Lycaeum > Leda > Documents > NMDA Antagonist Neurotoxicity: Mechanism and Prevention

The New

What's New


About Leda


Hosted Sites


NMDA Antagonist Neurotoxicity: Mechanism and Prevention

What's Related >>

By J. W. Olney et al. From "Science" 1991 Dec 6;254(5037):1515-8.


Antagonists of the N methyl D aspartate (NMDA) subtype of glutamate receptor, including phencyclidine (PCP) and ketamine, protect against brain damage in neuro logical disorders such as stroke. However, these agents have psychotomimetic properties in humans and morphologically damage neurons in the cerebral cortex of rats. It is now shown that the morphological damage can be prevented by certain anticholinergic drugs or by diazepam and barbiturates, which act at the y aminobutyric acid (GABA) receptor channel complex and are known to suppress the psychotomimetic symptoms caused by ketamine. Thus, it may be possible to prevent the unwanted side effects of NMDA antagonists, thereby enhancing their utility as neuroprotective drugs.

Antagonist of the NMDA sub type of glutamate receptor are potentially useful for preventing neuronal degeneration in neurological disorders such as stroke (1). However, treatment of adult rats with noncompetitive (phencycli dine, MK 801, tiletamine, ketamine) or competitive [D 2 amino 5 phosphonopentanoate (D AP5)] NMDA antagonists causes neurotoxic side effects consisting of pathomorphological changes in neurons of the cingulate and retrosplenial cerebral cortices (2, 3). After low doses these changes may be reversible, but higher doses can cause irreversible neuronal necrosis (4). Therefore, it has been questioned whether NMDA antagonist therapy can be applied without incurring serious side effects. However, we now report that certain anticholinergic or GABAergic agents protect cerebrocortical neurons against the adverse side effects of NMDA antagonists.

The neurotoxic action of MK 801 in the adult rat cingulate cortex is potentiated by pretreatment with the cholinergic agonist pilocarpine (5). This potentiating effect was abolished by coadministration of scopolamine, a cholinergic muscarinic antagonist. These results suggested that activation of muscarinic receptors might be involved in the process by which MK 801 causes neurotoxic side effects. To explore this possibility, we administered scopolamine intraperitoneally (ip) in various doses (0.01 to 5 mg per kilogram of body weight) 10 min after a subcutaneous (sc) dose of MK 801 (0.4 mg/kg) that reliably causes neurotoxic side effects in 100% of treated rats. We examined the brains after 4 hours and found that scopolamine completely prevented MK 801 neurotoxicity at doses >0.25 mg/kg. Dose-response studies revealed that the ED50 (dose of scopolamine that prevented MK-801 neurotoxicity in 50% of treated animals) was 0.13 mg/kg. Additional anticholinergic compounds were also effective with an order of potencies correlating with their binding affinities for M1 muscarinic receptors (6).

To determine whether anticholinergic agents can protect against the neurotoxic effects of noncompetitive NMDA antagonists other than MK 801, we treated six rats with a neurotoxic dose (5 mg/kg sc) of PCP and six rats with this dose of PCP plus scopolamine (0.5 mg/kg ip) and killed the animals 4 hours later. All of the rats treated with PCP alone had conspicuous vacuolar changes in cingulate and retrosplenial cortical neurons, whereas none of the rats treated with PCP plus scopolamine had such changes.

In order for NMDA antagonists to be optimally useful as neuroprotective agents in conditions such as stroke, it may be necessary to use relatively large doses. Therefore, we conducted experiments to determine whether the dose of anticholinergic agent required to prevent neurotoxic side effects of a low dose of NMDA antagonist would also protect against a high dose. It required 0.25 mg/kg ip scopolamine to prevent the neurotoxic side effects in 100% of treated rats (n = 6) after a relatively low dose of MK 801 (0.4 mg/kg sc); therefore, we treated adult rats with this dose of scopolamine plus a high dose of MK 801 (5 mg/kg sc) and found that it prevented the neurotoxic side effect in all animals (n = 6), whereas all controls (n = 6) that received the high dose of MK 801 by itself had a severe vacuole reaction in cingulate retrosplenial cortical neurons.

Using a chick embryo retina assay, we conducted experiments to determine whether anticholinergic drugs interfere with the neuroprotective actions of NMDA antagonists. We have shown that the neurotoxic action of NMDA (120 mM) in the chick retina is prevented by adding 200 nM MK 801 to the incubation medium. Therefore, we incubated the chick retina in medium containing NMDA (120 mM) and MK 801 (200 nM) and added scopolamine in various concentrations from 10 to 50 mM. Scopolamine did not interfere with the ability of MK 801 to prevent NMDA neurotoxicity. Thus, a tissue concentration of scopolamine 250 times higher than that of MK 801 does not interfere with the neuroprotective properties of MK 801, whereas a dose of scopolamine only 1/20 as high as the MK 801 dose prevented the neurotoxic side effects of MK 801 in rodent cortex.

The competitive NMDA antagonist D AP5, when injected into the cingulate cortex, causes a neurotoxic reaction identical to that caused by systemic MK 801 or PCP (3). In the present study, we extended these experiments to include systemic administration of 3 [(+) 2 carboxypiperazin 4 yl]-propyl 1 phosphonic acid (CPP), a competitive NMDA antagonist that is more potent than D AP5. Intravenous (iv) administration of CPP in a dose of 50 mg/kg caused a vacuole reaction in six of six rats that was identical to the reaction in six of six positive controls that received MK 801 (0.4 mg/kg sc) and was not present in six of six controls treated with normal saline. Administration of scopolamine (0.5 mg/kg ip) 10 min after CPP (50 mg/kg iv) prevented this reaction in six of six rats.

The psychotomimetic effects of PCP include hallucinogens, agitation, and disorientation. Ketamine, a PCP receptor ligand, when used in humans as an anesthetic, causes similar psychotomimetic effects termed an "emergence" reaction (7). Because diazepam reduces the severity of the psychotomimetic side effects of ketamine and is used widely in human anesthesia for this purpose (8), we studied the effects of diazepam on NMDA antagonist neurotoxicity. Because barbiturates, such as diazepam, act at GABAA receptors, we also tested several barbiturates. At a dose of 1 mg/kg ip, diazepam provided up to 50% protection; this effect could not be exceeded by increasing the dose sevenfold. However, each of four barbiturates completely protected against NMDA antagonist neurotoxicity with a steep dose response curve. The protection conferred by barbiturates cannot be attributed to general anesthesia properties because the nonbarbiturate anesthetic halothane did not suppress NMDA antagonist neurotoxicity (9).

Although the neurotoxic reaction appears to be confined to specific neurons within the cingulate and retrosplenial cortices, the pathological action of NMDA antagonists may not be limited to these neuronal populations. These agents induce heat shock protein (HSP) not only in the specific neurons that undergo pathomorphological changes but also in several other types of forebrain neurons (10, 11). This HSP response is blocked by scopolamine or GABAergic agents. Thus, the HSP response may be triggered by the same toxic mechanism that causes vacuolization of cerebrocortical neurons and may serve as an alternate marker of neuronal susceptibility to the type of injury.

Our evidence that either competitive or noncompetitive NMDA antagonists cause identical neurotoxic side effects, and either muscarinic cholinergic antagonists or GABAergic agents block these side effects, suggests that the effect is triggered by suppression of NMDA receptor function but also involves muscarinic and GABAergic receptors.

The difference in efficacy between diazepam and barbiturates may relate to their different modes of action at the GABA receptor. Barbiturates act directly to open the chloride channel, even in the absence of GABA (12), whereas diazepam acts only to potentiate the action of GABA (13). Because blockade of the NMDA receptor would result in cessation of GABA release, there would be very little GABA in the synaptic cleft for diazepam to potentiate. Therefore, consistent with our findings, diazepam should have only a partial effect in contrast to barbiturates which, like anticholinergics, should provide complete protection against the pathomorphological effects of the NMDA antagonist.

Typically, diazepam has been found to be partially but not completely effective in eliminating emergence symptoms associated with ketamine anesthesia (8). We found no studies pertaining to the use of anticholinergics for this purpose, and only a single study (14) pertaining to barbiturates. In this study, a single dose of thiopental (2 to 3 mg/kg) provided more complete protection against ketamine emergence reactions than has been reported for diazepam (8, 14).

Because GABAergic agents can prevent both the morphopathological and psychopathological side effects of NMDA antagonists, a common mechanism may underlie both effects (15). If so, it is possible that either GABAergic or anticholinergic drugs may provide a simple and safe (16) method of preventing both psychotomimetic and neurotoxic side effects of NMDA antagonists without interfering (17) with the neuroprotective properties of these compounds.


1. S. M. Rothman, J. Neurosci. 4,1884 (1984); D. W. Choi, Neuron 1, 623 (1988); J. W. Olney, Biol. Psychiatry 26, 505 (1989).

2. J. W. Olney, J. Labruyere, M. T. Price, Science 24, 1360 (1989).

3. J. Labruyere, M. T. Price, J. W. Olney, Soc. Neurosci. Abstr. 15, 761 (1989).

4. H. L. Allen and L. L. lversen, Science 247,221 (1990).

5. J. W. Olney and M. T. Price, unpublished data.

6. R. E. Burke, Movement Disorders 1, 135 (1986); S. B. Freedman, M. S. Beer, E. A. Harley, Eur. J. Pharmacol. 156, 133 (1988).

7. B. E. Marshall and D. E. Longnecker, in Goodman and Gilman's The Pharmacological Basis of Therapeutics, A. G. Gilman, T. W. Rall, A. S. Nies, P. Taylor, Eds. (Pergamon, New York, 1990), p. 307.

8. D. L. Rcich and G. Silvay, Can.J. Anaesth. 36,186 (1989), E. N. Ayim and F. X. Makatia, East Afr. Med J. 53,377 (1976); K. Korttila and J. Levanen, Acta Anaesthesiol. Scand. 22, 640 (1978); D. L. Coppel, J. G. Bovill, J. W. Dundee, Anaesthesia 28, 293 (1973).

9. Rats (n = 6) were treated with a standard dose of MK 801 (0.4 mg/kg sc), then anesthetized with halothane, and maintained under halothane anesthesia for 4 hours. When killed at 4 hours, all rats in both groups had an abundant display of vacuolated neurons in the cingulate and retrosplenial cortices.

10. J. W. Sharp, S. M. Sagar, F. R. Sharp, Soc. Neurosci. Abstr. 16, 1122 (1990).

11. M. A. Sesma, M. T. Price, J. W. Olney, ibid., in press. We localized HSP immunocytochemically using a primary antibody directed against HSP72 (Amersham RPN 1197) and a modified Vector Elite avidin biotin peroxidase (ABC) method recommended by F. R. Sharp (9). An antisera free incubation served as an internal control for each case.

12. R. L. Macdonald and J. L. Barker, Science 200, 775 (1978), E. S. Levitan, L. A. Blair, V. E. Dionne, E. A Barnard, Neuron 1, 773 (1980); N. Akaike, T. Maruyama, N. Tokutomi, J. Physiol. (London) 394, 85 (1987), C. F. Zorumski and K. E. Isenberg, Am. J. Psychiatry 148, 162 (1991).

13. R. E. Study and J. L. Barker, Proc. Natl. Acad. Sci. U.S.A. 78, 7180 (1981).

14. J. A. O. Magbagbeola and N. A. Thomas, Can. Anaesth. Soc. J. 21, 321 (1974).

15. The proposal that the pathomorphological and psychotomimetic side effects of NMDA antagonists may be causally linked rests on the reasonable assumption that reversible injury confined to cingulate retrosplenial neurons might produce a temporary derangement in psychological functions mediated by any or all components of an extensive neural network with which these neurons communicate.

16. Behavioral side effects of competitive and noncompetitive NMDA antagonists are similar and consist of ataxia and hyperactivity at low to moderate doses, with a progressive increase in muscle tone at higher doses that causes the animals to lie on their sides with partially flexed limbs held in a rigid posture. Anticholinergic drugs were well tolerated; in fact they tended to relieve these symptoms, especially the muscular rigidity. Treatment with an NMDA antagonist plus a barbiturate was also well tolerared. Consistent with the barbiturate effect alone, the animals appeared heavily sedated, but there was no apparent potentiation by the barbiturate of the NMDA antagonisrt's effects or vice versa, and respiratory function was not compromised.

17. Certain barbiturates, especially thiamylal, effectively block both NMDA and non NMDA subtypes of glutamate receptor and can prevent ischemic neuronal degenerarion [J. Olney et al., Neurosri. Lett. 68,29 (1986); J. Olney, in Excitatory Amino Acids in Health and Disease, D. Lodge, Ed. (Wiley, London, 1988), pp. 337 352]. Thus, barbiturates are neuroprotective in a dual sense. They protect against isehemic neuronal degeneration in many brain regions by blocking glutamate (induding NMDA) receptors, while preventing NMDA antagonist neurotoxicity in the cingulate cortex by exerting GABAmimetic activity that is stronger than their NMDA antagonist activity.

18. G. Paxinos and C. Watson, The Rat Brain in Stereotaxic Coordinates (Academic Press, New York, 1982).

19. We thank F. R. Sharp for advice on immunohistochemical methods. Supported in part by a grant from the Huntington's Disease Foundation, PHS grants DA 53568 and AG 05681, National Institute of Mental Health Research Scientist Award MH 38894 (J.W.O.), a Weldon Spring Fund grant, and a Research Leave Award from the University of Missouri St. Louis (M.A.S.).

Created 11/16/2000 22:52:23
Modified 11/16/2000 22:52:23
Leda version 1.4.3