under depolarizing conditions
Normalized protein synthesis (% of control ratio of TCA ppt/sol [3H]leucine) expressed as means + S.E.M. for more than 4 experiments in duplicate. Protein synthesis was measured after a 30 rain exposure to the condition in the first column with or without simultaneous exposure to 10 mM glutamate. The inhibition due to the presence of glutamate under each condition has been calculated in the column headed 'Glutamate inhibition' by dividing the glutamate value by the no agonist value and subtracting from 100%. MK-801 was present, where indicated, at 100/tM. The data for glutamate in Krebs have been published in ref. 26. Ionic condition
No agonist
Glutamate
G/utamate inhibition
Krebs Krebs + MK-801
100 100
25.70+3.70- / -74% 27.17+3.15_~ -73% / KC1128 mM, NaCI26mM mMt r 30.40+5.20 13.80+1.70 / * -55% KC1128 mM, NaC126 / + MK-801 t_ 65.80+9.00 48.31+8.91 ~ -27% I__ ns--J Veratridine 100/~M + MK-801 50.01+11.65 20.80+4.40 -58% KC160 mM, added + MK-801 64.02+10.40 25.11+5.38 -61% Choline 120 mm, NaC126 mM + MK-801 128.41+2.01 18.45+4.62 -93% All values significantlydifferent from 100% (P < 0.001). There was a significant main effect of glutamate by ANOVA (P < 0.05). Selected subgroup analyses are shown by the lines in the table, ns not significant. *P < 0.01. ,p < 0.05.
medium, which was identical to the standard Krebs buffer except that it contained 128 mM KCI and 26 mM NaCI, inhibited protein synthesis by 70%. However, in the presence of MK-801 (100 /zM) during the entire experiment, the inhibition was reduced to only 34%. This protective effect of MK-801 suggested that depolarization itself induced an N M D A receptor-mediated inhibition of protein synthesis. As in our previous report 26, we focused on non-competitive antagonism of N M D A receptors, because of this uncertainty regarding the actual concentration of glutamate at the receptors, assuming that at the high concentration used, complete N M D A receptor antagonism would be achieved regardless of glutamate concentration. Glutamate, in normal Krebs buffer, produced a 74% inhibition of protein synthesis (second column of Table I). As we have reported previously, MK-801 (100/zM) failed to antagonize the effects of glutamate 26. In the depolarizing medium, glutamate still caused a significant inhibition of protein synthesis of 55%. However, when MK-801 (100/~M) was present during depolarization, the glutamate did not cause a significantly greater inhibition
of protein synthesis. Thus, not only did depolarization induce its own NMDA-receptor-mediated inhibition of protein synthesis, but under depolarizing conditions, the effect of glutamate also became partially NMDA-receptor-mediated, which is a marked change from the agonist effects observed under normal ionic conditions. As can also be seen in Table I, both a reduction in sodium and an increase in potassium were necessary to achieve maximum protection from the effects of glutamate. Addition of 60 mM KCI to the medium (with NaCI kept constant) was less effective (61% inhibition), as was veratridine (56% inhibition). A low sodium (26 mM) medium, with NaC1 replaced by choline-Cl, was without effect. Time course and dose-response
Time course studies of the effects of depolarization both alone and with glutamate added during depolarization are shown in Fig. 1. There appears to be a linear decline in protein synthesis with depolarizations of increasing duration (Fig. 1A). The addition of MK-801 (100/zM) during the depolarization reduces the inhibition. Glutamate exposures of increasing duration cause similar inhibition of protein synthesis in normal Krebs buffer and in the depolarizing medium (Fig. 1B). However, MK-801 under depolarizing conditions afforded clear protection against the effects of glutamate exposures of increasing duration. The curve for glutamate exposure during depolarization in the presence of MK801 is not significantly different from the curve for depolarization alone with MK-801 present. The protective effects of MK-801 with depolarization were most complete for brief glutamate exposures. As the duration of the insult was increased, there were more pronounced effects of depolarization and glutamate exposure in spite of the protective effects of MK-801. These experiments demonstrate that neither depolarization nor MK-801 alone is sufficient to prevent the 50% inhibition of protein synthesis caused by a 5 min exposure to glutamate, but the combined treatments result in protein synthesis at control levels. The relative completeness of the protection of the combination of depolarization and MK-801 against the effects of glutamate exposure are illustrated by the dose-response curves in Fig. 2. Under depolarizing conditions, the glutamate dose-response curve for a 5 min exposure is similar in shape to that observed in normal Krebs buffer 26. When MK-801 (100 /zM) is present under depolarizing conditions, however, the curve is clearly altered, with no significant effect after exposure to 10 mM glutamate. We investigated the effects of adding selective agonists under depolarizing conditions in the presence of MK-801
102
to determine whether non-NMDA receptor-mediated effects were modified by depolarization (Table II). When compared to the inhibition of protein synthesis caused by depolarization alone in the presence of MK-801 (100 /~M), there were only small additional decreases caused by AMPA, quisqualate or glutamate which did not achieve statistical significance. Thus, when slices were depolarized in the presence of MK-801, addition of non-NMDA receptor agonists was without effect, suggesting that depolarization either prevents or replaces the effects of non-NMDA receptor agonists.
Glutamate Dose Response
Under Depolarizing Conditions 150
125
.~ ~ ,00
J~ ~
75
oo O~
50
• o
Control MK801
25
0
~
........
In vitro 'ischemia'
, 1
........
Glutamate
Because our original hypothesis was that severe depo-
A. Effects of MK801 During Depolarization 120
, 1
.
.
.
.
.
.
.
.
,
10
(mM)
Fig. 2. Dose-response curves for glutamate under depolarizing conditions with and without MK-801. Normalized protein synthesis after exposure to increasing concentrations of glutamate in depolarizing medium in the presence or absence of MK-801 (100/~M). Glutamate produced a concentration-dependent inhibition of protein synthesis only in the absence of MK-801 (P < 0.01). Subgroup analysis reveals a significant difference between the two curves only for 10 mM glutamate as denoted by the asterisk.
180 oo
¢~ ~
60
20 •
,
.
,
5
-
,
10 Duration
•
,
15
.
20
,
.
25
,
30
(rnln)
of Exposure
Effects of Glutamate and MK801 During Depolarization
B. 120 ~
a
I I Glutamate
100
Alone
m Glutamate During Depolarization
• G,u!amat...0
'=~
80
J ~ ~o 0 0
larization reproduced the ionic conditions of ischemia, we compared the effects of depolarization with a simulation of 'ischemic conditions' in the slice by incubation in a medium depleted of oxygen and containing no glucose. For convenience, these conditions will be termed 'ischemic'. After the standard 30 min preincubation, the medium was replaced with Krebs buffer without glucose, previously equilibrated with 95%N2/5%CO 2. After 2 min, some slices were exposed to agonists for 5 min, an ischemic period of 7 min. This treatment is long enough to irreversibly block synaptic transmission in slices7. The effects of the ischemic conditions were similar to those of the high potassium, low sodium medium. In the absence of agonists, the ischemic conditions inhibited protein synthesis by 60% (Table III). There was a clear
40
TABLE II
20
Effects of agonists under depolarizing conditions with MK-801 present "
;
"
I0
"
I'5
"
2~0
"
2'5
"
3~0
Dunlllon ol E x p o s u r e (mln)
Fig. I. Effects of high potassium (120 mM), low sodium (26 mM) on glutamate inhibition of protein synthesis. A: normalized protein synthesis after exposure to depolarizing medium in the presence or absence of MK-801 (I00/~M) for periods of 5-30 min. There is a time-dependent inhibition of protein synthesis (P < 0.01) and a significant protective effect by MK-80I (P < 0.05). B: normalized protein synthesis after exposure to glutamate (10 mM) under the same conditions. Again, there is a time-dependent inhibition of protein synthesis (P < 0.01) and a significant protective effect by MK-801 (P < 0.05). The effect of glutamate under depolarizing conditions was not significantly different from that in normal Krebs buffer. There was no statistically significant effect of glutamate when slices were depolarized in the presence of MK-801 except for the 30 rain exposure. All values are means and standard errors for more than 4 independent experiments in duplicate.
Normalized values (% control) expressed as means + S.E.M. for more than 4 experiments in duplicate. All values were significantly different from 100% (P < 0.01). There was a significant main effect of 128 mM K by ANOVA (P < 0.01). Subgroup analysis revealed significant differences between Krebs and 128 mM K only in the presence of no agonist or glutamate (P < 0.01), not in the presence of quisqualate or AMPA. There was no significant effect of agonists within the 128 mM K condition. The data in the first column have been published in ref. 26.
Agonist None AMPA Quisqualate Glutamate
Krebs
% Agonist 128 mM K inhibition
100 47.99+7.58 -52% 56.41+11.51 -44% 27.17+3.15 -73%
65.80+9.00 54.00+5.09 51.65+9.15 48.31_+8.91
% Agonist inhibition -18% -22% -27%
103 protective effect of MK-801 (100 a M ) against ischemic inhibition of protein synthesis. The effects of APV, a competitive antagonist at the N M D A receptor, were smaller than those of MK-801, but still significant. More consistent protection was obtained by additional removal of Ca 2÷ from the medium during the ischemic insult. The period of ischemia was critical; 5 min of ischemia had no effect, while 6 and 7 min insults produced progressively larger inhibitions of protein synthesis. At 6 min, MK-801 (100/~M) and Ca2+-free medium restored protein synthesis to normal, but at 7 min, the protection conferred by these conditions was only partial. Seven minutes was chosen because both enhancements and reductions in protection could potentially be observed. When glutamate or specific agonists at non-NMDA receptors (AMPA or quisqualate) were added during the ischemic period, no further reduction in protein synthesis was observed (Table III). Similarly, in the presence of MK-801 (100/~M), APV or either antagonist in Ca2÷-free medium, there was no further inhibition produced by the
TABLE III Effects of addition of MK-801, A PV and glutamate receptor agonists on 'ischemic' inhibition of protein synthesis
Normalized protein synthesis after recovery from 'ischemia' as described in the text. Values are expressed as a percentage of control slice and as a percentage of the 'ischemia' condition without added agonist. In an overall 3-way ANOVA there was a significant (P < 0.001) main effect of protective treatment (MK-801 or APV, with or without calcium) and a significant main effect of agonist treatment (P < 0.05). Subgroup analysis by protective therapy revealed significant effects (P < 0.001) of all protective treatments compared to Krebs and a significant effect of low calcium. Within individual treatments there was no main effect of agonists. All values are means and standard errors for more than 4 independent experiments, n.d. = not determined. A gonist
Ischemia % Control
lschemia and No calcium % No % Control agonist
No agonist 37.05+5.37 n.d. MK-801 100/tM 66.97_+8.28 100% 76.98_+7.81 APV 100/~M 49.65+9.40 100% 70.19+5.55 Glutamate 10mM Glutamate + MK-801 Glutamate + APV
% No agonist
100% 100%
30.97_+4.65
n.d.
41.93_+8.86" 63%
74.27_+4.09
96%
51.13_+8.79 76%
55.43_+4.05
79%
AMPA 100gM 33.34_+6.38 AMPA + MK-801 57.02_+7.87 AMPA + APV 53.68_+1.51
n.d. 85% 80%
60.28_+9.31 59.58_+16.29
78% 85%
Quis 100#M 40.38_+3.73 n.d. Quis + MK-801 68.34_+8.48 102% 66.67_+11.52 Quis + APV 47.98_+4.36 72% 57.43_+6.92
87% 82%
addition of agonists. The effects of APV, a competitive antagonist at the N M D A receptor, were occasionally smaller than those of MK-801, but still significant. Thus, just as observed under depolarizing conditions, ischemic conditions induced an N M D A receptor-mediated inhibition of protein synthesis that could be blocked by MK-801. Further, the lack of effect of additional glutamate or non-NMDA agonists during ischemia was the same as that observed under depolarizing conditions. DISCUSSION The experiments reported here may provide some insight into the efficacy of N M D A receptor antagonists in preventing acute ischemic neuronal damage. The increase in extraceUular potassium and the neuronal depolarization that occur during ischemia may enhance neuronal sensitivity to glutamate, allowing N M D A receptormediated neurotoxicity to predominate. As assessed by inhibition of protein synthesis in our experiments, the toxic effects of glutamate were not N M D A receptormediated under conditions of normal glucose and oxygen. In contrast, under conditions of either high extracellular potassium or ischemia, the effects of glutamate could be reduced by N M D A receptor antagonists. As reported in our previous paper 26, the inhibition of protein synthesis caused by glutamate under normal ionic conditions was not decreased by the non-specific competitive glutamate receptor antagonists kynurenic acid (1 mM) or D N Q X (100 pM), either alone or in combination with MK-801. Thus, the receptor mechanisms mediating glutamate inhibition of protein synthesis under normal ionic conditions remains unclear. At high concentrations, glutamate might cause sufficient activation of N M D A receptors to produce neurotoxicity, but the N M D A receptor-mediated effects may be hidden by larger effects of glutamate at other sites which were not blocked by the competitive antagonists. Either glutamate acts at sites unaffected by receptor antagonists or these competitive antagonists are not effective at the high concentrations of glutamate necessary to overcome the protective effects of glutamate transport. Regardless of the actual site or sites of action of glutamate, we observed that, under normal physiological conditions, MK-801 was ineffective in protecting against inhibition of protein synthesis in hippocampal slices. These results differ from the many reports of experiments performed on dissociated neuronal cultures, in which N M D A receptor antagonists prevent glutamate neurotoxicity. It may be that the hippocampal slice is more sensitive to toxicity at sites other than the N M D A receptor. As discussed at length in our previous report 26, the preparation of slices entails both mechanical and
104 ischemic injury, which possibly sensitizes them to subsequent injurious events. Furthermore, experiments are restricted by the brief duration of survival of an acutely prepared slice, making acute physiologic and morphologic changes most amenable to study, and making correlation with long-term neuronal survival impossible. Thus, the differences in receptor sensitivity to glutamate toxicity under conditions of physiologic oxygen and glucose could, in part, be due to the different methodology. At the same time, NMDA receptor antagonists can protect hippocampal slices against the effects of conditions intended to mimic ischemia in vitro 5'7. We have now demonstrated that depolarization or ischemia allows MK-801 to protect against the inhibition of protein synthesis caused by glutamate exposure. Thus, in a more neuroanatomically complete brain tissue, additional factors such as membrane depolarization may influence the neurotoxicity caused by the extracellular accumulation of glutamate. Given our uncertainty regarding the site of actions of glutamate in normal Krebs buffer, the observation of NMDA receptor-mediated toxicity under conditions of depolarization and ischemia could result from a combination of increased NMDA-receptor-mediated toxicity and the suppression of other events. An increased sensitivity to NMDA-receptor-mediated toxicity during depolarization is consistent with our current understanding of this receptor-channel complex. Under normal physiological conditions, there is a voltage-sensitive block of the channel by magnesium ions 1, so that the response to NMDA will depend on membrane potential. The NMDA-mediated component of the EPSP occurs only during periods of sustained activity and membrane depolarization. However, since we previously observed that NMDA can cause neurotoxicity in normal Krebs buffer with Mg 2+ present and removal of Mg 2÷ does not alter glutamate mediated toxicity26, it seems unlikely that the primary effect of depolarization is to allow the NMDA-receptor-linked channel to open. Even though other factors may be more important, depolarization would, of course, still enhance NMDA receptor activation. The decreased effects of selective agonists at sites other than the NMDA receptor under depolarizing conditions or ischemia suggest that some of the effects of glutamate in normal Krebs buffer were suppressed by depolarization. Since the correlation between excitatory potency and the toxicity of glutamate analogues is poorer than was first thought when the term 'excitotoxicity' was coined 14'21"28, it is probable that neurotoxicity of both exogenous agonists and ischemia depend on specific mechanisms that can be dissociated from simple membrane depolarization. Thus, the 'excitotoxicity' of exces-
sive non-NMDA receptor activation was not reproduced by depolarization with high extracellular potassium, but actually appeared to be suppressed. Experiments in tissue culture suggest that the nonNMDA-mediated toxicity of glutamate is sodium dependent 3'e3. The quisqualate- and kainate-receptorgated conductances are carried by Na t predominantly and would be attenuated by the large depolarization. While the voltage sensitivity of glutamate receptor-gated conductances have been studied in isolation in vitro, the effects of high K ÷ conditions or of ischemia on receptor and channel properties have not been extensively studied in brain slices. A full investigation of neuronal physiology under conditions of depolarization and energy failure would be necessary to determine the relative currents through each type of receptor-gated channel. The decreased sensitivity to exogenous non-NMDA receptor agonists might appear to be the result of saturation of receptors with endogenous agonists released by depolarization. We believe that this is unlikely for several reasons. First, depolarization alone, which would release large amounts of glutamate does not fully reproduce the effects of glutamate alone. The degree of inhibition of protein synthesis caused by depolarization in the presence of MK-801 was much less than that produced by glutamate and MK-801 in normal Krebs buffer. Secondly, protein synthesis inhibition caused by glutamate in the presence of MK-801 was decreased in absolute level by adding depolarization, further suggesting that depolarization protected neurons from a component of toxicity, and did not simply replace the effects of exogenous glutamate with released endogenous glutamate. Finally, a large depolarization produced by elevated K ÷ with Na t concentration held constant would also release endogenous agonists, but did not decrease the effects of agonists. The effects of depolarization appear to be due to more than just release of endogenous agonists. We were interested in the possibility that depolarization could inhibit glutamate transport and thus affect glutamate toxicity. Under normal conditions, high capacity transport may reduce extracellular glutamate concentrations within the slice, protecting neurons from what would otherwise be toxic glutamate concentrations. Interestingly, we did not observe a shift in the doseresponse curve for glutamate under depolarizing conditions, which would be expected to inhibit activity of the sodium-dependent transporter 2'~3. This finding suggests that, even if sodium-dependent glutamate transport was reduced by depolarization, the total transport of glutamate was not sufficiently inhibited to increase glutamate potency. We extended our experiments to a model of ischemia
105 to determine whether its effects on the slice were similar to depolarization. The combination of hypoxia and glucose depletion produced an NMDA-receptor-mediated inhibition of protein synthesis, as did depolarization. In 'ischemic' slices, addition of glutamate or n o n - N M D A receptor agonists had no further effect. Since, at least under some conditions, protection against ischemia could be complete, an added inhibition of protein synthesis by these agonists could have been observed if the neurons had undiminished sensitivity to non-NMDA-receptor-mediated effects. If endogenous glutamate were released in sufficient amounts and glutamate uptake were inhibited, then it is possible that the glutamate receptors are actually saturated under these conditions. If receptors are indeed saturated, then the finding that MK-801 can completely protect slices suggests that the glutamate released during ischemia causes neurotoxicity solely through N M D A receptor activation. O u r observation of similarities in the pharmacological sensitivity of the hippocampal slice during depolarization with high potassium and during in vitro 'ischemia' suggests that ischemic depolarization is a critical event in both producing ischemic neuronal damage by releasing glutamate and in determining receptor mechanisms for glutamate neurotoxicity. During ischemia, energy reserves are rapidly depleted and synaptic activity ceases, REFERENCES 1 Ascher, P. and Nowak, L., A patch-clamp study of excitatory amino acid activated channels, Adv. Exp. Med. Biol., 203 (1986) 507-511. 2 Balcar, V.J. and Johnston, G.A.R., Glutamate uptake by brain slices and its relation to the depolarization of neurones by acidic amino acids, J. Neurobiol., 3 (1972) 295-301. 3 Choi, D.W., Ionic dependence of glutamate neurotoxicity, J. Neurosci., 7 (1987) 369-379. 4 Choi, D.W., Maulucci-Gedde, M. and Kriegstein, A.R., Glutamate neurotoxicity in cortical cell culture, J. Neurosci., 7 (1987) 357-368. 5 Clark, G.D. and Rothman, S.M., Blockade of excitatory amino acid receptors protects anoxic hippocampal slices, Neuroscience, 21 (1987) 665-671. 6 Dingledine, R., Brain Slices, Plenum Press, New York, 1984. 7 Dong, W., Schurr, A., Reid, K.H., Shields, C.B. and West, C.A., The rat hippocampal slice preparation as an in vitro model of ischemia, Stroke, 19 (1988) 498-502. 8 Ellren, K. and Lehmann, A., Calcium dependency of Nmethyl-D-aspartate toxicity in slices from the immature rat hippocampus, Neuroscience, 32 (1989) 371-379. 9 Gill, R., Foster, A.C. and Woodruff, G.N., MK-801 is neuroprotective in gerbils when administered during the post-ischaemic period, Neuroscience, 25 (1988) 847-855. 10 Hansen, A.J., Hounsgaard, J. and Jahnsen, H., Anoxia increases potassium conductance in hippocampal nerve cells, Acta Physiol. Scand., 115 (1982) 301-310. 11 Jones, B.N., Amino acid analysis by o-phthaldialdehyde precolumn derivatization and reverse-phase HPLC. In J.E. Shively (Ed.), Methods of Protein Microcharacterization, Humana Press, Clifton, NJ, 1986, pp. 121-151. 12 Kirino, T., Delayed neuronal death in the gerbil hippocampus
accompanied by neuronal hyperpolarization. After a variable period of this electrical silence, a large rapid depolarizing shift in m e m b r a n e potential occurs 5'1°. This depolarizing shiftl like the depolarization used in these experiments, is associated with both a change in membrane potential and a large rise in extracellular potassium. If the results of the current data are extrapolated to ischemia in vivo, the following ionic shifts might occur. The large depolarization and potassium efflux would reduce the sodium gradient, reducing the toxic effects of glutamate at sites other than the N M D A receptor. At the same time, calcium entry through the N M D A receptorgated channel would still occur, perhaps enhanced by release of the magnesium ion block of the channel, causing acute neuronal damage that is preventable with N M D A antagonists. The fact that, under some conditions, N M D A receptor antagonists can protect against neuronal degeneration even after the ischemic event, suggests that the increased glutamate release and heightened N M D A sensitivity may persist after the acute ischemic event.
Acknowledgements. This work was supported in part by National Institutes of Health Grants NS01310 to J.J.V. and NS13584 to J.T.C. MK-801 was generously provided by Merck, Sharpe and Dohme.
following ischemia, Brain Research, 239 (1982) 57-69. 13 Lajtha, A. and Sershen, H., Inhibition of amino acid uptake by the absence of Na ÷ in slices of brain, J. Neurochem., 24 (1975) 667-672. 14 Lehmann, J., Ferkany, J.W., Schaeffer, P. and Coyle, J.T., Dissociation between the excitatory and 'excitotoxic' effects of quinolinic acid analogs on the striatal cholinergic interneuron, Z Pharmacol. Exp. Ther., 232 (1985) 873-882. 15 Lipton, P. and Feig, S., Protein synthesis in guinea-pig hippocampal slices: an autoradiographic study using light and electron microscopy, Soc. Neurosci. Abstr., 13 (1987) 1145. 16 Lipton, P. and Whittingham, T.S., The effect of hypoxia on evoked potentials in the in vitro hippocampus, J. Physiol., 287 (1979) 427-438. 17 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J., Protein measurement with the Folin phenol reagent, J. Biol. Chem., 193 (1951) 265-275. 18 Phillips, L.L. and Steward, O., Protein synthesis by rat hippocampal slices maintained in vitro, J. Neurosci. Res., 21 (1988) 6-17. 19 Prenen, G.H.M., Go, G.K., Postema, E, Zuiderveen, E and Korf, J., Cerebral cation shifts in hypoxic-ischemic brain damage are prevented by the sodium channel blocker tetrodotoxin, Exp. Neurol., 99 (1988) 118-132. 20 Raley, K.M. and Lipton, P., The effects of in vitro 'ischemia' on macromolecular synthesis in rat hippocampal slices, Soc. Neurosci. Abstr., 12 (1986) 693. 21 Retz, K.C. and Coyle, J.T., The differential effects of excitatory amino acids on 45CAC12by slices from mouse striatum, Neuropharmacology, 23 (1984) 89-94. 22 Rothman, S.M., Synaptic release of excitatory amino acid neurotransmitter mediates anoxic neuronal death, J. Neurosci., 4 (1984) 1884-1891. 23 Rothman, S.M., The neurotoxicity of excitatory amino acids is
106 produced by passive chloride influx, J. Neurosci., 5 (1985) 1483-1489. 24 Siesj6, B.K. and Wieloch, T., Cerebral metabolism in ischemia: neurochemical basis for therapy, Br. J. Anaesth., 57 (1985) 47. 25 Simon, R.P., Swan, J.H., Griffiths, T. and Meldrum, B.S., Blockade of N-methyl-o-aspartate receptors may protect against ischemic damage in the brain, Science, 226 (1984) 850-852. 26 Vornov, J.J. and Coyle, J.T., Glutamate neurotoxicity and the
inhibition of protein synthesis in the hippocampal slice, ./. Neurochem., 56 (1991) 996-1006. 27 Whittingham, T.S., Lust, W.D. and Passonneau, J.V., An in vitro model of ischemia: metabolic and electrical alterations in the hippocampal slice, J. Neurosci., 4 (1984) 793-802. 28 Zaczek, R. and Coyle, J.T., Excitatory amino acid analogues: neurotoxicity and seizures, Neuropharmacology, 21 (1982) 1526.
99
BRES 16830
Enhancement of NMDA receptor-mediated neurotoxicity in the hippocampal slice by depolarization and ischemia James J. Vornov I and Joseph T. Coyle 2 1Department of Neurology and 2Departments of Psychiatry and Neuroscience, The Johns Hopkins School of Medicine, Baltimore, MD 21205 (U.S.A.) (Accepted 5 March 1991) Key words: Ischemia; Hippocampus; Glutamate; Protein synthesis; N-Methyl-r,-aspartate; Neurotoxicity; In vitro
Evidence from animal stroke models suggests that the proximate cause of neuronal degeneration after ischemia is massive release of glutamate and activation of NMDA receptors. However, in the physiologic presence of oxygen and glucose in the rat hippoeampal slice preparation, the neurotoxicity of glutamate, as measured by inhibition of protein synthesis, requires high concentrations and is not prevented by glutamate receptor antagonists. Thus, the NMDA receptor-mediated neurotoxic effects of extraceUular glutamate accumulation during isehemia might depend on additional factors, such as neuronal depolarization. In the experiments reported here, slices were exposed to glutamate in a medium intended to mimic the ionic conditions found during ischemia, high potassium (128 mM) and low sodium (26 raM). This depolarizing medium itself inhibited protein synthesis in a manner which was partially mediated by NMDA receptor activation, since it was significantly reversed by the noncompetitive NMDA antagonist, MK-801. Furthermore, the effect of glutamate under depolarizing conditions was also significantly decreased by MK-801, suggesting that glutamate was acting at NMDA receptors. Thus, depolarization appears to enhance the sensitivity of neurons to toxic NMDA receptor activation by glutamate. Under conditions that mimic ischemia, hypoxia plus hypoglycemia, a similar protective effect of NMDA receptor antagonists was observed. Depolarization and ischemia both appeared to attenuate the neurotoxicity of non-NMDA receptor agonists. It appears that under conditions of normal glucose and oxygen, high concentrations of bath applied glutamate inhibit protein synthesis at sites other than the NMDA receptor. However, when the Na + gradient is decreased, as occurs during ischemia, glutamate's NMDA effects predominate. These findings suggest that ionic shifts may play a central role in permitting NMDA receptor-mediated ischemic neuronal damage. INTRODUCTION Large amounts of glutamate and other excitatory amino acid neurotransmitters are released into the extraceUular space during ischemia and appear to contribute to neuronal injury by activation of N M D A receptors. The best evidence for the role of N M D A receptor activation in producing ischemic damage is that neuronal degeneration can be prevented by N M D A receptor antagonists like 2-amino-5-phosphonovalerate (APV) and MK-8019,25. In dissociated neuronal tissue culture, the N M D A receptor-mediated injury caused by hypoxia 22 can be reproduced by a brief exposure to glutamate 4. The similarity between the effects of hypoxia and exogenous glutamate on these neuronal cultures has been taken as further evidence for the role of glutamate release and subsequent N M D A receptor activation in producing neuronal injury after ischemia. This evidence supports the hypothesis that glutamate released during ischemia injures neurons by neurotoxic activation of N M D A receptors. Because of the many possible differences between
dissociated neuronal culture and the intact brain, we have chosen the acute hippocampal slice preparation as a model system in which to test this hypothesis in vitro 26. The hippocampal slice has proved to be a useful model of ischemic and excitotoxic damage 5`7,SASA6,2°,27, allowing the study of well characterized synaptic connections in vitro 6. While brain slice preparations have the advantage of preserving much of the normal anatomical and structural state of fully differentiated adult neurons in vitro, the limited survival of the preparation favors acute metabolic or functional changes as measures of injury. Since the histologic signs of injury may take days to develop after an ischemic insult in vivo, only severe, acute morphologic changes are amenable to study. For example, after a 5 min period of carotid artery occlusion in the gerbil, standard histological examination of the hippocampus does not reveal the full extent of neuronal degeneration for at least 3 days 12. We have previously demonstrated that the inhibition of protein synthesis is a useful functional measure of neuronal distress in the hippocampal slice preparation 26. We initially chose inhibition of protein synthesis as a
Correspondence: J.J. Vornov, Department of Neurology, Meyer 1-130, The Johns Hopkins Hospital, Baltimore, MD 21205, U.S.A.
I00 measure of injury because an inhibition of protein synthesis is one of the few acute signs of injury in the hippocampus in vivo after a brief ischemic insult, and so might serve as a sensitive, acute indicator of neuronal injury in the slice preparation. As measured by incorporation of [3H]leucine, the rate of protein synthesis can be determined conveniently in whole slices by scintillation counting or regionally by autoradiography. Effects of many treatments on large numbers of slices can be measured simultaneously, an advantage compared to electrophysiological observation of neuronal function. As we and others have shown, protein synthesis in the slice is predominantly neuronal, so that the inhibition is a specific, quantitative measure of neuronal metabolism 15, t8,e6. However, since the limited survival of the slice does not allow reliable long-term observation of neuronal degeneration, inhibition of protein synthesis must be considered a measure of toxicity or injury and is not necessarily a correlate of neuronal death. Brief bath application of glutamate to the acute hippocampal slice preparation acutely inhibits n e u r o n a l protein synthesis and causes histological signs of neurotoxicity about 5 h after exposure 26. However, these effects could not be shown to be mediated by N M D A receptor activation, since the N M D A receptor antagonists ketamine and MK-801, alone and in combination with the less selective antagonists kynurenic acid and D N Q X , had no effect on the inhibition of protein synthesis caused by glutamate. These results did not support the hypothesis that, during ischemia, selective N M D A receptor-mediated toxicity occurs following glutamate release into the extracellular space. Because we could not induce N M D A receptor-mediated toxicity with bath applied glutamate, it seemed that other events during ischemia might enhance the N M D A receptor-mediated toxicity of glutamate. This could occur by a combination of a suppression of toxicity mediated at sites other than the N M D A receptor and/or an enhancement of N M D A receptor mediated toxicity. During ischemia or hypoxia, neurons are known to undergo a rapid loss of m e m b r a n e potential, 'ischemic depolarization '5'1°'19'24 associated with a large shift of potassium ions. Since high extracellular potassium could conceivably result in both glutamate release and an e n h a n c e m e n t of n e u r o n a l sensitivity to N M D A receptor-mediated neurotoxicity, we examined the effect of incubating slices in m e d i u m that might mimic the ionic conditions of ischemia.
MATERIALS AND METHODS Slices were prepared as described previously26. Briefly, 450 #m thick transverse rat hippocampal slices were prepared with a
Macllwain tissue chopper using standard techniques. The slices were incubated in 6 glass scintillation vials, 2 slices per vial, each containing 5 ml of Krebs-bicarbonate buffer (NaCI 122.6 mM, NaHCO 3 26.2 mM, KCI 5.4 mM, MgSO4 1.2 mM, NaH2PO4 1.2 mM, CaCI2 2.0, D-glucose 10 raM) individually bubbled with 95%O2/5%CO2 in a 37 °C water bath. Bubbling was adjusted to keep the slices moving gently on the bottom of the vial. Each experiment consisted of 2 sets of 6 vials, each set prepared independently from one hippocampus. One vial of slices from each animal was always untreated to serve as a control for that set of slices. As in previously reported experiments 26 the protocol consisted of a 30 min preincubation period, followed by a 30 rain drug treatment. For treatments of less than 30 min, drugs were added to the slices during the final part of the 30 min to ensure that incubation and recovery times were identical. In experiments involving 'ischemia'. after the usual preincubation period, the slices were placed in a glucose-free Krebs buffer pre-equilibrated with 95%N2/5%CO2 and the vials were bubbled with this oxygen-free gas. In all experiments, slices recovered for 30 min and then were labeled for 30 min with tracer concentrations of [3H]leucine (spec. act. 60 Ci/mmol; 3 /~Ci/vial, 10 nM leucine). Slices were rinsed for 5 min in ice-cold Krebs buffer, which depletes tracer from the most rapidly exchanging compartment, considered to be the extracellular space. The slices were then blotted and individually placed in tared microcentrifuge tubes and weighed. The slices were then sonicated in 6% TCA, centrifuged for 10 min at 14,000 g, and the radioactivity of an aliquot of the supernatant was determined by liquid scintillation counting as a measure of TCA-soluble [3H]leucine. In certain experiments, the TCA-soluble material was analyzed by HPLC 11 to determine total slice leucine and its specific radioactivity. The radioactivity was recovered as a single peak corresponding to leucine. In all conditions tested, the total amount of leucine per/tg protein was unchanged by experimental manipulation. The specific radioactivity of leucine in the slice was therefore proportional to the intracellular TCA-soluble leucine, the precursor pool available for protein synthesis. The TCA-precipitable pellet, containing the labeled, newly synthesized protein, was washed by resuspending in TCA, recentrifuging and dissolving in 0.1 M NaOH. The radioactivity in this NaOH solute was measured and the protein was determined by the method of Lowry~7. As described previously, glutamate increases the uptake of [3H]teucine from the extracellular space, thus increasing the specific activity of the precursor pool for protein synthesis. To correct for precursor specific activity, protein synthesis was expressed as the ratio of TCA-precipitable to TCA-soluble radioactivity. In each experiment, the soluble and precipitable [3H]leucine/~g protein and [3H]leucine/mg wet weight were calculated and compared for consistency with the calculated ratio reported here. The two slices in each vial were treated as duplicates of the same experimental condition, averaged and expressed as a percentage of the control slices in the same experiment. Each independent experiment was treated as a single observation. All statistics were performed on the multiple conditions in each experiment with multi-way factorial ANOVAs with the Sheff6 F-test as a supplementary post hoc test for differences between individual means. Statistical significance was considered to be at the P < 0.05 level. RESULTS
Glutamate exposure under depolarizing conditions We first tested the hypothesis that ionic conditions like those found during ischemia can induce N M D A receptormediated toxicity by examining the effects of large depolarizations on slices. As detailed in the first column of Table I, a 30 rain incubation of slices in a depolarizing
101 TABLE I Effects of glutamate on protein synthesis under depolarizing conditions
Normalized protein synthesis (% of control ratio of TCA ppt/sol [3H]leucine) expressed as means + S.E.M. for more than 4 experiments in duplicate. Protein synthesis was measured after a 30 rain exposure to the condition in the first column with or without simultaneous exposure to 10 mM glutamate. The inhibition due to the presence of glutamate under each condition has been calculated in the column headed 'Glutamate inhibition' by dividing the glutamate value by the no agonist value and subtracting from 100%. MK-801 was present, where indicated, at 100/tM. The data for glutamate in Krebs have been published in ref. 26. Ionic condition
No agonist
Glutamate
G/utamate inhibition
Krebs Krebs + MK-801
100 100
25.70+3.70- / -74% 27.17+3.15_~ -73% / KC1128 mM, NaCI26mM mMt r 30.40+5.20 13.80+1.70 / * -55% KC1128 mM, NaC126 / + MK-801 t_ 65.80+9.00 48.31+8.91 ~ -27% I__ ns--J Veratridine 100/~M + MK-801 50.01+11.65 20.80+4.40 -58% KC160 mM, added + MK-801 64.02+10.40 25.11+5.38 -61% Choline 120 mm, NaC126 mM + MK-801 128.41+2.01 18.45+4.62 -93% All values significantlydifferent from 100% (P < 0.001). There was a significant main effect of glutamate by ANOVA (P < 0.05). Selected subgroup analyses are shown by the lines in the table, ns not significant. *P < 0.01. ,p < 0.05.
medium, which was identical to the standard Krebs buffer except that it contained 128 mM KCI and 26 mM NaCI, inhibited protein synthesis by 70%. However, in the presence of MK-801 (100 /zM) during the entire experiment, the inhibition was reduced to only 34%. This protective effect of MK-801 suggested that depolarization itself induced an N M D A receptor-mediated inhibition of protein synthesis. As in our previous report 26, we focused on non-competitive antagonism of N M D A receptors, because of this uncertainty regarding the actual concentration of glutamate at the receptors, assuming that at the high concentration used, complete N M D A receptor antagonism would be achieved regardless of glutamate concentration. Glutamate, in normal Krebs buffer, produced a 74% inhibition of protein synthesis (second column of Table I). As we have reported previously, MK-801 (100/zM) failed to antagonize the effects of glutamate 26. In the depolarizing medium, glutamate still caused a significant inhibition of protein synthesis of 55%. However, when MK-801 (100/~M) was present during depolarization, the glutamate did not cause a significantly greater inhibition
of protein synthesis. Thus, not only did depolarization induce its own NMDA-receptor-mediated inhibition of protein synthesis, but under depolarizing conditions, the effect of glutamate also became partially NMDA-receptor-mediated, which is a marked change from the agonist effects observed under normal ionic conditions. As can also be seen in Table I, both a reduction in sodium and an increase in potassium were necessary to achieve maximum protection from the effects of glutamate. Addition of 60 mM KCI to the medium (with NaCI kept constant) was less effective (61% inhibition), as was veratridine (56% inhibition). A low sodium (26 mM) medium, with NaC1 replaced by choline-Cl, was without effect. Time course and dose-response
Time course studies of the effects of depolarization both alone and with glutamate added during depolarization are shown in Fig. 1. There appears to be a linear decline in protein synthesis with depolarizations of increasing duration (Fig. 1A). The addition of MK-801 (100/zM) during the depolarization reduces the inhibition. Glutamate exposures of increasing duration cause similar inhibition of protein synthesis in normal Krebs buffer and in the depolarizing medium (Fig. 1B). However, MK-801 under depolarizing conditions afforded clear protection against the effects of glutamate exposures of increasing duration. The curve for glutamate exposure during depolarization in the presence of MK801 is not significantly different from the curve for depolarization alone with MK-801 present. The protective effects of MK-801 with depolarization were most complete for brief glutamate exposures. As the duration of the insult was increased, there were more pronounced effects of depolarization and glutamate exposure in spite of the protective effects of MK-801. These experiments demonstrate that neither depolarization nor MK-801 alone is sufficient to prevent the 50% inhibition of protein synthesis caused by a 5 min exposure to glutamate, but the combined treatments result in protein synthesis at control levels. The relative completeness of the protection of the combination of depolarization and MK-801 against the effects of glutamate exposure are illustrated by the dose-response curves in Fig. 2. Under depolarizing conditions, the glutamate dose-response curve for a 5 min exposure is similar in shape to that observed in normal Krebs buffer 26. When MK-801 (100 /zM) is present under depolarizing conditions, however, the curve is clearly altered, with no significant effect after exposure to 10 mM glutamate. We investigated the effects of adding selective agonists under depolarizing conditions in the presence of MK-801
102
to determine whether non-NMDA receptor-mediated effects were modified by depolarization (Table II). When compared to the inhibition of protein synthesis caused by depolarization alone in the presence of MK-801 (100 /~M), there were only small additional decreases caused by AMPA, quisqualate or glutamate which did not achieve statistical significance. Thus, when slices were depolarized in the presence of MK-801, addition of non-NMDA receptor agonists was without effect, suggesting that depolarization either prevents or replaces the effects of non-NMDA receptor agonists.
Glutamate Dose Response
Under Depolarizing Conditions 150
125
.~ ~ ,00
J~ ~
75
oo O~
50
• o
Control MK801
25
0
~
........
In vitro 'ischemia'
, 1
........
Glutamate
Because our original hypothesis was that severe depo-
A. Effects of MK801 During Depolarization 120
, 1
.
.
.
.
.
.
.
.
,
10
(mM)
Fig. 2. Dose-response curves for glutamate under depolarizing conditions with and without MK-801. Normalized protein synthesis after exposure to increasing concentrations of glutamate in depolarizing medium in the presence or absence of MK-801 (100/~M). Glutamate produced a concentration-dependent inhibition of protein synthesis only in the absence of MK-801 (P < 0.01). Subgroup analysis reveals a significant difference between the two curves only for 10 mM glutamate as denoted by the asterisk.
180 oo
¢~ ~
60
20 •
,
.
,
5
-
,
10 Duration
•
,
15
.
20
,
.
25
,
30
(rnln)
of Exposure
Effects of Glutamate and MK801 During Depolarization
B. 120 ~
a
I I Glutamate
100
Alone
m Glutamate During Depolarization
• G,u!amat...0
'=~
80
J ~ ~o 0 0
larization reproduced the ionic conditions of ischemia, we compared the effects of depolarization with a simulation of 'ischemic conditions' in the slice by incubation in a medium depleted of oxygen and containing no glucose. For convenience, these conditions will be termed 'ischemic'. After the standard 30 min preincubation, the medium was replaced with Krebs buffer without glucose, previously equilibrated with 95%N2/5%CO 2. After 2 min, some slices were exposed to agonists for 5 min, an ischemic period of 7 min. This treatment is long enough to irreversibly block synaptic transmission in slices7. The effects of the ischemic conditions were similar to those of the high potassium, low sodium medium. In the absence of agonists, the ischemic conditions inhibited protein synthesis by 60% (Table III). There was a clear
40
TABLE II
20
Effects of agonists under depolarizing conditions with MK-801 present "
;
"
I0
"
I'5
"
2~0
"
2'5
"
3~0
Dunlllon ol E x p o s u r e (mln)
Fig. I. Effects of high potassium (120 mM), low sodium (26 mM) on glutamate inhibition of protein synthesis. A: normalized protein synthesis after exposure to depolarizing medium in the presence or absence of MK-801 (I00/~M) for periods of 5-30 min. There is a time-dependent inhibition of protein synthesis (P < 0.01) and a significant protective effect by MK-80I (P < 0.05). B: normalized protein synthesis after exposure to glutamate (10 mM) under the same conditions. Again, there is a time-dependent inhibition of protein synthesis (P < 0.01) and a significant protective effect by MK-801 (P < 0.05). The effect of glutamate under depolarizing conditions was not significantly different from that in normal Krebs buffer. There was no statistically significant effect of glutamate when slices were depolarized in the presence of MK-801 except for the 30 rain exposure. All values are means and standard errors for more than 4 independent experiments in duplicate.
Normalized values (% control) expressed as means + S.E.M. for more than 4 experiments in duplicate. All values were significantly different from 100% (P < 0.01). There was a significant main effect of 128 mM K by ANOVA (P < 0.01). Subgroup analysis revealed significant differences between Krebs and 128 mM K only in the presence of no agonist or glutamate (P < 0.01), not in the presence of quisqualate or AMPA. There was no significant effect of agonists within the 128 mM K condition. The data in the first column have been published in ref. 26.
Agonist None AMPA Quisqualate Glutamate
Krebs
% Agonist 128 mM K inhibition
100 47.99+7.58 -52% 56.41+11.51 -44% 27.17+3.15 -73%
65.80+9.00 54.00+5.09 51.65+9.15 48.31_+8.91
% Agonist inhibition -18% -22% -27%
103 protective effect of MK-801 (100 a M ) against ischemic inhibition of protein synthesis. The effects of APV, a competitive antagonist at the N M D A receptor, were smaller than those of MK-801, but still significant. More consistent protection was obtained by additional removal of Ca 2÷ from the medium during the ischemic insult. The period of ischemia was critical; 5 min of ischemia had no effect, while 6 and 7 min insults produced progressively larger inhibitions of protein synthesis. At 6 min, MK-801 (100/~M) and Ca2+-free medium restored protein synthesis to normal, but at 7 min, the protection conferred by these conditions was only partial. Seven minutes was chosen because both enhancements and reductions in protection could potentially be observed. When glutamate or specific agonists at non-NMDA receptors (AMPA or quisqualate) were added during the ischemic period, no further reduction in protein synthesis was observed (Table III). Similarly, in the presence of MK-801 (100/~M), APV or either antagonist in Ca2÷-free medium, there was no further inhibition produced by the
TABLE III Effects of addition of MK-801, A PV and glutamate receptor agonists on 'ischemic' inhibition of protein synthesis
Normalized protein synthesis after recovery from 'ischemia' as described in the text. Values are expressed as a percentage of control slice and as a percentage of the 'ischemia' condition without added agonist. In an overall 3-way ANOVA there was a significant (P < 0.001) main effect of protective treatment (MK-801 or APV, with or without calcium) and a significant main effect of agonist treatment (P < 0.05). Subgroup analysis by protective therapy revealed significant effects (P < 0.001) of all protective treatments compared to Krebs and a significant effect of low calcium. Within individual treatments there was no main effect of agonists. All values are means and standard errors for more than 4 independent experiments, n.d. = not determined. A gonist
Ischemia % Control
lschemia and No calcium % No % Control agonist
No agonist 37.05+5.37 n.d. MK-801 100/tM 66.97_+8.28 100% 76.98_+7.81 APV 100/~M 49.65+9.40 100% 70.19+5.55 Glutamate 10mM Glutamate + MK-801 Glutamate + APV
% No agonist
100% 100%
30.97_+4.65
n.d.
41.93_+8.86" 63%
74.27_+4.09
96%
51.13_+8.79 76%
55.43_+4.05
79%
AMPA 100gM 33.34_+6.38 AMPA + MK-801 57.02_+7.87 AMPA + APV 53.68_+1.51
n.d. 85% 80%
60.28_+9.31 59.58_+16.29
78% 85%
Quis 100#M 40.38_+3.73 n.d. Quis + MK-801 68.34_+8.48 102% 66.67_+11.52 Quis + APV 47.98_+4.36 72% 57.43_+6.92
87% 82%
addition of agonists. The effects of APV, a competitive antagonist at the N M D A receptor, were occasionally smaller than those of MK-801, but still significant. Thus, just as observed under depolarizing conditions, ischemic conditions induced an N M D A receptor-mediated inhibition of protein synthesis that could be blocked by MK-801. Further, the lack of effect of additional glutamate or non-NMDA agonists during ischemia was the same as that observed under depolarizing conditions. DISCUSSION The experiments reported here may provide some insight into the efficacy of N M D A receptor antagonists in preventing acute ischemic neuronal damage. The increase in extraceUular potassium and the neuronal depolarization that occur during ischemia may enhance neuronal sensitivity to glutamate, allowing N M D A receptormediated neurotoxicity to predominate. As assessed by inhibition of protein synthesis in our experiments, the toxic effects of glutamate were not N M D A receptormediated under conditions of normal glucose and oxygen. In contrast, under conditions of either high extracellular potassium or ischemia, the effects of glutamate could be reduced by N M D A receptor antagonists. As reported in our previous paper 26, the inhibition of protein synthesis caused by glutamate under normal ionic conditions was not decreased by the non-specific competitive glutamate receptor antagonists kynurenic acid (1 mM) or D N Q X (100 pM), either alone or in combination with MK-801. Thus, the receptor mechanisms mediating glutamate inhibition of protein synthesis under normal ionic conditions remains unclear. At high concentrations, glutamate might cause sufficient activation of N M D A receptors to produce neurotoxicity, but the N M D A receptor-mediated effects may be hidden by larger effects of glutamate at other sites which were not blocked by the competitive antagonists. Either glutamate acts at sites unaffected by receptor antagonists or these competitive antagonists are not effective at the high concentrations of glutamate necessary to overcome the protective effects of glutamate transport. Regardless of the actual site or sites of action of glutamate, we observed that, under normal physiological conditions, MK-801 was ineffective in protecting against inhibition of protein synthesis in hippocampal slices. These results differ from the many reports of experiments performed on dissociated neuronal cultures, in which N M D A receptor antagonists prevent glutamate neurotoxicity. It may be that the hippocampal slice is more sensitive to toxicity at sites other than the N M D A receptor. As discussed at length in our previous report 26, the preparation of slices entails both mechanical and
104 ischemic injury, which possibly sensitizes them to subsequent injurious events. Furthermore, experiments are restricted by the brief duration of survival of an acutely prepared slice, making acute physiologic and morphologic changes most amenable to study, and making correlation with long-term neuronal survival impossible. Thus, the differences in receptor sensitivity to glutamate toxicity under conditions of physiologic oxygen and glucose could, in part, be due to the different methodology. At the same time, NMDA receptor antagonists can protect hippocampal slices against the effects of conditions intended to mimic ischemia in vitro 5'7. We have now demonstrated that depolarization or ischemia allows MK-801 to protect against the inhibition of protein synthesis caused by glutamate exposure. Thus, in a more neuroanatomically complete brain tissue, additional factors such as membrane depolarization may influence the neurotoxicity caused by the extracellular accumulation of glutamate. Given our uncertainty regarding the site of actions of glutamate in normal Krebs buffer, the observation of NMDA receptor-mediated toxicity under conditions of depolarization and ischemia could result from a combination of increased NMDA-receptor-mediated toxicity and the suppression of other events. An increased sensitivity to NMDA-receptor-mediated toxicity during depolarization is consistent with our current understanding of this receptor-channel complex. Under normal physiological conditions, there is a voltage-sensitive block of the channel by magnesium ions 1, so that the response to NMDA will depend on membrane potential. The NMDA-mediated component of the EPSP occurs only during periods of sustained activity and membrane depolarization. However, since we previously observed that NMDA can cause neurotoxicity in normal Krebs buffer with Mg 2+ present and removal of Mg 2÷ does not alter glutamate mediated toxicity26, it seems unlikely that the primary effect of depolarization is to allow the NMDA-receptor-linked channel to open. Even though other factors may be more important, depolarization would, of course, still enhance NMDA receptor activation. The decreased effects of selective agonists at sites other than the NMDA receptor under depolarizing conditions or ischemia suggest that some of the effects of glutamate in normal Krebs buffer were suppressed by depolarization. Since the correlation between excitatory potency and the toxicity of glutamate analogues is poorer than was first thought when the term 'excitotoxicity' was coined 14'21"28, it is probable that neurotoxicity of both exogenous agonists and ischemia depend on specific mechanisms that can be dissociated from simple membrane depolarization. Thus, the 'excitotoxicity' of exces-
sive non-NMDA receptor activation was not reproduced by depolarization with high extracellular potassium, but actually appeared to be suppressed. Experiments in tissue culture suggest that the nonNMDA-mediated toxicity of glutamate is sodium dependent 3'e3. The quisqualate- and kainate-receptorgated conductances are carried by Na t predominantly and would be attenuated by the large depolarization. While the voltage sensitivity of glutamate receptor-gated conductances have been studied in isolation in vitro, the effects of high K ÷ conditions or of ischemia on receptor and channel properties have not been extensively studied in brain slices. A full investigation of neuronal physiology under conditions of depolarization and energy failure would be necessary to determine the relative currents through each type of receptor-gated channel. The decreased sensitivity to exogenous non-NMDA receptor agonists might appear to be the result of saturation of receptors with endogenous agonists released by depolarization. We believe that this is unlikely for several reasons. First, depolarization alone, which would release large amounts of glutamate does not fully reproduce the effects of glutamate alone. The degree of inhibition of protein synthesis caused by depolarization in the presence of MK-801 was much less than that produced by glutamate and MK-801 in normal Krebs buffer. Secondly, protein synthesis inhibition caused by glutamate in the presence of MK-801 was decreased in absolute level by adding depolarization, further suggesting that depolarization protected neurons from a component of toxicity, and did not simply replace the effects of exogenous glutamate with released endogenous glutamate. Finally, a large depolarization produced by elevated K ÷ with Na t concentration held constant would also release endogenous agonists, but did not decrease the effects of agonists. The effects of depolarization appear to be due to more than just release of endogenous agonists. We were interested in the possibility that depolarization could inhibit glutamate transport and thus affect glutamate toxicity. Under normal conditions, high capacity transport may reduce extracellular glutamate concentrations within the slice, protecting neurons from what would otherwise be toxic glutamate concentrations. Interestingly, we did not observe a shift in the doseresponse curve for glutamate under depolarizing conditions, which would be expected to inhibit activity of the sodium-dependent transporter 2'~3. This finding suggests that, even if sodium-dependent glutamate transport was reduced by depolarization, the total transport of glutamate was not sufficiently inhibited to increase glutamate potency. We extended our experiments to a model of ischemia
105 to determine whether its effects on the slice were similar to depolarization. The combination of hypoxia and glucose depletion produced an NMDA-receptor-mediated inhibition of protein synthesis, as did depolarization. In 'ischemic' slices, addition of glutamate or n o n - N M D A receptor agonists had no further effect. Since, at least under some conditions, protection against ischemia could be complete, an added inhibition of protein synthesis by these agonists could have been observed if the neurons had undiminished sensitivity to non-NMDA-receptor-mediated effects. If endogenous glutamate were released in sufficient amounts and glutamate uptake were inhibited, then it is possible that the glutamate receptors are actually saturated under these conditions. If receptors are indeed saturated, then the finding that MK-801 can completely protect slices suggests that the glutamate released during ischemia causes neurotoxicity solely through N M D A receptor activation. O u r observation of similarities in the pharmacological sensitivity of the hippocampal slice during depolarization with high potassium and during in vitro 'ischemia' suggests that ischemic depolarization is a critical event in both producing ischemic neuronal damage by releasing glutamate and in determining receptor mechanisms for glutamate neurotoxicity. During ischemia, energy reserves are rapidly depleted and synaptic activity ceases, REFERENCES 1 Ascher, P. and Nowak, L., A patch-clamp study of excitatory amino acid activated channels, Adv. Exp. Med. Biol., 203 (1986) 507-511. 2 Balcar, V.J. and Johnston, G.A.R., Glutamate uptake by brain slices and its relation to the depolarization of neurones by acidic amino acids, J. Neurobiol., 3 (1972) 295-301. 3 Choi, D.W., Ionic dependence of glutamate neurotoxicity, J. Neurosci., 7 (1987) 369-379. 4 Choi, D.W., Maulucci-Gedde, M. and Kriegstein, A.R., Glutamate neurotoxicity in cortical cell culture, J. Neurosci., 7 (1987) 357-368. 5 Clark, G.D. and Rothman, S.M., Blockade of excitatory amino acid receptors protects anoxic hippocampal slices, Neuroscience, 21 (1987) 665-671. 6 Dingledine, R., Brain Slices, Plenum Press, New York, 1984. 7 Dong, W., Schurr, A., Reid, K.H., Shields, C.B. and West, C.A., The rat hippocampal slice preparation as an in vitro model of ischemia, Stroke, 19 (1988) 498-502. 8 Ellren, K. and Lehmann, A., Calcium dependency of Nmethyl-D-aspartate toxicity in slices from the immature rat hippocampus, Neuroscience, 32 (1989) 371-379. 9 Gill, R., Foster, A.C. and Woodruff, G.N., MK-801 is neuroprotective in gerbils when administered during the post-ischaemic period, Neuroscience, 25 (1988) 847-855. 10 Hansen, A.J., Hounsgaard, J. and Jahnsen, H., Anoxia increases potassium conductance in hippocampal nerve cells, Acta Physiol. Scand., 115 (1982) 301-310. 11 Jones, B.N., Amino acid analysis by o-phthaldialdehyde precolumn derivatization and reverse-phase HPLC. In J.E. Shively (Ed.), Methods of Protein Microcharacterization, Humana Press, Clifton, NJ, 1986, pp. 121-151. 12 Kirino, T., Delayed neuronal death in the gerbil hippocampus
accompanied by neuronal hyperpolarization. After a variable period of this electrical silence, a large rapid depolarizing shift in m e m b r a n e potential occurs 5'1°. This depolarizing shiftl like the depolarization used in these experiments, is associated with both a change in membrane potential and a large rise in extracellular potassium. If the results of the current data are extrapolated to ischemia in vivo, the following ionic shifts might occur. The large depolarization and potassium efflux would reduce the sodium gradient, reducing the toxic effects of glutamate at sites other than the N M D A receptor. At the same time, calcium entry through the N M D A receptorgated channel would still occur, perhaps enhanced by release of the magnesium ion block of the channel, causing acute neuronal damage that is preventable with N M D A antagonists. The fact that, under some conditions, N M D A receptor antagonists can protect against neuronal degeneration even after the ischemic event, suggests that the increased glutamate release and heightened N M D A sensitivity may persist after the acute ischemic event.
Acknowledgements. This work was supported in part by National Institutes of Health Grants NS01310 to J.J.V. and NS13584 to J.T.C. MK-801 was generously provided by Merck, Sharpe and Dohme.
following ischemia, Brain Research, 239 (1982) 57-69. 13 Lajtha, A. and Sershen, H., Inhibition of amino acid uptake by the absence of Na ÷ in slices of brain, J. Neurochem., 24 (1975) 667-672. 14 Lehmann, J., Ferkany, J.W., Schaeffer, P. and Coyle, J.T., Dissociation between the excitatory and 'excitotoxic' effects of quinolinic acid analogs on the striatal cholinergic interneuron, Z Pharmacol. Exp. Ther., 232 (1985) 873-882. 15 Lipton, P. and Feig, S., Protein synthesis in guinea-pig hippocampal slices: an autoradiographic study using light and electron microscopy, Soc. Neurosci. Abstr., 13 (1987) 1145. 16 Lipton, P. and Whittingham, T.S., The effect of hypoxia on evoked potentials in the in vitro hippocampus, J. Physiol., 287 (1979) 427-438. 17 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J., Protein measurement with the Folin phenol reagent, J. Biol. Chem., 193 (1951) 265-275. 18 Phillips, L.L. and Steward, O., Protein synthesis by rat hippocampal slices maintained in vitro, J. Neurosci. Res., 21 (1988) 6-17. 19 Prenen, G.H.M., Go, G.K., Postema, E, Zuiderveen, E and Korf, J., Cerebral cation shifts in hypoxic-ischemic brain damage are prevented by the sodium channel blocker tetrodotoxin, Exp. Neurol., 99 (1988) 118-132. 20 Raley, K.M. and Lipton, P., The effects of in vitro 'ischemia' on macromolecular synthesis in rat hippocampal slices, Soc. Neurosci. Abstr., 12 (1986) 693. 21 Retz, K.C. and Coyle, J.T., The differential effects of excitatory amino acids on 45CAC12by slices from mouse striatum, Neuropharmacology, 23 (1984) 89-94. 22 Rothman, S.M., Synaptic release of excitatory amino acid neurotransmitter mediates anoxic neuronal death, J. Neurosci., 4 (1984) 1884-1891. 23 Rothman, S.M., The neurotoxicity of excitatory amino acids is
106 produced by passive chloride influx, J. Neurosci., 5 (1985) 1483-1489. 24 Siesj6, B.K. and Wieloch, T., Cerebral metabolism in ischemia: neurochemical basis for therapy, Br. J. Anaesth., 57 (1985) 47. 25 Simon, R.P., Swan, J.H., Griffiths, T. and Meldrum, B.S., Blockade of N-methyl-o-aspartate receptors may protect against ischemic damage in the brain, Science, 226 (1984) 850-852. 26 Vornov, J.J. and Coyle, J.T., Glutamate neurotoxicity and the
inhibition of protein synthesis in the hippocampal slice, ./. Neurochem., 56 (1991) 996-1006. 27 Whittingham, T.S., Lust, W.D. and Passonneau, J.V., An in vitro model of ischemia: metabolic and electrical alterations in the hippocampal slice, J. Neurosci., 4 (1984) 793-802. 28 Zaczek, R. and Coyle, J.T., Excitatory amino acid analogues: neurotoxicity and seizures, Neuropharmacology, 21 (1982) 1526.
