Blockade of NMDA Receptors and Apoptotic Neurodegeneration in the Developing Brain

See allHide authors and affiliations

Science  01 Jan 1999:
Vol. 283, Issue 5398, pp. 70-74
DOI: 10.1126/science.283.5398.70


Programmed cell death (apoptosis) occurs during normal development of the central nervous system. However, the mechanisms that determine which neurons will succumb to apoptosis are poorly understood. Blockade of N-methyl-d-aspartate (NMDA) glutamate receptors for only a few hours during late fetal or early neonatal life triggered widespread apoptotic neurodegeneration in the developing rat brain, suggesting that the excitatory neurotransmitter glutamate, acting at NMDA receptors, controls neuronal survival. These findings may have relevance to human neurodevelopmental disorders involving prenatal (drug-abusing mothers) or postnatal (pediatric anesthesia) exposure to drugs that block NMDA receptors.

Glutamate promotes certain aspects of neuronal development, including migration, differentiation, and plasticity (1). In the first 2 weeks of neonatal life in the rat, the NMDA subtype of glutamate receptor undergoes a period of hypersensitivity, in which neurons bearing NMDA receptors are rendered highly sensitive to excitotoxic degeneration (2). During this period, NMDA receptors are primary mediators of glutamatergic fast excitatory neurotransmission in the brain (3). Although NMDA receptor activation can promote survival of cerebellar granule cells in vitro (4) or dentate granule neurons in vivo (5), evidence that even transient inactivation of NMDA receptors can be lethal for many neurons has not been described. We now report that during a specific stage in ontogenesis coinciding with the period of NMDA receptor hypersensitivity, transient blockade of NMDA receptors triggers a wave of apoptotic neurodegeneration in which large numbers of neurons are deleted from the developing brain.

Seven-day-old rats were injected intraperitoneally (ip) with vehicle or dizocilpine [(+)MK801] (0.5 mg per kilogram of body weight) at 0, 8, and 16 hours and the brains examined at 24 hours by TUNEL (terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling) and silver staining to detect degenerating cells (6, 7). In the vehicle-treated rats, degenerating neurons were sparsely distributed in the forebrain (Fig. 1, A and C, and Table 1). In contrast, in (+)MK801-treated rats there was a marked increase of degenerating neurons in specific regions (Fig. 1, B and D, and Table 1). Using an unbiased optical dissector method (8), we determined that MK801 caused an increased numerical density of degenerating neurons from 3-fold in the hypothalamus to 39-fold in the laterodorsal thalamus (Table 1). The density of degenerating neurons in vehicle-treated rats varied from 0.2 to 1.55% of the total neuronal density in the brain regions examined, whereas in (+)MK801-treated pups the density of degenerating neurons was as high as 15 to 26% of the total neuronal density in layer II of parietal, frontal, and cingulate cortices and 12% in the laterodorsal thalamus (Table 1). Examination of the degenerating cells by electron microscopy (9) revealed (Fig. 1, F and G) that the degenerative process was ultrastructurally indistinguishable from the physiological (programmed) cell death that occurs naturally in the developing brain (10, 11).

Figure 1

(A and B) Low-magnification (×9.8) images of transverse hemisections from the brains of 8-day-old rats treated with vehicle (A) or (+)MK801 (B) 24 hours previously. Both sections were stained by the TUNEL method. TUNEL-positive neurons (small dark dots) are abundant in the brain of the (+)MK801-treated rat (B), but present in insignificant amounts in the vehicle-treated control (A). (C) Enlargement (×98) of the boxed region in (A). (D) Enlargement (×98) of the boxed region in (B). (E) Magnified view (×98) of the superficial parietal cortex of an 8-day-old rat 24 hours after PCP treatment, showing that PCP triggers the same apoptotic response as (+)MK801 in layer II neurons of the parietal cortex. (E) was stained by the DeOlmos silver method (7) and shows that this silver method stains the same populations of degenerating neurons as the TUNEL method in the immature brain after PCP or (+)MK801 treatment. (F andG) Electron micrographs of degenerating neurons in (B). Ultrastructural changes after (+)MK801 treatment are similar to those seen in neurons undergoing programmed cell death (10,11). (F) Early stage of apoptosis. (G) A later stage in which the cell is more condensed and the nuclear membrane (arrows) has become fragmented and discontinuous. Throughout both stages the plasma membrane remains intact. Magnification, ×3275.

Table 1

The NMDA antagonist (+)MK801 increases the rate of apoptosis in the brain of 8-day-old rats. Seven-day-old rats received either vehicle or (+)MK801 (0.5 mg/kg ip) at 0, 8, and 16 hours and were killed at 24 hours on postnatal day 8 (P8). By use of an optical dissector method (8), the numerical cell densities (cells per cubic millimeter) and degenerating cell densities (degenerating cells per cubic millimeter) in 16 brain regions of vehicle- and (+)MK801-treated rats (n = 5 to 7) were estimated. Shown are numerical cell densities (mean ± SEM) in vehicle-treated 8-day-old rats. The extent of apoptosis in vehicle-treated and (+)MK801-treated rats is shown as the ratio of degenerating cell density to total cell density and is expressed as a percentage (mean ± SEM). **P < 0.01, ***P< 0.001 (Student's t test). CA1 HPC: CA1 hippocampus; DG: dentate gyrus; Thal ld: laterodorsal thalamus; Thal md: mediodorsal thalamus; Thal v: ventral thalamus; Hypothal vm: ventromedial hypothalamus; Fr: frontal cortex; Par: parietal cortex: Cing: cingulate cortex; Rspl: retrosplenial cortex; II, IV: layers 2 and 4.

View this table:

To determine whether nonneuronal cells were vulnerable to the proapoptotic effect of (+)MK801, we used TUNEL staining in combination with immunohistochemistry for glial fibrillary acidic protein (GFAP) to identify astrocytes (12). Examination of histological sections from 7-day-old rats treated with (+)MK801 revealed that cells staining positive for GFAP were distinct from those staining positive for TUNEL. Apoptotic cells that were examined by electron microscopy in the early stages of degeneration retained the morphological characteristics of neurons.

Administration of (+)MK801 (0.25 to 1 mg/kg ip) triggered apoptotic neurodegeneration in the 7-day-old rat brain that was dose-dependent (Fig. 2A). Rat pups that were killed at 4, 8, 12, 16, 24, and 48 hours after (+)MK801 (0.5 mg/kg) treatment showed an apoptotic response at 4 hours that increased progressively in the 12- to 24-hour interval (Fig. 2B).

Figure 2

(A) The effect of dose on the apoptotic response to (+)MK801 (•) or (−)MK801 (○) was studied by administering a single dose of (+)MK801 (0.05, 0.25, 0.5, 0.75, or 1 mg/kg ip) or (−)MK801 (0.05, 0.5, or 1 mg/kg ip) on P7 and killing the rat pups 24 hours later (n = 5 to 7). Degenerating neurons were sampled by the optical disector method (8) in silver-stained brain sections to provide a numerical density (degenerating neurons per cubic millimeter) for each of the 16 brain regions designated in Table 1. The numerical density counts from these brain regions were added to give a cumulative density count of degenerating neurons for each brain, and a mean was calculated for each treatment condition. Analysis of variance (ANOVA) revealed a significant effect of treatment with (+)MK801 [F(5,26) = 53.97; P < 0.0001)] with multiple comparisons showing that a dose of 0.05 mg of (+)MK801 per kilogram did not trigger apoptosis, whereas doses >0.25 mg/kg significantly increased apoptosis. Treatment of rat pups with (−)MK801 on P7 resulted in a weak neurodegenerative response that became significant compared with vehicle-treated rats (dashed line) only at the dose of 1 mg/kg. (B) The time course of the apoptosis response to (+)MK801 was studied by administering a single dose of (+)MK-801 (0.5 mg/kg ip) on P7 and killing the rat pups 4, 8, 12, 16, 24, or 48 hours later. The apoptotic response was evaluated as described in (A). ANOVA revealed a significant effect of posttreatment interval on severity of brain damage [F(6,28) = 34.43;P < 0.0001], with multiple comparisons showing significant increase at 4 hours and a more robust response at 16 and 24 hours. By 48 hours there were no remaining signs of apoptosis. Dashed line indicates spontaneous apoptosis.

To determine whether the proapoptotic effect of MK801 can be attributed to its NMDA receptor–blocking activity, we administered the NMDA antagonists phencyclidine (PCP), ketamine, and carboxypiperazin-4-yl-propyl-1-phosphonic acid (CPP). When administered by dosing regimens calculated to maintain a steady blockade of NMDA receptors for at least 8 hours (13), all three agents induced a pattern of cell degeneration similar to that induced by (+)MK801 (Tables 1 and 2and Fig. 1, E and F). Electron microscopy verified that the degenerative reaction induced by each agent was ultrastructurally indistinguishable from the programmed cell death that occurs naturally during brain development (10, 11). We also tested the (−) enantiomer of MK801, which is substantially less potent than (+)MK801 in blocking NMDA receptors. At a dose of 1 mg/kg, (−)MK801 triggered a weak apoptotic response compared with the same dose of (+)MK801 (Fig. 2A).

Table 2

NMDA antagonists trigger an increased rate of apoptosis in the rat brain on P0, P3, or P7. At these ages rat pups received either vehicle or (+)MK801 (0.5 mg/kg ip) at 0, 8, and 16 hours and were killed at 24 hours. The proapoptotic effect of PCP, ketamine, or CPP was tested in 7-day-old rats. Drugs were administered according to the dosing regimen described in (13). By use of an optical disector method (8), the numerical densities of degenerating neurons (degenerating neurons per cubic millimeter) in 14 brain regions of vehicle- and MK801-treated pups were estimated. Shown are mean numerical densities of degenerating neurons ± SEM (n = 5 to 7). Effects of NMDA antagonists on the rate of apoptosis at each age are presented as ratios of mean numerical densities of degenerating cells in drug-treated versus age-matched vehicle-treated rats [expressed as fold increase (↑)]. NMDA antagonists do not trigger apoptosis in brain regions displaying no physiological apoptosis, and vulnerability to the proapoptotic action of NMDA antagonists is dependent on developmental age. *P < 0.05, **P < 0.01, ***P < 0.001 (Student's t test). Abbreviations as in Table 1.

View this table:

We administered either vehicle or (+)MK801 to rats on postnatal days 0 (P0), P3, P7, P14, or P21. In addition, we treated pregnant rats on embryonic day 17 (E17), E19, or E21 with (+)MK801 [0.5 mg/kg subcutaneously (sc) at 0, 8, and 16 hours]. This dose regimen was chosen because it produced a maximal apoptotic response in our initial experiments with 7-day-old rats. Fetuses (removed by hysterotomy on E18 or E20) or postnatal pups were killed at 24 hours after the first treatment of the mother with (+)MK801.

Between P0 and P14, (+)MK801 triggered apoptosis, but the magnitude of the response was dependent on the developmental age (Tables 1 and 2). Sensitivity to (+)MK801-induced apoptosis was high on both P0 and P3, increased between P3 and P7, and then decreased sharply between P7 and P14. On P21 apoptotic neurons were undetectable in vehicle-treated rats and detectable at very low levels in (+)MK801-treated rats.

Distinct patterns of apoptotic neurodegeneration were observed at different developmental ages (Table 2). The rate of spontaneous apoptosis in vehicle-treated rats was high on P0, especially in the mediodorsal and ventral thalamic nuclei, the ventromedial hypothalamus, and caudate. Brain regions with an appreciable rate of spontaneous apoptosis on P0 also showed a robust response to MK801, which was most evident in the dentate gyrus (17-fold increase over the spontaneous rate). In contrast to all other brain regions studied, apoptosis was not detected in the cerebral cortex in either vehicle- or MK801-treated animals on P0. On P3 the rate of spontaneous apoptosis decreased markedly in the hypothalamus and mediodorsal and ventral thalamic nuclei, remained at a moderately high level in most other regions, and de novo increased to high levels in layer II of the frontal, parietal, cingulate, and retrosplenial divisions of the cerebral cortex. On P3, MK801 triggered apoptosis in all regions studied, notably the subiculum (8-fold) and laterodorsal thalamus (10-fold). On P7 the spontaneous rate declined to modest levels in all regions, but the MK801-induced increases remained high in many regions (14.6, 16.5, 18, 18, 24.2, 24.8, and 39.7 times the spontaneous rate in the frontal cortex, caudate, subiculum, retrosplenial, parietal, cingulate cortices, and laterodorsal thalamus, respectively). Regions that showed high spontaneous or induced rates on P0 (dentate gyrus, CA1 hippocampus, ventromedial hypothalamus, and ventral and mediodorsal thalamus) displayed low spontaneous and induced rates on P7.

The rate of (+)MK801-induced apoptosis in fetuses exposed in utero on E21 and killed on P0 (24 hours after the first treatment of the mother) was similar to that of animals treated as infants on P0 (27.3, 9.9, and 2.5 times higher than the spontaneous rate in the dentate gyrus, hippocampal CA1 subfield, and ventromedial hypothalamus, respectively). Fetuses exposed in utero on E19 showed no increases in regions examined, with the exception of the ventromedial hypothalamus (1.8-fold higher than vehicle-treated rats). Fetuses exposed in utero to either vehicle or MK801 on E17 showed neither spontaneous nor induced apoptosis in any of the regions examined.

To determine the specificity of the proapoptotic action of NMDA antagonists, we tested the effects of 6-nitro-7-sulfamoylbenzo[f]quinoxaline-2,3-dione (NBQX), an antagonist of non-NMDA glutamate receptors [20 mg/kg ip given at 0, 75, 150 min and 8 hours (n = 5)], and scopolamine hydrobromide, an antagonist of cholinergic muscarinic receptors [0.3 mg/kg ip given at 0, 4, and 8 hours (n = 5)]. In addition, we tested haloperidol, an antagonist of dopamine receptors [10 mg/kg ip given at 0 and 8 hours (n = 5)]. None of these agents reproduced the apoptosis-inducing action of NMDA antagonists.

Finally, because NMDA receptors are highly permeable to Ca2+ ions (13), we examined whether blockade of Ca2+ influx by other routes—that is, voltage-dependent Ca2+ channels—also cause apoptotic neurodegeneration in the brain. We treated 7-day-old rats with the Ca2+-channel blockers (14) nimodipine (50 mg/kg at 0 and 8 hours;n = 6) or nifedipine (10 mg/kg at 0 and 8 hours;n = 6) and examined the brains by TUNEL and silver staining at 24 hours after treatment. Neither of these agents caused significant apoptosis in the 7-day-old rat brain.

Our findings indicate that in the developing rat brain transient blockade of NMDA receptors causes sensitive neurons to die by a process that resembles the programmed cell death that occurs naturally in the developing brain (10, 11). The effect is caused by NMDA receptor blockade because agents that block NMDA receptors by different mechanisms triggered the response. The effect is age-dependent and is specific for NMDA receptors because it was not reproduced by blockade of other major excitatory systems in the brain (muscarinic and non-NMDA glutamatergic) or by blockade of a major inhibitory system (dopaminergic). Thus, it appears that many neurons in the mammalian brain undergo a stage during development when they are critically dependent on glutamate stimulation through NMDA receptors, and sustained deprivation of this input activates programmed cell death.

The neuronal populations vulnerable to degeneration induced by NMDA antagonists have abundant NMDA receptors (15). However, receptor density is not the sole determinant of neuronal vulnerability, because neurons in the CA1 hippocampal region express high levels of mRNA for the NR1 receptor subunit (15) on P7 and at this age are less sensitive than cortical layer II neurons to the proapoptotic action of NMDA antagonists. However, on E21 and P0, CA1 neurons are more vulnerable than cerebrocortical neurons to NMDA antagonists, which suggests that additional factors, such as the stage in synaptogenesis and the subunit composition of the NMDA receptor assembly, may govern neuronal vulnerability.

NMDA antagonists were most effective in triggering apoptosis in the rat forebrain on P7, the age at which the rat forebrain is most vulnerable to the excitotoxic effect of NMDA (2). Thus, during this developmental period survival of NMDA receptor–bearing neurons depends on glutamatergic input being regulated within narrow bounds. In the human fetal forebrain, expression of NMDA receptors peaks during weeks 20 to 22 of gestation (16), a period that marks the beginning of the brain growth spurt, which spans much of the third trimester of pregnancy and overlaps extensively into the postnatal period (17). In the rat, peak expression of NR1 (15) and the brain growth spurt (18) occur in the first week after birth. If peak vulnerability of the human forebrain to the proapoptotic action of NMDA antagonists corresponds to these developmental events in the rat, the window of vulnerability for humans would include the entire third trimester of pregnancy. Our findings may have relevance in a human context of drug abuse, where fetuses may be exposed in utero to agents such as PCP (angel dust), ketamine (special K), and ethanol (18, 19), all potent NMDA antagonists. Furthermore, human infants are sometimes exposed to ketamine and nitrous oxide, two NMDA antagonists widely used in pediatric anesthesia (20, 21).

It is important to consider the length of time for which NMDA receptors must be blocked for the neurodegenerative process to become operative. The threshold dose of (+)MK801-induced apoptosis in our study was 0.25 mg/kg, which produced behavioral symptoms indicative of NMDA receptor blockade lasting 4 to 6 hours. An increased rate of apoptosis was detected in the infant rat brain beginning 4 hours after administration of a single dose of (+)MK801 (0.5 mg/kg). Thus, blockade of NMDA receptors for ∼4 hours is sufficient to trigger apoptotic neurodegeneration in the developing mammalian brain.

Blockade of NMDA receptors gives rise to different patterns of neuronal loss depending on the stage of development at which the interference occurs. Such a mechanism could contribute to a variety of neuropsychiatric disorders.

  • * To whom correspondence should be addressed: E-mail: hrissanthi.ikonomidou{at}


View Abstract

Navigate This Article