Protein Kinase C Activation Decreases Cell Surface Expression of the GLT-1 Subtype of Glutamate Transporter

Na (cid:1) -dependent glutamate transporters are required for the clearance of extracellular glutamate and influ-ence both physiological and pathological effects of this excitatory amino acid. In the present study, the effects of a protein kinase C (PKC) activator on the cell surface expression and activity of the GLT-1 subtype of glutamate transporter were examined in two model systems, primary co-cultures of neurons and astrocytes that endogenously express GLT-1 and C6 glioma cells transfected with GLT-1. In both systems, activation of PKC with phorbol ester caused a decrease in GLT-1 cell surface expression. This effect is opposite to the one observed for the EAAC1 subtype of glutamate transporter (Davis, K. E., Straff, D. J., Weinstein, E. A., Bannerman, P. G., Correale, D. M., Rothstein, J. D., and Robinson, M. B. (1998) J. Neurosci. 18, 2475–2485). Several recom-binant chimeric proteins between GLT-1 and EAAC1 transporter subtypes were generated to identify domains required for the subtype-specific redistribution of GLT-1. We identified a carboxyl-terminal domain consisting of 43 amino acids (amino acids 475–517) that is required for PKC-induced GLT-1 redistribution. Mutation

Glutamate is the predominant excitatory neurotransmitter in the central nervous system (CNS) 1 (1,2) and is removed from the synaptic cleft by sodium-dependent glutamate transport activity. This activity is mediated by a family of five subtypes of transporters that share up to 60% sequence identity (for reviews, see Refs. [3][4][5]. Expression of these transporters is generally restricted to particular cell types and brain regions. Two subtypes, GLT-1 and GLAST, are astroglial and two others, EAAC1 and EAAT4, are neuronal. Expression of the fifth transporter, EAAT5, is restricted to the retina (for reviews, see Refs. 4 and 5). Both the neuronal and glial glutamate transporters control the amplitude and/or duration of synaptic responses (6,7) and prevent an extracellular accumulation of this potential excitotoxin (8 -10). For several different reasons, it is thought that the glial transporter, GLT-1, may represent the predominant route for the clearance of extracellular glutamate in forebrain (for review, see Ref. 11). Therefore, defining the mechanisms that regulate GLT-1 has the potential to impact our understanding of both the physiology and pathology of glutamate in the CNS.
Several different second messengers regulate the activity of GLT-1, including free radicals, arachidonic acid, and PKC (for reviews, see Refs. 4 and 5). However, the effects of PKC on GLT-1-mediated activity are varied. For example, activation of PKC causes an increase in activity when GLT-1 is expressed in HeLa cells using vaccinia virus (12) but has no effect on the activity of GLT-1 stably transfected into HeLa cells or two other peripheral cell lines (13). In a cell line that endogenously expresses the human variant of GLT-1 (Y-79 human retinoblastoma), activation of PKC causes a decrease in activity by increasing the K m value for transport (14). In a preliminary study, activation of PKC caused a decrease in GLT-1 activity in stably transfected Madin-Darby canine kidney cells (15). These studies suggest that the effects of PKC activation may be dependent on the levels of transporter expression and/or the presence of other cellular proteins that are not constitutively expressed in all cells.
Many types of membrane-bound proteins undergo a dynamic trafficking between the cell surface and intracellular compartments. Changes in cell surface expression can up-or downregulate the activity of membrane proteins much faster (within minutes) than can normally be achieved by altering the rate of protein synthesis. One well-characterized example is up-regulation of the GLUT4 subtype of glucose transporter, which is rapidly redistributed to the plasma membrane in response to insulin (for reviews, see Refs. 16 and 17). A classic example of down-regulation is agonist-activated internalization of G-protein-coupled receptors (for review, see Ref. 18). Several groups have recently shown that activation of PKC causes a rapid change in the cell surface expression of neurotransmitter transporters, including the GAT1 subtype of GABA transporter (19), the dopamine transporter (20,21), and the serotonin transporter (22). In view of these observations, it may not be unexpected that trafficking of glutamate transporters is also controlled. In fact, our previous studies indicate that the cell surface expression of the neuronal glutamate transporter, EAAC1, is increased by activating PKC or by activating the platelet-derived growth factor receptor (23,24).
In the present study, the effects of a PKC activator on GLT-1 cell surface expression were examined in co-cultures of neurons and astrocytes. In this model system, phorbol ester caused a rapid decrease in the cell surface expression of GLT-1. When GLT-1 was introduced into C6 glioma cells, a CNS-derived cell line that endogenously expresses the neuronal transporter, EAAC1, a PKC activator caused a decrease in the cell surface expression of GLT-1 and a decrease in GLT-1-mediated transport activity. In contrast, a PKC activator caused an increase in EAAC1 cell surface expression in these same experiments, indicating that the differential effects of PKC are related to differences in primary structure of these two highly homologous proteins. Using chimeras made of reciprocal domains of GLT-1 and EAAC1, we demonstrate that a 43-amino acid carboxyl-terminal domain is both necessary and sufficient for PKC-mediated redistribution of GLT-1. Following mutation of serine and threonine residues in this domain, we show that a non-conserved serine residue (serine 486) is partially responsible for the regulated decrease in GLT-1 cell surface expression. Although we found that a PKC activator increased incorporation of 32 P into wild type GLT-1, mutation of this single serine residue did not reduce this signal. During these studies, we also developed evidence that the same 43-mino acid domain of GLT-1 was required for functional expression of transporter activity in specific chimeras. Together, these studies provide evidence for PKC-dependent regulation of GLT-1 and identify a structural domain required for this effect.
Cell Culture-Neuron/astrocyte-mixed cultures were prepared from embryonic day 17-19 cortices of rat as described previously (8) with minor modifications. Cells were plated onto poly-D-lysine (50 g/ml)coated plastic dishes and maintained in a 7% CO 2 incubator at 37°C. The cultures were fed with a one-third medium exchange twice a week. After 7-10 days, neurons in these cultures sit on top of a monolayer of astrocytes and represent Ͻ40% of the cells. GLT-1 is endogenously expressed in these cultures, and most of the immunoreactivity codistributes with glial-specific markers (26).
C6 glioma cells were obtained from the American Type Culture Collection (Rockville, MD) and maintained as previously described (24). Briefly, cells were grown in DMEM supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin and maintained at 37°C in 5% CO 2 incubator. Cells were passaged less than 60 times. Under these conditions no gross changes in morphology or changes in experimental effects were observed. Cells were incubated with PMA or vehicle (Me 2 SO) for 30 min and used for experiments as described below.
Construction of Chimeric Transporters-Chimeric glutamate transporters were constructed using cDNAs encoding GLT-1 and EAAC1 rat transporter subtypes, provided by Drs. Kanner and Hediger, respectively (27,28). Chimeras were generated utilizing unique restriction sites contained within at least one of the transporters. If the restriction site was not contained in both transporters, it was introduced by PCR using appropriate oligonucleotides. For example, the Tth111I restriction site, present only in the GLT-1 cDNA, was introduced at the homologous site in EAAC1. In all cases, the restriction sites were introduced so that the resulting protein would contain the amino acid sequences from one of the two transporters. Twelve different chimeric transporters were constructed in this way. All chimeras that did not contain the carboxyl-terminal epitope recognized by the anti-GLT antibody (amino acids 557-573 of GLT-1) (25) were tagged with an HA epitope (YPYDVPDYA) that could be used to follow the trafficking. Wild type as well as chimeric transporters were subcloned into the mammalian expression vector pcDNA3.1(ϩ) (Invitrogen, Carlsbad, CA). The sequences of all chimeras were verified by sequencing at the molecular biology core at the Children's Hospital of Philadelphia.
Generation of Mutant GLT-1-All mutants were made using PCR and a 5Ј-mutagenic primer at a unique PshA1 site or a 3Ј-mutagenic primer at a unique Tth111I site to amplify 339-or 1559-bp fragments of GLT-1 between PshA1-XbaI or EcoRI-Tth111I restriction sites, respectively. Amplified fragments were subcloned back into the expression vector pcDNA-GLT-1 and sequenced to verify the mutations.
Transfection of C6 Glioma Cells-C6 glioma cells were grown to between 50 and 60% confluence in 10-cm dishes prior to introduction of cDNA plasmids using a cationic lipid-based procedure, GenePorter. Cells plated in 10-cm dishes were transfected with 8 -12 g of DNA mixed with 60 l of GenePorter. The amounts of cDNA/GenePorter were adjusted for surface area/numbers of cells in 12-well plates. After a 3-to 5-h incubation in 5 ml of DMEM, an additional 5 ml of modified medium (DMEM containing 20% fetal bovine serum, 4 mM glutamine, 200 units/ml penicillin, and 200 g/ml streptomycin) was added to the mixture for an additional 15 h. To produce stably transduced cells, calcium phosphate precipitation was used to introduce the cDNA, and stably transfected clones were selected with 600 g/ml G418.
Measurement of Na ϩ -dependent Transport Activity-Na ϩ -dependent L-Glu transport activity was measured as previously described (24). Cells were grown in a monolayer on 12-well plates and rinsed twice with 1 ml of warmed (37°C) sodium-or choline-containing buffer prior to incubation with radioisotope (0.5 M L-[ 3 H]Glu) in sodium-or cholinecontaining buffer for 5 min. After stopping uptake with ice-cold cholinecontaining buffer, cells were solubilized in 0.1 N NaOH. Aliquots of this lysate were analyzed for protein and radioactivity. Na ϩ -dependent uptake was defined as the difference between the signal observed in Na ϩ -containing and in choline-containing buffer.
Biotinylation of Cell Surface Proteins-The biotinylation procedure was performed as described previously with slight modifications (23,24). Briefly, cells were grown to between 60 and 90% confluence. Plates were rinsed with ice-cold phosphate-buffered saline containing 0.1 mM CaCl 2 and 0.1 mM MgCl 2 and then were incubated in this same solution supplemented with 1 mg/ml sulfosuccinimidobiotin for 20 min at 4°C. After incubation, cells were rinsed three times with phosphate-buffered saline-Ca/Mg containing 100 mM glycine and incubated in this buffer for 30 min at 4°C to quench the unreacted biotin. Cells were lysed with radioimmunoprecipitation assay lysis buffer with protease inhibitors. After removal of the cellular debris by centrifugation, biotinylated proteins were batch-extracted using avidin-coated Sepharose beads. After addition of SDS-PAGE loading buffer, cell lysate, biotinylated proteins (cell surface proteins), and non-biotinylated proteins (intracellular proteins) were frozen until analysis.
For immunoprecipitation of transporters, the supernatant of cell lysate was first precleared by incubation for 1 h at 4°C with agaroseconjugated rabbit serum (Sigma, St. Louis, MO). After centrifugation, the cleared lysate was incubated overnight at 4°C with anti-GLT-1 antibody at a dilution of 1:200 and precipitated by incubation with prewashed protein G-agarose (Invitrogen, Gaithersburg, MD) for 2 h at 4°C. The protein G-agarose beads were washed four times in lysis buffer and resuspended in 50 l of sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 0.1% bromphenol blue; 10% glycerol, 100 mM dithiothreitol). Samples were resolved by 10% SDS-PAGE. After drying, the gel was submitted to autoradiography. Western Blot Analyses-After resolution on 10% SDS-polyacrylamide gels, proteins were transferred to polyvinylidene fluoride membranes. After blocking in TBS-T (50 mM Tris, pH 8.0,150 mM NaCl, 0.2% Tween 20) containing 5% nonfat dry milk, membranes were incubated with anti-GLT-1 antibody (1:10,000) or anti-EAAC1 (1:75) and anti-actin antibody (1:4,000) for 1.5 h. The anti-HA antibody was used a 1:500 dilution. Following removal of the primary antibody, membranes were washed and then incubated with horseradish peroxidase-linked donkey anti-rabbit IgG (1:5,000), and proteins were visualized by enhanced chemiluminescence. For proteins labeled with 32 P, the dried gels were rehydrated in 50 mM ammonium bicarbonate for 30 min, and the proteins were transferred to a polyvinylidene fluoride membrane.

Effects of a PKC Activator on Cell Surface Expression and
Activity of GLT-1-It has recently been reported that the cell surface availability of several neurotransmitter transporters can be rapidly regulated by activation of PKC (for reviews, see Refs. 29 -31). The effects of PKC activators on the activity of GLT-1 are varied and may depend on cellular environment (12)(13)(14)(15). Therefore, we initially examined the effect of shortterm activation of PKC on the cell surface expression of GLT-1 using primary co-cultures of neurons and astrocytes derived from embryonic rat cortex. The effects of the PKC activator, PMA, on the cell surface distribution of GLT-1 were examined by Western blot analysis of biotinylated cell surface proteins. Although PMA had no effect on the total amount of GLT-1 immunoreactivity (lysate), the amount of cell surface GLT-1 decreased to ϳ70% of control ( Fig. 1, A and B). In these same cultures, PMA caused an increase in the amount of biotinylated EAAC1 immunoreactivity. 2 PMA also causes an increase in EAAC1 cell surface expression in neuron-enriched cultures (32). Because GLT-1 and EAAC1 can be preferentially expressed in different cell populations (astrocytes and neurons, respectively), either the effect of PKC activators are dependent on the presence of additional proteins that are selectively expressed in astrocytes and/or neurons or these two homologous transporters contain different specific structural domains that dictate the response to PKC activation.
To directly determine if structural differences between these transporters may explain the divergent effects of PKC activation, GLT-1 was transfected into C6 glioma, a cell line that endogenously expresses the EAAC1 subtype of transporter (24,33). In C6 glioma transiently transfected with GLT-1, a 30-min incubation with PMA caused a decrease in the cell surface expression of GLT-1, an increase in the intracellular pool of GLT-1, and no change in total GLT-1 immunoreactivity (Fig. 1, C and D). Because phorbol esters are known to interact with targets other than PKC (34), the effect of a PKC-selective antagonist, Bis II, on GLT-1 cell surface expression was examined. In these studies, Bis II had no effect in the absence of PMA but completely blocked the PMA-induced decrease in GLT-1 cell surface expression (Fig. 2).
Because transient transfection only results in expression in 20 -30% of C6 glioma (data not shown, determined using green fluorescent protein-expressing vector), the studies described above do not rule out the possibility that the differential effects For these studies, actin was visualized on a different immunoblot. The percentage of biotinylated actin in these studies was 18 Ϯ 3%, and there was no effect of PMA on the percent actin biotinylated (16 Ϯ 3%). As has been previously reported (44), we observe bands that are consistent with monomers (approximately 70 kDa) and multimers (approximately 210 kDa). B, quantitation of data from three independent experiments (mean Ϯ S.E.) that were performed in triplicate and are expressed as a percentage of the values observed in vehicle (Me 2 SO)treated cells. C, Western blot showing effect of PMA on cell surface expression of GLT-1 in C6 glioma transiently transfected with GLT-1 cDNA. C6 glioma were transfected with pcDNA3 as a control; no GLT-1 immunoreactivity was detected. D, the quantitation of ten independent experiments (mean Ϯ S.E.). No biotinylated actin was observed in these studies. *, p Ͻ 0.05 compared with vehicle; ***, p Ͻ 0.001 compared with vehicle. of a PKC activator on EAAC1 and GLT-1 occur in different populations of cells. To rule out this possibility, GLT-1 was also stably transfected into C6 glioma and several clones were isolated, providing a homogenous population of cells that express both transporters. A clone (C6-GLT-1) that expresses ϳ50% more activity than that observed in untransfected cells was chosen for further studies (V max ϭ 914 pmol/mg of protein per minute compared with 589 pmol/mg of protein per minute in control, n ϭ 2). In this cell line, PMA (100 nM for 30 min) reduced the cell surface expression of GLT-1 (to 25 Ϯ 11% of control, n ϭ 4) and increased the cell surface expression of EAAC1 in the same experiments (to 160 Ϯ 14% of control, Fig.  3, A and B). These studies provide compelling evidence that the differential effects of PMA on EAAC1 and GLT-1 cell surface expression are related to structural differences.
Effects of a PKC Activator on GLT-1-mediated Activity-In previous studies, several groups have shown that the excitatory amino acid analog, dihydrokainate (DHK), is more potent as an inhibitor of GLT-1-mediated activity (IC 50 values, 3-50 M) than as an inhibitor of EAAC1-mediated transport activity (IC 50 values, 1 to Ͼ3 mM) (Refs. 13, 35, and 36, for review, see Ref. 37). Therefore, low concentrations of DHK should selectively inhibit GLT-1-mediated activity. In initial studies, the effects of increasing concentrations of DHK on activity were examined in stably transfected and untransfected C6 glioma. In these experiments, inhibition of transport activity by DHK was consistent with a single population of sites with an IC 50 value of 1100 M (n ϭ 2). In C6-GLT-1 cells, the data for inhibition of transport activity were best fit to two sites with IC 50 values of 16 and 1300 M; 31% of the sites were of higher affinity (n ϭ 2). This percentage of high affinity sites (GLT-1mediated activity) is consistent with the higher V max value observed in these cells compared with controls (see previous paragraph).
This C6-GLT-1 clone was used to determine if PMA reduces the level of DHK-sensitive transport. The effects of PMA on total transporter activity and DHK-sensitive activity were compared in untransfected and stably transfected cell lines. In untransfected cells, PMA increased Na ϩ -dependent L-[ 3 H]-Glu transport activity, and 100 M DHK had essentially no effect on activity in both vehicle and PMA-treated cells (Fig. 3C). In C6-GLT-1 cells, PMA had no effect on total transporter activity, but PMA decreased DHK-sensitive transport activity (Fig. 3D). The reduction in DHK-sensitive transport activity, to ϳ40% of control, is in close agreement with the reduction in GLT-1 cell surface expression observed in this cell line (Fig. 3B). These data suggest that PMA decreases GLT-1-mediated activity and demonstrate that this decrease in activity is associated with a decrease in cell surface expression.
Generation and Functional Characterization of Chimeric Transporters-Because EAAC1 and GLT-1 respond differently to PMA in the same cellular milieu, differences in primary structure should account for the differential effects of PKC. To identify structural determinants required for the PKC-induced redistribution of GLT-1, a family of chimeras made by reciprocal exchange of domains between GLT-1 and EAAC1 were generated (Fig. 4). The transporters were dissected into four segments. To simplify the nomenclature, E or G was used to indicate if the amino acid sequences in each portion of a given chimera were derived from EAAC1 or GLT-1, respectively. The first segment contains the 446 (GLT-1) or 415 (EAAC1) aminoterminal amino acids, the second segment contains 28 amino acids, the third segment contains 43 amino acids, and the fourth segment contains the final 56 (GLT-1) or 37 (EAAC1) amino acids. In chimeras that lack the final carboxyl-terminal domain of GLT-1, and therefore would not be recognized by anti-GLT-1 antibodies, a hemagglutinin tag was introduced into the amino terminus to provide an epitope that could be used to follow the trafficking of these transporters. The same epitope tag was introduced into the amino terminus of wild type GLT-1 to confirm that it does not interfere with regulated trafficking. Each of these constructs was transiently transfected into C6 glioma cells to determine if they mediate Na ϩdependent L-[ 3 H]Glu transport activity. Untransfected and mock transfected C6 glioma cells were used as controls. Transient transfection of most of these chimeras resulted in significant increases in transporter activity, but four were not functional (Table I).
Evaluation of Non-functional Chimeras-Transfection of all of the chimeras that contained the first 415 amino acids of EAAC1 resulted in an increase in Na ϩ -dependent transport activity (Table I). In contrast, only three of the chimeras that contained the first 446 amino acids of GLT-1 were functional, GGGE, GEGG, and GEGE (Table I). The only common structural feature of this set of chimeras is the simultaneous presence of the first and third regions of GLT-1. Chimeras HA-GEEE, GEEG, HA-GGEE, and GGEG, which do not contain the third region of GLT-1, were not functional. This may sug-gest that subtype-specific interactions between multiple regions of GLT-1 are required for transport activity. Such interaction may be required for the proper folding of the transporter, for assembly of functional multimers, and/or possibly trafficking to the cell surface. To test this possibility, we analyzed the cell surface expression of GE chimeras in further experiments. In these studies, the percentage of the transporter on the cell surface was much lower than that observed with all of the functional constructs (Table I and Fig. 5). In addition to having a smaller percentage of transporter expressed on the cell surface, the molecular mass of the band that is consistent with a monomer was ϳ5 kDa smaller than other functional transporters (see GEGG in Fig. 5). A band of the same size was also observed in the lysate and intracellular fractions obtained from cells transfected with functional chimeras when the immunoblots were overexposed (see GEGG in Fig. 5). This smaller species was observed for all four non-functional chimeras. A similar observation has been made by others (38 -40), and it has been suggested that these proteins represent "immature" unglycosylated forms of the transporters.
Effects of PMA on Cell Surface Expression of Chimeric Transporters-With each transporter, the percentage of immunoreactivity targeted to the plasma membrane and the effect of PMA on the cell surface expression was examined using a membrane-impermeant biotinylation reagent (Table I). Initially, we confirmed that the epitope tag does not interfere with normal transporter behavior. As was observed with wild type GLT-1, PMA decreased the cell surface expression of the HAtagged GLT-1 ( Table I). The effects of PMA on HA-tagged EAAC1 were also examined. PMA increased the cell surface expression of HA-EAAC1 ( Fig. 6B and Table I). This effect of PMA, increasing cell surface expression to ϳ130% of control, was somewhat smaller than that observed in earlier studies (24) and than that observed in studies using C6 glioma stably transfected with GLT-1 (Fig. 3B). We have also examined the effects of PMA on endogenous EAAC1 cell surface expression in many of the experiments in which other chimeras were transfected into C6 glioma. In these experiments, we found that PMA increased EAAC1 cell surface expression to ϳ130% of control (see Table I). These studies suggested that transfection per se may partially block the effects of PMA on EAAC1 cell FIG. 4. Schematic diagram of wild type and chimeric transporters of GLT-1 and EAAC1. A, molecular composition of wild type and chimeric transporters. White bars correspond to GLT-1 and black bars to EAAC1. Numbers in the box indicate the numbers of amino acids in each domain. In the epitope-tagged constructs, an HA sequence YPYDVPDYA was introduced immediately after the methionine start codon. B, schematic of the proposed topological model of GLT-1 (adapted from Ref. 39). The arrows indicate the approximate locations of the three restriction sites that were used to generate the chimeric transporters. These sites occur or were introduced to provide the junctions depicted in A.
surface expression, and in preliminary studies we have found that mock transfection with pcDNA significantly diminished the effect of PMA on EAAC1 cell surface expression (data not shown).
Treatment with an activator of PKC decreased the cell surface expression of chimeric transporters that contained the third region of GLT-1, including EGGG, EEGG, HA-GGGE, HA-GEGE, GEGG, and HA-EEGE ( Fig. 6 and Table I). In contrast, chimeras that did not contain this region were unaffected by PKC activation (EEEG) or even slightly increased on the cell surface (HA-EGEE) in response to PKC activation.
Together, these studies suggest that the third region, amino acids 475-517, of GLT-1 is necessary for PMA to decrease GLT-1 cell surface expression. Importantly, PMA decreased the cell surface expression of HA-EEGE ( Fig. 6 and Table I), indicating that this domain of GLT-1 is sufficient and that additional GLT-1-specific residues are not required for the PKC-dependent redistribution of GLT-1. Because the HA-EEGE chimera is internalized in response to PKC activation, it may suggest that the residues required for EAAC1 redistribution to the plasma membrane may lie in the same 43-amino acid domain. It is unfortunate that GGEG is not functional, because TABLE I Activity of chimeric transporters and effect of PMA on cell surface expression C6 glioma were transiently transfected with different cDNAs. Na ϩ -dependent transport activity was examined and expressed as a percentage of that observed in untransfected controls; notice that mock transfection with vector (pcDNA3) had no effect on transport activity. All functional constructs significantly increased transport activity (p Ͻ 0.001 compared to mock transfected controls). The amount of biotinylated/cell surface transporter was examined in vehicle and PMA-treated cells. The amount of transporter that is expressed on the cell surface after vehicle treatment is expressed as a percentage of total transporter immunoreactivity. The effect of PMA on cell surface expression of functional constructs was also examined and expressed as a percentage of that observed in vehicle treated cells. Data are presented as the mean Ϯ S.E., and the numbers of independent observations are included in parentheses.   , ϩ), cell surface proteins were labeled and batch-extracted. Immunoblots were incubated with anti-GLT-1 and anti-actin antibodies. Note that very little of the non-functional chimeras are expressed on the cell surface. Also, the predominant band observed for the monomer forms of the non-functional chimeras was ϳ5 kDa less than that observed for a functional chimera (GEGG). A lighter exposure of this gel reveals that the multimers of these non-functional transporters are also smaller than their functional counterparts (data not shown). Similar data were obtained with the two other non-functional chimeras (see Table  I). These experiments have been repeated at least three times. it would be interesting to know if this transporter behaves like wild type EAAC1.
Effects of PMA on Cell Surface Expression and Phosphorylation of Mutant Transporters-The 43-amino acid domain that is apparently critical for PKC-dependent internalization of GLT-1 contains 5 serine and 3 threonine residues and differs from EAAC1 in 24 amino acids (see Fig. 7). Because this effect is PKC-dependent, we initially focused on mutation of serine and threonine residues as potential phosphorylation sites that may be important for trafficking. GLT-1 cDNAs containing mutations of these residues to alanine were expressed in C6 glioma cells by transient transfection and analyzed for cell surface expression in response to PKC activation. Simultaneous mutation of both unique threonine (502, 516) residues to alanine had no effect on the PMA-induced redistribution of GLT-1 (Fig. 7A). Simultaneous mutation of all five serine residues in this domain resulted in partial attenuation of the effects of PMA. Therefore, individual serine residues were mutated to alanine to determine if one or more of these 5 serine residues are required for the PMA-induced internalization. Four of the mutants responded to PMA-like wild type GLT-1, whereas the effects of PMA on the fifth mutant (S486A) were partially blunted and essentially identical to those observed with the mutant that lacked all 5 serine residues (Fig. 7). There is evidence that regulation of the cell surface expression of the GAT1 subtype of GABA transporter is dependent upon phosphorylation of tyrosine residues (41). Therefore, the single tyrosine residue that is present in this 43-amino acid domain was also mutated to alanine. This mutant also responded to PMAlike wild type GLT-1 (Fig. 7), suggesting that this tyrosine residue is not required for redistribution of GLT-1. Transient transfection of each of these mutants resulted in increased activity in C6 glioma cells.
Various structural domains that include phosphorylated serine/threonine residues are important in regulated trafficking of membrane proteins (42,43). Although serine 486 does not appear at a classic PKC consensus sequence, the effects of PMA on incorporation of 32 P into immunoprecipitable GLT-1 were examined. A consistent PMA-induced increase in 32 P incorporation into an immunoprecipitable band of ϳ70 kDa was observed in C6 glioma transiently transfected with GLT-1 but not in mock transfected cells (Fig. 8A). This provides strong evi-  Table I for the numbers of independent observations). The arrows indicate residues that were mutated. To verify that these mutants were functional, C6 glioma were transfected with these constructs and Na ϩ -dependent transport activity was measured (pcDNA3, 103 Ϯ 3 (10); GLT-1, 193 Ϯ 27 (10); all mutants increased activity by at least 67%, n Ն 3). Data are presented as a percentage of the activity observed in untransfected cells and are the mean Ϯ S.E. The numbers of observations are presented in parentheses.
dence that PKC phosphorylates GLT-1 when it is expressed in C6 glioma. Although mutation of all five serine residues in the 43-amino acid domain resulted in diminished PMA-induced incorporation of 32 P into GLT-1, mutation of the single serine residue (S486A) that was partially responsible for redistribution had no effect on the level of 32 P incorporation. As a control, each of the gels used to examine 32 P incorporation were rehydrated, and the proteins were transferred to Immobilon-P membranes and probed for GLT-1 immunoreactivity. Comparable levels of GLT-1-immunoreactivity were observed in C6 glioma transfected with the three different GLT-1 constructs, indicating that the differences in phosphorylation, or lack thereof, cannot be attributed to differences in protein expression (Fig. 8B). Together, these studies suggest that either the difference in phosphorylation of wild type and the serine 486 mutant is not detectable by this method or that phosphorylation of this residue is not required for PMA-induced internalization of GLT-1. Furthermore, these studies suggest that PMA causes incorporation of 32 P into one of the 5 serine residues in this domain, but at present it is unknown if this phosphorylation has functional effects on GLT-1.

DISCUSSION
Of the five glutamate transporters that have been identified only three, GLT-1, GLAST, and EAAC1, are found in significant quantities in the mammalian forebrain (for reviews, see Refs. 4 and 5). Several lines of evidence suggest that GLT-1 mediates the bulk of glutamate transport in this tissue. For example, immunoprecipitation of GLT-1 from rat forebrain eliminates more than 90% of the reconstitutable activity (44). The pharmacological properties of GLT-1 match those observed in forebrain synaptosomes (for review, see Ref. 37), and antisense knock down of GLT-1 suggests a major contribution of GLT-1 to total transport activity (9). Electrophysiological studies indicate that astrocytic GLT-1-mediated activity is important for limiting activation of glutamate receptors and may also contribute to synaptic plasticity (45)(46)(47). Finally, genetic deletion of GLT-1 in mice reduces cortical transport activity to less than 10% of control (10). These mice display spontaneous seizures, have altered excitatory signaling, and most die by 13 weeks of age. Together, these studies provide compelling evidence that GLT-1 is critical for the maintenance of normal glutamate homeostasis and suggest that regulation of GLT-1 may have important implications for both normal and abnormal brain function.
In the present study, we demonstrated that treatment with a PKC activator causes a decrease in GLT-1 cell surface expression in mixed cultures of neurons and astrocytes. Moreover, this PKC activator also caused a decrease in the cell surface expression of GLT-1 in transiently or stably transfected C6 glioma. In stably transfected cells, the PKC activator increased cell surface expression of EAAC1, indicating that the differential effects of PKC activation are due to differences in the primary structure rather than differences in cellular milieu. Using chimeric transporters consisting of reciprocal domains between these homologous transporters, we identified a 43amino acid region of GLT-1 that is both necessary and sufficient for a PKC-induced decrease in cell surface expression. This segment is thought to include the last transmembrane domain and part of the carboxyl-terminal intracellular tail (38,39,48). The cell surface expression of all chimeras carrying this domain was decreased in response to treatment with a PKC activator. This decrease in GLT-1 cell surface expression was associated with increased incorporation of inorganic phosphate into immunoprecipitable GLT-1. Mutation of two unique threonine residues in this region had no effect on the PKC-dependent redistribution of GLT-1. Simultaneous mutation of 5 serine residues in this portion of GLT-1 partially attenuated the PMA-induced decrease in cell surface expression and reduced the PKC-dependent phosphorylation. These data strongly suggest that GLT-1 is phosphorylated at one or more of these 5 serine residues. It is not clear if this phosphorylation alters the properties of the transporter, but it does not appear to be required for the PKC-induced redistribution of GLT-1. Although mutation of a single serine residue in this region (486) to alanine attenuated the PKC-dependent decrease in cell surface expression to the same extent as mutation of all 5 serine residues, this mutation had no discernable effect on phosphorylation of the transporter.
The cell surface expression of several different membrane proteins is controlled, presumably to provide a mechanism to regulate activity faster than would be possible by de novo protein synthesis. One of the more extensively studied examples of this type of regulation is that of G-protein-coupled receptors. The activation of many of these receptors is decreased by either homologous or heterologous desensitization. Homologous desensitization is caused by prolonged exposure to agonist followed by activation of a specific kinase (G-proteincoupled receptor kinase), whereas heterologous desensitization can be caused by activation of other kinases, including PKC. It is generally believed that both forms of desensitization are the result of direct phosphorylation of the receptor followed by internalization through an endocytic pathway (Refs. 49 and 50; for review, see Ref. 18), but there is evidence that direct receptor phosphorylation is not required using a truncated opioid receptor (51). After internalization, the receptors are dephosphorylated and either targeted for degradation or recycled to the plasma membrane (for review, see Ref. 18). A similar model of direct phosphorylation has been proposed for the CD3␥ subunit of the T cell receptor (for recent discussion, see Ref. 43).
Recent studies have shown that the activity and cell surface expression of several of the neurotransmitter transporters is decreased by PKC. Activation of PKC decreases the cell surface expression of the serotonin transporter (22), the dopamine transporter (20,21), and in some systems the GAT1 subtype of GABA transporter (19). It is possible that homologous PKC-dependent mechanisms are responsible for these effects, but it is not clear if direct transporter phosphorylation is required. There is a close temporal correlation between the time course FIG. 8. Effect of PMA on phosphorylation of wild type and mutant variants of GLT-1. C6 glioma cells were transiently transfected with pcDNA3, wild type GLT-1, mutant GLT-S486A, or GLT-5S/5A variants. After metabolic labeling with 0.25 mCi/ml [ 32 P]inorganic phosphate (H 3 PO 4 ) for 3.5 h, cells were treated with PMA (100 nM, ϩ) or vehicle (Me 2 SO, Ϫ) for 30 min. Cells were lysed, and transporters were immunoprecipitated as described under "Experimental Procedures." Samples were resolved by 10% SDS-PAGE. A, after drying, the gels were submitted to autoradiography. Note the absence of a band at ϳ70 kDa in mock (pcDNA3) transfected cells providing evidence that the band observed in transfected cells reflects GLT-1. B, this same gel was rehydrated in 50 mM ammonium bicarbonate, and the proteins were transferred to a polyvinylidene fluoride membrane. GLT-1 immunoreactivity was visualized. This experiment has been performed in three independent experiments with similar results.
for phosphorylation and internalization of the serotonin transporter (52). However, elimination of serine residues within PKC consensus sequences had no effect on PKC-dependent inhibition of serotonin transport activity, but neither phosphorylation nor internalization of the transporter were examined, so it is unclear if these studies are related (53). The dopamine transporter is also phosphorylated in response to activation of PKC. Mutation of all serine residues at PKC consensus sequences abolishes PKC-dependent phosphorylation of this transporter but has no apparent effect on PKC-dependent internalization (54). Recent studies have suggested that tyrosine phosphorylation controls the rate of GAT1 internalization (41). In the present study, mutation of a single serine residue partially abolished the PKC-dependent redistribution of GLT-1 but did not decrease phosphorylation of GLT-1. This suggests that either direct transporter phosphorylation is not required for redistribution or that the amount of phosphate incorporated is not sufficient to detect a change upon mutation of this site. Finally, we cannot rule out the possibility that serine 486 is phosphorylated and that upon mutation PKC phosphorylates the transporter at another residue that is not phosphorylated in the wild type transporter. If direct phosphorylation is not required for transporter redistribution, this would suggest that an accessory protein involved in sorting of GLT-1 may be a target of PKC.
Many membrane proteins are continuously internalized through clathrin-or caveolin-dependent endocytosis with a relatively short half-life (on the order of minutes) and recycled back to the plasma membrane through translocation from intracellular vesicles (for review, see Ref. 55). With the exception of the dopamine transporter, which appears to be internalized in response to PKC activation (20,21), very little is known about whether PKC accelerates endocytosis of these transporters or decreases translocation of the transporter back to the cell surface. Several different short amino acid peptide sequences have been identified as sorting sequences important for internalization (for review, see Ref. 55). These sequences are usually found on the amino-or carboxyl-terminal tails of the proteins and, as might be expected, are intracellular. The domain identified as being both necessary and sufficient for GLT-1 internalization is predicted in topological models to contain approximately 9 amino acids of the last (carboxyl-terminal) transmembrane domain and ϳ34 amino acids of the intracellular carboxyl-terminal tail (see Fig. 4) (39,48). Comparison of this sequence to the corresponding region in EAAC1 reveals 24 amino acid differences, and of these 24 differences 3 are conservative substitutions (Leu-483 for Val, Glu-501 for Asp, and Val-503 for Ile). The single serine residue that upon mutation to alanine attenuates the effects of PKC is 5 amino acids upstream of an isoleucine-valine. It is thought that a serine residue followed by a dileucine motif 4 or 5 amino acids downstream may serve as a consensus sequence for internalization of several membrane proteins (for recent discussion, see Ref. 56). In the present study, we have not examined the effects of mutation of the isoleucine and valine residues to determine if this abolishes internalization. Two other consensus sequences that have been identified for the clathrin pathway (Asn-Pro-X-Tyr or Tyr-polar-polar-bulky side chain) are not found in the region of GLT-1 that is required for PKC-dependent redistribution. In future studies, it will be important to determine if these effects of PKC are dependent upon accelerated endocytosis or decreased translocation to the plasma membrane and to further define the specific structural features of GLT-1 that are required for this internalization.
In summary, we show that PKC activation results in a rapid redistribution of GLT-1 from the cell surface in both primary cultures derived from rat brain and in C6 glioma cells. This effect of PKC is opposite to that observed with EAAC1. Using a series of chimeras between EAAC1 and GLT-1, we identified a domain in GLT-1 that is required for this internalization. Furthermore, we have identified a serine residue within this domain that appears to be partially responsible for this redistribution. Defining the mechanisms that regulate these transporters may have important implications for understanding both the physiological and pathological consequences of transporter regulation.