- Email: [email protected]
Appearance of NMDA receptors triggered by anoxia independent of voltage in vivo and in vitro
EXPERIMENTAL
NEUROLOGY
112,
304-311
(19%)
Appearance of NMDA Receptors Triggered by Anoxia Independent of Voltage in Viva and in Vitro N. HoRI,**~
N. DOI,+ S. MNAHARA,~
Y. SHINODA,?
AND D. 0. CARPENTER*
*Wadsworth Center for Laboratories and Research, New York State Department of Health and School of Public Health, Albany, New York 12201-0509; iDepartment of Pharmacology, Kyushu University Faculty of Dentistry, Fukuoka 812, Japan; and *Department of Biology, Kyushu University Faculty of Science, Fukuoka 812, Japan
and CA3 show characteristic changes in response to synaptic activation, and finally neurons lose electrical activity (delayed neuronal death, DND). There are consistent differences in susceptibility to DND in these three hippocampal areas, with CA1 being the most sensitive and area dentata the least (9). The mechanisms that are responsible for DND are incompletely understood, and a variety of factors such as excessive activity of excitatory amino acids (3,6,22), accumulation of intracellular calcium (2), and depletion of intracellular energy stores (12,18) have been considered. The observation that different populations of morphologically quite similar neurons in the same brain region have different vulnerabilities to DND presents the possibility of a preparation in which one may study factors that contribute to DND. Such study is of importance not only for the understanding of the mechanisms of (and potentially ways of preventing) direct anoxic cell death in humans, but also because of the possibility that similar mechanisms may underlie neuronal cell death in a variety of human diseases characterized by selective neuronal loss, such as Alzheimer’s, amyotrophic lateral sclerosis, and Parkinson’s Disease. We have compared in viva and in vitro models of anoxic damage and provide evidence in support of the postulate that one can study DND in an in vitro brain slice preparation. Such a preparation may have considerable utility for investigating the mechanisms of DND. In this report we document the models and present preliminary evidence for the involvement of N-methyl-D-aspartate (NMDA) receptor activation in DND in both systems.
Using rat hippocampus we have studied the pattern of neuronal death, abnormal discharge and loss of electrical excitability in slices prepared from animals subjected to bilateral, four-vessel cerebral anoxia and in slices prepared from normal animals that are subjected to anoxia in the recording chamber. As others have reported, pyramidal neurons in area CA1 are lost first after anoxia, while CA3 neurons have an intermediate sensitivity, and those in dentate are relatively anoxiaresistant. After anoxic damage to the intact animal, neurons in both CA1 and CA3 show abnormal bursting discharges in response to synaptic activation for several days, and then the response in CA1 decreases in amplitude and finally the area become unexcitable. While antagonists for N-methyl-D-aspartate (NMDA) receptors have essentially no effect on synaptic responses in control animals, they reduce the bursting responses and greatly depress the small responses in CA1 as neurons are becoming unexcitable after anoxia. With intracellular recording CA1 neurons from animals made transiently anoxic, in contrast to controls, show prolonged synaptic responses, the later components of which are blocked by NMDA antagonists. When slices from normal animals are subjected to anoxia such that excitability is totally lost over a period of about 10 min, there is no significant membrane depolarization during the anoxic episode and recovery of excitability occurs with reoxygenation. However, a period of hyperexcitability and bursting follows and electrical excitability is lost in CA1 but not CA3 neurons after about SO min. Both the bursting and loss of CA1 excitability are prevented by application of amino phosphono-valeric acid (APV) or phencyclidine (PCP) during the anoxic episode. With intracellular recording CA1 neurons from slices made transiently anoxic but not from control slices show prolonged, APV-sensitive synaptic potentials. These observations suggest that in both in vivo and in vitro preparations a critical effect of transient anoxia is the induction of previously absent NMDA receptors and that the trigger for this induction is not voltage dependent. 0 1991
Academic
Press,
METHODS Rats (200-250 g) were used in all experiments. In one series of experiments 26 animals were subjected to transient forebrain ischemia (occlusion of both carotids and both vertebral arteries for 10 min) 1 to 7 days prior to preparation of brain slices for electrophysiological study using the method of Pulsinelli and Brierley (20). Other studies were done by preparing slices from 38 normal animals, then subjecting the slices to anoxia in the experimental chamber. In both studies animals were killed by cervical dislocation under light ether anesthe-
Inc.
INTRODUCTION Two or three days after transient forebrain ischemia, hippocampal pyramidal neurons in area dentata, CAl, 0014-4886/91 Copyright All rights
$3.00 0 1991 by Academic Press, of reproduction in any form
304 Inc. reserved.
NMDA
RECEPTORS
AND
DELAYED
sia and the brains were quickly removed and placed in iced Krebs-Ringer. Slices (450 pm) were cut perpendicular to the long axis of the hippocampus using a simple vibratome (Nomiyama, Japan) and were preincubated in oxygenated Krebs-Ringer (NaCl, 126 n&f; KCl, 5 mM; KH,PO,, 1.26 mA4; MgSO,, 1.3 n&f; CaCl,, 2.4 mM; NaHCO,, 26 mM; and glucose, 10 mM) saturated with 95% 0, and 5% CO, for about 2 h at 35°C prior to mounting in the recording chamber. The slices were submerged on a plexi mesh in the chamber and perfused with Krebs-Ringer at a rate of 3-5 ml/min at 35°C. A monopolar electrode was used for stimulation of synaptic inputs to area dentata, CAl, and CA3. For activation of area dentata the electrode was placed on the perforant path, for CA1 the electrode was placed on the Shaffer collaterals in CA3, while for CA3 activation the electrode was placed on the mossy fibers. Stimulation pulses were 50 ps in duration, about 15 V, and at intervals of 5 s. Recordings were made in the cell body layers of area dentata, CAl, and CA3 with extracellular electrodes (20-50 pm tip diameter) filled with saline. In those studies where anoxia was instituted in the slice, responses were first recorded in normal oxygenated Krebs-Ringer, then the perfusion solution was changed to one equilibrated with 5% O,, 90% N, and 5% CO,. RESULTS
Figure 1 illustrates the histological changes induced by a transient occlusion of the four arteries to the brain 7 days prior to fixation. In the animals exposed to anoxia there is a massive loss of pyramidal neurons in CAl, while those in CA3 appear to be normal. Thus, this stimulus is sufficient to induce a selective and near total death of neurons in CAL Histological sections were obtained from more than 30 animals. While the degree of neuronal loss varied from animal to animal the pattern of the loss was constant, in that CA1 was always the most sensitive and area dentata always the least. Figure 2 illustrates recordings obtained from CA1 and CA3 upon activation of the Shaffer collateral and mossy fiber afferent pathways, respectively. The recordings were made in hippocampal slices obtained from control animals and from animals subjected to four-artery occlusion 24 h and 7 days prior to preparation of the slices. Although these recordings were obtained from three different slices, the results are typical of those obtained from 20 control animals, 14 slices from animals studied at 24 h and 10 at 7 days postischemia. In control animals the synaptic response in both CA1 and CA3 showed a single biphasic response followed by a prolonged tail. By 24 h postischemia, the response at both sites was very abnormal, showing a prolonged negative wave on which were superimposed multiple population repetitive discharges. The number of such waves was always larger in CA3. Their magnitude was usually greater in CA3 than
NEURONAL
DEATH
305
CA1 and varied with thickness (larger in thin slices). An increase in peak amplitude and oscillatory responses after a stimulus which in the control always gave a single peak was seen in about 70% of recordings from CA1 at 24 to 48 h (18 slices from 11 animals). There was, however, some variability from animal to animal in the exact time course of hyperexcitability. At 7 days the response in CA3 still showed hyperexcitability, but there was no response remaining in CAl. This loss of electrical activity correlates with the disappearance of neurons in the histological section in Fig. 1. There was a total loss of electrical activity in CA1 in slices prepared from 9 of 10 animals studied at 7 days postanoxia. The responses from CA3 remained hyperexcitahle in all animals studied over 7 days. In preliminary experiments on only a few animals, CA3 neurons ultimately became unexcitable also, but only after periods of about 2 weeks to 2 months. The reason for the latter times for DND in CA3 will be the subject of further studies. Results from another animal from which a slice was prepared 2 days after anoxia are shown in Fig. 3. In this preparation recordings were made from area dentata, CA3, and CAl. The population response in area dentata was essentially normal, while that in CA3 was very hyperexcitable, in the same fashion as shown in Fig. 2 at 24 h. The response in CA1 was very much reduced as compared with normal controls. The lower records show this CA1 response at 10 times higher gain and the effect on this response of amino phosphono-valeric acid (APV), an antagonist at NMDA receptors. APV (5 X lop5 M) reversibly reduced the amplitude of this small response to about 20% of its original peak. This observation suggests that, while the response to stimulation of the Shaffer collaterals is very much reduced after anoxia, the major part of what response remains reflects activation of NMDA receptors. Figure 4 shows extracellular and intracellular recordings from CA1 obtained from a slice from an animal subjected to four-vessel ischemia 28 h prior to slice preparation. The extracellular response shows hyperexcitability, with multiple oscillation on the late phase. With intracellular recording a prolonged depolarization is seen, a finding never seen in control recordings. When APV (5 x lo-’ M) was applied for 10 min, oscillations and the late intracellular depolarization were considerably reduced. This effect was reversible with washing. Similar results were obtained in four neurons. We instituted experiments in slices from normal animals to determine whether we could detect differential sensitivity of neurons in CA1 and CA3 to anoxia applied to the slice when studied over a time period realistic for maintaining a slice preparation. Results from such a study are shown in Fig. 5. In both regions effectively all electrical responses except a small afferent presynaptic volley were blocked at the end of the lo-min anoxic period using 5% 0,. Upon reinstituting the oxygenated Krebs-Ringer there was a recovery of electrical re-
306
HORI
ET
AL.
sections of hippocampi of control animals and animals subjected to four-vessel occlusion for a period of 10 min 7 days FIG. 1. Histological prior to removal from the animal and fixation. The brain was removed and fixed in 2.5% glutaraldehyde in Krebs-Ringer overnight and frozen sections were cut at 10 gm and stained with cresyl violet. Note the loss of neurons in CA1 in the hippocampus subjected to anoxia (indicated by arrows).
sponses within about 2-5 min. However, the responses were usually larger than the controls and often exhibited multiple population peaks in both CA1 and particularly in CA3 as illustrated at 2 h postanoxia. After 6.5 h
there was some recovery of the CA3 response nearer to the control, but the response in CA1 was essentially totally gone. Thus, the late effects of anoxia applied to the in vitro brain slice show a similar difference in regional
NMDA CONTROL
RECEPTORS
AND
DELAYED
7 DAYS
24 HOURS
NEURONAL CONTROL
307
DEATH APV (5 x 1O-5 M)
WASH
CA3
-Y
CA1 -
---I-
i r--
2 rn” p‘
--l 40 msec
FIG. 2. Population EPSP recordings from CA1 and CA3 from three different animals. The control responses are from slices prepared from normal animals. The response consists of a sharp negative wave, reflecting the population discharge of the postsynaptic neurons, followed by a positive wave which reflects the slower EPSP in the population. In all control slices there is little or no indication of additional population discharges on the falling phase of the slow wave. The other slices were prepared from animals subjected to hilatera1 four-vessel occlusion for 10 min applied 24 h and 7 days, respectively, before slice preparation. In all preparations the stimulus strength was 10 V and 50 ps duration. Note the multiple discharge spikes in both slices at 24 h and in CA3 at 7 days. No electrical postsynaptic response was obtained from CA1 at 7 days.
susceptibility for cell death, as indicated by loss of electrical excitability to that seen when anoxia is applied to the whole animal. Such studies were performed on 20 slices. While not all slices remained viable long enough to show such a dramatic difference as is illustrated in
c11
FIG. 3. Responses obtained by synaptic activation of the dentate gyrus (R.DG), CA3, and CA1 recorded from an animal subjected to bilateral four-vessel anoxia 2 days prior to slice preparation. The dentate response is effectively normal, while that in CA3 is hyperexcitable and that in CA1 is much smaller than normal. The lower records show the CA1 response at higher magnification and the effects of a lo-min application of 5 X 10v6 M APV. While the control responses of CA1 are not APV-sensitive, this reduced response was reversibly reduced by about 75%.
FIG. 4. Extracellular (A-C) and extracellular (D-E) recordings from CA1 from an animal exposed to four-vessel anoxia 28 h prior to slice preparation. During recordings B and E APV (5 X lo-’ M) was perfused for 10 min. Records C and F were taken after washing with normal Krebs-Ringer for 20 min.
Fig. 3, at 2 h postanoxia greater than 90% of the slices showed a decrease in CA1 as compared to control without a comparable decrease in CA3. In nine neurons the response in CA1 was reduced to 10% or less of control at a time when the response in CA3 was equal to or larger than control and showed multiple oscillations. Figure 6 shows the effect of APV and phencyclidine (PCP), which is known to block the ion channel associated with the NMDA receptor (l), on responses elicited from CA1 before, during, and after a lo-min anoxic episode. Neither NMDA response antagonist had any effect on the control response, indicating that there was little, if any, activation of NMDA receptors under control conditions. During anoxia there was a transient hyperexcitability in the control as well as in presence of APV or
CONTROL
ANOXIA
(10 min)
WASH (2 hours)
WASH
(6.5 hours]
FIG. 6. Effects of anoxia applied to normal slices on responses in CA1 and CA3. The CA1 and CA3 recordings are from different experiments. After control responses were obtained the slices were subjected to perfusion for 10 min of Krebs-Ringer equilibrated with 5% 0x-95% N,. Almost all indication of electrical activity was abolished at the end of 10 min in both regions. After perfusion again with normally oxygenated Krebs-Ringer a hyperexcitable response was elicited from both regions. When the slice was maintained for a prolonged period (6.5 h) the response obtained from CA3 remained hyperexcitable but returned toward normal, while CA1 deteriorated and became essentially unexcitable.
308
HORI
FIG. 6.
Effects of NMDA receptor antagonists on the sequence of synaptic events in CA1 during and after anoxia in the slice. Recordings are from three different slices; the numbers above the traces indicate the time in minutes from the start of perfusion with the anoxic solution. At 2 min all slices show hyperexcitability, as evidenced by the increase in the fast peak response. With further exposure to the anoxic soIution responses declined. Excitability was restored after reoxygenation. At the end of 1 h the response in the control had begun to decrease, while that in slices exposed to both APV and PCP during the anoxic episode remained hyperexcitable. These particular slices did not show bursting discharges, but the hyperexcitability is indicated by the increase in peak amplitude to the synaptic response.
PCP at 2 min. By 4 min the responses in all three slices had fallen to much lower values and decreased even further over the lo-min exposure to anoxia. It is noteworthy that the response was essentially abolished in both APV and PCP, but not in the control. This observation resembles the results of Fig. 2 even though taken at a different time point in the response to anoxia. Nevertheless, these results indicate that the residual response of these neurons late in anoxia results from activation of NMDA receptors, even though these receptors do not appear to contribute significantly to the response in the control. After washing for 1 h there was a return of responsiveness in all three preparations. However, the response in the slice not treated with a NMDA receptor or channel antagonist was already reduced as compared to control. These preparations always continued to deteriorate and became electrically unexcitable (n = 20). In contrast, the slices perfused with AVP (n = 5) and PCP (n = 5) during the anoxia showed a transient hyperexcitability, after which the response returned slowly to the control values. In none of the slices studied with the NMDA antagonists did loss of electrical excitability occur within the 2 to 7 h in which they were studied. The appearance of a NMDA component to the response to Schaffer collateral stimulation late in the lomin anoxic exposure suggests the possibility that the CA1 neurons are extremely depolarized, and consequently the Mg2+ blockade of NMDA receptors, which normally occurs at resting potential, is relieved. In order to test whether this was the case, intracellular record-
ET
AL.
FIG. 7. Intracellular recordings from a pyramidal neuron during anoxia applied in the slice. Record C is the control and shows the voltage change caused by a constant current pulse of 0.2 nA and the response elicited by activation of the Shaffer collaterals in CA3. The next records were obtained at 4, 8, and 12 min, respectively,after beginning perfusion with Krebs-Ringer equilibrated with 5% O,-95% N,. W20 is a trace taken 20 min after reperfusion with oxygenated solution.
ings were made of 10 CA1 neurons during the anoxic episode, and one example is shown in Fig. 7. Also shown are the responses to a hyperpolarizing current pulse of 0.2 nA and activation of the Schaffer collaterals. In this neuron the low oxygen Krebs-Ringer was perfused for 12 min before the cell became unexcitable. Several points are clear. There was no significant depolarization of the membrane potential, which actually hyperpolarized by about 3 mV. Resting resistance of the membrane decreased somewhat and recovered even more than the control after reoxygenation. However, after reoxygenation the response to synaptic stimulation became hyperexcitable. In none of the 10 neurons studied was there significant depolarization observed during the anoxic episode. Thus, the appearance of NMDA responses after anoxia cannot be due to depolarization and removal of Mg2+ blockade of NMDA receptor channels. The results in Fig. 8 indicate that the hyperexcitability seen in CA1 neurons postanoxia is to a great degree the result of activation of NMDA receptors. The records illustrated are intracellular recordings from a CA1 neu-
FIG. 8. Intracellular recording made 45 min after anoxia from the same pyramidal neuron in CA1 as illustrated in Fig. 7. With application of lo-’ M APV for 10 min the late component of the burst was reversibly abolished.
NMDA
RECEPTORS
AND
DELAYED
ron in a slice that had been exposed to anoxia 45 min earlier. The response to stimulation of the Shaffer collaterals in this cell is a prolonged depolarization, reflecting the hyperexcitability following anoxia. In order to determine whether or not this hyperexcitability reflected NMDA receptor activation, the slice was perfused with 10e4 M APV for 10 min. As seen, APV abolished most of the late depolarization reversibly. Similar results were obtained from 12 of 15 CA1 neurons examined. As in CA1 neurons, the hyperexcitability in CA3 neurons was also found to be sensitive to APV. Since hyperexcitability postanoxia was not unique to CA& the presence of NMDA-mediated hyperexcitability does not, by itself, explain why CA1 neurons are uniquely sensitive to DND even though these results strongly suggest that activation of the NMDA receptors on CA1 neurons is required for the appearance of DND. DISCUSSION
Comparison of the In Viuo and In Vitro Model Systems The major difference between in uiuo and in vitro conditions in the model of anoxic neuronal death is the time course. In one preparation electrical excitability is lost in a matter of hours, while in the other it occurs over days. The common features probably include a loss of excitability during the initial anoxic injury (although this has not been directly demonstrated in the in uiuo preparation) and a prolonged period of hyperexcitability prior to the loss of excitability in CAl. It appears likely, on the basis of our experience with these two models, that in both preparations the time course of loss of excitability by DND can be varied by changing the duration of the anoxic episode as well as other parameters such as glucose and oxygen concentrations. In the slice experiments we have carefully determined an oxygen and glucose concentration that will cause the immediate loss of excitability to occur slowly and be complete only at the end of 10 min. In these circumstances, there is always a recovery of excitability with reoxygenation and there is relatively little variation in the occurrence and time course of late sequelae. The most convincing features indicating that the slice model of DND resembles the DND seen in the in uiuo model are the greater sensitivity of CA1 neurons and the sensitivity of the hyperexcitable CA1 responses to the NMDA receptor antagonist, APV. While these results do not prove that the slice offers an exact model system, and indeed it is probably impossible to prove such a premise, our results suggest sufficient similarities to justify use of this preparation, which is technically very much easier. Use of an in vitro preparation for exposure to anoxia has advantages in allowing control over a variety of factors difficult to regulate in the intact
NEURONAL
DEATH
309
animal, such as temperature, pH, ionic composition of solutions, and drug application. Thus, this preparation has considerable potential for future studies. Our results showing the relative lack of sensitivity of area dentata to anoxic damage as compared to CA3 and especially CA1 are compatible with previous observations (4). The reason for the differential sensitivity is, at present, unknown. Various possible factors may include the density of NMDA receptor sites, cellular differences in ability to regulate and buffer rises in intracellular calcium concentration, cellular differences in intracellular energy stores, or levels of cellular activity of calcium-activated proteases. Mechanisms Responsible for Postanoxic Hyperercitability While NMDA receptor activation does not appear to contribute significantly to the response of CA1 neurons to Schaffer collateral stimulation, both the in uiuo and in vitro anoxic exposures resulted in late abnormal responses in CAl, for which a considerable portion of the response was sensitive to NMDA antagonists. This feature, shown in Fig. 3 in the intracellular response of a neuron exposed to anoxia in uiuo and also in the slice 45 min earlier in Figs. 6 and 7, is characteristic of the reduced CA1 response from animal exposure to anoxia 2 days earlier. The reason for the appearance of NMDA responses is not obvious. Unlike the kainate and quisqualate excitatory amino acid receptorlionophores, which are assumed to be responsible for the APV-insensitive portion and almost all of the control responses, the response to the NMDA receptor is voltage-dependent as a result of a voltage-dependent blockade of the associated channel by Mg2+ (16,19). When the neuron is excessively depolarized, the blockade of the channel by Mg2+ is removed and the channel becomes conductive. While Na+ and K+ are known to carry most of the current for all three types of excitatory amino-acid-activated channels, the NMDA channel, unlike the kainate and quisqualate channels, is permeable to Ca2+ (14). It is likely that the differences in functional effect of NMDA channel activation, as compared to the other excitatory amino acid receptors, is a result of the entry of calcium into cells. In some circumstances, such as in long-term potentiation, the entry of calcium through NMDA channels is assumed to have positive consequences (13,E). In other circumstances, such as with anoxic neuronal damage, it is likely that the entry of calcium exceeds the buffering capacity of the cell (9), leading to a series of destructive actions, at least in part mediated by activation of calcium-activated proteases that proceed to degrade intracellular proteins (17). The appearance of a large NMDA response might be secondary to sufficient depolarization of the CA1 neu-
310
HORI
rons so as to relieve Mg 2+ block of the channel. This does not, however, appear to be the explanation for the observed results. The neurons recorded in Figs. 6 and 7 show perfectly normal resting potentials and yet, unlike the controls, the response to Schaffer collateral stimulation has a large, APV-sensitive component. The change in response might reflect an increase in the number of NMDA receptors, a change in their voltage dependency, or a change in some other modulatory feature of the receptor/channel complex. The reason for the appearance of the NMDA response postanoxia is one of the most important barriers to our understanding of the mechanism of the events leading to death of neurons. There is a very similar problem in explaining the acute events occurring during exposure to anoxia in the slice preparations. In the presence of APV and PCP all indications of electrical excitability are lost after an initial period of hyperexcitability. There is considerable debate about the reason for loss of the responses to the quisqualatefkainate receptors, but it is likely that much of the block is due to depression of presynaptic release of transmitter (11). The fact that there is much less response after about 4 min of anoxia in the presence of NMDA antagonists than in the controls indicates that NMDA receptors become active. This is despite lack of a detectable effect of APV or PCP on the field potential in the control preparation. The appearance of NMDA responses might also result from excessive depolarization of the neuron, but the results of Fig. 6 show no such depolarization. Leblond and Krnjevic (11) have shown that CA1 neurons are hyperpolarized by brief periods of anoxia in slices. They (10) also demonstrated that a number of ionic currents, including calcium and potassium currents, were reduced and suggested that this reduction might be secondary to ATP depletion. Thus, the appearance of NMDA responses in these circumstances is surprising and not explicable on the basis of present knowledge of the voltage dependence of the NMDA response (7). There is an accumulating body of evidence that NMDA antagonists protect against anoxic neuronal cell death in whole animals (5,21,25), brain slices (3,8), and neuronal cell cultures (2). Anoxia has also been shown to trigger a proteolysis of hippocampal spectrin, which can be prevented by use of NMDA receptor blockers (23). Since spectrin is a substrate of the calcium-activated protease, calpain (24), the proteolysis is presumed to result from a rise in internal calcium concentration subsequent to activation of the NMDA receptor. Our results are compatible with a major role for the NMDA receptor in the cellular response to anoxic injury. However, they strongly suggest that the increase in NMDA receptor activity is a consequence of a process as yet not understood.
ET
AL.
ACKNOWLEDGMENTS These studies were supported thank Mrs. Charlene McAuliffe
by NIH Grant ROl NS23807. for secretarial assistance.
We
REFERENCES 1. ANIS, N. A., S. C. BERRY, N. R. BURTON,
2.
AND D. LODGE. 1983. The dissociative anesthetics, ketamine andphencyclidine, selectively reduce excitation of central mammalian neurons by Nmethyl-aspartate. Br. J. Phnrmacol. 79: 565-575. CHOI, D. W. 1988. Calcium-mediated neurotoxicity: Relationship to specific channel types and role in ischemic damage. Trends Neurosci. 11: 465-469.
3.
CLARK, G. D., AND S. M. ROTHMAN. 1987. Blockade amino acid receptors protects anoxic hippocampal science 21: 665-671.
4.
DIENEL, G. A., W. A. PULSINELLI, AND T. E. DUFFY. 1980. Regional protein synthesis in rat brain following acute hemispheric ischemia. J. Neurochem. 35: 1216-1226.
5.
GILL, R., A. C. FOSTER, AND G. N. WOODRUFF. neuroprotective in gerbils when administered ischemic period. Neuroscience 25: 847-855.
6.
HAJOS, F., C. GARTHWAITE, AND J. GARTHWAITE. 1986. Reversible and irreversible neuronal damage caused by excitatory amino acid analogues in rat cerebellar slices. Neuroscience 18: 417-436.
7.
HORI, N., N. DOI, S. MNAHARA, Y. SHINODA, AND D. 0. CARPENTER. 1989. NMDA responses on hippocampal neurons change after exposure to anoxia and hypoglycemia. Neurosci. Abstr. 15: 769.
8.
KASS, I. S., G. CHAMBERS, AND J. E. COTTRELL. 1989. The Nmethyl-D-aspartate antagonists aminophosphonovaleric acid and MK-801 reduce anoxic damage to dentate granule and CA1 pyramidal cells in the rat hippocampal slice. Erp. Neural. 103: 116-122. KASS, I. S., AND P. LI~ON. 1986. Calcium and long-term transmission damage following anoxia in dentate gyrus and CA1 regions of the hippocampal slice. J. Physiol. London 378: 313-334. KRNJEVIC, K., AND J. LEBLOND. 1989. Changes in membrane currents of hippocampal neurons evoked by brief anoxia. J. Neurophysiol. 62: 15-30. LEBLOND, J., AND K. KRNJEVIC. 1989. Hypoxic changes in hippocampal neurons. J. Neurophysial. 62: 1-14.
9.
10.
11.
of excitatory slices. Neuro-
1988. MK-801 is during the post-
12.
LIPTON, P., AND T. S. WHITTINGHAM. 1982. Reduced ATP concentration as a basis for synaptic transmission failure during hypoxia in the in vitro guinea pig hippocampus. J. Phystil. London 325: 51-65.
13.
LYNCH, G., J. LARSON, S. KELSO, G. BARRIONUEVO, AND F. SCHO?TLER. 1983. Intracellular injections of EGTA block induction of hippocampal long-term potentiation. Nature (London) 305: 719-721.
14.
MACDERMOTT, A. B., M. L. MAYER, G. L. WESTEIROOK, S. J. SMITH, AND J. L. BARKER. 1986. NMDA-receptor activation increases cytoplasmic calcium concentration in cultured spinal cord neurones. Nature fLondon) 321: 519-522. MALENKA, R. C., J. A. KAUER, R. S. ZUCKER, AND R. A. NICOLL. 1988. Postsynaptic calcium is sufficient for potentiation of hippocampal synaptic transmission. Science 242: 81-84.
15.
16.
MA~R,
M.
L., G. L. WESTBROOK,
AND P. B. GUTHRIE.
1984.
NMDA
17.
18.
19.
20.
RECEPTORS
AND DELAYED
Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones. Nature (London) 309: 261-263. MELLGREN, R. L. 1987. Calcium-dependent proteases: An enzyme system active at cellular membranes? FASEB J. 1: llO115. NOVELLI, A., J. A. REILLY, P. G. LYSKO, AND R. C. HENNEBERRY. 1988. Glutamate becomes neurotoxic via the NMDA receptor when intracellular energy levels are reduced. Brain Res. 451: 205-212. NOWAK, L., P. BREGESTOVSKI, P. ASCHER, A. HERBERT, AND A. PROCHIANTZ. 1984. Magnesium gates glutamate-activated channels in mouse central neurons. Nature (London) 307: 462-465. PULSINELLI, W. A., AND J. B. BRIE-Y. 1979. A new model of bilateral hemispheric ischemia in the unanesthetized rat. Stroke 10:267-272.
NEURONAL
DEATH
311
21. ROMAN, R., H. BARTKOWSKI, AND R. SIMON. 1989. The specific NMDA receptor antagonist AP-7 attenuates focal ischemic brain injury. Neurosci. Lett. 104: 19-24. 22. ROTHMAN, S. M., AND J. W. OLNEY. 1986. Glutamate and the pathophysiology of hypoxic-ischemic brain damage. Ann. Neural. 19:105-111. 23. SEUBERT, P., K. LEE, AND G. LYNCH. 1989. Ischemia triggers NMDA receptor-linked cytoskeletal proteolysis in hippocampus. Brain Res. 492: 366-370. 24. SIMAN, R., M. BAUDRY, AND G. LYNCH. 1984. Brain fodrin: Substrate for the endogenous calcium-activated protease, calpain I. Proc. Natl. Acad. Sci. USA 81: 3572-3576. 25. STEINBERG, G. K., J. SALEH, R. DELAPAZ, D. K~Is, AND S. R. ZARNEGAR. 1989. Pretreatment with the NMDA antagonist dextrorphan reduces cerebral injury following transient focal ischemia in rabbits. Brain Res. 497: 382-386.
NEUROLOGY
112,
304-311
(19%)
Appearance of NMDA Receptors Triggered by Anoxia Independent of Voltage in Viva and in Vitro N. HoRI,**~
N. DOI,+ S. MNAHARA,~
Y. SHINODA,?
AND D. 0. CARPENTER*
*Wadsworth Center for Laboratories and Research, New York State Department of Health and School of Public Health, Albany, New York 12201-0509; iDepartment of Pharmacology, Kyushu University Faculty of Dentistry, Fukuoka 812, Japan; and *Department of Biology, Kyushu University Faculty of Science, Fukuoka 812, Japan
and CA3 show characteristic changes in response to synaptic activation, and finally neurons lose electrical activity (delayed neuronal death, DND). There are consistent differences in susceptibility to DND in these three hippocampal areas, with CA1 being the most sensitive and area dentata the least (9). The mechanisms that are responsible for DND are incompletely understood, and a variety of factors such as excessive activity of excitatory amino acids (3,6,22), accumulation of intracellular calcium (2), and depletion of intracellular energy stores (12,18) have been considered. The observation that different populations of morphologically quite similar neurons in the same brain region have different vulnerabilities to DND presents the possibility of a preparation in which one may study factors that contribute to DND. Such study is of importance not only for the understanding of the mechanisms of (and potentially ways of preventing) direct anoxic cell death in humans, but also because of the possibility that similar mechanisms may underlie neuronal cell death in a variety of human diseases characterized by selective neuronal loss, such as Alzheimer’s, amyotrophic lateral sclerosis, and Parkinson’s Disease. We have compared in viva and in vitro models of anoxic damage and provide evidence in support of the postulate that one can study DND in an in vitro brain slice preparation. Such a preparation may have considerable utility for investigating the mechanisms of DND. In this report we document the models and present preliminary evidence for the involvement of N-methyl-D-aspartate (NMDA) receptor activation in DND in both systems.
Using rat hippocampus we have studied the pattern of neuronal death, abnormal discharge and loss of electrical excitability in slices prepared from animals subjected to bilateral, four-vessel cerebral anoxia and in slices prepared from normal animals that are subjected to anoxia in the recording chamber. As others have reported, pyramidal neurons in area CA1 are lost first after anoxia, while CA3 neurons have an intermediate sensitivity, and those in dentate are relatively anoxiaresistant. After anoxic damage to the intact animal, neurons in both CA1 and CA3 show abnormal bursting discharges in response to synaptic activation for several days, and then the response in CA1 decreases in amplitude and finally the area become unexcitable. While antagonists for N-methyl-D-aspartate (NMDA) receptors have essentially no effect on synaptic responses in control animals, they reduce the bursting responses and greatly depress the small responses in CA1 as neurons are becoming unexcitable after anoxia. With intracellular recording CA1 neurons from animals made transiently anoxic, in contrast to controls, show prolonged synaptic responses, the later components of which are blocked by NMDA antagonists. When slices from normal animals are subjected to anoxia such that excitability is totally lost over a period of about 10 min, there is no significant membrane depolarization during the anoxic episode and recovery of excitability occurs with reoxygenation. However, a period of hyperexcitability and bursting follows and electrical excitability is lost in CA1 but not CA3 neurons after about SO min. Both the bursting and loss of CA1 excitability are prevented by application of amino phosphono-valeric acid (APV) or phencyclidine (PCP) during the anoxic episode. With intracellular recording CA1 neurons from slices made transiently anoxic but not from control slices show prolonged, APV-sensitive synaptic potentials. These observations suggest that in both in vivo and in vitro preparations a critical effect of transient anoxia is the induction of previously absent NMDA receptors and that the trigger for this induction is not voltage dependent. 0 1991
Academic
Press,
METHODS Rats (200-250 g) were used in all experiments. In one series of experiments 26 animals were subjected to transient forebrain ischemia (occlusion of both carotids and both vertebral arteries for 10 min) 1 to 7 days prior to preparation of brain slices for electrophysiological study using the method of Pulsinelli and Brierley (20). Other studies were done by preparing slices from 38 normal animals, then subjecting the slices to anoxia in the experimental chamber. In both studies animals were killed by cervical dislocation under light ether anesthe-
Inc.
INTRODUCTION Two or three days after transient forebrain ischemia, hippocampal pyramidal neurons in area dentata, CAl, 0014-4886/91 Copyright All rights
$3.00 0 1991 by Academic Press, of reproduction in any form
304 Inc. reserved.
NMDA
RECEPTORS
AND
DELAYED
sia and the brains were quickly removed and placed in iced Krebs-Ringer. Slices (450 pm) were cut perpendicular to the long axis of the hippocampus using a simple vibratome (Nomiyama, Japan) and were preincubated in oxygenated Krebs-Ringer (NaCl, 126 n&f; KCl, 5 mM; KH,PO,, 1.26 mA4; MgSO,, 1.3 n&f; CaCl,, 2.4 mM; NaHCO,, 26 mM; and glucose, 10 mM) saturated with 95% 0, and 5% CO, for about 2 h at 35°C prior to mounting in the recording chamber. The slices were submerged on a plexi mesh in the chamber and perfused with Krebs-Ringer at a rate of 3-5 ml/min at 35°C. A monopolar electrode was used for stimulation of synaptic inputs to area dentata, CAl, and CA3. For activation of area dentata the electrode was placed on the perforant path, for CA1 the electrode was placed on the Shaffer collaterals in CA3, while for CA3 activation the electrode was placed on the mossy fibers. Stimulation pulses were 50 ps in duration, about 15 V, and at intervals of 5 s. Recordings were made in the cell body layers of area dentata, CAl, and CA3 with extracellular electrodes (20-50 pm tip diameter) filled with saline. In those studies where anoxia was instituted in the slice, responses were first recorded in normal oxygenated Krebs-Ringer, then the perfusion solution was changed to one equilibrated with 5% O,, 90% N, and 5% CO,. RESULTS
Figure 1 illustrates the histological changes induced by a transient occlusion of the four arteries to the brain 7 days prior to fixation. In the animals exposed to anoxia there is a massive loss of pyramidal neurons in CAl, while those in CA3 appear to be normal. Thus, this stimulus is sufficient to induce a selective and near total death of neurons in CAL Histological sections were obtained from more than 30 animals. While the degree of neuronal loss varied from animal to animal the pattern of the loss was constant, in that CA1 was always the most sensitive and area dentata always the least. Figure 2 illustrates recordings obtained from CA1 and CA3 upon activation of the Shaffer collateral and mossy fiber afferent pathways, respectively. The recordings were made in hippocampal slices obtained from control animals and from animals subjected to four-artery occlusion 24 h and 7 days prior to preparation of the slices. Although these recordings were obtained from three different slices, the results are typical of those obtained from 20 control animals, 14 slices from animals studied at 24 h and 10 at 7 days postischemia. In control animals the synaptic response in both CA1 and CA3 showed a single biphasic response followed by a prolonged tail. By 24 h postischemia, the response at both sites was very abnormal, showing a prolonged negative wave on which were superimposed multiple population repetitive discharges. The number of such waves was always larger in CA3. Their magnitude was usually greater in CA3 than
NEURONAL
DEATH
305
CA1 and varied with thickness (larger in thin slices). An increase in peak amplitude and oscillatory responses after a stimulus which in the control always gave a single peak was seen in about 70% of recordings from CA1 at 24 to 48 h (18 slices from 11 animals). There was, however, some variability from animal to animal in the exact time course of hyperexcitability. At 7 days the response in CA3 still showed hyperexcitability, but there was no response remaining in CAl. This loss of electrical activity correlates with the disappearance of neurons in the histological section in Fig. 1. There was a total loss of electrical activity in CA1 in slices prepared from 9 of 10 animals studied at 7 days postanoxia. The responses from CA3 remained hyperexcitahle in all animals studied over 7 days. In preliminary experiments on only a few animals, CA3 neurons ultimately became unexcitable also, but only after periods of about 2 weeks to 2 months. The reason for the latter times for DND in CA3 will be the subject of further studies. Results from another animal from which a slice was prepared 2 days after anoxia are shown in Fig. 3. In this preparation recordings were made from area dentata, CA3, and CAl. The population response in area dentata was essentially normal, while that in CA3 was very hyperexcitable, in the same fashion as shown in Fig. 2 at 24 h. The response in CA1 was very much reduced as compared with normal controls. The lower records show this CA1 response at 10 times higher gain and the effect on this response of amino phosphono-valeric acid (APV), an antagonist at NMDA receptors. APV (5 X lop5 M) reversibly reduced the amplitude of this small response to about 20% of its original peak. This observation suggests that, while the response to stimulation of the Shaffer collaterals is very much reduced after anoxia, the major part of what response remains reflects activation of NMDA receptors. Figure 4 shows extracellular and intracellular recordings from CA1 obtained from a slice from an animal subjected to four-vessel ischemia 28 h prior to slice preparation. The extracellular response shows hyperexcitability, with multiple oscillation on the late phase. With intracellular recording a prolonged depolarization is seen, a finding never seen in control recordings. When APV (5 x lo-’ M) was applied for 10 min, oscillations and the late intracellular depolarization were considerably reduced. This effect was reversible with washing. Similar results were obtained in four neurons. We instituted experiments in slices from normal animals to determine whether we could detect differential sensitivity of neurons in CA1 and CA3 to anoxia applied to the slice when studied over a time period realistic for maintaining a slice preparation. Results from such a study are shown in Fig. 5. In both regions effectively all electrical responses except a small afferent presynaptic volley were blocked at the end of the lo-min anoxic period using 5% 0,. Upon reinstituting the oxygenated Krebs-Ringer there was a recovery of electrical re-
306
HORI
ET
AL.
sections of hippocampi of control animals and animals subjected to four-vessel occlusion for a period of 10 min 7 days FIG. 1. Histological prior to removal from the animal and fixation. The brain was removed and fixed in 2.5% glutaraldehyde in Krebs-Ringer overnight and frozen sections were cut at 10 gm and stained with cresyl violet. Note the loss of neurons in CA1 in the hippocampus subjected to anoxia (indicated by arrows).
sponses within about 2-5 min. However, the responses were usually larger than the controls and often exhibited multiple population peaks in both CA1 and particularly in CA3 as illustrated at 2 h postanoxia. After 6.5 h
there was some recovery of the CA3 response nearer to the control, but the response in CA1 was essentially totally gone. Thus, the late effects of anoxia applied to the in vitro brain slice show a similar difference in regional
NMDA CONTROL
RECEPTORS
AND
DELAYED
7 DAYS
24 HOURS
NEURONAL CONTROL
307
DEATH APV (5 x 1O-5 M)
WASH
CA3
-Y
CA1 -
---I-
i r--
2 rn” p‘
--l 40 msec
FIG. 2. Population EPSP recordings from CA1 and CA3 from three different animals. The control responses are from slices prepared from normal animals. The response consists of a sharp negative wave, reflecting the population discharge of the postsynaptic neurons, followed by a positive wave which reflects the slower EPSP in the population. In all control slices there is little or no indication of additional population discharges on the falling phase of the slow wave. The other slices were prepared from animals subjected to hilatera1 four-vessel occlusion for 10 min applied 24 h and 7 days, respectively, before slice preparation. In all preparations the stimulus strength was 10 V and 50 ps duration. Note the multiple discharge spikes in both slices at 24 h and in CA3 at 7 days. No electrical postsynaptic response was obtained from CA1 at 7 days.
susceptibility for cell death, as indicated by loss of electrical excitability to that seen when anoxia is applied to the whole animal. Such studies were performed on 20 slices. While not all slices remained viable long enough to show such a dramatic difference as is illustrated in
c11
FIG. 3. Responses obtained by synaptic activation of the dentate gyrus (R.DG), CA3, and CA1 recorded from an animal subjected to bilateral four-vessel anoxia 2 days prior to slice preparation. The dentate response is effectively normal, while that in CA3 is hyperexcitable and that in CA1 is much smaller than normal. The lower records show the CA1 response at higher magnification and the effects of a lo-min application of 5 X 10v6 M APV. While the control responses of CA1 are not APV-sensitive, this reduced response was reversibly reduced by about 75%.
FIG. 4. Extracellular (A-C) and extracellular (D-E) recordings from CA1 from an animal exposed to four-vessel anoxia 28 h prior to slice preparation. During recordings B and E APV (5 X lo-’ M) was perfused for 10 min. Records C and F were taken after washing with normal Krebs-Ringer for 20 min.
Fig. 3, at 2 h postanoxia greater than 90% of the slices showed a decrease in CA1 as compared to control without a comparable decrease in CA3. In nine neurons the response in CA1 was reduced to 10% or less of control at a time when the response in CA3 was equal to or larger than control and showed multiple oscillations. Figure 6 shows the effect of APV and phencyclidine (PCP), which is known to block the ion channel associated with the NMDA receptor (l), on responses elicited from CA1 before, during, and after a lo-min anoxic episode. Neither NMDA response antagonist had any effect on the control response, indicating that there was little, if any, activation of NMDA receptors under control conditions. During anoxia there was a transient hyperexcitability in the control as well as in presence of APV or
CONTROL
ANOXIA
(10 min)
WASH (2 hours)
WASH
(6.5 hours]
FIG. 6. Effects of anoxia applied to normal slices on responses in CA1 and CA3. The CA1 and CA3 recordings are from different experiments. After control responses were obtained the slices were subjected to perfusion for 10 min of Krebs-Ringer equilibrated with 5% 0x-95% N,. Almost all indication of electrical activity was abolished at the end of 10 min in both regions. After perfusion again with normally oxygenated Krebs-Ringer a hyperexcitable response was elicited from both regions. When the slice was maintained for a prolonged period (6.5 h) the response obtained from CA3 remained hyperexcitable but returned toward normal, while CA1 deteriorated and became essentially unexcitable.
308
HORI
FIG. 6.
Effects of NMDA receptor antagonists on the sequence of synaptic events in CA1 during and after anoxia in the slice. Recordings are from three different slices; the numbers above the traces indicate the time in minutes from the start of perfusion with the anoxic solution. At 2 min all slices show hyperexcitability, as evidenced by the increase in the fast peak response. With further exposure to the anoxic soIution responses declined. Excitability was restored after reoxygenation. At the end of 1 h the response in the control had begun to decrease, while that in slices exposed to both APV and PCP during the anoxic episode remained hyperexcitable. These particular slices did not show bursting discharges, but the hyperexcitability is indicated by the increase in peak amplitude to the synaptic response.
PCP at 2 min. By 4 min the responses in all three slices had fallen to much lower values and decreased even further over the lo-min exposure to anoxia. It is noteworthy that the response was essentially abolished in both APV and PCP, but not in the control. This observation resembles the results of Fig. 2 even though taken at a different time point in the response to anoxia. Nevertheless, these results indicate that the residual response of these neurons late in anoxia results from activation of NMDA receptors, even though these receptors do not appear to contribute significantly to the response in the control. After washing for 1 h there was a return of responsiveness in all three preparations. However, the response in the slice not treated with a NMDA receptor or channel antagonist was already reduced as compared to control. These preparations always continued to deteriorate and became electrically unexcitable (n = 20). In contrast, the slices perfused with AVP (n = 5) and PCP (n = 5) during the anoxia showed a transient hyperexcitability, after which the response returned slowly to the control values. In none of the slices studied with the NMDA antagonists did loss of electrical excitability occur within the 2 to 7 h in which they were studied. The appearance of a NMDA component to the response to Schaffer collateral stimulation late in the lomin anoxic exposure suggests the possibility that the CA1 neurons are extremely depolarized, and consequently the Mg2+ blockade of NMDA receptors, which normally occurs at resting potential, is relieved. In order to test whether this was the case, intracellular record-
ET
AL.
FIG. 7. Intracellular recordings from a pyramidal neuron during anoxia applied in the slice. Record C is the control and shows the voltage change caused by a constant current pulse of 0.2 nA and the response elicited by activation of the Shaffer collaterals in CA3. The next records were obtained at 4, 8, and 12 min, respectively,after beginning perfusion with Krebs-Ringer equilibrated with 5% O,-95% N,. W20 is a trace taken 20 min after reperfusion with oxygenated solution.
ings were made of 10 CA1 neurons during the anoxic episode, and one example is shown in Fig. 7. Also shown are the responses to a hyperpolarizing current pulse of 0.2 nA and activation of the Schaffer collaterals. In this neuron the low oxygen Krebs-Ringer was perfused for 12 min before the cell became unexcitable. Several points are clear. There was no significant depolarization of the membrane potential, which actually hyperpolarized by about 3 mV. Resting resistance of the membrane decreased somewhat and recovered even more than the control after reoxygenation. However, after reoxygenation the response to synaptic stimulation became hyperexcitable. In none of the 10 neurons studied was there significant depolarization observed during the anoxic episode. Thus, the appearance of NMDA responses after anoxia cannot be due to depolarization and removal of Mg2+ blockade of NMDA receptor channels. The results in Fig. 8 indicate that the hyperexcitability seen in CA1 neurons postanoxia is to a great degree the result of activation of NMDA receptors. The records illustrated are intracellular recordings from a CA1 neu-
FIG. 8. Intracellular recording made 45 min after anoxia from the same pyramidal neuron in CA1 as illustrated in Fig. 7. With application of lo-’ M APV for 10 min the late component of the burst was reversibly abolished.
NMDA
RECEPTORS
AND
DELAYED
ron in a slice that had been exposed to anoxia 45 min earlier. The response to stimulation of the Shaffer collaterals in this cell is a prolonged depolarization, reflecting the hyperexcitability following anoxia. In order to determine whether or not this hyperexcitability reflected NMDA receptor activation, the slice was perfused with 10e4 M APV for 10 min. As seen, APV abolished most of the late depolarization reversibly. Similar results were obtained from 12 of 15 CA1 neurons examined. As in CA1 neurons, the hyperexcitability in CA3 neurons was also found to be sensitive to APV. Since hyperexcitability postanoxia was not unique to CA& the presence of NMDA-mediated hyperexcitability does not, by itself, explain why CA1 neurons are uniquely sensitive to DND even though these results strongly suggest that activation of the NMDA receptors on CA1 neurons is required for the appearance of DND. DISCUSSION
Comparison of the In Viuo and In Vitro Model Systems The major difference between in uiuo and in vitro conditions in the model of anoxic neuronal death is the time course. In one preparation electrical excitability is lost in a matter of hours, while in the other it occurs over days. The common features probably include a loss of excitability during the initial anoxic injury (although this has not been directly demonstrated in the in uiuo preparation) and a prolonged period of hyperexcitability prior to the loss of excitability in CAl. It appears likely, on the basis of our experience with these two models, that in both preparations the time course of loss of excitability by DND can be varied by changing the duration of the anoxic episode as well as other parameters such as glucose and oxygen concentrations. In the slice experiments we have carefully determined an oxygen and glucose concentration that will cause the immediate loss of excitability to occur slowly and be complete only at the end of 10 min. In these circumstances, there is always a recovery of excitability with reoxygenation and there is relatively little variation in the occurrence and time course of late sequelae. The most convincing features indicating that the slice model of DND resembles the DND seen in the in uiuo model are the greater sensitivity of CA1 neurons and the sensitivity of the hyperexcitable CA1 responses to the NMDA receptor antagonist, APV. While these results do not prove that the slice offers an exact model system, and indeed it is probably impossible to prove such a premise, our results suggest sufficient similarities to justify use of this preparation, which is technically very much easier. Use of an in vitro preparation for exposure to anoxia has advantages in allowing control over a variety of factors difficult to regulate in the intact
NEURONAL
DEATH
309
animal, such as temperature, pH, ionic composition of solutions, and drug application. Thus, this preparation has considerable potential for future studies. Our results showing the relative lack of sensitivity of area dentata to anoxic damage as compared to CA3 and especially CA1 are compatible with previous observations (4). The reason for the differential sensitivity is, at present, unknown. Various possible factors may include the density of NMDA receptor sites, cellular differences in ability to regulate and buffer rises in intracellular calcium concentration, cellular differences in intracellular energy stores, or levels of cellular activity of calcium-activated proteases. Mechanisms Responsible for Postanoxic Hyperercitability While NMDA receptor activation does not appear to contribute significantly to the response of CA1 neurons to Schaffer collateral stimulation, both the in uiuo and in vitro anoxic exposures resulted in late abnormal responses in CAl, for which a considerable portion of the response was sensitive to NMDA antagonists. This feature, shown in Fig. 3 in the intracellular response of a neuron exposed to anoxia in uiuo and also in the slice 45 min earlier in Figs. 6 and 7, is characteristic of the reduced CA1 response from animal exposure to anoxia 2 days earlier. The reason for the appearance of NMDA responses is not obvious. Unlike the kainate and quisqualate excitatory amino acid receptorlionophores, which are assumed to be responsible for the APV-insensitive portion and almost all of the control responses, the response to the NMDA receptor is voltage-dependent as a result of a voltage-dependent blockade of the associated channel by Mg2+ (16,19). When the neuron is excessively depolarized, the blockade of the channel by Mg2+ is removed and the channel becomes conductive. While Na+ and K+ are known to carry most of the current for all three types of excitatory amino-acid-activated channels, the NMDA channel, unlike the kainate and quisqualate channels, is permeable to Ca2+ (14). It is likely that the differences in functional effect of NMDA channel activation, as compared to the other excitatory amino acid receptors, is a result of the entry of calcium into cells. In some circumstances, such as in long-term potentiation, the entry of calcium through NMDA channels is assumed to have positive consequences (13,E). In other circumstances, such as with anoxic neuronal damage, it is likely that the entry of calcium exceeds the buffering capacity of the cell (9), leading to a series of destructive actions, at least in part mediated by activation of calcium-activated proteases that proceed to degrade intracellular proteins (17). The appearance of a large NMDA response might be secondary to sufficient depolarization of the CA1 neu-
310
HORI
rons so as to relieve Mg 2+ block of the channel. This does not, however, appear to be the explanation for the observed results. The neurons recorded in Figs. 6 and 7 show perfectly normal resting potentials and yet, unlike the controls, the response to Schaffer collateral stimulation has a large, APV-sensitive component. The change in response might reflect an increase in the number of NMDA receptors, a change in their voltage dependency, or a change in some other modulatory feature of the receptor/channel complex. The reason for the appearance of the NMDA response postanoxia is one of the most important barriers to our understanding of the mechanism of the events leading to death of neurons. There is a very similar problem in explaining the acute events occurring during exposure to anoxia in the slice preparations. In the presence of APV and PCP all indications of electrical excitability are lost after an initial period of hyperexcitability. There is considerable debate about the reason for loss of the responses to the quisqualatefkainate receptors, but it is likely that much of the block is due to depression of presynaptic release of transmitter (11). The fact that there is much less response after about 4 min of anoxia in the presence of NMDA antagonists than in the controls indicates that NMDA receptors become active. This is despite lack of a detectable effect of APV or PCP on the field potential in the control preparation. The appearance of NMDA responses might also result from excessive depolarization of the neuron, but the results of Fig. 6 show no such depolarization. Leblond and Krnjevic (11) have shown that CA1 neurons are hyperpolarized by brief periods of anoxia in slices. They (10) also demonstrated that a number of ionic currents, including calcium and potassium currents, were reduced and suggested that this reduction might be secondary to ATP depletion. Thus, the appearance of NMDA responses in these circumstances is surprising and not explicable on the basis of present knowledge of the voltage dependence of the NMDA response (7). There is an accumulating body of evidence that NMDA antagonists protect against anoxic neuronal cell death in whole animals (5,21,25), brain slices (3,8), and neuronal cell cultures (2). Anoxia has also been shown to trigger a proteolysis of hippocampal spectrin, which can be prevented by use of NMDA receptor blockers (23). Since spectrin is a substrate of the calcium-activated protease, calpain (24), the proteolysis is presumed to result from a rise in internal calcium concentration subsequent to activation of the NMDA receptor. Our results are compatible with a major role for the NMDA receptor in the cellular response to anoxic injury. However, they strongly suggest that the increase in NMDA receptor activity is a consequence of a process as yet not understood.
ET
AL.
ACKNOWLEDGMENTS These studies were supported thank Mrs. Charlene McAuliffe
by NIH Grant ROl NS23807. for secretarial assistance.
We
REFERENCES 1. ANIS, N. A., S. C. BERRY, N. R. BURTON,
2.
AND D. LODGE. 1983. The dissociative anesthetics, ketamine andphencyclidine, selectively reduce excitation of central mammalian neurons by Nmethyl-aspartate. Br. J. Phnrmacol. 79: 565-575. CHOI, D. W. 1988. Calcium-mediated neurotoxicity: Relationship to specific channel types and role in ischemic damage. Trends Neurosci. 11: 465-469.
3.
CLARK, G. D., AND S. M. ROTHMAN. 1987. Blockade amino acid receptors protects anoxic hippocampal science 21: 665-671.
4.
DIENEL, G. A., W. A. PULSINELLI, AND T. E. DUFFY. 1980. Regional protein synthesis in rat brain following acute hemispheric ischemia. J. Neurochem. 35: 1216-1226.
5.
GILL, R., A. C. FOSTER, AND G. N. WOODRUFF. neuroprotective in gerbils when administered ischemic period. Neuroscience 25: 847-855.
6.
HAJOS, F., C. GARTHWAITE, AND J. GARTHWAITE. 1986. Reversible and irreversible neuronal damage caused by excitatory amino acid analogues in rat cerebellar slices. Neuroscience 18: 417-436.
7.
HORI, N., N. DOI, S. MNAHARA, Y. SHINODA, AND D. 0. CARPENTER. 1989. NMDA responses on hippocampal neurons change after exposure to anoxia and hypoglycemia. Neurosci. Abstr. 15: 769.
8.
KASS, I. S., G. CHAMBERS, AND J. E. COTTRELL. 1989. The Nmethyl-D-aspartate antagonists aminophosphonovaleric acid and MK-801 reduce anoxic damage to dentate granule and CA1 pyramidal cells in the rat hippocampal slice. Erp. Neural. 103: 116-122. KASS, I. S., AND P. LI~ON. 1986. Calcium and long-term transmission damage following anoxia in dentate gyrus and CA1 regions of the hippocampal slice. J. Physiol. London 378: 313-334. KRNJEVIC, K., AND J. LEBLOND. 1989. Changes in membrane currents of hippocampal neurons evoked by brief anoxia. J. Neurophysiol. 62: 15-30. LEBLOND, J., AND K. KRNJEVIC. 1989. Hypoxic changes in hippocampal neurons. J. Neurophysial. 62: 1-14.
9.
10.
11.
of excitatory slices. Neuro-
1988. MK-801 is during the post-
12.
LIPTON, P., AND T. S. WHITTINGHAM. 1982. Reduced ATP concentration as a basis for synaptic transmission failure during hypoxia in the in vitro guinea pig hippocampus. J. Phystil. London 325: 51-65.
13.
LYNCH, G., J. LARSON, S. KELSO, G. BARRIONUEVO, AND F. SCHO?TLER. 1983. Intracellular injections of EGTA block induction of hippocampal long-term potentiation. Nature (London) 305: 719-721.
14.
MACDERMOTT, A. B., M. L. MAYER, G. L. WESTEIROOK, S. J. SMITH, AND J. L. BARKER. 1986. NMDA-receptor activation increases cytoplasmic calcium concentration in cultured spinal cord neurones. Nature fLondon) 321: 519-522. MALENKA, R. C., J. A. KAUER, R. S. ZUCKER, AND R. A. NICOLL. 1988. Postsynaptic calcium is sufficient for potentiation of hippocampal synaptic transmission. Science 242: 81-84.
15.
16.
MA~R,
M.
L., G. L. WESTBROOK,
AND P. B. GUTHRIE.
1984.
NMDA
17.
18.
19.
20.
RECEPTORS
AND DELAYED
Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones. Nature (London) 309: 261-263. MELLGREN, R. L. 1987. Calcium-dependent proteases: An enzyme system active at cellular membranes? FASEB J. 1: llO115. NOVELLI, A., J. A. REILLY, P. G. LYSKO, AND R. C. HENNEBERRY. 1988. Glutamate becomes neurotoxic via the NMDA receptor when intracellular energy levels are reduced. Brain Res. 451: 205-212. NOWAK, L., P. BREGESTOVSKI, P. ASCHER, A. HERBERT, AND A. PROCHIANTZ. 1984. Magnesium gates glutamate-activated channels in mouse central neurons. Nature (London) 307: 462-465. PULSINELLI, W. A., AND J. B. BRIE-Y. 1979. A new model of bilateral hemispheric ischemia in the unanesthetized rat. Stroke 10:267-272.
NEURONAL
DEATH
311
21. ROMAN, R., H. BARTKOWSKI, AND R. SIMON. 1989. The specific NMDA receptor antagonist AP-7 attenuates focal ischemic brain injury. Neurosci. Lett. 104: 19-24. 22. ROTHMAN, S. M., AND J. W. OLNEY. 1986. Glutamate and the pathophysiology of hypoxic-ischemic brain damage. Ann. Neural. 19:105-111. 23. SEUBERT, P., K. LEE, AND G. LYNCH. 1989. Ischemia triggers NMDA receptor-linked cytoskeletal proteolysis in hippocampus. Brain Res. 492: 366-370. 24. SIMAN, R., M. BAUDRY, AND G. LYNCH. 1984. Brain fodrin: Substrate for the endogenous calcium-activated protease, calpain I. Proc. Natl. Acad. Sci. USA 81: 3572-3576. 25. STEINBERG, G. K., J. SALEH, R. DELAPAZ, D. K~Is, AND S. R. ZARNEGAR. 1989. Pretreatment with the NMDA antagonist dextrorphan reduces cerebral injury following transient focal ischemia in rabbits. Brain Res. 497: 382-386.