Glutamate activates GFAP gene promoter from cultured astrocytes through TGF-β1 pathways

Authors

  • Luciana Ferreira Romão,

    1. Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
    Search for more papers by this author
  • Vivian de Oliveira Sousa,

    1. Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
    Search for more papers by this author
  • Vivaldo Moura Neto,

    1. Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
    Search for more papers by this author
  • Flávia Carvalho Alcantara Gomes

    1. Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil
    Search for more papers by this author

Address correspondence and reprint requests to Flávia Carvalho Alcantara Gomes, Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro – Centro de Ciências da Saúde, Bloco F, Ilha do Fundão – 21941-590 – Rio de Janeiro, RJ, Brazil. E-mail: fgomes@anato.ufrj.br

Abstract

Glial cells are currently viewed as active partners of neurons in synapse formation. The close proximity of astrocytes to the synaptic cleft suggests that these cells might be potential targets for neuronal-released molecules although this issue has been less addressed. Here, we evaluated the role of the excitatory neurotransmitter, glutamate, in astrocyte differentiation. We recently demonstrated that cortical neurons activate the gene promoter of the astrocyte maturation marker, GFAP (glial fibrillary acidic protein) of cerebral cortex astrocytes by inducing TGF-β1 (transforming growth factor beta 1) secretion in vitro. To access the effect of glutamate on GFAP gene, we used transgenic mice bearing the β-Galactosidase (β-Gal) reporter gene under the regulation of the GFAP gene promoter. We report that 100 μM glutamate activates the GFAP gene promoter of astrocytes from cerebral cortex revealed by a significant increase in the number of β-Gal positive astrocytes. Neutralizing antibodies against TGF-β completely prevented glutamate and neuronal-induction of GFAP gene, thus indicating that this event is mediated by TGF-β1. Further, induction of GFAP gene in response to glutamate was followed by nuclear translocation of the Smad transcription factor, a hallmark of TGF-β1 pathway activation. The antagonist of the metabotropic glutamate receptor, MCPG, inhibited neuronal effect suggesting that neuronal activation of GFAP gene promoter involves glutamate metabotropic receptors. MAPK (PD98059) and PI3K (LY294002) inhibitors fully prevented activation of GFAP gene promoter by all treatments. Surprisingly, these inhibitors also abrogated TGF-β1 direct action on GFAP gene although they did not inhibit Smad-2 phosphorylation, suggesting that TGF-β1-induced GFAP gene activation might involve cooperation between the canonical and non-canonical TGF-β pathways. Together, our results suggest that glial metabotropic glutamate 2/3 receptor activation by neurons induces TGF-β1 secretion, leading to GFAP gene activation and astrocyte differentiation and involves Smad and MAPK/PI3K pathways. Our work provides evidence that astrocytes surrounding synapses are target of neuronal activity and might shed light into the role of glial cells into neurological disorders associated with glutamate neurotoxicity.

Abbreviations used:
β-Gal

β-Galactosidase

CM

conditioned medium

DMEM

Dulbecco’s modified Eagle’s medium

GFAP

glial fibrillary acidic protein

GLAST

glutamate/aspartate transporter

MAPK

mitogen-activated protein kinase

MCPG

(S)-methyl-4-carboxyphenylglycine

mGlu

metabotropic glutamate

PBS

phosphate-buffered saline

PI3K

phosphatidylinositol-3-kinase

RG

radial glia

TGF

transforming growth factor

Glial cells have traditionally been considered supportive, satellite cells of the nervous system. Works over the past decade however have revealed that neuron-glia interactions play key roles in several events of brain development, such as neuronal proliferation and differentiation (Lim and Alvarez-Buylla 1999; Lie et al. 2005), neuronal migration (Hatten 2002); axonal guidance (Garcia-Abreu et al. 1995; Martinez and Gomes 2002, 2005); synapse formation (Christopherson et al. 2005; Stevens et al. 2007) and glial maturation (Gomes et al. 1999a,b; De Sampaio e Spohr et al. 2002; Schmid et al. 2003; Sousa et al. 2004; Barnabé-Heider et al. 2005; Stipursky and Gomes 2007).

Recently, astrocytes were shown to be active partners of pre- and post-synaptic terminals during synapse establishment. Astrocytes derived signals induce the formation of new synapses and modulate their number and strength in several systems in rodents and humans (Pfrieger and Barres 1997; Mauch et al. 2001; Christopherson et al. 2005; Johnson et al. 2007; Stevens et al. 2007). The intimate relationship between astrocytes and synaptic terminals in vivo (Ventura and Harris 1999) suggests that besides affecting synapse formation, astrocytes are also potential targets for neuronal-derived molecules such as neurotransmitters. In the present work, we analyze the role of glutamate, the major CNS excitatory neurotransmitter, in astrocyte differentiation.

Astrocytes express functional ionotropic and metabotropic glutamate receptors (mGluRs), and regulate glutamate uptake and inactivation. The mGluRs are classified into three groups on the basis of their sequence homology, pharmacological profile and transduction pathways: Group-I includes metabotropic glutamate 1 and 5 (mGlu1 and 5) receptors, which are coupled to polyphosphoinositide hydrolysis; Group-II (mGlu2 and 3) and Group-III (mGlu4, 6, 7, and 8) receptors are negatively coupled to adenylyl cyclase activity in heterologous expression systems (Pin and Duvoisin 1995). Activation of group II metabotropic glutamate receptor in astrocytes is associated with a neuroprotective action provided by synthesis and secretion of TGF-β1 (Bruno et al. 1998; Corti et al. 2007).

The TGF-β superfamily is constituted by multifunctional polypeptide members, which perform critical functions in tissue repair and development (Shi and Massagué 2003). TGF-β1 is involved in several CNS pathologies, where it has been implicated in organization of the glial scar in response to injury and in several neurodegenerative diseases (Zhu et al. 2002; Vivien and Ali 2006). Emerging evidences have suggested a key role for this factor in several CNS developmental processes such as cell adhesion, neuronal migration and differentiation, synaptogenesis and glial differentiation (Bottner et al. 2000; De Sampaio e Spohr et al. 2002; Brionne et al. 2003; Miller 2003; Packard et al. 2003; Gomes et al. 2005; Massagué and Gomis 2006; Stipursky and Gomes 2007; Siegenthaler and Miller 2008).

By using transgenic mice bearing 2 kbp of the 5′ flanking region of the GFAP gene linked to the β-Galactosidase (β-Gal) reporter gene, we previously demonstrated that cortical neurons activated the GFAP gene promoter and induced astrocyte differentiation in vitro (Gomes et al. 1999b). This event was mediated by synthesis and secretion of TGF-β1 by astrocytes in response to factors secreted by neurons (De Sampaio e Spohr et al. 2002). The identity of these neuronal-derived molecules remained unidentified. In the present paper, we investigated the role of glutamate in GFAP gene promoter activation and astrocyte differentiation in vitro. We now report that activation of glial group-II mGluRs enhances the synthesis of TGF-β1 and induces GFAP gene promoter through the activation of three signaling pathways: the mitogen-activated protein (MAP) kinase, the phosphatidylinositol (PI)-3-K and the Smad pathways.

Materials and methods

Astrocyte primary cultures and cocultures

Astrocyte primary cultures were prepared from transgenic mice bearing part of the 5′ flanking region of the murine GFAP gene linked to the Escherichia coli β-Galactosidase (β-Gal) reporter gene (lacZ) as previously described (Sousa et al. 2004). Cultures were prepared from cerebral cortex (Cc), midbrain (M) and cerebellum (Cb) derived from newborn transgenic mice. All animals were kept under standard laboratory conditions according to NIH guidelines. Briefly, mice were anaesthetized by hypothermia, decapitated, brain structures were removed and the meninges were carefully stripped off. Dissociated cells were plated onto 15.5 mm diameter wells (24-well plates; Corning Inc., NY, USA), previously coated with polyornithine (1.5 μg/mL, molecular weight 41 000; Sigma Chemical Co., St Louis, MO, USA), in Dulbecco’s modified Eagle’s medium/F12 (DMEM-F12) medium supplemented with 10% fetal calf serum (Invitrogen, Carlsbad, CA, USA). For immunocytochemistry assays, cells were plated on polyornithine-treated glass coverslips. The cultures were incubated at 37°C in 5% CO2 incubator for 10 days until reaching confluence. Glial monolayers were then incubated for an additional day in serum-free medium and used as substrate for coculture assays. Neurons (2 × 106 cells/well) freshly dissociated from 14 embryonic days old (E14) Swiss mice were plated onto the transgenic glial monolayer carpets and cocultures kept for 24 h according to Gomes et al. 1999b.

Synaptosomal preparations

Synaptosomes were prepared on a discontinuous Percoll gradient according to the method of Dunkley et al. (1988). In brief, cerebral structures pooled from four mice were weighed and homogenized in 0.32 M sucrose, 1 mM EDTA, and 0.25 mM dithiotreitol, pH 7.4 (SED), using a glass homogenizer. The preparation was centrifuged at 1000 g for 10 min, the pellet was discarded, and the supernatant was gently transferred to a four-step gradient of 3%, 7%, 15%, and 23% Percoll in SED solution. Tubes were centrifuged at 30 000 g for 13 min and the synaptosome fractions were collected from the interface of the 15% and 23% Percoll steps. The fraction was washed twice by centrifugation at 30 000 g for 18 min and resuspended in DMEM-F12. All synaptosomes were used on the same day of preparation. Synaptosomes conditioned medium was prepared in DMEM-F12 serum-free for 16 h at 37°C in a 5% CO2 incubator. The medium was then recovered and centrifuged at 1500 g for 5 min to eliminate eventual debris and used immediately.

Detection of β-Gal activity and quantitative analysis of β-Gal positive astrocytes

Detection of β-Gal activity was performed according to Gomes et al. (1999a). Briefly, glial cultures and cocultures were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) and stained for β-Gal with 0.4 mg/mL of 5-bromo-4-chloro-3-indolyl-β-D-galactoside (X-Gal, U.S. Biochemical Corp., Cleveland, OH) as substrate in 4 mM potassium ferricyanide, 4 mM potassium ferrocyanide, 2 mM MgCl2 and 0.001% Tween 20. Staining was allowed to occur for 16–20 h at 37°C. Development of the β-Gal reaction was stopped after several washes with PBS. After β-Gal detection, transgenic cultures were analyzed for β-Gal-positive astrocytes under a Nikon Eclipse TE300 microscope. Most of the cells considered positive for β-Gal reaction presented a dark blue staining of the nucleus, however, all of blue-nuclei cells were quantified independently of the intensity of the labeling, which ranged from slight to dark blue. At least five fields were counted per well and approximately 5 × 103 cells were counted per experiment. The experiments were done in triplicate and each result represents the mean of three independent experiments. Statistical analyses were performed by anova.

Conditioned medium preparation

Neuronal conditioned medium (CM) was prepared as previously described (Gomes et al. 1999b; De Sampaio e Spohr et al. 2002). Briefly, neurons derived from E14 mice were cultured in DMEM-F12 serum-free for 24 h at 37°C in a 5% CO2 incubator. The culture medium was then recovered, centrifuged at 1500 g for 10 min to eliminate eventual cellular debris and used immediately or stored in aliquots at −20°C for further use. Adherent cells on the coverslips were fixed with 4% paraformaldehyde and immunoreacted with antibody to the neuronal marker β-tubulin III (mouse anti-human β-tubulin III antibody; Promega Corporation; 1 : 500). Approximately 95% of the cells stained with the antibody, attesting their neuronal phenotype. No GFAP-positive cells were found under these conditions. CM from cocultures was prepared by culturing E14 neurons with newborn astrocytes for 24 h. After recovering of coculture CM, the medium was centrifuged and processed as described above for neuronal CM. Use of coculture or neuronal CM yielded similar results.

Treatment of astrocyte monolayers with conditioned medium, TGF-β1 and glutamate

Astrocyte monolayers derived from newborn transgenic mice cortices were prepared as previously described. After reaching confluence, cultures were incubated for an additional day with serum-free medium and then incubated in the following conditions: (a) in the presence of TGF-β1 (10 ng/mL) (R&D Systems, Buckinghamshire, UK), or (b) glutamate (Merck) (10 μM, 100 μM and 1 mM), or (c) neuronal conditioned medium. Cultures were kept for an additional 24 h at 37°C in a 5% CO2 incubator and then stained with X-Gal as described.

Intracellular cascades inhibition and antibody blocking assays

Astrocyte monolayers were concomitantly incubated in the presence of TGF-β1 (10 ng/mL), glutamate (100 μM), neurons, or CM and specific signaling pathway inhibitors for 24 h, accordingly to the previously described protocol (De Sampaio e Spohr et al. 2002). The following inhibitors were used: PD98059, MAPK-specific inhibitor (50 μM); LY294002, PI3K-specific inhibitor (5 μM); (S)-Methyl-4-carboxyphenylglycine, group I/II mGluRs antagonist (MCPG-500 μM). Inhibitors concentration was used accordingly to Martinez and Gomes 2002 (PD98059, LY294002) and Strasser et al. 1998 (MCPG). All inhibitors were purchased from Calbiochem (La Jolla, CA, USA) and diluted in dimethylsulfoxide (DMSO, Sigma). For neutralization assays, astrocyte monolayers were cultured simultaneously in the presence of glutamate (100 μM) or neurons and 10 μg/mL of a neutralizing antibody against transforming growth factor-β1 (TGF-β1) (chicken anti-human TGF-β1 antibody; R&D Systems) or a control unrelated immunoglobulin.

Immunocytochemistry

Immunocytochemistry was performed as previously described (Sousa et al. 2004). Briefly, cells were fixed with 4% paraformaldehyde for 30 min and permeabilized with 0.2% Triton X-100 for 5 min at 25°C. After permeabilization, cells were blocked with 10% normal goat serum (Vector Laboratories, Inc., Burlingame, CA, USA) in PBS (blocking solution) for 1 h and incubated overnight at 4°C with the anti-rabbit GFAP antiserum (Dako, Carpinteria, CA, USA; 1/500) diluted in blocking solution. Previous to the primary antibody incubation, endogenous peroxidase activity was abolished with 3% H202 for 10 min, followed by washing with PBS. After primary antibody incubation, cells were extensively washed with PBS/10% normal goat serum and incubated with secondary antibody conjugated with horseradish peroxidase (goat anti-rabbit; Sigma Chemical, 1/200) for 1 h at 25°C. Peroxidase activity was revealed with 3,3′-diaminobenzidine (DAB peroxidase substrate kit; Vector Laboratories, Inc.). Analysis of Smad nuclear translocation was performed by immunofluorescence as previously described by Stipursky and Gomes (2007). Primary antibody was mouse anti-Smad 4 (Santa Cruz Biotechnology; Santa Cruz, CA, USA; 1 : 5) and secondary antibody was alexa fluor 594 goat anti-mouse IgG, (Molecular Probes, 1 : 300). Nuclei were counterstained with DAPI (4′,6-Diamidino-2-phenyindole, dilactate; Sigma Chemical). After immunostaining, cells were visualized using a TE 300 Nikon microscope. Negative controls were performed by omitting the primary antibody during staining. In all cases no reactivity was observed when the primary antibody was absent.

Western blotting analysis

Biochemical characterization of proteins was performed according to Stipursky and Gomes (2007). Astrocyte cultures were lysed in 2X loading buffer (100 mM Tris-Cl [pH6.8]; 4% of sodium dodecyl sulfate; 0.2% of bromophenol blue; 20% of glycerol; 200 mM of dithiotreitol) and then boiled for 5 min before loading in the gel. Approximately 20 μg of protein per lane were submitted to eletrophoresis in a 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis mini gel and electrically transferred onto a Hybond-P polyvinylidene difluoride membrane (Amersham Biosciences) for 1 h. Membranes were blocked in 5% PBS-milk and primary antibodies were added for 1 h at 25°C followed by peroxidase-conjugated secondary antibody incubation for 1 h at 25°C. Proteins were visualized using the enhancing chemiluminescence’s detection system (SuperSignal West Pico Chemiluminescent Substrate-Pierce). The following primary antibodies were used: rabbit anti-Phospho-Smad2 (Cell Signaling) and mouse anti-α-Tubulin (Sigma Chemical; 1 : 1000). The following secondary peroxidase-conjugated antibodies were used: goat anti-rabbit IgG and goat anti-mouse IgG (Amersham Biosciences; 1 : 3000).

Results

Glutamate activates GFAP gene promoter of transgenic astrocytes

We previously demonstrated that neurons activated GFAP gene promoter from cortical astrocytes by secreting soluble molecules (Gomes et al. 1999b). In order to evaluate if this effect is due to neurotransmitters from neurons we employed synaptosome assays, which have been considered a useful tool to study neurotransmitters (Dunkley et al. 1988). When cortical astrocytes derived from GFAP-lacZ mice were cultured in the presence of neurons, synaptosome preparations or synaptosomes conditioned medium there was an increase in the number of β-Gal positive astrocytes (Fig. 1a). Since synaptosome preparations are known to be mainly constituted by neurotransmitter vesicles we sought to analyze the involvement of glutamate, the main excitatory neurotransmitter secreted by central neurons.

Figure 1.

 Glutamate activates GFAP gene promoter of transgenic astrocytes. Cerebral cortex astrocytes from newborn transgenic mice were cultured for 24 h in the following conditions: alone (Control; a,b,c); in the presence of increasing concentrations of synaptosome (Syn 2 μg; Syn 20 μg; Syn 50 μg; a), embryonic cortical neurons (Neuron, a), conditioned medium derived from neurons (CM neuron, a) or from synaptosome (CM syn, a); in the presence of increasing concentrations of glutamate (Glu 10 μM; Glu 100 μM; Glu 1 mM; b,c) After β-Gal activity detection, the number of β-Gal positive cells was quantified (a,c). Each point represents the average of three independent experiments done in triplicate. In (a) and (c), all groups are statistically different from the control (p < 0.01). Scale bar: 100 μm.

Treatment of astrocyte monolayers with increasing concentrations of glutamate (10 μM, 100 μM and 1 mM) increased the number of β-Gal positive astrocytes by approximately 125–150% (Fig. 1b and c). No statistic difference was observed between the three different concentrations of glutamate. These results suggest that this neurotransmitter might be a positive modulator of GFAP gene and thus of astrocyte differentiation in the cerebral cortex.

Activation of GFAP gene by glutamate is mediated by TGF-β1

Activation of group II mGlu2/3 receptor induces synthesis and secretion of TGF-β1 in astrocyte (Bruno et al. 1998; Corti et al. 2007). We previously demonstrated that induction of the GFAP gene promoter by cortical neurons is mediated by TGF-β1 secreted by astrocytes in response to neuronal soluble factors. Here, we sought to analyze the role of TGF-β1 as mediator of glutamate effect on GFAP gene promoter.

We thus, cultured cortical astrocytes from newborn transgenic mice alone or with 100 μM of glutamate, cortical neurons or 10 ng/mL of TGF-β1, in the presence or absence of 10 μg/mL of TGF-β1 neutralizing antibody (anti-TGF-β1) or 500 μM of mGlu2/3 receptor antagonist (MCPG). After 24 h, β-Gal positive cells were analyzed. Glutamate, neurons and TGF-β1 greatly increased the number of β-Gal positive astrocytes (Fig. 2). Addition of neutralizing antibody against TGF-β1 fully prevented this increase (Fig. 2), whereas addition of a control unrelated immunoglobulin had no effect (data not shown). Likewise, addition of the glutamate receptor antagonist, MCPG, fully prevented glutamate and neurons effects on GFAP gene, whereas it had no effect on TGF-β1 activation of the GFAP gene promoter (Fig. 2). Effects of glutamate and TGF-β1 in the levels of endogenous GFAP protein resembled those obtained for β-Gal quantification (data not shown).

Figure 2.

 Activation of GFAP gene by glutamate is mediated by TGF-β1. Cerebral cortex astrocytes from newborn transgenic mice were cultured alone (a–c) or in the presence of TGF-β1 (10 ng/mL) (d–f), glutamate (100 μM) (g–i) or E14 neurons, in presence or absence of 10 μg/mL of neutralizing antibodies against TGF-β1 (α-TGF-β1) or 500 μM of the mGlu2/3 receptor antagonist, MCPG. After 24 h of culture, β-Gal astrocyte number was analyzed (j). Each point represents the average of three independent experiments carried out in triplicate *p < 0.001 (in comparison to the control). Glutamate, neurons and TGF-β1 dramatically increased β-Gal astrocyte number whereas addition of α-TGF-β1 or MCPG prevented this effect. Simultaneous addition of MCPG and TGF-β1 did not inhibit TGF-β1 induction of GFAP gene (j). Scale bar: 100 μm.

Since GFAP gene promoter of astrocytes from different brain regions are distinctly modulated by growth factors and cell interactions (Sousa et al. 2004), we investigated glutamate influence on GFAP gene from astrocytes derived from different structures. Astrocytes from the cerebral cortex, midbrain and cerebellum of newborn transgenic mice were cultured alone and in the presence of 100 μM of glutamate or 10 ng/mL of TGF-β1 for 24 h. Glutamate and TGF-β1, which greatly increased the number of β-Gal positive astrocytes in cerebral cortex monolayers as previously discussed (125–150% increase), had no effect on cerebellar astrocytes, and only glutamate had an effect on midbrain astrocytes (75% increase in the number of β-Gal midbrain positive astrocytes) (Fig. 3). Taken together, these results suggest that GFAP gene might be differently modulated by glutamate in distinct brain regions.

Figure 3.

 GFAP genes from distinct brain regions are differently modulated by glutamate. Cerebral cortex, cerebellum and midbrain astrocytes from newborn transgenic mice were cultured alone, in the presence of glutamate (100 μM) or TGF-β1 (10 ng/mL) for 24 h. After β-Gal activity detection, β-Gal-positive astrocytes were quantified. Each point represents the average of three independent experiments done in triplicate.*p < 0.05; **p < 0.01 (in comparison to the control) (mean ± SD). Glutamate and TGF-β1 differently affected distinct subpopulations of astrocytes. Whereas both neuromodulators had effect on cortical astrocytes, neither of them activated cerebellar GFAP gene and only glutamate induced GFAP gene from midbrain.

To fully demonstrate that TGF-β1 is a mediator of glutamate effect on GFAP gene from cerebral cortical astrocytes, we analyzed Smad 4 nuclear translocation, a hallmark of TGF-β1 pathway activation, in response to glutamate exposure. As observed in Fig. 4(a), control astrocytes present Smad 4 predominantly in the cytoplasm. Treatment with TGF-β1 (Fig. 4b) or glutamate (Fig. 4c) for 24 h induced nuclear translocation of this transcription factor demonstrating that glutamate activates the TGF-β1/Smad pathway in cortical astrocytes. Further, addition of MCPG impaired glutamate-induced-Smad 4 translocation (Fig. 4d). These data together suggest a key role for TGF-β1 as a downstream mediator of glutamate and neurons effect on GFAP gene promoter.

Figure 4.

 Activation of GFAP gene by glutamate involves the TGF-β1/Smad pathway. Cerebral cortex astrocytes from newborn transgenic mice were cultured alone (a) or in the presence of TGF-β1 (10 ng/mL, b), glutamate (100 μM, c) or glutamate in the presence of 500 μM of the mGlu2/3 receptor antagonist, MCPG (Glutamate + MCPG, d). Immunolabeling for Smad 4 revealed a cytoplasmic localization in control cultures (a) and nuclear distribution in TGF-β1 (b) and glutamate-treated (c) cultures. Addition of MCPG impaired glutamate-induced Smad 4 nuclear translocation (d). Insets show DAPI nuclear staining. Scale bars: 30 μm.

Activation of the GFAP gene promoter involves MAPK/PI3K pathways

Since activation of mGluR has been associated with induction of MAPK/PI3K pathways in cortical astrocytes leading to TGF-β1 secretion (D’Onofrio et al. 2001), we investigated the signaling pathways involved in GFAP gene activation by applying MAP and PI3 kinase inhibitors. Cortical astrocytes from newborn transgenic mice were cultured alone or with 100 μM of glutamate, neurons, conditioned medium (CM) from cocultures or 10 ng/mL of TGF-β1, in the presence or absence of 50 μM of PD98059 (MAPK inhibitor), 5 μM of LY294002 (PI3K inhibitor) or both. As shown, in Fig. 5, neurons, CM, glutamate and TGF-β1 greatly increased the number of β-Gal positive astrocytes. Addition of MAPK or/and PI3K inhibitors fully prevented such increase in all conditions studied (Fig. 5), consistent with previous data that glutamate induction of synthesis and secretion of TGF-β1 involves MAP/PI3K pathways (D’Onofrio et al. 2001). Addition of PD98059 and LY294002 to control astrocytes had no effect on cell viability as revealed by cell number and integrity (data not shown). Surprisingly, addition of MAPK and PI3K inhibitors also abrogated TGF-β1 direct action on GFAP gene promoter.

Figure 5.

 Activation of the GFAP gene promoter by glutamate or TGF-β1 itself is mediated by MAPK/PI3K pathways. Cerebral cortex astrocytes from newborn transgenic mice were cultured alone (a,b), in the presence of glutamate 100 μM (c,d), E14 neurons, conditioned medium (CM), or TGF-β1 (10 ng/mL) (e,f); either in the presence (b,d,f) or absence (a,c,e) of PD98059 (50 μM) + LY294002 (5 μM). GFAP promoter-directed expression of lac-Z was revealed by X-Gal (blue nuclei) prior to anti-GFAP immunocytochemistry (gray staining). After 24 h of culture, β-Gal astrocyte number was analyzed (g). Each point represents the average of three independent experiments carried out in triplicate *p < 0.001 (in comparison to the control). Glutamate, neurons, CM and TGF-β1 increased β-Gal astrocyte number. Addition of MAPK or PI3K inhibitors fully prevented the effect of glutamate, neurons (or CM) and TGF-β1 on GFAP gene. Scale bar: 50 μm.

Since MAP/PI3K pathways have been reported to modulate the Smad pathway we sought to investigate if the observed effect was due to secondary blockage of the canonical pathway by the inhibitors (Derynck and Zhang 2003; Dziembowska et al. 2007). We thus analyzed the levels of Smad 2 phosphorylation by Western blotting assays. Treatment of astrocytes with 10 ng/mL of TGF-β1 for 30 min increased the levels of P-Smad 2 in 2,8 times. Such increase was sustained even in the presence of 50 μM of PD98059 (MAPK inhibitor) and 5 μM of LY294002 (PI3K inhibitor) (Fig. 6a and b). Consisting with Western blotting assays, treatment of astrocytes with PD98059 and LY294002 did not impair Smad nuclear translocation, suggesting that MAPK and PI3K do not modulate TGF-β1-induced Smad 2 phospholylation (Fig. 6f). Addition of PD98059 and LY294002 to astrocytes cultures in the absence of TGF-β1 had no effect on the levels of P-Smad 2.

Figure 6.

 TGF-β1 phosphorylation of Smad 2 is independent of MAPK/PI3K pathways in astrocytes. Cerebral cortex astrocytes from newborn mice were cultured alone (Control), in the presence of TGF-β1 (10 ng/mL) (TGF-β1); PD98059 (50 μM) + LY294002 (5 μM) (PD + LY); or simultaneously, in the present of TGF-β1 and both inhibitors (TGF + PD + LY) for 30 min. (a) Representative western blot and (b) graphic analysis showing levels of phosphorylated Smad 2 (P-Smad 2) content: Equal amounts (20 μg/lane) of total protein of astrocytes cultures were resolved in 10% poliacrylamide gel and analyzed by immunobloting for P-Smad 2. The levels of the Smad protein immunoreactivity are expressed relative to the levels of the housekeeping gene protein, α-Tubulin. Inhibition of the non-canonical TGF-β1 pathways, MAPK and PI3K pathways, did not block Smad 2 phosphorylation in response to TGF-β1. *p < 0.05 (in comparison to the control). (c–f) Immunolabeling for Smad 4 revealed a cytoplasmic localization in control cultures (c) and nuclear distribution in TGF-β1 (d). Addition of PD98059 and LY294002 did not affect TGF-induced Smad 4 nuclear translocation (f). Scale bar: 100 μm.

Together, these data show that glutamate/TGF-β1 induction of the GFAP gene promoter involves either the Smad-canonical pathway and the non-canonical MAPK/PI3K pathways.

Discussion

In the present work, we demonstrated for the first time that the excitatory neurotransmitter glutamate activates the GFAP gene promoter of cerebral cortical astrocytes through induction of TGF-β1 signaling pathway. Our work provides strong evidence that synaptically associated astrocytes besides controlling synaptic function are also targets of neuronal activity. Further, our data might provide new insights into the role of glial cells in neurodegenerative disorders associated with glutamate neurotoxicity.

We previously demonstrated that neurons activate GFAP gene promoter and induce astrocyte differentiation in vitro by inducing synthesis and secretion of TGF-β1 by astrocytes (De Sampaio e Spohr et al. 2002; Sousa et al. 2004). Here, we report that this event is triggered by activation of mGluR in astrocytes. Our results are consistent with previous data that 1) activation of group-II mGluRs in astrocytes leads to increased formation and release of TGF-β (Bruno et al. 1998; Corti et al. 2007); 2) neurotransmitters might modulate astrocyte morphology and GFAP expression (Matsutani and Yamamoto 1997; Runquist and Alonso 2003). Our work is the first, however, to show a link between these biological events, by reporting that the excitatory neurotransmitter glutamate activates the GFAP gene promoter and triggers astrocyte differentiation, through TGF-β pathways.

Most of the astrocytes from the cerebral cortex arise from a population of specialized progenitor cells, called radial glia (RG) cells, which act as a scaffold for newborn migrating neurons (Rakic 1971). After neurogenesis and neuronal migration are completed, most of RG cells transform into astrocytes (Culican et al. 1990; De Azevedo et al. 2003). The transition from RG phenotype to astrocytes in the rodent CNS is characterized by replacement of RG-markers by the astrocyte differentiation marker, GFAP (Dahl 1981; Pixley and de Vellis 1984). Although this event is well recognized, molecules and mechanisms involved are not completely identified.

The fact that glutamate antagonists impair RG cell development and neuronal migration, together with described expression of glutamate receptors isoforms in the embryonic brain greatly suggest that these cells might be target for glutamate action (Luján et al. 2005). We recently showed that neurons activate the GFAP gene promoter in RG cells by activating the TGF-β1 pathway (Stipursky and Gomes 2007). Increased expression of mGlu3 in the late embryonic, gliogenic period, together with the fact that RG expresses TGFR (Miller 2003;Sousa et al. 2006; Stipursky and Gomes 2007) opens the question that glutamate/TGF-β homeostase might be key elements in RG-astrocyte transition during cerebral cortex development. Additional support for a cross-talk between GFAP/astrocyte differentiation and glutamate signaling emerged from the recent demonstration that expression of GFAP is essential to anchor the glutamate transporter GLAST in the astrocyte plasma membrane thus enhancing GLAST-mediated transport (Sullivan et al. 2007). This is consistent with the observation that GFAP knockout mice exhibit reduced glutamate clearance (Hughes et al. 2004). These works, together with ours, suggest that changes in GFAP gene expression and glutamate homeostasis might mutually influence each other and support the hypothesis that glutamate/TGF-β1 cross-talk might be a key step to ensure that astrogliogenesis and astrocyte differentiation temporally fit synapse establishment period.

Transforming growth factor-β signaling is mediated mainly by two serine threonine kinase receptors, TGFRI and TGFRII, which activate Smad 2/3 and Smad 4 transcription factors. Phosphorylation and activation of these proteins is followed by formation of Smads 2/3-4 complex, which translocates to the nucleus regulating transcriptional responses to TGF-β (Shi and Massagué 2003). In the present work, we demonstrated that glutamate priming of cerebral cortical astrocytes induced Smad nuclear translocation. Further, it also implicated the MAPK and PI3K pathways. Addition of the specific inhibitor of MEK1/2 kinase, PD98059, or the PI3K inhibitor, LY294002, fully prevented activation of GFAP gene promoter in response to glutamate. This is in close agreement with previous data that astrocyte TGF-β1 synthesis in response to activation of glial metabotropic receptors is mediated by MAP and PI3 kinases (D’Onofrio et al. 2001; Corti et al. 2007). Surprisingly, these inhibitors also abrogated TGF-β1 direct action on GFAP gene promoter, suggesting that besides the Smad pathway, non-canonical pathways are also implicated in TGF-β1-induced GFAP gene activation.

Our data are supported by evidences that TGF receptors might act through multiple intracellular pathways such as protein kinase A, Jun N-terminal kinase and p38 kinases, and MAP and PI3 kinases, besides the classical Smad-mediated pathway (Zhu et al. 2002, 2004; Derynck and Zhang 2003). As pharmacological inhibition of MAPK or PI3K completely abrogated the effect of TGF-β1 in GFAP gene promoter, it is conceivable that transactivation of TGFR leads to 1) induction of PI3K followed by MAPK pathway activation (either PI3K upstream or downstream) or 2) alternatively, TGFR may activate two separate cascades, a PI3K dependent pathway and the classical MAPK pathway (Fig. 7). The fact that administration of LY294002 and PD98059 alone is sufficient to completely inhibit TGF-β effect and concomitant addition does not yield additive inhibition (data not shown) calls in favor of converging, rather than independent pathways, as previously shown in other systems (Carballada et al. 2001; Yart et al. 2001; Kim et al. 2002; Yu et al. 2002).

Figure 7.

 Cross-talking between glutamate and TGF-β1 signaling pathways. Astrocyte differentiation induced by the neurotransmitter glutamate is mediated by group-II glial metabotropic receptors 2 and 3 (mGlu2/3) and requires glial production of transforming growth factor-β1 (TGF-β1). Through activation of MAPK and PI3K pathways, glutamate increases production and release of TGF-β1 by astrocytes, which in turn protects neighboring neurons against excitotoxic death (1). Additionally, after binding to its receptor, TGFR (2), TGF-β1 activates GFAP gene and induces astrocyte differentiation through the non-canonical MAPK/PI3K (3) and canonical Smad (4) TGF-β1 pathways.

We previously demonstrated that activation of GFAP gene promoter by TGF-β1 in RG-like cells and astrocytes is triggered by the Smad-dependent pathway (Stipursky and Gomes 2007). Although this pathway is the main mediator of TGF-β signaling, TGF responses have been showed to be the result of a balance between other intracellular signaling pathways. An alternative possibility would be that TGF-β-induced GFAP gene activation, as reported here, involves interaction between Smad-dependent and -independent pathways (Derynck and Zhang 2003). Such cross-talk has been observed for example, in human glioblastoma cells, in which Smad 2 phosphorylation is decreased by p38/MAPK inhibitor (Dziembowska et al. 2007). Our data contrast with those since MAPK and PI3K inhibitors did not interfere with Smad 2 phosphorylation. The fact that inhibition of non-canonical pathway completely inhibited TGF effect on GFAP suggests that the Smad canonical pathway is not sufficient to assure GFAP gene activation in astrocytes in response to TGF-β1.

The Smad complexes binds DNA sequences termed Smad-binding elements that contain a minimal four-nucleotide domain AGAC also called ‘CAGA’ box (Massagué 2000). These domains have not been identified within the GFAP gene promoter and Smad effects on GFAP gene have been suggested to be the result of cooperation between Smad proteins and other molecules such as STAT in neural progenitors (Nakashima et al. 1999; Li and Grumet 2007). It remains to be determined, however, the molecular mechanism by which Smad activates the GFAP gene in astrocytes (direct binding or indirect binding).

Our data show that TGF-β1-activation of GFAP gene involves two separate cascades: the canonical and non-canonical pathways. Elucidation of the molecular mechanisms implicating PI3K/MAPK and Smad pathways in GFAP gene activation and astrocyte differentiation await further experiments (Fig. 7).

We reported here that astrocytes derived from different brain regions present distinct responses to the glutamatergic signaling. Many evidence now support the concept that astroglia is formed by a heterogeneous population of cells that markedly vary in their responsiveness to several agents such as hormone, growth factors, neurotransmitters, and pattern of cell interaction (Dennis-Donini et al. 1984; Garcia-Abreu et al. 1995; Lima et al. 1998; Perego et al. 2000; De Sampaio e Spohr et al. 2002; Schlüler et al. 2002; Matthias et al. 2003; Sousa et al. 2004). We previously reported that cortical neurons do not activate the GFAP gene promoter of transgenic astrocytes derived from midbrain and cerebellum although they induced synthesis of TGF-β1 by these cells (Sousa et al. 2004). Such unresponsiveness to neurons was not related to different expression of members of the TGF-β pathway since we observed wild distribution of TGF-β1 and its receptor in subpopulations of astrocytes derived from all regions (Sousa et al. 2004). The fact that cerebellar astrocytes do not respond to glutamate by activating lacZ-GFAP gene calls in favor of TGF-β as a mediator of glutamate effects on GFAP gene since TGF-β1 has already been shown not to activate GFAP gene from these cells (Sousa et al. 2004). It is intriguing, however, that although TGF-β1 did not activate the lacZ-GFAP gene from midbrain astrocytes (Sousa et al. 2004; data from the present paper), glutamate did. A possibility is that activation of group-II mGluRs might be associated with modulation of other factors, rather than TGF-β in midbrain, as suggested for other systems (Ciccarelli et al. 1999).

Glutamate signaling is a balance between their receptors and transporters, which untimely determines the extracellular concentration of the neurotransmitter. Previous work provided evidence that glutamate signaling is under local CNS region-specific modulation (Schlüler et al. 2002; Regan et al. 2007). Whereas the pituitary adenylate cyclase activating peptide, TGF-α and epidermal growth factor act as potent regulators of the expression of glutamate transporters in striatal glia, these factors completely failed to affect glutamate transporter-1 and GLAST protein levels in astroglial cultures from cerebellum, mesencephalon and spinal cord (Schlüler et al. 2002). Further, the expression of mGlu receptor subtypes presents distinct regional and temporal profile, which might account for the diversity of glutamatergic signaling found in the brain (Porter and McCarthy 1997) and in the present work.

Assembly of synapses requires coordination between pre- and pos-synaptic neurons and the surround glial cells (Allen and Barres 2005). TGF-β signaling has been implicated in synapse formation and stabilization in invertebrate models (Packard et al. 2003). The data presented here address as an attractive issue, the investigation of the role of TGF-β1 and astrocytes in development of vertebrate glutamatergic synapses.

It has been suggested that dysfunctional glial glutamate transporters and receptors are associated with several neurological disorders such as amyotrophic lateral sclerosis, Alzheimer’s disease and epilepsy (Seifert et al. 2006; Oberheim et al. 2008). High levels of TGF-β1 are observed in several diseases where glutamate metabolism dysfunctions are described (Tesseur and Wyss-Coray 2006). The cross-talk between TGF-β1 and glutamate signaling might offer a new strategy to increase the local production of neurotrophic factors in the CNS. Understanding the cellular components (astrocytes and neurons) and molecules (TGF-β and neurotransmitters) underlying synapse functions not only contribute to go deeper into neuro-glia interactions during CNS development but might create new strategies towards clinical neuroprotection trials with mGluR-selective compounds.

Acknowledgements

We thank Ismael Gomes, Adiel Batista do Nascimento, and Rosenilde Afonso for technical assistance. We also thank Daniela Uziel-Rozenthal (Federal University of Rio de Janeiro) for carefully reading the manuscript. This work was supported by grants from FAPERJ, CNPq, PRONEX, and CAPES.

Ancillary