Address correspondence and reprint requests to Lortet Sylviane, IC2N-CNRS, 31 chemin Joseph Aiguier, 13402 Marseille Cedex 20, France. E-mail: firstname.lastname@example.org
A co-ordinated regulation between neurons and astrocytes is essential for the control of extracellular glutamate concentration. Here, we have investigated the influence of astrocytes and glia-derived cholesterol on the regulation of glutamate transport in primary neuronal cultures from rat embryonic cortices. Glutamate uptake rate and expression of the neuronal glutamate transporter EAAC1 were low when neurons were grown without astrocytes and neurons were unable to clear extracellular glutamate. Treatment of the neuronal cultures with glial conditioned medium (GCM) increased glutamate uptake Vmax, EAAC1 expression and restored the capacity of neurons to eliminate extracellular glutamate. Thus, astrocytes up-regulate the activity and expression of EAAC1 in neurons. We further showed that cholesterol, present in GCM, increased glutamate uptake activity when added directly to neurons and had no effect on glutamate transporter expression. Furthermore, part of the GCM-induced effect on glutamate transport activity was lost when cholesterol was removed from GCM (low cholesterol-GCM) and was restored when cholesterol was added to low cholesterol-GCM. This demonstrates that glia-derived cholesterol regulates glutamate transport activity. With these experiments, we provide new evidences for neuronal glutamate transport regulation by astrocytes and identified cholesterol as one of the factors implicated in this regulation.
A bi-directional signalling between neurons and astrocytes has been widely demonstrated during the development of the mammalian central nervous system (Sivron et al. 1993; Le Roux and Reh 1994; Travis 1994; Wang et al. 1994; Wu and Barish 1994; Li et al. 1999). It is noteworthy that astrocytes and neurons co-differentiate during the late embryonic and early post-natal life corresponding to the period of synaptogenesis. Recently, a possible role of glial cells in synapse development was indicated by a series of studies on rat retinal cells showing that astrocytes profoundly increase synapse number and are required for their maintenance (Pfrieger and Barres 1997; Ullian et al. 2001). These data suggest a co-ordinated development of neurons and astrocytes. Such interactions could be of primary interest for the elaboration of the glutamatergic synapse because glutamate released in the synaptic cleft is known to be cleared by glutamate transporters expressed by both neurons and astrocytes. Indeed, in the rat brain, three main glutamate transporters have been identified on postsynaptic neurons (EAAC1, Kanai and Hediger 1992) or on nearby glial cells (GLT1, Storck et al. 1992; GLAST, Pines et al. 1992). A correct elimination of glutamate is thus critical to maintain effective synaptic transmission and to prevent the accumulation of glutamate to toxic levels (Choi et al. 1987).
There is now a consensus to suggest that the glial transporters play the major role in the prevention of excitotoxicity (Rothstein et al. 1996; Peghini et al. 1997). The neuronal transporter might be implicated in the fine regulation of neuronal excitability and plasticity at glutamate-mediated synapses. As an example of the possible implication of EAAC1 in memory formation, Levenson et al. (2002) showed that glutamate uptake was up-regulated in the hippocampus during the early phase of long-term potentiation due to a redistribution of EAAC1 on the plasma membrane.
Actually, glutamate transport is a highly regulated phenomenon both at transcriptional and post-translational levels (for reviews, see Sims and Robinson 1999; Danbolt 2001). Recent studies have shown that soluble factors secreted by neurons enhance the expression of GLT1 and GLAST, suggesting that neurons are involved in the regulation of expression of the glial glutamate transporters (Gegelashvili et al. 1997; Swanson et al. 1997; Schlag et al. 1998). In contrast, a possible influence of astrocytes on EAAC1 expression or neuronal glutamate uptake rate has never been explored. Moreover, the relationship between glutamate transporters and synapse elaboration must also be documented. For example, we have no evidence of the effect of cholesterol, known to be secreted by astrocytes and to promote synapse development (Mauch et al. 2001), on glutamate transporter activity and expression.
In this context, the present study was undertaken to determine whether or not astrocytes can modulate the expression and activity of EAAC1. Our experimental model consists of two neuronal cultures from embryonic cortices of rats, without astrocyte (pure neuronal culture) and with few astrocytes (neuron-enriched culture). We measured glutamate transporter expression, glutamate uptake rate and extracellular glutamate concentrations in neuronal cultures treated or not with medium conditioned by astrocytes. We determined that cholesterol was contained in that medium and analysed the effect of cholesterol on glutamate transporter activity and expression by neuronal cultures.
Materials and methods
Wistar rats were from Iffa Credo (L'Arbresle, France). Culture medium (DMEM-F12, DMEM), glutamine and penicillin-streptomycin were purchased from Bio Whittaker (Verviers, Belgium). MK801 [(5R,10S)-(+)-5-methyl-10,11-dihydro-5H-dibenzo(a,d)cyclohepten-5,10-imine] and DL-threo-β-benzyloxyaspartic acid (TBOA) was from Tocris Cookson Ltd. (Bristol, UK). l-[3H]glutamate (specific activity 15–25 Ci/mmol) was from NEN Life Science Products (Zaventem, Belgium). Mevastatin, mouse anti-microtubule associated protein type 2 (MAP2), fluoroscein isothiocyanate (FITC) conjugated goat anti-rabbit and rabbit anti-actin antibody were from Sigma-Aldrich (St Quentin Fallavier, France). Rabbit anti-glial fibrillary acidic protein was from Dako (Glostrup, Denmark), Rhodamine conjugated goat anti-mouse from Immunotech (Marseille, France) and Fluorsave from Calbiochem (San Diego, CA, USA). Protease inhibitors (complete tablets) were from Roche Diagnostics (Xeylan, France). The Hybond ECL nitrocellulose membrane was from Amersham Pharmarcia Biotech (Freiberg, Germany). The affinity purified anti-EAAC1 antibody was from Alpha Diagnostic International (San Antonio, USA), the two glial glutamate transporter antibodies were kindly provided by G. Pietrini (Perego et al. 2000). All other chemicals were from Sigma-Aldrich.
Primary cortical neuronal cultures were prepared as previously reported (Lortet et al. 1999) from cortices of 18-day-old embryos from pregnant Wistar rats. Cells were mechanically dissociated [dissociation medium, 1 : 1 mixture of Dulbecco's modified Eagle's medium (DMEM) and Ham's F12 (F12) medium, supplemented with 2 mm glutamine, 50 IU/mL penicillin and 50 μg/mL streptomycin, and 10% heat-inactivated fetal bovine serum] by means of heat-tapered Pasteur pipettes. After centrifugation (400 g, 10 min), cells were suspended in chemically defined culture medium [DMEM-F12 supplemented with glucose (33 mm), insulin (25 μg/mL), transferrin (100 μg/mL), putrescine (100 μm), progesterone (20 nm), sodium selenite (30 nm), oestradiol (2 pm), arachidonic acid (1 μg/mL), glutamine (2 mm), penicillin (50 IU/mL) and streptomycin (50 μg/mL)]. The number of viable cells was determined by trypan blue exclusion and the cells were plated at a density of 105 cells/cm2 in plastic multi-well plates (costar, Dutscher, Brumath, France) previously coated with poly ornithine 10 μg/mL for 30 min at 37°C. Culture dishes were incubated at 37°C in a humidified 5% CO2-95% air atmosphere. The culture medium was changed at 3 day in vitro (DIV). This culture was named neuron-enriched culture and contained 90% neurons and 10% astrocytes at 14 DIV. To obtain a pure neuronal culture, cytosine arabinoside (AraC) 10 μm was added after 1 day of culturing and for 2 days in order to stop the proliferation of the astrocytes. After 3 DIV, the medium was changed and remained the same therafter. At 14 DIV, the pure neuronal culture contained 99.5% neurons.
Primary cortical astroglial cultures were prepared by seeding cell suspensions from the cortices of newborn rats (P0). Cells were mechanically dissociated from five cortices in 10 mL culture medium (DMEM supplemented with 4 mm glutamine, 50 IU/mL penicillin, 50 μg/mL streptomycin and 10% heat-inactivated fetal bovine serum). Cells were suspended in culture medium at a density of one cortex for 36 mL of medium, plated in plastic multi-well plates (falcon, Dutscher) and incubated at 37°C in a humidified 5% CO2-95% air atmosphere. The medium was changed every 5 days until cells were confluent (around 14 DIV). They were then placed for 1 week in defined medium [DMEM supplemented with 4 mm glutamine, 50 IU/mL penicillin and 50 μg/mL streptomycin, 5 μg/mL insulin and 0.5 mg/mL fatty acid free bovine serum albumin (BSA)].
Treatment of neuronal cultures
Glial conditioned medium was the medium conditioned for 1 week by confluent astrocytes in chemically defined medium. It was stored at −20°C before utilization and added to neurons at 50% of the initial volume of the neuronal culture medium. In control conditions, the same volume of medium used to cultivate astrocytes but not exposed to these cells, was added to neurons.
Low cholesterol-glia conditioned medium (LC-GCM) was obtained, as previously reported by Mauch et al. (2001), by adding mevastatin (10 μm) to confluent astroglial cultures for 1 week. Before addition to neuronal cultures, residual mevastatin was removed by re-buffering LC-GCM to glial culture medium (NAP10 column, Amersham Pharmacia Biotech.).
Cholesterol (1, 5 or 10 μg/mL) from ethanol stock was added directly to neurons, with the same final concentration of ethanol (0.5%). We previously verified that ethanol 0.5% had no significant effect on glutamate uptake rate.
GCM, LC-GCM and cholesterol were added at 4 DIV and remained until the end of the culture.
TBOA (25 μm), alone or together with GCM, was added to neurons at 4 DIV and remained until the time of experiment.
Measurement of cholesterol content
Extraction of cholesterol from GCM was performed with chloroform : isopropanol (3 : 1) and, after centrifugation, clear supernatant was decanted. The solvent was evaporated under nitrogen and total cholesterol (free cholesterol and cholesteryl esters) concentration determined by enzymatic assay (cholesterol enzymatique from Onyx, Vieille-Brioude, France) using a cholesterol standard from Sigma.
Mitochondrial respiration was assessed by MTT colorimetric assay. MTT is converted to an insoluble blue formazan product by mitochondrial dehydrogenases in living cells but not in dying cells. Cultures were incubated with MTT (500 μg/mL) at 37°C for 3 h and then washed and incubated in HCl 0.08 N/isopropanol to dissolve the blue formazan product. Optical density (OD) read at 570 nm was subtracted from OD read at 690 nm. The mitochondrial activity of the cells was utilized as an index of cellular viability.
Glutamate uptake was measured as described previously (Lortet et al. 1999). After two washes of the cells with the uptake buffer (pH 7.4, 10 mm glucose, 5 mm KCl, 127 mm NaCl, 10 mm Hepes, 2.5 mm CaCl2, 1.2 mm MgSO4, 1.3 mm KH2PO4), uptake assays were started by adding l-[3H]glutamate at 1 μm final concentration (specific activity 15–25 Ci/mmol). For the kinetic analysis, glutamate concentrations ranging from 0.1 to 250 μm were used. Incubations were performed at 37°C for 2 min. The reaction was stopped by adding 1 mL of cold, sodium-free buffer, and followed by two washes with the same cold medium. To dissolve the cells, 1 N NaOH was added to the culture dishes and the radioactivity was assessed by liquid scintillation counting. To determine the part of radioactivity not due to sodium-dependent glutamate transport, l-[3H]glutamate uptake was assessed in physiological medium in which sodium was omitted and replaced by choline. Protein content was determined by the method developed by Lowry et al. (1951) with bovine serum albumin as the standard. The values of l-[3H]glutamate incorporation were divided by protein content. We previously established that the rate of glutamate uptake is linear with time of incubation up to 5 min and quantity of protein used in the assay.
Measurement of extracellular glutamate
The extracellular glutamate concentration was measured by HPLC with a fluorometric detection. An aliquot of culture supernatant was deproteinisated by 0.5 N perchloric acid and centrifuged at 13 000 g for 10 min at 4°C. The supernatant was analyzed after pre-column derivatization with o-phtaldialdehyde. Each sample was diluted (1/20) in a 0.15 m sodium borate buffer and directly injected into a C-18 column (Spherisorb 5 μm ODS2 150 × 3 mm). The mobile phase consisted of 0.1 m potassium acetate (pH 5.8) in methanol (v/v; 1/7). Elution was performed with a methanol gradient ranging from 12.5 to 56%. Glutamate concentration in the culture medium was calculated with the Millenium program from Waters (St Quentin en Yvelines, France).
Cells were grown on double-coated coverslips and fixed in 4% paraformaldehyde. They were then pre-incubated for 30 min in lysine (18.3 mg/mL) and for 30 min in blocking buffer (6% BSA) containing 0.1% Triton X100. The incubation with the primary antibodies: mouse anti-microtubule associated protein type 2 (MAP2, 1/500) and rabbit anti-glial fibrillary acidic protein (GFAP, 1/500) was performed overnight at 4°C. Rhodamine-conjugated goat anti-mouse and fluoroscein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG antibodies were used to detect primary antibodies (1/250, 1 h). Finally, the slides were mounted in Fluorsave before examination with a Leica microscope (Leica Microsystems). The number of cells was estimated by counting MAP2+-cells (neurons) and GFAP+-cells (astrocytes) on three distinct fields (× 40 magnification) per culture.
Western blot analysis
After one wash in cold PBS, primary cortical cells were harvested in a lysis buffer (50 mm Tris HCl, pH 7.5; 1 mm EDTA; 1% triton X100) with protease inhibitors and centrifuged for 5 min at 18 000 g at 4°C. Supernatants were aliquoted and stored at −80°C until used for western analysis. Protein samples were diluted in loading buffer 5 ×[Tris HCl 0.3M pH 6.8; 4% sodium dodecyl sulfate (SDS); 50% glycerol; dithiothreitol 0.5 m], loaded (80 μg per lane for EAAC1 and 40 μg for GLAST and GLT1) on a 10% SDS polyacrylamide gel and then transferred to an Hybond ECL nitrocellulose membrane with a semi-dry apparatus (Bio-Rad, Munich, Germany). After the transfer, membranes were immersed for 1 hour in blocking buffer [5%-non-fat dry milk; 0.1% Tween 20; 50 mm Tris-buffered saline (TBS)] and then probed overnight at 4°C with the primary rabbit antibodies diluted in 5% milk TBS. The primary antibodies used were the anti-actin antibody (1/300), the affinity purified anti-EAAC1 antibody (1/250; 0.5 mg/mL), the affinity purified anti-GLAST antibody (GLAST N: 1/4000; 0.35 mg/mL) and affinity purified anti-GLT1 antibody (GLT1 C: 1/5000; 1.5 mg/mL). The two glial glutamate transporter antibodies were kindly provided by G. Pietrini (Perego et al. 2000). After washing, the blots were incubated for 1 h at room temperature (20°C) with horseradish peroxydase-conjugated goat anti-rabbit IgG diluted 1 : 10000 in 5% milk TBS. After the last three additional washes in TBS, the peroxydase signal was detected with an ECL Kit from Pierce (Montluçon, France). The densitometric analysis was performed with Scion Image software on digital camera-acquired images of blots. Values from densitometric analysis of glutamate transporter proteins were normalized to those of actin and were expressed as a percentage of the value found in the untreated neuron-enriched culture.
Data and statistics
Each experiment was performed in duplicate on at least three independent cultures. Statistical analysis was performed by repeated measures anova with the Fisher's test for multiple group comparisons. p < 0.05 was considered to denote significance.
Characterization of the neuronal cultures
In these experiments, we studied two cortical neuronal cultures: a neuronal culture with few astrocytes (neuron-enriched culture) and a pure neuronal culture. Immunostaining at 14 DIV (Fig. 1) with anti-MAP2 antibody (red) and anti-GFAP antibody (green) shows that some astrocytes were present in the neuron-enriched culture and their number increased after GCM treatment. In the pure neuronal culture were no astrocytes, even under GCM treatment. Neurons had extended dendritic processes and had morphologic features similar to those of neurons cultivated with astrocytes. We performed cell counting at 14 DIV. The neuron-enriched culture had an astrocyte ratio ranging from 9.0 ± 1.6% in non-treated to 18.0 ± 1.5% after GCM treatment. In the pure neuronal culture, the percentage of astrocytes was always less than 0.5% (0.5 ± 0.2 and 0.2 ± 0.1 after GCM treatment) confirming the neuronal purity of this culture.
Figure 2 shows the evolution of MTT metabolism in the neuronal cultures, treated or not with GCM. During the first week in culture, no significant difference in MTT metabolism was found between the two neuronal cultures. MTT metabolism increased, reflecting the cell growth. Neurons from both cultures showed a similar pattern of development in the presence or absence of astrocytes. During the second week, MTT values of the neuron-enriched culture remained stable, GCM increasing MTT metabolism at 14 DIV (+23%). In the pure neuronal culture, MTT values decreased during the second week in vitro (−52% at 14 DIV as compared with 7 DIV). GCM treatment increased MTT values of the pure neuronal culture during the second week of culturing. Nevertheless, MTT values remained significantly lower than those of the neuron-enriched culture of the same age. Cell counting, that really reflects cell viability as compared with MTT assay that reflects both viability and growth, was preformed at 14 DIV and confirmed MTT assay results: cell number was significantly reduced in the pure neuronal culture (−61%) and this effect was attenuated after GCM treatment (−28%).
Extracellular glutamate concentration is dependent on cultures and treatment
DMEM-F12 culture medium (Fig. 3) contained 50 μm glutamate which was diluted at 4 DIV (2 : 1, 30 μm) with glutamate-free glial culture medium (conditioned or not by glia depending on the treatment). Glutamate was rapidly eliminated by neuron-enriched culture ([Glu]ec < 5 μm at 4 DIV) and remained at a value close to 5 μm up to the end of the culture. The same evolution of glutamate concentration was observed in the neuron-enriched culture treated by GCM. By contrast, in the pure neuronal culture, glutamate concentration remained at a value close to 30 μm. This high extracellular concentration might explain the decrease in cell viability observed during the second week in vitro (Fig. 2). We have therefore tested whether glutamate receptor blockade with MK801 (1 μm), a-non-competitive NMDA receptor antagonist, improve neuronal survival and have found that, indeed, MTT value was doubled in the pure neuronal culture treated by MK801 (OD: 0.223 ± 0.027 vs. 0.452 ± 0.040 after MK801 treatment).
Treatment of this pure neuronal culture with GCM caused a significant decline in extracellular glutamate concentration within 3 days. After 72 h of GCM treatment, glutamate concentration reached a value close to that found in the neuron-enriched culture (around 5 μm). However, when the glutamate uptake inhibitor TBOA (25 μm) was added together with GCM, glutamate concentration remained at a value not different from that of the untreated-pure neuronal culture (Table 1). We therefore propose that the effect of GCM on glutamate concentration is due to an increase in glutamate transport activity.
Table 1. Effect of TBOA (25 μ m ) on extracellular glutamate concentration in the pure neuronal culture, at 7 DIV
TBOA 25 μm
Cells were treated at 4 DIV with TBOA (25 μm) alone or together with GCM. We first verified that TBOA (25 μm) blocked glutamate uptake (−70% at 7 DIV, data not shown). Extracellular glutamate concentration (μM) was measured by HPLC at 7 DIV. Results are means ± SEM of two independent experiments performed in triplicate each. ap < 0.05 versus untreated pure neuronal culture.
Glutamate uptake rate was increased by glial soluble factors
Figure 4 shows the changes in 1 μ m l -[ 3 H]glutamate uptake rate during the development of the neuronal cultures. Glutamate uptake was nearly stable in the neuron-enriched culture during the first 2 weeks in vitro (37.7 pmol/mg/min at 14 DIV). Treatment of these cells with GCM at 4 DIV increased the glutamate transport velocity with a delay of 72 h. Glutamate uptake rate remained significantly elevated during the 2 weeks in culture (+25% at 14 DIV). After short-term GCM incubation (1, 6 and 12 h) glutamate uptake rate was not increased (data not shown). On the contrary, glutamate uptake rate was lower in the pure neuronal culture as compared with the neuron-enriched culture (−77% at 14 DIV). GCM treatment of the pure neuronal culture increased glutamate uptake rate (+142% at 14 DIV) that remained significantly lower than the values found in the neuronal enriched culture. Similarly with the neuron-enriched culture, GCM had no effect after 1, 6, 12 h (data not shown) and 24 h of treatment. Kinetic analysis was performed at 7 DIV ( Fig. 5 ) . Mean Vmax and Km values were calculated from the means of Vmax and Km found in each independent experiment. In the neuron-enriched culture, Vmax = 573.4 ± 106.6 pmol/mg/min and Km = 9.8 ± 0.9 μ m ; after GCM treatment: Vmax = 1031 ± 227 pmol/mg/min and Km = 15.2 ± 3.0 μ m . In the pure neuronal culture, the kinetic parameters were: Vmax = 127.2 ± 25.1 pmol/mg/min and Km = 10.6 ± 1.4 μ m ; after GCM treatment: Vmax =196.6 ± 19.0 pmol/mg/min and Km = 11.5 ± 2.7 μ m . These results revealed that GCM treatment increased significanty the apparent Vmax of glutamate transport and did not significantly affect the apparent Km in both neuronal cultures.
Glial soluble factors increased EAAC1 expression
We measured the level of expression of the three glutamate transporters, EAAC1, GLAST and GLT1 by western blotting in the neuronal cultures, treated or not with GCM (Fig. 6). These measurements were performed at both 7 and 14 DIV. Value from densitometric analysis was normalized to that of actin, and was expressed as a percentage of the value found in the neuron-enriched culture.
In the neuron-enriched culture, cells expressed EAAC1 (70 kDa), GLAST (64 kDa) and GLT1 (69 kDa). GCM treatment significantly increased the expression of EAAC1 (+20 and +30% at 7 and 14 DIV, respectively). More surprisingly, GCM treatment also increased GLT1 expression in the neuron-enriched culture (+32% at 7 DIV and +45% at 14 DIV) and had no effect on GLAST expression.
In the pure neuronal culture, we detected almost exclusively EAAC1 expression. As compared with the neuron-enriched culture, EAAC1 expression was low (−24% at 7 DIV and −28% at 14 DIV). Upon exposure to GCM, EAAC1 expression was markedly increased, reaching significantly higher levels than those of the non-treated cultures (+55% at 7 DIV and +90% at 14 DIV vs. untreated pure neuronal culture). GLAST expression was never detected in the pure neuronal culture, treated or not by GCM. GLT1 expression was barely detectable at 7 DIV and detected no more at 14 DIV. GLT1 was not induced by GCM treatment in this pure neuronal culture.
Cholesterol, present in GCM, increased glutamate uptake rate without affecting EAAC1 expression
We measured the cholesterol content in GCM and found a mean value of 4.9 ± 0.9 μg/mL. We then tested the effect of cholesterol (1, 5 or 10 μg/mL) added to the neuronal culture from day 4 on glutamate uptake rate. Figure 7 shows that cholesterol improves in a dose-dependent manner the velocity of glutamate transport in the neuron-enriched culture. Glutamate transport was significantly increased after 10 μg/mL cholesterol treatment at 7 DIV (+41%) and 14 DIV (+29%). In the pure neuronal culture, the velocity of glutamate uptake was increased after 5 and 10 μg/mL cholesterol treatment at 7 DIV (+41 and +37%, respectively) and after 10 μg/mL cholesterol treatment at 14 DIV (+88%). In both neuronal cultures, cholesterol (10 μg/mL) added together with GCM at 4 DIV had no additive effect on glutamate uptake rate as compared with GCM alone (Figs 8a–b). Low cholesterol-GCM was obtained by treatment of astrocyte culture with mevastatin, an inhibitor of cholesterol synthesis as previously described by Mauch et al. (2001). Glutamate uptake rate was significantly decreased when cholesterol was eliminated from GCM (LC-GCM) as compared with GCM (Figs 8a–b). The addition of cholesterol (10 μg/mL) to LC-GCM increased glutamate uptake rate, reaching a value not significantly different from GCM. The extracellular glutamate concentration, low in the neuron-enriched culture, was not changed by the different treatments (Fig. 8c). In contrast, in the pure neuronal culture, cholesterol reduced glutamate concentration from 25 μm (untreated) to 12 μm. This decrease is less important than with GCM (5 μm). The removal of cholesterol from GCM (LC-GCM) altered the GCM-induced decrease in glutamate concentration that, nevertheless, remained significantly reduced as compared with untreated culture (13 μm). Furthermore, when cholesterol was added to LC-GCM, glutamate concentration reached a value not different from GCM. Altogether, these results demonstrate that cholesterol, present in GCM, enhances the neuronal ability to take up glutamate.
In order to test whether or not the increase in glutamate uptake rate by cholesterol was due to an increase in glutamate transporter expression, we tested the effect of 10 μg/mL cholesterol on glutamate transporter expression at 14 DIV. Results in Fig. 9 show that the expression of the glutamate transporters, EAAC1, GLT1 and GLAST were not significantly modified by cholesterol treatment in both types of neuronal cultures.
In this study, we have tested the effect of astrocytes on the expression and activity of the neuronal glutamate transporter EAAC1 in embryonic cortical neurons. For this purpose, neurons were cultivated in the presence (neuron-enriched culture) or in the absence (pure neuronal culture) of astrocytes, and were supplemented with medium that had been conditioned by astrocytes.
The major finding of the present study is that, during the development of neurons, factors released by astrocytes can modulate the expression of EAAC1, concomitantly with an enhancement of glutamate uptake rate by neurons. We have further shown that cholesterol is one of the factors implicated in glutamate transport regulation by GCM. These results provide further arguments for neuron–astrocyte interactions, involving soluble factors, during CNS development, and new evidences for regulatory mechanisms of glutamate transport.
Glutamate transporter expression and activity was dependent on glial cells
In the neuron-enriched culture (around 10% astrocytes), the three glutamate transporters, EAAC1, GLAST and GLT1 are expressed. We reported in a previous paper (Guillet et al. 2002) that EAAC1 was expressed in neurons and GLAST in astrocytes during the whole time in culture. GLT1 was expressed by astrocytes from the second week in vitro. Cells had a great capacity to incorporate extracellular glutamate, as shown by the rapid elimination of extracellular glutamate present in the medium.
In the pure neuronal culture, the only glutamate transporter detected was EAAC1. GLAST has never been observed. Recent papers (Chen et al. 2002; Schmitt et al. 2002) described that neurons in culture can express GLT1 and/or a variant form of GLT1 (GLT1v). In the present experiments, we used an antibody that recognizes GLT1 and not its variant (Perego et al. 2000). GLT1 was very slightly expressed and detected only at 7 DIV. We obtained the same data with an antibody directed against a region shared by GLT1 and GLT1v (unpublished observation), kindly provided by N. C. Danbolt (anti-GLT1 B12, peptide 12–26; Lehre et al. 1995). Therefore, in these experiments we have no evidence for a neuronal expression of GLT1/GLT1v.
Thus, in the pure neuronal culture, neurons are able to express a basal level of EAAC1 without the need of astrocytes. However, the expression of EAAC1 measured at 7 and 14 DIV was reduced as compared with the neuronal enriched culture (−28% at 14 DIV) suggesting that astrocytes might up-regulate the expression of EAAC1. In the pure neuronal culture, glutamate uptake rate was low. The Vmax of glutamate transport in this pure culture represented only 22% the Vmax recorded in the neuron-enriched culture. This might be due both to the down regulation of EAAC1 and to the absence of the glial transporters. It is important to notice that astroglial cells have a great capacity to transport glutamate [for a comparison, the rate of 1 μm glutamate uptake was 256 ± 36 pmol/mg prot/min (n = 3) in a cortical astrocyte culture as compared with 40 pmol/mg prot/min in the neuron-enriched culture and 10 pmol/mg prot/min in the pure neuronal culture]. Probably as a consequence of the decrease in glutamate transport, neurons failed to clear glutamate present in the medium that remained at a concentration of around 30 μm. It is noticeable that aspartate, present in DMEM-F12 at the same concentration as glutamate (50 μm), transported by the same transporter as glutamate but not released by neurons during synaptic activity, showed the same evolution as glutamate in the pure neuronal culture (result not shown). Therefore, the high glutamate concentration in this culture is more likely due to the deficiency of glutamate transport and not to glutamate release. According to Durkin et al. (1997), such a glutamate concentration (30 μm) might not alter cell viability during the first week in vitro but became cytotoxic after 7 DIV. Indeed, we showed that cell viability was altered during the second week in culture and that the NMDA glutamate receptor antagonist MK801 partly protected neurons during that period of time. Therefore, the decrease in cell viability observed during the second week in culture in the pure neuronal culture is due to the lack of glutamate elimination.
Glial soluble factors increased glutamate transport activity and EAAC1 expression
GCM treatment of the neuron-enriched culture induced an increase in glutamate uptake rate, glutamate transport Vmax and EAAC1 and GLT1 expressions. In the case of EAAC1, our result suggests that glial factors that might be released by astrocytes are able to up-regulate the expression of the neuronal glutamate transporter. To our knowledge, this is the first evidence for such a regulation. As regards GLT1, the increase in expression observed in our experiments is in line with previous reports showing that neurons influence GLT1 expression (Gegelashvili et al. 1997; Swanson et al. 1997; Schlag et al. 1998). Perego et al. (2000) reported that the level of GLT1 expression in astrocytes depends on differentiation/maturation stage and activity of the neurons and Ullian et al. (2001) have shown that soluble factors released by astrocytes increase the number of mature, functional synapses. Therefore, GCM treatment might increase neuronal maturity and/or activity that in turn increased GLT1 expression by astrocytes. As we cannot detect a neuronal expression of GLT1, the effect of GCM on GLT1 in the neuron-enriched culture was probably not because of an effect on the putative GLT1 expressed in neurons but, as previously stated, to an increase in the astrocytic expression of GLT1. GLAST expression was not changed by GCM treatment. Altogether, these experiments suggest that GCM treatment of the neuron-enriched culture led to an increase in glutamate transport rate that might be explained by enhanced EAAC1 and GLT1 expressions.
GCM treatment of the pure neuronal culture led to an increase in neuronal glutamate uptake rate by 93% that occurred not immediately, but with a delay of more than 24 h of GCM treatment. Concommitantly, glutamate concentration was reduced to control value (< 5 μm), suggesting that the increase in glutamate transport rate allowed neurons to eliminate glutamate present in the medium. The relationship between glutamate transport activity and the extracellular glutamate steady state level is further demonstrated by experiments with the glutamate uptake inhibitor TBOA that abolished the decline in glutamate concentration observed after GCM treatment. Thus, GCM, by increasing glutamate uptake rate, allowed neurons grown without astrocytes to clear glutamate from the extracellular space.
Furthermore, GLT1 was not induced by GCM and we therefore conclude that GCM acted specifically on the regulation of EAAC1.
The increase in glutamate transport activity by GCM can reflect the regulation of transporter kinetics, surface expression, subcellular localization and clustering and/or protein expression. We have shown previously (Lortet et al. 1999) that glutamate transport activity in neuronal cultures is rapidly, within 30 min, regulated by protein kinases A and C. Similarly, Davis et al. (1998) reported the mobilisation of EAAC1 from intracellular storage elements to the cell membrane by a 30-min PKC-mediated activation. The fact that it took more than 1 day to enhance glutamate transport activity after GCM treatment rules out the possibility of a fast modulatory effect. In accordance with a more delayed phenomenon, we show here that GCM treatment up-regulated EAAC1 expression (+90% vs. untreated pure neuronal culture) and also induced an increase in glutamate transport Vmax with no change in affinity of the transporter for glutamate. We can therefore conclude that GCM increased the total number of glutamate transporter proteins and the number of glutamate transport sites at the cell surface.
Altogether, these experiments confirmed the involvement of glial soluble factors in the regulation of the activity of neuronal glutamate transport and EAAC1 expression.
Cholesterol present in GCM increased glutamate transport rate
Evidently, the next step of our study would be the identification of the glial factor(s) involved in this regulatory process. Conversely, the identification of neuronal factor(s) regulating the expression of glial glutamate transporters is still under investigation and, to date, several compounds have been implicated: glutamate (Poitry-Yamate et al. 2002), PACAP (Figiel and Engele 2000). Among all the possible candidates for the glial factor(s), we decided to focus on cholesterol because we showed that cholesterol was contained in GCM and, interestingly, new reports demonstrated the implication of astrocytes-derived cholesterol in synapse formation (for a review see Pfrieger 2003). Pfrieger and Barres (1997) showed that retinal ganglion cells cultured in the presence of astrocyte-conditioned medium have 10 times higher levels of synaptic activity, and Ullian et al. (2001) demonstrated that astrocytes increased the number of mature, functional synapses sevenfold. Finally, Mauch et al. (2001) identified the glial factor responsible for synapse development as cholesterol and they showed that the effect of cholesterol on synaptogenesis was not immediate but with a delay of 24 h. It is noteworthy that GCM in our experiments has a similar delayed effect. Therefore, a specific involvement of cholesterol in EAAC1 regulation might be suggested as EAAC1 expression is regulated by glial factors and because this neuronal glutamate transporter is localized close to the synaptic sites (Shashidharan et al. 1997; Conti et al. 1998), possibly contributing to the functioning of glutamatergic synapses.
To test the possible involvement of cholesterol in GCM effect on glutamate transport regulation, we first confirmed the cholesterol content of GCM of around 5 μg/mL, close to that previously reported by Mauch et al. (2001). We then tested the effect of cholesterol added to neuronal cultures (with and without glia) and found that cholesterol increased glutamate transport (with a minor effect than GCM), but with no effect on glutamate transporter expression. The effect of GCM and cholesterol were not additive, suggesting that they share the same pathway. This is further confirmed by the fact that low cholesterol-GCM had a reduced effect on glutamate transport as compared with GCM, and glutamate transport was increased when cholesterol was added to LC-GCM. Cholesterol did not fully mimic GCM effect and is therefore probably one of the several factors present in GCM responsible for the regulation of glutamate uptake. It is possible that cholesterol modifies the membrane composition and/or the subcellular localization of EAAC1 and thereby its activity. In favour of the second hypothesis, there is now a lot of evidence that EAAC1 can be regulated by a redistribution of the protein from a subcellular compartment to the cell membrane (Davis et al. 1998). Such an effect of GCM and cholesterol on the localization of several presynaptic vesicular proteins has indeed been previously reported (Mauch et al. 2001 and Ullian et al. 2001). In line with this, Hering et al. (2003) presented evidences that lipid rafts, composed primarily of cholesterol and shingolipids, exist in dendrites of neurons where they are associated with a set of postsynaptic proteins. They are involved in localized signalling at the membrane and trafficking of membranes and proteins. Hering et al. demonstrated that depletion of cholesterol led to a loss of synapses and dendritic spines.
We therefore propose that EAAC1 might be associated with cholesterol-rich domains. Cholesterol might then enhance glutamate transport activity, either by concentrating EAAC1 and regulatory proteins in lipid rafts, thereby facilitating the formation of active complexes, or by regulating the trafficking of EAAC1 to the cell membrane. Indeed, preliminary experiments from Butchbach et al. (2003) reported the association of glutamate transporters, including EAAC1, in lipid rafts.
In conclusion, in the CNS, formation of a complete functional glutamatergic synapse includes the biosynthesis and recruiting of glutamate transporters, so cues for expression of glutamate transporter activity is of great interest. Using neuronal cultures with and without astrocytes, we are able to show that glial soluble factors and cholesterol increased the activity of glutamate transport. Furthermore, glial-derived signals up-regulated EAAC1 expression in neurons. It is highly possible that the effect of GCM on glutamate transporter activity and expression in neurons is multifactorial. We identified cholesterol, present in GCM, as a factor. Glia-derived cholesterol might feed the production of lipid rafts in neurons regulating thereby glutamate transport activity. Together, these data argue that signals from glial cells play a crucial role in glutamate transporter expression by neurons and indicate a new form of glutamate transporter regulation. These in vitro results might be relevant to the in vivo post-natal development of cortex. Indeed, Furuta et al. (1997) demonstrated that EAAC1 expression reached a maximum level during post-natal development of the cortex of the rat (around P5-P16), corresponding to the end of the period of astrocyte multiplication and the beginning of synaptogenesis. We also provide some more evidence for the regulation of the glial glutamate transporter GLT1 by neurons. Altogether, these experiments add new arguments for the cross regulation of glutamate transporters by neurons and astrocytes.
A neuroprotective or deleterious effect of glial cells on neurons has been suggested repeatedly, with important implications for neurological diseases. We provide here new evidence for a possible neuroprotecive effect of glia by increasing neuronal glutamate transporter expression, that might also be deleterious if this transporter functions in a reverse way. Such data also stress the importance of a better understanding of neuron–astrocyte interaction, not only in brain development but also possibly in neurodegenerative diseases.
The authors wish to thank Denise Samuel for HPLC measurements, Diane Re for glial cultures and Lydia Kerkerian Le-Goff for helpful discussions and critical reviews. Grazzia Pietrini is gratefully acknowledged for generously providing us with glial glutamate transporters antibodies. Niels Chr Danbolt is also gratefully acknowledged for his kind gift of the antibody anti-GLT1 (GLT1 B12). This work was supported by grant number 0034052004707501 from French DGA/DSP. Benoit Canolle is a fellow of DGA/DSP.