Developmental Expression of N-Methyl-d-Aspartate (NMDA) Receptor Subunits in Human White and Gray Matter: Potential Mechanism of Increased Vulnerability in the Immature Brain

Abstract

The pathophysiology of perinatal brain injury is multifactorial and involves hypoxia-ischemia (HI) and inflammation. N-methyl-d-aspartate receptors (NMDAR) are present on neurons and glia in immature rodents, and NMDAR antagonists are protective in HI models. To enhance clinical translation of rodent data, we examined protein expression of 6 NMDAR subunits in postmortem human brains without injury from 20 postconceptional weeks through adulthood and in cases of periventricular leukomalacia (PVL). We hypothesized that the developing brain is intrinsically vulnerable to excitotoxicity via maturation-specific NMDAR levels and subunit composition. In normal white matter, NR1 and NR2B levels were highest in the preterm period compared with adult. In gray matter, NR2A and NR3A expression were highest near term. NR2A was significantly elevated in PVL white matter, with reduced NR1 and NR3A in gray matter compared with uninjured controls. These data suggest increased NMDAR-mediated vulnerability during early brain development due to an overall upregulation of individual receptors subunits, in particular, the presence of highly calcium permeable NR2B-containing and magnesium-insensitive NR3A NMDARs. These data improve understanding of molecular diversity and heterogeneity of NMDAR subunit expression in human brain development and supports an intrinsic prenatal vulnerability to glutamate-mediated injury; validating NMDAR subunit-specific targeted therapies for PVL.

Introduction

Although neonatal care has dramatically improved the survival rates after premature birth, there still remains significant neurological morbidity associated with prematurity. Preterm deliveries make up more than 500 000 or ∼12.5% of all infant births in the United States (Berman and Stith Bitler 2007). Approximately 5–10% of infants born <32 weeks gestational age and weighing <1500 g (very low birth weight, VLBW) have major motor deficits and up to 60% have neurocognitive deficits and/or behavioral disabilities as long-term sequelae (Holsti et al. 2002; Woodward et al. 2005). A unique pattern of injury observed in more than 50% of VLBW infants is periventricular leukomalacia (PVL), involving focal periventricular necrotic lesions and more diffuse cerebral white matter damage with injury and dysfunction of premyelinating oligodendrocytes (Volpe 2008, 2009a, 2009b). More recent detailed pathologic and radiologic investigations have shown that this white matter injury is often associated with neuronal and axonal abnormalities in cortical and subcortical gray matter structures, a constellation termed encephalopathy of prematurity (EOP) (Volpe 2005, 2009a, 2009b). Currently, there is no cure for the brain injury associated with either PVL or EOP.

Compared with injury to the premature brain, hypoxic-ischemic (HI) insults occurring near term result in different neuropathological patterns, mortality rates, and long-term neurologic morbidity (Takenouchi et al. 2012). At term, brain injury due to HI is associated with marked neuronal damage, particularly within the cerebral cortex, thalamus, caudate putamen, hippocampus, and brain stem, although white matter injury also occurs (Miller et al. 2005; Miller and Ferriero 2009). Death or permanent disability is observed in 25% of cases of moderate encephalopathy and 76% of babies with severe encephalopathy exhibit severe deficits of cognition and significant impairments in quality of life (Ronen et al. 2007; Johnson et al. 2009; Modi et al. 2009). To date, no treatment has been developed for this injury. However, rodent studies implicate glutamate receptors as potential therapeutic targets (Jensen 2002).

While the pathophysiology of perinatal HI brain injury is very complex, one of the major mechanisms underlying the selective maturational, regional, and cell-type- specific vulnerability is excitotoxicity via the over activation of ionotropic glutamate receptors (Follett et al. 2000; Jensen 2002, 2005; Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006). Ionotropic glutamate receptors include the α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid glutamate receptors (AMPARs), N-methyl-d-aspartate receptors (NMDARs), and kainate receptors. Of particular importance is their high permeability to Ca²⁺ ions, which confers a central role in learning, memory, and synaptic plasticity under physiological conditions, and cell death under excitotoxic pathological conditions (Paoletti and Neyton 2006). Glutamate-mediated neurotransmission is required for normal brain development and injury response, and glutamate receptor function is tightly regulated under normal conditions for maintenance of homeostasis. Recent literature on plasticity and depression, schizophrenia and other affective disorders, indicates that both too little and too much activation of the NMDAR subtype can be detrimental (Parsons et al. 2007; Weickert et al. 2012). Further, alterations in NMDAR function may be a potential link between early life brain injury and later life neuropsychiatric disorders (Mednick et al. 1988; Fatemi and Folsom 2009; Weickert et al. 2012).

NMDARs are widely expressed throughout the central nervous system, and their number, localization, and subunit composition are strictly regulated and differ in a cell- and synapse-specific manner (Sanz-Clemente et al. 2013). Similar to AMPARs, the developmental expression of NMDARs on neurons and glia has been well documented in rodents, and we recently described the maturational profiles for the AMPAR subunits GluR1, GluR2, GluR3, and GluR4 in the perinatal human brain (Follett et al. 2000, 2004; Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006). To date however, there is still little information regarding the expression of NMDARs during the same window of development. This is a significant gap in knowledge, as the NMDAR may be a therapeutic target for perinatal brain injury. Like AMPAR subunits, brain region, developmental stage, and level of synaptic activity are some of the factors that regulate NMDARs and the expression of their individual subunits (Sanz-Clemente et al. 2013), and pharmacotherapies with use-dependent properties or action at specific subunits could have clinical potential. Specifically, NMDAR subunit composition determines numerous functional properties of the channel, such as Ca²⁺ permeability and conductance, and developmental changes in individual subunits may alter susceptibility to excitotoxic injury. Previously, we showed in a small number of cases that NMDARs are present on immature oligodendrocytes in the developing human brain during the window of peak vulnerability to PVL (Manning et al. 2008). Here, using a large number of human tissue cases, we examined the protein expression of 6 NMDAR subunits, NR1, NR2A, NR2B, NR2C, NR2D, and NR3A in normal developing human white and gray matter samples from 20 postconceptional weeks (PCW) through adulthood. We then compared the subunit expression patterns in these brains without injury to those observed in cases of confirmed PVL. We hypothesized that the specific vulnerability of the developing brain to injury caused by HI and the manifestation of PVL is in part mediated by excitotoxicity conveyed through maturation-specific NMDAR subunit composition. We predicted that NMDAR subunits expression would be developmentally regulated, with overall upregulation of all subunits prior to birth and expression patterns unique to white and gray matter. We sought to understand whether NMDAR subunit composition exclusive to restricted developmental periods of gestation and cell types render the immature brain more sensitive to excitotoxicity and convey vulnerability to PVL. These data significantly improves the understanding of the molecular diversity of NMDAR subunit protein expression changes in human brain development and alterations in cases of PVL.

Materials and Methods

Human Subjects

Brain specimens were collected prospectively from the fetal, neonatal, pediatric, and adult autopsy populations of the Boston Children's Hospital and Brigham and Women's Hospital. Additional cases were obtained from the University of Maryland NICHD Brain and Tissue Bank for Developmental Disorders. The samples were obtained from standard diagnostic postmortem examinations, and all procedures and experiments were conducted under guidelines approved by the Clinical Research Committee at each institution. The case demographics, causes of death, and relevant clinical and neuropathological data are listed in Tables 1 and 2. The age of the subjects ranged from 20 PCW to 63 years of age (n = 68; 42 males, 24 females, 2 unknown). The postconceptional age (PCA) was calculated as gestational age (GA) plus postnatal age (PNA). In fetuses, the age was determined by foot measurement, as previously described (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006; Jantzie et al. 2011). Control subjects fell into the following clinically relevant categories: midgestation (20–24 PCW, n = 13), preterm (25–37 PCW, n = 13), term and post-term neonates (38–45 PCW, n = 11), infancy and early childhood (52–364 PCW, n = 13), and adulthood (38–63 years, n = 8). Subjects with PVL fell into the following 2 categories: preterm (n = 3) and term/post-term neonates (n = 7) and were grouped together in all analyses. In all control subjects, death was unrelated to neurological injury or disease, and the diagnosis at autopsy is listed in Table 1. In all PVL subjects, death was associated with significant systemic congenital abnormalities, respiratory insufficiency, and/or cardiac abnormalities concomitant with significant neurological injury. The study cases were divided into those with no recognizable abnormalities on microscopic examination (n = 49), those with minimal white matter changes (mild diffuse white matter gliosis [DWMG]; n = 9) and those with a primary diagnosis of PVL (n = 10). For the majority of cases, the postmortem interval (PMI) was ≤24 h (n = 54). In a small number, the PMI ranged between 25 and 62 h (n = 10). This information was not available for 4 cases. Regression analyses determined that there was no statistically significant effect of PMI on any of the subunit expression levels examined in either the white matter or the cortex.

Table 1.

Clinical characteristics of control sample population

Case	PCA (weeks)	GA (weeks)	PNA (weeks)	PMI (h)	Gender	Autopsy diagnosis	Neuropathology abnormalities
1	20	20	0	N/A	NA	Chorioamnionitis	Ø
2	20	20	0	24	M	Extreme prematurity/severe IUGR	Ø
3	20	N/A	N/A	6	NA	Extreme prematurity	Ø
4	20	20	0	11	M	Extreme prematurity	Ø
5	21	21	0	30	F	Extreme prematurity	Ø
6	21	21	0	11	F	Extreme prematurity	Ø
7	23	23	0	4	F	Miscarriage	Ø
8	23	23	0	42	M	Extreme prematurity	Ø
9	23	22	1	62	M	Extreme prematurity	Ø
10	24	24	0	46	M	MCA	Ø
11	24	24	0	7	F	IUGR/umbilical cord abnormality	Ø
12	24	24	0	18	M	Extreme prematurity/MCA	Ø
13	24	24	0	32	M	Extreme prematurity	Ø
14	25	25	0	25	M	Extreme prematurity	Ø
15	26	25	1	24	M	Prematurity/sepsis	Ø
16	27	27	0	60	M	Prematurity/lung Immaturity	Ø
17	29	29	0	23	M	Prematurity/PH	Ø
18	30	30	0	10	M	Prematurity/respiratory insufficiency	Ø
19	31	31	0	4	M	Stillborn	DWMG
20	31	31	0	6	M	Stillborn	DWMG
21	32	29	3	22	M	NEC/sepsis	DWMG
22	34	34	0	31	F	Stillborn	Ø
23	35	35	0	20	M	Renal agenesis	DWMG
24	35	35	0	10	M	Stillborn	DWMG
25	37	37	0	25	M	Stilborn	Ø
26	37	37	0	N/A	M	Stillborn/renal agenesis	DWMG
27	38	36	2	21	F	Multi organ failure, gastroschisis	DWMG
28	38	37	1	5	M	CHD	Ø
29	38	38	0	N/A	F	Hypoplastic left heart syndrome	Ø
30	38	38	0	8	M	Multicystic dysplastic kidneys	DWMG
31	39	39	0	2	F	Hypoplastic left heart syndrome	Ø
32	40	40	0	22	M	Trauma	Ø
33	40	40	0	NA	M	SIDS	Ø
34	41	41	0	24	M	CHD	DWMG
35	43	40	3	13	F	CHD/PH	Ø
36	44	N/A	N/A	9	F	Cardiac Arrhythmia/abnormalities	Ø
37	45	40	5	7	M	Suffocation	Ø
38	52	40	12	24	F	Positional asphyxia	Ø
39	53	40	13	20	F	Bronchopneumonia	Ø
40	54	40	14	16	M	CHD	Ø
41	57	40	17	22	M	Bronchopneumonia	Ø
42	72	N/A	N/A	20	M	Trauma	Ø
43	76	NA	NA	20	M	Pediatric cocaine toxicity	Ø
44	83	40	43	18	M	Hyperthermia	Ø
45	87	40	47	10	F	Bronchiolitis	Ø
46	92	40	52	15	M	Drowning	Ø
47	168	40	128	20	F	Trauma	Ø
48	210	36	174	11	F	Drowning	Ø
49	208	N/A	4 years	21	F	Lymphocytic myocarditis	Ø
50	364	N/A	7 years	12	M	Drowning	Ø
51	38 years	N/A	38 years	17	M	Myelogenous leukemia	Ø
52	40 years	N/A	40 years	9	M	Hemopericardium/aortic dissection	Ø
53	43 years	N/A	43 years	22	F	Hodgkin's disease	Ø
54	50 years	N/A	50 years	17	M	Complications of diabetes mellitus	Ø
55	51 years	40	51 years	9	F	Respiratory failure	Ø
56	57 years	N/A	57 years	11	M	Leukemia	Ø
57	62 years	N/A	62 years	6	M	Trauma	Ø
58	63 years	40	63 years	19	F	Lung cancer	Ø

CHD, congenital heart disease/defect; DWMG, diffuse white matter gliosis; F, female; GA, gestational age; GM, germinal matrix; IUGR, intrauterine growth restriction; MCA, multiple congenital anomalies; M, male; N/A, information not available; NEC, necrotizing enterocolitis; PCA, postconceptional age; PH, pulmonary hypoplasia; PMI, postmortem interval; PNA, postnatal age.

Table 2.

Clinical characteristics of periventricular leukomalacia sample population

Case	PCA (weeks)	GA (weeks)	PNA (weeks)	PMI (h)	Gender	Autopsy diagnosis	Neuropathology abnormalities
1	33	33	0	16	M	PH	PVL
2	37	36	1	16	F	MCA	PVL
3	37	36	1	24	F	MCA/CHD	PVL
4	38	38	0	21	M	Multiple cardiac abnormalities/respiratory insufficiency	PVL
5	39	37	2	6	M	Hypoplastic left heart syndrome	PVL
6	39	39	0	39	F	Ascending intrauterine infection/bronchopneumonia	PVL
7	40	38	2	20	F	Hypoplastic left heart syndrome	PVL
8	41	41	0	3	F	Cerebral edema/chorioamnionitits	PVL
9	45	41	4	13	M	Complex heterotaxy/MCA	PVL
10	45	37	8	24	M	Multiple cardiac abnormalities	PVL

CHD, congenital heart disease/defect; F, female; GA, gestational age; M, male; MCA, multiple congenital anomalies; NA, information not available; PCA, postconceptional age; PH, pulmonary hypoplasia; PMI, Postmortem interval; PNA, postnatal age; PVL, periventricular leukomalacia.

Standard Neuropathologic Evaluations

Neuropathologic examinations were performed on all specimens and included macroscopic evaluations of the brain in situ and after removal, as well as microscopic examination, as previously described (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006). For control cases in the normal development arm of this investigation (Table 1), cases with gray matter necrosis or white matter injury, such as PVL, hypoxic-ischemic encephalopathy (HIE), brain malformations, or any other diagnostic lesion of the central nervous system, were specifically excluded. Thus, control cases had no evidence of gross brain abnormalities, nor were lesions detected by macroscopic examination of cerebral cortex, white matter, deep gray matter structures, hippocampus, brainstem, and cerebellum. Microscopically all control cases were diagnosed as histologically normal or presented minimal prominence of DWMG. The PVL cases presented a combination of cerebral white matter focal periventricular necrotic lesions associated with a more severe DWMG (Kinney 2009; Volpe 2009a, 2009b, Andiman et al. 2010) (Table 2).

Immunoblotting

Western blot analysis was performed on tissue homogenates of dissected white and gray matter. Adult cases were used as indices of maturity (n = 8) and as the standard for protein expression, as loading controls such as GAPDH, β-tubulin, or β-actin vary considerably in their expression levels across development and are inappropriate for our studies (Henson et al. 2008). At the time of autopsy, the unfixed brain was examined, and parieto-occipital lobe tissue was flash frozen immediately and stored at −80 °C. Subsequently, gray and white matter was dissected and homogenized separately for protein extraction. Membrane protein extraction and immunoblotting was then carried out as per our previously published protocols (Talos, Fishman et al. 2006; Talos, Follett et al. 2006; Jantzie et al. 2011; Talos et al. 2012). Briefly, after isolation of membrane proteins, a Bradford protein assay was performed to determine total protein amount, and 30 µg of protein was loaded on precast 7.5% Tris HCl gels (BioRad). Following transfer to polyvinylidene difluoride, membranes were blocked for 1 h in 5% nonfat dry milk in Tris-buffered saline solution containing 0.1% Tween-20. Membranes were then incubated in antibodies against NMDAR subunits overnight at 4 °C. All antibodies against the NMDAR subunits were purchased from Millipore, Billerica, MA, with the exception of anti-NR2D that was obtained from Santa Cruz Biotechnology, Santa Cruz, CA, USA. The following dilutions were utilized in this study: anti-NR1 (1:500; 120 kDa); anti-NR2A (1:100; 180 kDa); anti-NR2B (1:250; 180 kDa); anti-NR2C (1:100; 140 kDa); anti-NR2D (1:100; 165 kDa); anti-NR3A (1:100; 130 kDa). An HRP conjugated anti-rabbit IgG secondary antibody was then applied for 1 h and after appropriate washes protein bands were visualized using SuperSignal West Femto Maximum Sensitivity ECL substrate. Relative optical density was measured for each band using FujiFilm software.

Data Analysis

As previously published, densitometry of bands were obtained for each receptor subunit and were normalized to adult standards run on each blot (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006). Relative protein levels for individual NMDAR subunits were then plotted as a function of PCA. Outliers were identified according to the ROUT method. Regression analyses with linear and quadratic terms were used to model the relationship between PCA and relative NR1, NR2A, NR2B, NR2C, NR2D, and NR3A expression in cases between 20 and 45 PCW. Regression analyses were also performed to determine the effect of PMI, gender, or pH on NMDA subunit expression level, controlling for PCA. Significant differences in overall NMDAR expression levels among control and PVL cases between 33 and 45 PCW were determined using a Mann–Whitney U-test. Regression analyses were performed to model changes in NMDAR expression pattern with PCA in the control and PVL groups, simultaneously. Interaction terms between group and age were included to allow for differences in expression patterns with age. The adjusted R² was used to determine the best fitting model and F-tests were used to compare the expression patterns in the 2 groups. Analyses were performed with SPSS (IBM, Somers, NY, USA) and SAS v9.3 (SAS Institute, Inc., Cary, NC, USA). In all analyses, statistical significance was achieved with P ≤ 0.05.

Immunostaining

At the time of autopsy, blocks of parietal lobe cortex and underlying white matter, collected at the level of the atrium, were immersed in fresh 4% paraformaldehyde in phosphate-buffered saline (PBS; pH 7.4) and maintained at 4 °C for at least 48–72 h, before cryoprotected in 30% sucrose in PBS for 24 h. Additional double label immunocytochemistry was carried out on snap-frozen tissue blocks lightly fixed for 1 week at −20 °C in a paraformaldehyde-containing solution, as previously described (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006). Free-floating sections (50 or 120 µm thick) were double labeled with antibodies against the NMDAR subunits and cell-specific markers (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006; Jantzie et al. 2011). Sections were rinsed and blocked for 2 h in a 5% normal goat serum solution. Primary antibodies against late OL precursors (anti-O4, 1:500, gift from Dr Pfeiffer, Farmington, CT, USA), immature OLs (anti-galactocerebroside C/GalC, 1:100, Millipore), mature OLs (anti-myelin basic protein/SMI 99, 1:1000, Covance, Dedham, MA, USA), neurons (anti-neuronal nuclei/NeuN, 1:100, Millipore), astrocytes (anti-glial fibrillary acidic protein/SMI 22, 1:1000, Covance), and microglia/macrophages (biotinylated Tomatolectin, 1:200, Vector Laboratories, Burlingame, CA, USA) were then applied and sections were allowed to incubate overnight at 4 °C. The following day, sections were washed, and a species-appropriate fluorescent secondary antibody conjugated to Alexa-Fluor 568 /488 (1:1000, Invitrogen/Life Technologies, Grand Island, NY, USA) was applied for 1 h. Tomatolectin was used in combination with Avidin D conjugated to Texas Red (1:2000, Vector). Sections were then washed, blocked, and incubated overnight at 4 °C with anti-NR1 (1:100), anti-NR2A (1: 500), anti-NR2B (1:500), and anti-NR3A (1:500) antibodies. Subsequently, either a biotinylated anti-rabbit secondary antibody (1:200, Vector) followed by a fluorescent Avidin D conjugate (1:2000, Vector), or a rabbit fluorescent secondary antibody conjugated to Alexa-Fluor 568/488 (1:1000, Invitrogen) was applied for 1 h each. Sections were then washed and coverslipped with an anti-fade mounting media (Fluoromount-G, SouthernBiotech, Birmingham, AL, USA). Appropriate negative controls (no primary antibody) were run in parallel and as per the procedure described above.

Results

Developmental Regulation of NMDAR Subunits in Human Parieto-Occipital White Matter

The temporal sequence of NR1, NR2A, NR2B, NR2C, NR2D, and NR3A subunit expression in human white matter was clearly dependent upon age. Figure 1 shows the normalized values of NMDA receptor subunits (expressed as percentage of adult levels) for individual cases, from midgestation through early childhood. The overall expression levels of each subunit were markedly higher during development compared with adult expression (dotted line), with NR1 567%, NR2A 305%, NR2B 205%, NR3A 515%, NR2C 195%, and NR2D 345% of adult.

Figure 1.

NMDA receptor subunit expression in white matter changes significantly with postconceptional age. Normalized values of NMDA receptor subunits (expressed as percentage of adult levels) for individual cases, from midgestation through early childhood, are shown in scatter plots and demonstrate the overall expression levels of NR1 (A), NR2A (B), NR2B (C), NR3A (D), NR2C (E), and NR2D (F) are markedly higher during development compared with adult expression (dotted line) in white matter. Regression analyses (insets) reveal that between 20 and 45 postconceptional weeks (PCW), NR1 (A, P = 0.002) and NR2B (C, P = 0.04) expression was maximal during the preterm period, whereas NR2A expression was highest at midgestation (B, P = 0.03). The overall levels of NR3A, NR2C, and NR2D, although higher than adult throughout the investigated time course were not significantly different between 20 and 45 postconceptional weeks.

Regression analyses (Fig. 1, insets) reveal that between 20 and 45 PCW NR1 and NR2B expression in the white matter increases gradually from low levels at midgestion to maximal expression during the preterm period (Fig. 1A, P = 0.002 and Fig. 1C, P = 0.04 respectively). Unlike NR1 and NR2B, NR2A expression was highest at midgestation and exhibited a constant decline with advancing PCA between 20 and 45 PCW (Fig. 1B, P = 0.03). By term, levels of NR1, NR2B, and NR2A subunit expression were decreasing to the adult, although still higher than adult through the first years of childhood. The overall levels of the remaining subunits, NR3A, NR2C, and NR2D were not significantly different between 20 and 45 PCW (Fig. 1D,E,F). A summary of the NMDAR subunit changes between 20 and 45 PCW and relative expression peaks in white matter is shown in Table 3. There were no statistically significant effects of PMI, tissue homogenate pH, or gender on any of the investigated subunits, controlling for PCA. A summary of the statistical interaction between NMDAR subunits and gender are shown in Supplementary Table 1.

Table 3.

Summary of NMDAR subunit changes in human white and gray matter between 20 and 45 postconceptional weeks

Subunit	Relative expression peak
White matter
NR1	*Preterm
NR2A	*Midgestation
NR2B	*Preterm
NR2C	P > 0.05
NR2D	P > 0.05
NR3A	P > 0.05
Gray matter
NR1	P > 0.05
NR2A	*Term
NR2B	P > 0.05
NR2C	P > 0.05
NR2D	*Midgestation
NR3A	*Term

*P < 0.05.

Developmental Regulation of NMDAR Subunits in Human Parieto-Occipital Gray Matter

As in the white matter, we began our analyses of NMDAR subunits in the cortex by examining relative protein expression levels of NR1 (Fig. 2A), NR2A (Fig. 2B), NR2B (Fig. 2C), NR3A (Fig. 2D), NR2C (Fig. 2E), and NR2D (Fig. 2F). As shown in these scatter plots, mean values of NR1 (1259%), NR2B (234%), NR3A (674%), NR2C (335%), and NR2D (183%) were significantly higher during development compared with adult (dotted line). In contrast to all other subunits, overall NR2A expression through development was lower than adult, at only 74% of adult expression (Fig. 2B).

Figure 2.

NMDA receptor subunit expression in gray matter correlates significantly with postconceptional age. Normalized values of NMDA receptor subunits (expressed as percentage of adult levels) for individual cases, from midgestation through early childhood, demonstrate the overall expression levels of NR1 (A), NR2A (B), NR2B (C), NR3A (D), NR2C (E), and NR2D (F) compared with adult expression (dotted line) in gray matter. As shown in scatter plots, mean values of NR1 (1259%), NR2B (234%), NR3A (674%), NR2C (335%), and NR2D (183%) were significantly higher during development compared with adult (dotted line). Overall NR2A expression was lower than adult, at only 74% of adult expression (Fig. 4B). As shown in insets, regression analyses performed on cases ranging from 20 to 45 PCW demonstrated significant developmental changes dependent on age for NR2A, NR3A, and NR2D. NR2A expression increased linearly with postconceptional age (B, P = 0.01). Between 20 and 45 PCW, NR3A expression increased significantly with maturation (D, P = 0.02), with a relative peak at term and in the immediate postnatal period. In contrast, NR2D decreased with development between 20 and 45 PCW (F, P = 0.03). Regression analyses for NR1, NR2B, and NR2C revealed no significant changes in expression between 20 and 45 PCW (P > 0.05).

Regression analyses performed on cases ranging from 20 to 45 PCW (Fig. 2, insets) demonstrated significant developmental changes dependent on age for NR2A, NR3A, and NR2D. Specifically, NR2A expression increased linearly with PCA (Fig. 2B, P = 0.01). Between 20 and 45 PCW, NR3A expression increased significantly with maturation (P = 0.02, Fig. 2D), with a relative peak at term and in the immediate postnatal period. In contrast, NR2D decreased with development between 20 and 45 PCW (Fig. 2F, P = 0.03). Similar analyses performed for NR1 (Fig. 2A), NR2B (Fig. 2C), and NR2C (Fig. 2E) revealed no significant changes in expression between 20 and 45 PCW (P > 0.05). A summary of the NMDAR subunit changes between 20 and 45 PCW in gray matter is shown in Table 3. Additionally, similar to white matter, there were no statistically significant effects of PMI, pH, or gender on any of the investigated subunits in the cortex, controlling for PCA (data not shown).

Cellular Distribution of NMDAR Subunits in the Developing Human White and Gray Matter

A characteristic feature of human developing white matter is the presence of immature glia and of abundant subcortical neurons (Kinney et al. 2012). During the second half of gestation, white matter is mainly populated by pre-myelinating OLs (pre-OLs, including both O4+O1− late OL precursors and O4+O1+ immature OLs), while only few MBP+ myelin fibers are present, mostly in the deeper white matter (Back et al. 2001; Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006). At the same time, there is an abundance of microglia, which present mostly an intermediate or amoeboid activated phenotype (Billiards et al. 2006). As previously described, vimentin+ radial glial cells predominate over ramified astrocytes at midgestation, but during the preterm period are gradually replaced with mature GFAP+ astrocytes (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006).

The expression of the obligatory NMDAR subunit NR1 was very low within the white matter at midgestation, and almost exclusively confined to a few pre-OL and radial glia processes (data not shown). However, in the preterm period, NR1 was highly expressed on all cellular elements of human white matter, corresponding with the peak expression demonstrated by western blot. At preterm (25–37 PCW), NR1 was abundantly expressed in O4+ pre-OL cell bodies and processes (Fig. 3A1–C1), in Tomatolectin+ amoeboid microglia (Fig. 3D1–F1), and in NeuN+ subcortical neurons (3J1–L1), and to a lesser extent in GFAP+ astrocytes (Fig. 3G1–I1). At term (38–45 PCW, Fig. 3A2–L2), NR1 expression was relatively decreased in pre-OLs (Fig. 3A2–C2), but more robust in astrocytes (Fig. 3G2–I2), while remaining unchanged in individual amoeboid microglia (Fig. 3D2–F2) and subcortical neurons (Fig. 3J2–L2).

Figure 3.

NR1 is expressed on oligodendrocytes, microglia, astrocytes, and neurons at preterm and term. At preterm (25–37 PCW, A1–L1), NR1 was highly expressed on all cellular elements including O4+ pre-OL cell bodies and processes (A1–C1), Tomatolectin+ amoeboid microglia (D1–F1), and NeuN+ subcortical neurons (3J1–L1). NR1 was expressed to a lesser extent on GFAP+ astrocytes (G1–I1). At term (38–45 PCW, A2–L2), NR1 expression was relatively decreased in pre-OLs (A2–C2) compared with preterm expression levels, but more robust in astrocytes (G2–I2), while remaining unchanged in individual amoeboid microglia (D2–F2) and subcortical neurons (J2–L2).

White matter NR2A and NR2B receptor subunits demonstrated not only maturational changes, but also marked differences in cell-specific expression. The NR2A subunit was abundantly expressed on glial cells starting as early as midgestation, and continuing during preterm and term periods (38–45 PCW, Fig. 4). NR2A was highly expressed in pre-OLs cell bodies and processes (Fig. 4A–C), in amoeboid microglia (Fig. 4G–I), and astrocytes (Fig. 4J–L), but appeared lower in mature MBP+ OLs (Fig. 4D–F). While abundantly expressed in developing glial cells, NR2A remained undetectable in subcortical neurons during the entire late gestation period (25–37 PCW, Fig. 5A1–C1 and 38–45 PCW, Fig. 5A2–C2). In contrast, NR2B subunit was highly expressed on subcortical neurons at all ages (Fig. 5D1–F1 and Fig. 5D2–F2), with highest levels observed during the preterm period (Fig. 5D1–F1).

Figure 4.

Glial NR2A expression at term (38–45 PCW) is characterized by expression on oligodendrocytes, microglia, and astrocytes. The NR2A subunit at term was highly expressed in pre-OLs cell bodies and processes (A–C), compared with in mature MBP+ OLs (D–F). NR2A expression at term was also noted in Tomatolectin+ amoeboid microglia (G–I) and GFAP+ astrocytes (J–L).

Figure 5.

NR2A and NR2B exhibit differential expression patterns in subcortical neurons at preterm and term. Throughout preterm (25–37 PCW, A1–F1) and term (38–45 PCW, A2–F2), NR2A expression is low in subcortical NeuN+ neurons compared with NR2B. Although expressed on subcortical neurons at all ages, the highest levels of NR2B were observed during the preterm period (D1–F1).

The NR3A subunit was expressed within the developing white matter, both during the preterm and term periods (38–45 PCW, Fig. 6A–L). Robust NR3A expression was particularly observed in pre-OLs (Fig. 6A–C), mature OLs (Fig. 6D–F), amoeboid microglia (Fig. 6G–I), and subcortical neurons (Fig. 6J–L, arrowhead), and to lesser extent in astrocytes (Fig. 6J–L, small arrow).

Figure 6.

NR3A is widely expressed on glia in the developing white matter at term. From 38 to 45 PCW, robust NR3A expression is observed in pre-OLs (A–C), mature OLs (D–F), and amoeboid microglia (G–I). NR3A is also observed on subcortical neurons (J–L, arrowhead), and to lesser extent in astrocytes (J–L, small arrow).

In the cortex, NR1, NR2A, and NR2B immunocytochemistry confirmed the expression of a mature NMDAR configuration during childhood (92–364 PCW or 1–4 years, Fig. 7A–I), with NR1 (Fig. 7A–C) and NR2A (Fig. 7D–F) highly expressed on most layer V cortical neurons, concomitant with a more restricted NR2B (Fig. 7G–I) expression pattern.

Figure 7.

NR1 and NR2A are highly expressed on layer V cortical neurons during childhood. Immunohistochemistry confirms the expression of a mature NMDAR configuration during childhood (92–364 PCW or 1–4 years, A–I), whereby NR1 (A–C) and NR2A (D–F) are highly expressed on layer V pyramidal neurons, concomitant with a restricted NR2B (G–I) expression pattern.

NR2A is Significantly Increased in White Matter From Postmortem Cases of Periventricular Leukomalacia

In addition to investigating normal developmental NMDAR subunit expression patterns, levels of NR1, NR2A, NR2B, and NR3A were analyzed in white and gray matter from infants with PVL (cases were 33–45 PCW). Mean subunit expression levels in the PVL cases were compared with the mean control expression levels of the control cases in the same developmental window. In white matter, PVL cases had significantly increased NR2A expression, exhibiting 447% of adult expression, compared with only 139% in the control samples (Fig. 8B, P = 0.02). Similar analyses of NR1, NR2B, and NR3A (Figs 8A,C,D) in the white matter revealed no statistically significant differences in overall expression levels between the injured brains and noninjured controls. In the cortex (Fig. 9), NR1 (Fig. 9A) and NR3A (Fig. 9D) were decreased (P = 0.05) in PVL cases, while other subunits remained unchanged. A summary of the statistical interaction between NMDAR subunits and gender are shown in Supplementary Table 1. As with the control cases, we found no statistical relationship between gender and NR1, NR2A, NR2B, and NR3A in the gray matter and NR1, NR2B, and NR3A in white matter in PVL cases. Our analyses did however reveal an interaction between gender and NR2A expression in the white matter of a small number PVL cases (Supplementary Table 1).

Figure 8.

NR2A is significantly elevated in the white matter samples from brains with periventricular leukomalacia (PVL). In cases of confirmed PVL, NR1 (A), NR2B (C), and NR3A (D) expression levels are similar to those present in white matter from uninjured brains. However, NR2A expression (B) is significantly elevated in PVL white matter compared with age-matched uninjured brains and represents a significant deviation from normal developmental expression (P = 0.02).

Figure 9.

NR1 and NR3A expression levels are reduced in gray matter samples from brains with periventricular leukomalacia (PVL). In cases of PVL, NR1 (A) and NR3A (D) expression levels are decreased in gray matter compared with those observed in uninjured brains (P = 0.05 for both). Levels of NR2A (B) and NR2B (C) are unchanged in gray matter from PVL compared with uninjured brains.

Discussion

Rodent studies have shown that NMDARs are present on both neurons and glia (Karadottir et al. 2005; Salter and Fern 2005; Manning et al. 2008). In neurons, activation of NMDARs is critical for learning, memory, and synaptic plasticity under physiological conditions (Bliss and Collingridge 1993; Malenka and Nicoll 1993; Nicoll and Malenka 1999; Cull-Candy et al. 2001; Nakanishi et al. 2009; Matta et al. 2011), while their overactivation has been implicated in the pathophysiology of stroke and other neurodegenerative disorders. In vitro and in vivo investigations indicate that the synaptic and extrasynaptic expression of NMDAR subunits differentially contribute to excitotoxic injury (Sattler et al. 2000; Papadia et al. 2008; Wroge et al. 2012). As little is known about developmental expression patterns in the human brain, especially very early in development, we examined NMDAR subunits in both the developing white and gray matter of the human brain to understand the age-specific role that individual subunits may play in the selective vulnerability to perinatal brain injury and the pathophysiology of PVL. Knowledge of the developmental regulation of human NMDAR subunits is crucial for understanding normal brain development and rational design of age-appropriate pharmacotherapies for diseases with excitotoxicity as a main component of their pathophysiology, including PVL, HIE, and neonatal seizures. Here, we establish for the first time, a comprehensive developmental expression profile of the 6 major NMDAR subunits throughout gestation and in the early postnatal period in a large postmortem data set.

Subunit Composition Dictates NMDAR Pharmacology, Function, and Biophysical Properties

Glutamate homeostasis is central to proper brain development and injury response, and the balance can be shifted towards excitotoxicity following brain insults (Jensen 2002, 2005). Our data, and those of others, indicate that individual subunits are subject to strict mechanisms of regulation (Monyer et al. 1992; Henson et al. 2008). While NR1 is the obligatory subunit, the specific properties of functional NMDARs are shaped by the combination of NR1 with NR2 and/or NR3 subunits (Monyer et al. 1994; Flint et al. 1997; Sasaki et al. 2002; Henson et al. 2008). NR2 and NR3 subunits determine the biophysical and pharmacological properties, and influence NMDAR assembly and synaptic targeting. Specifically, NR2A and NR2B have different kinetic properties, subcellular localization, and trafficking. The NR2A subunit is responsible for shorter excitatory postsynaptic current (EPSC) duration and in the adult brain is located at synapses, while NR2B confers longer EPSCs and is enriched at extrasynaptic sites (Flint et al. 1997; Massey et al. 2004; Hardingham and Bading 2010). Both NR2A and NR2B subunits generate high-conductance channel openings with high sensitivity to block by Mg²⁺, while NR2C and NR2D give rise to low-conductance openings and lower sensitivity to Mg²⁺ (Cull-Candy et al. 2001). Inclusion of the NR3A subunit suppresses the receptor activity, decreases its Ca²⁺ permeability and confers a relative insensitivity to Mg²⁺ (Das et al. 1998; Sasaki et al. 2002; Tong et al. 2008; Zhou et al. 2009), which allows receptor activation in the absence of membrane depolarization.

NMDAR Subunits Are Highly Expressed in Immature White Matter and Exhibit Unique Glial Expression Patterns

We demonstrate for the first time that in the human developing white matter, all NMDAR subunits are expressed at higher levels compared with adult, providing evidence to support an intrinsic prenatal vulnerability to NMDAR mediated injury. Midgestation is characterized by the highest NR2A levels, specifically on glial cell populations and not neurons, while NR1 and NR2B expression peaks during the preterm period. Enhanced NR2B function increases the Ca²⁺ permeability of the NMDAR during a time of rapid myelination and subcortical neuronal growth. Similarly, we found increased expression of receptor subunits with lower Mg²⁺ sensitivity, NR2C, NR2D, and NR3A, in the white matter throughout development compared with adult expression. Lower Mg²⁺ sensitivity results in a lower threshold for NMDAR activation, and coincides with the overall peak in NMDAR expression in the developing human white matter. Notably, this is the same time period where other glutamate receptors are upregulated, including the Ca²⁺-permeable GluR2-deficient AMPARs (Follett et al. 2000, 2004; Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006) and Group I metabotropic glutamate receptors (Jantzie et al. 2011). Together, these characteristics may in part underlie the vulnerability of white matter at this stage of development and the common emergence of PVL.

There is evidence from animal models that functional NMDARs are expressed on OLs, localize predominantly to the processes, and confer vulnerability to OL injury (Karadottir et al. 2005; Salter and Fern 2005; Micu et al. 2006; Manning et al. 2008). In rodents, the NR1 subunit is highly expressed on O4+/O1+ immature OLs, and as shown here, has a similar expression pattern in preterm infants (Manning et al. 2008). We demonstrate that human OLs express multiple NMDAR subunits, such as NR2A and NR3A. Given their abundant expression on the processes and the reduced volume of these structures, even small rises in Ca²⁺ through NMDAR channels could be excitotoxic. Indeed, activation of NMDARs predominantly injures the OL processes in animals (Karadottir et al. 2005; Salter and Fern 2005; Micu et al. 2006), and there is a striking lack of OL processes within human necrotic PVL foci (Billiards et al. 2006). We and others have previously shown that high expression of GluR2-lacking AMPA receptors (Follett et al. 2000, 2004; Itoh et al. 2002; Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006) on immature OLs at a point when they initiate myelination, increases their sensitivity to glutamate-mediated excitotoxicity and Ca²⁺ overload. Interestingly, the deletion of NMDARs from developing OLs induces a compensatory increase in surface expression of Ca²⁺-permeable AMPARs, which emphasizes the critical role of glutamate-evoked Ca²⁺ signaling in OL development (De Biase and Bergles 2011), consistent with previous observations in neurons (Wang and Gao 2010).

We also found that NMDARs, mainly composed of NR1 and NR2B subunits, are highly expressed on subcortical neurons. The high expression of NR2B-containing receptors, which coincides with co-expression of GluR2-lacking AMPARs during this developmental period (Talos, Fishman, et al. 2006; Talos, Follett, et al. 2006) may render this neuronal cell population sensitive to excitotoxicity. Subcortical neurons likely represent remnant subplate neurons and late migrating GABAergic neurons (Xu et al. 2011; Kinney et al. 2012), which are critical for proper cortical network development, including thalamo-cortical connections and inhibitory circuit formation. Excitotoxic injury to subcortical neurons and damage to the subplate may contribute to improperly integrated inhibitory circuits, cognitive dysfunction, and seizure susceptibility, as observed in survivors of prematurity.

In this study, we found that amoeboid microglia express high levels of NMDARs in the preterm and term period, especially composed of NR1, NR2B, and NR3A subunits. Their overactivation under HI conditions may increase the production of proinflammatory cytokines and chemokines, leading to subsequent injury to neurons and OLs, as has been demonstrated in developing rodents (Murugan et al. 2011; Kaindl et al. 2012). We also observed human astrocytes to express NMDARs, containing predominantly NR1 and NR2A subunits, concordant with previous demonstration of functional NMDARs on rodent astrocytes (Lalo et al. 2006; Palygin et al. 2011).

NR2A and NR3A Expression Levels Peak in Gray Matter at Term and in the Early Postnatal Period

Here, we determined that expression levels of NR1, NR2B, NR2C, NR2D, and NR3A were significantly higher in the immature cortex compared with adult levels across the entire investigated developmental time course. Subanalyses of expression levels during the midgestation, preterm, and term and the immediate postnatal period (20–45 PCW) indicated that NR2A and NR3A significantly increase with PCA, with a relative NR3A peak at term, together with a constant decrease in NR2D levels with maturation. These data are similar to the developmental expression of NMDAR subunits in the rat brain, where NR2D mRNA is highly expressed prenatally, while NR2A mRNA is first highly detected near birth and protein expression appearing around postnatal day 7 (P7), and NR3A peaking at P8 (Monyer et al. 1994; Wong et al. 2002; Kim et al. 2005; Rinaldi et al. 2007). High NR1, NR2B, and NR3A expression observed in human neonatal cortex are indicative of overall cortical immaturity and reflect a critical period of neuronal development and synaptic formation (Webb et al. 2001; de Graaf-Peters and Hadders-Algra 2006). Corroborating prior reports, gray matter expression levels presented here indicate a clear shift in expression of NR2B and NR2A receptors with advancing PCA. Our immunohistochemistry confirms the presence of NR1, NR2A, and NR2B on layer V cortical neurons, and is also reflective of an NR2B to NR2A switch with maturity. Importantly, LTP induction in the neonate is known to acutely drive the switch of synaptic NMDARs from NR2B to NR2A containing (Bellone and Nicoll 2007; Matta et al. 2011).

NMDAR Subunits Are Altered in PVL

In order to identify significant deviations from normal development and gain insight into the putative role of NMDAR subunit composition in the pathophysiology of developmental brain injury and to enhance clinical translation of prior rodent data (Manning et al. 2008), we acquired white and gray matter from 10 cases between 32 and 45 PCW with PVL. In the white matter of these PVL brains, NR2A expression was significantly elevated compared with age-matched uninjured controls. Interestingly, NR2A subunits have faster kinetics consistent with lower channel maximum open probability time compared with NR2B (Gielen et al. 2009) and a relationship between the NR2A gene GRIN2A genetic polymorphisms and attention deficit hyperactivity disorder (Turic et al. 2004) have been documented in human studies linking this subunit to cognition. Importantly, the increase in NR2A in PVL cases documented here is distinctly opposite the developmental pattern observed in noninjured developing brains. It is possible that these alterations may be compensatory as it increases Mg²⁺ sensitivity, and could result in relative protection from excitotoxicity. However, the increased NR2A protein is not necessarily direct evidence of increased NR2A subunit incorporation into functional receptors. In the absence of robust evidence of increased NR1 and therefore NMDAR densities, it is likely that these increases reflect significantly altered NMDAR composition in the injured developing brain (Turnock-Jones et al. 2009). In gray matter, NR1 and NR3A expression levels were also decreased in the PVL cohort, and could represent compensatory changes that would decrease NMDA receptor function and reduce excitotoxicity as well. However, decreased NR1 and NR3A may also signify developmental delay in injured cortex or be reflective of cell loss. Taken together, the changes in NMDAR subunit expression observed in the white and gray matter of infants with PVL, while observed in infants that died of their injuries, indicate selective abnormalities within the glutamatergic system and deviations from normal developmental trajectories.

Potential Study Limitations

We attempted to carefully address the confounding factors that are inherent to all human tissue studies, including PMI, pH, and gender (Henson et al. 2008). While the control arm of our investigation encompasses a relatively large cohort, we were limited in this investigation by a smaller number of PVL cases. We found no statistical relationship or interaction between individual subunit expression and gender in white or gray matter without evidence of brain injury. In contrast, our analyses revealed an interaction between gender and NR2A expression in the white matter of PVL cases; however, our study is not powered appropriately to comment definitively on gender differences in NMDAR subunit expression in these cases. Certainly, future investigations in larger PVL cohorts would be useful to understand the relationship between NMDAR subunit expression, gender, and vulnerability to excitotoxicity in the developing central nervous system.

The small number of PVL cases, and the overall amount of tissue obtained from injured young brains also limited immunohistochemical cellular localization of individual subunits. Investigation of NMDAR subunit expression in clinically relevant rodent models of PVL, including those modeling in utero and prenatal HI (Robinson et al. 2005; Mazur et al. 2010) would be useful in addressing these issues and are warranted. NMDAR subunit expression on individual cell types, as performed in the development arm of this investigation, was not feasible in the PVL samples, due to the overall poor integrity of the PVL tissue in combination with reduced antigenicity and fixation-sensitive receptor subunit antibodies. As subunit composition and compartmental localization of NMDARs affect channel activity and downstream signaling (Kohr 2006), assessment of NMDARs at synaptic, perisynaptic, and extrasynaptic sites would have yielded substantial information on NMDAR function during development and following brain injury. Given that this study was performed exclusively in postmortem human brain, operative analysis of individual combinations of receptor subunits via electrophysiology, Ca²⁺ imaging, or double labeling for multiple receptor subunits within a specific cell type was not possible and is an inherent limitation.

Conclusions

The present data reveal unique patterns of NMDAR subunit expression in the premature human brain that may intrinsically predispose it to glutamate-mediated HI injury, including the presence of highly calcium permeable NR2B-containing and magnesium-insensitive NR3A-containing NMDARs. These patterns have been well documented in rodent tissue, and have been implicated in the heightened excitability necessary for the normal critical period of synaptogenesis, but may also represent a “double-edged sword” of vulnerability that renders the perinatal brain susceptible to excitotoxic injury. In PVL, regional NMDAR subunit expression changes documented in NR2A, NR1, and NR3A may be compensatory or protective against excitotoxicity and simultaneously impair critical period plasticity. Indeed, rodent studies have revealed that NMDAR antagonists such as memantine are protective following injury and that brief treatment is not associated with adverse effects on brain development (Chen et al. 1998; Manning et al. 2008, 2011). These data suggest that therapeutic efficacy and clinical utility of NMDAR-directed therapies will depend on targeting the most appropriate NMDAR subtype, will likely require tight regulation of timing of administration in restricted developmental windows (Costa et al. 2010), and support further consideration of brief NMDAR antagonist therapy in human premature brain injury.

Supplementary Material

Supplementary Material can be found at http://www.cercor.oxfordjournals.org/.

Funding

This work was supported by the National Institutes of Neurological Disorders and Stroke at the National Institutes of Health (NS 031718 and DP1 OD003347 from the Office of the Director, to F.E.J), the Heart and Stroke Foundation of Canada (to L.L.J), and Alberta Innovates Health Solutions (to L.L.J). This work was also supported by the Boston Children's Hospital Intellectual and Developmental Disabilities Research Center Cellular Imaging Core (P30 HD18655).