Direct Protein Kinase C-dependent Phosphorylation Regulates the Cell Surface Stability and Activity of the Potassium Chloride Cotransporter KCC2*

The potassium chloride cotransporter KCC2 plays a major role in the maintenance of transmembrane chloride potential in mature neurons; thus KCC2 activity is critical for hyperpolarizing membrane currents generated upon the activation of γ-aminobutyric acid type A and glycine (Gly) receptors that underlie fast synaptic inhibition in the adult central nervous system. However, to date an understanding of the cellular mechanism that neurons use to modulate the functional expression of KCC2 remains rudimentary. Using Escherichia coli expression coupled with in vitro kinase assays, we first established that protein kinase C (PKC) can directly phosphorylate serine 940 (Ser940) within the C-terminal cytoplasmic domain of KCC2. We further demonstrated that Ser940 is the major site for PKC-dependent phosphorylation for full-length KCC2 molecules when expressed in HEK-293 cells. Phosphorylation of Ser940 increased the cell surface stability of KCC2 in this system by decreasing its rate of internalization from the plasma membrane. Coincident phosphorylation of Ser940 increased the rate of ion transport by KCC2. It was further evident that phosphorylation of endogenous KCC2 in cultured hippocampal neurons is regulated by PKC-dependent activity. Moreover, in keeping with our recombinant studies, enhancing PKC-dependent phosphorylation increased the targeting of KCC2 to the neuronal cell surface. Our studies thus suggest that PKC-dependent phosphorylation of KCC2 may play a central role in modulating both the functional expression of this critical transporter in the brain and the strength of synaptic inhibition.

gradients (1). Adult mammalian neurons maintain low intracellular Cl Ϫ concentrations, which arise principally from the activity of the potassium chloride cotransporter-2 (KCC2). The maintenance of such low levels of intracellular Cl Ϫ ions is responsible for hyperpolarizing Cl Ϫ currents upon activation of GABA A and Gly receptors, which are responsible for fast synaptic inhibition in the adult central nervous system (2)(3)(4)(5). Molecular studies have demonstrated that KCC2 is a member of a CCC superfamily and that these transporters are composed of 12-transmembrane domains with N-and C-terminal cytoplasmic domains (2,6,7). KCC2 is expressed exclusively in neurons throughout the adult brain. Developmentally KCC2 is first detected around 10 days in vitro in cultured rat neurons, which is coincident with the emergence of hyperpolarizing GABA A receptor-mediated Cl Ϫ currents (4,8). Gene knock-out of KCC2 has revealed that ablating the expression of this protein results in early postnatal death. Neurons derived from these animals exhibit compromised GABA A receptor-mediated synaptic inhibition (9).
Under pathological conditions such as epilepsy or ischemic brain injury, deficits in the expression of KCC2 are evident together with decreased efficacy of GABAergic inhibition and with the emergence of depolarizing GABA A receptor-mediated currents that reflect decreased neuronal Cl Ϫ extrusion (10). These changes in functional expression are believed in part to be transcriptional (11,12), but post-translational modification of KCC2 is also likely to be of central importance. Intriguingly the activity of a number of protein kinases, including WNK3, WNK4, brain-type creatine kinase, TrkB receptors, tyrosine kinases, and PKC, have all been reported to influence KCC2 activity (13). However, it remains to be established whether these varying kinase activities actually directly regulate KCC2 phosphorylation and whether altered levels of phosphorylation modulate the function or membrane trafficking of this key transporter.
To further address the role of phosphorylation in regulating KCC2 we have assessed whether this protein is directly phosphorylated and whether this covalent modification alters transporter functional expression. Our studies demonstrated that KCC2 is directly phosphorylated by PKC activity on Ser 940 within the major C-terminal intracellular domain of this protein. PKC-dependent phosphorylation of Ser 940 increased KCC2 cell surface stability and activity by decreasing endocy-tosis from the plasma membrane. Endogenous KCC2 expressed in hippocampal neurons was also phosphorylated by PKC activity and, in common with our recombinant studies activation of PKC, increased the accumulation of KCC2 on the neuronal plasma membrane. Together our studies suggest a critical role for PKC-mediated phosphorylation of Ser 940 in KCC2 in regulating the functional expression of this transporter in the brain.

EXPERIMENTAL PROCEDURES
Antibodies-Monoclonal mouse anti-KCC2 antibody clone N1/12 was purchased from the University of California Davis/ NINDS/NIMH NeuroMab facility. Polyclonal rabbit anti-KCC2 antibody was purchased from Upstate.
Biotinylation and Endocytosis Assays-Cells were washed twice with 1ϫ PBS containing 0.5 mM MgCl 2 and 1 mM CaCl 2 (PBS-CM) and then incubated with 2 ml of 1ϫ PBS-CM containing 1 mg/ml sulfo-NHS-SS-biotin for 30 min for biotin labeling. After labeling, the biotin reaction was quenched by washing three times with 1ϫ PBS-CM containing 50 mM glycine and 0.1% bovine serum albumin (14,15). Cells were then lysed in 10 mM NaPO 4 , 2% Triton X-100, 0.5% deoxycholate, 10 mM sodium pyrophosphate, 25 mM NaF, 1 mM sodium orthovanadate, 100 M phenylmethylsulfonyl fluoride, 10 g of aprotinin, leupeptin, and pepstatin, 5 mM EDTA, EGTA, and 100 mM NaCl, and biotinylated proteins were purified on immobilized avidin eluted in SDS-PAGE sample buffer. KCC2 levels were then measured by immunoblotting. To measure endocytosis, cells were labeled on ice as described above and incubated at 37°C for varying time periods. The remaining cell surface biotin was then cleaved by exposure to 2 mM reduced glutathione for 10 min on ice (14). Cells were lysed, and biotinylated proteins were purified as detailed above. Data were then correct for the efficiency of glutathione S-transferase cleavage (biotin remaining at time 0), and the proportion of the total cell surface population of KCC2 internalized over time was then calculated.
Expression and Purification of KCC2 Fusion Proteins-The respective nucleotides encoding amino acids 1-102 and 645-1116 of KCC2 (3) were amplified from rat brain cDNA using the following primer pairs: AAGTCGACCATGCTCAACAACC-TGACGGACTGCGAG/AAGCGGCCGCTCAGGAGTAGA-TGGTGATGACCTCTCGGC and CAAGGATCCGATCCG-AGGCCTGTCTCTCAGTGCAGC/CAACTCGAGTCAGG-AGTAGATGGTGATGACCTCTCG, respectively. They were then cloned into pTrcHis2C (Invitrogen) to yield His-N⅐KCC2 and His-C⅐KCC2. After DNA sequencing fusion proteins were expressed in E. coli strain BL21. Exponentially growing cultures were treated with 100 M isopropyl 1-thio-␤-D-galactopyranoside for 3 h, and bacterial pellets were lysed using 6 M guanidine hydrochloride (pH 7.8) and sonication. The fusion protein was then bound to ProBond TM resin (Invitrogen) in 8 M urea (pH 7.8) for 30 min and washed with decreasing pH of buffer. Elution of fusion protein was then carried out by washing the resin with a buffer at pH 4.0. The eluted protein was dialyzed extensively in PBS and stored at Ϫ80°C before use.
Expression of KCC2 in HEK-293 Cells-Wild-type and mutant KCC2 cDNAs were cloned into the mammalian expression vector PRK5, which utilizes the human cytomegalovirus promoter for transgene expression (16). Cells were transfected using electroporation with a total of 10 g of DNA and utilized 24 -48 h after transfection (17).
Immunoblotting-Cells were lysed by lysis buffer containing 10 mM Na 2 HPO 4 , 2% Triton X-100, 0.5% deoxycholate, 10 mM sodium pyrophosphate, 25 mM NaF, 1 mM sodium orthovanadate, 100 M phenylmethylsulfonyl fluoride, 10 g of aprotinin, leupeptin, and pepstatin, 5 mM EDTA, EGTA, and 100 mM NaCl. Insoluble proteins were removed by centrifugation at 13,200 rpm for 10 min. 50 g of protein was loaded into a 6% acrylamide gel and resolved by electrophoresis by 160 V for 1 h. The resolved proteins were then electrotransferred at 50 mA onto nitrocellulose membrane over 16 h. The protein blots were blocked by 5% skim milk for 1 h. KCC2 protein was recognized by monoclonal anti-KCC2 antibody (from University of California Davis) at 1 g/ml concentration diluted in 5% skim milk. horseradish peroxidase-conjugated donkey antimouse secondary antibody was used to recognize the anti-KCC2 antibody at a concentration of 0.2 g/ml. Chemiluminescence was generated by a Visiglo TM horseradish peroxidase plus substrate kit from Amresco and detected by an LAS-3000 image reader from Fujifilm. Quantification of the chemiluminescence signal was carried out by Multi Gauge version 3.0 from Fujifilm.
Immunofluorescence-HEK-293 cells transfected with KCC2 plasmids or 4-week-old hippocampal neuron cultures were grown on 1-cm-diameter glass coverslips coated with 1 mg/ml poly-L-lysine. Cells were fixed with 4% paraformaldehyde in PBS for 15 min, washed three times with 1ϫ PBS, and blocked with 1ϫ PBS containing 5% skim milk and 0.2% Triton X-100 for 10 min. The cells were incubated with monoclonal anti-KCC2 antibody for 2 h, washed five times with 1ϫ PBS, incubated with TRITC-conjugated polyclonal anti-mouse antibody in blocking buffer for 1 h, washed five times with 1ϫ PBS, and mounted on glass slides with 3 l of Vectashield mounting medium. The prepared slides were imaged with a confocal microscope after 24 h. Acquisition of confocal images was carried out using Laser Sharp 2000 software from Bio-Rad. Quantification of fluorescence images was carried out using Meta-Morph software from Universal Imaging Corp.
In Vitro Kinase Assay-0.5 g of fusion protein was incubated with 10 Ci of [␥-32 P]ATP and 1-50 ng of purified PKC or cAMP-dependent protein kinase (PKA) from Calbiochem in a buffer containing 20 mM HEPES (pH 7.5), 10 mM MgCl 2 , and 50 M ATP for 10 min at 30°C and terminated by the addition of 2ϫ SDS-PAGE sample buffer (18,19). The reaction mixture was then analyzed by SDS-PAGE. A phosphorimaging device was used to quantify the incorporation of 32 P into the proteins.
Neuronal Cultures-In brief, rat embryos at embryonic day 18 were removed and decapitated into 1ϫ Hanks' balanced salt solution (HBSS; from Invitrogen) on ice. Brain tissues from the embryos were removed and transferred into fresh HBSS on ice, and hippocampal regions were further dissected out. Dissected hippocampi were placed in 0.25% trypsin at 37°C for 15 min with gentle shaking. Hippocampi were washed with HBSS two times for 5 min and passed through Pasteur pipettes 10 times to dissociate. Nondissociated debris was allowed to settle to the bottom for 10 min. Dissociated hippocampal neurons were counted by hemocytometer and plated on a 60-mm culture dish at a density of 1 million/dish in attachment medium containing 10% fetal bovine serum, 1 mM sodium pyruvate, and 25 mM glucose in minimum Eagle's medium (Invitrogen). 4 h after plating, the attachment medium was replaced with warm maintenance medium containing 2% B-27 neural supplement, 2% fetal bovine serum, and 0.5 mM glutamine in Neurobasal medium (Invitrogen). Hippocampal cultures were kept in an incubator conditioned at 37°C and 5% CO 2 . 0.5 ml of maintenance medium was added into the culture dish every 3 days to replenish the loss of medium from evaporation.
Peptide Mapping and Phosphoamino Acid Analysis-Gel slices excised from SDS-polyacrylamide gels and washed and digested with 0.1 mg/ml trypsin were subjected to two-dimensional mapping and visualized by autoradiography. The resulting phosphopeptides were also hydrolyzed using 6 N HCl, and the resulting phosphoamino acids along with phosphoamino acid standards were then separated by thin layer chromatography and visualized by autoradiography as detailed previously (18,19).
Whole-cell Metabolic Labeling and Immunoprecipitation-HEK-293 cells or 28 -35 days in vitro hippocampal neurons in 60-mm dishes were labeled with 0.5-1.0 mCi/ml [ 32 P]orthophosphoric acid for 4 h in phosphate-free media or with 200 mCi/ml [ 35 S]methionine (16). Cell were then lysed in lysis buffer containing 10 mM NaPO 4 , 5 mM EDTA and EGTA, 100 mM NaCl, 10 mM sodium pyrophosphate, 25 mM NaF, 2% Triton X-100, 0.5% deoxycholate, 1 mM sodium orthovanadate, 100 M phenylmethylsulfonyl fluoride, and 10 g of aprotinin, leupeptin, and pepstatin (16). The supernatant was collected after centrifugation. 2 g of antibody and 40 l of protein A-Sepharose (1:1 slurry) were then added to the supernatant and incubated for 2 h at 4°C with constant agitation. The protein A-Sepharose was then washed extensively in lysis buffer supplemented with 500 mM NaCl and analyzed by SDS-PAGE.

In Vitro Analysis of KCC2 Phosphorylation by PKC-Molec-
ular studies have demonstrated that the consensus site for phosphorylation by a number of classical second messengerdependent protein kinases, including both PKC and PKA, are evident within both the major intracellular domains of KCC2 (3). To commence our studies we expressed the major N-terminal (amino acids 1-102; His-N⅐KCC2) and C-terminal (amino acids 645-1116; His-C⅐KCC2) intracellular cytoplasmic domains of KCC2 as His-tagged fusion proteins in E. coli (Fig.  1A). Purified fusion proteins were then subjected to in vitro kinase assays. This revealed that the His-C⅐KCC2 fusion protein was phosphorylated by purified PKC to a final stoichiometry of ϳ0.6 mol/mol, whereas His-N⅐KCC2 was not phosphorylated under similar conditions (Fig. 1B). In contrast, neither of the fusion proteins was phosphorylated by purified PKA, but the cytoplasmic tail of the GABA B R2 subunit, a previously identified of substrate of this enzyme (20), was phosphorylated under the same conditions (Fig. 1B). Peptide mapping and FIGURE 1. Ser 940 is the major site of PKC phosphorylation within KCC2. A, the major intracellular N-and C-terminal domains of KCC2 were cloned into pTrcHis2C vector to generate His-tagged fusion proteins His-N⅐KCC2 and His-C⅐KCC2, respectively. B, His-N⅐KCC2 and His-C⅐KCC2 fusion proteins and the intracellular domain of the GABA B R 2 subunit (CR2) fused to glutathione S-transferase were subjected to in vitro kinase assays in the presence of [␥-32 P]ATP followed by SDS-PAGE. Phosphorylation was then visualized using a phosphorimaging device (the upper panel represents 32 P, and the lower panel shows Coomassie (Com) staining of the same gel). C, 32 P-labeled fusion proteins were subjected to phosphoamino acid analysis followed by autoradiography. The migration of phosphoserine (pS), -threonine (pT), and -tyrosine (pY) standards is indicated. D, 32 P-labeled fusion proteins and 32 P-KCC2 immunopurified from neurons were digested with trypsin, and the resulting peptides were applied to TLC plates and subjected to electrophoresis followed by ascending chromatography as indicated. The small arrows indicate the origins and the letters the major phosphopeptides A-C, respectively. E, wild-type (WT) and mutant fusion proteins were phosphorylated in vitro by purified PKC in the presence of [␥-32 P]ATP and subjected to SDS-PAGE. Phosphorylation was then visualized using a phosphorimaging device (the upper panel represents 32 P, and the lower panel shows Coomassie staining of the same gel). The level of phosphorylation for each mutant was then compared with that seen for control wild type, which was assigned a value of 100%. *, significantly different from control (p Ͻ 0.01; Student's t test, n ϭ 3).
phosphoamino acid analysis revealed that His-C⅐KCC2 was primarily phosphorylated on serine residues within two major phosphopeptides labeled A and B, respectively (Fig. 1, C and D).
To further analyze KCC2 phosphorylation, site-directed mutagenesis was utilized to convert candidate serine residues within His-C⅐KCC2 to alanines. Based on the consensus for PKC phosphorylation of (R/K)X (1-4) (S/T)X (1-3) (R/K) where X is any amino acid (21,22), mutant fusion proteins were produced in which Ser 728 , Ser 940 , and Ser 1034 were individually and sequentially mutated to alanines. The phosphorylation of the resulting purified proteins by PKC was then compared with that seen for His-C⅐KCC2. Although mutation of Ser 728 and Ser 1034 did not significantly alter phosphorylation, mutation of Ser 940 reduced phosphorylation to 32.5 Ϯ 2.5% of control. Moreover mutation of Ser 940 in combination with either Ser 728 or Ser 1034 had very similar effects on PKC-dependent phosphorylation compared with mutation of Ser 940 alone (Fig. 1E).
The phosphorylation of the 32 P-His-C S940A /KCC2 was further analyzed using peptide mapping and phosphoamino acid analysis. Significantly, this revealed that mutation of Ser 940 ablated peptide A seen on PKC-dependent phosphorylation of His-C⅐KCC2 (Fig. 1D) and that the remaining sites of phosphorylation for this kinase in His-C S940A /KCC2 are serine residues (Fig. 1C). Together these results suggest that the major site for PKC phosphorylation within His-C⅐KCC2 is Ser 940 .
Ser 940 Is a Major Site for PKC-dependent Phosphorylation of KCC2 when Expressed in HEK-293 Cells-To analyze the relevance of our in vitro observations, we transiently expressed KCC2 in HEK-293 cells. KCC2 expression was first analyzed by immunoprecipitation with anti-KCC2 antibodies after metabolic labeling with [ 35 S]methionine. This resulted in the isolation of two bands with approximate molecular masses of 125 and 130 kDa from cells expressing wild-type KCC2 but not from cells expressing empty vector. Similar bands were also immunoprecipitated from cells expressing a mutant form of KCC2 in which Ser 940 (KCC2 S940A ) was mutated to an alanine ( Fig. 2A). The phosphorylation of KCC2 was examined by immunoprecipitation after metabolic labeling with [ 32 P]orthophosphoric acid. A major band of 130 kDa was evident from cells expressing wild-type KCC2 that was not seen in those expressing vector alone, demonstrating basal phosphorylation of this protein in HEK-293 cells. Activation of PKC with phorbol dibutyrate (PDBu) for 10 min produced a significant increase (p Ͻ 0.01) in KCC2 phosphorylation of 195.7 Ϯ 5.6% of that evident under basal conditions (Fig. 2B). Phosphorylation of KCC2 S940A was analyzed using similar methodology, and this mutant construct exhibited robust levels of basal phosphorylation. However in contrast to wild-type KCC2, PDBu did not significantly enhance the phosphorylation of KCC2 S940A (Fig. 2B). These results are consistent with our in vitro experiments and strongly suggest that the major site for PKC phosphorylation within the C-terminal intracellular domain of this transporter is Ser 940 .
Phosphorylation of Ser 940 Enhances KCC2 Cell Surface Expression Levels-To address the functional consequences of KCC2 phosphorylation, we first compared the proportion of KCC2 and KCC2 S940A expressed on the cell surface of HEK-293 cells using biotinylation (14,15). This revealed that 22.9 Ϯ 4.5% of wild-type KCC2 was present on the cell surface at steady state, and this could be significantly increased to 40.6 Ϯ 5.4% upon activation of PKC over a 10-min time period, but the total levels of KCC2 expression were not altered under the same conditions (Fig. 2C). In contrast to this 38.7 Ϯ 4.6% of KCC2 S940A was present on the plasma membrane, a level significantly higher (p Ͻ 0.01) than that of wild-type KCC2 (22.9 Ϯ 4.5%; Fig. 2C). However, PDBu treatment did not significantly increase the cell surface expression level of KCC2 S940A (Fig. 2C).
To confirm the results of our biotinylation assays, we assessed the effects of activating PKC on cell surface expression levels of KCC2 using immunohistochemistry. To do so cells expressing KCC2 or KCC2 S940A were treated with PDBu, permeabilized, and stained with anti-KCC2 antibodies. Confocal images were then recorded from these cells, and the pixel intensity across the entire cell was measured in the presence and absence of PBDU treatment (Fig. 3A). The relative level of membrane expression was determined by calculating the ratio of fluorescence signal associated with the cell periphery and the cytoplasm. Notably treatment of cells expressing wild-type KCC2 with PBDU significantly increased the ratio fluorescence associated with the cell periphery by 182 Ϯ 7.2% of control untreated cells (Fig. 3, B and C). Similar experiments were performed on cells expressing KCC2 S940A ; however, in these cells PDBu did not significantly alter the distribution of KCC2 immunoreactivity (Fig. 3, A-C). Together these biochemical and imaging experiments indicate that phosphorylation of KCC2 on Ser 940 increases transporter cell surface expression levels. The level of phosphorylation for each construct was then normalized to that seen under basal conditions (ϪPDBu ϭ 100%). C, cells expressing KCC2 constructs were treated with (ϩ) and without (Ϫ) 100 nM PDBu and then subjected to biotinylation. The purified cell surface (CS) and total fractions (T) were then immunoblotted with KCC2 antibodies as illustrated in the upper panel. The proportion of KCC2 expressed on the cell surface was then determined for each construct, and data were normalized to those seen under basal conditions (ϪPDBu ϭ 100%) *, significantly different from control (p Ͻ 0.01; Student's t test, n ϭ 3-4).

PKC-dependent Phosphorylation of Ser 940 Decreases KCC2
Endocytosis-To further evaluate the mechanism underlying PKC-dependent modulation of KCC2 cell surface stability, the possible role of this enzyme in regulating transporter endocytosis was evaluated. To do so transfected HEK-293 cells were labeled with NHS-SS-biotin and incubated at 37°C for up to 20 min in the presence of leupeptin to prevent lysozomal degradation of any internalized protein (14). After cleavage of remaining cell surface NHS-SS-biotin with reduced glutathione, cells were lysed, and internalized biotinylated proteins were isolated on avidin and immunoblotted with KCC2 antibodies. After controlling for the efficiency of cleavage (remaining biotin at 0 min) this approach revealed that the entire cell surface population of KCC2 was endocytosed within 10 min under basal conditions. This process was linear over the initial 5-min period of the assay (Fig. 4A). Under control conditions 80.7 Ϯ 8.2% of the total cell surface population of KCC2 was internalized within 5 min, whereas in the presence of PDBu internalization was significantly reduced (p Ͻ 0.01) to 30.5 Ϯ 4.5% (Fig. 4B).
We used the KCC2 S940A mutant to assess the role of Ser 940 in mediating the effects of PKC on KCC2 endocytosis. Compared with wild-type KCC2 the endocytosis of this mutant appeared to be significantly decreased over a time course of 20 min in the presence or absence of PKC activation (Fig. 4A). To quantify these results, we compared the level of endocytosis of KCC2 and KCC2 S940A after a 5-min incubation at 37°C in the pres-  ence and absence of PDBu. Significantly lower levels (p Ͻ 0.01) of KCC2 S940A endocytosis were evident compared with wildtype KCC2 under these conditions (15.6 Ϯ 3.2 versus 80.7 Ϯ 8.2%, respectively; Fig. 4B). However, in contrast to KCC2, PDBu treatment did not significantly alter the endocytosis of KCC2 S940A (Fig. 4B). These results strongly suggest that the phosphorylation of Ser 940 acts to modulate cell surface stability of KCC2 by slowing its endocytosis.
The Activity of KCC2 Is Increased by PKC Activation in Cultured Cells-To measure the functional effects of PKC activity on KCC2 function, expressing cells were treated with bumetanide and ouabain to inhibit endogenous Na ϩ -K ϩ -2Cl Ϫ cotransporter and Na ϩ -K ϩ -ATPase, respectively. KCC2 activity was measured by basal furosemide-sensitive 86 Rb ϩ influx. In cells expressing KCC2, significant levels of 86 Rb ϩ accumulation were evident over a 3-5-min time course that were not seen with cells expressing empty vector (Fig. 5). Influx was significantly enhanced (p Ͻ 0.01) by a 15-min preincubation with N-ethylmaleimide (NEM), an accepted activator of KCC2 (2) (Fig. 5). Pretreatment of cells with phorbol 12-myristate-13acetate (PMA) produced a large and highly significant increase (p Ͻ 0.01) in KCC2 activity as measured by furosemide-sensitive 86 Rb ϩ influx (Fig. 5). To test the role that direct phosphorylation of KCC2 plays in this modulation, the effects of PMA on a number of KCC2 mutants were analyzed. Significantly, the ability of PKC to modulate the activity of KCC2 was blocked by mutation of Ser 940 , but this treatment did not alter the ability of NEM to stimulate transporter activity (Fig. 5). In contrast mutation of Ser 728 or Ser 1034 in KCC2 was without effect on either PKC or NEM-dependent stimulation of KCC2 activity, but interestingly mutation of Ser 728 appeared to increase constitutive activity of KCC2 (Fig. 5). Together this series of experiments revealed that activation of KCC2 by PKC was dependent upon Ser 940 , the major site of phosphorylation for this kinase within this protein.
PKC Activity Modulates the Phosphorylation and Cell Surface Stability of Endogenous KCC2 in Cultured Neurons-To examine the relevance of our recombinant studies, we examined the phosphorylation of KCC2 in hippocampal neurons. To initiate these experiments, we first examined the expression of KCC2 in these cells via immunoblotting. A major band of ϳ130 kDa was seen with anti-KCC2 antibodies that was first detected at 2 weeks and reached maximal expression levels between 4 -5 weeks in culture (Fig. 6A) consistent with other studies on the developmental expression of KCC2 in culture (4,(23)(24)(25). To examine phosphorylation, 4 -5-week cultures were labeled with [ 32 P]orthophosphoric acid and subjected to immunoprecipitation with KCC2 antibody. Under control conditions a major phosphoprotein of 130 kDa was seen immunoprecipitating with KCC2 antibodies but not with control IgG, demonstrating that KCC2 is basally phosphorylated in hippocampal neurons (Fig. 6B). Treatment of neurons with PDBu produced a highly significant increase (p Ͻ 0.01) in KCC2 phosphorylation (p Ͻ 0.01) to 295.5 Ϯ 22.3% of control, an effect reduced by co-application of the PKC inhibitor calphostin (Cal). In addition, inhibition of PKC alone significantly reduced (p Ͻ 0.01) the basal level of KCC2 phosphorylation to 15.5 Ϯ 3.5% of control (Fig. 6B).
KCC2 phosphorylation in this system was further evaluated using peptide mapping and phosphoamino acid analysis. Under basal conditions KCC2 was principally phosphorylated on serine and threonine residues, and that activation of PKC also induced tyrosine phosphorylation (Fig. 6C). Intriguingly peptide map analysis of 32 P-KCC2 from neurons after activation of PKC revealed the presence of two phosphopeptides (A and B in Fig. 1D), which showed very similar migration to those evident FIGURE 5. PKC-dependent phosphorylation of Ser 940 modulates KCC2 activity. K ϩ -Cl Ϫ cotransport activity was monitored 48 h after transfection by measuring the furosemide-sensitive 86 Rb ϩ influx in cells expressing a range of KCC2 constructs or vector (VT). Influx of 86 Rb ϩ was linear for 10 min in all KCC2 constructs, and 3-5-min influxes were routinely performed. HEK-293 cells were pretreated with 2 M bumetanide prior to the influx assay to inhibit endogenous Na ϩ -K ϩ -2Cl Ϫ cotransport activity and 100 M ouabain to inhibit endogenous Na ϩ -K ϩ -ATPase. The level of 86 Rb ϩ influx is indicated under basal conditions (Ⅺ) or in the presence of 100 nM PMA (f) or 50 M NEM (u) over a time course of 3-5 min. *, significantly different from control (p Ͻ 0.01; Student's t test, n ϭ 3). on mapping PKC-phosphorylation 32 P-His-C⅐KCC2 together with a neutral peptide (C in Fig. 1D). Together these results suggest that Ser 940 is likely to represent a site of PKC phosphorylation in neuronal KCC2. We also examined the effects of PKC activation on the cell surface stability of KCC2 using biotinylation. This revealed that activation of PKC with PDBu produced a highly significant increase (p Ͻ 0.01) in the proportion of KCC2 expressed on the neuronal cell surface (CS) to 295.6 Ϯ 9.8% of control, an effect that was blocked by a specific PKC inhibitor (Fig. 6D). To confirm our biotinylation experiments, we measured the effects of PKC activation on the targeting of KCC2 to the plasma membrane using immunohistochemistry followed by confocal microscopy (Fig. 7A). The relative level of membrane expression was determined by measuring the ratio of fluorescence signals associated with the cell surface and the cytoplasm of individual proximal dendrites (Fig. 7B). Treatment of neurons with PDBu significantly increased (p Ͻ 0.01) the level of KCC2 immunoreactivity at the periphery of dendrites compared with control untreated neurons, an effect that was decreased via inhibition of PKC (Fig. 7C). Therefore in common with our recombinant studies these observations demonstrated that KCC2 in its native environment is subjected to PKC-dependent phosphorylation and that this covalent modification increases the targeting of KCC2 to the neuronal cell surface.

DISCUSSION
The potassium chloride cotransporter KCC2 is the major determinant of Cl Ϫ transmembrane gradients in adult neurons. The activity of KCC2 results in low intracellular Cl Ϫ concentrations that are responsible for hyperpolarizing responses of GABA A and Gly receptors in fast synaptic inhibition within the central nervous system (4,8). Deficits in KCC2 activity are of importance in epilepsy and other central nervous system pathologies (26); therefore comprehending the cellular mechanisms neurons use to regulate the activity of this protein is of significance.
Here we have begun to examine the molecular sites of phosphorylation for individual protein kinases within KCC2 and the role that these residues play in regulating its functional expression. There are a number of studies that have analyzed the regulation of KCC2 by agents that modify the activity of protein kinases and phosphatases (6, 20 -27); however, it remains to be demonstrated that KCC2 is actually phosphorylated in neurons. We commenced our analysis by expressing the major N-(residues 1-102) and C-terminal intracellular domains (residues 645-1116) of KCC2 as fusion proteins in E. coli and analyzed their phosphorylation in vitro. This revealed that the C-terminal but not the N-terminal fusion protein was selectively and stochiometrically phosphorylated by PKC. Peptide mapping and phosphoamino acid analysis revealed that this phosphorylation occurred on serine residues within two major phosphopeptides. Site-specific mutagenesis of Ser 940 with the C-terminal fusion reduced PKC phosphorylation to ϳ25% of control and ablated one of the major peptides seen on phosphorylation of the wild-type protein. Consistent with our in vitro analysis, full-length KCC2 exhibited high levels of basal phosphorylation when expressed in HEK-293 cells as measured via immunoprecipitation after metabolic labeling with [ 32 P]orthophosphoric acid. Moreover, phosphorylation of KCC2 was robustly increased upon activation of PKC, an effect that was ablated by mutation of Ser 940 . Therefore our combined in vitro and in vivo analysis demonstrated that KCC2 is a substrate of PKC and that the major site of phosphorylation within the major intracellular domain of this transporter is Ser 940 .
In addition to our recombinant experiments, we directly analyzed the phosphorylation of KCC2 in hippocampal neurons using immunoprecipitation. This revealed that under basal conditions in 28 -35 days in vitro hippocampal neurons, KCC2 was basally phosphorylated on serine and threonine residues. Activation of PKC produced a dramatic increase in the stoichiometry of its phosphorylation to ϳ400% of that seen under basal conditions. Enhanced phosphorylation upon the activation of PKC occurred on serine, threonine, and tyrosine residues. Consistent with our result in hippocampal neurons, we found that KCC2 is phosphorylated on serine/threonine residues in cortical neurons (not shown). In addition, oxidative stress and/or high metabolic activity have recently been reported to induce phosphorylation of KCC2 on tyrosine residues (27). To directly compare our recombinant and neuronal studies of KCC2 phosphorylation, we used peptide mapping. This revealed that KCC2 in hippocampal neurons was phosphorylated within three major tryptic peptides after activation of PKC. Significantly, two of these peptides were also evident on tryptic peptide mapping of the C terminus of KCC2 after PKC phosphorylation in vitro. These results together with our recombinant studies suggest that Ser 940 is phosphorylated in neuronal KCC2 molecules upon the activation of PKC.
The effects of PKC activity on KCC2 functional expression were evaluated using both biochemical and functional FIGURE 7. The subcellular distribution of KCC2 in hippocampal neurons is modulated by PKC activity. A, 4-week-old cultured hippocampal neurons were treated with (ϩ) or without (Ϫ) 100 nM PDBu and/or 10 mM Cal for 10 min. Neurons were fixed, permeabilized, and stained with anti-KCC2 antibody coupled to a rhodamine-conjugated secondary. Confocal images were then acquired from the bottom to the top of proximal dendrites. In the enlargements (insets) the region used for quantification are indicate by the lines, and the numbers 1 and 2 represent the points at which data acquisition commenced and terminated, respectively. B, the pixel intensity per unit length (0.1 m) was then calculated for cells treated with PDBu (gray line) and PDBu/ Cal (broken black line) and for control cells (solid black line). C, the ratio of plasma membrane to cytosol signals was calculated, and values were then compared with levels seen under control conditions (ϪPDBu; 100%). *, significantly different from control (p Ͻ 0.01; Student's t test, n ϭ 10 cells from three differing transfections). approaches. As measured by biotinylation, activation of PKC increased the cell surface expression levels of KCC2 by ϳ200% relative to control over a 10-min time period. This effect of PKC on KCC2 steady state cell surface expression levels was critically dependent on Ser 940 , as mutation of this residue abrogated the effect of activating PKC upon transporter cell surface stability. Likewise immunohistochemistry revealed that activation of PKC increased the targeting of PKC to the periphery of HEK-293 cells, an effect that was also dependent upon Ser 940 . To analyze the mechanism underlying the effects of Ser 940 phosphorylation on cell surface stability, the effects of PKC activity on KCC2 transporter endocytosis were measured. Under basal conditions the entire cell surface population of KCC2 was internalized over a time course of 10 min. Activation of PKC dramatically slowed the endocytosis of KCC2 compared with control, an effect that was critically dependent on Ser 940 , consistent with the higher steady state levels expression of this mutant. Previous experiments using Xenopus oocytes have illustrated that the activation of PKC decreases cell surface expression levels and activity of KCC2 (6). The varying effects of PMA may result from differing trafficking itineraries between oocytes and HEK-293 cells or the varying incubation temperatures of 18 and 37°C, respectively. However, and in keeping with our results, the effects of PMA in oocytes was dependent on Ser 940 , suggesting a critical role for this residue in regulating KCC2 functional expression via PKC activity. As Ser 940 is not conserved in other CCCs (3), these results suggest a unique role for this residue in regulating KCC2 function.
The effects of PKC activity on the cell surface stability of KCC2 on the neuronal plasma membrane were examined. Using both biotinylation and imaging, it was evident that activation of PKC in hippocampal neurons that had been cultured for 4 -5 weeks produced a large and highly significant increase in the cell surface expression levels of KCC2 within 10 min. Whether these effects were mediated via altered endocytosis remains to be ascertained; however, it is interesting to note that studies in hippocampal neurons have previously shown that under basal conditions 50% of cell surface KCC2 is degraded within 20 min. Enhancing PKC-dependent phosphorylation may thus provide a rapid and dynamic mechanism to enhance the cell surface and activity of KCC2. In keeping with this potential mechanism, Fiumelli et al. (28) have demonstrated that E Cl shifts caused by changing [Ca 2ϩ ] i are dependent on PKC activity. However if these effects are mediated by changes in the stochiometry of KCC2, phosphorylation has not been demonstrated.
Our studies have shown that KCC2 is phosphorylated directly by PKC on Ser 940 within the cytoplasmic C-terminal domain of this critical transporter and that phosphorylation of this residue acts to increase KCC2 functional expression by slowing its endocytosis. Therefore, cell signaling molecules that activate PKC signaling pathways may have profound effects on neuronal Cl Ϫ homeostasis by regulating the phosphorylation and functional expression of KCC2. Given the critical role KCC2 plays in regulating Cl Ϫ homeostasis, this phospho-dependent functional modulation may have significant consequences for the efficacy of synaptic inhibition mediated by GABA A and Gly receptors.