Protein Kinase C-mediated Phosphorylation of the BGT1 Epithelial γ-Aminobutyric Acid Transporter Regulates Its Association with LIN7 PDZ Proteins

The Na/Cl-dependent BGT1 transporter has osmoprotective functions by importing the small osmolyte betaine into the cytosol of renal medullary epithelial cells. We have demonstrated previously that the surface localization of the transporter in Madin-Darby canine kidney cells depends on its association with the LIN7 PDZ protein through a PDZ target sequence in the last 5 residues of the transporter (-KETHL). Here we describe a protein kinase C (PKC)-mediated mechanism regulating the association between BGT1 and LIN7. Reduced transport activity paralleled by the intracellular relocalization of the transporter was observed in response to the PKC activation induced by 12-O-tetradecanoylphorbol-13-acetate (TPA) treatment. This activation caused clathrin-dependent internalization of the transporter and its targeting to a recycling compartment that contains the truncated transporter lacking the LIN7 binding motif (BGTΔ5) but not the LIN7 partner of BGT1. The decreased association between BGT1 and LIN7 was demonstrated further by coimmunoprecipitation studies and in vitro binding to recombinant LIN7 fusion protein. The TPA treatment induced phosphorylation of surface BGT1 on serine and threonine residues. However, a greater increase in phosphothreonines than phosphoserines was measured in the wild type transporter, whereas the opposite was true in the BGTSer mutant in which a serine replaced the threonine 612 in the LIN7 association motif (-KESHL). No similar increase in relative phosphoserines or phosphothreonines was found in the BGTΔ5 transporter. Moreover, phosphorylation of threonine 612 in a BGT COOH-terminal peptide impaired its association with recombinant LIN7. Taken together, these data demonstrate that the post-translational regulation of BGT1 surface density is a result of transporter phosphorylation and that threonine 612 is an essential residue in this PKC-mediated regulation.

The Na/Cl-dependent BGT1 transporter has osmoprotective functions by importing the small osmolyte betaine into the cytosol of renal medullary epithelial cells. We have demonstrated previously that the surface localization of the transporter in Madin-Darby canine kidney cells depends on its association with the LIN7 PDZ protein through a PDZ target sequence in the last 5 residues of the transporter (-KETHL). Here we describe a protein kinase C (PKC)-mediated mechanism regulating the association between BGT1 and LIN7. Reduced transport activity paralleled by the intracellular relocalization of the transporter was observed in response to the PKC activation induced by 12-O-tetradecanoylphorbol-13-acetate (TPA) treatment. This activation caused clathrindependent internalization of the transporter and its targeting to a recycling compartment that contains the truncated transporter lacking the LIN7 binding motif (BGT⌬5) but not the LIN7 partner of BGT1. The decreased association between BGT1 and LIN7 was demonstrated further by coimmunoprecipitation studies and in vitro binding to recombinant LIN7 fusion protein.
The TPA treatment induced phosphorylation of surface BGT1 on serine and threonine residues. However, a greater increase in phosphothreonines than phosphoserines was measured in the wild type transporter, whereas the opposite was true in the BGTSer mutant in which a serine replaced the threonine 612 in the LIN7 association motif (-KESHL). No similar increase in relative phosphoserines or phosphothreonines was found in the BGT⌬5 transporter. Moreover, phosphorylation of threonine 612 in a BGT COOH-terminal peptide impaired its association with recombinant LIN7. Taken together, these data demonstrate that the post-translational regulation of BGT1 surface density is a result of transporter phosphorylation and that threonine 612 is an essential residue in this PKC-mediated regulation.
Renal medulla is normally the only hypertonic tissue in mammals, and its hypertonicity is fundamental to the ability of the kidney to concentrate urine. The BGT1 member of the renal ␥-aminobutyric acid (GABA) 1 transporter family is a betaine GABA transporter localized on the basolateral surface of the inner medulla, where it mediates the uptake and intracellular accumulation of the small betaine osmolyte, thus protecting the cells from hypertonicity. The transporter was cloned from Madin-Darby canine kidney (MDCK) cells (1); BGT1 transport activity is low in MDCK cells maintained in isotonic medium but greatly induced in response to prolonged hypertonic stress. Transcription plays a major role in this stimulation, leading to increased BGT1 mRNA levels and consequently increased transport activity (1,2). However, it has been suggested that the post-transcriptional regulation of BGT1 protein also plays a role in cell adaptation to hypertonic stress (3).
The single PDZ domain of LIN7 is a type I PDZ domain that selectively binds ligands such as the BGT1 COOH terminus (-ETHL). The association of BGT1 and LIN7 is not required for targeting BGT1 to the basolateral surface of MDCK cells but, it is necessary to prevent its internalization, thus mediating transporter accumulation on the lateral junctional surface (5). This need for the association with LIN7 to prevent internalization may provide the basis for a regulatory mechanism involving the rapid modulation of surface BGT1 expression.
Transporter phosphorylation has been linked to regulatory protein trafficking, and it has been shown that various members of the sodium-dependent neurotransmitter transporter family (including GABA transporters) are acutely regulated in response to the activation of serine and threonine kinases (13)(14)(15)(16). A number of consensus sites for these are located in the intracellular domains of BGT1, but no data are available concerning their phosphorylation. However, it has been shown that the phosphorylation of residues in the COOH-terminal sequences binding PDZ domains (17,18), including serine at position Ϫ2 of the PDZ binding sites for type I PDZ domains (19), and the phosphorylation of a serine residue in a PDZ domain (20) abolish the association of target proteins with their PDZ partners. The phosphorylation-mediated regulation of these interactions controls the surface distribution of AMPA and ␤-adrenergic receptors (17,21).
Because the interaction of BGT1 with the PDZ domain of LIN7 is a key factor determining its surface density, we investigated whether BGT1 phosphorylation events dynamically regulate BGT1-LIN7 interactions, thus affecting transporter activity and distribution. Analysis of the effects of activation of protein kinase A (PKA) and protein kinase C (PKC) on BGT1 showed that PKC (but not PKA) phosphorylates BGT1 and disrupts its interaction with LIN7, thus causing a clathrinmediated relocalization of the transporter to an intracellular recycling compartment and a parallel decrease in its activity. Our findings demonstrate a novel mechanism for the regulation of transporter activity based on the PKC-mediated modulation of PDZ protein interactions with target sequences.

Plasmid Construction
The cDNAs of canine BGT1 carrying a single base mutation that produces a protein with serine instead of threonine in position 612 (BGTSer) were obtained by QuikChange site mutagenesis kit (Stratagene) using pCB6-mycBGT as template (4), with the forward primer 5Ј-GTTGGGGAGAAGGAGAGCCACTTGTAGGATGTGGCCAGC-3Ј carrying a single base substitution (G instead of C, underlined in the sequence) and its complementary sequence as the reverse primer. The fragment corresponding to nucleotides 1886 -1947 of the BGT cDNA was digested with KpnI-ClaI and subcloned to similarly digested pCB6-mycBGT. The sequence of the resulting clones was confirmed by DNA sequencing (PRIMM).

Cell Cultures and Transfections
MDCK-BGTSer cell lines were obtained by calcium phosphate transfection and G418 selection (4). MDCK strain II cells, or MDCK-BGTSer, MDCK-mycBGT1 (4), and MDCK-mycBGT⌬5 (5) cell lines, were cultured as described previously (4). They were plated at a density of 2.5 ϫ 10 4 cells/cm 2 on glass coverslips for immunocytochemistry, Transwell filter inserts (0.4-m pores; Costar) for GABA uptake experiments, or tissue culture dishes (Falcon) for biochemistry, and cultured for 60 h to reach confluence before treatment. In the case of transient transfections, the MDCK cells were cotransfected by the calcium phosphatemediated method 24 h after plating using cDNA coding for untagged BGT1 and myc-tagged BGT⌬5; the MDCK-BGT1 cell lines were transfected 48 h after plating with cDNA coding for wild type HA-tagged dynamin1 or K44A (a gift from Dr. Schmid) (22).

Cell Treatment
After incubation for 30 min at 37°C in Dulbecco's modified Eagle's medium (DMEM) containing 25 mM HEPES, the cells were treated in the same medium with 1 M TPA (tumor promoter 12-O-tetradecanoylphorbol-13-acetate) (Sigma) for 30 min to stimulate PKC, or 10 M forskolin (FSK) (Calbiochem) to activate PKA, or 2 M GF109203X (Tocris) for 20 min before TPA treatment as a means of specifically preventing PKC activation. To inhibit clathrin-mediated internalization, the cells were pretreated for 30 min in hypertonic medium consisting of DMEM/HEPES supplemented with 0.45 M sucrose. Clathrinindependent endocytosis was blocked by maintaining the cells for 2 h at 37°C in DMEM supplemented with 200 M genistein. After treatment, the cells were processed for immunofluorescence or biochemical studies as indicated.

Primary Antibodies
Rabbit polyclonal antibodies were raised against a synthetic peptide consisting of amino acids 598 -614 of dog BGT1; coupling and immunization were performed as described previously (23). To obtain the BGT17 and KETHL antibodies, the serum was affinity purified on columns of synthetic peptides consisting, respectively, of the last COOH-terminal 17 and 5 amino acids of BGT1 immobilized on CNBr-activated Sepharose 4B (Amersham Biosciences) as described previously (24).
To optimize the yield in the immunoprecipitation experiments, the BGT17 antibody was used to recognize the wild type transporter, whereas the BGT-KLH polyclonal antibody raised against an internal peptide (5) was used to immunoprecipitate mutant transporters. Qualitatively similar results were obtained when wild type and mutant transporters were immunoprecipitated with the 9E10 monoclonal antibody directed against the c-myc epitope (MBL, Santa Cruz), but the low efficiency prevented its use in these experiments. In the immunofluorescence experiments the wild type transporter was revealed by BGT17 or anti-myc antibodies, and the mutant transporters were localized using BGT-KLH or anti-myc antibodies. In the coexpression experiments, the purified rabbit polyclonal KETHL and the anti-myc antibodies were used, respectively, to distinguish BGT1 and the myc-epitope tagged BGT⌬5. LIN7 PDZ protein was immunorevealed by a rabbit polyclonal antiserum raised against amino acids 44 -207 of mouse LIN7 fused to a sequence of 6 histidines (histidine-LIN7 fusion protein) (12). The following antibodies were also used: monoclonal anti-␤-catenin (Transduction Laboratories), anti-GST (Santa Cruz), and anti-HA epitope (Roche Applied Science), polyclonal antibodies against the bovine mannose 6-phosphate receptor as a late endosomal marker (a gift from Dr. Hoflack) (25), and human catepsin D as a lysosomal marker (a gift from Dr. Isidoro) (26).

Immunocytochemistry
The cells were fixed in 4% paraformaldehyde or ice-cold methanol and permeabilized with 0.5% Triton X-100. Immunostaining with primary antibodies was followed by incubation with rhodamine-conjugated anti-rabbit and fluorescein isothiocyanate (FITC)-conjugated antimouse IgG (Jackson Immunoresearch).
For wheat germ agglutinin (WGA) labeling, the MDCK-mycBGT1 clones were incubated on ice for 1 h with FITC-WGA (Sigma) in phosphate-buffered saline ϩ 0.1 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ ϩ 1% bovine serum albumin 60 h after plating. The cells were washed to remove any unbound labeling lectin and then cultured for 30 min in DMEM/HEPES supplemented with 1 M TPA before fixation in 4% paraformaldehyde and staining for BGT1.
The confocal images were obtained using a Bio-Rad MRC-1024 confocal microscope.

Immunoprecipitation and Western Blot Analysis
The MDCK cell lines were washed twice with cold phosphate-buffered saline and solubilized in lysis buffer (25 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM MgCl 2 , 1% Triton X-100, 5 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, and a mixture of protease inhibitors with or without a mixture of phosphatase inhibitors). After centrifugation at 12,000 ϫ g for 20 min, the lysates were incubated overnight at 4°C with antibodies preincubated for 2 h at 4°C with 20 l of protein A-Sepharose. The immunocomplexes bound to the beads were recovered by centrifugation, washed three times with lysis buffer, and loaded onto an 11% SDS-polyacrylamide gel. The gels were blotted onto nitrocellulose membrane at 120 mA overnight, and the blots were probed for BGT1 and LIN7 using the primary antibodies described above, with protein A conjugated to peroxidase as the secondary reagent, and visualized by means of ECL (PerkinElmer Life Sciences).

Affinity Chromatography Assays
Binding of Recombinant GST-LIN7 to Immobilized Peptides-The expression of mLIN7 fused to GST and inserted into pGEX-1 vector (5) was induced in Escherichia coli DH5␣. Bacterial lysates containing ϳ2 g of GST-LIN7 were incubated with ϳ20 g of peptides immobilized on CNBr-activated Sepharose 4B (Amersham Biosciences). In competition experiments, 60 g of free peptides was preincubated with the bacterial lysate for 2 h at room temperature before overnight incubation with the Sepharose-bound peptides, and the mixture was washed four times at 4°C using 1.5 ml of phosphate-buffered saline containing 0.5% Triton X-100 (50% of the detergent concentration in the bacterial lysate). The proteins bound to the Sepharose beads were solubilized, loaded onto a 10% SDS-polyacrylamide gel, and immunoblotted.
Binding of MDCK Cell Line Lysates to Immobilized GST-LIN7 Fusion Protein-The expression of mLIN7 fused to GST was induced in E. coli DH5␣. MDCK-mycBGT1 cell lines treated or not with TPA were solubilized in lysis buffer supplemented with phosphatase inhibitors and incubated overnight with ϳ4 g of the GST-LIN7 fusion protein previously immobilized on glutathione-Sepharose 4B (Amersham Biosciences). The bound transporter was resolved by 10% SDS-PAGE, transferred to a nitrocellulose membrane, and immunostained using the polyclonal BGT17.

Phosphorylation Experiments
Metabolic Labeling-Metabolic labeling with 32 P i was carried out by preincubating cells in phosphate-free medium for 3 h (27) and then incubating them for 30 min at 37°C with phosphate-free medium containing 32 P i (final specific activity 0.15-0.25 mCi/ml). The labeled cells were treated as described above, washed in the presence of phosphatase inhibitors, and lysed as described above. The transporters were immunoprecipitated, and the immunocomplexes were loaded onto 10% SDS-PAGE. The gels were dried and autoradiographed. The signal intensity was quantified densitometrically using NIH Image 1.59 software, and statistical significance was determined by means of Student's t test.
Phosphoamino Acid Analysis-Immunoprecipitated phosphorylated transporters were electrophoresed on 10% gels at 50 mV overnight, and after drying and autoradiography, the region of the gel containing the transporter was excised, and the protein was eluted overnight at 37°C in 50 mM ammonium bicarbonate, pH 8, with 50 g of Pronase (Sigma). After being evaporated by a vacuum concentrator (SpeedVac), the samples were hydrolyzed in 0.2 ml of 6 N HCl at 110°C for 2 h and evaporated again. The hydrolyzed samples were dissolved in a pH 1.9 buffer (7.8% acetic acid, 2.5% formic acid) with a mixture of cold phosphoamino acid standards (phosphoserine, phosphothreonine, and phosphotyrosine, 25 nM each), spotted onto cellulose thin layer plates (Schleicher & Schuell), and separated by two-dimensional electrophoresis: 750 V for 2 h with pH 1.9 buffer in the first dimension, and 500 V for 1.5 h at pH 3.5 (0.5% pyridine and 5% acetic acid) in the second dimension. The standards were visualized with ninhydrin, and the plates were autoradiographed at Ϫ70°C in a cassette fitted with an intensifying screen (Kodak).
Analysis and Quantification of Phosphoamino Acid-To visualize the TPA-mediated relative increase in phosphothreonines compared with phosphoserines in the transporters having an identical amount of serines (BGT1 and BGT⌬5) and the relative increase in phosphoserines compared with phosphothreonines in the transporters having identical threonines (BGT⌬5 and BGTSer), autoradiograms with comparable phosphoserines (BGT1 and BGT⌬5) or phosphothreonines (BGT⌬5 and BGTSer) were quantitated using the NIH Image program. The results were expressed as the -fold increase in phosphothreonines or phosphoserines in response to TPA normalized to the corresponding increase in phosphoserines or phosphothreonines (i.e. the -fold increase in phosphothreonines in BGT1 ϭ the TPA-induced increase in phosphothreonines/the TPA-induced increase in phosphoserines).

GABA Influx Assay
GABA influx was assayed according to Yamauchi et al. (2), with modifications (23). Briefly 10 ϫ 10 3 cells were plated on 6.5-mm Transwell filters, grown for 60 h, and treated as described above. [ 3 H]GABA (PerkinElmer Life Sciences) in incubation buffer was applied to the apical and basolateral sides at a final concentration of 10 M for 10 min at 37°C. Uptake was terminated by aspirating the medium, and the cells were washed three times with ice-cold incubation buffer. After removing the filters from the supports, the cells were solubilized in 0.2 ml of 1% SDS, and the samples were counted using a ␤-counter in 5 ml of scintillation solution (Ultima Gold, PerkinElmer Life Sciences).

PKC Activation Decreases BGT1 Activity and Relocalizes the Transporter in Endocytotic
Compartments-To verify whether serine/threonine kinase proteins regulate BGT1, we measured its activity in response to the activation of PKC and PKA in a MDCK-mycBGT1 clone by evaluating [ 3 H]GABA uptake, because BGT1 transports GABA with even higher affinity than betaine. An ϳ60% reduction in GABA influx was observed after TPA treatment, thus suggesting a PKC-mediated mechanism (Fig. 1A). The dramatic effect of PKC activation on GABA influx was dose-dependent and reached a peak when the cells were treated for 30 min with 1 M TPA (Fig. 1B). This effect was largely but not completely prevented by the PKC specific inhibitor GF109203X because a reduction of ϳ25-30% was still measured in the presence of the inhibitor (Fig. 1A). The reduction in GABA transport activity in response to TPA treatment was accompanied by the intracellular relocalization of BGT1 (Fig. 1C, compare the lateral surface staining in a with the intracellular staining in b), which appeared to be completely prevented by the PKC inhibitor GF109203X (c). A 25-30% reduction in transport activity unrelated to the intracellular relocalization of the transporter was also measured after FSK treatment (Fig. 1, A and C, d), thus suggesting that alternative mechanisms independent of transporter relocalization are responsible for the inhibition of transport activity after FSK and TPA.
To establish whether BGT1 is internalized in response to TPA treatment, we first performed an endocytosis morphological assay, in which glycoproteins were labeled with FITC-WGA at 0°C, chased at 37°C for 30 min in the presence of 1 M TPA, and then fixed and stained for BGT1. As shown in Fig. 2A, many of the intracellular structures containing BGT1 also con-tained internalized WGA-labeled glycoproteins, thus indicating that BGT1 is in endocytotic compartments.
To identify the internalization pathway followed by BGT1, we analyzed the effect of endocytosis inhibition on transporter distribution. It has been reported widely that 0.45 M hypertonic sucrose treatment specifically inhibits endocytosis through clathrin-coated pits (28,29), whereas 200 M genistein inhibits clathrin-independent endocytosis (30). We found that sucrose prevented the TPA-mediated internalization of wild type BGT1, as revealed by the shift in the distribution of BGT1 from intracellular compartments to the cell surface (Fig. 2B). Genistein treatment did not prevent the TPA-induced relocalization of BGT1, thus indicating a clathrin-dependent mechanism underlying the TPA-mediated regulation of BGT1 distribution.
We also used a dominant negative mutant of dynamin1 (K44A) to block clathrin-mediated and dynamin1-dependent internalization (22). MDCK-mycBGT1 cell lines were transiently transfected with wild type dynamin1 or the mutant K44A. The latter blocked the TPA-mediated internalization of BGT1, thus indicating that the process requires the functional budding of clathrin-coated endocytotic vesicles, but, when the cells were transfected with wild type dynamin1, TPA still in- duced the translocation of BGT1 to endosomes (Fig. 2C). A truncated transporter lacking the PDZ association motif (BGT⌬5) constitutively recycles from the intracellular endocytotic compartments to the plasma membrane, but immunofluorescence staining detects the transporters in the intracellular compartment where it accumulates (Fig. 2B) (5). The truncated transporter staining appeared to be unchanged after TPA stimulation, but the sucrose-induced inhibition of clathrin-dependent endocytosis led to its surface accumulation. Taken together, these data document a clathrin-dependent internalization of wild type and truncated BGT1.
TPA Treatment Targets Internalized Transporter to the BGT⌬5-containing Recycling Compartment-We next investigated the identity of the intracellular compartments containing BGT1 by first examining whether endogenous MDCK late endosomal and lysosomal markers colocalized with BGT1. After 30-min incubation with TPA, virtually no BGT1 colocalization with the mannose 6-phosphate receptor late endosomal marker or the cathepsin D lysosomal marker was detected by the specific antibodies (Fig. 3A). To verify whether BGT1 colocalizes with BGT⌬5 in the perinuclear recycling compartment of MDCK cells after TPA treatment, wild type BGT1 was coexpressed with myc-tagged BGT⌬5. A rabbit polyclonal KETHL antibody raised against the last 5 amino acids of BGT1 was used to distinguish the wild type transporter from the truncated form, and the latter was detected using an anti-myc monoclonal antibody. Colocalization of BGT1 and BGT⌬5 in the intracellular compartment was clearly revealed only after TPA treatment (Fig. 3B). Taken together, these data indicate that PKC activation leads to the internalization and targeting of surface BGT1 into a recycling compartment containing the transporter lacking the LIN7 binding site.
TPA Treatment Disrupts the Association between BGT1 and LIN7-The results described above suggest that PKC activation causes BGT1 to behave in the same manner as the truncated BGT⌬5 mutant, which is incapable of interacting with LIN7, a hypothesis that was explored by first analyzing the distribution of LIN7 by means of immunofluorescence staining. Under normal conditions, the majority of BGT1 colocalized with LIN7 on the lateral surface of MDCK cells, but in response to TPA stimulation, the transporter accumulated in LIN7-free intracellular structures, thus further suggesting that the bind-ing of BGT1 to LIN7 is affected by TPA treatment (Fig. 4A). Staining with structural markers of the surface junctional domain excluded a generalized effect of PKC activation on the relocalization of these proteins (see ␤-catenin staining in Fig. 4B).
The effect of TPA on the association of the transporter and LIN7 was shown further by immunoprecipitation experiments using BGT1 antibodies; LIN7 was detected in the immunoprecipitates obtained under normal conditions but not after TPA treatment (Fig. 4C). The different localization of BGT1 and LIN7 in TPA-treated cells and the results of the coimmunoprecipitation experiments indicate a decreased association between the two proteins. To establish whether changes in the transporter were responsible for this decreased affinity, we performed affinity chromatography experiments. The TPA treatment of MDCK-mycBGT1 cells caused an ϳ50% reduction in BGT1 recovery from the GST-LIN7-conjugated affinity column (Fig. 4D), thus demonstrating a direct effect of PKC on the ability of BGT1 to bind LIN7 and suggesting that the transporter is phosphorylated by the kinase.
Phosphorylation of BGT1 by PKC-To test whether PKC activation specifically mediates BGT1 phosphorylation in intact cells, MDCK cells overexpressing myc-tagged BGT1 were metabolically labeled with 32 P i and then stimulated with TPA to activate PKC or FSK to activate PKA. The activation of PKC led to a prominent band of phosphorylated wild type transporter (a 10-fold increase compared with control), whereas PKA activation or preincubation with the specific PKC inhibitor GF109203X before TPA treatment did not induce significant levels of phosphorylated BGT1 (Fig. 5A). Interestingly, we found a higher basal level of phosphorylated transporter lacking the LIN7 binding motif (BGT⌬5) than wild type BGT1 and a smaller increase after PKC stimulation (Fig. 5B).
The different phosphorylation behavior of the truncated BGT1 might have been caused by its intracellular localization. We therefore wondered whether the internalization and relocalization of BGT1 is triggered by the direct phosphorylation of the transporter or whether BGT1 phosphorylation is subsequent to its internalization. When transporter internalization was inhibited by hyperosmotic sucrose, TPA treatment caused an ϳ2and 3-fold increase in the phosphorylation of, respectively, the truncated and wild type BGT1 compared with TPA Having demonstrated that PKC mediates the phosphorylation of BGT1 before its internalization, we investigated whether threonine 612, an essential residue for BGT1 binding to LIN7, is phosphorylated in response to TPA treatment. Because position Ϫ2 in the binding site for type I PDZ proteins can be a threonine or a serine residue, we tested whether a BGTSer peptide (containing serine instead of threonine at position 612) retained its binding site for LIN7. A smaller amount of GST-LIN7 protein was retained by BGTSer than BGT peptides similarly conjugated to Sepharose columns, and free BGT better competed for the GST-LIN7 binding to the BGTSer column than the BGTSer peptides (Fig. 6A). The BGTSer mutant therefore apparently had a lower affinity for LIN7 in vitro. We generated a cell line stably expressing the cDNA encoding a mutant transporter in which threonine 612 was replaced by a serine (MDCK-BGTSer), and its distribution behavior before and after TPA treatment matched that of the wild type BGT1 (Fig. 6B). The mutant transporter localized on the plasma membrane before treatment and relocated intracellularly in response to TPA. Associated with the lower affinity of BGTSer for LIN7, the basal level of phosphorylated mutant transporter was higher than in wild type BGT1, and it increased 2-fold in response to TPA (Fig. 6C).
To identify the phosphoamino acid residue(s) of BGT1 involved in its PKC-mediated internalization, we performed simultaneous phosphoamino acid analyses of wild type and mutants BGT1 immunoisolated from basal and TPA-stimulated cells and found that BGT1 phosphorylation occurs on serine moieties, but PKC activation induced phosphorylation on threonines, accounting for ϳ15% of the phosphoserine signal (Fig. 7A). The relative increase of phosphothreonines in BGT1 and the decreased binding affinity of BGT1 for LIN7 suggest that an essential residue such as threonine 612 in the PDZ target motif is phosphorylated and that this causes the transporter internalization. When threonine 612 was replaced by serine (BGTSer), we found a greater increase in phosphoserines than phosphothreonines in response to TPA, as expected in the case of phosphorylation of serine 612 by a serine/ threonine kinase. A truncated transporter lacking the 612 site (BGT⌬5) showed similar increases in phosphoserines and phos- phothreonines. No phosphotyrosine was found under any conditions. When these data were expressed as a -fold increase after TPA treatment, a large relative reduction in phosphothreonines was revealed in the mutants lacking threonine 612, whereas the presence of serine at position 612 caused a relative increase in phosphoserines in the BGTSer (Fig. 7B). Taken together, these data indicate that TPA mediates the phosphorylation of threonine or serine in position 612 and suggest that the phosphorylated residue plays a primary role in regulating the transporter-LIN7 association.
To assess directly whether phosphorylation of threonine at position Ϫ2 affects the ability of BGT1 to interact with LIN7, we analyzed the in vitro binding of LIN7 to threonine 612phosphorylated or unphosphorylated BGT COOH-terminal peptides (Fig. 7C). Bacterial cell lysates expressing LIN7 fused to GST were incubated with BGT peptide-conjugated affinity columns, and the bound proteins were eluted and analyzed by immunoblotting. When the BGT COOH-terminal peptide was phosphorylated at threonine 612, GST-LIN7 no longer interacted with it. Moreover, free nonphosphorylated, but not phosphorylated, BGT peptide competed for the GST-LIN7 binding to the column. Phosphorylation of the transporter is therefore induced by the action of PKC on both serine and threonine residues, but the phosphorylation of threonine 612 is sufficient to abolish the in vitro binding of BGT1 to LIN7. DISCUSSION The expression of the BGT1 epithelial GABA transporter is strongly induced by hypertonic stress (1, 2). The up-regulation of BGT1 protein and transport activity is largely controlled by transcriptional mechanisms (31,32), but it has been suggested that a post-transcriptional mechanism relying on the insertion of preexisting BGT1 into the plasma membrane plays a role during cell adaptation to hypertonic stress (3). We have shown previously a mechanism regulating surface transporter density based on a PDZ-mediated association with LIN7, a PDZ protein that forms a heterotrimeric complex with the PDZ domaincontaining proteins LIN2 (CASK) and LIN10 (Mint, X11); LIN7 binding retains BGT1 on the basolateral surface of MDCK cells, whereas a truncated mutant BGT1 lacking the COOHterminal PDZ-interacting motif is internalized and targeted to a recycling compartment (5).
The results presented here provide evidence of the PKCdependent modulation of BGT1 activity through the regulation of BGT1-LIN7 binding. The effects of PKC are mediated by a mechanism involving the phosphorylation of BGT1 which increases clathrin-and dynamin1-dependent BGT1 internalization and targets the internalized transporter to a recycling compartment. Our findings therefore demonstrate a post-transcriptional mechanism for the regulation of transporter activity based on the PKC-mediated modulation of LIN7 PDZ protein interactions.
In particular, we here show that phosphorylation of a target carrier alters its interaction with a PDZ domain-containing protein, thus destabilizing the transporter from the plasma membrane and causing an increased rate of endocytosis. A first indication that the transporter internalizes in response to TPA was obtained by means of a morphological assay for endocytosis, which showed the colocalization of BGT1 with internalized WGA-labeled surface glycoproteins, and further confirmation of the PKC-mediated internalization of BGT1 in endocytic compartments was given by the results of endocytosis-inhibiting treatment. Because sucrose and a dynamin1 mutant (but not genistein) blocked the TPAinduced intracellular relocalization of BGT1, we conclude that the transporter is internalized in response to TPA via a clathrin-and dynamin1-dependent pathway.
Our data indicate that BGT1 is not targeted to degradation after TPA-mediated internalization because total transporter levels are maintained (Fig. 4, C and D), and there is no colocalization of BGT1 with degradative pathway markers (Fig.  3A). On the contrary, we did find good colocalization of BGT1 with the truncated transporter. We have demonstrated previously BGT⌬5 enrichment in transferrin-positive recycling endosomes under basal conditions (5), and here we show the translocation of wild type BGT1 to BGT⌬5-positive recycling endosomes during PKC stimulation, thus suggesting that unbound wild type BGT1 constitutively recycles to the cell surface. The intracellular accumulation of BGT1 after PKC activation may be caused by a combination of accelerated internalization and reduced recycling, as has been documented in the case of the DAT dopamine transporter (33,34), but the accumulation of BGT⌬5 in the recycling compartment regardless of TPA treatment suggests that this is an intrinsic prop-erty of unbound transporters rather than an indirect effect of TPA on both internalization and recycling pathways.
Because the members of the sodium-dependent family of transporters have multiple consensus sites for phosphorylation in their amino acid sequences, it has been speculated that their activity may be regulated by direct phosphorylation. The regulation of BGT1 activity by PKC and PKA activators has been documented previously, and predicted consensus sites for PKC and PKA phosphorylation have been identified in the amino acid sequence of BGT1 (14,16), but no data were available concerning in vivo BGT1 phosphorylation or the amino acid residues involved in this post-translational modification. We found that in vivo BGT1 phosphorylation was specifically mediated by PKC because PKA activation with FSK, or TPAinduced PKC activation in the presence of its specific inhibitor, had no significant effect on transporter phosphorylation. An alternative explanation is that the phosphorylation was below the level of detection in these experiments. However, decreased transport activity unaccompanied by intracellular relocalization and transporter phosphorylation was measured after FSK and TPA treatment, as indicated by the 20% reduction in transport measured in the presence of the PKC inhibitor, which suggests that additional mechanisms may be responsible for impaired transport activity after PKA and PKC stimulation. We here document a TPA-mediated mechanism responsible for were prelabeled with 0.15 mCi/ml 32 P i and then treated with 1 M TPA at 37°C for 30 min (ϩTPA). The transporters were immunoprecipitated using the affinity-purified antibodies described in "Experimental Procedures," and their phosphorylation was analyzed by SDS-PAGE followed by autoradiography. The 97-kDa molecular mass standard is indicated on the left. Quantitation of the experiment is expressed as -fold values over basal BGT1 phosphorylation. transport phosphorylation and decreased transporter surface density which accounts for at least half of the 60% decrease in transport activity measured in response to TPA.
Under basal conditions, phosphoserines were detected but not phosphothreonines, whereas after PKC activation, phosphothreonines were clearly observed together with increased phosphoserines in the wild type transporter and accounted for 15% of the phosphoserines signal. Comparison of the TPAinduced increase in phosphoamino acid in the wild type and mutant transporters showed a considerable increase in phosphothreonines in the wild type transporter and a clear increase in phosphoserines in BGTSer, whereas no similar increase in phosphoserines or phosphothreonines was observed in the truncated transporter. These results can be expected in the case of the phosphorylation of the residues in position 612 (whether it is a serine as in BGTSer or a threonine as in BGT1) and in the absence of the 612 phosphorylation site.
Our data do not show whether PKC directly phosphorylates the transporter, but they do indicate the involvement of a serine/threonine kinase in transporter phosphorylation. Moreover, the basal phosphorylation of BGTSer serines and threonines is insufficient to induce internalization and targeting to the recycling compartment because no intracellular staining was ever observed in the MDCK-BGTSer cell line. BGTSer relocalizes intracellularly in response to TPA, and this translocation is associated with the greater increase in phosphoserines in BGTSer than in BGT⌬5 (which lacks the 612 serine residue) or in BGT1, which has a threonine in position 612. Taken together, these data indicate that TPA stimulates the phosphorylation of serines and threonines in the transporter and that phosphorylation of threonine 612 in the wild type BGT1 in response to TPA is necessary to internalize and target the transporter to the recycling compartment. In line with the involvement of the phosphorylation of this residue in the dissociation of BGT1 from LIN7, the phosphorylation of threonine 612 in the BGT peptide completely abolished its binding to recombinant LIN7.
The reduced association of BGT1 with LIN7 in response to PKC-mediated phosphorylation events has been widely documented: (i) the intracellular relocalization of the transporters in compartments not containing LIN7; (ii) the absence of LIN7 in immunoprecipitates obtained using the BGT1 antibody; and (iii) the clathrin-dependent internalization of BGT1 and its localization in a recycling compartment, like its counterpart BGT⌬5 lacking the LIN7 binding motif. In addition, the reduced ability of immobilized bacterial GST-LIN7 to retain BGT1 from a FIG. 7. The threonine 612 residue of BGT1 is phosphorylated after PKC activation. A, phosphoamino acid analysis of the wild type and mutant transporters by two-dimensional separation. MDCK cells stably expressing myc-BGT1, myc-BGT⌬5, or BGTSer (cell line 7) were prelabeled with 0.25 mCi/ml 32 P i and then treated with 1 M TPA at 37°C for 30 min (ϩTPA). The labeled transporters were immunoprecipitated, resolved by gel electrophoresis, and autoradiographed. The bands corresponding to the transporters were excised from the gels and submitted to Pronase digestion followed by acid hydrolysis. The hydrolysates were mixed with cold phosphoamino acid standards and separated by twodimensional thin layer electrophoresis. The plates were autoradiographed, and the phosphoamino acid standards visualized by means of ninhydrin staining (dotted circles). Two experiments were carried for BGT1 and one experiment for each mutant transporter; the autoradiograms correspond to different exposure times to show comparable intensity of the phosphoserine spots between BGT1 and BGT⌬5, and the phosphothreonines spots between BGT⌬5 and BGTSer. B, the results of A were quantitated using the NIH Image program, with the results for each transporter being expressed as the -fold increase in relative phosphothreonines (P-Thr) or phosphoserines (P-Ser) (TPA-mediated -fold increase in phosphothreonines in relation to the increase in phosphoserines, or vice versa). C, phosphorylation of BGT1 on threonine 612 affects the BGT1-LIN7 interaction. The asterisk indicates phosphorylation on threonine 612 in the amino acid sequence of the BGT synthetic peptide immobilized on CNBr-activated Sepharose 4B and used in the interaction assays. A lysate from bacteria expressing the GST-LIN7 fusion protein was incubated with the indicated immobilized peptides. Bound GST-LIN7 was detected using an anti-GST monoclonal antibody (IB:GST). 10% of the bacterial lysate used for the assays (Lys) was used as control. The 65-kDa molecular mass standard indicates the retained fusion protein; the lower bands correspond to the degradation products of GST-LIN7. Recombinant GST-LIN7 was efficiently retained by the BGT synthetic peptide (Seph-BGT), but not by the BGT-P peptide containing phosphorylated threonine 612 (Seph-BGT-P). Free unphosphorylated BGT peptide, but not phosphorylated BGT-P peptide, competed for the GST-LIN7 binding to the column. lysate of MDCK-mycBGT1 cells treated with TPA clearly demonstrates that BGT1 phosphorylation is a necessary and sufficient modification to decrease its binding affinity to LIN7.
The basal phosphorylation of the transporters not bound to LIN7 (BGT⌬5) or having a lower binding affinity for it (BGT-Ser) seems to be more efficient than that of the bound wild type BGT1. These data suggest an equilibrium between bound and unbound transporter which is shifted toward the unbound state by PKC-mediated phosphorylation (probably on the unbound form and especially on residue 612), thus targeting transporter internalization.
Evidence supporting the PKC-mediated regulation of proteinprotein associations has been described in relation to the GAT1 neuronal GABA transporter, in which PKC regulates the interactions between GAT1 and syntaxin1A (35). In this case, other syntaxin1A-binding partners rather than the transporter are substrates for the PKC-induced modifications regulating the availability of syntaxin1A to interact with GAT1.
PKC-mediated regulation of PDZ domain-mediated interactions by means of the direct in vivo phosphorylation of PDZ ligands has been documented in the AMPA receptor GluR2 subunit involved in synaptic plasticity (17), in which phosphorylation of a serine in the GluR2 binding site decreases the affinity of the subunit to the PDZ-containing domain GRIP (glutamate receptor-interacting protein) but not PICK1 (protein interacting with C kinase). It has also been shown that the PICK1 PDZ protein interacts directly with the DAT dopamine transporter (36), and this appears to stabilize the transporter on the cell surface, similarly to LIN7 in the case of BGT1. Although PKC regulation of the DAT-PICK1 interaction has been proposed, no precise data are yet available. It is interesting to note that, as in the case of BGT1, PKC activation increases the clathrin-mediated endocytosis of DAT and its targeting to recycling pathways (33).
The sodium-dependent family of plasma membrane transporters (including GAT and DAT) bind other proteins, regulate their activity in response to PKC activation, and change their distribution between the plasma membrane and recycling compartments (15,33,37). However, no single mechanism linking these events has been documented. Here we demonstrate that in one member of this family (the epithelial GABA transporter BGT1), the mechanism by means of which PKC regulates the transport activity is transporter phosphorylation, which affects its association with the LIN7 protein and thus causes its intracellular redistribution to a recycling compartment. Regulation by means of changing cell surface expression may account for the rapid and efficient fine tuning of neurotransmission in the case of neurotransmitter transporters and (in the case of BGT1) cell volume regulation during antidiuresis or the acute onset of diuresis, which are processes that require, respectively, the intracellular accumulation or transient loss of osmolytes.
In conclusion, we here document a novel mechanism of BGT1 transport activity regulation based on TPA-mediated phosphorylation of surface BGT1 which causes the transporter internalization and targeting to a recycling compartment. Moreover, phosphorylation of a threonine or a serine residue essential for BGT1-LIN7 association is a key event in the regulation of BGT1 transport activity.