Dynamic changes in N-methyl-d-aspartate receptors after closed head injury in mice: Implications for treatment of neurological and cognitive deficits

Abstract

Traumatic brain injury is a leading cause of mortality and morbidity among young people. For the last couple of decades, it was believed that excess stimulation of brain receptors for the excitatory neurotransmitter glutamate was a major cause of delayed neuronal death after head injury, and several major clinical trials in severely head injured patients used blockers of the glutamate N-methyl-d-aspartate (NMDA) receptor. All of these trials failed to show efficacy. Using a mouse model of traumatic brain injury and quantitative autoradiography of the activity-dependent NMDA receptor antagonist MK801, we show that hyperactivation of glutamate NMDA receptors after injury is short-lived (<1 h) and is followed by a profound and long-lasting (≥7 days) loss of function. Furthermore, stimulation of NMDA receptors by NMDA 24 and 48 h postinjury produced a significant attenuation of neurological deficits (blocked by coadministration of MK801) and restored cognitive performance 14 days postinjury. These results provide the underlying mechanism for the well known but heretofore unexplained short therapeutic window of glutamate antagonists after brain injury and support a pharmacological intervention with a relatively long (≥24 h) time window easily attainable for treatment of human accidental head injury.

Head trauma is a leading cause of mortality and morbidity among young people in the western world (1). Traumatic and ischemic brain injury triggers a large, transient increase in excitatory amino acid transmitter efflux in the brain of experimental animals and human subjects (2-6). Glutamate activation of the N-methyl-d-aspartate (NMDA) receptor (NMDAR), which is a ligand-gated ion (calcium and sodium) channel, results in channel opening and ion influx into the cell. It has been suggested that this process mediates delayed excitotoxic neuronal death after brain ischemia and trauma (7, 8), although the concept is not universally accepted (9, 10). Support for the involvement of NMDAR activation in neuronal death after brain injury has come from numerous studies showing that NMDAR antagonists reduce cell death and improve outcome in animal models of traumatic brain injury (TBI) and stroke. NMDAR antagonists appear to be most efficacious when given before or immediately after the insult and lose efficacy if administered >30-60 min postinjury (11-15).

Microdialysis studies of extracellular glutamate in human TBI and stroke patients suggested that the increase in glutamate in humans is more sustained [6 h to several days (7, 8)] than in rodents, where it only lasts minutes (2-6). This result may have contributed to a decision to administer NMDAR antagonists in clinical trials of head injury for several days rather than once after severe nonpenetrating injury. All clinical trials of NMDAR antagonists to date failed to show efficacy. Furthermore, some of these trials had to be stopped prematurely because of increases in mortality and morbidity in the drug arm of the stroke trials (16-18), suggesting that prolonged blockade of NMDAR may actually be harmful in the posttraumatic or postischemic patient. Studies of energy metabolism after human and rodent TBI demonstrate large dynamic changes occurring within the first hour after TBI, such that a hypermetabolic state lasting only 30 min (in rats) to a few hours (in humans) is followed by a profound depression lasting 5 and ≥30 days in rats and humans, respectively (19-21). Epileptic activity and reductions in ATP content are also restricted to the acute phase after brain injury in rats (22, 23). These combined observations led us to hypothesize that NMDAR may undergo similarly large and dynamic changes in availability and responsiveness after TBI. Stimulation of NMDAR is thought to be crucial for memory formation in the mammalian brain (24), and cognitive impairments are extremely common and often long lasting after TBI (25, 26), suggesting a functional deficit rather than excessive stimulation of NMDAR in the chronic stage after TBI. Excessive activation followed within a relatively short time by desensitization and loss of functional NMDAR could explain the preclinical and clinical experience with NMDAR antagonists and suggest alternative modes of treatment. This hypothesis was tested in a mouse model of closed head injury (CHI) (27, 28) produced by a free-falling weight impacting the intact skull, in combination with quantitative autoradiography of the radiolabeled NMDAR noncompetitive antagonist MK801 (29, 30). Because noncompetitive NMDAR antagonists (e.g., TCP and MK801) are thought to be “use-dependent” ligands, i.e., they bind to a site inside the channel made accessible when the receptor is activated by glutamate (31), an increase in binding of agents of this class can be used as a functional marker of excessive NMDAR activation in the brain (32, 33).

Materials and Methods

Trauma Model. We used 8- to 12-week-old male Sabra mice (30-35 g) kept under controlled temperature and light conditions with food and water available ad libitum. The study was approved by the Institutional Animal Care Committee of the Hebrew University. Experimental CHI was induced by using a modified weight-drop device developed in our laboratory (27, 28). Briefly, after induction of ether anesthesia, a midline longitudinal incision was performed, the skin was retracted and the skull was exposed. The left anterior frontal area was identified and a Teflon tipped cone (2-mm diameter) was placed 1 mm lateral to the midline, in the midcoronal plane. The head was held in place manually and a 75-g weight was dropped on the cone from a height of 18 cm, resulting in a focal injury to the left hemisphere. After trauma, the mice received supporting oxygenation with 95% O₂ for no longer than 2 min and were then returned to their cages. Sham controls received anesthesia and skin incision only. For autoradiographic studies, groups of four to five animals per time point were killed 15 min, 1, 4, 8, or 24 h, or 7 days postinjury and brains were quickly removed and frozen on dry ice.

Autoradiography of Activated (Open Channel State) NMDAR. Autoradiography for activated NMDAR distribution was performed on freshly frozen, unwashed brain sections as described by Porter and Greenamyre (30) with small modifications. Six parallel series of consecutive cryostat sections (10 μm, cut at -15°C and thaw-mounted onto coated glass slides) were produced in the coronal plane and collected at 100- to 200-μm intervals from the prefrontal cortex to the cerebellum. The preincubation stage (meant to facilitate removal of endogenous glutamate and other water-soluble coactivators or blockers) was intentionally omitted (29, 30). Sections were incubated directly with a very small volume (10-100 μl, depending on the size of the section) of 10 nM [³H]MK801 in 50 mM Tris-acetate buffer at pH 7.4 for 4 h at room temperature, without addition of exogenous glutamate and glycine to the incubation mixture. Binding density under these conditions is proportional to the actual density of activated (open channel) NMDAR in the individual brain and region in situ. Nonspecific binding was measured on a second series by incubating the radioactive ligand in the presence of excess (5 μM) unlabeled MK801, and sections were washed, dried, and apposed to film with calibrated tritium micro scales (Amersham Pharmacia) as described (30, 34). The resulting autoradiograms of sections and standards were digitized simultaneously by using a large-bed UMAX scanner. A third series was stained with cresyl violet for anatomical verification of the lesion site and regions of interest. Sections from the remaining three series were labeled with iodoMK801 (35) and used for experiments involving manipulations of incubation conditions (Supporting Materials and Methods, which is published as supporting information on the PNAS web site).

Image Analysis. Scanned images were analyzed quantitatively (36, 37) by using nih image software. Anatomical regions underlying the autoradiographic images were identified on the consecutive brain sections stained with cresyl violet in reference to a mouse brain atlas (38). Regions of interest were grouped in three levels by distance from the impact, as follows: level 1 (anterior striatal level, Bregma 1.7 to 0.02) contained sections in the direct path of the impact; level 2 (posterior striatal level, Bregma -0.22 to -0.82) contained regions directly posterior to the impact; and level 3 (dorsal hippocampal level, Bregma -1.2 to -2.06) contained regions more remote (posterior) from the site of impact. Gray levels of seven to eight regions per anatomical level were measured in triplicates on the left and right separately; the gray levels were converted to nCi/mg (1 Ci = 37 GBq) via the calibrated standard curve. Nonspecific binding was similarly measured for each region and animal, and the values of nonspecific binding subtracted from total binding to yield specific binding. The mean specific binding values in the various regions of sham treated animals were then compared statistically to the mean values in CHI animals killed at various times after the injury.

Drug Treatments. The NMDAR agonist NMDA and the noncompetitive antagonist MK801 were dissolved in saline and injected i.p. in groups of eight to nine animals per treatment. Three groups of animals were used to test the effect of NMDAR activation on neurological recovery: (i) head injury followed by saline vehicle 1 and 2 days later (controls); (ii) head injury followed by NMDA 20 mg/kg, 1 and 2 days later; or (iii) head injury followed by NMDA 20 mg/kg combined with MK801 1 mg/kg, both administered 1 and 2 days after the injury. The three groups were repeatedly evaluated for neurobehavioral deficits over a 2-week period (see below). The NMDA dose was chosen from a dose-response experiment in intact mice in which we tested doses between 20 and 100 mg/kg. The lowest dose produced hyperactivity and some stereotypy but no convulsions or mortality, whereas significant mortality was observed at doses above 60 mg/kg. Cognitive function was evaluated in head-injured and intact rats administered with either NMDA 20 mg/kg or saline 24 and again 48 h postinjury and tested 14 days later for performance in the object recognition test (see below).

Neurobehavioral Evaluation. The neurological severity score (NSS) is a 10-point scale that assesses the functional neurological status of mice based on the presence of various reflexes and the ability to perform motor and behavioral tasks such as beam walking, beam balance, and spontaneous locomotion (39). Animals are awarded one point for failure to perform one item, such that scores can range from zero (healthy uninjured animals) to a maximum of 10, indicating severe neurological dysfunction, with failure at all tasks. The NSS obtained 1 h after trauma reflects the initial severity of injury and is inversely correlated with neurological outcome. Animals were evaluated 1 h after CHI, and 1, 2 or 3, 7, and 14 days later. Mice were randomized to the three groups after the initial evaluation to ensure similar initial severity values. Each animal was assessed by an observer who was blinded to treatment. The extent of recovery (dNSS) was calculated as the difference between the NSS at 1 h and at any subsequent time point. Thus, a positive dNSS reflects recovery, a 0 reflects no change, and a negative dNSS reflects neurological deterioration.

Evaluation of Performance in the Object Recognition Test. The object-recognition test was performed as originally described by Ennaceur and Delacour (40). In the first part of the test (14 days after CHI) mice were placed in the testing cage (a glass aquarium-like transparent box of 60 × 60 cm) for 1 h habituation. On the following day they were put back into the same cage with two identical objects. The cumulative time spent by the mouse at each of the objects was recorded manually during a 5-min interval by an observer blinded to the treatment received. Four hours later, the mice were reintroduced into the cage, where one of the two objects was replaced by a new one. The time (of 5 min total) spent at each of the objects was recorded. The basic measure is the percent of the total time spent by mice in exploring an object during the testing period, whereby normal healthy rodents will spend relatively more time exploring a new object than a familiar, i.e., “memorized” object.

Results

Morphological Consequences of CHI. The injury resulted in time-dependent pathological changes in brain morphology, including intraparenchymal bleeding and edema, as described (27, 28, 39, 41-43). Animals killed 15 min to 24 h after CHI had some intracranial bleeding at the site of impact but no lesions. By day 7, all animals had a distinct cavitation lesion surrounded by dense gliosis corresponding to the area in the immediate vicinity of the impact. The brain on the side contralateral to the injury did not present bleeding or a lesion.

Dynamic Posttraumatic Changes in MK801 Binding to NMDAR. The density of activated NMDAR measured by quantitative autoradiography of MK801 in freshly frozen, unwashed brain sections (30) showed time-dependent changes in binding that varied with the brain region analyzed and the distance from the focal injury. Significant, bilateral increases in binding, presumably indicative of increased receptor activation and channel opening by increased efflux of endogenous glutamate, were measured 15 min postinjury in many regions posterior (in the “penumbra”) to the impact (Table 1). The largest increases (>50%) were seen in the hippocampus, especially the CA1 field and the dentate gyrus. The hippocampus contains the highest concentrations of NMDAR in the brain (29) (Fig. 1) and is intimately involved in memory function (44-46). Additional regions showing significant increases included the substantia innominata, amygdala, and several cortical and subcortical regions posterior and ventral to the impact (Table 1). In contrast, brain regions at close proximity to the impact showed a significant bilateral decrease in binding at this time point (Table 2 and Fig. 1). NMDAR open-channel binding in all regions declined progressively over time between 60 min and 8-24 h postinjury and remained low 1 week postinjury (Tables 1 and 2). The regions most affected by CHI, showing >50% reduction in binding, were the cortical regions closest to the impact (Table 2). However, significant decreases of 30-50% were also measured in the hippocampus, perirhinal cortex, and other regions posterior and ventral to the injury (Table 1). Seven days postinjury, binding was significantly lower on the injured (ipsilateral) compared to the contralateral side, in the brain regions closest to the impact (Table 2). These lateralized effects were not observed at earlier time points or in regions posterior and ventral to the impact (Table 1).

Table 1. Effect of injury and time on regional distribution of NMDAR posterior to the impact.

Region	Sham	15 min	1 h	4 h	8 h	24 h	7 days
Amygdala
Left	6.7 ± 0.6	8.67 ± 0.8	5.4 ± 0.6	4.8 ± 0.5*	4.5 ± 0.7*	5.5 ± 0.5	6.1 ± 0.4
Right	6.9 ± 0.6	9.0 ± 1.0†	5.7 ± 0.4	5.2 ± 0.5	4.5 ± 0.7*	5.7 ± 0.3	6.2 ± 0.6
Anterior thalamus
Left	5.8 ± 0.37	6.8 ± 0.29	4.4 ± 0.45*	3.5 ± 0.56*	5.5 ± 0.32	4.9 ± 0.10*	5.0 ± 0.46*
Right	6.0 ± 0.39	6.5 ± 0.42	4.4 ± 0.34*	3.5 ± 0.57*	5.5 ± 0.32	4.7 ± 0.32*	5.1 ± 0.48*
Dentate gyrus
Left	16.7 ± 3.2	27.2 ± 4.4†	12.7 ± 1.3	10.2 ± 1.9	7.7 ± 1.1*	12.7 ± 1.9	12.0 ± 1.1
Right	16.4 ± 3.5	26.3 ± 3.7†	11.7 ± 2.2	10.4 ± 2.0	7.2 ± 1.2*	12.1 ± 1.5	11.2 ± 0.8
Dorsolateral striatum
Left	5.7 ± 0.34	6.7 ± 0.53†	4.3 ± 0.33*	3.9 ± 0.71*	5.4 ± 0.38	4.6 ± 0.35*	4.3 ± 0.42*
Right	5.7 ± 0.45	7.0 ± 0.54†	4.0 ± 0.45*	3.8 ± 0.54*	5.4 ± 0.39	4.4 ± 0.20*	4.8 ± 0.43*
Hippocampus CA1
Left	19.9 ± 4.0	33.0 ± 5.0†	13.7 ± 1.3	13.8 ± 2.3	9.2 ± 1.2*	14.4 ± 2.0	14.7 ± 1.4
Right	21.1 ± 4.9	36.5 ± 5.5†	12.1 ± 2.4	13.2 ± 3.8	9.5 ± 2.0*	14.4 ± 1.6	14.2 ± 1.2
Hippocampus CA3
Left	13.8 ± 2.1	19.0 ± 2.1†	11.2 ± 1.0	8.7 ± 1.2*	6.8 ± 1.1*	10.4 ± 1.2	9.8 ± 0.6*
Right	15.6 ± 3.1	20.2 ± 3.0	14.4 ± 4.4	10.7 ± 1.3	6.2 ± 1.0*	10.4 ± 1.1	9.7 ± 0.5*
Hypothalamus
Left	3.8 ± 0.5	5.3 ± 0.4†	3.4 ± 0.6	3.4 ± 0.4	2.8 ± 0.7	2.9 ± 0.2	3.4 ± 0.15
Right	4.0 ± 0.7	5.3 ± 0.27†	3.14 ± 0.3	3.3 ± 0.2	3.1 ± 0.7	2.8 ± 0.1#	3.4 ± 0.2
Motor cortex
Left	9.7 ± 0.84	11.2 ± 1.3†	5.8 ± 0.56*	6.6 ± 1.0*	6.9 ± 0.79*	8.3 ± 0.56*	5.3 ± 0.52*
Right	9.7 ± 0.64	11.4 ± 1.1†	6.0 ± 0.49*	6.4 ± 1.0*	6.7 ± 0.69*	7.8 ± 0.45*	6.2 ± 0.50*
Perirhinal cortex
Left	8.0 ± 0.6	9.16 ± 1.4	6.4 ± 0.3	5.1 ± 0.8*	4.5 ± 1.0*	5.4 ± 0.4*	6.1 ± 0.4*
Right	8.1 ± 0.85	9.0 ± 1.3	7.5 ± 0.3	NA	4.5 ± 1.0*	5.6 ± 0.4*	5.7 ± 0.4
Piriform cortex
Left	7.9 ± 0.45	9.0 ± 1.3	5.8 ± 0.61*	5.1 ± 0.83*	6.0 ± 0.45*	6.0 ± 0.23*	5.3 ± 0.48*
Right	7.9 ± 0.37	9.2 ± 0.75	5.9 ± 0.62*	5.1 ± 0.66*	6.2 ± 0.53*	6.1 ± 0.36*	5.7 ± 0.52*
Somatosensory cortex
Left	10.1 ± 0.64	11.7 ± 1.3†	6.2 ± 0.56*	6.8 ± 1.2*	6.9 ± 0.72*	8.2 ± 0.51*	5.2 ± 0.54*
Right	10.1 ± 0.71	12.2 ± 0.87†	6.7 ± 0.63*	6.8 ± 1.1*	7.3 ± 0.86*	8.4 ± 0.50*	6.6 ± 0.53*
Substantia innominata
Left	3.7 ± 0.30	5.0 ± 0.32†	3.5 ± 0.40	2.2 ± 0.36*	4.5 ± 0.20*	2.3 ± 0.31*	3.1 ± 0.44
Right	3.5 ± 0.31	4.8 ± 0.22†	3.6 ± 0.43	2.0 ± 0.32*	4.5 ± 0.11*	2.5 ± 0.27*	3.1 ± 0.42
Ventromedial striatum
Left	6.1 ± 0.27	6.9 ± 0.50†	4.5 ± 0.40*	4.1 ± 0.75*	5.4 ± 0.37	4.9 ± 0.58*	4.6 ± 0.43*
Right	5.9 ± 0.41	7.0 ± 0.52†	4.3 ± 0.42*	4.1 ± 0.63*	5.3 ± 0.35	4.6 ± 0.37*	4.9 ± 0.46*
Ventral thalamus
Left	5.97 ± 0.5	6.2 ± 0.47	4.3 ± 0.4*	4.6 ± 0.5*	4.9 ± 0.3	4.3 ± 0.3*	5.3 ± 0.35
Right	5.8 ± 0.7	6.3 ± 0.51	4.46 ± 0.5	4.3 ± 0.4*	4.1 ± 0.5*	4.3 ± 0.2*	4.9 ± 0.4

Brain sections from four to five animals per time point were collected at coronal levels posterior to the impact (levels ii and iii in Materials and Methods) and incubated with [3H]MK801 without washing or addition of glutamate and glycine. The table shows the mean ± SEM of density of specifically bound radioactivity in nCi/mg in regions in the uninjured (right) and injured (left) hemispheres. ANOVA (time) with repeated measures (region) showed a significant effect of time and region on MK801 binding density. Individual regions where than tested by one-way ANOVA. Significance is defined as P < 0.05 by ANOVA followed by post-hoc Fisher's PLSD test.

^*Lower than sham, P < 0.05.

^†Higher than sham, P < 0.05.

Fig. 1.

Dynamic changes in NMDAR open channel availability after CHI. Brain sections at the level of dorsal hippocampus from a noninjured animal (Top), a CHI mouse killed 15 min postinjury (Middle), and an injured animal killed 8 h postinjury (Bottom) illustrate the dynamic nature of changes in NMDAR after injury. The original autoradiograms were pseudocolored by using the extended rainbow color scale shown above the top section (red and white representing the highest values). Note the bilateral increase in MK801 binding throughout the hippocampus, cortex, and thalamus of the animal killed 15 min postinjury and the profound decline in the same regions in an animal killed 8 h after CHI.

Table 2. Effect of injury and time on regional distribution of NMDAR, coronal level of impact.

Region	Sham	15 min	1 h	4 h	8 h	24 h	7 days
Corpus callosum
Left	4.8 ± 1.1	3.6 ± 0.8	3.0 ± 0.65	2.1 ± 0.22	2.6 ± 0.72	1.9 ± 0.26	3.7 ± 0.20
Right	5.0 ± 1.1	3.5 ± 0.9	3.0 ± 0.7	2.0 ± 0.28	2.5 ± 0.67	1.9 ± 0.19	3.9 ± 0.12
Cingulate cortex
Left	14.1 ± 2.1	8.3 ± 2.2	7.3 ± 0.9	6.3 ± 1.2	4.7 ± 0.95	8.1 ± 0.90	5.9 ± 0.31
Right	14.4 ± 2.0	8.6 ± 2.4	7.2 ± 0.8	6.2 ± 1.1	4.6 ± 0.89	8.2 ± 0.93	6.6 ± 0.25
Dorsolateral striatum
Left	8.7 ± 1.8	6.2 ± 1.6	4.9 ± 0.5	4.1 ± 0.56	4.0 ± 0.64	5.2 ± 0.66	5.0 ± 0.20
Right	8.9 ± 1.6	6.5 ± 1.7	4.8 ± 0.7	4.4 ± 0.71	4.2 ± 0.79	4.7 ± 0.52	5.3 ± 0.15
Frontal motor cortex
Left	14.8 ± 2.2	8.5 ± 2.3	7.4 ± 0.83	6.6 ± 1.3	4.7 ± 0.83	8.2 ± 0.82	5.3 ± 0.24*
Right	14.6 ± 2.2	8.6 ± 2.4	7.6 ± 0.86	6.3 ± 1.2	4.7 ± 0.82	8.0 ± 0.78	6.9 ± 0.36
Somatosensory cortex
Left	15.1 ± 2.5	8.6 ± 2.4	7.1 ± 1.2	6.4 ± 1.5	4.8 ± 0.89	8.3 ± 0.80	5.4 ± 0.41*
Right	14.8 ± 2.6	8.8 ± 2.6	7.4 ± 1.1	6.0 ± 1.1	5.0 ± 0.97	8.0 ± 0.83	7.0 ± 0.34
Piriform cortex
Left	11.4 ± 2.0	6.7 ± 1.7	6.5 ± 0.88	4.7 ± 0.78	4.3 ± 0.76	5.6 ± 0.58	5.6 ± 0.23
Right	11.1 ± 2.2	7.1 ± 1.9	6.3 ± 0.84	4.4 ± 0.55	4.5 ± 0.76	5.7 ± 0.68	5.9 ± 0.25
Ventromedial striatum
Left	9.3 ± 1.7	6.6 ± 1.7	5.0 ± 0.70	4.5 ± 0.61	4.2 ± 0.81	5.3 ± 0.63	5.2 ± 0.26
Right	9.4 ± 1.6	6.6 ± 1.7	5.0 ± 0.84	4.4 ± 0.80	4.3 ± 0.78	5.1 ± 0.54	5.5 ± 0.25

Brain sections were collected from the level of impact (level i in Materials and Methods) and treated as described. Data show the mean ± SEM of density of specifically bound radioactivity in nCi/mg in regions in the uninjured (right) and injured (left) hemispheres. ANOVA with repeated measures showed a significant reduction (P < 0.05) in NMDA receptor density of all regions at all time points when compared to sham, with the exception of the corpus callosum at 7 days (P = 0.088). Significance is defined as P < 0.05 by ANOVA followed by post hoc PLSD test.

^*Left (injured) lower than right, P < 0.05.

Effects of NMDA Treatment on Neurological and Cognitive Recovery After CHI. The above observations led us to speculate that starting as early as a few hours after CHI, the injured animals are no longer likely to be in a hyperexcited state, because glutamate levels are reportedly no longer higher than normal (2-6, 9, 10), whereas NMDAR appear to be significantly hypo functional, possibly due at least in part to inhibition by an endogenous factor (30, 47, 48). If such a blockade as well as frank loss of receptors were indeed contributing to the reduction in NMDAR binding and the neurological and cognitive deficits observed at these times, it could conceivably be overcome by increasing the levels of agonist stimulation through administration of a nontoxic dose of NMDA (49, 50). To test this hypothesis, we evaluated the effects of NMDA on motor and cognitive function after CHI.

To achieve groups of animals with comparable trauma, the NSS was initially evaluated 1 h after CHI and animals randomized into three groups with similar mean initial NSS (6.14 ± 0.14, 6.4 ± 0.2, and 6.3 ± 0.3), then assigned to receive vehicle, NMDA, or NMDA+MK801, respectively. NSS just before treatment initiation (24 h) was also similar in the three groups. Neurological recovery 7 and 14 days postinjury was significantly affected by NMDAR activation (χ² of 12.9 and 15.4, df of 2 and 2, and P = 0.002 and P < 0.0001 at 7 and 14 days postinjury, respectively). Comparison of the individual groups showed that recovery was significantly accelerated in the NMDA-treated mice when compared to saline 7 and 14 days postinjury (P = 0.05 and 0.016, respectively; Fig. 2). Coadministration of the antagonist completely reversed the beneficial effect of the agonist (Fig. 2). Recovery in the NMDA+MK801 group was not only significantly worse than in the NMDA-alone group (P = 0.001 and 0.0001 at 7 and 14 days, respectively), it was also significantly worse compared to the vehicle-treated mice (P = 0.017 and 0.05 at 7 and 14 days, respectively; Mann-Whitney test; Fig. 2).

Fig. 2.

Activation of NMDAR through systemic agonist administration improves neurological recovery after CHI. Mice subjected to CHI were treated with NMDA, NMDA+MK801, or vehicle and dNSS calculated as described in Materials and Methods. dNSS was significantly higher in the NMDA-treated animals (top curve, open symbols) compared to vehicle-treated CHI mice (middle curve) at 7 days (P = 0.05) and 14 days (P = 0.016) by the Mann-Whitney nonparametric t test. Administration of MK801 1 mg/kg in addition to NMDA obliterated the benefi-cial effect of the agonist (bottom curve). NMDA+MK801 animals had signifi-cantly lower dNSS values (P < 0.001) at 7 (P = 0.001) and 14 (P < 0.0001) days compared to NMDA alone. Recovery was also significantly worse when compared to vehicle at 7 (P = 0.017) and 14 (P = 0.05) days.

NMDA-treated animals also performed significantly better than vehicle-treated controls in the object recognition test 14 days after CHI, at a dose which had no effect on performance in naïve animals. All four groups of mice (naïve, naïve+NMDA, CHI, and CHI+NMDA) spent a similar proportion (≈50%) of time exploring two objects in an observation cage at baseline (Fig. 3). Four hours later, the mice were reintroduced into the cage in which one of the two “old” objects was replaced by a novel object. The vehicle-treated CHI mice spent a similar proportion of their time with the old and new object (Fig. 3), with no significant preference for the novel object, whereas the NMDA-treated injured mice significantly increased their exploration of the novel object compared to the familiar object (Fig. 3), to the same level (≈70%) as intact untreated mice and intact mice treated with NMDA.

Fig. 3.

Activation of NMDAR through systemic agonist administration improves cognitive performance in the object recognition test after CHI. Results are means and standard errors of eight to nine animals per treatment group. Mice were subjected to the object recognition test 14 days after CHI. Injured, vehicle-treated animals (filled bars) lost the ability to recognize the new object, whereas NMDA-treated CHI animals (gray bars) spent a significantly higher percentage of their exploration time near the novel object (^*, P < 0.0001; Student's t test) and were indistinguishable from intact untreated animals (open bars). NMDA at the dose administered in this study had no effect on the performance of intact animals (hatched bars).

Discussion

The results of the above studies shed light on several persistent inconsistencies regarding the role of glutamate in the long-term outcome of brain injury. Traumatic and ischemic brain injuries, although different in etiology and in some clinical manifestations, have both been shown to trigger increased release of excitatory amino acid neurotransmitters in animals and in man. This increase was described as relatively short-lived in several different animal models of ischemic and traumatic brain injury (2, 3, 4, 9, 10), whereas the duration in humans is controversial. Some studies report elevations of extracellular glutamate for days after injury, and others report normalization within 6 h (5, 6). These observations led to the assumption that human TBI and stroke patients would require prolonged (3-7 days) NMDAR blockade to achieve therapeutic efficacy, despite the fact that animal models clearly showed loss of efficacy within 1 h postinjury (11-15). However, for released glutamate to have a calcium influx-dependent deleterious effect on neuronal survival, it needs to activate its receptors. The extent and duration of NMDAR activation in animal and human posttraumatic or postischemic brains have not been reported to date. Using modifications of the standard autoradiographic procedure for measuring NMDAR total density (29, 30, 36) that were designed to preserve the endogenous milieu [i.e., omission of tissue washing and exogenous glutamate and glycine (30)], we have been able to measure the density of endogenously activated (open channel) NMDAR with radioactive MK801. CHI did indeed increase the density of MK801 binding under these conditions. However, the increase was short-lived, corresponding to the reported therapeutic window of NMDAR antagonists in models of brain injury, and region-specific, being most pronounced in the hippocampus, a region subserving memory functions. Regions at the level of impact had decreased, rather than increased, MK801 binding even at 15 min postinjury, probably reflecting an earlier onset and even shorter (<15 min) duration of activation at the site of impact. To our knowledge, this is the first report of acute (<1h) changes in MK801 binding in the posttraumatic brain, and we also suggest that these early changes are largely reversible, because washing the brain sections in buffer for 60 min at room temperature, a procedure apt to remove endogenously released glutamate, reversed the effect (Fig. 4, which is published as supporting information on the PNAS web site).

Sixty minutes postinjury, we observed a significant bilateral decrease in activated NMDAR at the level of impact and at more posterior levels. Persistent and progressive decreases were observed within the first 24 h after the injury. These reductions were only partially reversible by washing the tissue sections for 60 min at room temperature (Fig. 5, which is published as supporting information on the PNAS web site), a procedure apt to remove not only glutamate but also endogenous, water-soluble inhibitors of NMDAR channels as described by other investigators (30, 47, 48). The earliest decreases in functional NMDAR described in this study are compatible with a CHI-induced desensitization and early increase in the level or activity of this inhibitory factor, which does not appear to be magnesium (47), and may constitute one of the first lines of defense mounted by the brain against the excessive stimulation of NMDAR after injury. The progressive decrease in functional NMDAR between 1 and 24 h postinjury most likely reflects contributions from additional mechanisms, such as reductions in NMDAR density and gene expression, which have been reported to occur within a few hours and last up to several days after different models of inflammatory, ischemic, or traumatic brain injury (34, 51-53).

In accordance with our hypothesis, we found that administration of the NMDAR agonist NMDA to animals with CHI 1 and 2 days after the injury significantly improved general neurological and cognitive function assessed 2 weeks postinjury. Moreover, coadministration of MK801 and NMDA resulted not only in abrogation of the beneficial effects of NMDA, but in prolongation and aggravation of neurological deficits seen in injured, vehicle-treated mice, probably because of blockade of endogenous glutamate as well as the administered agonist. Similarly, Barth and colleagues (54) have demonstrated a long-lasting (7 days) exacerbation of brain injury-related deficits induced by a delayed administration of MK801: although early administration of MK801 was beneficial (54), as reported also in other injury models including ours (11-15), delaying the initiation of MK801 administration for a couple of days, until the animals achieved some spontaneous recovery, resulted in an increase in the severity of deficits (54).

We also show here that the beneficial effects of NMDA treatment outlast the treatment period by more than a week, suggesting that the accelerated recovery is not dependent on chronic exposure to the agonist. Several research groups have demonstrated similar long-lasting beneficial effects of amphetamine on functional recovery from brain injury in animals and Man (55-57), suggesting a common mechanism related to improvement in blood flow and facilitation of repair mechanisms.

The effect of CHI on working memory, assessed by the object recognition test (40), appears to be both pervasive and persistent, because at 14 days postinjury all of the vehicle-treated CHI mice showed a profound deficit in task performance. Recently, it was shown that performance of the object recognition task can be disrupted by blockade of NMDAR in the dorsal hippocampus (46). Our results demonstrating a performance deficit in CHI animals with reduced NMDAR in the hippocampus and reversal of this deficit by treatment with the agonist NMDA are consistent with these findings and suggest that TBI related cognitive deficits might be especially responsive to NMDAR activation within a relatively wide time window after injury.

Conclusion

Taken together, our findings strongly suggest that administration of NMDAR blockers to brain-injured subjects beyond the acute postinjury period is not only useless (as has been shown by both animal and clinical studies) but actually counterproductive (as suggested by the results of clinical trials). Furthermore, brain-injured subjects may stand to benefit from stimulation of NMDAR in the subacute postinjury phase, although the duration of this period needs to be defined in humans, possibly through the use of new NMDAR open channel imaging agents (33, 58).