GLAST/EAAT1-induced Glutamine release via SNAT3 in Bergmann glial cells: evidence of a functional and physical coupling


Address correspondence and reprint requests to Arturo Ortega, Ph.D., Departamento de Genética y Biología Molecular, Cinvestav-IPN, Apartado Postal 14-740, México DF 07000, México.



Glutamate, the major excitatory transmitter in the vertebrate brain, is removed from the synaptic cleft by a family of sodium-dependent glutamate transporters profusely expressed in glial cells. Once internalized, it is metabolized by glutamine synthetase to glutamine and released to the synaptic space through sodium-dependent neutral amino acid carriers of the N System (SNAT3/slc38a3/SN1, SNAT5/slc38a5/SN2). Glutamine is then taken up by neurons completing the so-called glutamate/glutamine shuttle. Despite of the fact that this coupling was described decades ago, it is only recently that the biochemical framework of this shuttle has begun to be elucidated. Using the established model of cultured cerebellar Bergmann glia cells, we sought to characterize the functional and physical coupling of glutamate uptake and glutamine release. A time-dependent Na+-dependent glutamate/aspartate transporter/EAAT1-induced System N-mediated glutamine release could be demonstrated. Furthermore, D-aspartate, a specific glutamate transporter ligand, was capable of enhancing the co-immunoprecipitation of Na+-dependent glutamate/aspartate transporter and Na+-dependent neutral amino acid transporter 3, whereas glutamine tended to reduce this association. Our results suggest that glial cells surrounding glutamatergic synapses may act as sensors of neuron-derived glutamate through their contribution to the neurotransmitter turnover.

Abbreviations used



analysis of variance


Bergmann glial cells


bovine serum albumin






(1R,2R)-3-[(1S)-1-amino-2-hydroxy-2-oxoethyl]cyclopropoane-1,2-dicarboxylic acid




DL-threo-b-Benzyloxyaspartic acid


Dulbecco's modified Eagle's medium



EAAT 1-5

excitatory amino acid transporters


enhanced chemiluminescence reagent


ethylenediaminetetraacetic acid


Na+-dependent glutamate/aspartate transporter




glutamate transporter-1


glutamine synthetase




kainate binding protein


l-(+)-2-amino-4-phosphonobutyric acid






Nonidet P-40


phosphate-buffer saline


phosphate-buffer saline containing 0.3% Triton


sodium dodedecyl sulfate polyacrylamide gel electrophoresis


Na+-dependent neutral amino acid transporter 2-3-5


Tris-buffered saline




(+/-)-1-aminocyclopentane-trans-1,3-dicarboxilic acid

l-Glutamate (l-Glu) is the major excitatory amino acid neurotransmitter in the vertebrate brain. Membrane-specific l-Glu receptors expressed in neurons and glial cell mediate most, albeit not all, of its effects. Ionotropic receptors of the α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA), N-methyl-D-aspartate (NMDA), and kainate (KA) subtypes are proteins composed of different subunits (Sager et al. 2009). Metabotropic Glu receptors have been subdivided in terms of sequence similarity and signaling mechanisms in three groups. Signaling through these receptors is carried out by the phospholipase C and Ca2+ for Group I and inhibition of adenylate cyclase for Groups II and III (Wieronska and Pilc 2009).

l-Glu extracellular levels are tightly regulated through its uptake, mostly into glial cells, by a family of Na+-dependent l-Glu transporters known as excitatory amino acid transporters of which EAAT-1, also known as Na+-dependent glu/aspartate transporter (GLAST) and EAAT-2 or Glu transporter-1 (Glt-1) are mainly expressed in glia cells (Danbolt 2001). Once this amino acid has been removed, it is rapidly converted to l-glutamine (l-Gln) through the action of Gln synthetase (GS) (Shank and Campbell 1984). Neutral amino acid transporters mediate both the glial release and the neuronal uptake of l-Gln. Several transporter systems have been described for these amino acids, based not only in sequence identity but also on their kinetic properties (Dolinska et al. 2000, 2003; Hagglund et al. 2011). In general terms, it has been assumed that l-Gln release from astrocytes (Albrecht 1989). One of the transporters that carries out this release is the Na+-dependent neutral amino acid transporter 3 (SNAT3), a member of System N, a family of l-Gln transporters capable of acting in a reversed fashion (Broer et al. 2004). In contrast, neuronal l-Gln uptake has been postulated to be carried out by members of System A transporters (SNAT2) (Jenstad et al. 2009), although this proposal has been questioned at least for neocortical neurons (Grewal et al. 2009).

Bergmann glial cells (BGC) extend their processes through the molecular layer of the cerebellar cortex surrounding excitatory and inhibitory synapses (Somogyi et al. 1990). The tripartite relationship between pre-synapse, post-synapse, and surrounding glia as a source of neuroactive substances like ATP and D-serine has also been acknowledged (Henneberger et al. 2010). Therefore, it has been postulated that BGCs are involved in neuronal communication and that they might even constitute a neuronal reservoir (Hansson and Ronnback 1995; Malatesta et al. 2003; Anthony et al. 2004). When cultured, BGC become an excellent model in which the molecular and cellular basis of glial–neuronal signaling can be analyzed in the context of glutamatergic transmission (Lopez-Bayghen et al. 2007). In such preparations, l-Glu acting through its receptors, changes gene expression at the transcriptional and translational levels (Gonzalez-Mejia et al. 2006; Rosas et al. 2007). In addition, it has become evident that plasma membrane l-Glu transporters are also involved in glutamatergic signaling. For example, it has been demonstrated that l-Glu uptake activity triggers glucose influx (Magistretti 2009) and increases GS activity (Lehmann et al. 2009) in astrocytes. Furthermore, in BGC cultures, GLAST/EAAT-1, the sole l-Glu transporter expressed in these cells (Regan et al. 2007) is linked to the mammalian target of rapamycin (mTOR) activation and thus protein expression regulation (Martinez-Lozada et al. 2011). With these evidences in mind, we decided to explore the putative role of l-Glu transporters in the regulation of l-Gln release. To reach this goal, we first characterized the L-[3H] Gln uptake activity in our culture system and evaluated if GLAST/EAAT-1 substrates such as D-Aspartate (D-Asp) would induce a significant l-Gln release. Once we were able to detect such an effect, we hypothesized that both transporters are present in a macromolecular complex and that D-Asp treatment would be linked to the appearance of such complex. Indeed, we were able to detect a coupling between GLAST and SNAT3.

These results suggest that glia cells associated to glutamatergic synapses behave as l-Glu sensors and that these cells may control, to a greater extent than assumed before, excitatory transmission. Moreover, our findings strengthen the notion of the active participation of glia cells in synaptic transmission.

Materials and methods


Tissue culture reagents were obtained from GE Healthcare (Carlsbad, CA, USA). DHPG, 6,7-Dinitroquinoxaline-2,3-dione, AMPA, NMDA, DL-threo-b-Benzyloxyaspartic acid, threo-β-hydroxyaspartate and l-Glu were all obtained from Tocris-Cookson (St. Louis, MO USA). KA was obtained from Ocean Produce International (Shelburne, Nova Scotia, Canada). L-[3H] Glutamic acid (specific activity 40 Ci/mmol) and L-[3H] Glutamine (specific activity 50.3 Ci/mmol) were obtained from Perkin Elmer (Waltham, MA, USA). The AG 1X-8 anion exchange resin was purchased from Bio-Rad (Hercules, CA, USA). The antibodies used were anti- SNAT3 (Santa Cruz, CA, USA), anti-calbindin (Sigma-Aldrich, St. Louis, MO, USA), anti-kainate binding protein (KBP), and an anti-peptide (GLIQALVTALGTSSSSAT) GLAST anti-serum (produced and characterized in our laboratories). Specificity of the anti-peptide anti-GLAST antibodies was performed as follows. A 1 : 1000 dilution of the sera was pre-absorbed with 10 and 50 ng of the peptide for 30 min at 4°C and used for Western blot analysis of chick cerebellum and cultured Bergmann glia extracts (see below). Horseradish peroxidase-linked anti-mouse or anti-rabbit antibodies, and the enhanced chemiluminescence reagent, were obtained from Amersham Biosciences (Buckinghamshire, UK). All other chemicals were purchased from Sigma (St. Louis, MO, USA).

Cell culture and stimulation protocol

Primary cultures of cerebellar BGC were prepared from 14-day-old chick embryos as previously described (Ortega et al. 1991). Cells were plated in plastic culture dishes in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, 2 mM Gln, and gentamicin (50 μg/mL) and used on the fourth or fifth day after culture. Before any treatment, confluent monolayers were switched to non-serum DMEM media containing 0.5% bovine serum albumin for 30 min and then treated as indicated. Antagonists or inhibitors were added 30 min before agonists. The cells were treated with the l-Glu analogs added to culture medium for the indicated times; after that, medium was replaced with DMEM/0.5% albumin.

Staining procedures

BGC primary cultures were grown on poly-L-lysine-treated (0.01 mg/mL) glass coverslips following the procedure described above. Cells were fixed by exposure for 10 min to methanol at −20ºC and washed twice with phosphate-buffer saline (PBS) containing 0.5% Triton X-100 (washing solution). Non-specific binding was prevented by incubation with 1% bovine serum albumin in PBS (blocking solution) for 1 h. Cells were exposed 1 h to the primary antibodies anti-SNAT3 in blocking solution at 25°C. Then, cells were washed three times with washing solution and incubated with a 1 : 100 dilution of the fluorescent-labeled secondary antibodies dissolved in blocking solution. After washing out secondary antibody, cell preparations were mounted with Vectashield (Vector Laboratories, Burlingame, CA, USA) and examined with an inverted fluorescence microscope (Zeiss Axioscope 40, Gottingen, Germany).

For immunohistochemistry, P0 chick cerebella was removed and placed immediately in cold PBS. The tissue was washed once in cold PBS to remove blood and placed in 4% paraformaldehyde in PBS (pH 7.4) for one hour. The fixative was changed once and the tissue was left at 4°C during 48 h. The cerebella were cryoprotected successively in 10%, 20%, and 30% sucrose in PBS and sagittally sectioned at 50 μm with a cryostat (Microm International GmbH, Walldorf, Germany). For immunohistochemistry, tissue was washed profusely in PBS to remove excess aldehydes and then incubated 10 min in 1.8% hydrogen peroxide solution to remove endogenous peroxidase activity. Non-specific antibody binding was blocked by incubating the sections in blocking solution (3% goat serum in PBS containing 0.3% Triton, PBT) for 1 h at 25°C. Sections were incubated at 4°C for 48 h with primary antibodies; mouse anti-calbindin (1 : 5000), anti-KBP (1 : 2500), and anti-SNAT3 (1 : 2500) in blocking solution. The sections were washed three times in PB, and placed in biotinylated anti-mouse secondary antibodies (1 : 1000; Vector Laboratories, Inc., Burlingame, CA, USA) for 3 h (for calbindin), anti-rabbit secondary antibodies for KBP, and anti-guinea pig secondary antibodies for SNAT3. A 1 h incubation with the avidin-biotinylated horseradish peroxidase complex (1 : 250; Vector Labs) followed. The antibody-peroxidase complexes were revealed with a solution containing 0.05% diaminobenzidine, nickel sulfate (10 mg/mL; Fisher Scientific, Pittsburg, PA, USA), cobalt chloride (10 mg/mL; Fisher Scientific), and 0.01% hydrogen peroxide, which produced a black–purple precipitate. Sections were mounted onto gelatin-subbed slides, dehydrated, and cleared in Hemo-De (Fisher Scientific); then, cover slips were collocated with Permount. The sections were analyzed in an Olympus BX41 microscope.

L-[3H]Gln Uptake and release

Confluent BGC monolayers seeded in 24-well plates were washed three times to remove all non-adhering cells with 0.5 mL aliquots of solution A containing 25 mM HEPES-Tris, 130 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2. 0.8 mM MgCl2, 33.3 mM glucose, and 1 mM NaH2PO4 at pH 7.4. When indicated, NaCl was replaced by LiCl. The time course of L-[3H] Gln influx was initiated at = 0 by the addition of 0.5 ml solution A containing 0.5 μCi/mL of L-[3H] Gln solution A, 200 μM Gln. The reaction was stopped by aspirating the radioactive medium and washing each well within 15 s with 0.5 mL aliquots of an ice-cold solution A. For the determination of the kinetic parameters, the cold l-Gln concentration was modified to a final 0.5, 3, 5, 7.5, and 10 mM concentration and the uptake time was 30 min. The uptake was stopped as described above. The cells in the wells were then exposed for 2 h at 37°C to 0.5 mL NaOH and an aliquot of that solution counted in a Beckmann 7800LS scintillation counter. A minimum of three experiments in quadruplicates was done for each condition.

For the release experiments, BGC were seeded on 60 mm dishes and loaded for 3 h with 0.5 μCi/mL of L-[3H] Gln in solution A. The medium was replaced every 2 min and a total of 15 fractions were collected. The various stimuli were added in fractions 6-10. The radioactivity associated to each fraction was determined by liquid scintillation counting and expressed as percentage of the total radioactivity in the experiment.

L-[3H] Glutamine metabolism

The extent of L-[3H] Gln remaining after the 3 h loading period and the percentage of L-[3H] Gln released was determined by means of anion exchange separation of L-[3H] Gln, using a AG 1X-8 200/400 resin essentially as described previously (Mongin et al. 2011). The system was standardized with L-[3H] Glu and L-[3H] Gln, that were eluted with HCl and H2O, respectively.

Immunoprecipitation and Western blots

Cells from confluent monolayers were harvested with PBS (10 mM K2HPO4/KH2PO4, 150 mM NaCl, pH 7.4) containing protease inhibitors (1 mM phenylmethylsufonyl fluoride, 1 mg/mL aprotinin, 1 mg/mL leupeptin). Tissues or cells suspensions were lysed with RIPA buffer (50 mM Tris-HCl, 1 mM ethylenediaminetetraacetic acid, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycolate, pH 7.4). Cell lysates were pre-absorbed with 15 μL of protein G coupled to Sepharose 4B for 25 min at 4°C. The cleared lysates (1 mg of protein) were incubated with agarose-coupled anti-GLAST or anti-SNAT3 antibodies for 10 h, 4°C and then immunoblotted.

For Western blots, immunoprecipitates or cell lysates were denaturized in Laemmli's sample buffer, resolved through 10% sodium dodedecyl sulfate polyacrylamide gel electrophoresis and then electroblotted to nitrocellulose membranes. Blots were stained with Ponceau S stain to confirm that protein content was equal in all lanes. Membranes were soaked in PBS to remove the Ponceau S and incubated in Tris-buffered saline containing 5% dried skimmed milk and 0.1% Tween 20 for 60 min to block the excess of non-specific protein binding sites. Membranes were then incubated overnight at 4°C with the particular primary antibodies indicated in each figure, followed by the adequate secondary antibodies. Immunoreactive polypeptides were detected by chemiluminescence and exposed to X-ray films. Densitometric analyses were performed and data analyzed with the Prism, GraphPad Software (San Diego, CA, USA).

Statistical analysis

Data are expressed as the mean values (average) ± the standard error (SE). Statistical analyses were performed with a one-way anova, and Tukey post hoc comparisons were considered statistically significant when p ≤ 0.05.


l-Gln transporter SNAT3 is expressed in cultured chick cerebellar BGC

As a first step in the characterization of a plausible coupling between l-Glu uptake and l-Gln release, we decided to explore the biochemical nature of l-Gln uptake activity in BGC. Taking into consideration that of the l-Gln transporters described thus far in glial cells, the N family is capable to function in a reverse mode, we concentrated in this system and particularly in SNAT3, because it has been described that the highest levels of this transporter are present from post-natal day 7 in the rat cerebella, which corresponds roughly to chick embryonic day 14 (middle stage of cerebellar development) (Boulland et al. 2003). To gain insight into SNAT3 expression in BGC, we performed immunolabeling of P0 chick cerebella with an anti-SNAT3 specific antibody along with staining for specific cell markers, calbindin, KBP, and GLAST/EAAT1 to identify the layers of the cerebellar cortex.

It is important to mention that the anti-GLAST/EAAT1 antibody used in this study was generated in our lab against a peptide sequence specific of the chick protein, that it is known to possess a divergent C-terminus from the rat, mouse, and human sequence (Espinoza-Rojo et al. 2000). The specificity of this anti serum is presented in Fig. 1a. Pre-absorption of a 1 : 1000 dilution of the antibodies with 50 ng of the peptide prevents completely the recognition of the characteristic 60 kDa band. Figure 1b illustrates the localization of the Purkinje cell bodies and their arborizations after immunolabeling with the Purkinje's cell marker, calbindin. BGC were localized through the use of anti-KBP antibodies, and by GLAST/EAAT1 immunoreactivity. KBP is expressed exclusively in chick cerebellum BGC (Somogyi et al. 1990) and GLAST/EAAT1 is the major cerebellar glia transporter (Ottersen et al. 1997). SNAT3 immunostaining was abundantly found in the pial border lining the molecular cell layer as well as in Bergmann glia processes, in a similar fashion as KBP labeling, in contrast to calbindin that strongly labels the Purkinje cell soma and arborizations. Note that the pial cerebellar surface corresponds to BGC terminal end-feet (Fig. 1b). Immunofluorescence experiments in our cultured cells demonstrate that SNAT3 is present in cultured BGC both in the cytoplasm and in the plasma membrane (Fig. 1c). Finally, in Fig. 1(d), the 60-kDa immunoreactive band that corresponds to SNAT3 was detected. Taken together, these results show that SNAT3 is present in Bergmann glia both in situ as well as in primary cultures.

Figure 1.

Na+-dependent neutral amino acid transporter 3 (SNAT3) expression in Bergmann glia. (a) Specificity of anti Na+-dependent glutamate/aspartate transporter (GLAST)/EAAT1 anti-peptide antibodies. Total protein extracts from cultured chicken Bergmann glial cells (BGC) (50 μg) and cerebellum (75 μg) were assayed by Western blot with and without pre-absorption with 50 ng of the C-terminus chick GLAST peptide. (b) SNAT3 expression in chick cerebellum. Immunohistochemical characterization of chick cerebellar sections. Calbindin (Calb) labels Purkinje cells; kainate binding protein (KBP) and GLAST/EAAT1 are the BGC markers. Bar size: 50 μm. (c) Cells staining with anti-SNAT3 antibodies in cultured BGC (green). Nuclei were counter-stained with DAPI (blue). (d) Total protein extracts from chick BGC cultures and cerebellum were assay by Western blot with anti-SNAT3 antibodies. A representative Western blot is shown.

Characterization of L-[3H]Gln uptake activity in cultured Bergmann glia

To establish a functional coupling between l-Glu uptake and l-Gln release, we measured the kinetic parameters of L-[3H] Gln uptake in our culture system. Since it is known that System N is the only l-Gln uptake system capable to function with Li+, we decided to use a LiCl-containing solution A. The results are presented in Fig. 2. A time-dependence of L-[3H] Gln accumulation in LiCl-buffer was observed (Fig. 2a). By increasing l-Gln concentrations, we were able to determine the kinetic parameters of L-[3H] Gln uptake. As depicted in Fig. 2b, the Li+-tolerant component displays a KM of 2.925 mM and a Vmax of 1.818 μmol/min*mg. In addition, we made an amino acid competition experiment, we used MeAIB, the specific inhibitor of system A, alanine as a competitor for System Asc, A and N; for system N we used histidine and the mix of leucine and threonine as competitors for systems Asc and L. As shown in Fig. 2c, His reduces approximately 50% of Gln uptake in BGC, which is similar to the amount of uptake that remains in Li+ containing assay solution. Overall these results suggest that l-Gln transport in BGC is mediated in an approximate 40% by SNAT3.

Figure 2.

System N kinetic parameters in cultured Bergmann glial cells (BGC). (a) Time course of L-[3H] Gln uptake in BGC. The uptake was carried out in assay solution in which NaCl was replaced by LiCl (see Methods). A fixed amount of 0.5 μCi/mL L-[3H] Gln, 200 μM l-Gln final concentration was present in the uptake assay for the indicated time periods. (b) L-[3H] Gln uptake assay was performed with increasing l-Gln concentrations (0 mM, 0.5 mM, 3 mM, 5 mM, 7.5 mM, 10 mM). Transport kinetics Vmax (μmol/min per mg protein) and KM (mM) were determined by non-linear regression with the Prism 5 software (GraphPad) (c) Effect of different amino acids or the replacement of NaCl by LiCl on L-[3H] Gln uptake in BGC. l-Gln was added at 200 μM concentration, the A-System specific substrate MeAIB was added at a 0.5 mM concentration, while other amino acids were added at a 10 mM concentration. Values represent mean ± SEM of a three independent experiments in quadruplicate, reproduced three times with similar results. Data were compared with non-stimulated of normal solution A by a Student's t-test (*** p < 0.001).

Functional coupling of GLAST with SNAT3

In BGC, l-Glu uptake is carried out by GLAST/EAAT1 (Ruiz and Ortega 1995) and D-Asp is capable to be transported by this carrier in an electrogenic fashion, with a net Na+ influx. Under such circumstances, we reasoned that the uptake of L-[3H] Gln, that relies on the Na+ gradient would have to be inhibited. The results are presented in Fig. 3a, D-Asp at saturating concentrations (60 and 120 μM), prevents L-[3H] Gln uptake. As expected, this effect is dose dependent with an IC50 of 0.873 μM (Fig. 3b). It should be noted that this value is only indicative of a specific transporter mediated effect and by no means reflects the affinity of D-Asp toward GLAST/EAAT-1 as established by D-[3H] Asp uptake experiments. It should be noted a common property of any signaling cascade is amplification (Seger and Krebs 1995). These results prompted us to evaluate if the exposure of L-[3H]-Gln-loaded BGC to l-Glu or D-Asp resulted in an increase in the presence of radioactivity in the medium, consequence of a reversed transport of the tracer through SNAT3. Indeed this is the case, as depicted in Fig. 4: l-Glu treatment resulted in an induced L-[3H] Gln release. The identity of the released material as L-[3H] Gln was determined by ion exchange chromatography in a subset of experiments. Each fraction collected under the stimulus (2 mL) of these experiments (three independent release experiments), as well as the final lysate, were added onto an activated 2 ml AG 1X-8 200/400 anion exchange column to separate L-[3H] Gln from its metabolites. The column content was eluted with 2 ml volumes of H2O, followed by three 2 ml volumes of 0.1 M HCl. Water elution removed uncharged L-[3H] Gln, while subsequent acid elution extracted negatively charged metabolites such as l-Glu, α-ketoglutarate and other tricarboxylic acid intermediates (Mongin et al. 2011). The system was standardized with the application of a 150 μL aliquot of 10 μΜ L-[3H] Glu aliquot and a 150 μL sample of 10 μΜ L-[3H] Gln, that were eluted with HCl and H2O, respectively. In all the fractions tested, roughly 70% of the radioactivity passed through the column as the L-[3H] Gln standard. Moreover, this percentage was also found in the final cell extract, suggesting that most of the material released (> 70%) corresponds to l-Gln (Deitmer et al. 2003).

Figure 3.

Effect of D-Aspartate (D-Asp) in l-Gln uptake. (a) Effect of two different D-Asp concentrations (60 and 120 μM) in BGC L-[3H] Gln uptake activity. Data were compared with non-stimulated cultures by a Student's t-test (*p < 0.05). (b) IC50 calculation of the D-Asp effect on L-[3H] Glu uptake. Results are the mean ± SEM of three independent experiments. GraphPad Prism 5 was used to calculate the IC50 value.

Figure 4.

l-glutamate (l-Glu) induces L-[3H] Gln release from Bergmann glial cells (BGC). Cells were loaded with 0.5 μCi/mL of L-[3H] Gln for 3 h. The media was removed and replaced every 2 min, a total of 15 fractions were collected. Fractions 1–5 and 11–14 contained solution A (see methods section), while in fractions 6–10 the following stimulus were added, D-Aspartate (D-Asp) 1 mM or 50 μM, kainate (KA) 0.5 mM, l-Glu 1 mM in sodium-containing or sodium-free assay buffer. Fraction 15 represents the cell lysates. The radioactivity associated to each fraction was determined by liquid scintillation counting and expressed as percentage of the total radioactivity in the experiment. A typical experiment is shown, reproduced three times with similar results.

Note that exposure to either 50 μM or 1 mM D-Asp, also results in a stimulus-mediated release. In contrast, the AMPA receptors agonist, KA, did not evoke any significant L-[3H] Gln release. These results suggest that l-Glu uptake is likely linked to l-Gln release.

Physical interaction between GLAST and SNAT3

The results described above were suggestive of a plausible physical interaction between GLAST and SNAT3. To test this possibility, we performed immunoprecipitation assays coupled to Western blot identification. The blots presented in Fig. 5a show that immunoprecipitation of BGC lysates with anti-GLAST antibodies and Western blot analysis with anti-SNAT3 antibodies enables us to detect the latter transporter in the immune complexes. As expected, SNAT3 immunoprecipitates contain GLAST. At this point, we decided to explore the possibility that GLAST activity would modulate its interaction with SNAT3. To this end, we exposed confluent BGC cultures to 1 mM l-Glu and/or 2 mM l-Gln. As depicted in Fig. 5b, l-Glu but not l-Gln favor GLAST/SNAT3 physical interaction. We next explored if the transport activity of these two proteins would affect their putative association. To this end, we exposed the cultured cells to 60 μM D-Asp, 200 μM l-Gln, or a combination of both. The results are depicted in Fig. 5c: D-Asp induces GLAST/SNAT3 association (line 2), while l-Gln reduces the D-Asp effect (line 4). Note that 200 μM l-Gln induces a non-significative (compared to control) increase in GLAST/SNAT3 co-immunoprecipitation.

Figure 5.

Na+-dependent neutral amino acid transporter 3 (SNAT3)/Na+-dependent glutamate/aspartate transporter (GLAST) interaction in Bergmann glial cells (BGC). (a) BGC extracts were immunoprecipitated with anti-GLAST antibodies and the immunoprecipitate assayed by Western blot analysis with SNAT3 antibodies. The opposite experiment was done, total BGC proteins were immunoprecipitated with anti-SNAT3 and then subject to Western blot analysis with anti-GLAST serum. (b and c) BGC monolayers were exposed to the indicated stimulus; l-glutamate (l-Glu) 1 mM or l-Gln 2 mM (b), D-Asp 60 μM and l-Gln 200 μΜ (c). Total extracts were inmmunoprecipitated with anti-GLAST and then subjected to Western blot analysis with SNAT3 antibodies. Western blot anti-GLAST was used as loading control. (*p < 0.05) by a Student's t-test. A typical autoradiogram of three independent experiments is shown.


Over the last decade, the concept of the tripartite synapse has been widely supported (Araque et al. 1999; Volterra and Meldolesi 2005; Halassa et al. 2007; Haydon et al. 2009). Glia cells express a battery of neurotransmitter receptors and transporters that enable them to respond to neuronal activity. Within glutamatergic synapses, the fundamental role of astrocytes in the recycling of the neurotransmitter has long been acknowledged (Bak et al. 2006). The l-Glu/Gln shuttle provides good evidence about the capacity of glial cells to respond to synaptic activity through the modification of their cellular functions such as GS expression and activity (Lehmann et al. 2009). Another example of a tight coupling of glia cells to glutamatergic synaptic activity is the astrocyte/neuron lactate shuttle (Magistretti 2009). Furthermore, depolarization of the cerebellar parallel fibers activates Bergmann glial l-Glu receptors and transporters (Balakrishnan and Bellamy 2009).

Using as a model system cultured BGC from chick cerebellum, we characterized several signaling cascades involved in l-Glu-dependent transcriptional regulation (Lopez-Bayghen et al. 2007). We have been able to demonstrate that l-Glu signaling is also supported by the unique l-Glu transporter expressed in these cells, GLAST/EAAT1 (Martinez-Lozada et al. 2011). All these findings, as well as the lack of a complete biochemical characterization of the l-Glu/Gln shuttle, led us to explore a plausible interaction between l-Glu uptake and l-Gln release.

The first issue that we had to tackle before any biochemical experiment could even be designed, was the identification of the expression of l-Gln transporters that function in reverse mode within BGC. Since System N members could fulfill this requirement, and SNAT3 expression is prominent in the days just before birth and in the adulthood in the rat, we restricted our approach to SNAT3. However, we must mention that SNAT5 has been shown to be present in rat Bergmann glia cells (Cubelos et al. 2005). The results presented in Fig. 1, demonstrate, in the one hand, the specificity of our anti-peptide anti-GLAST/EAAT1 antibodies (Fig. 1a) as well as that SNAT3 is expressed in chick cerebellar BGC, both in situ as well as in our culture system. It is important to note the similar labeling of the Bergmann glia processes with anti-GLAST, anti-KBP, and anti-SNAT3 antibodies. Unfortunately, all these antibodies were generated in goat, preventing us to do double immunolabeling experiments. Nevertheless, if one compares the decoration with anti-calbindin antibodies, a Purkinje cells marker, it remains clear that SNAT3 co-localizes with KBP and GLAST.

The fact that the kinetic parameters detected for SNAT3 in our culture system are in the mM range as has been reported for other glia cells, made us confident that indeed, Bergmann glia express SNAT3 transporters. One could argue that several reports settle a Km value of approximately 1 mM (Kilberg et al. 1980; Chaudhry et al. 2001; Broer et al. 2004) for SNAT3 and that the ones we report here are higher (2.9 mM). It should be noted, however, that the reported constants were obtained in heterologous glial cultures, whereas we use an enriched BGC culture (Ortega et al. 1991). Besides, our model system is avian origin in contrast to rodent cells used in the referred studies. Whether the apparent discrepancy is related to the cell population examined or the species used, it is not known at this moment. At this stage, we cannot rule out a plausible involvement of SNAT5 in our measurements of L-[3H] Gln uptake or release. A pertinent observation could also be that if we expect that this carrier would function in a reverse manner, why we would be interested in the characterization of the uptake activity. The answer is simple: it is easier to measure uptake than release, and for kinetic purposes it is valid approach (Broer et al. 2004).

In any event, it became important to demonstrate a l-Glu-induced l-Gln release. Given the fact that in our experimental conditions, > 70% of the L-[3H] Gln loaded in the cells remains as L-[3H] Gln as judged by ion exchange chromatography, the experiments described in Fig. 4 show that D-Asp is capable to trigger l-Gln release in a more efficient manner than l-Glu. A plausible explanation to this finding is that l-Glu binds to receptors and transporters, whereas D-Asp would only binds to GLAST/EAAT1. These results strengthen our confidence in the hypothesis of a complex between these transporters. A detectable immunoreactive SNAT3 polypeptide is present in anti-GLAST immunoprecipitates and vice versa (Fig. 5a). Whether this is a direct protein–protein interaction between these two transporters, it is not known at this moment, work in progress in our lab is aimed at that direction. In support to our hypothesis, it is important to mention that recently the group of Billups, using electrophysiological recordings, has reported that astrocytes juxtaposed to the glutamatergic calyx of Held synapse in the rat medial nucleus of the trapezoid body release l-Gln as a consequence of l-Glu transport activation (Uwechue et al. 2012).

What could it be the purpose of a GLAST/SNAT3 association? One could postulate that in order for SNAT3 to work in reverse mode a significant rise in intracellular Na+ within the vicinity of the transporter should be taking place. A functional coupling with GLAST would provide that local rise in Na+ concentration, if this interpretation is correct then, first of all, D-Asp influx should prevent any L-[3H] Gln uptake, and as demonstrated in Fig. 3, this is exactly the case. The fact that this prevention in L-[3H] Gln uptake is dose dependent suggests that this is an specific effect. A valid argument in the interpretation of these results is that the reported Km of BGC GLAST/EAAT1 is 62 μM and the IC50 values for the prevention in L-[3H] Gln uptake is of 0.873 μM. A plausible explanation for this apparent discrepancy lies in the signal amplification inherent to cell signaling. For example, in BGC, a 1 mM l-Glu has to be applied to record a Na+ inward current (Bennay et al. 2008), whereas the EC50 for l-Glu obtained through Oct-2/DNA binding activity is 164 μM (Mendez et al. 2004). Furthermore, it could be expected that D-Asp should increase GLAST/SNAT3 association whereas l-Gln not. Indeed, the results presented in Fig. 5c support this vision. A model of our interpretation of the results is shown in Fig. 6.

Figure 6.

Schematic representation of the Na+-dependent glutamate/aspartate transporter (GLAST)/Na+-dependent neutral amino acid transporter 3 (SNAT3) complex. l-Glu released from the parallel fibers is taken up into Bergmann glia cells through GLAST. The increase in [Na+]i favors the operation, in reverse mode, of SNAT3 with the consequent release of l-Gln. Glu, l-glutamate; Gln, glutamine; GLU-R, Glutamate receptors; GLAST, sodium-dependent glutamate/aspartate transporter; SNAT, sodium-dependent neutral amino acid transporter.

Taken together, our results provide a support to the critical involvement of glia cells in the modulation of l-Glu-mediated neurotransmission in what is nowadays known as the tripartite synapse.


This study was supported by grants from Conacyt-Mexico to A.O. (79502,188138) and PROMEP/SEP to A.R (UAQ-FOFI2012-FCQ201216). Z.M-L, A.M.G. and M.F-M are supported by Conacyt-Mexico fellowships. The technical assistance of Blanca Ibarra is acknowledged. The authors certify that there is no conflict of interest with any financial organization regarding the material discussed in the manuscript.