l-Glutamate is a major excitatory neurotransmitter that binds ionotropic and metabotropic glutamate receptors. Cerebral endothelial cells from many species have been shown to express several forms of glutamate receptors; however, human cerebral endothelial cells have not been shown to express either the N-methyl-d-aspartate (NMDA) receptor message or protein. This study provides evidence that human cerebral endothelial cells express the message and protein for NMDA receptors. Human cerebral endothelial cell monolayer electrical resistance changes in response to glutamate receptor agonists, antagonists, and second message blockers were tested. RT-PCR and Western blot analysis were used to demonstrate the presence of the NMDA receptor. Glutamate and NMDA (1 mM) caused a significant decrease in electrical resistance compared with sham control at 2 h postexposure; this response could be blocked significantly by MK-801 (an NMDA antagonist), 8-(N,N-diethylamino)-n-octyl-3,4,5-trimethyoxybenzoate (an intracellular Ca2+ antagonist), and N-acetyl-l-cystein (an antioxidant). Trans(±)-1-amino-1,3-cyclopentanedicarboxylic acid, a metabotropic receptor agonist (1 mM), did not significantly decrease electrical resistance. Our results are consistent with a model where glutamate, at excitotoxic levels, may lead to a breakdown in the blood brain barrier via activation of NMDA receptors.
l-glutamate is an excitatory neurotransmitter (33) that functions through its binding to several classes of receptors: 1) dl-α-amino-3-hydroxy-5-methylisoxazole propionic acid and kainic acid receptors, 2) metabotropic receptor, and 3) N-methyl-d-aspartate (NMDA) receptors (NMDAR) (15). Glutamate is present in plasma at concentrations ranging from 18 to 25 μM (11) and as high as 3 mM in the parenchymal cell stores (8, 9, 12). However, under ischemic or traumatic conditions, the glutamate concentration levels in the brain interstitial space can increase 55-fold (3). Extracellular glutamate is primarily cleared by the action of a family of Na+-dependent high-affinity transporters located on glutamatergic neuronal processes as well as on glial cells (12). However, during ischemia, the ATP depletion that occurs causes a marked flooding of glutamate into the extracellular space increasing the glutamate concentration to as high as 1 mM (2).
During stroke and trauma, neural injury within the cerebrum is provoked by cerebral oxygen and glucose deprivation, resulting in the excessive release of stored synaptic glutamate. This excessive synaptic release is due to the loss of ATP stores and dissipation of membrane ion gradients, leading to potassium efflux and membrane depolarization. Eventually, anoxia triggers a massive depolarization and the opening of voltage-dependent sodium channels. As a result, glutamate is released by synaptic exocytosis and trapped in the interstitium due to the reversal of the glutamate transporters flooding the synaptic space with glutamate. This response, in turn, leads to the massive overstimulation of NMDA receptors, referred to as “glutamate excitotoxicity” (10, 19, 25, 27). In addition to neuronal excitotoxicity, increased extracellular glutamate may also contribute to “vasogenic edema” (characterized by an increase in microvascular solute permeability) (1, 14, 30).
The overabundance of glutamate during these forms of stress activates ionotropic NMDA receptors (ligand-gated calcium channels). The resulting calcium influx leads to the production of reactive oxygen species (ROS), causing a further release of intracellular glutamate (20). Glutamate excitotoxicity therefore represents a vicious physiological cycle that occurs in and among neurons (19, 25).
While the mechanisms by which excitotoxicity leads to “cytotoxic edema” (neuronal cell swelling) have been examined in depth (16, 17, 30), the process though which excitotoxicity might lead to “vascular edema” is not known (30, 39). However, because the NMDA receptor blockers MK-801 and ifenprodil have been shown to block the vascular edema formation seen in trauma or neurotoxic injury (10, 24) and because porcine (28) and rat (18) cerebral endothelial cells have been shown to express NMDA receptors, excessive glutamate receptor activation might have a direct deleterious effect on cerebral endothelial functions, like barrier.
It has recently been shown that polymorphonuclear leukocytes, on stimulation by inflammatory stimuli in the cerebral circulation, will release glutamate and, through stimulation of metabotropic glutamate receptors (mGluR), causes a loss of the endothelial barrier (7). Collard et al. (7), also describe that the activation of the mGluRI or III results in a time-dependent loss of phosphorylated vasodilator-stimulated phosphoprotein (VASP), suggesting that VASP phosphorylation mediates glutamate-dependent increases in endothelial permeability, illustrating that the cerebral endothelium is capable of responding to glutamate directly.
Therefore, this current study tests the hypothesis that “excitotoxic” levels of glutamate lead to an NMDA or metabotropic glutamate receptor-dependent cerebral endothelial barrier dysfunction. Our data are most consistent with the activation of the NMDA family of glutamate receptors (rather than metabotropic receptors) and suggest that activation of the NMDA receptor may contribute to glutamate-dependent loss of cerebral endothelial barrier function.
Reagents. Medium 199, insulin, transferrin, selenium, heparin, HEPES, MK-801, l-glutamic acid, 8-(N,N-diethylamino)-n-octyl-3,4,5-trimethoxybenzoate (TMB-8), NMDA, trans-(±)-1-amino-1,3-cyclopentanedicarboxylic acid (tACPD), and N-acetyl-l-cysteine (NAC) were purchased from Sigma (St. Louis, MO). All experiments were performed at 37°C, with 7.5% CO2, and with the use of HBSS plus glucose (catalog no. H1387, Sigma) and 15 mM HEPES (pH 7.4).
Cell culture. Human brain endothelial cells (HBEC) were purchased from ScienCell Research Laboratories (San Diego, CA) and were maintained in human brain endothelial medium (ScienCell). HBEC were used for RT-PCR analysis only. Immortalized HBEC (IHEC) were supplied by Dr. D. Stanimirovic and were maintained in medium 199 (with 10% FCS, 1% antibiotic/antimycotic, 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenium, and 600 USP units/1 heparin) and grown to 100% confluency. IHEC were seeded to 8-μm pore inserts (24-well format, VWR; West Chester, PA) for barrier studies. All endothelial cultures were used on confluency.
Western blot analysis. IHEC protein samples (25 μg each) were separated on 7.5% SDS-PAGE, transferred to nitrocellulose membranes, and blocked overnight in 5% milk. Membranes were then incubated for 24 h with anti-human NMDAR1 and NMDAR2A/B (Chemicon; Temecula, CA) polyclonal antibodies at a 1:250 dilution in 0.1% milk. The membranes were then washed three times (5 min each wash) and developed for 24 h in goat anti-mouse alkaline phosphatase secondary antibody (Sigma; St. Louis, MO) was added at a 1:1,000 dilution in 0.1% milk. The membranes were then reacted with nitroblue tetrazolium-5-bromo-4-chloro-3-indolyl phosphate chromogen to visualize the proteins.
RT-PCR. The presence of NMDAR1 mRNA was determined in both primary HBEC and IHEC. DNAse-treated RNA (1 μg) was converted to cDNA with the use of RT and amplified with the following primer set: (NMDAR1) sense primer, 5′-GATGTCTTCCAAGTATGCGGA-3′, and antisense primer, 5′-GGAATCTCCTTCTTGACCAG-3′. The PCR mixture was amplified for 30 cycles with the use of a three-step protocol: denaturation at 94°C (1 min), annealing at 50°C for 1 min, and elongation at 72°C for 1 min. The corresponding 667-bp (NMDAR1) product was separated in a 1.5% agarose gel stained with ethidium bromide and viewed and analyzed with the use of Alpha Innotech gel documentation system (San Leandro, CA). RNA extracts from the human brain (cerebrum) were used as the positive control for all experiments. Negative controls were performed in which the RT enzyme was replaced with RNAse-free water.
Immunoprecipitation. A RIPA lysis buffer was placed on the cultured IHEC and incubated for 30 min on ice. The lysate was then transferred to a conical tube and spun at 10,000 g for 10 min at 4°C. The supernatant was then transferred to a fresh tube and incubated with normal mouse serum for 1 h on ice. The lysate was then used to resuspend 100 μl of packed cell volume of fixed Staphylococcus aureus Cowan I (SAC), and this SAC/lysate slurry was incubated for 30 min on ice. The SAC/lysate mixture was then spun for 15 min at 10,000 g at 4°C. The supernatant was removed and put into a clean tube, and 1:50 of antibody (supplied by Dr. R. P. Yasuda, Department of Pharmacology, Georgetown University) was added and incubated overnight at 4°C. After the overnight incubation, 100 μl of protein A bead suspendion (10% vol/vol in lysis buffer) were added to the antibody-antigen reaction and incubated for 1 h on ice. We then collected the beads by centrifugation at 10,000 g for 15 s at 4°C. The immune complexes were then washed three times with lysis buffer. A sample buffer was added and heated to 85°C for 10 min and the sample buffer/bead solution was spun at 10,000 g for 15 s. The supernatant was then loaded onto a 7.5% gel, and Western blot analysis was performed.
Transendothelial resistance analysis: barrier function. All barrier function experiments were done in HBSS plus 15 mM HEPES at 37°C and 7.5% CO2. An epithelial voltohmeter (EVOM, World Precision Instruments; Sarasota, FL) was used to analyze IHEC cultures on 8-μm pore inserts for changes in electrical barrier. Cell-coated chambers were placed into a Boyden chamber and allowed to adjust for 1 h to the HBSS + HEPES solution, and then changes in electrical resistance, a model of barrier function (26), were recorded in response to glutamate (at 0.01, 0.1, and 1 mM), NMDA (1 mM), and tACPD (1 mM), up to time = 3 h. All test agents were added to their final concentration at time 0 (end of the 1-h adjustment period). In some experiments cells were pretreated with NAC (2 mM) or MK-801 (10 μM) for 10 or 20 min, respectively. All experiments were performed at least n = 4 times and baseline resistance of the monolayers had to be a minimum of 150 Ω/0.33 cm2 before the insert was used. Typically, the monolayer resistance of this cell line ranged from 150 to 269 Ω/0.33cm2 with an average of 183.89 ± 24.875 Ω/0.33cm2 (means ± SD). Data were expressed as “percent baseline” resistance, which allowed cumulative statistical analysis to be done and account for unpreventable batch-to-batch variations.
Statistical analysis. All values are expressed as means ± SD. Data were analyzed using one-way ANOVA with Bonferroni's correction for multiple comparisons. Significance was accepted at P < 0.05.
Glutamate barrier studies. All barrier experiments were performed in quadruplicate. The electrical resistance of the IHEC monolayers was tested in response to the amino acid glutamate (at 0.01, 0.1, and 1 mM concentrations). It has been demonstrated that the 1 mM concentration is acceptable to model stroke or excitotoxic events (8, 9, 12, 22, 38). It was observed that over 3 h after glutamate exposure, there was a decrease in the monolayer electrical resistance at all three concentrations; however, only the 1 mM dose became significant compared with control (n = 4, P < 0.001; Fig. 1). Glutamate at the 1 mM concentration produced a time-dependent decrease in the electrical resistance, which became significant by 2 h (P < 0.001.
NMDA and tACPD on barrier. On the basis of the observation that 1) glutamate appears to contribute to vascular edema when at excitotoxic levels (1, 14, 30) and 2) NMDA receptors have been shown to be present on porcine (28) and rat cerebral endothelium (18), we hypothesized that the observed decrease in endothelial resistance was due to stimulation/activation of an NMDA-like receptor. Brain endothelial barrier electrical resistance was similarly decreased in response to both 1 mM glutamate and 1 mM NMDA (n = 3, P < 0.001, Fig. 2) compared with untreated controls at 3 h, becoming significant by 2 h (n = 3, P < 0.001; Fig. 2). Treatment with either 1 mM glutamate or NMDA was not significantly different from one another (Fig. 2). This decrease in resistance is not likely to be a result of cell death or toxicity because lactate dehydrogenase release did not change within 3 h after 1 mM glutamate treatment (sham control = 3.94% of total lactate dehydrogenase release and 1 mM glutamate treatment = 4.6%).
However, it has also been reported that metabotropic glutamate receptors are present and functional on human cerebral microvascular cells (7); therefore, we treated the cells with tACPD (a pan metabotropic glutamate receptor agonist) (34). tACPD (1 mM) did not cause a decrease in electrical resistance (n = 4, Fig. 2).
HBEC and IHEC RT-PCR and Western blot analysis for NMDAR1. Some of the previous studies, which came to the conclusion that NMDAR1 was not present on human endothelial cells, used primer sets that were specific for rats but not for humans (23) and also used oligo(dT) instead of random hexamer. With the use of primer sets designed for human NMDAR1 and random hexamer instead of oligo(dT), we were able to demonstrate that HBEC and IHEC both contained message for NMDAR1 (Fig. 3). A sequence analysis showed that our PCR product was 100% homologous with human NMDAR1.
On the basis of the controversial literature surrounding the RT-PCR results for the NMDAR subunits, we decided it was necessary to demonstrate the presence of these subunits by both Western blot analysis and immunoprecipitation. Western blot analysis demonstrated that IHEC express protein for both the NMDAR1 and its subunit NMDAR2A/B [band 1 is the expected band for NMDAR1 at ∼116 kDa and band 2 is the N-deglycosylated NMDAR1 band at ∼103 kDa, which have been described by Genever et al. (13)] (Fig. 4) and immunoprecipitation verified the NMDAR1 Western blot result (Fig. 4).
Barrier disruption is NMDAR-Ca2+ and ROS induced. To test whether the glutamate-mediated decrease in transendothelial resistance (TER) was due to changes in intracellular Ca2+ and ROS concentrations, both associated with NMDA receptor activation during excitotoxicity, monolayers were pretreated with TMB-8, an intracellular Ca2+ antagonist (21), and the glutathione-repleting antioxidant NAC (36, 40). TMB-8 (0.1 mM) significantly blocked the decrease in electrical barrier caused by 1 mM glutamate implicating calcium in this response (n = 3, Fig. 5). NAC (2 mM) significantly blocked the decrease in TER associated with 1 mM glutamate exposure (n = 3, Fig. 5). MK-801, a selective noncompetitive NMDAR1 blocker, was also effective in blocking the barrier dysfunction caused by 1 mM glutamate (n = 3, Fig. 5).
At least three different families of receptors bind glutamate and mediate responses to glutamate: 1) dl-α-amino-3-hydroxy-5-methylisoxazole propionic acid/kainic acid, 2) NMDA, and 3) metabotropic receptors. The objective of the present study was to evaluate the effect of “excitotoxic” levels of glutamate on brain endothelial barrier function (15–17). As stated earlier, Collard et al. (7) reported that leukocytes, on stimulation, will release glutamate, and that when this stimulation occurs at the cerebral vasculature, metabotropic receptors located on the brain endothelial cells are activated. Collard et al. (7), by using immunofluorescence and an in vitro endothelial permeability model, showed that activation of metabotropic glutamate receptors (mGluR I and III) leads to the phosphorylation of VASP and diminish cerebral endothelial barrier (7).
In this study, we modeled brain endothelial barrier with the use of a modified voltohmeter (26), which permits the continuous, instantaneous measurement of endothelial electrical resistance changes produced by pharmacological stimulants (NMDA, glutamate, tACPD, and MK-801) and several second mechanism-based inhibitors (NAC and TMB-8).
Our data demonstrate a loss of brain endothelial barrier function in response to both glutamate and NMDA (1 mM), but not in response to the metabotropic glutamate receptor agonist tACPD (1 mM, Fig. 2). However, tACPD transiently and significantly increased monolayer electrical resistance compared with untreated controls (Fig. 2). This brief increase in barrier mediated by tACPD returned to baseline control values between 120 and 180 min, suggesting that cerebral endothelial cells express functional metabotropic glutamate receptors. Because tACPD did not decrease the electrical barrier, but glutamate and NMDA (a specific NMDA receptor agonist) did reduce the barrier (Figs. 1 and 2), our data suggest that, in our cell model, loss of brain endothelial barrier is more consistent with an NMDA-type glutamate receptor-mediated effect rather than a metabotropic glutamate receptor effect (7).
To further characterize this response, monolayers were pretreated with MK-801, a highly selective and noncompetitive NMDAR1 blocker, before treatment with glutamate. MK-801 (10 μM) significantly attenuated the loss in endothelial electrical barrier produced by 1 mM glutamate (Fig. 5), which also argues for participation of NMDAR1 receptors.
NMDA receptors are ionotropic receptors and act as ion channels (Ca2+ >> K+ and Na+) on glutamate/NMDA binding. Stimulation of NMDAR1 receptors will increase the intracellular calcium concentration of cells expressing this receptor. With respect to junctional barrier, it has previously been shown that increased cytosolic calcium can lead to 1) changes in junctional organization, 2) changes in cell tension, and 3) changes in cell morphology, all of which can lead to junctional permeability (31). Consequently, we used TMB-8, an intracellular Ca2+ antagonist, to reduce Ca2+-dependent glutamate-mediated barrier failure. Pretreatment of monolayers with 10–5 M TMB-8 significantly reduced changes in barrier produced by 1 mM glutamate (Fig. 5). Therefore, Ca2+ appears to play some role in the endothelial response to glutamate and is also consistent with NMDAR1-mediated effects.
Persistent stimulation of NMDA receptors during excitotoxicity is associated with the production of ROS (20). Because elevated cytoplasmic Ca2+ can trigger mitochondrial production of ROS (superoxide, peroxynitrite, and hydroxyl radical), disorganize endothelial junctions, and reduce cerebral endothelial barrier (4, 5, 29, 35), we pretreated brain endothelial monolayers with the glutathione-repleting antioxidant NAC before glutamate stimulation. NAC blocked the decrease in resistance caused by glutamate (Fig. 5) indicating that endothelial responses to glutamate are at least partially due to the development of an intracellular oxidant stress.
Therefore, Ca2+ and ROS both appear to contribute to this response. As stated before, barrier changes are Ca2+ dependent and may reflect 1) changes in junctional organization, 2) changes in cell tension, and 3) changes in cell morphology. Reciprocally, ROS can mobilize cell Ca2+ stores and lead to the opening of calcium channels, both of which can drive endothelial barrier changes (32, 37). Currently, because both TMB-8 and NAC block changes in electrical resistance with a similar time course, we cannot elucidate the exact order of these events.
Of the three receptors that bind glutamate, only the NMDA family of glutamate receptors is activated by NMDA. Because monolayers exposed to NMDA show a significant loss of electrical barrier (identical to that seen with equimolar glutamate), our data are consistent only with decreased endothelial barrier function as being dependent on an NMDA-type receptor.
We could not conclude that stimulation of NMDA receptors on endothelial cells necessarily mediated this response because the IHEC cell line (used in this study) had previously been reported (23) to not express ionotropic glutamate NMDAR subunits (1, 2A, and 2B). It is critical to note that in that particular study, the PCR primers were not optimized for human samples (23), and a rat primer set was used to amplify human NMDAR1 transcripts.
In our study, by using a random hexamer priming strategy instead of oligo(dT) and human NMDAR1 optimized primer sets in the RT-PCR process, we were now able to detect message for NMDAR1 (Fig. 3). This amplified message corresponds to the anticipated size product for NMDAR1, which was also seen in human cerebral tissue and primary human cerebral endothelial cells. This was confirmed by Western blot analysis, which showed protein expression for both NMDAR1 and NMDAR2A/B (Fig. 4), but two bands were detected on the Western blots (a band at ∼103 kDa and the anticipated band at ∼116 kDa). To demonstrate that we were observing the presence of NMDA receptor protein, we performed immunoprecipitation and only the 103-kDa band was seen (Fig. 4). According to Genever et al. (13), the band at ∼103 kDa represents the N-deglycosylated form of the NMDA receptor, which is the primary form present in human tissue. Genever et al. (13) suggest that “deglycosylation” of the receptor allows for enhanced stability and specific orientation and localization within the tissue.
Our results are consistent with a model where glutamate (1 mM), at “excitotoxic” levels, decreases the electrical barrier of brain microvascular endothelial cells. This effect appears to be due to the activation of NMDARs on these cells. The fact that there was not a loss of barrier in response to tACPD treatment suggests that metabotropic glutamate receptors do not play a role in our model of glutamate excitotoxicity; however, this may be culture dependent or may vary within different anatomical regions of the brain (23).
Because our cells respond similarly to NMDA and glutamate (1 mM), but not tACPD, and are blocked by MK-801, we believe that our data support a role for NMDAR1 in the cerebral endothelial barrier dysfunction seen during excitotoxicity. This observation suggests that glutamate may contribute to forms of brain injury through an NMDA receptor-mediated loss of endothelial barrier and that endothelial NMDA receptors may be an important prophylactic and therapeutic target in stroke and other associated cerebral traumas.
The authors thank Dr. Danica Stanimirovic for supplying the brain endothelial cells used in this study and Dr. Robert Yasuda for insights and antibody contributions.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2003 by the American Physiological Society