Abbreviations used: E, embryonic day; EM, electron microscopy; FITC, fluorescein isothiocyanate; GFAP, glial fibrillary acidic protein; MAP2, microtubule-associated protein 2; P, postnatal day; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; SDS, sodium dodecyl sulfate; TTX, tetrodotoxin.
Address correspondence and reprint requests to Dr. G. Pietrini at CNR Cellular and Molecular Pharmacology Center, Via Vanvitelli 32, 20129 Milano, Italy. E-mail: GraziaP@Farma.csfic.mi.cnr.it
Abstract: The GLT-1 and GLAST astroglial transporters are the glutamate transporters mainly involved in maintaining physiological extracellular glutamate concentrations. Defects in neurotransmitter glutamate transport may represent an important component of glutamate-induced neurodegenerative disorders (such as amyotrophic lateral sclerosis) and CNS insults (ischemia and epilepsy). We characterized the protein expression of GLT-1 and GLAST in primary astrocyte—neuron cocultures derived from rat hippocampal tissues during neuron differentiation/maturation. GLT-1 and GLAST are expressed by morphologically distinct glial fibrillary acidic protein-positive astrocytes, and their expression correlates with the status of neuron differentiation/maturation and activity. Up-regulation of the transporters paralleled the content of the synaptophysin synaptic vesicle marker p38, and down-regulation was a consequence of glutamate-induced neuronal death or the reduction of synaptic activity. Finally, soluble factors in neuronal-conditioned media prevented the down-regulation of the GLT-1 and GLAST proteins. Although other mechanisms may participate in regulating GLT-1 and GLAST in the CNS, our data indicate that soluble factors dependent on neuronal activity play a major regulating role in hippocampal cocultures.
Glutamate is the major excitatory neurotransmitter in the CNS and is involved in several brain functions, such as learning and memory (Bliss and Collingridge, 1993). However, the excessive accumulation of extracellular glutamate leads to neuronal death and is involved in the pathophysiology of ischemic brain damage (Schousboe and Frandsen, 1995). Furthermore, increased levels of extracellular glutamate have also been implicated in various neurodegenerative diseases, such as amyotrophic lateral sclerosis (Rothstein et al., 1995), Alzheimer's disease (Greenamyre and Young, 1989), and Huntington's disease (Perry and Hansen, 1990).
Several studies have demonstrated that the astroglial GLT-1 and GLAST transporters play the major role in maintaining physiological extracellular glutamate levels (Rothstein et al., 1996), and, given their crucial role in protecting neurons against excitotoxicity, their expression and activity must be strictly regulated.
Several aspects therefore remain to be clarified, and we also wondered whether the signals exchanged by neurons and astroglia to regulate glutamate concentration may vary according to the brain region because differential GLT-1 and GLAST regulation has been shown in different areas of the brain (Lehre et al., 1995; Ullensvang et al., 1997).
In this study, we examined the neuronal-induced regulation of astroglial glutamate transporters in primary astrocyte—neuron cocultures derived from the hippocampal region, a brain area in which glutamate transmission plays a fundamental role.
MATERIALS AND METHODS
Primary hippocampal astrocyte—neuron cocultures
The hippocampal cells were prepared from Sprague—Dawley rats at postnatal day (P) 1 or P3. Single-cell suspensions were obtained by means of mechanical dissociation in Hanks' balanced salt solution containing 2.5 mg/ml trypsin type XI and DNase I type IV (0.1 mg/ml). The cells were recovered by centrifugation and plated onto poly-L-lysine-coated Petri dishes (Corning; 35 mm in diameter) at a density of 150,000-200,000/ml. The culture media contained minimal essential medium, 0.6% glucose, 1 mM glutamine, 2.2 g/L NaHCO3, 10 μg/ml bovine transferrin, 25 μg/ml insulin, and 10% fetal calf serum (heat-inactivated for 30 min at 56°C). The cultures were maintained at 37°C in a 5% CO2 humidified incubator. Half of the culture media was replaced 1 day after plating and then, unless otherwise indicated (new media), maintained until the end of the experiments. In some experiments, the media were supplemented with cytosine arabinoside (3 mM) after 4 days to prevent the overgrowth of the astroglial cells (this treatment did not affect the expression of the transporters).
Immunofluorescence and immunogold-electron microscopy (EM)
The cultures were plated onto sterile glass coverslips coated with poly-L-lysine, rinsed briefly in warm phosphate-buffered saline (PBS), and fixed with 4% paraformaldehyde and 4% sucrose in PBS for 20 min at room temperature before being processed for immunofluorescence as previously described (Cameron et al., 1991). The GLT-1 and GLAST astroglial glutamate transporters were localized by means of rabbit polyclonal sera produced against synthetic peptides consisting of amino acids 1-16 (GLAST N) and 529-543 (GLAST C) at the N and C terminus, respectively, of rat GLAST and 496-512 (GLT-1 nC) and 562-573 (GLT-1 C) at the C termini of rat GLT-1 (the peptide sequences are shown in Fig. 1), which were coupled to keyhole limpet hemocyanin and individually subcutaneously injected into one rabbit each. The resulting sera (given the same name as the peptides) were affinity-purified using the immunogenic peptides immobilized on CNBr-activated Sepharose (Pharmacia) and used at a final concentration of 5 μg/ml in the western blot and immunofluorescence experiments. Monoclonal antibodies against glial fibrillary acidic protein (GFAP; Boehringer) and polysialogangliosides [A2B5; kindly provided by Dr. F. Aloisi (Aloisi et al., 1988)] were used for the double immunofluorescence labeling of the transporters with astroglial markers. The acquisition of neuronal polarity was tested using an anti-microtubule-associated protein 2 (MAP2; Boehringer) monoclonal antibody. Fluorescein isothiocyanate (FITC)-conjugated anti-mouse IgG and Texas Redconjugated anti-rabbit IgG from Jackson Immunoresearch (West Grove, PA, U.S.A.) were used as secondary reagents. In the double staining shown in Fig. 2C the cells were first stained using the GLT-1 C antibody, followed by Texas Red-conjugated anti-rabbit IgG, and then stained using the GLAST N antibody followed by FITC-conjugated anti-rabbit. To avoid binding of the FITC-conjugated secondary antibody to free GLT-1 antibody, the cells were incubated in an excess of anti-rabbit Fab before the GLAST staining.
For the immunogold-EM experiments cells plated on 10-cm-diameter Petri dishes were scraped with a rubber policeman into PBS, collected by centrifugation, and fixed for 30 min at room temperature with 4% paraformaldehyde and 0.2% glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.3. The cell pellets were equilibrated in concentrated sucrose, frozen, sectioned, incubated with antibodies, postfixed, stained, and embedded as previously described (Villa et al., 1993). The sections were first incubated with anti-GFAP (used as a marker of type 1 and 2 astrocytes) and anti-transporter antibodies, followed by anti-mouse IgG 15-nm-diameter and antirabbit IgG 5-nm-diameter gold particle conjugates. The stained sections were observed and photographed using a Hitachi model H7000 electron microscope.
Western blot analyses
The cultures in the 30-mm-diameter Petri dishes were washed twice in warm PBS and solubilized in 125 μ1 of buffer containing 0.45 M Tris-HCl (pH 8.9) and 4.5% sodium dodecyl sulfate (SDS), and their protein content was assayed following the microtiter plate protocol of the BCA protein assay reagent kit (Pierce). The solubilized samples were diluted 1:2 with a solution containing 60% glycerol, 30% β-mercaptoethanol, and 10% bromophenol blue, and 20 μg of each sample was loaded onto 10% SDS-polyacrylamide gel electrophoresis (PAGE). Total brain homogenates were prepared in 4 volumes (wt/vol) of buffer containing 320 mM sucrose and 4 mM HEPES-NaOH (pH 7.3) (Huttner et al., 1983), and their protein content was assayed as described above. The SDS-treated brain extracts were prepared from the homogenates of embryonic [embryonic day (E) 18], neonatal (P1), or adult (A) rats; 30 μg of each extract was loaded onto 10% SDS-PAGE, and the proteins were blotted onto nitrocellulose at 120 mA overnight as previously described (Borgese and Pietrini, 1986). The blots were immunostained with 125I-protein A (Amersham) as the secondary reagent (Borgese and Pietrini, 1986). In most experiments, the blots were probed with a polyclonal antibody raised against synaptophysin (p38) [kindly provided by Dr. R. Jahn (Fletcher et al., 1991)] and a monoclonal antibody raised against the marker of type 1 and 2 astroglial cells (GFAP; Boehringer). To analyze the radioimmunoblotting experiments quantitatively, radioactive bands were excised from nitrocellulose and counted in an auto-γ-counter.
Differential expression of GLT-1 and GLAST in astroglial cells
To study the protein expression of the astroglial transporters, we raised anti-peptide antibodies against rat GLT-1 and GLAST (the sequences of the peptides and the names of the resulting antibodies are indicated in Fig. 1). The antibodies were affinity-purified, and their specificity was tested by western blots of brain membranes and immunofluorescence of the neuron—astrocyte cocultures.
On western blotting, the affinity antibodies recognized electrophoretic mobility bands corresponding to those reported for the two transporters: In particular, the GLAST antibodies recognized a lower band (∼55 kDa) than the GLT-1 antibodies (∼60 kDa) (Fig. 1). Immunostaining of blots containing SDS-treated brain extracts from embryonic (E18), neonatal (P1), and adult (A) rats revealed the differential expression of the two transporters during brain development: The peak expression of both transporters occurs in mature brains, but GLAST expression is already high by E18, and that of GLT-1 is low in both embryonic and neonatal brains.
The same antibodies were then used to localize the astroglial glutamate transporters in the astrocyte—neuron cocultures, with similar results being obtained regardless of whether we used the GLAST N or GLAST C and the GLT-1 nC or GLT-1 C antibodies. All of the figures shown in this article were obtained using the GLAST N and the GLT-1 C antibodies. The 10-day-old cocultures contained a network of neurons lying on a bed of astroglial cells. The majority of the astroglial cells were flat polygonals, but there were also cells with a neuron-like morphology that had processes of various thicknesses and lengths; both of these cell types were stained by the antibodies against GFAP, the marker of type 1 and 2 astrocytes. Indirect immunofluorescence staining revealed that both transporters were expressed by the GFAP-positive cells, but whereas GLAST was particularly enriched in the flat polygonal cells (Fig. 2A), the GLT-1 antibody stained the process-bearing cells (Fig. 2B). Double staining with anti-GLT-1 and anti-GLAST antibodies confirmed the differential distribution of the two transporters in morphologically distinct cell types (Fig. 2C).
The two cell types also had distinct behaviors: The process-bearing astrocytes were always isolated cells that often grew on top of the flat cell layer or in areas free of flat cells, whereas the flat polygonal cells started to make connections with each other soon after plating and, after 2 weeks of culture, reached confluency and formed a monolayer of adjacent cells. The cell surface distribution of GLAST was not uniform: In isolated cells, it accumulated at the tips of short processes; in subconfluent and confluent cells, it concentrated in the membrane domains contacting adjacent cells (Fig. 3A). The surface localization of GLAST in the GFAP-positive cells was further confirmed by EM, which revealed GLAST antibodies (small gold particles) decorating the cell surface of GFAP-containing cells (large gold particles) in ultrathin frozen sections (Fig. 3B). EM was also performed in the case of GLT-1 antibodies, but their antigenicity was not sufficiently preserved to allow the transporter to be clearly detected.
Because the GLT-1 expressing cells had the characteristic morphology of type 2 astrocytes, we tested the colocalization of GLT-1 using a marker of type 2 astrocytes (the A2B5 antibody): No colocalization of the transporter was observed in the A2B5-positive cells. However, we cannot exclude the possibility that GLT-1-expressing cells are type 2 astrocytes because the cells positive for A2B5 (or the polysialogangliosides recognized by A2B5) progressively disappear from the cultures (Aloisi et al., 1988), and in young cultures GLT-1 expression was below the level of detection (data not shown).
The neurons identified by the MAP2 or p38 markers revealed virtually no coexpression of GLAST, whereas weak GLT-1 expression was detected with the GLT-1 antibodies in the neuronal processes (data not shown).
Up-regulation of GLT-1 and GLAST proteins accompanies neuronal differentiation and maturation
It has been shown that neuron-free primary astroglial cultures express very low levels of GLT-1 but that the presence of neurons in the cultures induces the expression of GLT-1 and increases that of GLAST (Gegelashvili et al., 1997; Swanson et al., 1997; Schlag et al., 1998). However, as these studies did not indicate whether transporter expression was dependent on the status of neuronal differentiation and maturation, we assessed these aspects biochemically by measuring the content of the synaptophysin synaptic vesicle marker p38 and morphologically by following the formation of the typical neuronal network of mature, fully differentiated, active neurons (by means of inverted microscope visual inspection and immunofluorescence staining with an antibody against p38). The acquisition of neuron polarity was also assessed immunofluorescently by following the compartmentalization of MAP2 in the somatodendritic compartments. These analyses revealed that the neurons differentiated and formed synaptic contacts by day 6 of culture and that longer culture times only led to a more complex neuronal network (data not shown).
Glial transporter protein expression was assayed by immunoblotting cultures grown at various times. In agreement with the brain total homogenate western blot results, the expression of GLT-1 and GLAST correlated with the maturation of the cultures. The expression of GLAST was already high during the first few days of culture, whereas GLT-1 expression was low; the expression of both transporters increased over time (differentiation/maturation). There was a greater than sixfold increase in GLT-1 content in 10-day-old cultures and an approximately twofold increase in the expression of GLAST. These increases were independent of glial proliferation (measured by GFAP expression) and were similar to the increased levels of the synaptic vesicle marker p38 (Fig. 4).
To distinguish the effect of neuron maturity from that of astrocyte maturity, we measured the expression of the transporters in cocultures in which the levels of the p38 neuronal marker were undetectable. The cultures were prepared from rats at P3 and, 3 days after plating, contained few neurons as revealed by the absence of p38 antibody staining (Fig. 4C). These cultures had a higher transporter content than cultures of the same age derived from P1, but this expression did not increase with the time in culture.
To determine whether neurons are also required to maintain transporter expression levels, we induced neuronal death in ≥8-day-old cultures (expressing high levels of astroglial transporters) by treating them with 500 μM glutamate for 15 min; the effects of the treatment were measured 24 h later. During the 24 h following glutamate treatment, the cells were cultured in a new culture medium or neuronal-conditioned medium (the medium collected from the culture before neuronal killing) and then processed for immunoblotting. Protein expression radioimmunoblotting analysis showed that the glutamate treatment reduced the p38 content by 80-90% but had no significant effect on the GFAP content. The effect of neuronal death on glutamate transporter expression depended on the culture medium: When the cells were cultured in new medium, there was an ∼50% reduction in comparison with the untreated samples, and the reduction in GLT-1 and GLAST levels was also observed 3 days after glutamate-induced neuronal death (data not shown). A clear decrease in transporter expression was also detected by means of immunofluorescence experiments (data not shown). However, when the cells were cultured in the neuronal-conditioned medium, the down-regulation of GLT-1 and GLAST was largely prevented (Fig. 5). Neuronal cells are therefore required to induce and sustain the expression of the astroglial glutamate transporters, and soluble factors are involved in their neuronal-induced regulation in hippocampal cocultures.
Synaptic activity regulates the expression of GLT-1 and GLAST
Our results indicate a relationship between neuronal maturation/differentiation and the expression of GLT-1 and GLAST and suggest the involvement of neuronal activity in their regulation. The involvement of synaptic activity in transporter regulation was directly tested by culturing the cells with compounds expected to reduce neuronal excitability and synaptic transmission without altering neuronal maturation (Craig et al., 1994; Verderio et al., 1994). Three-day-old cocultures were grown with medium containing tetrodotoxin (TTX) or the noncompetitive NMDA receptor antagonist MK-801 (to block Na+ channels or NMDA receptors), with the compounds being added to the culture medium every 2 days. The expression of the transporters, as well as that of the neuronal and glial markers, was assayed 3 or 6 days later by means of radioimmunoblotting. The treatments had no significant effect on the expression of the glial (GFAP) and neuronal (p38) markers, but there was a large reduction in the protein expression of both GLT-1 and GLAST (Fig. 6A). The expression of Na+ channels and NMDA receptors in astrocytes has been previously documented (Araque et al., 1998; O'Connor et al., 1998), and so we tested whether the down-regulation of the transporters was due to a direct effect of the TTX and MK-801 treatment on astrocyte channels and receptors. We observed no effect on transporter expression when cultures derived from P3 rats, and in which the level of p38 was undetectable, were treated with TTX or MK-801 (Fig. 6B).
Our study shows that astroglial GLT-1 and GLAST glutamate transporters in rat hippocampal-derived cocultures of astrocytes and neurons are selectively expressed on morphologically distinct GFAP-positive astrocytes and that neurons regulate the expression of both in a similar manner; our data also indicate that GLT-1 and GLAST protein levels depend on the level of differentiation/maturation and activity of the neurons.
It has been reported that cortical astroglial cells cocultured with neurons change from polygonal (undifferentiated) to stellate (differentiated) cells and that these changes are associated with the induction of GLT-1 expression (Swanson et al., 1997; Schlag et al., 1998). Although more intense GLT-1 staining has been observed in the differentiated than in the polygonal undifferentiated astrocytes (Schlag et al., 1998), the differential expression of the two astroglial transporters in morphologically distinct GFAP-positive glial cells has never been previously described. Our results may be explained by a switch from GLAST to GLT-1 expression during the differentiation of hippocampal astrocytes: however, our data indicate that GLAST expression is up-regulated and not down-regulated by neuronal-dependent soluble factors, thus suggesting that only a subset of the flat polygonal cells differentiates into process-bearing cells expressing GLT-1, or that the process-bearing cells are not differentiated forms. We favor the latter hypothesis because we never found any proof of the isoform switch in the cocultures but did find many process-bearing cells in freshly plated cells lacking both GLAST and GLT-1 expression.
It has been shown that culture conditions may lead to high levels of functional GLT-1 expression in neurons (Brooks-Kayal et al., 1998; Mennerick et al., 1998). Under our culture conditions, both of the GLT antipeptide-specific antibodies detected the expression of GLT-1 protein in neurons, although at much lower levels than those observed in the process-bearing astrocytes. Its expression in neurons therefore contributes to the total amount of GLT-1 protein measured by western blot, but this contribution does not seem to be relevant because the neuronal-conditioned medium maintained ∼90% of GLT-1 protein expression after neuronal killing.
Our characterization of the GLAST-expressing cells revealed an accumulation of the transporter in the surface domains mediating cell—cell contact. This particular localization is consistent with recent data suggesting protein—protein interactions between GLAST and a PDZ protein (Marie and Attwell, 1999) because it is known that PDZ proteins also accumulate in junctional domains (Bredt, 1998). The accumulation of the transporter in the processes contacting adjacent cells is reminiscent of the in vivo localization of the transporters in astroglial processes closely apposed to glutamatergic and GABAergic synapses (Chaudhry et al., 1995; Lehre and Danbolt, 1998).
In our hippocampal-derived primary cocultures, we observed that expression of both GLT-1 and GLAST was similarly dependent on neuronal factors. Our data indicate (a) a link between neuronal differentiation and the up-regulation of both astroglial transporters, (b) that treatments negatively affecting neuronal survival or activity lead to transporter down-regulation, and (c) that neuronal soluble factors prevent the down-regulation of GLT-1 and GLAST owing to neuronal death. As data obtained from cortical cocultures have shown that neuronal factors differentially regulate GLT-1 and GLAST (Gegelashvili et al., 1997; Schlag et al., 1998), it therefore seems that neuronal factors may determine the particular expression of GLT-1 and GLAST in different brain areas.
Our data are consistent with in vivo observations demonstrating that the up-regulation of both GLT-1 and GLAST follows the maturation of the CNS (synapse formation) (Sutherland et al., 1996; Furuta et al., 1997; Ullensvang et al., 1997) and that their down-regulation is a consequence of glutamatergic denervation (Ginsberg et al., 1995; Levy et al., 1995). They also indicate that synaptic activity up-regulates astroglial glutamate transporters as down-regulation of these transporters was measured under conditions of reduced synaptic transmission and excitability obtained by culturing the cells with TTX or MK-801. We excluded the possibility that the measured down-regulation of the transporters was due to a blockade of the astrocyte Na+ channels and NMDA receptors because these compounds had no effect on the cultures containing undifferentiated and/or few neurons.
Although it is possible that other mechanisms may cooperate in regulating the CNS expression of the transporters, the association between neuronal activity and the expression of the GLT-1 and GLAST astroglial glutamate transporters suggests the existence of a feedback regulatory loop tuning glutamate homeostasis in the brain and that interference with this association may cause or aggravate neuronal degeneration.