SEARCH

SEARCH BY CITATION

Keywords:

  • central nervous system;
  • L-glutamate pharmacology;
  • endothelial glutamate;
  • efflux;
  • blood-brain barrier model

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The concentration of the excitotoxic amino acid, L-glutamate, in brain interstitial fluid is tightly regulated by uptake transporters and metabolism in astrocytes and neurons. The aim of this study was to investigate the possible role of the blood-brain barrier endothelium in brain L-glutamate homeostasis. Transendothelial transport- and accumulation studies of 3H-L-glutamate, 3H-L-aspartate, and 3H-D-aspartate in an electrically tight bovine endothelial/rat astrocyte blood-brain barrier coculture model were performed. After 6 days in culture, the endothelium displayed transendothelial resistance values of 1014 ± 70 Ω cm2, and 14C-D-mannitol permeability values of 0.88 ± 0.13 × 10−6 cm s−1. Unidirectional flux studies showed that L-aspartate and L-glutamate, but not D-aspartate, displayed polarized transport in the brain-to-blood direction, however, all three amino acids accumulated in the cocultures when applied from the abluminal side. The transcellular transport kinetics were characterized with a Km of 69 ± 15 μM and a Jmax of 44 ± 3.1 pmol min−1 cm−2 for L-aspartate and a Km of 138 ± 49 μM and Jmax of 28 ± 3.1 pmol min−1 cm−2 for L-glutamate. The EAAT inhibitor, DL-threo-ß-Benzyloxyaspartate, inhibited transendothelial brain-to-blood fluxes of L-glutamate and L-aspartate. Expression of EAAT-1 (Slc1a3), −2 (Slc1a2), and −3 (Slc1a1) mRNA in the endothelial cells was confirmed by conventional PCR and localization of EAAT-1 and −3 in endothelial cells was shown with immunofluorescence. Overall, the findings suggest that the blood-brain barrier itself may participate in regulating brain L-glutamate concentrations. © 2012 Wiley Periodicals, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Homeostasis of the neurotransmitter and excitotoxic amino acid, L-glutamate, in the brain interstitial fluid (ISF) is maintained by the concerted action of transporters and metabolic pathways (Danbolt,2001). Removal of L-glutamate from the ISF via sodium-dependent excitatory amino acid transporters (EAATs) is necessary to maintain the concentration within the lower micromolar range for neuronal signalling to occur and toxicity to be avoided (Frandsen and Schousboe,1993). It has been firmly established that astrocytes express EAATs and that these play a significant role in the removal of L-glutamate from the ISF (Schousboe and Waagepetersen,2005). L-glutamate is taken up by astrocytes mainly via EAAT-1 (Slc1a3) and EAAT-2 (Slc1a2), although EAAT-3 (Slc1a1) has also been shown in cortical astrocytes (Bunch et al.,2009; Conti et al.,1998). In the astrocytes L-glutamate is either metabolized to the nontoxic amino acid, L-glutamine, or enters the tri-carboxylic acid cycle (Danbolt,2001). Neurons also express EAAT-3 and EAAT-4 (Slc1a6) on the postsynaptic membrane, thus keeping glutamate concentrations low in the synaptic cleft (Massie et al.,2001; Rothstein et al.,1994). The importance of EAAT-mediated removal of L-glutamate has been underlined by invivo studies, which showed that homozygous EAAT-2 knockout mice had 50% mortality at 6 weeks of age and an increased frequency of seizures resembling NMDA-induced seizures (Tanaka et al.,1997).

The role of the blood-brain barrier (BBB) in L-glutamate homeostasis is somewhat controversial. The BBB is formed by the small capillaries of the brain and acts as a dynamic barrier for the entrance of xenobiotics and nutrients into the central nervous system (CNS) (Reese and Karnovsky,1967). Part of this barrier function is due to tight junctions, which prevent paracellular transport and a range of transporters that regulate transcellular BBB permeation (Brightman and Reese,1969; Crone and Olesen,1982; Wolburg et al.,2009).

The concentrations of most amino acids in the cerebrospinal fluid are between 2 and 18% of that found in plasma (Abbott et al.,2010; Hawkins et al.,2006). Teichberg and coworkers reviewed studies concerning BBB efflux of L-glutamate and suggested that the BBB most likely plays a much larger and more active role than previously anticipated (Teichberg et al.,2009). An active efflux of L-aspartate and L-glutamate has been demonstrated after cerebral microinjection in rat brain in vivo (Hosoya et al.,1999). Moreover, reducing the blood concentration of L-glutamate in vivo, by accelerating degradation through glutamate-oxaloacetate transaminase in the blood, decreases brain L-glutamate concentration and size of infarct volume after middle cerebral artery occlusion in rats (Campos et al.,2011a). The transendothelial L-glutamate efflux has been suggested to be caused by EAATs on the abluminal membrane of the capillaries of the BBB (Hawkins,2009; O'Kane et al.,1999), with conflicting results regarding subtype expression pattern (Lyck et al.,2009; Uchida et al.,2011), and a not yet characterized luminal efflux pathway. Recently, polarized abluminal-to-luminal transport of L-glutamate was demonstrated in a porcine brain endothelium/astrocyte coculture model, however, the general EAAT inhibitor, DL-threo-ß-Benzyloxyaspartate (TBOA), showed only limited inhibition of the transport (Cohen-Kashi-Malina et al.,2012).

The aim of this study was to investigate the transendothelial transport of L-glutamate and L-aspartate, and to estimate the kinetics and capacity of the EAATs under physiological relevant circumstances.

Active transport of L-glutamate and L-aspartate was observed in the abluminal-to-luminal direction and inhibited by TBOA. D-aspartate was not effluxed, but accumulated in the endothelial cells when applied from the abluminal side. Transendothelial brain-to-blood L-glutamate and L-aspartate fluxes followed Michaelis–Menten kinetics in combination with passive paracellular diffusion. RT-PCR and immunofluorescence demonstrated the presence of EAAT-1 and EAAT-3 in the endothelial cells. This study thus demonstrates for the first time the kinetics and capacity of transendothelial abluminal to luminal efflux of excitatory amino acids in an in vitro BBB model, and supports previous studies, showing the presence of EAATs in the abluminal membrane of brain endothelial cells. This further supports the notion that the BBB participates in the regulation of brain ISF L-glutamate levels during pathologic events, where brain L-glutamate concentrations are increased.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Materials

The radioisotopes 3H-L-aspartic acid (specific activity 12.9 Ci/mmol), 3H-D-aspartic acid (specific activity 11.3 Ci/mmol), 3H-L-tryptophan (specific activity 17.9 Ci/mmol), 3H-L-glutamic acid (specific activity 49.6 Ci/mmol) and 14C-D-mannitol (specific activity 58.5 mCi/mmol) were purchased from Perkin–Elmer (Hvidovre, Denmark). All other chemicals and reagents were bought from Sigma-Aldrich (Rødovre, Denmark) unless otherwise stated.

Isolation of Primary Bovine Brain Endothelial Cells and Rat Astrocytes

Brain capillary endothelial cells (BCEC) and astrocytes were isolated and cultured as described previously (Helms et al.,2010). In brief, bovine brains were acquired from calves (below 12 months of age) from two slaughterhouses (Mogens Nielsen Kreaturslagteri A/S, Herlufmagle, Denmark and Aarhus Slaughterhouse, Aarhus, Denmark) and transported in phosphate buffered saline on ice. Meninges were removed, and the gray matter was isolated, homogenized in Dulbecco's Modified Eagles Medium-AQ supplemented with 10% foetal bovine serum, 10 mL/L nonessential amino acids (×100) and 100 U/mL/100μg/mL penicillin/streptomycin solution (DMEM). The homogenizer was a 40 mL Dounce tissue grinder (Wheaton Science Products, Millville). The homogenate was filtered through 160 μm mesh filters (Millipore, Copenhagen, Denmark). Capillaries were resuspended in DMEM and digested for 1 hour in an enzyme mix of DNAse I (170 U/mL), Collagenase type III (200 U/mL) and Trypsin TRL (90 U/mL) (Worthington Biochemical Corporation, Lakewood). The suspension was filtered through 200 μm mesh filters (Merrem and La Porte, Zaltbommel, The Netherlands), resuspended in foetal bovine serum:dimethylsulfoxide (9:1) and frozen overnight in a −80°C freezer. The capillaries were subsequently transferred to liquid nitrogen for long time storage.

Astrocytes were isolated from 3 to 4 days old Sprague Dawley rats (Taconic, Ejby, Denmark). Cerebella and midbrains were removed and the cerebral cortices were homogenized by trituration and incubated in Trypsin-EDTA in DMEM without serum and NaHCO3 with TES (50 mM). The suspension was filtered through 120 μm and 45 μm mesh filters (Merrem and La Porte, Zaltbommel, The Netherlands), and the filtrate was seeded in T75 flasks (1 flask for every 2 pups) and cultured until confluence (37°C, 10% CO2). When confluent, the flasks were shaken overnight at room temperature to detach oligodendrocytes. The confluent astrocytes were passaged to Poly-D-Lysine coated flasks (split ratio 1:3) and cultured for 2 weeks. Astrocyte conditioned medium (ACM) was collected three times a week. The astrocytes were passaged with Trypsin-EDTA, resuspended in foetal bovine serum:dimethylsulfoxide (9:1) and frozen overnight in a −80°C freezer (two vials pr. flask). The astrocytes were moved to liquid nitrogen for long time storage.

Coculture of Bovine Endothelial Cells and Rat Astrocytes

T75 flasks were coated with collagen type IV and fibronectin. Frozen bovine brain microvessels were thawed, seeded in the flasks and cultured for 4 days (37°C, 10% CO2) in DMEM:ACM (1:1) supplemented with 125 μg/mL heparin. The endothelial cells were passaged with a brief trypsinization and seeded at a density of 100,000 cells/filter insert on collagen/fibronectin coated Transwell polycarbonate filter inserts (Area = 1.12 cm2, pore radius = 0.4 μm, Corning Life Sciences, New York), where astrocytes had been seeded at the bottom 2 days before (130,000 cells/filter insert). The coculture model was cultured for 3 days in DMEM supplemented with 125 μg/mL heparin followed by 3 days of culture in DMEM without NaHCO3 (Gibco, Breda, The Netherlands), supplemented with 8-(4-CPT)-cyclic adenosine monophosphate (312.5 μM), dexamethasone (0.5 μM), RO-20-1724 (17.5 μM), and TES (50 mM). A schematic model illustration of the final coculture model is below (Fig. 1).

thumbnail image

Figure 1. A schematic representation of the blood-brain barrier model. Endothelial cells (squares) are cultured on permeable filter supports with astrocytes (stars) cultured on the bottom of the inserts. Transendothelial transport experiments were performed as indicated with arrows, abluminal-to-luminal and luminal-to-abluminal.

Download figure to PowerPoint

TEER-Measurements, Transport, and Uptake Studies

The transendothelial electrical resistance (TEER) was measured before the experiments, using an Endohm 12 cup electrode chamber (World Precision Instruments, Sarasota, Florida) connected to a Millicell-ERS device (Millipore, Massachusetts). The pooled value of TEER from all the conducted experiments was 1014 ± 70 Ω × cm2 (n = 23, N = 4–19). TEER in individual batches, cultured separately varied from 618 ± 225 (N = 5) to 1986 ± 164 Ω × cm2 (N = 4). Cultures which displayed resistance values below 500 Ω × cm2 were discarded.

Transport and uptake experiments were performed in the culture media unless otherwise stated, in order not to disrupt the fragile cocultures. 1 μCi/mL of either 3H-L-aspartic acid, 3H-D-aspartic acid, 3H-L-tryptophan or 3H-L-glutamic acid were added to either the luminal or abluminal solution together with 1 μCi/mL 14C-D-mannitol. The solutions of 3H-L-aspartic acid and 3H-L-glutamic acid were supplemented with nonlabeled compound to obtain the concentrations stated for the individual experiments. In some experiments TBOA (Tocris, Bristol, United Kingdom) or methionine sulfoximine (MSO) was added to give final concentrations of 50 μM and 10 mM, respectively. Culture trays were placed on a temperature-controlled shaking table at 37°C and 90 rpm. Receiver samples were taken after 30, 60, 90, 120, and 150 minutes and donor samples after 150 minutes. Samples were transferred to Ultima Gold scintillation fluid (Perkin–Elmer, Hvidovre, Denmark). The permeable supports with the cocultures were washed thrice in ice cold Hanks balanced salt solution, removed from the inserts with a scalpel, and transferred to scintillation fluid. Radioactivity was counted in a Tri-Carb 2100 TR Liquid Scintillation Analyzer (Packard Instrument Company, Meriden).

Transport studies were performed in both luminal to abluminal (L-A) and abluminal to luminal (A-L) direction. The overall mean mannitol permeability in all the experiments conducted was 0.88 ± 0.13 × 10−6 cm s−1 (n = 21, N = 4–19). The mean mannitol permeability of different batches varied from 0.28 ± 0.042 × 10−6 cm s−1 (N = 4) to 2.83 ± 1.76 × 10−6 cm s−1 (N = 5).

Accumulation studies were performed under sodium-free conditions. The culture medium was replaced with Hanks balanced salt solution with choline chloride as a replacement for sodium chloride before addition of radiolabeled substrate. Experiment was run for 150 minutes, where after the radioactive solution was withdrawn and filter inserts were treated as described earlier.

Quantitative Analysis of Amino Acid Concentrations in the Culture Media

Culture media from both luminal and abluminal chambers after 6 days of culture were analyzed for concentrations of amino acids, before addition of radiolabeled compounds. Samples were taken before the experiments and added acetonitrile in a 1:3 ratio. The extracted amino acids were subjected to precolumn o-phthaldialdehyde derivatization and separated on an Agilent Eclipse AAA column (4.6 mm × 150 mm, particle size 5 μm) using a Shimadzu 10A VP as described by Geddes and Wood (Geddes and Wood,1984). The derivatized amino acids were separated using a linear gradient of 5% acetonitrile in citrate phosphate buffer pH 5.9 (mobile phase A) and 90% acetronitrile (mobile phase B) at a flow rate of 1 mL/min. The gradient of mobile phase B increased from 5 to 50% in 22 minutes with a total run time of 38 minutes. Samples were quantified with a Shimadzu RF-10AXL fluorescence detector (excitation 350 nm; detection 450 nm). Data analyses were performed employing Microsoft Excel 2007 and GraphPad Prism v5.01 software's. The concentrations of L-aspartic acid were 1.8 ± 1.1 μM in the luminal medium and 1.3 ± 0.2 μM in the abluminal medium (N = 3). L-glutamate concentrations were 10.7 ± 3.4 μM in the luminal medium and 8.5 ± 0.1 μM in the abluminal medium, respectively (N = 3). The L-glutamate and L-aspartate concentrations given in the figures and tables do not include this contribution.

Reverse Transcriptase-Polymerase Chain Reaction and Immunocytochemistry

Total RNA was isolated from rat astrocytes, bovine brain cortex homogenate, freshly isolated bovine brain capillaries, endothelial cells after passage from culture flasks and endothelial cells after 6 days of coculture, using Total RNA Isolation Reagent according to the manufacturer's protocol (ABgene, Epsom, United Kingdom). Genomic DNA was removed with DNAse I Amplification grade according to manufacturer's protocol (Sigma-Aldrich, Steinheim, Germany) and reverse transcription was performed with MMLV high performance reverse transcriptase according to manufacturer's protocol (Epicentre, Maddison, Wisconsin). Polymerase chain reactions were carried out using HotStarTaq Plus DNA Polymerase according to manufacturer's protocol (Qiagen, Copenhagen, Denmark). The polymerase was activated at 95°C for 5 minutes and subsequently PCR was run using 35 cycles with 60 seconds annealing, 60 seconds extension and 30 seconds denaturation. Primers designed to match the bovine homologues of EAAT-1, EAAT-2, and EAAT-3 were purchased from Invitrogen (Hellerup, Denmark). The resulting primers are shown in Table 1. Reaction products were run on 1.5% agarose gels together with a 50 bp DNA ladder (Invitrogen, Hellerup, Denmark) and amplicons were visualized using ethidium bromide (0.5 μg/mL) in a fluorchemQ image station (Alpha Innotech/Cell Biosciences, Santa Clara, CA) to ensure that the primer products had the predicted size.

Table 1. Overview of Applied Primers
Primers
TargetForward strandReverse strandProduct size
EAAT-1 (Slc1a3)GCTTGCTCATCCATGCTGTTGAAGGTGATGGGTAGGGTG150
EAAT-2 (Slc1a2)TGCTCCTTATTCTGACGGCTTGGTGTCCAGCTCAGACTTG162
EAAT-3 (Slc1a1)CAGCAACACTGCCTGTCACTCAACGCTCAAGTCCAAATCA174
GFAP (Gfap)GACAGGAAGCAGATGAAGCCCCAGTTTGACGTTGAGCAGA555

The cocultured endothelial cells were treated with antibodies staining against EAAT-1 (Ab41751, Abcam, Cambridge, United Kingdom) (1:50) and EAAT-3 ((C-20)sc-7761, Santa Cruz, Heidelberg, Germany) (1:20) as well as Alexa 488 conjugated phalloidin (Molecular Probes, Leiden, The Netherlands) (1:200). The cells were fixed and permeabilized on the filter inserts with 4% paraformaldehyde + 0.2% Triton X-100 and subsequently blocked in PBS supplemented with 3% bovine serum albumin and 0.1% Tween 20. The filter inserts were removed, treated with RNAse and incubated with the relevant antibody overnight. The cells were incubated 30 minutes with goat anti-rabbit igG (5 μg/mL) (EAAT-1), rabbit anti-goat igG (5 μg/mL) (EAAT-3) or phalloidin coupled with Alexa 488 and subsequently briefly with propidium iodide (1.5 μM) (Molecular Probes, Leiden, The Netherlands) before confocal laser scanning microscopy examination with a Zeiss LSM 510 laser confocal microscope (Carl Zeiss, Jena, Germany).

Data Analysis

The transport data was plotted as total amount transported per cm2 against time. Fluxes (J) were calculated from the steady state slopes of the straight lines (accumulated amino acid as a function of time) divided by the cross-sectional area of transport. Apparent permeability values were calculated using Eq. (1).

  • equation image(1)

where Papp is the apparent permeability, J is the observed steady-state flux and Cdonor is the added concentration to the donor compartment.

Transendothelial L-A fluxes of L-glutamate and L-aspartate had a carrier-mediated component and a passive component, which could be described by a Michaelis–Menten-type equation combined with Ficks law [Eq. (2)].

  • equation image(2)

where J is transendothelial flux, Jmax is the maximal carrier-mediated flux, Km is the Michaelis constant of the transendothelial carrier-mediated transport pathway and P is the permeability of the passive paracellular flux. To determine Km and Jmax values of the A-L carrier-mediated L-glutamate and L-aspartate fluxes, the passive paracellular component of the total flux value was estimated from L-A fluxes. The passive flux component was subsequently subtracted from A-L fluxes to estimate Km and Jmax.

Data are reported as means ± standard error (SE) when results are obtained from different batches. n denotes the number of experiments on different batches. N denotes the number of filter inserts with cells, used in each experiment. In some experiments, mean values were calculated from a single batch with minimum three individual filter inserts and are thus represented as mean ± standard deviation (SD).

When appropriate, means were compared using a two-tailed student's t-test. Accumulated radiolabel inside the cells was calculated and compared using ANOVA followed by Dunnett's multiple comparison test with the abluminal accumulation as the control column.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The Endothelial/Astrocyte Cocultures Expressed EAAT-1, EAAT-2, and EAAT-3 mRNA and Protein at Day 6 of Culture

The expression and localization of EAAT-1 and EAAT-3 proteins were investigated in freshly isolated bovine brain capillaries and in endothelial cells after 6 days of coculture (Fig. 2).

thumbnail image

Figure 2. Immunofluorescence staining of freshly isolated bovine brain capillaries (A, B) and bovine brain endothelial cells on day 6 of coculture (CE). Green = EAAT-1 (Slc1a3) (A+C), EAAT-3 (Slc1a1) (B+D) or filamentous actin (E). Red = cell nuclei. Bars = 2 μm (capillaries, left panel), 5 μm (capillaries, right panel), 50 μm (endothelial cells, upper panel) or 5μm (endothelial cells, lower panel).

Download figure to PowerPoint

The capillaries showed a positive signal for EAAT-1 (Fig. 2A), whereas no signal was observed for EAAT-3 (Fig. 2B). The surrounding brain tissue showed positive EAAT-3 staining, which indicated that the antibody was specific and able to recognize the bovine homologue of EAAT-3. However, positive immunostaining for both EAAT-1 (Fig. 2C) and EAAT-3 (Fig. 2D) was observed in the endothelial cells on day 6 of coculture. EAAT-1 was distributed throughout the cells, but intense vesicular staining was observed in the proximity of the nuclei. XZ scanning could not determine if the signal was localized at the luminal or abluminal membrane because of the thin endothelial cells (approximately 2–3 μm) (lower panel). EAAT-3 was also distributed throughout the cells, and showed a strong intensity at the junctional regions. At present we have no explanation for the junctional localization of EAAT-3, but similar staining patterns for breast cancer resistance protein (BCRP) and the transferrin receptor have been observed (data not shown). Staining of filamentous actin confirmed that the stained cells were endothelial cells and not contaminated by astrocytes, which would have presented a different morphology. The actin fibers were concentrated at cell borders and the cells showed the typical endothelial morphology with large, flat cells in a tight monolayer (Fig. 2E). The differences between EAAT-3 protein expression in capillaries and endothelial cells in coculture indicate an effect of the coculture procedure on the EAAT expression pattern. This was further examined by conventional RT-PCR (Fig. 3).

thumbnail image

Figure 3. DNA products from RT-PCR with the primers indicated in Table 1. cDNA was isolated from rat astrocytes, freshly isolated capillaries, endothelial cells after passage from culture flasks (EC day 0), cocultures on day 6 and cortex homogenate. PCR was run for 35 cycles and bands were detected on a 1.5% agarose gel stained with ethidium bromide.

Download figure to PowerPoint

EAAT-1, EAAT-2, and EAAT-3 were all present in the brain homogenate, which served as a positive control for the primers. The capillaries showed high presence of EAAT-1, and EAAT-2, but only a faint band from EAAT-3, correlating with the lack of signal from the EAAT-3 antibody. The capillaries could, however, be associated with pericytes or astrocytes, which could give false high signals for EAAT-1 and EAAT-2. We therefore measured the expression of excitatory amino acid transporter mRNA in endothelial cells cultured for 4 days in flasks and subsequently trypsinized, a procedure which yields pure endothelial cells as previously demonstrated using staining with von Willebrands factor (Helms et al.,2010). Endothelial cells obtained immediately after trypsinization showed a high EAAT-1, mRNA expression and low EAAT-2, and EAAT-3 expression, which indicate that the high EAAT-2 expression in the capillaries might originate from astrocytic endfeet (Fig. 3). The endothelial cells showed an up-regulation of EAAT-3 mRNA expression after 6 days in coculture, consistent with the immunofluorescence investigation, which further supports that the EAAT-3, expression could be induced by the culture. RT-PCR using mRNA from rat astrocytes resulted in low intensity products from EAAT-1 and EAAT-2. The intensity was markedly lower in the astrocytes than in the coculture, whereby astrocyte contamination cannot be the main cause for the observed reaction products from the coculture mRNA.

L-Aspartate and L-Glutamate, but not D-Aspartate, Displayed Polarized Transport Across the Blood-Brain Barrier in the Brain to Blood Direction

The transendothelial transport of the radiolabeled excitatory amino acids 3H L-aspartate and 3H L-glutamate, as well as 3H D-aspartate, 3H L-tryptophan and 14C D-mannitol across the endothelial/astrocyte coculture was measured in the L-A and A-L directions (Fig. 4).

thumbnail image

Figure 4. (A, B) Transport of 3H labeled L-aspartate and L-glutamate across the endothelial/astrocyte coculture as a function of time, in the luminal-to-abluminal (▪) and abluminal-to-luminal direction (▴). (A) Transport of L-aspartate (5 μM), (B) Transport of L-glutamate (5 μM). Data are means ± SEM for L-aspartate (n = 6–10, N = 2–3) and L-glutamate (n = 2–5, N = 2–3). (C) Flux ratios (A-L/L-A) of 3H-labeled amino acids L-aspartate (Asp), D-aspartate (D-Asp), L-glutamate (Glu), L-tryptophan (L-Trp) and 14C labeled D-mannitol (D-Man). Data are estimated from the mean flux values derived from Figs. 3A,B for Asp and Glu, whereas D-Asp and L-Trp are means from one experiment (n = 1, N = 3) and D-Man is the mean from all experiments including mannitol (n = 15, N = 2–3).

Download figure to PowerPoint

A-L fluxes of L-aspartate were markedly higher than the corresponding L-A fluxes showing a polarized transport of L-aspartate across the endothelial/astrocyte coculture. A-L fluxes of L-aspartate were not constant during the experiment (Fig. 4A). This was due to a high transport rate, which caused a time-dependent decrease in the concentration gradient of L-aspartate across the endothelium. Fluxes were therefore extrapolated from the linear part of the accumulation curves.

The L-A permeability of L-aspartate was estimated to be 2.34 ± 0.44 × 10−6 cm s−1, while the A-L permeability was 20.7 ± 5.8 × 10−6 cm s−1, which constitutes a significant polarization of the transendothelial transport favouring the brain to blood direction (P = 0.0003). The efflux ratio was ∼9 (Fig. 4C). In total 4.6% ± 2.6% (n = 5) of the applied 3H L–aspartate was transported L-A, while 16.7% ± 7.8% (n = 9) was transported A-L (isotope recoveries ranging from 75 to 104%).

L-glutamate displayed vectorial transport comparable to that observed for L-aspartate (Fig. 4B). Permeability values were of 0.97 ± 0.2 × 10−6 cm s−1 (L-A) and 4.7 ± 0.5 × 10−6 cm s−1 (A-L) giving an efflux ratio of ∼5. This demonstrated a polarized transport of L-glutamate in the brain-to-blood direction (P < 0.0001). 1.8% ± 0.6% (n = 3) of the applied 3H L-glutamate was transported in the L-A direction during the course of the experiments (150 minutes), while 4.2% ± 1.0% (n = 5) was transported in the A-L direction (isotope recoveries ranging from 76 to 96%).

A-L and L-A fluxes of L-glutamate and L-aspartate were measured in the presence of 10 mM of the specific glutamine synthetase inhibitor L-methionine sulfoximine (MSO). Presence of the inhibitor did not significantly affect the fluxes of L-glutamate or L-aspartate (data not shown) indicating that conversion to glutamine was not a prerequisite for the A-L transport to occur.

The bi-directional transport of D-aspartate, D-mannitol, and L-tryptophan was investigated to validate that the observed vectorial transport for L-aspartate and L-glutamate was not due to a general polarization of transport across the in vitro model (Fig. 4C). D-aspartate, D-mannitol, and L-tryptophan had flux ratios around 1–1.5. It can thus be concluded that L-aspartate and L-glutamate, but not D-aspartate, have A-L polarized efflux across the BBB in vitro.

Transendothelial Transport of L-Aspartate and L-Glutamate Across Endothelial/Astrocyte Cocultures had Passive and Carrier Mediated Flux Components

The concentration dependency of L-aspartate and L-glutamate transcellular fluxes was examined to characterize the kinetic parameters of transendothelial transport.

The transendothelial fluxes of L-aspartate and L-glutamate are shown in Fig. 5. The L-A fluxes of both amino acids showed a linear relationship between flux and concentration, indicating a nonsaturable transport pathway in the concentration range investigated, the hallmark of passive permeation. A-L fluxes showed a saturable as well as a nonsaturable component in the concentration range investigated, indicative of carrier-mediated transport in combination with passive permeation. The nonsaturable components were calculated by curve fitting from the L-A transport and subtracted from the total A-L fluxes. These are shown in Fig. 5B,D. From these curves Km and Jmax were determined to be 69 ± 15 μM and 44 ± 3 pmol min−1 cm−2 for L-aspartate and 138 ± 49 μM and 28 ± 3 pmol min−1 cm−2 for L-glutamate.

thumbnail image

Figure 5. (A) Steady state transendothelial flux of L-aspartate as a function of added L-aspartate concentrations in the donor compartment. (▪) luminal-to-abluminal direction, (▴) abluminal-to-luminal direction . (B) Luminal-to-abluminal L-aspartate flux corrected for passive diffusion. Results are means ± standard deviation from data from two individual experiments pooled together into one (N = 3–5). (C) Steady state transendothelial flux of L-glutamate as a function of added L-glutamate concentrations in the donor compartment. (▪) Luminal-to-abluminal direction, (▴) Abluminal-to-luminal direction. (D) Luminal-to-abluminal L-glutamate flux corrected for passive diffusion. Results are mean ± SEM from (n = 3, N = 2–3) for abluminal to luminal transport and means ± standard deviation (n = 1, N = 3) for luminal-to-abluminal direction.

Download figure to PowerPoint

Intracellular Accumulation of L-Glutamate, L-Aspartate, and D-Aspartate Took Place from the Abluminal but not Luminal Side via a Carrier-Mediated and Sodium-Dependent Mechanism

The transendothelial flux data showed that the A-L flux of L-glutamate and L-aspartate was polarized, indicative of basolateral transport via EAATs. However, D-aspartate is a well-known substrate of the EAATs but was not effluxed across the endothelium. Lack of transcellular transport of D-aspartate could be caused by either lack of uptake across the abluminal membrane or lack of passage across the luminal membrane of the endothelial cells. Accumulation in the endothelial cells was therefore measured 150 min after application of radiolabeled substrate in the abluminal or luminal chamber. To estimate whether the astrocytes contributed to the observed accumulation, a parallel experiment was performed with permeable filter inserts covered with astrocytes on the abluminal side but without endothelial cells.

Accumulation of all radiolabeled substrates was significantly higher from the abluminal side than from the luminal side (P < 0.05 for L-aspartate and P < 0.01 for L-glutamate and D-aspartate), and could be inhibited almost completely by removal of sodium from the assay medium. Uptake of radiolabel into astrocytes was low when compared with the total uptake into the cocultures, indicating that the overall contribution of astrocytes to the accumulation was negligible. Abluminal uptake of D-aspartate was significantly higher than that of L-glutamate (P < 0.01) and L-aspartate (P < 0.001) (Dunnett's test with D-aspartate as control column). Thus all three amino acids were taken up from the abluminal solution into the endothelial cells. L-aspartate and L-glutamate were transported through the luminal membrane, as demonstrated in the transendothelial flux experiments (Figs. 4 and 5), whereas D-aspartate was not transported transcellularly but accumulated in the endothelial cells. These data indicates a transport pathway with an abluminal carrier-mediated and sodium dependent uptake, which mediates uptake of both L-glutamate, L-aspartate, and D-aspartate, and a luminal efflux step which mediates cell-to lumen transport of L-glutamate and L-aspartate, but not D-aspartate.

Transendothelial Transport of L-Aspartate and L-Glutamate was Inhibited by TBOA, an Inhibitor of the Excitatory Amino Acid Transporters EAAT-1, EAAT-2, and EAAT-3

The accumulation experiments indicated the presence of sodium dependent polarized transporters at the abluminal membrane of the endothelial cells. The A-L transendothelial flux of L-aspartate and L-glutamate were examined in the presence of the general EAAT inhibitor, TBOA, 1 mM unlabeled L-aspartate or 1 mM L-glutamate and when compared with the fluxes of tracer alone (Fig. 6).

thumbnail image

Figure 6. Transendothelial transport of 3H-L-aspartate (5 μM) (A) and 3H-L-glutamate (5 μM) (B) in the abluminal-to-luminal direction, as a function of time. (▴) control, (•) 1 mM unlabeled L-aspartate added to the abluminal compartment (▾), 1 mM unlabeled glutamate added to the abluminal compartment, (♦), 50 μM TBOA added to the abluminal compartment. Data are overall mean values ± SEM from three experiments with 2–3 filter inserts in each.

Download figure to PowerPoint

The L-A flux of L-aspartate was inhibited by 50 μM TBOA, L-aspartate, and L-glutamate (Fig. 6A). Apparent permeability coefficients were lowered from 20.7 ± 5.8 × 10−6 cm s−1 to 2.04 ± 0.3 × 10−6 cm s−1, 5.56 ± 1.9 × 10−6 cm s−1 and 8.44 ± 1.5 × 10−6 cm s−1 by L-aspartate, L-glutamate, and TBOA, respectively. Same inhibition pattern was observed for L-glutamate (Fig. 6B), where the trace permeability of 4.66 ± 0.5 × 10−6 cm s−1 was inhibited to 1.13 ± 0.3 × 10−6 cm s−1, 1.04 ± 0.04 × 10−6 cm s−1 and 2.45 ± 0.6 × 10−6 cm s−1 by L-aspartate, L-glutamate, and TBOA, respectively. The inhibition with TBOA constituted approximately 60–70% of total inhibition (100% inhibition = L-A passive permeability), which is close to the expected theoretical inhibition of ∼83% calculated from the Km and Jmax values from Fig. 5 as well as the Ki value of TBOA of 2.2–10 μM (Bunch et al.,2009).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The amino acid L-glutamate is essential for neuronal signalling but also neurotoxic in high concentrations in brain ISF. This necessitates a strict regulation of brain ISF glutamate levels. This study demonstrates that the endothelial cells of the BBB have the capacity to participate in this regulation by mediating active transport of L-glutamate from the abluminal to the luminal solution. The transport system operated in the micromolar range and recognized L-glutamate and L-aspartate as substrates. Overall, this study suggests that BBB-mediated glutamate efflux could play a significant role in the regulation of brain ISF glutamate levels, especially under pathophysiological conditions.

The study of expression and function of EAAT transporters at the BBB and their possible role in glutamate homeostasis can be facilitated by the use of BBB in vitro models. During the last decades a number of in vitro BBB models have been developed. However, common problems with these models have either been the lack of tightness, which leads to large paracellular fluxes of small molecules or the low expression of transport proteins (Deli et al.,2005; Garberg et al.,2005). Recently, our group developed a bovine endothelial/rat astrocyte coculture protocol, which was shown to improve the endothelial monolayer tightness (Helms et al.,2010). The modified culture protocol thus resulted in endothelial monolayers showing high TEER values and low mannitol permeability, as well as expression of transporters such as MDR-1, BCRP, LAT-1, and GLUT-1 (Helms et al.,2010). This has enabled studies of vectorial transcellular transport of small molecules and their transendothelial transport kinetics in a simple and isolated system when compared with in vivo.

The Glutamate Efflux System of the Blood-Brain Barrier has a Large Transport Capacity

In this study, we characterized the transendothelial transport kinetics of the BBB efflux system for L-glutamate and L-aspartate. Glutamate efflux by the neurovascular unit has been suggested to play a significant role in the ISF glutamate handling following ischemic events, and in vitro studies in a pig endothelium/astrocyte coculture model has indeed shown efflux of L-glutamate (Cohen-Kashi-Malina et al.,2012; Teichberg et al.,2009). It has also been demonstrated in a number of papers that removal of glutamate from blood decreases brain ISF glutamate, causes neuro-protection in animal models and that blood expression levels of the glutamate scavenger glutamate-oxaloacetate transaminase correlates with improved outcome following acute ischemic stroke in humans (Campos et al.,2011a, b; Gottlieb et al.,2003; Teichberg et al.,2009; Zlotnik et al.,2007). The overall Km-value of the transendothelial glutamate efflux was ∼138 μM in our study. Brain ISF glutamate levels have been estimated to be approximately 0.8–14.7 μM (Abbott et al.,2010; Lerma et al.,1986; Teichberg et al.,2009), well below the Km of the BBB glutamate efflux system described here. Under healthy conditions, the primary role of the BBB glutamate efflux system would be to maintain the glutamate gradient between ICF and blood. During pathophysiological conditions, where the brain glutamate levels increase, the system would gain an increasing importance and possibly serve as a mechanism to decrease ICF glutamate levels. Calculation of the quantitative contribution of BBB mediated efflux of glutamate to overall brain ISF glutamate homeostasis would demand knowledge of the uptake capacity and time course of all glutamate-handling processes in the brain as well as spatiotemporal considerations taking diffusion distances into accounts. However, some rough estimates regarding the role of the glutamate efflux system in ischemia can be made. During ischemic events the brain ISF glutamate increases with estimates ranging from eightfold in rat (Benveniste et al.,1984) up to more than 100-fold in humans (Kanthan et al.,1995). At brain glutamate concentrations above 200 μM the BBB-efflux system would be saturated and capable of removing 2.8 μmol glutamate per minute. Assuming a brain volume of 1.2 L and an ISF volume of 20% of this volume, a total ISF volume of 0.24 L can be calculated. In a theoretical example where we assume that the ISF has an initial concentration of 300 μM of glutamate (which equals 72 μmol in the ISF) the BBB efflux system would lower the ISF glutamate concentration from 300 μM to 200 μM within 9 minutes, indicating that BBB efflux of glutamate may play a significant role in control of ISF glutamate levels. The transport capacity of the system is thus in theory sufficiently high to influence brain L-glutamate homeostasis, given that the in vitro system is comparable to the brain endothelium in vivo. However, as the present glutamate efflux data are obtained using bovine endothelial cells, species differences may cause differences in absolute values. It might also be argued that the in vitro culture overexpress EAATs since we find expression of EAAT-3, in cocultures grown for 6 days, but not in freshly isolated capillaries. However, the L-glutamate efflux phenomenon has been observed in in vitro models of porcine tissue and in in vivo rat brain. This study thus supports the hypothesis that L-glutamate efflux might be a general property of the blood-brain barrier.

Intracellular Accumulation of Excitatory Amino Acids from the Abluminal Side Occurred via EAATs

L-glutamate and L-aspartate, as well as D-aspartate, was accumulated in the endothelial cells when added from the abluminal side. Concentrations of the three substrates in the endothelial cells were well above chemical equilibrium with the abluminal medium (22-fold for L-aspartate, 150-fold for L-glutamate, and 400-fold for D-aspartate), indicating a strong energy coupling in the transport process. Removal of the sodium gradient across the abluminal membrane caused the cellular concentration of L-glutamate to drop ∼30-fold, approximating the same concentration as that in the abluminal solution. However, this could also be a result of the cells losing their general polarization because of the lost sodium gradient. Addition of 1 mM unlabeled L-asparate and L-glutamate reduced the intracellular accumulation, which indicates a carrier-mediated pathway is involved in the abluminal uptake.

The transcellular fluxes of L-glutamate and L-aspartate were inhibited by ∼60% and 67%, respectively by the addition of 50 μM TBOA, which is known to be a general inhibitor of EAAT-1, −2, and −3 (Shimamoto et al.,1998; Shimamoto et al.,2000) with a Ki-value of 2.2–10 μM (Bunch et al.,2009). This indicates that the uptake across the abluminal membrane is mediated by EAATs in concordance with the results from Cohen-Kashi-Malina et al. (Cohen-Kashi-Malina et al.,2012). It has been suggested that the neutral amino acid transporter, ASCT2, could also be involved in basolateral uptake of both L-glutamate and L-aspartate into the endothelial cells of the BBB (Cohen-Kashi-Malina et al.,2012; Tetsuka et al.,2003). However, in our experiments the basolateral uptake of radiolabeled L-aspartate and L-glutamate would have to compete with the neutral amino acid substrates present in the medium, L-cysteine (520 μM), L-serine (400 μM) and L-threonine (800 μM) all well above their Km at ASCT2 of ∼20 μM (Utsunomiya-Tate et al.,1996), whereby ASCT2 is not expected to play a significant role.

The presence of excitatory amino acid transporters on the abluminal membrane has been somewhat controversial. We demonstrated the presence of mRNA of EAAT-1, EAAT-2 and EAAT-3 in cultured bovine endothelial cells and presence of EAAT-1 and EAAT-3 by immunofluorescence, however, in noncultured capillaries only EAAT-1 was present. This suggests that EAAT-3 could be up-regulated during culture. Recently published proteomics data from Uchida et al. shows that EAAT-1 is one of the most abundantly expressed transporters in human brain capillaries on protein level, but EAAT-3 is not present above detection limit (Uchida et al.,2011). Moreover, other investigations showing clear EAAT-3 expression on protein or mRNA level have been performed in cultured endothelial cells (Campos et al.,2011a; Cohen-Kashi-Malina et al.,2012; Lyck et al.,2009). These reports further support that EAAT-3 is not expressed to a high degree in capillaries but is subsequently up-regulated during culture. PCR analysis indicated a high expression of EAAT-2 in freshly islolated capillaries. However, the freshly isolated capillaries also showed presence of GFAP mRNA (data not shown), which shows that the astrocytes were still associated with the capillaries. It is therefore likely that the EAAT-2 signal mainly derived from astrocytic endfeet, which is further supported by the reduced expression in the cultured endothelial cells, where no GFAP signal was obtained (data not shown). Another possibility is that the lower EAAT-2 expression in endothelial cells and cocultures were caused by a down-regulation during culture. Indeed Lyck et al. showed that EAAT-2 mRNA was down-regulated when PECAM1-purified mouse brain microvessel endothelial cells were cultured in single or noncontact coculture with astrocytes, however, they only found a low expression of EAAT-2 in noncultured endothelial cells, which supports that the higher expression in our capillaries were indeed from astrocytes (Lyck et al.,2009). It can be concluded that there still exists controversy regarding the exact EAAT subtype expression pattern at the BBB, and that the influence of culture conditions should be investigated further with regards to up- and down-regulation of EAAT-2 and EAAT-3.

Apical Efflux of Excitatory Amino Acids was Mediated by a Transporter Which Excluded D-Aspartate

Results in this study show, in concordance with earlier in vivo brain efflux studies (Hosoya et al.,1999), that transcellular A-L permeability was highest for L-aspartate, lower for L-glutamate, and very low for D-aspartate. Hosoya et al. concluded that the efflux of glutamate and aspartate was not caused by EAATs, as D-aspartate is also a substrate for these. However, our results indicate that the abluminal uptake of L-glutamate, L-aspartate and D-aspartate was indeed caused by EAATs, whereas the following passage of the apical membrane is likely to be caused by a transport mechanism that accepts L-aspartate and L-glutamate but not D-aspartate.

Previously, an aspartate/glutamate (Aspartate/glutamate transporter-1) was found and shown to transport L-glutamate and L-aspartate but not D-aspartate (Matsuo et al.,2002). Moreover, it was inhibited to a smaller degree by cysteine, which can explain the cysteine inhibition observed by Okane et al.. Kinetic parameters for L-aspartate were found in oocytes to be Km = 25.5 ± 5.9 μM and Vmax = 0.67 ± 0.013 pmol/min/oocyte, which are in the same order of magnitude as the Km of transendothelial glutamate/aspartate efflux. However, Northern blot analysis of tissue distribution revealed only expression in the kidney and we could not detect significant mRNA levels of the transporter in neither capillaries nor endothelial cells (results not shown). The Cystine/Glutamate antiporter Xc could in theory be responsible for the apical efflux of L-glutamate, but would not be able to account for the efflux of L-aspartate (Bridges et al.,2012). The molecular identity of the luminal cell-to-blood transport step thus remains to be characterized.

Conversion of Glutamate to Glutamine was not a Necessary Step for Transendothelial Efflux of Glutamate

It has previously been proposed that the observed glutamate efflux could be caused by glutamine shuttling between astrocytic endfeet and endothelial cells, where after the glutamine was converted to glutamate in the endothelial cells and excreted to the blood (Teichberg et al.,2009). However, Cohen-Kashi-Malina et al. recently reported increased glutamate permeability across a pig endothelium/astrocyte coculture model upon addition of MSO, which selectively inhibits the glutamine synthetase (Cohen-Kashi-Malina et al.,2012; Dadsetan et al.,2011). Our study showed no significant effects on L-glutamate or L-aspartate A-L permeabilities. Furthermore, CalceinAM staining of astrocytes on the bottom of the filter inserts revealed that there were no astrocyte projections and thereby no direct cell contacts through the filter pores (data not shown), which has also been shown recently in a mouse blood-brain barrier coculture model (Shayan et al.,2011). The studies by Hosoya et al. and Gottlieb et al. also indicated capillary uptake and BBB efflux of intact glutamate (Gottlieb et al.,2003; Hosoya et al.,1999).

The abluminal uptake of D-aspartate, as well as L-glutamate and L-aspartate furthermore supported that EAAT-mediated uptake was the abluminal entry step in the transendothelial transport pathway, since D-aspartate cannot be metabolized in astrocytes like glutamate and aspartate (Davies and Johnston,1976). The suggested transendothelial pathway is depicted in Fig. 7.

thumbnail image

Figure 7. Transport of excitatory amino acids across the blood-brain barrier. L-glutamate, L-aspartate, and D-aspartate are taken up at the abluminal side of the endothelial cells by EAATs. Cellular uptake is inhibited by the subtype unspecific EAAT inhibitor, DL-threo-β-Benzyloxyaspartate (TBOA). L-glutamate and L-aspartate may subsequently exit into the blood stream via one or more transporters (X) in the luminal membrane.

Download figure to PowerPoint

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The authors thank the laboratory technician Lene Vigh for expert technical assistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES