Vulnerability of white matter towards antenatal hypoxia is linked to a species-dependent regulation of glutamate receptor subunits

Abstract

White-matter damage is a leading cause of neurological handicap. Although hypoxia-ischemia and excitotoxicity are major pathogenic factors, a role for genetic influences was suggested recently. Thus, protracted gestational hypoxia was associated with white-matter damage (WMD) in rat pups but not in mouse pups. Indeed, microglial activation and vessel-wall density on postnatal days (P)1 and P10 were found increased in both mouse and rat pups, but cell death, astrogliosis, and myelination were only significantly altered in hypoxic rat pups. We investigated whether this species-related difference was ascribable to effects of antenatal hypoxia on the expression of glutamate receptor subunits by using immunocytochemistry, PCR, and excitotoxic double hit insult. Quantitative PCR in hypoxic mouse pups on P1 showed 2- to 4-fold down-regulation of the AMPA-receptor subunits -1, 2, and -4; of the kainate-receptor subunit GluR7; and of the metabotropic receptor subunits mGluR1, -2, -3, -5, and -7. None of the glutamate-receptor subunits was down-regulated in the hypoxic rat pups. NR2B was the only NMDA-receptor subunit that was down-regulated in hypoxic mice but not in hypoxic rat on P1. Ifenprodil administration to induce functional inhibition of NMDA containing NR2B-subunit receptors prevented hypoxia-induced myelination delay in rat pups. Intracerebral injection of a glutamate agonist produced a larger decrease in ibotenate-induced excitotoxic lesions in hypoxic mouse pups than in normoxic mouse pups. Gestational hypoxia may regulate the expression of specific glutamate-receptor subunits in fetal mice but not in fetal rats. Therefore, genetic factors may influence the susceptibility of rodents to WMD.

Periventricular (P) WMD and subsequent cortical damage are the leading causes of cerebral palsy (CP) limited to preterm birth (1, 2). Animal models have been developed to unravel the mechanisms underlying these brain lesions. Factors that seem involved in the pathophysiology of CP in these models include hypoxia and ischemia, infection and inflammation, excitotoxicity, accumulation of reactive oxygen species, and deficiencies in growth factors (3, 4). These factors seem to act in combination to cause damage to the developing white matter.

Glutamate accumulation may be a mechanism common to many risk factors for CP. Glutamate, the major excitatory neurotransmitter, acts by means of several groups of receptors, namely, NMDA, AMPA, kainate, and metabotropic receptors (mGluRs). Excessive activation of glutamate receptors may cause cell vulnerability, in part as a result of intracellular calcium influx (5, 6). Intracerebral injection of glutamate agonists into the neocortex and white matter of newborn rodents produces histological lesions that mimic the brain damage observed in preterm neonates (7–10).

Several studies suggest that genetic factors may influence the susceptibility of very preterm infants to PWMD. The occurrence of CP may depend, at least in part, on polymorphisms in genes for TNF-α and mannose-binding lectin (11–13). The genetic regulation of glutamate-receptor expression also has a major role throughout development and responds to changes in the cerebral environment (14, 15).

We recently showed that protracted prenatal hypoxia in rats caused dramatic WMD in the pups, with microgial activation and oligodendrocyte death leading to deficient myelination (16). To investigate a potential role for genetic factors, we compared brain damage in mouse and rat pups subjected to similar levels of antenatal hypoxia. Damage was considerably milder in the mouse pups than in the rat pups. Mice and rats descend from a common ancestor and share many reproductive, developmental, morphological, and anatomical similarities, despite substantial genomic differences. Therefore, we investigated whether the effects of gestational hypoxia on the regulation of glutamate-receptor subunits differed between mice and rats. We found that antenatal hypoxia affected the subunit composition of glutamate receptors in mice, thereby reducing brain-damage severity. This effect was not found in rats. Among NMDA receptor subunits, NR2B was found to have a crucial role in the difference between mice and rats regarding susceptibility to hypoxia-induced WMD.

Results

Gestational Hypoxia Leads to Inflammation Without White-Matter Damage in Mouse Pups but Causes WMD in Rat Pups.

Pups were studied on 3 time points, synchronizing brain development stages between the 2 species according to the cortical myelin density, the principal endpoint in this study (see Materials and Methods).

We first investigated whether protracted gestational hypoxia produced similar brain abnormalities in mouse and rat pups. No difference in body temperature was found between mouse pups subjected to prenatal hypoxia and in controls (data not shown). Hematoxylin-eosin staining of sections from hypoxic and normoxic mouse pups showed comparable findings, with no white-matter lesions. In contrast, white-matter lesions were seen in rat pups (data not shown). Vessel-wall density in white matter as assessed by using glucose-transporter-1 (GLUT1) immunostaining was significantly increased in rats [supporting information (SI) Fig. S1A] and mice (Fig. S2A and Fig. S3 A and B) during the first week of life; density was normal on P10 in mice and P14 in rats (data not shown). Cingular cortex and basal ganglia from mouse pups exhibited a similar transient increase in neonatal angiogenesis, suggesting that gestational hypoxia led to fetal hypoxemia and, therefore, that absence of fetal hypoxemia was not the reason for absence of WMD in mouse pups. Tomato lectin (TL) immunostaining produced similar findings with an increased density of activated macrophages in the white matter on both hypoxic rat and mouse pups (Figs. S1B, S2B, and S3 C–F). Gestational hypoxia in mice was not associated with increased cell death in the cortex or white matter (Fig. S2C). Conversely, hypoxia in rats was associated with increased cell death in the white matter at the early stage (Fig. S1C).

Glial Phenotype After Gestational Hypoxia Differs Between Mouse and Rat Pups.

Astrogliosis as assessed by using GFAP immunostaining was significant in cingular white matter from intermediate and late stages rat pups subjected to gestational hypoxia (Fig. S1D). In contrast, no significant astrogliosis was detected in cingular white matter from hypoxic mouse pups at the same stages (Fig. S2D). Similarly, no difference in astrocyte density was noted throughout the cortex and basal ganglia obtained from hypoxic mouse pups, compared with controls (data not shown). White-matter myelination evaluated at the intermediate and late stages by using MBP immunohistochemistry was normal in hypoxic mouse pups (Fig. 1 A–E), but altered in hypoxic rat pups (Fig. 1 F–J). We failed to demonstrate any difference between male and female both in myelination defect observed in rat pups subjected to prenatal hypoxia. Similar findings were obtained by quantitative real-time PCR for MBP transcripts (Fig. S4). These findings were observed in cingular white matter, external capsule or lateral corpus callosum. Fig. 2 compared the panel of cellular markers assessed between rat and mouse pups normalized to respective controls. We found a statistically significant difference in astrogliosis and myelination defect between the 2 species emphasizing different phenotypes in mouse and rat pups exposed to hypoxia during gestation.

Fig. 1.

Gestational hypoxia induces a myelination defect in rats but not in mice. (A) OD quantification of MBP staining in the lateral corpus callosum from control and hypoxic mouse pups at the intermediate and late stages. (B and C) Microphotographs showing MBP-immunoreactive oligodendrocytes on coronal sections from control (C) and hypoxic (D) mouse pups at the late stage. The Insets delineate the regions where OD was measured. (D and E) Microphotographs at higher magnification from control (E) and hypoxic (F) mouse pups at the late stage. (F) OD quantification of MBP-staining in the lateral corpus callosum from control and hypoxic rat pups at the intermediate and late stages (**, P < 0.01). (G and H) Microphotographs showing MBP-immunoreactive oligodendrocytes on coronal sections from control (I) and hypoxic (J) rat pups at the late stage. The Insets delineate the regions where OD was measured. (I and J) Microphotographs at higher magnification from control and hypoxic rat pups at the late stage.

Fig. 2.

Comparison of cellular markers between mouse and rat brains compared with respective controls. Quantification of GLUT1-immunoreactive vessel walls, TL-immunoreactive microglial cells, GFAP-immunoreactive astrocytes, and MBP optical density in the WM of hypoxic rat pups at different stages, normalized to normoxic controls. Statistical analyses show significant differences in astrogliosis and myelination defects between the 2 species (*, P < 0.05 and ***, P < 0.001).

Expression of Specific Glutamate-Receptor Subtypes Is Down-regulated in Mouse Pups Exposed to Gestational Prenatal Hypoxia.

Because excitotoxicity and genetic regulation of glutamate-receptor expression are known to have a key role in brain damage (8, 14, 17, 18), we investigated whether glutamate-receptor expression was altered by hypoxia, and whether these alterations differed between rats and mice, potentially explaining the phenotypic differences noted between these 2 species. We assessed the expression of glutamate-receptor subtypes in rat and mouse at early, intermediate, and late stages of brain development. Quantitative real-time PCR in hypoxic mice at the early stage demonstrated a significant 2- to 4-fold down-regulation of the AMPA-receptor subunits GluR1, -2, and -4, the kainate receptor subunit GluR7, and the metabotropic receptor subunits mGluR1, -2, -3, -5, and -7 in at least 6 animals in 2 separate experiments, compared with normoxic controls (Table S1). Testing at later time points showed that the main differences were largest during the first few days after birth and resolved by the intermediate stage. Only metabotropic receptor subunit mGluR1 was down-regulated at the early and the late stage and kainate receptor KA2 at the late stage. In contrast, gestational hypoxia in mice induced no changes in the expression of the AMPA/kainate subunits GluR3, -5, and -6 or in the metabotropic receptor subunits mGluR4 and -6. In hypoxic rats, at all stages, none of the AMPA, kainate, or metabotropic receptor subunits showed differences in expression levels compared with control rats (data not shown).

Among the NMDA receptor subtypes, NR2B was the only down-regulated subunit at the early stage in mice (Table S1 and Fig. 3A). The NR2C subunit was down-regulated only at the later stage. In rats, none of the NMDA receptor subunits was down-regulated (data not shown), even NR2B at the early stage (Fig. 3B), when most of the WMD occurred. No difference was observed between male and female pups. We also performed analysis of each NMDA receptors subunits, at the 3 developmental stages and in 3 areas of the developing brain, including motor (M), somatosensori (SS), and visual (V) cortices. These data did not shown any difference in rat pups subjected to antenatal hypoxia compared with the controls (Fig. S5).

Fig. 3.

NR2B subunit ontogenic neocortical pattern of mRNAs, Western blot analysis of NR2B protein expression, and ³H-MK-801 binding in control and hypoxic mouse and rat pups. (A and B) Expression of NR2B subunits mRNAs in control and hypoxic mouse (A) and rat (B) pups cortex including white matter at the 3 stages (**, P < 0.01). (C) NR2B protein expression in control and hypoxic mouse and rat pups at the early stage. NR2B protein expression was down-regulated in hypoxic mouse pups but not in hypoxic rat pups. (D) H-MK-801 binding in control and hypoxic mouse pups at the early and intermediate stages.

Western blot analysis for NR2B protein at the early stage also confirmed down-regulation at plasmatic membranes in hypoxic mouse pups (Fig. 3C). Performing similar analysis for NR2B and NR1 subunits failed to demonstrate any down-regulation in hypoxic rat pups at any developmental stage (Fig. 3C and Fig. S6).

To obtain further evidence that changes in glutamate-receptor expression were involved in the ability of mouse pups to withstand gestational hypoxia, we measured NMDA-receptor density by using binding of tritiated dizocilpine (³H-MK-801) in membrane-enriched cortex preparation from mouse pups at the early and intermediate stage. No difference was found between the 2 groups at either age in 2 separate experiments (Fig. 3D). This result suggests that the role for antenatal hypoxia in modulating WMD may involve qualitative rather than quantitative regulation of the ability of glutamate to bind to NMDA sites.

Functional Inhibition of NR2B Receptors Prevented Delayed Myelination Induced by Hypoxia in Rat Pups.

Because NR2B was the only NMDA-receptor subunit that showed early down-regulation in response to gestational hypoxia in mice, we investigated the potential role for NR2B in the myelination defect seen in rat pups subjected to prenatal hypoxia. Rats were given i.p. ifenprodil hemitartrate (2.5 mg/kg; Tocris), from embryonic day (E)19 to P3, to inhibit heteromeric NMDA receptors containing NR2B subunits. Ifenprodil fully reversed the myelination defect in hypoxic rats, which exhibited similar white-matter myelination to that seen in mice (Fig. 4), in both the lateral corpus callosum and the cingular white matter.

Fig. 4.

Ifenprodil administration reverses the myelination defect in rats at the late stage. OD quantification of MBP staining in the lateral corpus callosum from late stage control and hypoxic rat pups treated with ifenprodil (ifen) (***, P < 0.01).

Gestational Hypoxia Prevented Excitotoxic-Induced Lesions in Mouse Pups.

We investigated the impact of NR2B-subunit down-regulation on the response of the developing brain to an excitotoxic challenge. Ibotenate, a glutamate agonist that binds to NMDA and metabotropic subtypes, was injected intracerebrally to mouse and rat pups on P2. The size of the resulting excitotoxic lesion in the white matter and cortical plate was measured on P6. In both the white matter and the cortex, the size of ibotenate-induced lesions was significantly reduced in hypoxic mouse pups (either male or female), compared with normoxic mouse pups, but no significant reduction in lesion size was noted in the hypoxic rat pups (Fig. 5). At a later time point (P21), we found that this preconditioning effect has been lost in mice (Fig. S7A). Comparison (normalized to controls) of ibotenate-induced lesion size in cortex and white matter between hypoxic rat and mouse pups was shown in Fig. 5. This preconditioning effect of antenatal hypoxia was not observed in hypoxic rat pups but was again observed in both cortex and white matter when rat pups were injected with ifenprodil. This result suggests that NR2B down-regulation (in hypoxic mice) or inhibition in (treated rats) was, at least in part, responsible for the species-dependent effect of antenatal hypoxia on excitotoxic-induced brain lesion.

Fig. 5.

Hypoxia protects newborn mice, but not newborn rats, from excitotoxic-induced brain lesion. Gestational hypoxia reduced lesion size both in the cortex and WM in mice given intracerebral ibotenate (Ibo). Gestational hypoxia did not reduce lesion size in rats given intracerebral ibotenate but added to i.p. injections of ifenprodil (ifen), gestational hypoxia reduces ibotenate-induced cortical lesion in neonatal rats (*, P < 0.05; **, P < 0.01; and ***, P < 0.001 compared with normoxic controls; §, P < 0.05).

NMDA intracerebral injection in mouse demonstrated a nonsignificant reduction in lesion size of both cortical and WMD (Fig. S7B). These data suggest that antenatal hypoxia-induced changes in metabotropic receptors expression, shown in Table S2, should have an additional role in the preconditioning effect of antenatal hypoxia.

Discussion

In this study, we investigated the mechanism by which protracted gestational hypoxia results in milder brain damage in mice than in rats. Gestational hypoxia was associated with increased angiogenesis and an inflammatory response in mouse pups during the first postnatal week. Surprisingly, mouse pups exhibited no increase in cell death leading to histological lesions, no astrogliosis, and no myelination defect, all of which were found in rat pups. This difference between these 2 species suggests that specific animal models may be relevant to delineate discrete mechanisms involved in PWMD. Mice and rats have significant genomic differences between each other and with humans, and it has been shown that genetic factors may influence the susceptibility to PWMD and subsequent CP in very preterm infants (11, 12, 17). Specific genetic polymorphisms for endothelial nitric oxide synthase (eNOS) or plasminogen activator inhibitor-2 in human infants may contribute to spontaneous preterm birth followed by a diagnosis of CP. CP was associated with the beta-2 adrenergic receptor gln27glu in white non-Hispanic children and with the plasminogen activator inhibitor-1 in Hispanic children (17, 18). Similarly, common genetic variants in genes encoding pro or antiinflammatory cytokines may influence the risk for spontaneous preterm birth. To our knowledge, no evidence has been reported that the genetic background may influence animal models of WMD. Results reported by the Rat Genome Sequencing Consortium in 2004 suggest that several aspects of genome evolution might be accelerated in rats compared with mice. Rats have a number of genes that are not found in mice, including genes involved in immunity, the production of pheromones, the breakdown of proteins, and the detection and detoxification of chemicals. Also, ≈50 chromosomal rearrangements occurred in each of the rodent lines after divergence from their common ancestor. However, ≈90% of rat genes have counterparts in the mouse and human genomes; and the rat, mouse, and human genomes encode similar numbers of genes (19–23). Therefore, we suggest that differences in genetic background between mice and rats may be one possible explanation for the neuroprotective effect of hypoxia in mice but not in rats.

Here, we show that one explanation for the decreased susceptibility to hypoxia-induced WMD in mice, compared with rats, may consist in down-regulation of specific glutamate-receptor subunits, able to decrease the vulnerability to excitotoxic insults. Several lines of evidence point to a role for neuronal glutamate receptors in many neurological disorders in neonates and adults (5, 8, 24–27). Here, we documented down-regulation just after birth of the AMPA-receptor subunits 1, 2, and 4; of the kainate-receptor subunit 7; and of the metabotropic receptor subunits 1, 2, 3, 5, and 7 in hypoxic mouse pups but not in hypoxic rat pups. Among NMDA-receptor subunits, NR2B was markedly down-regulated on P1 in mice, and the subsequent changes in glutamate-receptor expression were consistent with previous data (28). Such transient change may severely impact cortex and white-matter development in the brain challenged by adverse events occurring during perinatal period (1, 16). Over the last few years, an increasing number of studies have established that NR2B is involved in various of synaptic signaling events and protein–protein interactions. Activation of NR2B-subtype NMDA receptors triggers various of disorders such as schizophrenia, Parkinson's disease, Alzheimer's disease, and Huntington's disease (29). NR2B is expressed during processes involving oligodendrocytes, and NMDA receptors are extremely sensitive to ischemia (25). Also, activation of NR2B-containing NMDA receptors results in excitotoxicity, whereas activation of NR2A-containing NMDA receptors promotes neuronal survival (15). In accordance with these findings, NR2B was the only NMDA subunit that showed early significant down-regulation in mouse pups subjected to gestational hypoxia. Also, ifenprodil completely reversed the myelination deficiency in hypoxic rat pups, suggesting that NR2B may have a key role in hypoxia-induced WMD. The promoters of genes that encode the NR1, NR2B, NR2C, GluR1, GluR2, and KA2 subunits share several characteristics, including multiple transcriptional start sites and neuron-selective expression. In most cases, the promoter regions include elements that guide expression in neurons (14). The 5′ promoter region of mouse NR2B has been cloned by 2 independent groups (30, 31). Here, genomic DNA from mouse and rat-tail samples was amplified, and NR2B promoter in each species was sequenced to identify regulatory elements and factors that may control NR2B expression in the brain (Fig. S8). Direct sequencing identified a difference in a neuron cis-acting element between rats (CCAGGAG) and mice (CGAGAGA) (Fig. S8 Inset 1). This element may have a role in combination with other cis-acting elements to confer neuron-specific gene expression (30, 32). Therefore, we suggest that the mutation in this motif in rat NR2B promoter may contribute to explain the transcriptional difference between rats and mice after gestational hypoxia. Conversely, no difference was detected between the 2 species in a sequence known as the neuron-restrictive silencer element (RE1 or NRSE; Fig. S8 Inset 2), which represses transcription by associating with a repressor protein called REST or NRSF. This result suggests that a potential regulatory phenomenon in NR2B promoter may be independent from REST in our model of gestational hypoxia.

Last, we demonstrated that prenatal hypoxia was neuroprotective against cortical and white-matter lesions induced by intracerebral ibotenate injection on P2 in mouse pups but not in rat pups. In various animal models, a short period of hypoxia (preconditioning) may prevent ischemia- or stroke-related brain injury (33–35). To our knowledge, we provide the first evidence prenatal hypoxic preconditioning induces down-regulation of specific glutamate-receptor subunits, leading to a neuroprotective effect also detectable after pharmacological inhibition of NR2B subunit.

In conclusion, this study documented a previously undescribed biological effect of gestational hypoxia on glutamate-receptor regulation in mice. It establishes that genetic factors affect the susceptibility of the developing white matter to injury, in addition to other recognized risk factors. Further studies are needed to identify glutamate-receptor promoter fragments that may be candidates for designing gene-targeting constructs to deliver transgenes and other therapeutic agents to the developing brain.

Materials and Methods

Animal Models.

All experimental protocols and animal housing procedures complied with Institut National de la Santé et de la Recherche Médicale guidelines and with the Policies on the Use of Animals and Humans in Neuroscience Research (revised and approved by the Society for Neuroscience in January 1995).

Gestational hypoxia.

Pregnant Sprague–Dawley rats and Swiss, BALB/c/j, DBA/2, or CBA mice (Janvier) were placed in normoxic or hypoxic (10% O₂-90% N₂), gas chambers (Biospherix), and monitored as previously described (16). Hypoxia exposure lasted from E5 to E19 in rats and from E5 to E18 in mice. After delivery, normoxic (control) and hypoxic pups were studied on 3 time points, considered to better synchronize brain development stages between the 2 species, according to the cortical myelin density: P1 (early), P7 (intermediate), and P14 (late) for rat pups and P1 (early), P10 (intermediate), and P21 (late) for mouse pups.

Excitotoxic brain lesions.

Ibotenate (Tocris) is a glutamate analogue that activates both NMDA and metabotropic receptors; it does not activate AMPA or kainate receptors. Ibotenate (5 μg) or NMDA (2 μg) was injected intracerebrally on P2 or P21 to mouse and rat pups, as previously described (7, 8, 36–38). At least 12 animals of each treatment group were killed by decapitation 4 days after the injection (i.e., on P6 or P25), and the brains were processed as previously described (37). In all of the experiments described below, 2 investigators blinded to treatment group determined the size of the lesion in each pup.

Immunohistochemistry.

For the detection of myelin basic protein (MBP) detection, brains from rats pups killed at the intermediate or the late stage after gestational exposure to normoxia or hypoxia were fixed in 4% formaldehyde for 5 days. After embedding in paraffin, 10-μm sections were cut coronally from the frontal to the occipital pole. Sections were incubated with primary antibody diluted 1:1,000 (Sigma). Labeled antigens were detected by using the streptavidin-biotin peroxidase complex (Amersham) containing the chromogen diaminobenzidine.

Normoxic or hypoxic mouse and rat pups were perfused transcardially with 4% paraformaldehyde in phosphate buffer (0.12 M, pH 7.4) under inhaled isoflurane anesthesia (Abbott). The brains were postfixed, cryoprotected in sucrose, frozen, and cut into coronal sections 10 μm in thickness. The sections were incubated overnight with primary antibodies (Table S2). The primary antibodies were visualized by using streptavidin-biotin-peroxidase complex as described above. At least 6 animals were included in each experimental group. Experiments were repeated 3 times. Adjacent sections were used as negative controls, in which incubation with the primary antibody was omitted. Two investigators blinded to treatment group independently estimated the number of positive cells seen in cross-sections in an area measuring 0.0625 mm² (magnification, 40×) or 1 mm² (magnification, 10×).

Quantification of Myelination.

The OD of MBP-immunoreactive fibers was measured in the cingulum of coronal sections from mice and rats at the later stage (+2.16 to −0.36 mm from the bregma). OD was measured at 100× magnification by using image-analysis software (ImageJ, http://rsb.info.nih.gov/ij/) that read OD as gray levels, calculated the area and pixel value statistics of user-defined selections, and generated density histograms. Nonspecific background densities were measured in a region devoid of MBP immunostaining at each brain level and were subtracted from the cingulum values. OD was analyzed in at least 10 animals per group (normoxic or hypoxic) from 3 separate experiments and in at least 4 sections per brain.

Quantitative Real-Time PCR.

DNA-free total RNA from normoxic and hypoxic brain cortex including white matter or from M, SS, or V cortices was obtained by using a protocol adapted from Chomczynski and Sacchi (39). Primers for real-time RT-PCR were designed by using M-fold and Oligo6 software, based on published cDNA sequences encoding mouse and rat glutamate receptors and the myelination marker MBP. The nature of the amplified DNA was confirmed by sequencing. To standardize gene expression across samples, we first compared the expression levels of 4 well known housekeeping genes [GAPDH, β2-microglobulin, hypoxanthine-guanine phosphoribosyltransferase (HPRT), and β-glucoronidase] within the samples. For reverse transcription, we used 600 ng of total RNA and the Iscript cDNA synthesis kit (Bio-Rad). Real-time PCR was set up by using sybergreen-containing supermix (Bio-Rad) for 50 cycles with a 3-step program (25-sec denaturation at 96°C, 30-sec annealing at 60°C, and 30-sec extension at 72°C). Amplification specificity was assessed by melting curve analyses. Each experiment was run twice with a least 6 animals per group, and in both cases, samples were assessed in triplicate.

Binding.

NMDA receptor density was evaluated based on the binding of tritiated dizocilpine (³H-MK-801; Perkin–Elmer) in membrane-enriched cortex preparation (40). Briefly, ³H-MK-801 (2.5–17 nM) was incubated for 2 h at room temperature with 150–200 μg of proteins in 0.5 ml of Hepes (10 mM, pH 7.4) buffer containing L-glutamate (100 μM) and glycine (30 μM). Nonspecific binding was evaluated by adding 30 μM of unlabeled MK-801 (ICN). Incubations were stopped by rapid filtration through glass-fiber filters (GF/B; Whatman, Sigma). Binding capacity and K_d were estimated by Scatchard analysis in 2 separate experiments.

Western Blot Analysis.

Membrane proteins were extracted from forebrain cortex, including white matter, taken from mouse and rat pups. Extraction was achieved by homogenization in Hepes buffer containing protease inhibitors from Sigma, according to the manufacturer's instructions. We loaded 50 μg of protein from each sample and membranes were incubated overnight with a specific polyclonal antibody against NR2B or NR1 (Alomone Labs) diluted 1:200 or with an anti-β-actin goat antibody (Santa Cruz Biotechnology) diluted 1:10,000. Western blot analysis experiments were run in triplicate.

Statistical Analysis.

The data were analyzed by using Student's t test or one-way ANOVA. When a main effect was significant by ANOVA, post hoc multiple-comparison tests were performed (Bonferroni, Dunnett, or Kruskal-Wallis; Graph-Pad Prism version 4.01 for Windows).