Annexin A2 Mediates Apical Trafficking of Renal Na+-K+-2Cl− Cotransporter*

Background: Regulated membrane expression of the renal furosemide-sensitive Na+-K+-2Cl− cotransporter (NKCC2) determines essential functions of the kidney. Results: Annexin A2 (AnxA2) is a new interaction partner of NKCC2 with respect to its adluminal trafficking and activation. Conclusion: Discussion focuses on trafficking, membrane association, and phosphorylation of NKCC2 and its cellular ligand AnxA2. Significance: AnxA2 is relevant for the activation of NKCC2. The furosemide-sensitive Na+-K+-2Cl− cotransporter (NKCC2) is responsible for urine concentration and helps maintain systemic salt homeostasis. Its activity depends on trafficking to, and insertion into, the apical membrane, as well as on phosphorylation of conserved N-terminal serine and threonine residues. Vasopressin (AVP) signaling via PKA and other kinases activates NKCC2. Association of NKCC2 with lipid rafts facilitates its AVP-induced apical translocation and activation at the surface. Lipid raft microdomains typically serve as platforms for membrane proteins to facilitate their interactions with other proteins, but little is known about partners that interact with NKCC2. Yeast two-hybrid screening identified an interaction between NKCC2 and the cytosolic protein, annexin A2 (AnxA2). Annexins mediate lipid raft-dependent trafficking of transmembrane proteins, including the AVP-regulated water channel, aquaporin 2. Here, we demonstrate that AnxA2, which binds to phospholipids in a Ca2+-dependent manner and may organize microdomains, is codistributed with NKCC2 to promote its apical translocation in response to AVP stimulation and low chloride hypotonic stress. NKCC2 and AnxA2 interact in a phosphorylation-dependent manner. Phosphomimetic AnxA2 carrying a mutant phosphoacceptor (AnxA2-Y24D-GFP) enhanced surface expression and raft association of NKCC2 by 5-fold upon low chloride hypotonic stimulation, whereas AnxA2-Y24A-GFP and PKC-dependent AnxA2-S26D-GFP did not. As the AnxA2 effect involved only nonphosphorylated NKCC2, it appears to affect NKCC2 trafficking. Overexpression or knockdown experiments further supported the role of AnxA2 in the apical translocation and surface expression of NKCC2. In summary, this study identifies AnxA2 as a lipid raft-associated trafficking factor for NKCC2 and provides mechanistic insight into the regulation of this essential cotransporter.

Once expressed on the cell surface, a major proportion of NKCC2 is found within cholesterol-and glycosphingolipid-enriched membrane microdomains or lipid rafts (21,22). Partitioning of the cotransporter into lipid rafts appears to interfere with its regulated cellular translocation, including apical trafficking and endocytic retrieval (21)(22)(23). Molecular determinants involved in raft-dependent trafficking events have, however, barely been characterized so far.
Annexins have been shown to mediate lipid raft-dependent trafficking of several transmembrane proteins (24 -26). Members of the annexin protein family are known to associate with membrane phospholipids and facilitate membrane fusion of cytoplasmic vesicles with the plasma membrane (27)(28)(29). One particular member, annexin A2 (AnxA2), has specific roles in the kidney, where it mediates cAMP-induced trafficking of aquaporin 2 in the collecting duct (30). AnxA2 is expressed in TAL as well, where cAMP-induced fusion of NKCC2-carrying vesicles with the luminal plasma membrane permits rapid surface recruitment and activation of apically stored NKCC2 (12,19,(31)(32)(33)(34). We assumed a functional relation between AnxA2 and NKCC2 based on our present yeast two-hybrid assay data. AnxA2 may bind raft lipids to facilitate the clustering of cholesterol and glycosphingolipids and may thereby promote raft formation and support membrane fusion events, e.g. apical vesicles carrying the NKCC2 transporter (35). To do so, AnxA2 may act as a monomer or as a heterotetrameric complex together with its binding partner, the S100A10 protein (also termed p11) (27,36).
Here, we have identified AnxA2 as a component involved in NKCC2 trafficking. AnxA2 directly and selectively interacts with the N-terminal cytoplasmic domain of the cotransporter in its nonphosphorylated state and facilitates the apical translocation of NKCC2 in response to activation of TAL transport capacity by AVP or low chloride hypotonic stimulation.

EXPERIMENTAL PROCEDURES
Animals and Cells-Adult male Sprague-Dawley rats and Brattleboro rats with diabetes insipidus (DI; all rats 10 -14 week old) received standard diet and tap water ad libitum. DI rats received an intraperitoneal injection of the V2 receptor agonist, 1-deamino-8-D-arginine vasopressin (dDAVP; 1 g/kg body weight), or vehicle (0.9% saline; n ϭ 8 rats per group). For histochemical analysis, rats were anesthetized with Nembutal (100 mg/kg body weight); the abdominal cavity opened, and kidneys were fixed by retrograde perfusion through the abdominal aorta using 3% paraformaldehyde in PBS (37). Kidneys were then processed for cryostat and paraffin sectioning. For biochemical studies, rats were anesthetized with isoflurane and killed by cervical dislocation; kidneys were then removed, dissected into cortical and medullary portions, and processed for the biochemical evaluation.
Immunohistochemistry-Tissue sections were produced from paraffin-embedded or frozen material. Paraffin sections were dewaxed and boiled for 6 min in citrate buffer, pH 6, for antigen retrieval. Fixed cultured cells were treated for 10 min at room temperature with 0.5% Triton X-100, 1% bovine serum albumin (BSA) in PBS. The following general blocking step was performed on tissue sections with 5% skim milk in PBS for 30 min or on cells with 1% BSA in PBS. The respective primary antibodies were applied overnight at 4°C and rinsed in PBS, and suitable secondary Cy2-or Cy3-conjugated antibodies (Dianova) were then incubated for 1 h. Tissue was then mounted in glycerol/PBS, and coverslips were mounted on glass slides with Fluoromount-G (SouthernBiotech) and viewed in a confocal laser microscope (LSM 5 Exciter, Zeiss).
Tissue Preparation and Fractionation-Renal tissue samples were homogenized in homogenization buffer (250 mM sucrose, 10 mM triethanolamine, and protease inhibitors, pH 7.5) by grinding and subsequent ultrasonication. Cell debris and nuclei were removed by centrifugation at 1,000 ϫ g for 15 min at 4°C. Protein concentrations of the obtained postnuclear lysates were determined by the Bradford method using the BCA-protein assay reagent kit (Pierce). Plasma membrane-enriched fractions were separated by additional centrifugation at 16,000 ϫ g for 30 min at 4°C. The remaining supernatant fraction was thus enriched in cytoplasmic structures and vesicles. 50 g of protein were loaded per lane for immunoblotting.
Isolation of Lipid Rafts by Triton X-100 and Sucrose Gradient Fractionation-Lipid raft extraction was performed as described previously (41). Briefly, outer medullary postnuclear lysates were centrifuged at 120,000 ϫ g for 1 h at 4°C to obtain membranes and vesicles. The obtained pellets were resuspended in ice-cold homogenization buffer supplemented with 1% Triton X-100 and incubated for 1 h at 4°C under agitation, followed by ultracentrifugation at 120,000 ϫ g for 1 h at 4°C. The resulting pellets containing the detergent-insoluble fractions were resuspended in 2 ml of homogenization buffer with 40% sucrose and overlaid with 5 ml of 30% sucrose and 5 ml 5% AnxA2 Regulates Apical NKCC2 Trafficking sucrose to create a discontinuous sucrose gradient required for floating assay. The gradient was centrifuged overnight at 40,000 rpm and 4°C. Twelve 1-ml fractions were removed from the top of the gradient and analyzed by immunoblotting.
Immunoprecipitation of NKCC2 and AnxA2-NKCC2 and AnxA2 were immunoprecipitated from medullary lysates or lipid raft-enriched fractions of the discontinuous sucrose gradient (see above). Anti-NKCC2 (Millipore) or anti-AnxA2 (Santa Cruz Biotechnology) antibodies were coupled to Dynabeads according to the manufacturer's guidelines (Dynabeads M-270 Epoxy; antibody coupling kit (Invitrogen)). Lipid raftenriched fractions of the floating assay were identified by the presence of the typical raft proteins such as flotillin-1 and Tamm-Horsfall protein, pooled, and centrifuged at 40,000 ϫ g. The resulting pellets were dissolved in immunoprecipitation (IP) buffer (TBS containing 0.5% Tween 20 and proteinase inhibitor). Postnuclear rat outer medullary lysates (10 g/l) or lipid raft-enriched extracts (5 g/l) were incubated with antibody-coupled beads overnight at 4°C upon gentle agitation. The beads were washed three times with IP buffer followed by a stringent washing step (300 mM NaCl in IP buffer). Beads were boiled in 1ϫ Laemmli buffer to elute the precipitated proteins, and eluates were analyzed by immunoblotting.
Yeast Two-hybrid-Yeast two-hybrid screening was performed in an automated interaction-mating-matrix approach as described (44). DNA coding for two overlapping N-terminal fragments (AS (40 -140) and AS(77-177)) of human NKCC2 were amplified by PCR using a full-length clone for human NKCC2 as template (DKFZp686O14110Q or IRATp970G1058D (ImaGenes)). For generation of bait plasmids, PCR fragments were cloned into a pBTM117c yeast expression vector to generate fusion proteins containing a LexA binding domain and one of the NKCC2 fragments. As a selection marker, the vector contains the gene for phosphoribosylanthranilate isomerase (trp1), an enzyme required for tryptophan biosynthesis (45). Haploid L40ccMATa yeasts were transformed with bait plasmids and subsequently mated to a matrix of 16,888 individual L40ccMAT␣ yeast clones transformed with pACT-4 vectors that code for a set of nonredundant human proteins fused to a GAL4 activation domain and ␤-isopropylmalate dehydrogenase (Leu2) as a selection marker. Diploid cells were grown on SDII media deficient for tryptophan and leucine to select for the presence of both bait and prey plasmids. Physical interaction of the NKCC2 fragment with the respective prey protein results in the reconstitution of the GAL4 transcription factor that drives the expression of (lexAop)4-HIS3, (lexAop)8-URA3, and ␤-galactosidase reporter genes and thus allows growth on SDIV selection media lacking tryptophan, leucine, histidine, and uracil. All clones were also screened for their ␤-galactosidase activity using LacZ staining as described previously (46). To increase the stringency of the assay, the SDIV and LacZ approaches were run in triplicate and evaluated in a blinded fashion. Only clones that grew three times on SDIV medium and were at least positive twice for LacZ were considered valid hits. Identified candidates were subsequently ranked based on their TAL expression. To this end mRNA serial analysis of gene expression (SAGE) tags detected in the human nephron segments were mapped to human coding cDNA sequences. Nephronspecific SAGE tags were obtained as described previously (32,47). Human cDNA data were obtained from ENSEMBL (48) with the BIOMART tool (49). Using scripting software developed in-house, we mapped the SAGE tags to the cDNA and obtained for each gene a score for the nephron parts as follows: glomerulus; proximal convoluted tubule; proximal straight tubule; medullary thick ascending limb; cortical thick ascending limb; distal convoluted tubule; cortical collecting duct; outer medullary collecting duct, total. The ratio of (MTALϩCTAL)/total was used to indicate TAL-specific expression. A list of genes was produced that indicated positive yeast two-hybrid screening results and a high TAL expression according to the SAGE tag analysis.
Mass Spectrometry-NKCC2 and control IgG immunoprecipitates from lipid raft-enriched outer medullary fractions were subjected to SDS-PAGE and Coomassie Brilliant Blue staining. The bands produced by NKCC2 IP were excised, and the gel plugs were washed and equilibrated with ammonium bicarbonate in acetonitrile and digested with 0.02 g of trypsin at 37°C for 24 h. The resulting peptides were desalted and concentrated utilizing the ZipTip C18 (Millipore) technology and eluted directly onto the MALDI target (Bruker Daltonik) using ␣-cyano-4-hydroxycinnamic acid as matrix. The subsequent mass spectrometry (MS) analyses were carried out using a matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS, MALDI-Lift fragment massspectrometry (MALDI-TOF-TOF MS; Bruker)). Calibrated and annotated spectra were subjected to a database search (Swiss-Prot) utilizing the Bruker Bio-Tool 3.2 and the Mascot 2.2 search engine, which compare the experimental MALDI-TOF MS and MALDI-TOF-TOF MS data set with the calculated peptide masses for each entry into the sequence database and use empirically determined factors to assign a statistical weight to each individual peptide match.
DNA Constructs-Full-length mouse AnxA2 (NM_ 007585.3) and S100A10 (NM_009112.2) were amplified from mouse kidney cDNA using corresponding primer pairs and cloned into pEGFP-N1 or pGEX-6P1 plasmids for transfection experiments or binding assays. The complete N-terminal domain of NKCC2 (coding triplets 1-175; NM_001270617.1) was amplified from rat kidney cDNA and cloned into pGEX-6P1 vector. The mutant NKCC2 N-terminal constructs mimicking constitutive phosphorylation (substitution of Thr or Ser for Asp) or dephosphorylation (substitution of Thr or Ser for Ala) of conserved phosphoacceptor sites (Thr-96, Thr-101, Thr-114, and Ser-126) as well as mutant AnxA2 constructs mimicking constitutive phosphorylation (substitution of Tyr-24 or Ser-26 for Asp) or dephosphorylation (substitution of Tyr-24 for Ala) were generated using site-directed mutagenesis. All constructs were confirmed by sequencing (primer sequences are available on request).
GST Pulldown-After transformation of Escherichia coli TOP10 with pGEX-6p1 plasmids containing wild-type or mutant NKCC2 N termini, bacteria were incubated with 0.2 mM isopropyl ␤-D-thiogalactopyranoside (Thermo Scientific) for 15 h at room temperature to induce protein expression. FIGURE 1. NKCC2 is colocalized with AnxA2 and S100A10 in TAL. A and B, confocal images of rat medullary kidney TAL (mTAL) and MD double-stained with AnxA2 or S100A10 (red) and NKCC2 (green). Merge images show nuclei counterstained in blue; scale bars, 20 m. NKCC2 colocalization with AnxA2 and S100A10 is observed in mTAL. MD also shows colocalization with more cytoplasmic staining of AnxA2 and S100A10 which is absent from cortical TAL (cTAL). C, cytoplasmic view of luminal plasma membrane sheets from cultured rat TAL cells (raTAL) with EM immunogold double labeling of NKCC2 (black arrows) and AnxA2 or S100A10 (white arrowheads); scale bars, 40 nm. All signals are largely restricted to dark areas, whereas light areas show little signal. Below, quantification of the individual signals is shown. Data are the means Ϯ S.D.; *, p Ͻ 0.05; **, p Ͻ 0.01. D, same view, with EM immunogold double-labeling of NKCC2 (black arrowheads) and flotillin-1 or transferrin receptor (TFR; white arrowheads); scale bars, 40 nm. Although NKCC2 and flotillin-1 signals are colocalized in the dark areas, NKCC2 and transferrin receptor show distinct localization. Below, quantification of the individual signals is shown. Data are the means Ϯ S.D.; *, p Ͻ 0.05; **, p Ͻ 0.01.

Results of yeast two-hybrid assay
Yeast two-hybrid assay identified 36 candidate proteins as potential interacting partners with N-terminal fragments of NKCC2. Subsequent ranking of the candidates according to their expression in the TAL identifies AnxA2 among the proteins with the highest expression levels. Top hits of positive interacting genes of the yeast two-hybrid screen sorted according to the score indicating the relative TAL expression (last column). This score derives from the cTAL and mTAL SAGE hits. Base count refers to the cDNA length of the transcript. The 2nd column refers to the Ensemble database gene identifier and 3rd and 4th columns to yeast two-hybrid screening results.

AnxA2 Regulates Apical NKCC2 Trafficking
After that, bacteria were lysed with lysozyme (Sigma) and Triton X-100. Fusion proteins were purified by overnight incubation of lysates with Glutathione magnetic beads (Pierce) and subsequent washing of beads in pulldown buffer (50 mM HEPES, 200 mM NaCl, 10 mM CaCl 2 , and protease inhibitors, pH 7.5). HEK293 cells were transfected with wild-type or mutant pEGFP-N1-AnxA2 or pEGFP-N1-S100A10 or pEGFP-N1-AnxA1 and harvested in ice-cold pulldown buffer containing 1% Triton X-100; the resulting HEK293 lysates (10 -15 g/l) were incubated with GST fusion protein-coupled beads at 4°C overnight upon agitation. The beads were washed three times with pulldown buffer and boiled once in Laemmli buffer to elute bound proteins, and the eluates were evaluated by immunoblotting.
Peptide Spot Arrays-GST-fused N-terminal tails of wildtype NKCC2 as well as its phosphomimetic or nonphosphomimetic mutants (5 nM) were spotted onto nitrocellulose membranes and incubated with 5% skim milk in TBS adjusted with 1% Tween 20 to block unspecific bindings. After that, wild-type or phospho-AnxA2 mutants (Y24D or S26D) or S100A10 (each 0.1 M in blocking solution) were applied overnight at 4°C. Interactions were evaluated by immunoblotting using anti-AnxA2 or anti-S100A10 antibodies. GST was used as negative control, and AnxA2 or S100A10 was spotted on the membrane as positive controls.
Rip/Flip Immunoelectron Microscopy-Preparation of plasma membrane sheets was performed as described (50,51). Briefly, raTAL cells were grown to confluence on glass coverslips, fixed for 15 min in 0.5% paraformaldehyde/PBS, washed in PBS, and subsequently inverted on glow-discharged nickel electron microscopy grids coated with poly-L-lysine. Gentle pressure was applied to the coverslip for 15 s using a fine pair of forceps. The coverslips were then lifted, leaving portions of the upper cell surface adherent to the poly-L-lysine-coated grid. Membranes were fixed in 2% paraformaldehyde for 20 min at room temperature. Immunogold labeling of AnxA2, S100A10, flotillin-1, or transferrin receptor was performed using the respective primary antibodies and 5-or 10-nm gold-conjugated secondary antibodies (Abcam). For double-labeling of the grids, anti-NKCC2 antibody and 5-or 10-nm gold-conjugated secondary antibody (Abcam) were sequentially applied thereafter. Grids were then fixed in 2% glutaraldehyde in PBS, contrasted in 1% aqueous tannic acid and 1% aqueous uranyl acetate, washed with distilled H 2 O, and examined by transmission electron microscopy (EM; Zeiss E905).
RNA Interference-Small interference RNA (siRNA) for AnxA2 (Santa Cruz Biotechnology) or scramble siRNA (negative control) was repeatedly transfected (two transfections with 8-h time intervals) in 40 -50% confluent raTAL cells using HiPerFect (Qiagen). After that, cells were grown to confluence for another 50 h. For a rescue experiment, cells were transiently transfected with pEGFP-N1-AnxA2 (JetPEI) and incubated for 24 h. Cells were then lysed for RNA isolation followed by quantitative real time PCR and immunoblotting evaluations.
mRNA Extraction, cDNA Synthesis, and Quantitative PCR (qPCR) Analysis-RNA extraction from cellular or tissue lysates was performed using the TRIzol reagent (Invitrogen) according to the manufacturer's protocol, and cDNA was syn-thesized by reverse transcription (BioScript, Bioline). qPCR was performed using HOT FIREPol EvaGreen qPCR Mix Plus (Solis BioDyne) and primer pairs specific for AnxA2 (NM_007585.3), NKCC2 (NM_001270618.1), and GAPDH (forward, 5Ј-tgcaccaccaactgcttagc-3Ј; reverse, 5Ј-ggcatggactgtggtcatgag-3Ј) in the Real Time PCR 7500 system (Applied Biosystems). All samples were determined in triplicate. Threshold cycle (C t ) values were quantified and normalized against GAPDH (⌬C t ), and the calibrator was subtracted (⌬⌬C t ). Finally, to perform relative quantification, an exponentiation was calculated by use of 2 to the power of Ϫ⌬⌬C t (comparative C t method) (52).
Analysis of Data-All values are the means Ϯ S.D. Cohorts were compared with Student's two-tailed unpaired t test. p Ͻ 0.05 was accepted as statistically significant.

Identification of AnxA2 as an Interaction Partner of NKCC2
by a Yeast Two-hybrid Screen-To identify candidates that may interact with NKCC2, we first performed yeast two-hybrid screens. Baits consisting of two overlapping N-terminal fragments of human NKCC2 were used to screen a human cDNA library. The choice of the baits was based on the presence of multiply conserved, functionally relevant NKCC2 phosphoacceptor sites (Thr-100, Thr-105, Thr-118, and Ser-130). Yeast two-hybrid screening revealed 36 potential interaction partners of NKCC2 (Table 1). Subsequent ranking of the candidates by their renal expression levels identified AnxA2 as the product with the strongest expression in rat and human TAL. Because annexins may serve as trafficking factors for various transmem- brane proteins (29,(53)(54)(55), we hypothesized that AnxA2 may be involved in the trafficking of NKCC2.
Distribution of NKCC2 and AnxA2 in Lipid Rafts-The cellular distribution of AnxA2 was analyzed in rat kidney with a focus on the TAL. Because we have shown before that trafficking of NKCC2 in TAL depends on its distribution in rafts (21), we further analyzed its distribution together with AnxA2 in lipid rafts. There was substantial apical colocalization of AnxA2 with NKCC2 in medullary (m)TAL and macula densa (MD) cells, whereas cortical (c)TAL exhibited weak or no AnxA2 signal by confocal microscopy (Fig. 1, A and B). S100A10 protein, also termed p11 or AnxA2 light chain, mediates protein interactions with AnxA2 in a scaffolding manner (36,56,57). S100A10 was colocalized as well with NKCC2 in mTAL and MD but was absent from cTAL cells (Fig. 1, A and B). To study whether these components are also colocalized at a high resolution level and whether they are coexpressed within lipid rafts, sheets of the apical membrane were obtained from cultured raTAL cells using the rip/flip technique. We then used EMimmunogold staining of the sheets and confirmed an intimate colocalization of AnxA2 and S100A10 with NKCC2 in dark, electron-dense subdomains. Numerical evaluation confirmed that signals were largely restricted to the dark domains throughout (Fig. 1C). To further characterize the distribution of NKCC2 within the dark domains, which have earlier been referred to as rafts (51,58), double staining with the established raft marker flotillin-1 (59, 60) showed predominant colocalization of the two products in the dark domains, whereas the nonraft protein, transferrin receptor (61,62), was found outside of the dark domains in the light surroundings, mostly separated from NKCC2 signals. Quantitative evaluation underlined that the majority of NKCC2 signal was restricted to the dark domains likely to represent lipid rafts (Fig. 1D). To corroborate these data biochemically, detergent-resistant membrane rafts were isolated from rat kidney medullary homogenate using extraction with Triton X-100 and subsequent sucrose gradient centrifugation. The resulting detergent-insoluble fractions contained NKCC2, AnxA2, and S100A10 as well as flotillin-1 and another raft protein, glycosylphosphatidylinositol-anchored Tamm-Horsfall protein (Fig. 2, A and B) (63). In aggregate, these data thus support a codistribution of AnxA2, S100A10, and NKCC2 within apical membrane rafts. . NKCC2 and AnxA2 physically associate in rafts. A, representative immunoblots showing IP of NKCC2 and AnxA2 from medullary rat kidney homogenates. AnxA2 and S100A10 co-IP with NKCC2 as bait, and mutually, NKCC2 and S100A10 co-immunoprecipitate with AnxA2 as bait. IgG indicates control immunoprecipitation with nonspecific IgG. B, immunoblots after NKCC2 IP from Triton X-100-insoluble fractions of discontinuous sucrose gradient show co-IP band for AnxA2 and S100A10. Flotillin-1 serves as positive raft marker. C, densitometric evaluation of co-immunoprecipitated AnxA2 and S100A10 signals normalized to respective input signals revealed ϳ3-fold stronger interaction of NKCC2 with AnxA2 as compared with S100A10. Data are the means Ϯ S.D.; **, p Ͻ 0.01. D and E, MALDI-TOF mass spectra (left) and MALDI-TOF/TOF-MS/MS spectra (right) obtained by mass spectrometric evaluation of the eluates after NKCC2 IP from the insoluble gradient fractions confirm the presence of NKCC2 (D) and AnxA2 (E), supporting their physical association within rafts. Arrows point to NKCC2-or AnxA2-specific peptides (left panels). AU, arbitrary units.

NKCC2 N Terminus Directly
Interacts with AnxA2-Rat medullary kidney extracts or derived raft-enriched fractions were used for co-immunoprecipitation (co-IP) experiments to study potential interactions of NKCC2 with AnxA2 or S100A10 (Fig. 3, A-E). These experiments revealed substantial physical interaction of NKCC2 with AnxA2 but less so with S100A10 (Fig. 3, A-C). To specify interactions at the molecular level, the recombinant N terminus of NKCC2 and the recombinant, complete AnxA2 or S100A10 proteins were studied for interaction using peptide spot assay (Fig. 4, A-D). Resulting signals indicated binding between the N terminus of NKCC2 with AnxA2 but not S100A10 (Fig. 4A). To specify the role of N-ter-minal NKCC2 phosphorylation (Thr-96, Thr-101, Thr-114, and Ser-126) in its interaction with AnxA2, phosphomimetic and nonphosphomimetic mutants were tested in spot assays; AnxA2 selectively bound to the nonphosphomimetic mutants (Fig. 4A). These data were confirmed by GST pulldown assays using the wild-type and mutant N termini of NKCC2 as baits (Fig. 4, E and F).
Short Term Administration of Vasopressin Increases Surface Expression and Lipid Raft Association of NKCC2 and AnxA2-AVP is a potent endocrine activator of NKCC2 that promotes its apical trafficking (12,21,33). To study the effects of AVP on AnxA2 and S100A10, AVP-deficient Brattleboro rats with central DI were treated with the V2 receptor agonist, desmopressin (dDAVP, 30 min). Immunoblots after tissue fractionation revealed shifts of AnxA2 and NKCC2 by signal increases in the plasma membrane-enriched fractions (ϩ29% for AnxA2 and ϩ28% for NKCC2) compared with decreases in the cytoplasmic fractions (Ϫ29% for AnxA2 and Ϫ27% for NKCC2; p Ͻ 0.05); changes of S100A10 were absent (Fig. 5A). Confocal evaluation confirmed apical shifting of the AnxA2 signal, whereas the S100A10 signal was practically unaltered. Because of its strictly subapical localization at baseline, a shift of NKCC2 to the surface could not be visualized (Fig. 5B). dDAVP further induced parallel shifts of NKCC2 and AnxA2 into the low density sucrose gradient fractions from medullary extracts (ϩ13% for NKCC2 and ϩ19% for AnxA2; p Ͻ 0.05), suggesting enhanced raft association of both products (Fig. 6A). The interaction of NKCC2 with AnxA2 was increased 2-fold in the raft fractions as revealed by co-IP (Fig. 6B). Activation of the V2 receptor thus concomitantly stimulated luminal trafficking and raft association of NKCC2 and AnxA2 and corroborated their interaction. Because AnxA2 appears to exclusively interact with nonphosphorylated NKCC2, we suggest that AVP-induced trafficking events precede the phosphorylation of the cotransporter.
AnxA2 Facilitates Membrane Recruitment of NKCC2 upon Low Chloride Hypotonic Stress in Cultured Cells-Low chloride hypotonic stress is an established condition to promote apical trafficking and surface expression of NKCC2 in heterologous models (3,12,14,15). We therefore tested whether under this condition trafficking of endogenous NKCC2 was affected by the presence of AnxA2 or S100A10. To this end, cultured MD cells (MMDD1) were transfected with vectors coding for AnxA2-GFP, S100A10-GFP, or GFP alone and stimulated with low chloride hypotonic stress for 1 h. Cells were then fractionated, and the resulting plasma membrane pool was further extracted with cold detergent. The obtained raft fractions revealed markedly increased signals for NKCC2 upon transfection with AnxA2-GFP (ϩ239%; p Ͻ 0.01), whereas S100A10-GFP or GFP alone had no effects (Fig. 7, A and B). The specific role of AnxA2 was further illustrated cytochemically. Upon stimulation, a clear redistribution from the cytosol to the plasma membrane was observed for AnxA2-GFP but less so for S100A10-GFP, and GFP alone showed no effect (Fig. 7C). Data therefore suggest that AnxA2 facilitates luminal trafficking and raft association of NKCC2 upon stimulation. To further illustrate the role of AnxA2 in NKCC2 trafficking, AnxA2 deficiency was induced by RNAi-mediated knockdown of its expression. In contrast to MMDD1 cells, which had low endog- . AnxA2 but not S100A10 directly and selectively interact with nonphosphorylated NKCC2 species. A, representative peptide spot assays of interactions between the GST-fused, purified N-terminal tails of wild-type (WT) NKCC2 as well as its phosphomimetic or nonphosphomimetic mutants with AnxA2 or S100A10. WT and mutant NKCC2 N termini were spotted to nitrocellulose membranes, incubated with AnxA2 or S100A10, and detected by specific antibodies to AnxA2 or S100A10. AnxA2 selectively binds to WT NKCC2 and the nonphosphomimetic mutants. Note the lack of interactions between WT or mutant NKCC2 N termini and S100A10. B, specificity of the assays and the lack of antibody cross-reactivity were tested in control spots. C, Coomassie Blue staining of recombinant WT AnxA2 and S100A10. D, control immunoblot of GST-fused recombinant WT NKCC2 N terminus and its phosphomimetic and nonphosphomimetic mutants produced in E. coli. E, control immunoblot showing GFP-tagged AnxA2, S100A10, or AnxA1 produced in HEK293 cells. F, immunoblots of eluates after GST pulldown assays using phosphomimetic and nonphosphomimetic NKCC2 mutants as baits and lysates from HEK293 cells transfected with GFP-tagged AnxA2, S100A10, or AnxA1 (asterisk). Note strong interactions of AnxA2 with the WT and nonphosphomimetic NKCC2 mutants but less so with the phosphomimetic NKCC2 mutants. NKCC2 N termini show no interaction with S100A10 or AnxA1.
enous AnxA2 expression, cultured raTAL cells spontaneously displayed substantial abundance of AnxA2 and were therefore more suitable. Transfection with AnxA2 siRNA resulted in a significant knockdown of AnxA2 (Ϫ67% at the mRNA level; p Ͻ 0.01; and Ϫ77% at the protein level; p Ͻ 0.01) after 72 h; the abundance of NKCC2 was unaffected under this condition (Fig.   8, A and B). Low chloride hypotonic stress for 1 h led to efficient trafficking of NKCC2 to the surface of control cells, whereas in AnxA2-deficient cells, NKCC2 distribution was unchanged (Fig. 8C). In addition, control rescue experiments were performed using transient transfection of AnxA2-GFP subsequent to its knockdown. The resulting overexpression of AnxA2-GFP markedly increased AnxA2 abundance and normalized the cellular response to the low chloride hypotonic stress (Fig. 8, AnxA2-NKCC2 Interaction Depends on AnxA2 Phosphorylation-Generally, interactions of AnxA2 with other proteins depend on phosphorylation and/or dephosphorylation at its Tyr-24 residue or at residues Ser-12 and Ser-26 (24, 57, 64 -66). We have therefore mutated Tyr-24 to mimic its constitutive phosphorylation (Y24D) or dephosphorylation (Y24A), as well as Ser-26 to mimic its phosphorylation (S26D), and we analyzed these mutants for their ability to interact with NKCC2 in MMDD1 cells. Compared with control GFP transfection, overexpression of AnxA2-Y24D-GFP strongly facilitated surface expression and raft association of NKCC2 in response to low chloride hypotonic stress for 1 h (ϩ510% in Triton-insoluble fractions of plasma membrane; p Ͻ 0.05), whereas AnxA2-S26D-GFP and AnxA2-Y24A-GFP had only minor effects (Fig.  9, A and B). Next, interactions between the AnxA2 mutants and the NKCC2 N terminus were analyzed by peptide spot and GST pulldown assays using its phosphomimetic and nonphosphomimetic mutants (Fig. 10, A-E). As expected, AnxA2-Y24D readily interacted with the WT NKCC2 N terminus and its nonphosphomimetic mutants but not with its phosphomimetic FIGURE 5. V2 receptor stimulation of rats with diabetes insipidus stimulates trafficking of NKCC2 and AnxA2 but not S100A10. A and B, rats were analyzed after 30 min of treatment with desmopressin (dDAVP) or vehicle. A, immunoblots of plasma membrane (pm)-enriched and cytoplasmic (cyt) vesicle containing fractions from medullary kidney homogenates of dDAVP-or vehicle-treated rats. Note significant dDAVP-induced luminal shifts of AnxA2 and additional NKCC2 as reflected by their increased signals in the plasma membrane fractions; signals in cytoplasmic fractions are concomitantly decreased. Distribution of S100A10 signals is unchanged. Data are the means Ϯ S.D.; *, p Ͻ 0.05. B, confocal images show increase of apical AnxA2 signal, whereas S100A10 is unchanged. Merge images show nuclei counterstained in blue; scale bars, 10 m.

FIGURE 6. V2 receptor stimulation enhances the interaction between
AnxA2 and NKCC2 and corroborates their raft association. A, immunoblots showing distribution of NKCC2 and AnxA2 along discontinuous sucrose gradient from the Triton X-100-insoluble fractions obtained from outer medullary (OM) extracts of vehicle-or dDAVP-treated DI rats. Densitometry shows significant shifts from the higher (unframed) to the lower density fractions (red frames). B, immunoblots showing IP of NKCC2 and co-IP of AnxA2 from the low density fractions. The experiment was controlled by IP using nonspecific IgG. Densitometry of AnxA2-to-NKCC2 signal ratio reflects the strength of their interaction. The data are the means Ϯ S.D.; *, p Ͻ 0.05. mutants (Fig. 10, A and D). In contrast, AnxA2-S26D and AnxA2-Y24A mutants did not exhibit significant interactions either with WT or mutant N-terminal tails of NKCC2 (Fig. 10,  B and D). These results were corroborated by detection of phosphorylated AnxA2 with an anti-phospho-Tyr-24-AnxA2 antibody (pY24-AnxA2) in eluates from GST pulldown assays with rat kidney medulla lysates. Here, the WT NKCC2 N terminus, as well as the mutants mimicking its dephosphorylation, bound the phosphorylated AnxA2, whereas no interactions between the pY24-AnxA2 and the NKCC2 phosphomimetic mutants were detected (Fig. 10F). Overall, these results suggest that phosphorylation of AnxA2 at its Tyr-24 residue promotes its interaction with dephosphorylated NKCC2, which in turn facilitates the apical trafficking of the transporter.

DISCUSSION
Activity of a cell membrane transporter or channel, such as NKCC2, principally depends on its surface expression. Our previous work has shown that the distribution of NKCC2 in apical lipid rafts facilitates its anterograde trafficking to the cell surface (21). A role for lipid rafts during endocytic retrieval of NKCC2 has been suggested as well (22); because clathrin, otherwise relevant for its endocytosis, is not a raft-associated protein, this process appears to require uncoupling of at least a proportion of the cell membrane's NKCC2 pool from clathrin (67). Partitioning of NKCC2 to rafts thus appears to substan-tially determine steps of its surface expression and membrane half-life (21,23). The detailed impact of NKCC2's raftophilicity in the context of its cellular trafficking was, however, little understood. A potential key to better understand this lies in the fact that lipid rafts may as well serve as scaffolding platforms to facilitate interactions of NKCC2 with other proteins involved in its trafficking (21,23). Here, we have established AnxA2 as a permissive element in the apical trafficking and surface expression of NKCC2, with both products codistributed in lipid rafts. Furthermore, AnxA2 promoted the association of NKCC2 with rafts, which was probably related with the known property of annexins to bind to negatively charged phospholipids, primarily phosphoinositides, and to stabilize lipid rafts (35, 68 -70).
Mechanisms of anterograde trafficking of NKCC2 have been addressed previously. In particular, the role of conserved C-terminal motifs of NKCC2 have been elucidated with respect to its apical sorting (71)(72)(73). Here we have shown that the N-terminal tail of NKCC2 is relevant for its surface expression as well, and that intact trafficking requires a direct interaction of its N-terminal domain with AnxA2. This domain encloses several conserved, functionally relevant phosphoacceptor sites (13)(14)(15). It has been assumed earlier that phosphorylation of the cotransporter may interfere with its surface expression, although it was unclear whether NKCC2 becomes phosphorylated before, by the time of, or after its insertion in the apical membrane (14,FIGURE 7. Low chloride hypotonic stress enhances raft association of NKCC2 and AnxA2. A, immunoblots showing cellular distribution of transiently transfected AnxA2-GFP, S100A10-GFP, and GFP and endogenous NKCC2 in MMDD1 cells under normal chloride condition or low chloride hypotonic stress; the cytoplasmic (cyt) and the Triton X-100-soluble (TS) and -insoluble (TI) membrane fractions are shown. Densitometry demonstrates that AnxA2 but not S100A10 or GFP are shifted to the Triton-insoluble fraction. GAPDH was detected as marker of the cytoplasmic fractions, and flotillin-1 was used as a raft marker. Data are the means Ϯ S.D.; *, p Ͻ 0.05. B, densitometric quantification of NKCC2 abundance in the Triton-insoluble membrane fraction from AnxA2-GFP-, S100A10-GFP-, or GFP-transfected cells kept under normal chloride condition or low chloride hypotonic stress. AnxA2-GFP enhances raft association of NKCC2 more than S100A10-GFP or GFP alone. Data are the means Ϯ S.D.; *, p Ͻ 0.05; **, p Ͻ 0.01 for differences between low and normal chloride conditions; #, p Ͻ 0.01 for differences between the transfections. n.s., no significance. C, confocal images of the transfected MMDD1 cells under the different chloride conditions. Note increases of plasma membrane-associated AnxA2 (arrowheads) but not of S100A10 signals upon low chloride hypotonic stress; scale bars, 10 m. 15). Our present data show that, as far as its interaction with AnxA2 is concerned, trafficking of NKCC2 requires that its N-terminal residues be nonphosphorylated, which implies that NKCC2 membrane insertion precedes its phosphorylation (Fig.  11). This interpretation agrees with the observation that a predominant proportion of phosphorylated NKCC2 is localized in the apical membrane, where it is also colocalizing and interacting with its activating kinases, SPAK and OSR1 (21,74,75). The fact that NKCC2 surface expression was unaltered in transgenic mouse models presenting a dramatic decrease or increase of NKCC2 phosphorylation such as SORLA-deficient or SPAKdeficient mice, respectively, further suggests that membrane insertion of the cotransporter does not require its phosphorylation and is thus compatible with AnxA2's permissive role in NKCC2 apical trafficking (74 -76).
In the context of raft formation and stabilization, AnxA2 was further proposed to direct cytoplasmic vesicles or domains of organelles to the cytoskeleton to prepare for or to permit their intracellular translocation (69,77,78). The present fine structural analysis demonstrates colocalization of AnxA2 and NKCC2 within the same membrane microdomains that were identified as lipid rafts. Along with the additional biochemical evidence of their codistribution in lipid rafts and the evidence of their physical interaction, a role for AnxA2 may as well be suggested in NKCC2 vesicular trafficking. In this respect, cellular translocation of several other transmembrane proteins, including ion transporters and signaling receptors, has been shown to depend on the availability of or the interaction with AnxA2 (24, 25, 79 -81). In particular, AnxA2 was shown to determine cAMP-induced trafficking of aquaporin 2 (30). Because both aquaporin 2 and NKCC2 are major targets of AVP (3,11,82,83), AnxA2 appears to have a significant impact in AVP signaling in the two segments harboring these products, TAL and collecting duct, thereby serving to promote urine concentration. The observed AVP-induced stimulation of the interaction intensity between AnxA2 and NKCC2 as well as the pronounced apical shift of AnxA2 upon administration of the extrinsic AVP agonist dDAVP in AVP-deficient DI rats has provided additional supporting evidence for this hypothesis (Fig.  11). Because AnxA2 may be functional in its monomeric form or as a heterotetrameric complex together with its binding partner, protein S100A10, we have tested whether protein S100A10 Relative mRNA values were obtained using the comparative C t method and normalization to GAPDH expression. Data are the means Ϯ S.D.; **, p Ͻ 0.01. B, immunoblots from raTAL cell lysates and densitometric quantification. Note that NKCC2 abundance was generally unaffected. Tubulin serves as loading control; **, p Ͻ 0.01. C, confocal images of transfected raTAL cells kept under normal chloride condition or low chloride hypotonic stress; scale bars, 20 m. Representative confocal profiles of AnxA2 and NKCC2 signal intensities scored across two neighboring cells (lines) are shown at the right of the corresponding images. Note that the recruitment of NKCC2 into the plasma membrane under low chloride hypotonic stress was blunted after the AnxA2 knockdown but normalized by the rescue of AnxA2 (arrowheads).
plays a role in the present context as well. Our binding assays demonstrated that, despite the co-distribution of AnxA2, protein S100A10, and NKCC2 in lipid rafts, the interaction between AnxA2 and NKCC2 did not require the presence of protein S100A10. Whereas the physical interaction between AnxA2 and NKCC2 may facilitate trafficking and partitioning to rafts, the tetrameric form of AnxA2 may be functional in stabilizing apical rafts in TAL with the help of protein S100A10 (35,69,70,84,85).
binding of AnxA2 to the transporter. PKA is known to act upstream of calcineurin and thus mediates dephosphorylation of AnxA2 at Ser-26 whereby phosphorylation at Tyr-24 could be promoted (57, 86 -88). These results are in line with previously described stimulatory effects of PKA activation on the luminal trafficking of NKCC2 (19,89). It has been shown that PKA mediates effects of AVP by promoting the anterograde trafficking, as well as recycling of NKCC2 thus increasing its surface expression and activity (3,10,11). However, the components operating downstream of the PKA effects have not been elucidated so far (3). Here, we show that the AnxA2 mutant mimicking AnxA2 phosphorylation at Tyr-24 strongly facilitates membrane insertion of NKCC2 upon low chloride hypotonic stress, whereas the AnxA2 mutants mimicking Tyr-24 dephosphorylation or PKC-induced Ser-26 phosphorylation have no stimulatory effects on NKCC2 surface expression. It is therefore tempting to speculate that AnxA2 participates in stimulated apical trafficking/recycling of NKCC2 downstream of PKA activation. Thus, our data provide several lines of evidence that AnxA2 is involved in stimulated NKCC2 trafficking. At the same time, our AnxA2 knockdown or overexpression experiments induced only moderate changes of steady state NKCC2 surface expression suggesting that AnxA2 is not necessary for baseline NKCC2 function but plays a role in the rapid membrane recruitment of the transporter upon stimulation.
Acute increase of NKCC2 surface expression upon appropriate stimulation by AVP or low chloride may involve several synergistic mechanisms such as promotion of anterograde NKCC2 trafficking, rapid membrane insertion of NKCC2-containing vesicles from the apical pool, stimulation of NKCC2 recycling, and probably also inhibition of NKCC2 endocytosis (3,8,9). It is currently unclear which one of these processes depends on AnxA2; however, based on its apical distribution in TAL cells, its proposed PKA sensitivity, and its known ability to facilitate the fusion of intracellular vesicles with the FIGURE 10. Tyr-24 phosphorylation of AnxA2 promotes its direct interaction with NKCC2. A and B, peptide spot assays of interactions between the GST-fused, purified N-terminal phosphomimetic or nonphosphomimetic NKCC2 mutants and AnxA2 mutants mimicking its phosphorylation (Y24D or S26D). NKCC2 mutants were spotted to nitrocellulose membranes, incubated with AnxA2-Y24D (A) or AnxA2-S26D (B), and detected by anti-AnxA2 antibody. Specificity of the assays and lack of antibody cross-reactivity were verified by detection of the control spots using anti-AnxA2 antibody. A, note that the AnxA2-Y24D mutant selectively binds to the nonphosphomimetic NKCC2 mutants. B, AnxA2-S26D shows no interactions with either mutant of NKCC2. C, Coomassie Blue staining of recombinant AnxA2-Y24D and AnxA2-S26D. D, immunoblots of eluates after GST pulldown assays using WT N terminus of NKCC2 and its phosphomimetic and nonphosphomimetic mutants as baits and lysates from HEK293 cells transfected with either GFP-tagged AnxA2-Y24D, AnxA2-S26D, AnxA2-Y24A, or AnxA1 (asterisk). There are strong interactions of AnxA2-Y24D with the WT and nonphosphomimetic NKCC2 mutants but less so with the phosphomimetic mutants. AnxA2-S26D and AnxA2-Y24A mutants show only weak interactions with either NKCC2 mutant, whereas AnxA1 does not interact with NKCC2 N termini in this setting. E, control immunoblot showing GFP-tagged AnxA2 mutants (Y24D, S26D, and Y24A), S100A10, or AnxA1 produced in HEK293 cells. F, GST pulldown assay with NKCC2 N-terminal mutants and lysates from rat kidney medulla showing strong interactions of pY24-AnxA2 with WT NKCC2 N terminus and its nonphosphomimetic mutants but not with the phosphomimetic mutants.
In summary, we have identified AnxA2 as a novel factor for NKCC2 trafficking that facilitates surface expression of the cotransporter upon stimulation by AVP or low chloride hypotonic stress. While AVP is a major NKCC2 activating hormone, low chloride hypotonic stress leads to a decrease in luminal salt load as it may occur in TAL during physiologic fluctuations or pathophysiologic perturbations. We therefore believe that AnxA2 is involved in the dynamic adjustment of TAL transport function in response to endocrine stimuli or changes in luminal chloride load.