Abstract
During cortical development, oligodendrocyte precursor (OPC) attachment and detachment to microvessels play a crucial role in their positioning and differentiation. In the developing brain, endothelial cells are regionally diverse, and previous studies showed a peak in cortical endothelial NMDA receptor (eNMDAR) expression during perinatal life, coinciding with OPC migration along microvessels. This raises the hypothesis that eNMDAR might influence the fate of vessel-associated OPC. In this study, a Grin1lox/lox/VeCadCre mouse model was used to investigate in females and males the effects of endothelial GluN1 invalidation (eNMDAR−/−) on (1) positioning and differentiation of cortical oligodendrocytes and myelination, (2) OPC/microvessel association and endothelial MMP9-like activity, and (3) motor activity. Results showed that, from postnatal days (P) 2 to P15, PDGFRα expression was increased in eNMDAR−/− mice and returned to wild-type levels by P45. CNPase and MBP expression was reduced at P15 and remained low in adult eNMDAR−/− mice. Histological analysis revealed no change in OPC–microvessel association, but positioning was altered with increased density in layers VI and V at P15. Myelination was impaired, as evidenced by thinner corpus callosum, reduced myelin sheath thickness, and higher g-ratio. Axonal mitochondria density was significantly increased. Functional tests revealed that glutamate could not stimulate endothelial MMP9-like activity in eNMDAR−/− mice. Molecular, histological and functional changes were linked to sensorimotor disabilities. At P45, despite the absence of observable myelination defects, locomotor impairments persisted, suggesting that early OPC differentiation disruption contributes to lasting motor dysfunction. These findings offer new insights into OPC vulnerability in human preterm infants.
Significance Statement
During brain development, oligodendrocyte precursors (OPCs) integrate the neocortex by migrating along radial microvessels. Here, we show that targeted invalidation of the endothelial NMDA receptor delays the positioning and the differentiation of OPC in layers of the sensorimotor cortex resulting in sustainable underexpression of MBP, in reduced density of myelinated fibers, thinner myelin sheaths, and higher g-ratio values. At a functional level, invalidation of the endothelial NMDAR results in the inability for glutamate to stimulate MMP9-like activity. These molecular, cellular, and functional phenotypes are associated with neonatal and long-term motor impairments. Our findings highlight the contribution of the endothelial NMDA receptor on the differentiation of oligodendrocytes entering the sensorimotor cortex along microvessels.
Introduction
During cortical development, several cell types, including GABAergic interneurons (Won et al., 2013), astrocytes (Tabata et al., 2022), and oligodendrocytes (Tsai et al., 2016), require close interaction with microvessels to migrate and reach their final destination. The literature suggests that endothelial cells express and release molecules, such as matrix metalloproteinases (MMP; Léger et al., 2020a) and trophic factors like BDNF (Nakahashi et al., 2000), which actively contribute to the control of migrating cells, helping to position them correctly in the maturing brain (Henry et al., 2013). In line with this, recent studies have shown that while the attachment of oligodendrocyte precursor cells (OPCs) to microvessels is crucial for their migration, their detachment is equally important for their subsequent differentiation and the induction of myelination (Su et al., 2023). Moreover, growing evidence indicates that neurovascular dysfunction can lead to altered cell migration, potentially contributing to neurodevelopmental disorders (Wang et al., 2023). For instance, studies on fetal alcohol spectrum disorder (FASD) have shown that alcohol impairs the development of the cortical vasculature and endothelial cell activity (Jégou et al., 2012; Léger et al., 2020b). This detrimental effect on vascular integrity is associated with mispositioning of oligodendrocytes and long-term myelination defects (Brosolo et al., 2022; Marguet et al., 2022; Newville et al., 2022). Together, these findings support the hypothesis that vascular cues contribute to the cell fate of oligodendrocytes (Su et al., 2023).
At the molecular level, there is increasing evidence that endothelial cells in the developing brain are not phenotypically or regionally uniform (Tan et al., 2016; Porte et al., 2017). For example, transcriptomic and proteomic studies have revealed age-dependent specificities in cortical microvessels, which undergo significant remodeling during the first 2 weeks postnatally, involving changes in the extracellular matrix, cell adhesion, and junction protein expression (Porte et al., 2017). Notably, there is a peak in the expression of endothelial NMDA receptors (eNMDAR) during perinatal life, a period that coincides with the migration of OPC into the neocortex along radial microvessels (Nishiyama et al., 2021; Xia and Fancy, 2021). Furthermore, pharmacological studies conducted on cultured cortical slices from postnatal day 2 (P2) mouse neonates (Léger et al., 2020a) and primary cultures of dissociated P2 cortical mouse endothelial cells (Legros et al., 2009) have shown that glutamate stimulates endothelial intracellular calcium mobilization and the activity of metalloproteinase 9 (MMP9), a protease involved in regulating cell migration (Seeds et al., 1997; Barkho et al., 2008). These effects of glutamate are blocked by MK801, indicating an NMDAR-dependent mechanism (Legros et al., 2009). Given that the expression of eNMDAR in the developing cortex coincides both regionally and temporally with the migration of oligodendrocytes along radial cortical microvessels (Nishiyama et al., 2021; Xia and Fancy, 2021), it is tempting to speculate that eNMDAR may play a role in regulating the cell fate of oligodendrocytes.
In the present study, we hypothesized that eNMDAR are involved in the differentiation and positioning of oligodendrocytes entering the developing cortex through a vessel-associated process. Using a Grin1lox/lox/VeCadCre mouse strain, the objectives of this study were to characterize the impact of endothelial GluN1 NMDA receptor subunit inactivation on: (1) the expression of differentiation markers in the cortical oligodendrocyte lineage, (2) the cortical positioning of differentiating oligodendrocytes, (3) the development of the cortical vasculature and the density of vessel-attached oligodendrocyte precursors, (4) the cortical endothelial MMP9-like activity, (5) the myelination process, and (6) the associated neonatal and adult motor disorders, as assessed through sensorimotor tests and locomotor activity.
Materials and Methods
Chemicals
Bovine serum albumin (BSA), Hoechst 33258, isolectin-B4-TRITC, the cell permeant MMP-9 inhibitor II, phosphate-buffered saline (PBS), and Triton X-100 were from Sigma-Aldrich. SDS-PAGE Tris-glycine gel, renaturing buffer, developing buffer, and DQ-gelatin-FITC substrate were from Invitrogen. Artificial cerebrospinal fluid (aCSF) was from Hello Bio. Paraformaldehyde (PFA) was obtained from Labonord. Characteristics of the primary antibodies against CD31, CNPase, GAPDH, MBP, Olig2, PDGFRα, and β-actin are summarized in Supplementary Table S1. The secondary antibodies Alexa Fluor 488 donkey anti-rabbit (A-21206), Alexa Fluor 488 chicken anti-rat (A-21470), Alexa Fluor 488 donkey anti-goat (A-11055), and Alexa Fluor 594 donkey anti-rat (A-21209) used for immunohistochemistry were from Invitrogen. The horseradish peroxydase goat anti-rabbit (31460), rabbit anti-goat (305-035-045), and donkey anti-mouse (715-035-151) used for Western blot experiments were from Invitrogen.
Animals and housing
Wild-type C57Bl6/129 mice were purchased from Janvier. Floxed Grin1 mice (B6.129S4-Grin1tm2Stl/J; # 005246; Fig. S1) and VE-Cadherin-Cre mice [B6.FVB-Tg(Cdh5-cre)7Mlia/J; # 006137; Fig. S1] were from The Jackson Laboratory. Endothelial NMDA receptor knock-out mice (eNMDAR−/−) were obtained in the animal facility of the research building of the Rouen Faculty of Medicine by crossing Floxed Grin1 and VE-Cadherin-Cre strains. Animals were kept in temperature-controlled room (21 ± 1°C) with 12 h light/dark cycle (lights on 7:00 A.M. to 7:00 P.M.) and with food and water ad libitum. Mice were used according to the French Ethical Committee recommendations and European directives 2010/63/UE, and experiments were carried under the supervision of authorized investigators (authorization no. APAFIS#22136-2019092013438607).
Genotyping by polymerase chain reaction and validation of eNMDAR knock-out in endothelial cells
DNAs were extracted from mouse tails using NucleoSpin DNA Rapidlyse and were amplified on a thermal cycler with the following program: 2 min, 94°C; 20 s, 94°C; 15 s, 65–1.5°C per cycle decrease; 10 s, 68°C; repeats step 2–4 for 10 cycles; 15 s, 94°C; 15 s, 50°C; 10 s, 72°C; repeats step 5–7 for 28 cycles; 2 min, 72°C. Primers were provided by the Jackson Laboratory; Grin1 5′-AAACAGGGCTCAGTGGGTAA and 3′-GTGCTGGGATCCACATTCAT; Cre transgene 5′-GCGGTCTGGCAGTAAAAACTATC and 3′-GTGAAACAGATTGCTGTCACTT. Electrophoresis was realized on 3.5% of agarose gels and visualisation were performed with ethidium bromide (Fig. S1). Compared expression of Grin1 in cortical microvessels from wild-type (WT) and eNMDAR−/− mice was previously controlled by qRT-PCR using the primers Grin1f 5′-GAGATCGCCTACAAGCGACA and Grin1r 5′-TGGTACTGCTGCAGGTTCTT (Léger et al., 2020a). Visualization of the GluN1 subunit was done by double immunostaining with CD31 in WT and eNMDAR−/− mice at P2 (Fig. S1, Table S1). GluN1-positive puncta associated to microvessels were analyzed by 3D maps using the Imaris 10.2 software (Oxford Instruments). In VE-Cadherin-Cre mice, the Cre-recombinase is detected as early as embryonic day (E) 7.5 and progresses to almost full penetrance by E14.5 (Alva et al., 2006). Cre-recombinase was visualized by immunohistochemistry in cortical slices from VE-Cadherin-Cre mice (Fig. S1, Table S1).
Genotype and sex considerations
Because major findings of the present study highlighted a dysregulation of the oligodendrocyte differentiation and, in particular, of the myelination process in eNMDAR−/−, a control experiment targeting MBP expression was performed by Western blot in the four genotypes: Grin1+/+ VeCad+/+ (wild-type mice), Grin1lox/lox VeCad+/+ (Floxed Grin1 mice), Grin1+/+ VeCadCre (VE-Cadherin-Cre mice), and Grin1lox/lox VeCadCre (eNMDAR−/− mice; Fig. S1). The atypical MBP phenotype was only observed for the eNMDAR−/− genotype. In the main manuscript, data were expressed by comparing eNMDAR−/− mice with wild-type (WT) mice. Likewise, in order to research a gender effect, visualization and quantification of MBP-positive fibers invading the developing cortex were done in females and males (Fig. S6). Similarly, neonatal sensorimotor tests and locomotor tests were done from P2 to adulthood distinguishing between females and males (Fig. S7). Because no sex differences were found, data were pooled in the main manuscript and nonpooled data provided in supplementary figures (Figs. S6, S7).
Western blot
Cortices from WT and transgenic mice were rapidly microdissected at P2, P15, and P45. Tissues were homogenized and lysed in, respectively, 250 µl, 1 ml, and 2 ml of lysis buffer with 1% of protease and phosphatase inhibitors. After centrifugation, supernatants were collected and protein concentrations were determined by Bradford assay. One hundred micrograms of proteins were denatured at 100°C for 5 min in Laemmli 4× buffer (Tris-HCl 0.5 M; pH 6.8; SDS 8%; bromophenol blue 0.5%; glycerol 10%; 2-mercaptoethanol 10%; H2O). They were electrophoresed on 7.5, 10, or 12% acrylamide gel. Then, proteins were transferred to a nitrocellulose membrane. The membrane was incubated with blocking solution (Tris buffer saline with Tween 20 0.1%; milk 5%) for 45 min at room temperature and incubated overnight at 4°C with primary antibodies raised against Olig2, PDGFRα, CNPase, and MBP (Table S1). After incubation with the corresponding secondary antibodies coupled to horseradish peroxidase (Table S1), proteins were visualized using an enhanced chemiluminescence ECL Plus immunoblotting detection system (ECL; Bio-Rad Laboratories). The intensity of immunoreactive bands was quantified using a blot analysis software (Bio-Rad Laboratories). Depending of the protein of interest, GAPDH or β-actin was used as molecular weight standard.
Immunohistochemistry
After fixation of WT and transgenic mice at P2, P15, and P45 by intracardiac perfusion with PBS and PBS-PFA 4%, brains were extracted and additionally fixed in a PBS-PFA 4% solution for 12 h. Afterward, brains were placed in a PBS/sucrose 30% solution for 24 h. Then, brains were frozen in isopentane at −30°C and conserved at −80°C. Twenty-micrometer-thick transversal slices of brains were realized and incubated with blocking solution (BSA 1%; Triton X-100 0.3%; PBS) for 45 min before incubation with primary antibodies raised against Olig2, PDGFRα, MBP, and CD31 overnight at 4°C. Then, slices were rinsed three times with PBS and incubated with the adequate secondary antibody for 1.5 h. Cell nuclei were visualized by incubating slices for 1 min with Hoechst 33258 diluted at 1/5,000. Fluorescent signals were observed with a Thunder Tissue 3D Imaging system CTR5500 (Leica Microsystems).
Quantification of density and positioning of Olig2- and PDGFRα-positive cells in developing cortical layers
Measurement of the density of Olig2 and PDGFRα immunoreactive cells in cortical layers and corpus callosum (CC) was performed after immunostaining experiments on P2, P15, and P45 brain slices (Table S1). Images were acquired at 10× magnification, and regions of interest (ROI) were defined within the superficial cortical layers (SCL, future layers I–IV), the deep cortical layers (DCL, future layers V–VI), and CC at P2. At P15 and P45, the same procedure was used and ROI defined in cortical layers I–VI and in CC. Fluorometric analysis using the multipoint counting tool of the FIJI software (US National Institutes of Health) gave access to the number of immunoreactive cells present in the ROI. Density was then determined by a ratio between the number of cells and the ROI area. Ventro-dorsal distribution of the cell densities visualized the positioning of immunoreactive cells in the cortical layers. The analysis was repeated to cover the CC, the superficial and deep cortical layers, in both hemispheres and in three slices per animal (Brosolo et al., 2022).
Measurement of microvascular organization and of oligo-vascular association in the developing cortex
Analysis of the radial organization of cortical microvessels was performed at P2 after CD31 immunolabeling as previously described (Brosolo et al., 2022). Practically, images of the somatosensory cortical vasculature were acquired at 10× magnification and saved as Tiff files. Afterward, ROI were defined in the somatosensory cortex, and a frame of perpendicular lines to the cortical edge (radial orientation) was defined and applied for each ROI. For microvessels parallel to the frame, the Metamorph Software (Roper Scientific) attributed the angular value 0°. For microvessels perpendicular to the frame, the Metamorph Software attributed the maximal angle value 90°. Angle values of cortical microvessels were then classified and distributed in four classes [0–25°], [25–50°], [50–75°], and [75–90°] (Lecuyer et al., 2017). Validation of the OPC phenotype associated with radial cortical microvessels was done in P2 brain slices by triple immunolabeling targeting PDGFRα, Olig2, and CD31 (Table S1). Quantification of vessel-associated OPC was performed in mouse brain slices at P2 after double immunostaining with PDGFRα and CD31 antibodies (Table S1). Z-stack acquisitions were done at 40× magnification using a Leica Thunder Tissue 3D Imaging System CTR5500 in WT and eNMDAR−/− mice. Afterward, Z-stack series of images were loaded into the IMARIS imaging software 9.0.2 (Bitplane) for 3D reconstruction maps (Movies 1, 2). A given OPC was considered vessel-associated when the maximal distance between the center of PDGFRα-positive cell body and the center of the vessel lumen was 10 µm or less (Brosolo et al., 2022).
Visualization of OPC-vessel association by 3D reconstruction on the sensorimotor cortex of P2 WT mice. OPC are immunolabeled using a PDGFRα antibody (green signal) and vessels by CD31 immunostaining (red signal). [View online]
Visualization of OPC-vessel association by 3D reconstruction on the sensorimotor cortex of P2 eNMDAR−/− mice. OPC are immunolabeled using a PDGFRα antibody (green signal) and vessels by CD31 immunostaining (red signal). [View online]
Quantification of vascular and neuronal MMP9-like activity by in situ zymography
Cultured cross-slices from WT and eNMDAR−/− mouse brains were done in the frontal cortex between the 2.07 and 2.91 mm rostro-caudal positions (Paxinos et al., 2019). Brain slices were performed at P2, a developmental stage showing numerous vessel-associated OPC invading the developing cortex. P2 mice were killed via decapitation, and brains were rapidly dissected and immediately placed in ice-cold aCSF. Transverse slices (250 μm) were cut at 4°C using a Leica Vibratome , then transferred into a 35 mm glass-bottom dish (Thermo Scientific), and incubated at 37°C in a humidified incubator under a controlled atmosphere of 5% CO2/95% air in aCSF containing isolectin-B4-TRITC (1/50) to label cortical microvessels. The slices were then washed twice for 10 min with fresh aCSF and placed under an inverted fluorescence video-microscope system (DMI 6000B, Leica) under slow perfusion with aCSF containing the quenched gelatinase substrate DQ-gelatin-FITC (1/50) in the absence (control) or presence of glutamate (100 μM). The appearance of FITC fluorescence, indicative of cell-specific gelatinase activity in the slice, was recorded for time-lapse analysis with the Metamorph Software. Microphotographs were acquired at 488 and 540 nm every 15 min for 3 h to follow the DQ-gelatin-FITC degradation kinetics and to visualize the cortical microvessels, respectively. To discriminate between vascular and neuronal gelatinase activities, an overlay of MMP9-like activity (FITC) and isolectin-B4-TRITC signals was done before quantification of the fluorescence intensity with the Metamorph Software. A background level was defined in an adjacent negative region within the slice. MMP9 expression was controlled in both WT and eNMDAR−/− mice at P2 by using qRT-PCR (MMP9f 5′-CGGTCCTCACCATGAGTCC and MMP9r 5′-ACAAGTATGCCTCTGCCAGC) and scanline analysis of confocal-acquired immunohistochemistry to discriminate between nervous cells and microvessels (Fig. S2, Table S1).
Quantification of MBP-positive fibers and MBP-positive cell densities
Measurement of MBP-positive fiber density was realized at P15 and P45 in the CC and cortical layers of the somatosensory cortex from WT and eNMDAR−/− mice. After immunostaining of mouse brain slices with MBP antibody (Table S1), high-resolution images were acquired at 10× magnification using a Thunder Tissue 3D Imaging System. Images were loaded in the ImageJ software (US National Institutes of Health), and a thresholding of the immunoreactive MBP fibers was done in each ROI of the cortex, giving access to the thresholded fiber area/ROI area ratio. After noting the presence of ectopic MBP-positive cell bodies in eNMDAR−/− mice at P15, density of MBP-positive cell bodies was also quantified in cortical layers using the “Cell counter” plugin of the ImageJ software.
Transmission electron microscopy
Ultrastructural studies were carried out according to standardized protocols (Brosolo et al., 2022). Briefly, CC obtained from WT and eNMDAR−/− at P15 and P45 were fixed in a 2% glutaraldehyde solution, postfixed with 1% osmium tetroxide, and embedded in epoxy resin. Semithin sections were stained with toluidine blue, uranyl acetate, and lead citrate to enhance contrast and examined under photomicroscope to validate the ROI before proceeding to ultrathin sections. Grids were examined with a transmission electron microscope (H7650, Hitachi) at 80 kV in high contrast mode, equipped with a CCD camera Orius 11MPx (Gatan). After acquisition of ×25,000 magnification images using the Digital Micrograph software (Gatan), the axonal myelinated density, axon diameter, myelin sheath thickness, g-ratio, and mitochondria density were quantified using the FIJI software. Six ROIs were quantified per ultrathin section and two ultrathin sections were analyzed per animal.
Behavioral studies
Righting reflex test was realized from P2 to P8 (Feather-Schussler and Ferguson, 2016). Pups were placed on their back on a flat surface and time spent to turn over on their four limbs was measured. Two consecutive tries were realized and averaged. A time cutoff was fixed at 60 s. The negative geotaxis test was performed between P5 and P12 (Feather-Schussler and Ferguson, 2016). Pups were placed head down on a 30° inclined surface. Time spent by the pup to rotate at 180°, head to the top of the platform was recorded. Maximum time of 60 s was fixed for each trial. One try was realized per day. Locomotor activity was studied at P15 and P45 using the open field test (Tatem et al., 2014). Mice were isolated 5 min in individual cage before the test and placed in a box (50 × 50 × 38 cm). The test takes advantage of the naturel trend of rodent to explore their new environment. Along the 30 min of the test, the exploration of the environment by the mice was followed by videotracking using EthoVision XT software (Noldus) giving access to the total distance traveled. Motor coordination and skilled walking deficits were analyzed at P45 using the Locotronic device (IntelliBI Innovations). Wild-type and eNMDAR−/− mice were placed on a departure zone and should cross a horizontal ladder to get into their cages. Bars (3 mm diameter) are spaced by 7 mm. On both sides of the ladder, infrared sensors allow to record animal missteps distinguishing between front and rear paws. The time to cross the ladder was also recorded. Weight curves from P2 to P15 and weights at adulthood (P45) were provided in the Supplementary Figure S7.
Statistical analysis
Statistical analyses were realized using the GraphPad Prism (version 9.0.0). Statistical tests used for each experiment, n values, number of females and males, and p values were summarized in Supplementary Table S2. Data concerning the cortical microvascular organization were expressed as a distribution between angle classes and were analyzed by a chi-square test. All other data was expressed as mean ± SEM and analyzed using unpaired t test or one-way, two-way, and three-way ANOVA followed by Bonferroni’s post-test (Table S2).
Results
Effect of eNMDAR invalidation on the expression of differentiation markers of the oligodendrocyte lineage
At the perinatal period, OPC entering the developing cortex use radial microvessels as guides and oligo-vascular interactions contribute to the differentiation fate of OPC (Palhol et al., 2023). Because this time window coincides with high expression of endothelial NMDA receptors (Legros et al., 2009), we investigated the impact of eNMDAR invalidation on the differentiation profile of oligodendrocytes. Western blot analysis performed on P2 cortices showed that expression of Olig2, a marker of the whole oligodendrocyte lineage, was significantly increased in eNMDAR−/− mice (Fig. 1A; ***p < 0.001). Expression of PDGFRα, a marker of OPC, was also significantly increased in eNMDAR−/− mice (Fig. 1B; *p < 0.05). No effect was found regarding the expression of CNPase, a marker of pre- and mature oligodendrocytes (Fig. 1C). MBP was not detected by Western blot at this stage (Fig. 1D). At P15, no significant differences were found on Olig2 expression between WT and eNMDAR−/− mice (Fig. 1E). PDGFRα expression was still higher in eNMDAR−/− mice even if not statistically significant (Fig. 1F). CNPase expression was significantly lower in eNMDAR−/− mice (Fig. 1G; ***p < 0.001). MBP was also detected by Western blot at P15, and its expression level was drastically lower in eNMDAR−/− mice (Fig. 1H; ***p < 0.001). At P45, Olig2 was no more detected by Western blot (Fig. 1I). No difference on PDGFRα expression was found between WT and eNMDAR−/− mice (Fig. 1J). Even if less pronounced than at P15, the expression of CNPase (*p < 0.05) and MBP (**p < 0.01) at P45 remained significantly lower in cortical extracts from eNMDAR−/− than from WT mice (Fig. 1K,L). Comparison of the relative expression levels of PDGFRα, CNPase, and MBP at P2, P15, and P45 showed different expression profiles between WT and eNMDAR−/− mice (Fig. 1M). PDGFRα expression, which was higher in eNMDAR−/− at P2 and P15, reached WT levels at P45 (Fig. 1M, circles). CNPase was detected at the three developmental stages. While no difference was found between the two strains at P2, CNPase expression was significantly lower in eNMDAR−/− at P15 and P45 (Fig. 1M, squares). MBP was firstly detected at P15 and its expression level was much lower in eNMDAR−/− than in WT mice. As found for CNPase, this effect persisted at P45, even if less pronounced (Fig. 1M, triangles). Altogether, these data suggest that invalidation of eNMDAR durably impaired the expression of differentiation markers of mature and myelinating oligodendrocytes.
Effect of endothelial NMDA receptor invalidation on the expression of differentiation markers of the oligodendrocyte lineage. A–D, Quantification by Western blot of the expression of Olig2 (A), PDGFRα (B), CNPase (C), and MBP (D) on P2 cortical extracts from WT and eNMDAR−/− mice. E–H, Quantification by western blot of the expression of Olig2 (E), PDGFRα (F), CNPase (G), and MBP (H) on P15 cortical extracts from WT and eNMDAR−/− mice. I–L, Quantification by Western blot of the expression of Olig2 (I), PDGFRα (J), CNPase (K), and MBP (L) on P45 cortical extracts from WT and eNMDAR−/− mice. M, Time course representation of PDGFRα (circles), CNPase (squares), and MBP (triangles) relative expression in eNMDAR−/− versus WT mice (dotted line). Data are expressed as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 compared with WT. Statistical details are provided in Table S2.
Effect of eNMDAR invalidation on the distribution pattern of cortical differentiating oligodendrocytes
Immunohistochemistry experiments were performed in WT and eNMDAR−/− mice at P2, P15, and P45 using Olig2 and PDGFRα antibodies to visualize the whole oligodendrocyte lineage and OPC, respectively (Fig. 2). In addition, CD31 immunolabeling was used to visualize cortical microvessels. At P2, density of Olig2-positive cells gradually decreased from the CC to superficial cortical layers (SCL) in both WT and eNMDAR−/− mice (Fig. 2A). Regarding PDGFRα, numerous tangential and radial positive cells were, respectively, present in the CC and in the developing cortical layers of both WT and eNMDAR−/− mice (Fig. 2B). Two-way ANOVA analysis of cell densities showed a cortical layer effect for both Olig2-positive (Fig. 2C; F 69.93; ****p < 0.0001) and PDGFRα-positive cells (Fig. 2D; F 154.5; ****p < 0.0001) whereas no Genotype effect was found (Fig. 2C,D; Olig2, p = 0.3469; ns; PDGFRα, p = 0.7139; ns). Similar double immunohistochemistry experiments were performed at P15 to visualize and quantify the density of Olig2- and PDGFRα-positive cells in the developing cortex (Fig. 2E–H). For both markers, densities gradually decreased from the CC to layer V and remained low and stable from layers IV to I (Fig. 2E–H). Two-way ANOVA analysis showed a Cortical layer effect on Olig2 and PDGFRα cell densities (Fig. 2G,H; Olig2, F 140.4; ****p < 0.0001; PDGFRα, F 48.80; ****p < 0.0001). Moreover, a Genotype effect was found for Olig2- and PDGFRα-positive cells (Fig. 2G,H; Olig2, F48.03; ****p < 0.0001; PDGFRα, F 20.00; ****p < 0.0001). In particular, density of PDGFRα-positive cells was significantly higher in the cortical layer VI of eNMDAR−/− mice (Fig. 2H; *p < 0.05). At P45, a Cortical layer effect was still observed for Olig2- and PDGFRα-positive cell densities (Fig. 2I–L; Olig2, F 70.66; ****p < 0.0001; PDGFRα, F14.11; ****p < 0.0001) while no Genotype effect was found anymore between WT and eNMDAR−/− mice (Fig. 2K,L; Olig2, F 3.258; p = 0.0783; ns; PDGFRα, F 1.205; p = 0.2797; ns). Altogether, these data suggest that invalidation of eNMDAR delayed the positioning of OPC entering the developing cortex.
Effect of endothelial NMDA receptor invalidation on the density and positioning of Olig2- and PDGFRα-positive cells in the developing cortex. A, B, Visualization by immunohistochemistry of Olig2- (A) and PDGFRα-positive (B) cells in cortices from WT and eNMDAR−/− mice at P2. Cortical microvessels are visualized using a CD31 antibody. C, D, Quantification of the Olig2- (C) and PDGFRα-positive (D) cell density in the CC, the deep cortical layers (DCL), and the superficial cortical layers (SCL) of WT and eNMDAR−/− mice at P2. E, F, Visualization by immunohistochemistry of Olig2- (E) and PDGFRα-positive (F) cells in cortices from WT and eNMDAR−/− mice at P15. Cortical microvessels are visualized using a CD31 antibody. G, H, Quantification of the Olig2- (G) and PDGFRα-positive (H) cell density in the CC and cortical layers VI to I of WT and eNMDAR−/− mice at P15. I, J, Visualization by immunohistochemistry of Olig2- (I) and PDGFRα-positive (J) cells in cortices from WT and eNMDAR−/− mice at P45. Cortical microvessels are visualized using a CD31 antibody. K, L, Quantification of the Olig2- (K) and PDGFRα-positive (L) cell density in the CC and cortical layers VI to I of WT and eNMDAR−/− mice at P45. Data are expressed as mean ± SEM. Two-way ANOVA test analyzed the cortical layer and genotype interaction followed by the Bonferroni's post-test. *p < 0.05, **p < 0.01, ***p < 0.0001 compared with WT. Statistical details are provided in Table S2.
Effect of eNMDAR invalidation on the microvessel association of OPC entering the neocortex and on endothelial MMP9-like activity
The visualization of the cortical vasculature at P2 by using CD31 immunohistochemistry showed a preferential organization of microvessels, most of them having a radial orientation (Fig. 3A, arrows). Chi-square analysis of the angle value classes (radial microvessels belonging to the [0–25°] class) revealed no effect of eNMDAR invalidation on the cortical microvascular network (Fig. 3B; χ2 0.6152, df 3p = 0.8929; ns). Triple immunolabeling experiments targeting PDGFRα, Olig2, and CD31 showed that 94.2 ± 0.5% of PDGFRα-positive cells associated to radial cortical microvessels are Olig2-positive supporting that PDGFRα-positive cells entering the developing cortex along radial microvessels are OPC (Fig. S2A–E). 3D maps of PDGFRα-positive cells and cortical microvessels showed tight associations between PDGFRα-positive cells entering the developing cortex and radial microvessels in WT and eNMDAR−/− mice (Fig. 3C,D; Movies 1, 2). Two-way ANOVA analysis revealed a Cortical layer effect (F 5.816; *p < 0.05) with higher densities of vessel-associated PDGFRα-positive cells in DCL and SCL (Fig. 3E) whereas no Genotype effect was found (Fig. 3E; F 2.174; p = 0.1576; ns). Functional analysis of microvessel activity was investigated by quantifying endothelial MMP9-like activity using in situ zymography in P2 cultured slices (Fig. 3F–J). Time-lapse acquisitions showed a MMP9-like activity in radial microvessels (Fig. 3F–H, arrows) as well as in nervous cells (Fig. 3F–H, arrowheads). qRT-PCR and confocal imaging coupled to a scanline analysis were used to control that MMP9 expression was not impacted in eNMDAR−/− mice (Fig. S2). Overlay of MMP9-like activity (FITC signal) with isolectin labeling (TRITC signal) allowed to focus on the endothelial activity (Fig. 3F–H). In WT cultured slices, glutamate (100 µM) induced a marked increase of the vascular MMP9-like activity in DCL (Fig. 3I, Fig. S3). Two-way ANOVA analysis of time-lapse acquisitions indicated Time (F 2.287; *p < 0.05) and Glutamate treatment (F 81.93; ****p < 0.0001) effects (Fig. 3I). This effect of glutamate (100 µM) was significantly blocked by the cell permeant MMP9 inhibitor II (F 55.5; ****p < 0.0001; Fig. S2G). In contrast, in P2 eNMDAR−/− mice, glutamate (100 µM) had no stimulatory effect on the endothelial MMP9-like activity (Fig. 3I, Fig. S4) with no Time (F 0.8187; p = 0.6307; ns) and no Treatment (F 2.686; p = 0.1052; ns) effects. Statistical analysis comparing the two strains showed a Genotype effect (Fig. 3J; F 9.488; **p < 0.01). Similar analysis was done for the MMP9-like activity detected in nervous cells (Fig. S5). Double immunohistochemistry experiments indicated that the MMP9-like activity colocalized with NeuN-positive cells supporting that nervous cells presenting a MMP9-like activity belong to the neuronal lineage (Fig. S5A–I). No Treatment effect of glutamate (100 µM; F 0.02562; p = 0.8748; ns) and no Genotype effect (F 2.057; p = 0.1708; ns) were found (Fig. S5J,K). Altogether, these data support that invalidation of the eNMDAR did not modify the cortical vessel organization, the association of OPC to cortical microvessels, and the MMP9 expression but suppressed the capacity of glutamate to stimulate the endothelial MMP9-like activity.
Effect of endothelial NMDA receptor invalidation on the vessel association of PDGFRα-positive cells in the developing cortex. A, Visualization by CD31 immunolabeling of the cortical vasculature of a WT neonate at P2. Arrows point the preferential radial organization of the cortical microvessels. B, Distribution of the angular orientation of cortical microvessels in WT and eNMDAR−/− mice at P2. The angle class [0–25] corresponds to vessels with a radial orientation. Histograms represent proportion of vessels in each angle class. Statistical analysis was done using chi-square test (df0.6152 = 3, p = 0.8929; ns). C, D, Visualization of PDGFRα-positive cells entering the developing cortex in close association with radial microvessels in eNMDAR−/− neonate at P2 (C). The dotted insert indicates the ROI selected for the 3D Imaris map (D). E, Quantification of PDGFRα-positive cells associated to radial microvessels in WT and eNMDAR−/− mice at P2. Data are expressed as mean ± SEM. Two-way ANOVA test analyzed the cortical layer and genotype interaction followed by the Bonferroni's post-test. F–H, Visualization of endothelial (arrows) and neuronal (arrowheads) MMP9-like activities by in situ zymography in cultured organotypic slices from WT neonate at P2. Gelatinase activity (F) and microvessels (G) are visualized by incubation of the cultured slice with DQ-Gelatin-FITC and isolectin-TRITC, respectively. Fluorescent signals are then overlapped (H). I, Quantification by time-lapse acquisition of the effect of glutamate (100 µM) on the endothelial MMP9-like activity in organotypic slices from WT and eNMDAR−/− neonates at P2. Two-way ANOVA test analyzed the time and treatment interaction for a given genotype. J, Quantification of the effect of glutamate on the endothelial MMP9-like activity in organotypic slices from P2 WT and eNMDAR−/− neonates at 3 h. Two-way ANOVA test analyzed the genotype and treatment interaction at a given time point. *p < 0.05 compared with WT Ctrl. #p < 0.05 compared with WT Glut. Statistical details are provided in Table S2.
Effect of eNMDAR invalidation on the cortical network of MBP-positive fibers
Western blot experiments revealed a marked dysregulation of MBP expression in eNMDAR−/− mice at P15 (Fig. 1H). MBP being a structural protein essential for axonal myelination (Boggs, 2006), we first analyzed the cortical network of MBP-positive fibers at P15 (Fig. 4, Fig. S6). In P15 WT mice, a dense network of radial MBP-positive fibers extending from layers VI to IV was detected in the maturing cortex (Fig. 4A,B). Few MBP-positive cell bodies were observed (Fig. 4C, arrowhead). In eNMDAR−/− mice, the cortical network of MBP-positive fibers was strongly impacted with a low density of myelinated fibers particularly obvious in layers V and IV (Fig. 4D,E,G, layer IV, **p < 0.01). Moreover, whereas the whole cortical thickness was similar between WT and eNMDAR−/− mice (Fig. 4I), the unmyelinated cortical thickness was significantly increased in eNMDAR−/− mice (Fig. 4J, unmyelinated thickness, **p < 0.01) resulting in a marked increase of the unmyelinated/whole cortical thickness ratio (Fig. 4K, ***p < 0.001). Moreover, eNMDAR−/− mice were characterized by numerous isolated MBP-positive cell bodies observed from layers VI to IV (Fig. 4E,F,H, arrowheads; layer VI, *p < 0.05; layers V and IV, ***p < 0.001). The thickness of the CC was also quantified and appeared significantly reduced (Fig. 4L; **p < 0.01). No sex effect was found when females and males were considered separately (Fig. S6). The same morphometric analysis was performed at P45, and no significant differences were found anymore between the two genotypes (Fig. 5A–G). Altogether, these data support that, at the histological level, the cortical myelination process occurring in the inside-out direction was delayed in eNMDAR−/− mice.
Effect of endothelial NMDA receptor invalidation on the MBP-positive fiber network at P15. A, Low magnification microphotograph visualizing radial MBP-positive fibers in the somatosensory cortex of WT mice at P15. Note that the fiber network extends from layers VI to IV. The insert identifies the ROI shown at higher magnification in B. B, C, Microphotographs visualizing MBP-positive fibers at the interface between cortical layers V and IV. Note that at this stage, immunolabeling is essentially observed in fibers. Few cell bodies are observed (C; arrowhead). D, Low magnification microphotograph visualizing radial MBP-positive fibers in the somatosensory cortex of eNMDAR−/− mice at P15. Note that the fiber network extends from layers VI to mid-layer V. The insert identifies the ROI shown at higher magnification in E. E, F, Microphotographs visualizing MBP-positive fibers in cortical layer V. Note that at this stage, immunolabeling is observed in fibers and also in isolated cell bodies (D–F; arrowheads). G, Quantification of the density of MBP-positive fibers in the somatosensory cortex of WT and eNMDAR−/− mice at P15. Two-way ANOVA test analyzed the cortical layer and genotype interaction followed by the Bonferroni's post-test. H, Quantification of the density of MBP-positive cell bodies in the somatosensory cortex of WT and eNMDAR−/− mice at P15. Two-way ANOVA test analyzed the cortical layer and genotype interaction followed by the Bonferroni's post-test. I, Quantification of the whole thickness of the sensorimotor cortex in WT and eNMDAR−/− mice at P15. J, Quantification of the unmyelinated cortical thickness in WT and eNMDAR−/− mice at P15. K, Ratio of the unmyelinated cortical thickness with the whole cortical thickness in WT and eNMDAR−/− mice at P15. L, Quantification of the CC thickness in WT and eNMDAR−/− mice at P15. Data are expressed as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 compared with WT. Statistical details are provided in Table S2.
Effect of endothelial NMDA receptor invalidation on the MBP-positive fiber network at P45. A, B, Low magnification microphotographs visualizing the network of MBP-positive fibers in the sensorimotor cortex of WT (A) and eNMDAR−/− (B) mice at P45. Note that the fiber network extends from cortical layers VI to III in both genotypes. C, Quantification of the density of MBP-positive fibers in the somatosensory cortex of WT and eNMDAR−/− mice at P45. Two-way ANOVA test analyzed the cortical layer and genotype interaction followed by the Bonferroni's post-test. D, Quantification of the whole cortex thickness. E, Quantification of the unmyelinated cortex thickness. F, Ratio of the unmyelinated cortical thickness with the whole cortical thickness in WT and eNMDAR−/− mice at P45. G, Quantification of the CC thickness in WT and eNMDAR−/− mice at P45. Data are expressed as mean ± SEM. Statistical details are provided in Table S2.
Effect of eNMDAR invalidation on axonal myelination
Analysis of both cortical MBP-positive layers and CC thickness at P15 and P45 having shown marked developmental differences between both genotypes, we went to explore more precisely myelin sheath by transmission electron microscopy at both stages (Figs. 6, 7). In P15 WT mice, low magnification microphotographs showed that axons from the CC were myelinated (Fig. 6A). In eNMDAR−/− mice, whereas several axons were myelinated (Fig. 6B), images revealed the presence of numerous unmyelinated axons (Fig. 6B, arrowheads). Quantification indicated that the proportion of myelinated axons was significantly decreased in eNMDAR−/− mice (Fig. 6E; ***p < 0.001). Moreover, the diameter of myelinated axons (excluding myelin sheath) was also impacted (Fig. 6F). Whereas in WT mice most myelinated axons had a diameter between [0.4–0.6] µm, in eNMDAR−/− mice, the axonal diameter was significantly higher and ranging from [0.6–0.8] µm (chi-square analysis df16.49 = 5; **p < 0.01). Higher magnification images were used to visualize and analyze myelin sheath (Fig. 6C,D, dotted lines in brackets). Myelin sheath thickness was significantly reduced in eNMDAR−/− mice compared with WT mice (Fig. 6G; **p < 0.01). The effects of endothelial invalidation of NMDAR on axonal diameter and myelin sheath resulted in a significant increase of the g-ratio (Fig. 6H; ***p < 0.001). Correlation representation showed that, in eNMDAR−/− mice, axons with high diameters had thin myelin sheath thicknesses (Fig. 6I). Moreover, mitochondria density in axonal sections was significantly increased in eNMDAR−/− mice (Fig. 6J; **p < 0.01). At P45, transmission electron microscopy images showed very similar ultrastructural patterns between WT and eNMDAR−/− mice, quite all axons being myelinated (Fig. 7A,B). Quantitative analyses revealed no significant difference between the two genotypes regarding the density of myelinated axons and the diameters of myelinated axons (Fig. 7E,F). Using higher magnification images (Fig. 7C,D), no differences were found between WT and eNMDAR−/− mice regarding myelin sheath thickness, g-ratio, and mitochondria density (Fig. 7H–J). These data indicate that the decrease of MBP-positive fibers observed in eNMDAR−/− mice at P15 is associated with myelin sheath and axonal impairments. The absence of obvious ultrastructural defects at P45 supports a delayed maturation trajectory in eNMDAR−/− mice.
Effect of endothelial NMDA receptor invalidation on axonal myelination in the CC at P15. A, B, Low magnification microphotographs visualizing myelinated axons in CC of WT (A) and eNMDAR−/− (B) mice. Arrowheads indicate unmyelinated axons. Inserts identify ROI shown at higher magnification in C and D. C, D, Microphotographs visualizing myelinated axons in WT (C) and eNMDAR−/− (D) mice. Arrows indicate axonal mitochondria and dotted lines in brackets visualize myelin sheath thickness. E, Ratio of myelinated versus total axons in CC of WT and eNMDAR−/− mice at P15. F, Distribution of diameters of myelinated axons in CC of WT and eNMDAR−/− mice at P15. Axon diameters are distributed in six classes ranging from [0.2–0.4] to [>1.2] µm. Statistical analysis was done using chi-square test (df16.49 = 5; **p < 0.01). G, Quantification of the myelin sheath thickness in CC of WT and eNMDAR−/− mice at P15. H, Quantification of g-ratio in CC of WT and eNMDAR−/− mice at P15. I, Correlative representation of g-ratio and axon diameters in CC of WT and eNMDAR−/− mice at P15. J, Quantification of mitochondria density per myelinated axons in CC of WT and eNMDAR−/− mice at P15. Data are expressed as mean ± SEM. **p < 0.01, ***p < 0.001 compared with WT. Statistical details are provided in Table S2.
Effect of endothelial NMDA receptor invalidation on axonal myelination in the CC at P45. A, B, Low magnification microphotographs visualizing myelinated axons in CC of WT (A) and eNMDAR−/− (B) mice. Note that contrasting to P15, quite all axons are myelinated. Inserts identify ROI shown at higher magnification in C and D. C, D, Microphotographs visualizing myelinated axons in WT (C) and eNMDAR−/− (D) mice. Dotted lines in brackets visualize myelin sheath thickness. E, Density of myelinated axons in CC of WT and eNMDAR−/− mice at P45. F, Distribution of diameters of myelinated axons in CC of WT and eNMDAR−/− mice at P45. Axon diameters are distributed in six classes ranging from [0.2–0.4] to [>1.2] µm. Statistical analysis was done using chi-square test (chi-square, df = 0.2074, 5; p = 0.9990). G, Quantification of the myelin sheath thickness in CC of WT and eNMDAR−/− mice at P45. H, Quantification of g-ratio in CC of WT and eNMDAR−/− mice at P45. I, Correlative representation of g-ratio and axon diameters in CC of WT and eNMDAR−/− mice at P45. J, Quantification of mitochondria density per myelinated axons in CC of WT and eNMDAR−/− mice at P45. Data are expressed as mean ± SEM. ns, not statistically different. Statistical details are provided in Table S2.
Effect of eNMDAR invalidation on sensorimotor and locomotor activities
In human, neonatal white matter lesions can lead to motor disabilities ranging from sensorimotor deficits to cerebral palsy (Rumajogee et al., 2016). Because in the present study the molecular, morphometric, and functional analysis were performed in the sensorimotor cortex, we investigated whether the developmental phenotypes characterized in eNMDAR−/− mice were associated to sensorimotor troubles. In pups, the sensorimotor activity was evaluated by using the righting reflex and the negative geotaxis tests that require motor coordination (limb-limb coordination; Feather-Schussler and Ferguson, 2016). Regarding the righting reflex test, the latency for WT neonates to flip onto their feet from a supine position starts to decrease at P5 to reach a minimum at P8 (Fig. 8A). During the same tested period, the decrease of latency to flip was not observed in eNMDAR−/− mice (Fig. 8A; P6, *p < 0.05; P7 and P8, ****p < 0.0001). Two-way ANOVA analysis showed a Genotype effect (Fig. 8A; F 28.70; ****p < 0.0001). No sex effect was found when females and males were considered separately (Fig. S7C). The negative geotaxis test was performed from P5 to P12 (Feather-Schussler and Ferguson, 2016). Latency for WT neonates to move from a facing down position in a slope to a facing up position start to decrease at P8 (Fig. 8B). Two-way ANOVA analysis showed no Genotype effect (Fig. 8B; F 0.02527; p = 0.8738; ns) even if latency to move face up was significantly increased in eNMDAR−/− pups at P9 (Fig. 8B; ****p < 0.0001). No sex effect was found when females and males were considered separately (Fig. S7D). Spontaneous locomotor activity was assessed at P15 and P45 (Fig. 8C,D). At P15, the total distance traveled by WT mice in 30 min was 3,573.33 ± 191.50 cm (Fig. 8C, Movie 3). In contrast, the total distance covered by eNMDAR−/− mice was drastically lower (Fig. 8C; 347.65 ± 60.02 cm; ****p < 0.0001; Movie 4). No sex effect was found when females and males were considered separately (Fig. S7E). At P45, the total distance traveled by WT mice was higher than at P15 (Fig. 8D; 7,370 ± 253.4 cm; Movie 5). Similarly, the locomotor activity of eNMDAR−/− mice was higher at P45 than at P15 (Fig. 8D; 4,718 ± 259.4 cm; Movie 6) but remained significantly lower than that of WT mice at the same stage (Fig. 8D; ****p < 0.0001). No sex effect was found when females and males were considered separately (Fig. S7F). Motor coordination and skilled walking deficits were assessed at P45 using the horizontal ladder test (Fig. 8E–I; Metz and Whishaw, 2002). Typical data tables provided records on the number of paw errors and the time needed when crossing the ladder for WT (Fig. 8E) and eNMDAR−/− mice (Fig. 8F). The total number of step errors was significantly higher in eNMDAR−/− mice (Fig. 8G; ****p < 0.0001). Interestingly, in eNMDAR−/− mice, the number of missed steps was differently impacted between front and back paws (Fig. 8H; ***p = 0.0002). Similarly, the time needed to cross the ladder was significantly increased in eNMDAR−/− mice (Fig. 8I; ***p = 0.0002). No sex effects on both step errors and time to cross the ladder were found when females and males were considered separately (Fig. S7E–I). Altogether, these data indicate that invalidation of the endothelial NMDA receptor resulted in neonatal sensorimotor impairments and persisting motor disorders in adulthood.
Effect of endothelial NMDA receptor invalidation on sensorimotor tests in neonates and on locomotor activity in P15 and P45 mice. A, Quantification of the latency to restore a normal prone position (righting reflex) in neonates from WT (continuous line) and eNMDAR−/− (dotted line) mice. Two-way ANOVA test analyzed the age and genotype interaction; Bonferroni's post-test was performed. B, Quantification by the negative geotaxis test of the latency for WT (continuous line) and eNMDAR−/− (dotted line) neonates to move from a downward toward an upward position (180° rotation). Two-way ANOVA test analyzed the age and genotype interaction; Bonferroni's post-test was performed. C, Visualization and quantification using the open field test of the distance covered during 30 min by WT and eNMDAR−/− mice at P15. D, Visualization and quantification using the open field test of the distance covered during 30 min by WT and eNMDAR−/− mice at P45. E, F, Typical scoring tables obtained with the horizontal ladder test for a female WT (E) and a female eNMDAR−/− (F) mice. Bars indicate the total (orange), front (yellow), and back (blue) paw errors. The blue line indicates the total sensor activation time. G, Quantification of the total number of missed steps. H, Quantification of front and back paw errors in WT and eNMDAR−/− mice. Note that in eNMDAR−/− mice, most errors result from front paws. I, Quantification of the ladder crossing time for WT and eNMDAR−/− mice. Data are expressed as mean ± SEM. *p < 0.05, ***p < 0.001, ****p < 0.0001 compared with WT. Statistical details are provided in Table S2.
Recording of the open field test for P15 WT mice. [View online]
Recording of the open field test of P15 eNMDAR−/− mice. [View online]
Recording of the open field test of P45 WT mice. [View online]
Recording of the open field test of P45 eNMDAR−/− mice. [View online]
Discussion
Endothelial NMDA receptor and differentiation of oligodendrocytes
In humans, perinatal life constitutes a critical window of vulnerability for oligodendrocytes (Motavaf and Piao, 2021). In particular, OPCs are especially affected by hypoxic and/or inflammatory events, suggesting stage-dependent susceptibility (Back, 2017; Janowska et al., 2024). Mechanistically, in vitro and in vivo studies have demonstrated that glutamate receptors, including NMDA receptors, contribute to this vulnerability (Manning et al., 2008). In the developing cortex, previous studies have shown that OPCs require close interactions with radial microvessels (Brosolo et al., 2022), and their vascular attachment/detachment influences differentiation (Su et al., 2023). Moreover, neonatal endothelial cells from radial cortical microvessels express functional NMDA receptors (Fig. 9A; Legros et al., 2009; Léger et al., 2020a). Taken together, this prompted the hypothesis that the endothelial NMDA receptor (eNMDAR) regulates OPC positioning and differentiation in the neocortex. Using a CRE/LOX strategy targeting Ve-cadherin (Vestweber, 2007) and the Grin1 subunit of the NMDA receptor (Tsien et al., 1996), this study showed that eNMDAR deletion affected oligodendrocyte differentiation markers in an age-dependent manner. At P2, cortical expression of PDGFRα, an OPC marker, was higher in eNMDAR−/− mice than wild-type (WT) mice. This phenotype was reduced at P15 and disappeared by P45. In contrast, CNPase and MBP expressions, markers for immature and myelinating oligodendrocytes, respectively, were lower in eNMDAR−/− mice at P15, and this reduction persisted at P45, though attenuated. These results suggest that eNMDAR deletion disrupts oligodendrocyte differentiation, particularly the OPC to immature stage transition (Long et al., 2021).
Graphical abstract summarizing main findings. A, Previous study showed that neonatal cortical endothelial cells express functional eNMDAR (Legros et al., 2009). We developed Grin1lox/lox/VeCadCre (eNMDAR−/−) mice. B, Functional activity. Glutamate stimulates MMP9-like activity in radial cortical microvessels. This effect is abolished in eNMDAR−/− mice. C, Positioning. In WT mice, OPC entering the neocortex are associated to radial cortical microvessels. In eNMDAR−/− mice, positioning of OPC is delayed in deep cortical layers (DCL). D, Differentiation. In eNMDAR−/− mice, CNPase and MBP are underexpressed, and development of MBP-positive fiber network is delayed with isolated MBP cell bodies. E, Myelination. In eNMDAR−/− mice, myelin sheath thickness, g-ratio, and axonal mitochondrial density are impaired in CC. F, Motor tests in young mice. eNMDAR−/− mice have motor disabilities (righting reflex test and locomotor activity). G, Motor tests in adult mice. eNMDAR−/− mice present walking deficits at adulthood.
Endothelial NMDA receptor and microvessel–OPC interactions
Immunohistological data revealed a significant increase in the density of PDGFRα-positive cells in cortical layer VI of eNMDAR−/− mice at P15, a distribution pattern not observed at P45. As OPC migrate into the developing cortex along radial microvessels during the first postnatal week (Su et al., 2023), this suggests a role for eNMDAR in microvessel–OPC interactions. There is now considerable evidence that the vasculature supports proper cortical positioning of several cell types, including GABAergic interneurons (Won et al., 2013; Léger et al., 2020a) and oligodendrocytes (Tsai et al., 2016), which use microvessels during their migration. Recent studies have indicated that the vessel association/detachment contribute to oligodendrocytes differentiation (Su et al., 2023). Several molecules and signaling pathways have been shown to be involved in their attachment/detachment such as the endothelial cell-derived CXCL12 and its receptor CXCR4, guidance molecules like netrins, semaphorins, ephrins, Slit, and WNT proteins (Morimoto et al., 2023; Wälchli et al., 2023). In this study, 3D morphometric analysis showed neither vascular architecture nor the number of vessel-attached OPCs were altered by eNMDAR deletion, indicating that this receptor did not affect these two processes. However, MMP, which degrade the extracellular matrix, are a protein family actively involved in controlling cell migration in both physiological and pathological contexts (Luo, 2005; Van Hove et al., 2012). For example, it has been shown that during cortical development, MMP9 expression peaks perinatally (E17-P5; Bednarek et al., 2009) and that, in MMP9 null mice, the process outgrowth of oligodendrocytes is significantly retarded (Oh et al., 1999). Similarly, cerebellar granule cell precursor migration is delayed in MMP-9-deficient mice (Vaillant et al., 2003). In the present study, in situ zymography showed that, in eNMDAR−/− mice, glutamate failed to stimulate the endothelial MMP9 activity along radial microvessels (Fig. 9B). Altogether, these results are consistent with increased PDGFRα-positive cell density in cortical layer VI and reinforce the hypothesis of a delay in the positioning of OPC in the developing cortex of eNMDAR−/− mice at P15 (Fig. 9C). Interestingly, in a pathological context, NMDA receptor enhances MMP-like activity in glioma to promote dissemination (Ramaswamy et al., 2014) and glioma cells migrate along blood vessels (Pastorino et al., 2023). Consequently, these findings in a pathological context may reflect a dysregulated physiological function of the eNMDAR during development.
Endothelial NMDA receptor and cortical myelination
While vascular association is critical for OPC dispersal, their positioning and detachment appear equally important for initiating myelination (Su et al., 2023; Foerster et al., 2024). Given (1) the lack of glutamate-induced MMP9-like activity and (2) the transient increased of OPC density in layers VI and V at P15 concomitant with reduced expression of MBP, it was tempting to hypothesize that the myelination process was impaired in eNMDAR−/− mice. MBP immunohistochemistry revealed two key observations. First, in P15 eNMDAR−/− mice, the CC thickness and the MBP fibers density in upper layer V and layer IV were significantly reduced (Fig. 9D). Second, numerous isolated MBP-positive cells were observed in the upper layer V and layer IV supporting that the myelination process could be impaired (Fig. 9D). In mice, oligodendrocyte progenitors are generated embryonically but myelination begins postnatally (Doretto et al., 2011). Concurrently, axonal diameter progressively increases with postnatal development, and most axons with 0.5 µm diameter and higher are myelinated (Suminaite et al., 2019). In this study, myelinated axon diameter distribution shift from the range of 0.4–0.6 µm in WT mice to 0.6–0.8 µm in eNMDAR−/− mice. These results suggest that, in eNMDAR−/− mice, the myelination initiation started later, when axon diameters were larger. Consistent with this, the myelin sheath thickness was also significantly reduced in eNMDAR−/− mice. The g-ratio is a quantitative measure of relative myelin thickness, considering axonal diameter. It serves as a structural index of axonal myelination and reflects optimal axonal conductance (Chomiak and Hu, 2009). In 2009, Chomiak and Hu calculated the optimal g-ratio value in the rat CC to be between 0.75 and 0.8 (Chomiak and Hu, 2009). Here, the WT mice had a mean g-ratio value of 0.79 whereas in eNMDAR−/− mice, it was significantly higher indicating an imbalance between myelin sheath thickness and axonal diameter. Moreover, during axonal growth, mitochondrial trafficking ensures an adequate supply of ATP (Sheng, 2017), and a progressive decrease in axonal mitochondrial motility has been observed with the maturation of cortical axons, both in vitro and in vivo (Lewis et al., 2016). In this study, the mitochondria density in axonal sections was markedly increased in P15 eNMDAR−/− mice, suggesting that eNMDAR deletion resulted in impaired axonal maturation. At P45, myelination patterns normalized across genotypes. Taken together, the present results strongly support that eNMDAR deletion causes delayed axonal myelination (Fig. 9E), possibly leading to motor deficits.
Endothelial NMDA receptor and motor disorders
In human, preterm birth constitutes a major risk factor of white matter lesions, impairing myelination and leading to motor deficits ranging from developmental coordination disorders to cerebral palsy (Spittle et al., 2011). Preclinical models showed that preterm birth coincides with a vulnerable window for cortical oligodendrocyte differentiation (Back, 2017; Brosolo et al., 2022). Thus, we investigated whether impaired myelination observed in eNMDAR might result in motor disabilities by focusing on neonatal sensorimotor tests, locomotor activity, and motor coordination. During perinatal life, the righting reflex assesses a pup's ability to flip from supine. In this study, and in contrast to WT, eNMDAR−/− mice showed no decreased latency to flip during the testing period (P2–P8). Since the righting reflex is a spinal cord reflex potentiated by the cerebral cortex (Feather-Schussler and Ferguson, 2016), these results are consistent with molecular and cellular data showing an altered oligodendrocytes maturation and myelination (Fig. 9E). The negative geotaxis test is an unlearned response to directional movement against gravitational cues, used to evaluate early motor development and vestibular function. The average age for the appearance of the negative geotaxis reflex in rodents is P7 (Feather-Schussler and Ferguson, 2016). In contrast to the righting reflex test, no genotype effect was found between WT and eNMDAR−/− mice during the testing period. These differences between the righting and negative geotaxis tests suggest different neuronal circuit contribution: cerebellar in geotaxis (Wang et al., 2018) versus medullar/cortical in righting (Martes et al., 2024). At P15, the open field test revealed a drastic locomotor activity decrease in eNMDAR−/− mice (Fig. 9F), with partial recovery by P45. In addition, the horizontal ladder test, assessing motor coordination and skilled walking (Metz and Whishaw, 2002), revealed long-term disorders in eNMDAR−/− mice (Fig. 9G), especially in front paws. In human preterm neonates, white matter lesions disrupt myelination and motor coordination (Spittle et al., 2011). Altogether, these data suggest that eNMDAR would contribute to cerebral palsy pathophysiology by affecting oligodendrocyte trajectory entering the neocortex via vessel-associated mechanisms. Supporting this, a clinical study from Makki and coworkers showed upper limb deficits in 83% of cerebral palsy cases (Makki et al., 2014).
In conclusion, this study provides the first evidence that eNMDAR deletion in mice impairs oligodendrocyte differentiation, positioning, and myelination trajectories in the developing sensorimotor cortex. These defects are linked to endothelial dysfunction. Although most histological/ultrastructural defects resulting from the eNMDAR deletion resolve in adulthood, long-term motor impairments persist. These findings highlight eNMDAR's role during a critical neonatal window for oligodendrocyte maturation and its potential contribution to long-term motor disorders. These opens new avenues for research into eNMDAR's influence on cortical oligodendrocyte differentiation and offers new hypothesis to understand motor disabilities in human preterm neonates with white matter damage.
Footnotes
This work was supported by Rouen University, France; Institut National de la Santé et Recherche Médicale (INSERM, UMR1245), France; Rouen University Hospital, France; the Inserm US51 for access to Cell Imaging (PRIMACEN) and behavior (SCAC) platforms; Blood & Brain Institute Caen, France (OligVasc); France-BioImaging infrastructure (https://ror.org/01y7vt929) supported by the French National Research Agency (ANR-24-INBS-0005 FBI BIOGEN). A.B. is a recipient of a fellowship from the Normandy Region.
↵*B.J.G. and Ma.L. contributed equally to this work.
The authors declare no competing financial interests.
This paper contains supplemental material available at: https://doi.org/10.1523/JNEUROSCI.0199-25.2025
- Correspondence should be addressed to Bruno J. Gonzalez at bruno.gonzales{at}univ-rouen.fr.
