Characterization of the L-glutamate clearance pathways across the blood–brain barrier and the effect of astrocytes in an in vitro blood–brain barrier model

Abstract

The aim was to characterize the clearance pathways for L-glutamate from the brain interstitial fluid across the blood–brain barrier using a primary in vitro bovine endothelial/rat astrocyte co-culture. Transporter profiling was performed using uptake studies of radiolabeled L-glutamate with co-application of transporter inhibitors and competing amino acids. Endothelial abluminal L-glutamate uptake was almost abolished by co-application of an EAAT-1 specific inhibitor, whereas luminal uptake was inhibited by L-glutamate and L-aspartate (1 mM). L-glutamate uptake followed Michaelis–Menten-like kinetics with high and low affinity at the abluminal and luminal membrane, respectively. This indicated that L-glutamate is taken up via EAAT-1 at the abluminal membrane and exits at the luminal membrane via a low affinity glutamate/aspartate transporter. Metabolism of L-glutamate and transport of metabolites was examined using [U-¹³C] L-glutamate. Intact L-glutamate and metabolites derived from oxidative metabolism were transported through the endothelial cells. High amounts of L-glutamate-derived lactate in the luminal medium indicated cataplerosis via malic enzyme. Thus, L-glutamate can be transported intact from brain to blood via the concerted action of abluminal and luminal transport proteins, but the total brain clearance is highly dependent on metabolism in astrocytes and endothelial cells followed by transport of metabolites.

Introduction

Extracellular L-glutamate levels in the central nervous system (CNS) must be strictly controlled to maintain homeostasis.¹ Excitatory amino acid transporters (EAATs) play an important role in maintaining low levels of L-glutamate through uptake mainly into astrocytes,^2–4 but L-glutamate has also been suggested to be cleared from the brain interstitial fluid via the endothelial cells of the neurovascular unit constituting the blood–brain barrier (BBB) (for review see Cederberg et al.⁵). This has been demonstrated in perfused dog brains⁶ and in rats under different pathological conditions^7–11 and has led to the suggestion of a new treatment paradigm termed ”blood-glutamate scavenging,” where peripheral breakdown of L-glutamate is hypothesized to decrease L-glutamate concentrations in the brain interstitial fluid by accelerating brain-to-blood efflux.¹² This is an effective treatment in animal models, but the exact clearance pathways from the brain to the blood remain to be characterized.

EAAT expression has been shown in brain capillary endothelial cells from rats, mice, pigs, calves, and humans,^7,13–20 and two studies have shown vectorial transport of L-glutamate favoring the brain-to-blood direction with in vitro astrocyte/endothelial co-culture models.^13,20 However, the studies applied ³H L-glutamate to study the transcellular transport, but L-glutamate is highly metabolized in astrocytes and may also be metabolized in endothelial cells.^21,22 Thus, apparent transcellular transport of ³H L-glutamate may reflect an initial metabolism of L-glutamate followed by transport of labeled metabolites, for instance L-glutamine, intermediates derived from oxidative metabolism via the tri-carboxylic acid cycle (TCA cycle) and lactate.²³ Furthermore, it is unclear which EAAT subtypes are responsible for the L-glutamate uptake at the abluminal membrane, and the transporter responsible for the luminal transport step from endothelium-to-blood remains unknown at the molecular level.

The aim of this study was to characterize the brain-to-blood clearance of L-glutamate by investigating carrier-mediated transport across the abluminal and luminal membranes of endothelial cells as well as L-glutamate metabolism and transport of resulting metabolites in primary bovine endothelial/rat astrocyte co-cultures.

It was shown that intact L-glutamate can be transported across the BBB by concerted action of EAAT-1 and an unknown low affinity L-glutamate transporter. However, the main part of the applied L-glutamate was metabolized in the astrocytes and endothelial cells, and metabolites derived from oxidative metabolism in the TCA cycle were transported through the BBB.

Materials and methods

Materials

³H L-glutamic acid (51.1 Ci/mmol) and ¹⁴C D-mannitol (57.1 mCi/mmol) were from Perkin Elmer (Hvidovre, Denmark). L-[U-¹³C] glutamic acid, 99% enriched, was from Cambridge Isotopes Laboratories Inc. (Andover, MA, USA). All other chemicals and reagents were from Sigma-Aldrich (Rødovre, Denmark) unless otherwise stated.

Animal ethics

All animals were sacrificed without any prior treatment for primary cell isolation, thus no approval for animal experimentation was needed. Calves were slaughtered following Danish legislation on slaughtering of animals (BEK nr 135 af 14/02/2014, European legislation identifier/eli/lta/2014/135). Rats were euthanized according to Danish legislation on animal experimentation (BEK nr 12 af 07/01/2016, European legislation identifier/eli/lta/2016/12). The ARRIVE guidelines have been followed were relevant.

Isolation of primary bovine brain endothelial cells and rat astrocytes

The isolation and culture procedures are described elsewhere.²⁴ Bovine brains were acquired from calves below 12 months of age from a slaughterhouse (Mogens Nielsen Kreaturslagteri A/S, Herlufmagle, Denmark). The cortical gray matter was isolated and homogenized in Dulbecco’s Modified Eagles Medium-AQ (DMEM) using a 40 ml Dounce Tissue Grinder (Wheaton Science Products, Millville, USA). Capillaries were caught on 160-µm nylon filters (Millipore, Copenhagen, Denmark), and digested 1 h at 37℃ in DNAse I (170 U/ml), Collagenase type III (200 U/ml) and Trypsin TRL (90 U/ml) (Worthington Biochemical Corporation, Lakewood, USA) in DMEM supplemented with 10% fetal bovine serum, 1% (v/v) non-essential amino acid mixture, and 100 U/ml – 100 µg/ml penicillin–streptomycin solution (DMEM-Comp). The digested suspension was filtered through 200 -µm mesh filters, and the capillary pellet was resuspended in fetal bovine serum:dimethylsulfoxide (9:1), frozen in aliquots overnight at −80℃ and stored in liquid nitrogen.

Astrocytes were isolated from three to four days old Sprague Dawley rats (Taconic, Ejby, Denmark) as previously described.²⁵ Briefly, midbrains and olfactory bulbs were removed and cerebral cortices were pushed through 80 -µm nylon mesh into DMEM containing 20% fetal bovine serum and seeded in T75 culture flasks. The astrocytes grew to confluence, and the serum content was subsequently gradually reduced to 10% during three weeks culture. The astrocytes were passaged with Trypsin-EDTA, counted and resuspended in fetal bovine serum:dimethylsulfoxide (9:1) and frozen overnight at −80℃ (approximately 2 × 10⁶ cells/vial). The astrocytes were stored in liquid nitrogen. Astrocyte-conditioned medium (ACM) was collected three times a week during the last week of culture in 10% serum.

Culture of bovine brain endothelial cells and rat astrocytes

Frozen bovine brain capillaries were thawed and cultured for four days (37℃, 10% CO₂) in DMEM-Comp:ACM (1:1) supplemented with 125-µg/ml heparin in collagen type IV/fibronectin (1 µg/cm² of each)-coated T75 flasks. The endothelial cells were passaged and seeded at a density of 90,000 cells/cm² on collagen type IV/fibronectin-coated transwell polycarbonate permeable supports (area = 1.12 cm², pore radius = 0.4 µm, Corning Life Sciences, New York, USA). These were either empty supports (for non-contact co-cultures) or astrocytes had been seeded at the bottom of the supports two days before (120,000 cells/cm²) (for contact co-cultures). For non-contact co-cultures, astrocytes were seeded in the bottom of the wells two days prior to endothelial seeding (120,000 cells/cm²). The co-cultures were cultured for three days in DMEM-Comp supplemented with 125-µg/ml heparin, followed by three days in DMEM without NaHCO₃⁻ (Gibco, Breda, The Netherlands), supplemented with 10% fetal bovine serum, 1% (v/v) non-essential amino acid mixture, 100 U/ml–100 µg/ml penicillin-streptomycin solution, 312.5 µM 8 -(4-CPT)-cyclic adenosine monophosphate, 0.5 µM dexamethasone, 17.5 µM RO-20-1724 and 50 mM N-[tris(hydroxymethyl)methyl]-2-aminoethanesulfonic acid (TES) (DM + TES).

TEER-measurements and radioisotope studies

The transendothelial electrical resistance (TEER) was measured at 20–22℃ prior to all experiments, using an Endohm-12 cup electrode chamber (World Precision Instruments, Sarasota, Florida) connected to a Millicell-ERS device (Millipore, MA, USA).

Abluminal uptake experiments were performed in the non-contact co-cultures, while luminal uptake experiments were performed in contact co-cultures. Permeable supports were moved to new culture plates without astrocytes, and the DM + TES was changed to Hanks balanced salt solution with Ca²⁺/Mg²⁺ (HBSS) with HEPES (10 mM), 0.05% bovine serum albumin, 8 -(4-CPT)-cyclic adenosine monophosphate (312.5 µM), dexamethasone (0.5 µM) and RO-20-1724 (17.5 µM). 1 µCi/ml of ³H L-glutamic acid (0.02 µM) was added to the donor solution together with 1 µCi/ml ¹⁴C D-mannitol. Abluminal uptake was investigated in the presence of the general EAAT inhibitor, DL-threo-ß-Benzyloxyaspartate (TBOA) (Tocris, Bristol, United Kingdom) (100 µM), the EAAT-1 specific inhibitor, UCPH-101 (50 µM) (Tocris), the EAAT-2 specific inhibitor, dihydrokainic acid (100 µM), or unlabeled L-glutamate. Luminal uptake was examined in the presence of competing amino acid substrates (1 or 5 mM each); 100 -µM TBOA was added to the abluminal solution to inhibit EAAT-mediated uptake of L-glutamate leaking through the paracellular space. Culture trays were placed on a shaking table at 37℃ and 90 rounds/min. Uptake and remaining substrate was determined by withdrawing a donor sample, wash three times with ice cold HBSS, and cut out the supports. Samples were transferred to Ultima Gold scintillation fluid (Perkin-Elmer, Hvidovre, Denmark). Radioactivity was counted in a Tri-Carb 2100 TR Liquid Scintillation Analyzer (Packard Instrument Company, Meriden, USA).

¹⁴C D-Mannitol counts were used to estimate extracellular fluid volume in the uptake samples and the amounts of L-glutamate in the cells were corrected for this.

TEER was measured before and after change from DM + TES to HBSS in all uptake experiments. The change of medium caused an average resistance drop of 50 ± 6%. The average TEER prior to change of medium were 1543 ± 114 Ω·cm² (n = 11, total N = 170) in the non-contact co-cultures and 1491 ± 83 Ω·cm² (n = 29, total N = 439) in the contact co-cultures (P = 0.71).

Transcellular transport studies were performed on non-contact and contact co-cultures after six days of co-culture. Non-contact co-cultures were investigated both in the presence and absence of the astrocytes on the bottom of the well. For the transport experiments without astrocytes, the permeable supports and culture medium were transferred to new culture plates immediately prior to initiation of the experiment; 1 µCi/ml of ¹⁴C-D-mannitol and 1 µCi/ml ³H L-glutamate were added to the abluminal or luminal medium and culture trays were placed on a shaking table at 37℃ with stirring at 90 rounds/min. Receiver samples were taken after 30, 60, 90, 120, and 150 min and a donor sample after 150 min. Permeable supports were washed and collected and samples were counted as described above

Metabolism of L-glutamate and transport of metabolites

Metabolism studies were performed in non-contact co-cultures. TEER prior to the experiments was 1686 ± 114 Ω·cm² (n = 3, total N = 27). The luminal medium was kept unchanged, while the abluminal medium was replaced with a modified DM + TES with 100 µM [U-¹³C] L-glutamate without unlabeled L-glutamate/L-aspartate and fetal bovine serum. The cells were incubated 150 min with shaking as before. The luminal and abluminal media were withdrawn and the endothelial cells and astrocytes were lyzed with ethanol. Cell lysates and media samples were pooled in groups of three to obtain amino acid concentrations above detection limits. Samples were adjusted to pH 1–2 with HCl and dried under nitrogen flow. Organic extraction of analytes was performed based on a modified procedure after Mawhinney et al.²⁶ Standards containing unlabeled metabolites, cell lysates, and media were analyzed using a gas chromatograph (Agilent Technologies 7820A chromatograph, J&W GC column HP-5MS, parts no. 19091S-433) (Santa Clara, CA, USA) coupled to a mass spectrometer (Agilent Technologies 5977E). The isotopic enrichment of the metabolites was corrected for natural abundance of ¹³C using the unlabeled standards and calculated according to Biemann.²⁷

Amino acids were separated and quantified by reversed-phase high-performance liquid chromatography using an Agilent ZORBAX Eclipse plus C18 column (4.6 × 150 mm, particle size 3.5 µm, Agilent Technologies), pre-column online o-phthaldialdehyde derivatization and fluorescence detection (338 nm, 10 nm bandwidth, and reference wavelength 390 nm, 20 nm bandwidth). An Agilent 1260 Infinity system coupled to a 1260 Infinity fluorescence detector (Agilent Technologies) was employed.

Lactate concentrations were determined using the enzymatic L-lactic acid kit from Boehringer Mannheim/R-Biopharm (Mannheim, Germany) as previously described²⁸

Data analysis

TEER was multiplied with the surface area of the permeable supports (Ω·cm²). Cellular uptake values (pmol/cm²) were estimated by calculating the amount of substrate present in the cells and subtracting the contribution of isotope in the extracellular space, determined as the space accessible to ¹⁴C D-mannitol.

Cellular uptake data of ³H L-glutamic acid were further corrected for a non-displaceable background, which may represent unspecific binding to the cells and the permeable supports and arises because a fixed amount of ³H L-glutamate was applied in competition with increasing amounts of unlabeled L-glutamate. Disintegrations per minute (DPM) were plotted against the log concentration of L-glutamate (see inserts in Figure 1(b) and (d)). DPM approached a minimum value representing the passive non-displaceable background with increasing L-glutamate concentrations and specific cellular uptake was obtained by subtracting the non-displaceable DPM component from the observed DPM values.

Figure 1.

(a–e) Luminal (▪) and Abluminal (•) L-glutamate uptake in endothelial cells from contact (luminal) or non-contact (abluminal) co-cultures. (a) Uptake of ³H L-glutamate (0.02 µM) was examined as a function of time. Data points are mean ± SEM from three experiments (n = 3) with two permeable supports per condition (total N = 6).(b–c): Initial uptake velocity of ³H L-glutamate was measured after 5 min exposure to different abluminal L-glutamate concentrations. (b) shows the total uptake, while the insert shows disintegrations per minute (DPM) from L-glutamate as a function of log L-glutamate. DPM approximates a minimum value as L-glutamate concentration increases revealing a non-displaceable background count. (c) Initial uptake velocity after correction from the non-displaceable background found in the insert in figure (b). Data points are mean ± SEM for four experiments (n = 4) with two permeable supports per condition (total N = 8). (d–e) Initial uptake velocity of ³H L-glutamate was measured after 10 min exposure to different luminal L-glutamate concentrations. (d) shows the total uptake, while the insert shows DPM as in figure B. E shows the initial uptake velocity after correction from the non-displaceable background found in the insert in figure (d). Data points are mean ± SEM from four experiments (n = 4) with three permeable supports per condition (total N = 12).

Uptake rates were approximated using a Michaelis–Menten-like expression (equation (1))

$$V=\frac{V_{max}·[S]}{K_M+[S]}$$ (1)

where V was the uptake rate (pmol/(cm²·min)), [S] was the donor L-glutamate concentration (µM), V_max was the maximal uptake rate (pmol/(cm²·min)) and K_M was the Michaelis constant (µM). Where appropriate, results were compared using one-sided analysis of variance followed by Bonferronis multiple comparison test. One level of significance was applied throughout the statistical analyses (α = 0.05).

The transport data were plotted as total amount transported per cm² against time, and steady-state fluxes (J) were calculated from the slopes of the straight lines. Apparent permeabilities (P_app) were calculated using equation (2).

$$P_{\mathit{app}}=\frac{J}{C_{\mathit{donor}}}$$ (2)

C_donor is the added concentration to the donor compartment.

The data from the metabolism studies are presented as % labeling of M+X, where M is the mass of the unlabeled molecule and X is the number of labeled C-atoms in the metabolite. Absolute amounts of M+5 L-glutamate, M+5 L-glutamine, M+4 L-aspartate, and M+1/M+2 lactate were calculated by multiplying the % labeling with the total concentrations determined by HPLC-MS. Where appropriate, results were compared between cellular compartments using Student’s t-test. One level of significance was applied throughout the statistical analyses (α = 0.05).

Results

Abluminal L-glutamate uptake displayed high affinity resembling EAAT-mediated uptake, whereas luminal L-glutamate uptake displayed a lower affinity with similar capacity

³H L-glutamate uptake was investigated in the contact co-cultures to characterize luminal endothelial uptake, whereas the non-contact co-cultures were used to characterize the abluminal endothelial uptake, since the presence of astrocytes would influence this. The time dependent L-glutamate uptake was examined at both membranes to determine the linear phase of the uptake processes (Figure 1(a)). Abluminal and luminal L-glutamate uptake were linear for 15 min, after which they deviated and reached an apparent plateau (R²= 0.94 until 15 min, R²= 0.6–0.7 when 25 to 45 min are included). Based on this, uptake experiments were conducted for 5 (abluminal) or 10 (luminal) min.

Concentration dependent initial L-glutamate uptake rate at the abluminal membrane showed a pattern indicative of a combination of saturable and non-saturable uptake as the curve approximated linearity at concentrations above 250 µM (Figure 1(b)). The obtained DPM values were plotted against the log concentration of L-glutamate (Figure 1(b) insert). DPMs approximated 125 with increasing L-glutamate concentrations, representing the non-saturable uptake, which is a non-displaceable background rather than actual uptake. The apparent IC₅₀ value for displacement of L-glutamate was 43 µM (log IC₅₀ = 1.63, n = 4, total N = 8). Uptake rates were corrected to obtain the specific cellular uptake (Figure 1(c)). Cellular uptake was described by the Michaelis–Menten-like expression yielding a K_M of 63 ± 31 µM and a V_max of 12 ± 1.5 pmol/(cm²·min).

Luminal L-glutamate uptake showed the same pattern, with approximation of linearity at high concentrations (Figure 1(d)). The non-displaceable background was 242 DPM with an apparent IC₅₀ value of 337 µM (log IC₅₀ = 2.58, n = 4, total N = 12) (Figure 1(d) insert). This was subtracted from the total uptake data in Figure 1(d) to quantify the carrier-mediated component of the uptake (Figure 1(e)). Fitting the data to Michaelis–Menten-like kinetics resulted in a K_M value of 891 ± 453 µM and a V_max value of 15 ± 2.7 pmol/(cm²·min).

Abluminal L-glutamate uptake was highly inhibited by the specific EAAT-1 inhibitor, UCPH-101, whereas the luminal uptake was mainly inhibited by L-isomers of glutamate and aspartate

L-glutamate uptake was measured in the presence of L-glutamate, the general EAAT inhibitor, DL-threo-ß-Benzyloxyaspartate (TBOA), the EAAT-1 specific inhibitor, UCPH-101, the EAAT-2 specific inhibitor, dihydrokainic acid (DHK), a combination of UCPH-101 and DHK or the ASCT-1 and -2 substrate, L-serine (Table 1).

Table 1.

Abluminal uptake of ³H L-glutamate (0.02 µM) in endothelial cells from non-contact co-cultures in the presence of EAAT inhibitors or an ASCT substrate (L-serine).

Abluminal uptake (0.02 µM 3H L-glutamate)
Competing compound	Concentration (µM)	Uptake (% of control ± SEM)	Significance vs. control
L-glutamate	1000	9 ± 4	*
TBOA	100	13 ± 7	*
UCPH-101	50	22 ± 11	*
DHK	100	88 ± 19	NS
UCPH-101 +	50	24 ± 6	*
DHK	100
L-serine	500	90 ± 20	NS

Note: Uptake was estimated in the presence of unlabeled L-glutamate, DL-threo-ß-Benzyloxyaspartate (TBOA), UCPH-101, dihydrokainic acid (DHK), UCPH-101 + DHK and L-serine. Data points are mean ± SEM for three experiments (n = 3) with two to three filter inserts per condition (total N = 8). ASCT: alanine-serine-cysteine-threonine transporter; EAAT: excitatory amino acid transporter. NS = Non significant (P > 0.05).

^*P < 0.05.

³H L-glutamate uptake was significantly inhibited to 9–20% of control uptake by unlabeled L-glutamate, TBOA, and UCPH-101, whereas DHK and L-serine did not affect the uptake. The inhibition by TBOA, L-glutamate, and UCPH-101 did not differ significantly indicating that EAAT-1 is responsible for the majority of the L-glutamate uptake.

The luminal transporter was investigated for inhibiting compounds and conditions (Table 2).

Table 2.

Luminal uptake of ³H L-glutamate (0.02 µM) in endothelial cells from contact co-cultures in the presence of different amino acids and other relevant compounds.

Competing compound	Concentration (µM)	Uptake (% of control ± SEM)	Significance vs. control
Amino acids
L-alanine	1	128 ± 17	NS
L-asparagine	1	77 ± 25	NS
L-aspartate	1	50 ± 1	*
	5	32 ± 1	*
D-aspartate	1	80 ± 7	NS
	5	26 ± 1	*
L-cysteine	1	70 ± 10	NS
L-cystine	0.2	85 ± 6	NS
L-glutamine	1	94 ± 2	NS
L-glutamate	1	45 ± 4	*
	5	30 ± 4	*
D-glutamate	1	94 ± 2	NS
	5	71 ± 5	*
L-glycine	1	94 ± 3	NS
L-histidine	1	112 ± 12	NS
L-isoleucine	1	127 ± 33	NS
L-leucine	1	121 ± 27	NS
L-lysine	1	127 ± 30	NS
L-methionine	1	109 ± 10	NS
L-phenylalanine	1	128 ± 33	NS
L-proline	1	115 ± 20	NS
L-serine	1	104 ± 12	NS
L-threonine	1	104 ± 13	NS
L-tryptophan	1	107 ± 16	NS
L-tyrosine	1	140 ± 15	NS
L-valine	1	110 ± 48	NS
Other conditions
BCH	1	103 ± 10	NS
DMEM + TES	–	25 ± 4	*
L-homocysteic acid	1	86 ± 9	NS
	5	57 ± 4	*
TBOA	0.1	87 ± 8	NS
pH = 6	–	112 ± 7	NS

TBOA: DL-threo-ß-Benzyloxyaspartate; BCH = 2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid. NS: non-significant (P > 0.05), * = P < 0.05.

Note: All compounds showing significant inhibition were examined in three experiments (n = 3) with three permeable supports per condition (total N = 9), whereas amino acids showing no signs of inhibition were examined in one experiment with three permeable supports per condition (n = 1, total N = 3). Luminal uptake (0.02 µM ³H L-glutamate).

³H L-glutamate uptake was lowered to 45 ± 4% of the control uptake by 1 mM unlabeled L-glutamate and further to 30 ± 4% by 5 mM unlabeled L-glutamate. L-aspartate lowered the uptake to 50 ± 1% and 32 ± 1% in 1 mM and 5 mM concentrations, respectively. D-aspartate and D-glutamate only inhibited the uptake at concentrations of 5 mM to levels of 26 ± 1% and 71 ± 5%, respectively. The other amino acids did not affect the uptake. The LAT-1 inhibitor, 2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid (BCH) had no effect on the L-glutamate uptake. Likewise, TBOA had no effect on the luminal uptake indicating that the luminal uptake was not caused by EAATs at the luminal membrane. L-homocysteate inhibited L-glutamate uptake to 57 ± 4% at 5 mM. An acidification of the uptake medium to pH 6 did not affect the uptake.

L-glutamate was highly metabolized in both astrocytes and endothelial cells with subsequent transport of metabolites

L-glutamate metabolism and subsequent transport of metabolites were investigated by adding 100 µM [U-¹³C] L-glutamate to the abluminal medium in a non-contact co-culture. Labeling of potential metabolites was analyzed after 150 min in the abluminal medium, astrocytes, luminal medium, and endothelial cells (Figure 2). Figure 2(a) shows the possible metabolism products of direct metabolism of [U-¹³C] L-glutamate via glutamine synthetase, for synthesis of L-glutamine, or entry in the TCA cycle (labeling % for these pathways are shown in Figure 2(b)), whereas Figure 2(c) shows possible metabolism products after one turn in the TCA cycle (labeling % for these pathways are shown in Figure 2(d)). Intact [U-¹³C] L-glutamate (M+5) and derived metabolites were found in all four compartments (Figure 2(b) I–IV). L-glutamate in the non-metabolized M+5 state was mainly found in the abluminal medium and in the astrocytes, but it was also present in the luminal medium and in the endothelial cells.

Figure 2.

Metabolism of [U-¹³C] L-glutamate in the non-contact co-cultures. Uniformly ¹³C-labeled L-glutamate (100 µM) was initially added in the abluminal medium and the cultures were incubated for 150 min at 37℃. Cell extracts and media were analyzed using GC-MS to determine ¹³C-labeling in metabolites. (a) Via direct metabolism, [U-¹³C] L-glutamate (glu M+5) is converted to [U-¹³C] L-glutamine (gln M+5) or enters the tricarboxylic acid (TCA) cycle for conversion into intermediates such as α-[U-¹³C] ketoglutarate (α-KG M+5), [U-¹³C] succinate (suc M+4) and [U-¹³C] malate (mal M+4). [U-¹³C] aspartate (asp M+4) is formed due to the equilibrium with [U-¹³C] oxaloacetate. Alternatively, [U-¹³C] oxaloacetate condenses with unlabeled acetyl-CoA (Ac-CoA) and [3,4,5,6-¹³C] citrate (cit M+4) is formed. (b) labeling (%) in metabolites from direct metabolism of [U-¹³C] L-glutamate in the four compartments (I-IV). (c) [U-¹³C] oxaloacetate may be further metabolized in the TCA cycle (first turn) after condensation with unlabeled Ac-CoA and in successive steps, generates [1,2,3-¹³C] L-glutamate (glu M+3) and [1,2,3-¹³C] L-glutamine (gln M+3). Subsequent metabolism in the TCA cycle generates double labeled succinate (suc M+2), malate (mal M+2), and oxaloacetate. (d) labeling (%) in metabolites after the first turn of the TCA cycle in the four compartments (I-IV). Data are mean ± SEM from three experiments with nine permeable supports per experiment (n = 3, total N = 27).

[U-¹³C] L-glutamate was amidated to [U-¹³C] L-glutamine catalyzed by glutamine synthetase in the astrocytes, and it was released to the abluminal medium. However, almost no [U-¹³C] glutamine was found in the endothelial or in the luminal medium.

[U-¹³C] L-glutamate was converted to α-ketoglutarate and metabolized directly in the TCA cycle in both cellular compartments as evidenced by the presence of M+5 α-ketoglutarate, M+4 succinate, and M+4 malate (Figure 2(b)). The extent of labeling of these metabolites was higher in the astrocytes as compared to the endothelial cells (P < 0.05 for all TCA-cycle intermediates). However, the labeling of the precursor ¹³C L-glutamate in the endothelial cells (Figure 2(b) IV) was only 10% of that in astrocytes (Figure 2(b) II). Thus, the relative amounts of α-ketoglutarate, succinate, and malate as compared to the M + 5 L-glutamate were all significantly higher in the endothelial cell compartment than in the astrocytes (ratios varying from 1.5 to 7), indicating vivid oxidative metabolism in the endothelial cells. Labeled L-aspartate was found in all compartments (Figure 2(b) and (d)), mainly as M + 4 aspartate resulting from transamination between L-glutamate and M + 4 oxaloacetate, the latter formed from TCA cycle metabolism of M + 5 L-glutamate. M + 4 aspartate was found in the astrocytes and in the abluminal medium, whereas the labeling (%) was low in the luminal medium (Figure 2(b) and (d) III). However, when taking the total amount of aspartate in the luminal medium into account, the data indicate a considerable transfer of L-glutamate carbon as aspartate, either produced directly in the endothelial cells or produced in the astrocytes and transported across the endothelial cells.

The labeling of citrate was pronounced in the endothelial cells both as M + 4 and M + 2 (M + 4 citrate was relatively 1.5 times higher in the endothelial cells than in the astrocytes, P < 0.05) (Figure 2(b) IV and (d) IV). Furthermore, high levels of M + 5 and M + 6 citrate were found in the endothelial cells and in the luminal medium (Figure 3 III and IV). These are formed by re-cycling of labeled pyruvate, M + 3 or M + 1, formed from M + 4 or M + 2 malate in the direct metabolism and 1st turn TCA cycle metabolism of M + 5 L-glutamate and subsequent condensation of double- and mono-labeled acetylCoA with M + 4 oxaloacetate. This pathway requires the concerted action of malic enzyme and pyruvate dehydrogenase.

Figure 3.

Pyruvate recycling in the non-contact co-cultures. The cell cultures were treated as described in Figure 3. The labeling (%) in citrate and lactate that arises from metabolism of [U-¹³C] L-glutamate via malic enzyme combined with pyruvate dehydrogenase activity, i.e. pyruvate recycling, is presented. Data are mean ± SEM from three experiments with nine permeable supports per experiment (n = 3, total N = 27).

This is in line with the labeling of lactate (M + 1, M + 2) in the luminal medium (Figure 3 III).

The total concentrations of L-glutamate, L-glutamine, L-aspartate, and lactate were measured to convert the % labeling data into total amounts.

Approximately 0.5 nmol L-glutamate was transported intact across the endothelial cells, whereas 0.2 nmol was transported as M + 5 L-glutamine and 0.4 nmol as M + 4 L-aspartate. Approximately equal amounts of M + 5 L-glutamate were found on both sides of the endothelial cells (P = 0.75), whereas M + 5 L-glutamine showed around 300 times higher levels in the abluminal medium (P < 0.05). Together these data indicate that glutamate is transported across the BBB in its intact form but also metabolized in both the endothelial cells and astrocytes, and labeled metabolites can be transported across the BBB mainly as aspartate and lactate.

The presence of astrocytes significantly increased the apparent brain to blood transport of radiolabel after abluminal application of ³H L-glutamate

Apparent flux of ³H L-glutamate was investigated in non-contact co-cultures in the presence and absence of astrocytes to estimate the contribution of astrocytic TCA cycle-derived metabolites to the overall ³H flux (Figure 4(a)).

Figure 4.

(a) Apparent permeabilities of ³H and ¹⁴C label across the blood–brain barrier models in the luminal-to-abluminal (L–A) and abluminal-to-luminal (A–L) directions after application of ³H L-glutamate and ¹⁴C mannitol. Transport of radiolabel was examined in the presence and absence of the EAAT-1 inhibitor, UCPH-101. Data are means ± SEM from three experiments with three permeable supports for each condition (n = 3, total N = 9) except for mannitol, which was included in all experiments as control for paracellular permeability (n = 8, total N = 90). (b) Accumulation of ³H label in the endothelial cells after conclusion of the transport experiments shown in part A.

NCC: non-contact co-culture; CCC: contact co-culture; NS: not significant, *P < 0.05.

The permeability of the ¹⁴C labeled paracellular marker, D-mannitol, was 0.62 ± 0.1 × 10⁻⁶cm/s, which confirmed the tightness of the in vitro BBB models. In non-contact co-cultures without the presence of astrocytes, apparent luminal-to-abluminal (L–A) permeability for the ³H-label was significantly higher than the L–A permeability; however, only by a factor 1.5. A–L ³H permeability was significantly increased corresponding to an efflux ratio of 4, when astrocytes were present at the bottom of the culture wells. The A–L permeability across the contact co-culture model was not significantly different from this, indicating that direct cell–cell contact between endothelial cells and astrocytes did not influence the transport of the radiolabel. In the contact co-culture model, UCPH-101 lowered A–L permeability to the same levels as in the non-contact co-culture without astrocytes present, indicating that EAAT-1-mediated uptake into astrocytes plays an important role in the total ³H label transport. UCPH-101 did not significantly affect the permeability in either direction, when astrocytes were not present, indicating a minimal effect of EAAT-1 inhibition on efflux of intact L-glutamate. Intracellular accumulation was significantly higher in the endothelial cells when astrocytes were not present at the bottom of the well, indicating that the metabolism products are mainly transported and not accumulating in the endothelial cells (Figure 4(b)). The accumulation of ³H in the endothelial cells without astrocytes present was significantly reduced by addition of UCPH-101, demonstrating that EAAT-1 mediates uptake into the endothelial cells. Thus, L-glutamate can be taken up and transported across the endothelial cells, but the extent of this is low compared to the amounts being metabolized in astrocytes.

Discussion

Transport of L-glutamate from the brain interstitial fluid to the blood has been shown with in vivo^6,7,29 and in vitro studies^13,20 using either ³H L-glutamate,^13,20,29 quantifications of absolute L-glutamate concentrations in the brain and blood,⁷ or quantifications of L-glutamate in brain perfusate.⁶ However, the complete transport pathway including the influence of L-glutamate metabolism is not well understood.

Uptake across the abluminal membrane was mainly facilitated by EAAT-1

Abluminal L-glutamate uptake in brain endothelial cells followed Michalis Menten-like kinetics with a K_M of 63 ± 31 µM, which is within the range of reported affinities for EAATs.¹ TBOA and UCPH-101 significantly inhibited the uptake, whereas DHK and the ASCT-2 substrate, L-serine, had no effect. TBOA is a general EAAT inhibitor with K_I values of 2.2–10 µM at EAAT-1/-2 and -3,³⁰ but it does not inhibit ASCT-2.³¹ UCHP-101 is a specific inhibitor of EAAT-1. It has K_I values of 0.66 µM at EAAT-1 and >300 µM at EAAT-2 and -3.^32,33 DHK is known to specifically inhibit EAAT-2 with K_I values of 31 µM at EAAT-2 and >3000 µM at EAAT-1 and -3.³⁰ Thus, the inhibition pattern indicates that EAAT-1 is the main subtype responsible for the initial abluminal L-glutamate uptake in the BBB model. The abluminal uptake experiments were performed in non-contact co-cultures, since astrocyte presence at the abluminal side of the permeable supports could lead to a large overestimation of glutamate uptake because of astrocytic EAAT-1 and -2 expression. This may cause differences in EAAT expression in the endothelial cells. However, immunocytochemistry staining demonstrated that EAAT-1 was present with similar endothelial localization in both non-contact and contact co-cultures (supplementary Figure 1). Furthermore, we demonstrated in this study and previously that at least TEER and transferrin receptor expression levels are not altered by the localization of the astrocytes.³⁴

Luminal L-glutamate uptake was mediated by a low affinity L-glutamate transporter

Initial luminal L-glutamate uptake rates followed Michaelis–Menten-like kinetics with a K_M of 891 ± 453 µM. The kinetic characteristics and inhibitor profile may match a previously reported low affinity L-glutamate transporter,¹ which has been shown to transport L-glutamate with an affinity around 1 mM and to be inhibited by 5 mM L/D-aspartate and D-glutamate.^35–37 Benrabh et al.³⁸ have previously demonstrated L-glutamate uptake in perfused rat brains, which was inhibited by 1 mM L-aspartate or L-glutamate and 5 mM homocysteate, and Lee et al.³⁹ observed L-glutamate uptake with a K_M around 2.5 mM in luminal membrane vesicles from bovine brain capillaries. Thus, the transport of L-glutamate across the luminal membrane may be mediated by a low affinity transporter, at least in the lumen-to-cell interior direction. The low affinity of the transporter indicates that it will mainly function during pathological events with increased extracellular L-glutamate concentrations. However, EAATs are able to concentrate L-glutamate intracellularly because of the coupling to the sodium gradient.⁴⁰ Furthermore, glutamate affinity to the intracellular facing binding site may differ from the extracellular affinity examined in this study. Hence, the low affinity transporter may be able to transport L-glutamate from the endothelial cells to the blood under physiological conditions. The change from culture medium to HBSS in all uptake studies caused TEER to decrease by approximately 50%, which has been observed previously in the same model.⁴¹ However, the cultures retain sufficient tightness to enable polarized uptake studies. To confirm this, TEER prior to the uptake experiments, glutamate permeability and mannitol permeability in all uptake studies were compared to the glutamate uptake (see supplementary Figure 2). No correlation was observed between TEER and glutamate uptake, mannitol permeability and glutamate uptake or glutamate permeability and glutamate uptake, whereas the mannitol and glutamate permeability correlate with the measured TEER levels. This supports that paracellular leakage of L-glutamate is not a determining factor for the glutamate uptake.

L-glutamate is metabolized in the TCA cycle in both astrocytes and endothelial cells

Astrocytes are known to metabolize L-glutamate to glutamine via glutamine synthetase and to α-ketoglutarate via glutamate dehydrogenase.¹ Both compounds as well as downstream TCA cycle products were found in the astrocytes confirming both direct metabolism (for the M+5/M+4 compounds) and continued metabolism in the TCA cycle (for the M+3/M+2 compounds).⁴² The labeled metabolites, mainly L-aspartate, malate, and citrate were also found in the abluminal culture medium demonstrating release from the astrocytes as previously reported.^43–46 Endothelial L-glutamate metabolism is not thoroughly described. However, brain endothelial cells have a high density of mitochondria and may utilize L-glutamate as an energy substrate to fuel the ABC-efflux transporters.^22,47 Furthermore, mRNA expression of the genes encoding glutamine synthetase and glutamate dehydrogenase has been shown in mouse brain endothelial cells, although the absolute expression levels were far below those in astrocytes.⁴⁸ The brain endothelium has also been shown to express mitochondrial branched chain amino acid transferases, which could catalyze the metabolism of L-glutamate to α-ketoglutarate.²¹ In the present study, labeled TCA cycle intermediates as well as L-glutamine and L-aspartate were found in the endothelial cells. The luminal medium contained M+5 L-glutamate, which confirms that it can be transported intact across the BBB as previously shown.²⁰ Similarly, M+4 L-aspartate was found in the luminal medium, which could have been synthetized in the endothelial cells from M+4 oxaloacetate or taken up intact via EAAT-1 after release from the astrocytes. Smaller amounts of M+5 L-glutamine were transported to the luminal medium, especially considering the high donor concentration in the abluminal medium. This supports previous studies showing that glutamine synthetase inhibition (via methionine sulfoximine) had no effects on the apparent L-glutamate efflux.^13,20

Extensive M+5 and M+6 labeling of citrate shows concerted action of pyruvate dehydrogenase and malic enzyme, since this can only be formed via malic enzyme-catalyzed- malate conversion to pyruvate followed by re-entry in the TCA cycle of the labeled pyruvate via acetylCoA reaction with labeled oxaloacetate.⁴² This shows that L-glutamate is vividly metabolized in endothelial cells and thereby augments the capacity for endothelial cells to eliminate L-glutamate. The labeled citrate in the endothelial cells may also originate from metabolism in the astrocytes followed by release to the culture medium.^43,49 However, the degree of citrate labeling was higher in endothelial cells than in the astrocytes for both M+2, M+5, and M+6 citrate, which suggests local metabolism. Furthermore, the gene expression levels of the five SLC13 members, which are mainly responsible for uptake of TCA cycle intermediates are very low in brain endothelial cells.^48,50 This particular pathway for L-glutamate metabolism is essential for a complete oxidative degradation of the molecule to CO₂. If malate is not decarboxylated to pyruvate, the entrance of L-glutamate into the TCA-cycle serves an anaplerotic function.⁴² The luminal medium contained high amounts of M+1 and M+2 lactate. This again indicates that malic enzyme is active in the endothelial cells converting malate formed from L-glutamate to pyruvate, which is then converted to lactate catalyzed by lactate dehydrogenase. M+1 and M+2 lactate were present in substantially larger amounts than M+5 L-glutamate and M+4 L-aspartate in the luminal medium, whereas the presence of labeled lactate was low in all other compartments, indicating that the converted lactate is mainly excreted to the luminal side of the BBB, which could take place through MCT-1 mediated transport.^51,52 It has previously been suggested that oxidation of L-glutamate to lactate may be a major cataplerotic pathway for the brain in order to balance high extent of anaplerosis.²³ It is thus likely that a large fraction of the applied L-glutamate will be transported across the BBB as lactate, either formed in astrocytes and transported through the endothelial cells or formed directly in endothelial cells and released to the luminal medium.

The total BBB-mediated L-glutamate clearance takes place through a combination of glutamate transport and glutamate metabolism but is highly dependent on the presence of astrocytes

Absence of astrocytes in the non-contact co-culture model caused a significant decrease in the apparent L-glutamate brain-to-blood transport, down to levels approximating the blood-to-brain transport. Similar to this study, Cohen-Kashi-Malina et al.¹³ found significantly lower A-L transport of L-glutamate in non-contact co-culture as compared to contact co-culture.¹³ These differences could be caused by astrocyte mediated transport of L-glutamate to the endothelial cells via hemichannels or by a regulation of EAATs by the absence or presence of astrocytes; however, ³H L-glutamate transport was only examined in non-contact co-cultures without the astrocytes present. The data in the present study further show that it is the presence of astrocytes and not the localization (contact versus non-contact) that alters the apparent ³H label permeability, which points towards a significant contribution from metabolism in the astrocytes.

However, blood-glutamate scavenging in rats leads to a lowering of brain L-glutamate levels and improved outcome after different CNS pathologies,^7–11 which indicates a transport of intact L-glutamate from brain-to-blood. This is not necessarily dependent on EAAT-mediated efflux across the BBB, since L-glutamate may freely diffuse to the blood through a leaky BBB, which is observed during stroke, traumatic brain injury, and subarachnoid hemorrhage.^53–56 Alternatively, EAAT expression in brain endothelial cells may be upregulated during pathologic events, which would increase the capacity of the BBB to efflux L-glutamate to the blood.

The astrocyte-endothelial co cultures can be used for uptake, transport and metabolism studies but may not fully represent all the in vivo properties of the neurovascular unit

In general, in vitro systems can be useful to elucidate mechanisms at the cellular level. However, in vitro systems are simplifications and do not fully represent the physiological conditions. Pericytes, interneurons, and microglia are also parts of the neurovascular unit and may play a role in the L-glutamate clearance in vivo. Pericytes have previously been shown to have important effects on the development of the neurovascular unit in vivo;^57,58 however, their role in in vitro models is not clear. Beneficial effects of pericytes on TEER have been demonstrated in rat triple culture models,^59,60 whereas pig endothelial-pericyte co-cultures have shown both TEER decreasing and increasing effects.^61,62 Pericyte effects on EAAT expression and function are unknown, and inclusion of pericytes may provide more information on glutamate transport across the BBB. The model applied in this study expresses EAAT-1, which is also found in ex vivo bovine brain capillaries (see supplementary Figure 1) and in human brain capillaries.¹⁹ EAAT-1 staining was punctuate in both the capillaries and cultured cells and seemingly mainly localized in the cytosol and perinuclear. However, the dimensions of the endothelial cells where the distance between the abluminal and luminal membranes are often around 0.2 µm, make it difficult to clearly assess membrane localization, especially if the target protein is not uniformly expressed throughout the membrane, which seems to be the case for EAAT-1 in the brain endothelial cells.¹⁹ Furthermore, the uptake data clearly supported functional abluminal expression in the brain endothelial cells indicating that the model is suitable to examine cellular events in the overall process of BBB-mediated glutamate clearance.

Conclusion

The main pathways for L-glutamate clearance across the BBB are outlined in Figure 5.

Figure 5.

Suggested main metabolism and transport pathways for L-glutamate at the blood–brain barrier. L-glutamate in the brain interstitial fluid can be taken up via EAAT-1 into the brain endothelial cells, where accumulation can lead to facilitated transport across the luminal membrane via the unknown transporter, X, transporting mainly L-glutamate and L-aspartate. Transporter X has low affinity for L-glutamate and L-aspartate and is thus depending on high intracellular concentrations to facilitate transport. However, uncertainties exist about the kinetics of the luminal transporter since these determinations were made as luminal-to-cell uptake and not cell-to-lumen efflux. The transporter may have different properties for intracellular L-glutamate binding and transport. Simultaneously, L-glutamate is metabolized in astrocytes and endothelial cells to TCA cycle intermediates, as well as pyruvate, lactate, and aspartate, which can subsequently be transported to the blood.

α-KG: α-ketoglutarate; AAT: aspartate aminotransferase; AGC1: mitochondrial aspartate/glutamate carrier-1; EAAT: excitatory amino acid transporter; GDH: glutamate dehydrogenase; Gln: glutamine; Glu: L-glutamate; GS: glutamine synthetase; ISF: interstitial fluid; LAC: lactate; LDH: lactate dehydrogenase; MCT: monocarboxylic acid transporter-1; MPC1/2: mitochondrial pyruvate carrier-1/2; ME: Malic enzyme; Pyr: pyruvate; SNAT: small neutral amino acid transporter; TCA: tri-carboxylic acid cycle; X?: unknown luminal glutamate transporter. The figure is modified from Cederberg et al.⁵

L-glutamate is taken up at the abluminal membrane via EAAT-1 and an unidentified low-affinity transporter mediates the transport from the endothelial cell to the blood. However, L-glutamate is also metabolized in the endothelial cells and in astrocytes with subsequent transport of TCA cycle intermediates, aspartate, and lactate, which constitutes the majority of the BBB-mediated clearance of L-glutamate.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Blanca Aldana’s post-doctoral stay at the Department of Drug Design and Pharmacology is subsidized by the Grant from the Ministry, Technology and Innovation (SECITI) of Mexico. Birger Brodin acknowledges the funding from the Lundbeck Foundation Research Initiative on Brain Barriers and Drug Delivery (RIBBDD), the Predicting Drug Absorption Consortium and the Carlsberg Foundation.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

HCCH and BB prepared the manuscript outline. HCCH, BIA, SG, and MMJ performed the experiments. HCCH, BB, HSW, and CUN designed the experiments and BB, HSW, and CUN provided feedback to experimental work. HCCH and BIA wrote the manuscript. All authors participated in the feedback and writing process following the initial drafting of the manuscript.

Supplementary material

Supplementary material for this paper can be found at the journal website: http://journals.sagepub.com/home/jcb.