A Novel Regulatory Locus of Phosphorylation in the C Terminus of the Potassium Chloride Cotransporter KCC2 That Interferes with N-Ethylmaleimide or Staurosporine-mediated Activation*♦

Background: KCC2 is a potassium-chloride cotransporter essential for hyperpolarizing neurotransmission and is associated with multiple neurological disorders. Results: Thr934 and Ser937 are major regulatory sites of KCC2 activity, and their status influences other activation processes. Conclusion: Thr934 and Ser937 phosphorylation increases KCC2 transport kinetics. Significance: This study identifies a novel C-terminal KCC2 stimulatory phosphorylation site. The neuron-specific cation chloride cotransporter KCC2 plays a crucial role in hyperpolarizing synaptic inhibition. Transporter dysfunction is associated with various neurological disorders, raising interest in regulatory mechanisms. Phosphorylation has been identified as a key regulatory process. Here, we retrieved experimentally observed phosphorylation sites of KCC2 from public databases and report on the systematic analysis of six phosphorylated serines, Ser25, Ser26, Ser937, Ser1022, Ser1025, and Ser1026. Alanine or aspartate substitutions of these residues were analyzed in HEK-293 cells. All mutants were expressed in a pattern similar to wild-type KCC2 (KCC2WT). Tl+ flux measurements demonstrated unchanged transport activity for Ser25, Ser26, Ser1022, Ser1025, and Ser1026 mutants. In contrast, KCC2S937D, mimicking phosphorylation, resulted in a significant up-regulation of transport activity. Aspartate substitution of Thr934, a neighboring putative phosphorylation site, resulted in a comparable increase in KCC2 transport activity. Both KCC2T934D and KCC2S937D mutants were inhibited by the kinase inhibitor staurosporine and by N-ethylmaleimide, whereas KCC2WT, KCC2T934A, and KCC2S937A were activated. The inverse staurosporine effect on aspartate versus alanine substitutions reveals a cross-talk between different phosphorylation sites of KCC2. Immunoblot and cell surface labeling experiments detected no alterations in total abundance or surface expression of KCC2T934D and KCC2S937D compared with KCC2WT. These data reveal kinetic regulation of transport activity by these residues. In summary, our data identify a novel key regulatory phosphorylation site of KCC2 and a functional interaction between different conformation-changing post-translational modifications. The action of pharmacological agents aimed to modulate KCC2 activity for therapeutic benefit might therefore be highly context-specific.

These severe consequences of altered KCC2 function raised interest in mechanisms regulating its activity (5). On the cellular level, location in membrane rafts was shown to modify KCC2 transport activity (23,24). On the molecular level, interaction partners such as the ATPase subunit ␣2 (25), CIP1 (26), Neto2 (27), and several protein kinases, including brain-specific creatine kinase (28), SPAK (29), OSR1 (30), and WNKs (31) were identified. The regulatory role of phosphorylation is in line with previous pharmacological studies. The kinase inhibitors lavendustin A, genistein (32), or staurosporine (33) altered KCC2 transport activity in cultured hippocampal neurons, and calyculin (34) and okadaic acid (35), two potent phosphatase inhibitors, blocked activation of KCCs by cell swelling. The important role of phosphorylation was corroborated by the regulatory role of several identified phosphorylation sites in KCC2 (36). Phosphorylation of human threonines Thr 906 and Thr 1007 reduced the intrinsic transport activity of KCC2 (37,38). In contrast, phosphorylated Ser 940 increased surface expression and KCC2 transport function (39) as well as membrane clustering (40). Finally, the phosphorylation status of Tyr 903 and Tyr 1087 also regulates KCC2 activity, although the functional consequences of phosphorylation appear to be context-dependent and require further investigation (5,23,36,41,42).
Here, we wished to further examine the role of phosphorylation for KCC2 activity. Mining of large scale phospho-proteomics data revealed several phosphorylated serines not yet analyzed. To characterize their importance, these amino acids were systematically substituted by either alanine or aspartate in the rat KCC2b isoform to mimic the dephosphorylated state (alanine) or the phosphorylated state (aspartate), respectively. Expression and transport activity of the mutants were then determined in HEK-293 cells. This approach identified Ser 937 and the neighboring Thr 934 as major novel regulatory phosphorylation sites for KCC2 transport activity.

MATERIALS AND METHODS
Bioinformatic Analyses-To identify native KCC2 phosphorylation from proteomics approaches, the two databases PhosphoSitePlus (43) and PHOSIDA (44) were screened using the term KCC2 as entry. Both databases contain curated data of experimentally observed post-translational modifications, primarily of human and mouse proteins, which were obtained by high resolution mass spectrometric analyses.
KCC protein sequences for a diverse selection of organisms were obtained from a combination of BLAST searches against GenBank TM and data mining of the Ensembl database and the Joint Genome Institute (www.jgi.doe.gov). We used the protein sequences of human KCC1 (NP_005063.1), KCC2 (NP_065759.1), KCC3 (NP_598408.1), and KCC4 (NP_006589.2) as query. For each protein in each target species, we saved all sequences with an E-value of at least 10 Ϫ2 . These sequences were then reverseblasted (BLASTp or translated BLAST) against the Homo sapiens protein database, and only those protein sequences were retained that showed the same CCC protein sequence of H. sapiens that was used as a query sequence as the best hit (E-value of at least 10 Ϫ2 ). Each obtained sequence was then aligned at the amino acid level using the default settings in MUSCLE (45), as implemented in SeaView version 4.4.2 (46) and manually improved by eye thereafter.
Construction of Expression Clones-Wild-type rat KCC2b (GenBank TM accession number NM_134363) and HA-tagged mouse KCC2b (GenBank TM accession number NM_020333) expression clones with an HA tag at the N terminus or in the second extracellular loop were reported previously (47,48). Site-directed mutagenesis of KCC2b cDNA was performed according to the QuikChange mutagenesis system (Stratagene, Heidelberg, Germany). Forward oligonucleotides for the generation of the mutations are given in Table 1. All generated clones were confirmed by sequencing.
Determination of K ϩ -Cl Ϫ Cotransport-Transport activity was determined by measuring Cl Ϫ -dependent uptake of Tl ϩ in HEK-293 cells. Uptake measurements were done as described previously (49,50,47). Cells were transiently transfected with the respective construct, using TurboFect (Fermentas, Schwerte, Germany), according to the protocol provided. Briefly, 150 l of Opti-MEM (Invitrogen), 6 l of TurboFect (Fermentas, Karlsruhe, Germany), and ϳ3 g of DNA were mixed and incubated for 20 min at room temperature prior to transfection. 24 h after transfection, HEK-293 cells were plated in a blackwalled 96-well culture dish (Greiner Bio-One, Frickenhausen, Germany) at a concentration of 100,000 cells/well. The remainder of the cells were plated on a glass coverslip. After ϳ18 h, coverslips were processed for immunocytochemical analysis to determine transfection rates, which were routinely between 20 and 30% (Fig. 2). Cell cultures with lower transfection rates were omitted from subsequent flux measurements. The HEK-293 cells in 96-well culture dishes were processed for flux measurements by replacing the medium with 80 l of preincubation buffer (100 mM N-methyl-D-glucamine, 5 mM KCl, 2 mM CaCl 2 , 0.8 mM MgSO 4 , 5 mM glucose, 5 mM HEPES, pH 7.4) with or without 2 M FlouZin-2 AM dye (Invitrogen) plus 0.2% (w/v) Pluronic F-127 (Invitrogen). After incubation for 48 min at room temperature, cells were washed three times with 80 l GGAGATCCAGAGCATCACAGATGAAGATCGGGGCTCCATT T934A/S937Afor GGAGATCCAGAGCATCGCAGATGAAGCTCGGGGCTCC S1022Afor AGAACAAAGGCCCCGCTCCCGTCTCCTCGG S1022Dfor GAAGAACAAAGGCCCCGATCCCGTCTCCTCGGAG S1025Afor CCAGTCCCGTCGCCTCGGAGGGG S1025Dfor CCCCAGTCCCGTCGACTCGGAGGGGATC S1026Afor CAGTCCCGTCTCCGCGGAGGGGATCAA S1026Dfor CCCCAGTCCCGTCTCCGATGAGGGGATCAAGGACT mmT934Dfor CGGGAGATCCAGAGCATCGATGACGAGTCTCGGGGCTCC mmS937Dfor GGAGATCCAGAGCATCACAGATGAAGATCGGGGCTCCATT of preincubation buffer and incubated for 15 min with 80 l of preincubation buffer plus 0.1 mM ouabain to block Na ϩ /K ϩ ATPases. Thereafter, the culture dish was inserted into a fluorometer (Fluoroskan Accent, Thermo Scientific, Bremen, Germany), and the wells were injected with 40 l of 5ϫ thallium stimulation buffer (12 mM Tl 2 SO 4 , 100 mM N-methyl-D-glucamine, 5 mM HEPES, 2 mM CaSO 4 , 0.8 mM MgSO 4 , 5 mM glucose, pH 7.4). The fluorescence across the entire cell population in a single well was measured in a kinetic dependent manner (excitation 485 nm, emission 538 nm, 1 frame in 5 s in a 200-s time span). The activity was calculated with the initial values of the slope of Tl ϩ -stimulated fluorescence increase by using linear regression. At least two independent DNA preparations were used per construct, giving similar results. The effect of the thiol-alkylating reagent N-ethylmaleimide (NEM) or the kinase inhibitor staurosporine was determined by adding either 1 mM NEM or 8 M staurosporine to the preincubation buffer 15 min prior to flux measurements on the cells. For statistical analysis, data groups were compared using a Student's t test, and p Ͻ 0.05 was considered as statistically significant.
Immunocytochemistry-For immunocytochemistry, HEK-293 cells were seeded on 0.1 mg/ml poly-L-lysine-coated coverslips and incubated for 36 h. After fixation for 10 min with 4% paraformaldehyde in 0.2 M phosphate buffer and three washes in PBS, cells were incubated with blocking solution (0.3% Triton X-100, 3% bovine serum albumin, 10% goat serum in PBS) for 30 min. All steps were performed at room temperature. Primary antibody solution (anti-cKCC2, directed against the C-terminal part of KCC2, 1:1,000) (51) was added in blocking solution for 30 min. After three wash steps with PBS for 5 min, the secondary antibody was added, which was conjugated to a fluorescent probe (1:1,000; Alexa Fluor 488 goat anti-rabbit (Invitrogen)). After washing, cells were mounted onto glass slides with Vectashield Hard Set (Vector Laboratories, Burlingame, CA). Photomicrographs were taken using a Laser scanning microscope (Leica TCS SP2).
Cell Surface Labeling of KCC2-HA N-term Constructs-To determine the cell surface expression of mouse KCC2 WT-HA 2nd loop , KCC2 T934D-HA , or KCC2 S937D-HA , these constructs were expressed in HEK-293 cells. After 36 h, cells were kept at 4°C for 15 min and washed with a chilled washing solution containing 150 mM NaCl, 2.5 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 2.5 mM HEPES, and 2 g glucose/liter. Primary antibody (mouse anti-HA, 1:250 (Covance, Heidelberg, Germany)) was added for 25 min at 4°C, and cells were washed three times. The secondary antibody (Alexa Fluor 488 goat anti-mouse, 1:250 (Invitrogen)) was diluted in preheated (37°C) washing solution, and cells were incubated for 10 min at 37°C. After washing with preheated solution, cells were fixed with 4% paraformaldehyde. Fixed cells were stained as described above with anti-cKCC2 and an Alexa Fluor 594 goat anti-rabbit to detect all KCC2 protein.
For image analysis, 2048 square pixels containing 8-bit xyzconfocal pictures were taken using a Leica TCS SP5 device with an adjustable 20-fold immersion objective (oil n ϭ 1.514) and a pulsed white light laser with wavelengths adjusted to 498 nm (for Alexa Fluor 488, 10% power) and 590 nm (for Alexa Fluor 594, 4% power). The following settings were applied to all images: 100% detection of fluorescence emission with linear working HyD detectors from 507 to 548 nm (for Alexa Fluor 488) and from 610 to 646 nm (for Alexa Fluor 594) with gating restriction at 498 nm (0.30 and 6.50 ns) and 590 nm (0.30 and 6.50 ns) (Leica-microsystems.com); a pinhole of 300 m (recorded specimen thickness ϭ 13.3 m); and 2-fold line and frame averages. Ten to 21 images of each experiment were evaluated for mean intensity per pixel with pre-arranged FIJI software. Three to five biological replica were performed for each construct.
Immunoblot Analyses-To quantify the expression of KCC2 mutants, cells were lysed in a buffer containing 150 mM NaCl, 15 mM Tris, 1% dodecyl ␤-D-maltoside, and 0.4% iodoacetamide. After incubation for 5 min at 25°C, samples were centrifuged for 5 min at 125,000 ϫ g. Protein amount of the supernatant was determined using the Bradford assay. 10 g of each sample were loaded onto a 10% SDS-polyacrylamide gel system. After separation and electrotransfer onto PVDF membranes, membranes were incubated with anti-cKCC2 (dilution 1:2,000). After incubation for 2 h at room temperature, membranes were washed four times with TBS-T (20 mM Tris, 150 mM NaCl, 1% Tween, pH 7.5), and the secondary antibody donkey anti-rabbit IgG-HRP (Santa Cruz Biotechnology, Heidelberg, Germany) was applied for 1 h. After washing, bound antibodies were detected using an enhanced chemiluminescence assay (GE Healthcare) and a LAS-3000 documentation system (Fujifilm, Düsseldorf, Germany). Quantification of bands within the linear range of exposure was performed using the MultiGauge software version 3.1 (Fujifilm). Three biological replicas were performed for each construct. Data are given as means Ϯ S.E. Significant differences between the groups were analyzed by a Student's t test.

RESULTS
Database Mining and Phylogenetic Pattern of KCC2 Phosphorylation Sites-To identify bona fide phosphorylation sites in KCC2, the databases PhosphoSitePlus and PHOSIDA were searched for KCC2 entries. These databases contain experimentally observed phosphorylation sites identified largely by mass spectrometry-based proteomic studies (43,44). In addition to the known phosphorylation sites Thr 906 , Ser 940 , and Thr 1007 , Ser 25   databases (numbering according to the rat KCC2b protein) ( Table 2). Three of the amino acids, Ser 25 , Ser 26 , and Thr 34 , reside in the cytoplasmic N terminus, and the remainder are located in the C-terminal part (1). Here, we focused on the six phosphoserines Ser 25 , Ser 26 , Ser 937 , Ser 1022 , Ser 1025 , and Ser 1026 . Phylogenetic analyses of the KCC subfamily revealed that they cover different patterns of phylogenetic conservation. Ser 1022 , Ser 1025 , and Ser 1026 are highly specific to tetrapod KCC2 ( Fig. 1). Ser 937 is present only in vertebrate KCC2 and in non-therian KCC4. Finally, Ser 25 and Ser 26 showed the broadest occurrence with a presence in the most vertebrate KCC1, KCC2, and KCC4 family members (Fig. 1). Notably, the previously reported amino acids Thr 906 and Thr 1007 , which had been implicated in developmental activation of KCC2 during brain maturation (37), are highly conserved throughout metazoans (Fig. 1). This suggests that their phosphorylation does not only  play a role in vertebrate KCCs but also in those transporters from more ancient species like the choanoflagellate Monosiga brevicolis, which forms a sister group to the metazoan (52). Analysis of the neighboring amino acid sequences demonstrated conservation patterns very similar to those observed for the phosphorylated amino acid residues (Fig. 1). This indicates that the respective phosphorylation can also occur in other species.
Expression Analyses of KCC2 Phospho-mutants-To study the role of the phosphorylated serines Ser 25 , Ser 26 , Ser 937 , Ser 1022 , Ser 1025 , and Ser 1026 , we generated two mutants of rat KCC2b for each phosphorylation site. This isoform was chosen because it is the major KCC2 protein in the mature brain (53). In the following, we will refer to it as KCC2. Amino acid residues were mutated to either aspartate or alanine to mimic phosphorylated or dephosphorylated states of the respective serine. In addition, T906A and T1007A (corresponding to mouse and human Thr 906 and Thr 1006 ) mutants were generated. They were previously shown to result in increased KCC2 activity and served as positive controls (37,38). This resulted in 14 mutants. To check whether these mutations effect KCC2 expression, all constructs were transfected into HEK-293 cells. This cell line is most widely used to study KCC2 transport activity in a mammalian heterologous expression system (2,37,39,47,54). Immunocytochemical analyses revealed expression of all 14 mutants (Figs. 2 and 3). Immunoreactivity was detected for all mutants both at the plasma membrane and inside the cells (Fig. 3). Only the nucleus was spared. Compared with wildtype KCC2 (KCC2 WT ), no difference in subcellular localization was obvious upon visual inspection. Thus, all mutants were well expressed in HEK-293 cells.
Transport Activity of KCC2 Phospho-mutants-In a first series of experiments, we focused on the two N-terminal Ser 25 and Ser 26 phosphorylation sites and determined their activity in HEK-293 cells using Tl ϩ flux measurements. All four mutants, KCC2 S25A , KCC2 S25D , KCC2 S26A , and KCC2 S26D , as well as KCC2 WT mediated increased Tl ϩ flux compared with mocktransfected cells ( Fig. 4 and Table 3). Furthermore, 2 mM furosemide blocked most of the flux, demonstrating that the transport activity was largely mediated by KCC2 ( Fig. 4 and Table 3) (2).
No difference in activity was detected between KCC2 WT and KCC2 mutants or between alanine and aspartate substitutions. Previously, analysis of three N-terminal phosphorylated threonines in NKCC2 had revealed that only compound mutants resulted in altered behavior of the transporter (55). We therefore also generated the double mutants KCC2 S25A/S26A and KCC2 S25D/S26D . Both mutants are well expressed in HEK-293 cells (Figs. 2 and 3) and showed flux activities similar to KCC2 WT (Fig. 4 and Table 3). To ensure that our system detect differences in flux, we also determined the flux of KCC2 T906A and KCC2 T1007A . These mutants were shown to facilitate KCC2 activity (38). In keeping with the previous report, both mutants displayed a significantly increased transport activity, compared with KCC2 WT (Fig. 4 and Table 3). These results revealed that our system was well suited to detect altered KCC2 transport activity. The flux data therefore demonstrate that Ser 25 and Ser 26 single or combined substitutions do not influence KCC2 transport activity in HEK-293 cells.
We next focused on the C-terminal phosphorylation sites. Substitution of Ser 1022 , Ser 1025 , and Ser 1026 by alanine or aspartate resulted in no change in transport activity compared with KCC WT (Fig. 5 and Table 3). In contrast, KCC2 S937D mediated a furosemide-sensitive Tl ϩ flux, which was significantly increased compared with KCC2 WT (Fig. 5 and Table 3). The FIGURE 5. Transport activity of C-terminal KCC2 mutants. HEK-293 cells were transfected with KCC2 WT and KCC2 mutant constructs, and transport activity was determined by performing Tl ϩ flux measurements. The Tl ϩ flux mediated by KCC2 S1022A , KCC2 S1022D , KCC2 S1025A , KCC2 S1025D , KCC2 S1026A , and KCC2 S1026D was similar to the flux of KCC2 WT . In contrast, the phosphomimetic mutants KCC2 T934D and KCC2 S937D showed a significantly increased transport activity. The mutants KCC2 T934A , KCC2 S937A , and KCC2 T934A/S937A , which mimicked the dephosphorylated state, exhibited transport levels comparable with KCC2 WT . In the presence of furosemide, Tl ϩ flux was blocked. Graphs represent the mean Ϯ S.E. of at least three independent measurements, normalized to KCC2 WT . Statistical analysis is presented in Table 3.   showed no significant differences in transport activity. Confirming previous studies, KCC2 T906A and KCC2 T1007A displayed a significantly higher transport activity compared with KCC2 WT . In the presence of furosemide, the Tl ϩ flux was blocked. Graphs represent mean Ϯ S.E. of at least three independent measurements, normalized to KCC2 WT . Statistical analysis is presented in Table 3.

JOURNAL OF BIOLOGICAL CHEMISTRY 18673
phosphorylation-blocking mutant KCC2 S937A showed transport activity similar to KCC2 WT (Fig. 5 and Table 3). This result indicates that Ser 937 is not phosphorylated in KCC2 WT when expressed in HEK-293 cells. We noticed that the phosphopeptide harboring Ser 937 also contains Thr 934 in close proximity. In some mass spectrometric analyses of the corresponding phosphopeptide, the precise location of the phosphate group could not be determined, leading to an ambiguous status of Thr 934 as a phosphorylation site (Table 2) (56,57). BecauseThr 934 and Ser 937 demonstrated exactly the same degree of evolutionary conservation (Fig. 1), we mutated Thr 934 as well and determined the transport activity of KCC2 T934A and KCC2 T934D . Intriguingly, KCC2 T934Dmediated flux was significantly facilitated compared with KCC2 WT , whereas KCC2 T934A transport activity was similar to the control (Fig. 5 and Table 3). Thus, both Thr 934 mutants closely matched the results observed for the corresponding Ser 937 mutants (Fig. 5). This suggests that Thr 934 also represents a regulatory phosphorylation site. To exclude that transport activity of the single alanine mutants KCC2 T934A or KCC2 S937A was due to residual phosphorylation of the nonmutated Ser 937 or Thr 934 , respectively, we also generated the double mutant KCC2 T934A,S937A . The transport activity of this mutant was indistinguishable from that of the single mutants and of KCC2 WT (Fig. 5 and Table 3). These data support the notion that both amino acids are not phosphorylated in HEK-293 cells and that their phosphorylation does not contribute to the basal activity of KCC2 in this cell system. We conclude that phosphorylation of Thr 934 or Ser 937 represents the effective mechanisms to regulate KCC2 activity.
NEM and Staurosporine Effects Reverse upon Phosphorylation of Thr 934 or Ser 937 -NEM is one of the earliest described activators of KCC transport (2,58). Its mode of action is still unknown, but presumably it occurs via changes in C-terminal phosphorylation (58,59). To investigate whether Thr 934 or Ser 937 are involved in the NEM effect, we determined the transport activity of KCC2 T934A , KCC2 T934D , KCC2 S937A , KCC2 S937D , and KCC2 T934A,S937A in the presence of 1 mM NEM. In agreement with previous studies, this concentration of NEM significantly increased KCC2 WT transport activity ( Fig. 6 and Table 4). Similarly, all three phosphorylation-blocking alanine mutants KCC2 T934A , KCC2 S937A , and KCC2 T934A,S937A were significantly stimulated by NEM ( Fig. 6 and Table 4). This demonstrates that NEM is not acting via phosphorylation of Thr 934 or Ser 937 . Unexpectedly, the two phospho-mimetic mutants KCC2 T934D and KCC2 S937D were significantly inhibited by NEM and showed transport activities similar to KCC2 WT (Fig. 6 and Table 4). This suggests that the NEM effect is dependent on the conformational state of KCC2.
To test whether this observation is true for other stimuli as well, we applied staurosporine. This alkaloid acts as a kinase inhibitor that activates KCCs (33,60). Staurosporine closely mimicked the effect of NEM. KCC2 WT as well as all three alanine mutants, KCC2 T934A , KCC2 S937A , and KCC2 T934A,T937A , were significantly stimulated by this kinase inhibitor, whereas the transport activity of KCC2 T934D and KCC2 S937D was significantly reduced to the level of KCC2 WT (Fig. 6 and Table 4). From these results, we conclude that the effect of both agents is strongly dependent on the conformational state of KCC2. Furthermore, staurosporine does not act via Thr 934 or Ser 937 , because the transport activities of both the alanine and the FIGURE 6. Effect of NEM and staurosporine on the transport activity of Thr 934 and Ser 937 mutants. HEK-293 cells were transfected with KCC2 WT and KCC2 mutant constructs and transport activity was determined by performing Tl ϩ flux measurements. NEM and staurosporine were added 15 min prior to measurement. KCC2 WT , the single mutants KCC2 T934A and KCC2 S937A , and the double mutant KCC2 T934A/S937A were significantly activated by NEM and staurosporine. In contrast, the activity of the mutants KCC2 T934D and KCC2 S937D was significantly reduced in the presence of NEM and staurosporine. Graphs represent mean Ϯ S.E. of at least three independent measurements. Statistical analysis is presented in Table 4. aspartate mutants could still be modified by this kinase inhibitor. Finally, the concordant effects of NEM and staurosporine on the mutants suggest that both agents act via a similar mechanism. This is in agreement with the previous observation that both agents were unable to stimulate KCCs in ATP-depleted cells (61) and that deficiency of Src family kinases Fgr and Hck obliterates NEM stimulation of KCCs in red blood cells (62). Thus, NEM likely modifies KCC phosphorylation pattern in a way similar to staurosporine. No Change in Abundance and Cell Surface Expression of KCC2 T934D and KCC2 S937D Mutants-Finally, we addressed the underlying mechanism of increased transport activity of KCC2 T934D and KCC2 S937D . Altered phosphorylation might affect abundance, cell surface expression, or the transport kinetics of KCC2. To address the abundance of the KCC2 mutants, immunoblot analyses were performed for KCC2 T934A , KCC2 T934D , KCC2 S937A , and KCC2 S937D . Quantitative analysis of the immunoblots revealed no significant change in protein level (Fig. 7).
To investigate cell surface expression of the mutants, we used a previously reported KCC2 expression construct with an extracellular HA tag in the second extracellular loop between transmembrane domains 3 and 4 (48). We substituted Thr 934 or Ser 937 by aspartate resulting in the two new constructs KCC2 T934D-HA 2nd loop and KCC2 S937D-HA 2nd loop . Transfection of either of the two constructs into HEK-293 cells resulted in an increased KCC2 activity, compared with KCC2 WT-HA 2nd loop , demonstrating that the HA tag did not interfere with mutations in Thr 934 or Ser 937 (data not shown). Surface expression of the KCC2 variants was determined in HEK-293 cells by a sandwich protocol. As a control, we used the construct KCC2 WT-HA N-term with the HA tag at the intracellularly located N terminus. Surface-expressed KCC2 was labeled by incubating unfixed cells prior to permeabilization with antibodies against the extracellular HA tag for 25 min. Subsequently, cells were fixed and permeabilized, and total KCC2 was monitored using a KCC2 antibody. HEK-293 cells transfected with KCC2 WT-HA N-term showed only poor labeling with anti-HA, whereas anti-cKCC2 antibodies demonstrated expression of the protein (Fig. 8A). In cells transfected with the HA tag in the second extracellular loop, incubation with the anti-HA antibody resulted in cell surface staining for all three constructs, KCC2 WT , KCC2 T934D-HA , and KCC2 S937D-HA (Fig. 8A). Quantitative analyses of the signal ratio of surface staining to total expression revealed no difference between wild-type and mutant constructs (Fig. 8B), whereas significant differences in surface staining were obtained between KCC2 WT-HA N-term and KCC2 WT-HA 2nd loop (p ϭ 0.018), KCC2 T934D-HA 2nd loop (p ϭ 0.011), and KCC2 S937D-HA 2nd loop (p ϭ 0.027). From this observation, we conclude that the mutations do not affect surface expression but affect the intrinsic transport activity of KCC2.

DISCUSSION
Here, we report the identification of novel potent KCC2 phosphorylation sites encompassing Ser 937 and Thr 934 , which are involved in kinetic regulation of the transport activity. Phospho-mimetic substitution of these two amino acid residues inverted the action of staurosporine and NEM. These results hint to a cross-talk between different phosphorylation sites and reveal the impact of the state of the transporter on the action of post-translational modifications.
Modern mass spectrometry-based in vivo phospho-proteomics studies have identified thousands of bona fide phosphorylation sites of brain proteins (56,57,63). So far, these publicly available data have been poorly exploited. Here, we started to investigate the possible role of such experimentally verified phosphorylation sites for regulation of KCC2 transport activity. Mutational analyses in combination with flux measurements identified a hitherto unknown potent regulatory role of Thr 934 and Ser 937 (Fig. 5). Substitution of these amino acids by the phospho-mimetic amino acid aspartate significantly facilitated KCC2 transport in HEK-293 cells. Thus, phosphorylation of these amino acids will enhance KCC2 transport activ- FIGURE 7. Immunoblot analysis of KCC2 WT , KCC2 T934A/T934D , and KCC2 S937A/S937D . A, KCC2 constructs were transfected into HEK-293 cells, and proteins were isolated 36 h later. 10 g of protein were loaded onto a 10% SDS-polyacrylamide gel system, and the amount of KCC2 was determined by immunoblotting using an anti-cKCC2 antibody. KCC2-IR was detected for KCC2 WT and KCC2 mutants, but not for mock-transfected cells, whereas ␤-tubulin, serving as a loading control, was present in all lanes. The figure shows one experiment out of three biological replicates with similar results. B, KCC2 bands were quantified and normalized to the expression of ␤-tubulin. This revealed no differences in the expression level between KCC2 WT and KCC2 mutants in HEK-293 cells. Graphs represent mean Ϯ S.E. of three independent experiments.
ity. This is in line with the observation of phosphorylated Ser 937 and likely Thr 934 in adult brain tissue (57).
So far, phosphorylation mainly affected trafficking or the half-lives of KCC2 (5,36). In contrast to these findings, the increased activity of KCC2 T934D or KCC2 S937D was not paralleled by an increase in total amount or cell surface expression of the transporter (Figs. 6 and 7). These results reveal a kinetic up-regulation of the transport activity by phosphorylation of these two amino acid residues. Kinetic regulation of the intrinsic transport activity of KCC2 is also suggested for Thr 906 and Thr 1007 , based on studies of the homologous amino acid residues Thr 991 and Thr 1048 in KCC3 (37). Surface biotinylation experiments demonstrated that dephosphorylation of these two threonines activated KCC3 without an increase in plasma membrane localization (37). Taken together, these data suggest a model, in which the intrinsic transport activity of KCC2 is increased by phosphorylation of Thr 934 or Ser 937 (this study) and decreased by phosphorylation of Thr 906 and Thr 1007 (37).
The prevailing notion holds that phosphorylation activates NKCCs and inactivates KCCs (31,58,65,66). This elegant concept of reciprocal regulation of Cl Ϫ -inward and Cl Ϫ -outward transporters was originally based on the observations that cell swelling, NEM, staurosporine, and serine-threonine phosphatase 1 inhibited NKCC1 (67-69) and activated KCCs (33,58,67). This view was subsequently substantiated by the reciprocal effect of WNKs on KCCs and NKCC/NCC transport activities (31,70) and by the fact that dephosphorylation of Thr 906 and Thr 1007 facilitates KCC transport (37) and phosphorylation of Tyr 903 and Tyr 1087 increases KCC2 degradation in lysosomes (41). However, the recent characterization of the Ser 940 phosphorylation site of KCC2 already challenged this concept. Dephosphorylation of Ser 940 decreased cell surface expression of KCC2 (39) and concomitantly caused depolarizing action of GABA in neurons (71). Our observation of up-regulated activity in KCC2 T934D and KCC2 S937D supports the emerging concept that KCC2 can both be activated and deactivated by phosphorylation, depending on the amino acid residues involved. However, the staurosporine and NEM experiments indicate that the targeted amino acid residue is not the sole determinant for the effect of phosphorylation/dephosphorylation. Both agents can have opposite effects, depending on the phosphorylation state of Thr 934 and Ser 937 (Fig. 6). In the dephosphorylated state, mimicked by alanine substitutions, KCC2 transport was facilitated by these two agents, and in the phosphorylated state, represented by the aspartate mutants, the effect was the opposite. These results indicate that the conformational state of KCC2 has an impact on the functional consequences of phosphorylation/dephosphorylation of a given amino acid residue. This is reminiscent of red blood cells, where low concentrations of NEM at 0°C activated and high concentration of NEM at 37°C inhibited KCC-mediated transport (72).
Because of the rapid effects within minutes, NEM or staurosporine are assumed to regulate the intrinsic activity of KCCs, similar to phosphorylation of Thr 934 or Ser 937 (33,58,59). A likely explanation for the opposite effect of both agents on Thr 934 and Ser 937 mutants is because they induce conformational changes that differ depending on the conformational state of KCC2. In KCC2 WT , KCC2 T934A , and KCC2 S937A , they cause a state with high transport activity, whereas in KCC2 T934D and KCC2 S937D , which have already adopted the state of high transport activity, further conformational changes imposed by indirect action of NEM or staurosporine convert KCC2 into a state of basal activity. An alternative explanation is that the conformation of KCC2 T934D and KCC2 S937D deoccludes a hidden site, upon which NEM or staurosporine indirectly act in a manner different from their action in KCC2 WT (58). In either case, the data reveal a functional cross-talk between phosphorylation of Thr 934 and Ser 937 and another staurosporine/NEM-modulated phosphorylation site. Thus, the effect of phosphorylation can itself be modulated by phosphorylation of other sites. This might be important for finetuning the operational mode of KCC2 and to integrate different signaling pathways. These findings also imply that drugs aimed to modify KCC2 phosphorylation for therapeutic benefit (36,73) have to be investigated under different conditions as their effect might depend on the conformational state of KCC2 and therefore be highly context-specific. Functional cross-talk between different regions has also been suggested for NKCC1, where phosphorylation of the N terminus causes movement of the C terminus, which in turn entails altered interactions between different transmembrane domains (74).
Among the phosphorylated amino acid residues in KCC2, Thr 934 and Ser 937 showed an intermediate level of evolutionary conservation. The two residues are only present in the KCC2 subgroup of vertebrate KCCs and in non-therian vertebrate KCC4 family members (Fig. 1). This is in stark contrast to Thr 906 and Thr 1007 , which are strongly conserved throughout evolution. Phosphorylation of Ser 26 , which is present in most vertebrate KCC1, KCC2 and KCC4 family members, had no influence on KCC2 activity. Taken together, these data establish evolutionary conservation as a poor predictor for regulatory phosphorylation sites in KCCs. This might be expected, because subgroup-specific phosphorylation sites enable differential regulation of the diverse KCCs within the same cell. This is likely an important step favoring subfunctionalization of the different family members during vertebrate evolution (10), as they can be co-expressed in brain areas (75).
Our analyses failed to detect a regulatory role of Ser 25 , Ser 26 , Ser 1022 , Ser 1025 , and Ser 1026 in HEK-293 cells. Ser 25 and Ser 26 are located in the cytoplasmic N terminus of KCC2. Previous investigations have revealed an important role of this sequence for CCC transport activity. Phosphorylation of Ser 96 in KCC3 is involved in cell swelling-related activation of the transporter (76). Furthermore, N-terminal truncation disrupted transport activity of KCC1 (77) and KCC2 (9). In the recently identified KCC2a isoform, a SPAK-binding site with a yet unknown functional role is located in the N terminus (53). In NKCC1 and NKCC2, three N-terminally located phosphothreonines modulate transport activity (55). Interestingly, in NKCC2, mutation of either of the three threonines to alanine had no effect; mutations of two threonines resulted in reduced activity, and mutations of all three threonines blocked hypertonic activation (55). It might therefore be that mutation of more than one phosphorylated amino acid residue in the N terminus of KCC2 is required to observe a phenotype. However, the double mutants KCC2 S25D/S26D and KCC2 S25A/S26A demonstrated KCC2 WTlike activity as well (Fig. 4), indicating that phosphorylation of these two serines do not influence KCC2 activity in HEK-293 cells. This does not exclude a role in other processes such as trafficking of KCC2 in neurons. The C-terminal Ser 1022 , Ser 1025 , and Ser 1026 are located within the so-called ISO domain of KCC2, which confers KCC2 activity under isotonic conditions (78 -80). The observation that this isotonic activity of KCC2 is not influenced by the serine-threonine phosphatase inhibitor calyculin A in Xenopus laevis oocytes (78) supports our conclusion that phosphorylation of Ser 1022 , Ser 1025 , and Ser 1026 is not required for basal KCC2 transport activity.

CONCLUSION
At least three different regulatory states of KCC2 can exist in the plasma membrane as follows: an inactive conformation, a conformation conferring basal activity, and a conformation with high transport rate. Our data reveal that Thr 934 and Ser 937 participate in interconversion between these conformational states and, together with Thr 906 and Thr 1007 , may represent key residues for kinetic regulation of KCC2. Furthermore, the varying effect of staurosporine and NEM provide evidence for a cross-talk between different phosphorylation sites. This might have implications for drug research aimed to modify KCC2 function via changes in phosphorylation.