A Highly Conserved Motif at the COOH Terminus Dictates Endoplasmic Reticulum Exit and Cell Surface Expression of NKCC2*

Mutations in the apically located Na+-K+-2Cl− co-transporter, NKCC2, lead to type I Bartter syndrome, a life-threatening kidney disorder, yet the mechanisms underlying the regulation of mutated NKCC2 proteins in renal cells have not been investigated. Here, we identified a trihydrophobic motif in the distal COOH terminus of NKCC2 that was required for endoplasmic reticulum (ER) exit and surface expression of the co-transporter. Indeed, microscopic confocal imaging showed that a naturally occurring mutation depriving NKCC2 of its distal COOH-terminal region results in the absence of cell surface expression. Biotinylation assays revealed that lack of cell surface expression was associated with abolition of mature complex-glycosylated NKCC2. Pulse-chase analysis demonstrated that the absence of mature protein was not caused by reduced synthesis or increased rates of degradation of mutant co-transporters. Co-immunolocalization experiments revealed that these mutants co-localized with the ER marker protein-disulfide isomerase, demonstrating that they are retained in the ER. Cell treatment with proteasome or lysosome inhibitors failed to restore the loss of complex-glycosylated NKCC2, further eliminating the possibility that mutant co-transporters were processed by the Golgi apparatus. Serial truncation of the NKCC2 COOH terminus, followed by site-directed mutagenesis, identified hydrophobic residues 1081LLV1083 as an ER exit signal necessary for maturation of NKCC2. Mutation of 1081LLV1083 to AAA within the context of the full-length protein prevented NKCC2 ER exit independently of the expression system. This trihydrophobic motif is highly conserved in the COOH-terminal tails of all members of the cation-chloride co-transporter family, and thus may function as a common motif mediating their transport from the ER to the cell surface. Taken together, these data are consistent with a model whereby naturally occurring premature terminations that interfere with the LLV motif compromise co-transporter surface delivery through defective trafficking.

The Na-K-2Cl co-transporter, NKCC2, provides the major route for sodium/chloride transport across the apical plasma membrane of the thick ascending limb (TAL) 3 of the kidney (1). This co-transporter is critical for salt reabsorption, acid-base regulation, and divalent mineral cation metabolism (2). The prominent importance of NKCC2 in renal functions is evidenced by the effect of loop diuretics, which as pharmacologic inhibitors of NKCC2, are extensively used in the treatment of edematous states (2). Even more impressive, inactivating mutations of the NKCC2 gene in humans causes Bartter syndrome type 1 (BS1), a life-threatening renal tubular disorder for which the diagnosis is usually made in the antenatal-neonatal period, due to the presence of polyhydramnios, premature delivery, salt loss, hypokalemia, metabolic alkalosis, hypercalciuria, and nephrocalcinosis (3). Without appropriate treatment, patients with BS1 will not survive the early neonatal period (4). In congruence with the severity of the symptoms and the uniformity of the clinical picture, functional analysis of diverse NKCC2 mutants consistently revealed a loss of function effect of the tested mutations (5,6). However, regulatory characterizations of mutants NKCC2 were limited to Xenopus laevis oocytes. Indeed, studies aimed at understanding the post-translational regulation of NKCC2 have been hampered by the difficulty of expressing the co-transporter protein in mammalian cells (7,8). As a consequence, our knowledge of the molecular mechanisms underlying membrane trafficking of mutated NKCC2 proteins in mammalian cells is nil. Increasing our understanding of the molecular determinants underlying NKCC2 expression in renal cells is essential for elucidating the pathophysiology of BS1 and for improving the available treatments (9,10). Undeniably, only analysis of the expression such NKCC2 of mutants in renal cells would definitively establish their cellular fate.
NKCC2 belongs to the superfamily of electroneutral cationcoupled chloride (CCC) co-transporters (SLC12A) (1). The cation-chloride co-transporters (CCCs) family comprises two principal branches of homologous membrane proteins. One branch includes the Na ϩ -dependent chloride co-transporters composed of the Na ϩ -K ϩ -2Cl Ϫ co-transporters (NKCC1 and NKCC2) and the Na ϩ -Cl Ϫ co-transporter (NCC). The second branch includes the Na ϩ -independent K ϩ -Cl Ϫ co-transporters composed of at least four different isoforms: KCC1 KCC2, KCC3, and KCC4 (11). Within the families, the CCCs share 25-75% amino acid identity. All of these co-transporters exhibit similar hydropathy profiles with 12 transmembranespanning domains, an amino terminus of variable length, and a long cytoplasmic carboxyl terminus. Because the COOH-terminal domain of NKCC2 is the predominant cytoplasmic region, it is likely to be a major factor in the trafficking of the NKCC2 protein. Moreover, there have been several reports demonstrating that COOH-terminal residues are important for correct protein targeting (12)(13)(14). Occasionally, COOH-terminal mutations are known to cause genetic disorders (15)(16)(17). Although studies of other ion transporters support the importance of the COOH-terminal signals in protein stability, maturation, surface delivery, and ER export (18 -22), little is known about the role of COOH-terminal signals in the biogenesis of NKCC2.
We were recently able to express NKCC2 protein in mammalian cells (23), providing therefore a powerful tool to study and understand the molecular mechanisms underlying the cotransporter expression and regulation in renal cells. This allowed us, in this study, to take the advantage of the existence of natural mutants altering the COOH-terminal tail of the cotransporter to investigate the role of the COOH terminus in the biogenesis of NKCC2 and to explore possible mechanisms implicated in BS1. The results demonstrate the importance of the COOH terminus in normal maturation of the NKCC2 protein. Indeed, we identified a motif of three hydrophobic residues, 1081 LLV 1083 , highly conserved in the COOH-terminal tails of all members of the CCC family, that controls the rate of ER export and thus of surface expression of NKCC2. Loss of the motif disrupts glycosylation and plasma membrane localization of NKCC2. Therefore, we propose abnormal trafficking as a common BS1 mechanism associated with mutations depriving NKCC2 of its COOH terminus.

EXPERIMENTAL PROCEDURES
Materials-All chemicals were obtained from Sigma unless otherwise noted. Penicillin and streptomycin were from Invitrogen.
Subclonings were carried out with the following vectors: 1) pCMV-Myc and pCMV-HA (Clontech), containing c-Myc epitope or HA tag, a multiple cloning site, and an ampicillin resistance gene; 2) pEGFP-C2 (Clontech), which contains the green fluorescent protein (GFP) gene, a multiple cloning site, and a KANAr resistance gene; and 3) pcDNA3.1/V5-His-TOPO (Invitrogen), which contains epitope V5, a multiple cloning site, and an ampicillin resistance gene.
Cell Culture-Opossum kidney (OKP) cells were grown in DMEM complemented with 10% fetal bovine serum (Invitrogen), penicillin (100 units/ml), and streptomycin (100 units/ml) at 37°C in a humidified atmosphere containing 5% CO 2 . Human embryonic kidney (HEK) 293 cells were maintained in DMEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. For DNA transfection, cells were grown to 60 -70% confluence on plastic culture dishes and then were transiently transfected for 5 h with plasmids using the Lipofectamine Plus kit according to the manufacturer's instructions (Invitrogen). For protein degradation assays, transiently transfected cells were treated with MG132 (10 M) or leupeptin (100 M) for 6 -12 h prior to cell lysis.
Measurement of Intracellular pH and Na-K-2Cl Co-transporter Activity-Measurement of cytoplasmic pH (pH i ) was accomplished in cells grown to confluence on coverslips using the intracellularly trapped pH-sensitive dye 2Ј,7Ј-bis(carboxyethyl)-5,6-carboxyfluorescein. pH i was estimated from the ratio of fluorescence with excitation wavelengths of 495 and 450 nm and emission wavelength of 530 nm (Horiba Jobin Yvon, Longjumeau, France). Calibration of the 2Ј,7Јbis(carboxyethyl)-5,6-carboxyfluorescein excitation ratio was accomplished using the nigericin technique. Na-K-2Cl cotransport activity was measured as bumetanide-sensitive NH 4 influx, as previously described (23,24). Cells were first bathed at 37°C in a CO 2 -free Hepes/Tris-buffered medium containing 135 mM NaCl, 10 mM Hepes (pH 7.4), 5 mM KCl, 1 mM MgCl 2 , 0.8 mM KH 2 PO 4 , 0.2 mM K 2 HOP 4 , 1 mM CaCl 2 , and 10 mM BaCl 2 to measure base-line pH i . 20 mM NH 4 Cl was then added to the medium, inducing a very rapid initial cellular alkalinization followed by a pH i recovery. The initial rate of intracellular pH recovery (dpH i /dt) was measured over the first 20 s of records, as reported earlier (23,25). The ⌬pH i caused by NH 4 Cl addition was used to calculate the cell buffer capacity, which was not different between the studied groups (data not shown). There-fore, the Na-K-2Cl co-transporter activity is expressed as dpH i /dt. Protein Preparation, Immunoblotting, and Immunoprecipitation-Forty-eight hours post-transfection, cells were washed three times with cold PBS and lysed in 0.2 or 0.5 ml of lysis buffer (120 mM Tris/Hepes, pH 7.4, 150 mM NaCl, 5 mM EDTA, 3 mM KCl, 1% (v/v) Triton X-100) containing protease inhibitors (Complete 1697498; Roche Applied Science). Samples were harvested, sonicated, and centrifuged at 16,000 rpm for 15 min at 4°C. Protein expression levels were assessed after normalizing and loading equal amounts of total protein for 7.5% SDS-PAGE separation and Western blotting with anti-Myc, anti-HA, anti-V5, and anti-actin antibodies. For immunoprecipitation, cells were solubilized with lysis buffer containing 0.4 M NaCl, 0.5 mM EGTA, 1.5 mM MgCl 2 , 10 mM Hepes, pH 7.9, 5% (v/v) glycerol, 0.5% (v/v) Nonidet P-40 and protease inhibitors (Complete; Roche Applied Science). Immunoprecipitation was carried out using anti-Myc antibody followed by affinity purification using protein G-agarose beads (Dynal). After incubation with protein G-agarose beads for 1 h at room temperature, the immunocomplex was washed three times in PBS (Invitrogen). The protein samples were boiled in loading buffer, run on gradient 6 -20% SDS-polyacrylamide gels, and probed with primary antibodies of interest and horseradish peroxidase-conjugated secondary antibody, according to standard procedures. Proteins were visualized by enhanced chemiluminescence detection (PerkinElmer Life Sciences), following the manufacturer's instructions.
Biotinylation-Cells were placed on ice and rinsed twice with a cold rinsing solution containing PBS, pH 7.5, 1 mM MgCl 2 , and 0.1 mM CaCl 2 . Cells were then gently agitated at 4°C for 1 h in borate buffer, pH 9, containing 1 mg/ml NHS-biotin. Biotinylation was stopped by washing three times with ice-cold PBS supplemented with 100 mM glycine. Cells were then washed three times in PBS supplemented with 1 mM MgCl 2 and 0.1 mM CaCl 2 . Washed cells were lysed for 45 min at 4°C in solubilizing buffer (150 mM NaCl, 5 mM EDTA, 3 mM KCl, 120 mM Tris/ Hepes, pH 7.4, 1% (v/v) Triton X-100) containing protease inhibitors (Complete 1697498; Roche Applied Science). After taking an aliquot of the total cell extract from each sample to provide a measure of total NKCC2 expression, the rest of the cell lysates were incubated in avidin beads (Sigma) overnight at 4°C. After overnight incubation, samples were centrifuged at 16,000 rpm for 5 min, and the supernatant (the intracellular fraction) was removed. Avidin beads were then washed with solubilizing buffer and then centrifuged for 7 min at 16,000 rpm seven times. Pellets were incubated in solubilizing buffer and denaturating buffer for 10 min at 95°C and stored at Ϫ20°C. Each fraction was subjected to SDS-PAGE and Western blot analysis.
Immunocytochemistry-Forty-eight hours post-transfection, confluent cells were washed twice with PBS, pH 8, 1 mM MgCl 2 , and 0.1 mM CaCl 2 . cells were incubated at 4°C for 1 h in PBS ϩϩ (pH 8, 1 mM MgCl 2 , and 0,1 mM CaCl 2 ), containing 1 mg/ml NHS-biotin. Cells were rinsed three times in rinsing solution with 100 mM glycine and reincubated at 4°C in the same solution for 10 min. Then they were washed three times with PBS ϩϩ . Cells were fixed with 2% paraformaldehyde in PBS for 20 min at room temperature, incubated with 50 mM NH 4 Cl, permeabilized with 0.1% Triton X-100 for 1 min and incubated with DAKO (antibody diluent with background-reducing components) for 30 min to block nonspecific antibody binding. Fixed cells were incubated for 1 h at room temperature with the primary antibodies at the appropriate dilution in DAKO. Mouse anti-protein-disulfide isomerase (PDI), rabbit anti-Myc, and rabbit anti-calnexin were visualized with Texas Red-coupled secondary antibodies, Cy2-labeled anti-rabbit (green), and fluorescein isothiocyanate-coupled IgG antibodies, respectively. Mouse anti-Myc was visualized with Texas Red-coupled secondary antibodies or Cy2-labeled anti-mouse (green). Cells were then washed with PBS and mounted with Vectashield.
Pulse-Chase Assays-After 24 h, OKP and HEK 293 cells transiently transfected with plasmid DNA using Lipofectamine reagent (Invitrogen) were incubated in cysteine-and methionine-free DMEM starvation media for 1 h. Starvation medium was removed and replaced with DMEM labeling medium containing [ 35 S]methionine/cysteine labeling mix. After 1 h, cells were rinsed three times with PBS and another three times with normal growth medium and returned to normal growth medium for the duration of chase to the specified time points. Cells were washed twice with ice-cold PBS and incubated on ice for 1 h in lysis buffer with a mixture of protease inhibitors, after which solubilized extracts were collected for immunoprecipitation. Proteins were immunoprecipitated with monoclonal anti-Myc antibody, resolved with SDS-PAGE, blotted onto nitrocellulose, and revealed by autoradiography.
Statistical Analyses-Results are expressed as mean Ϯ S.E. Differences between means were evaluated using paired or unpaired t test or analysis of variance as appropriate. p Ͻ 0.05 was considered statistically significant.

NKCC2 COOH-terminal Removal Disrupts Glycosylation and Plasma Membrane Localization of the Co-transporter-
Previous analysis of the NKCC2 amino acid sequence suggested a protein comprising 12 transmembrane-spanning domains, with both NH 2 and COOH termini located intracellularly (Fig.  1A). A large number of mutations leading to BS1 are located at the COOH terminus of NKCC2, which is an indication of the importance of this domain in NKCC2 protein expression and function (5,6,26,27). In the present work, we studied the effect of the most distal mutation described to date in the NKCC2 COOH terminus (26), Y998X, on NKCC2 expression and function. This NKCC2 mutation introduces a premature stop codon, causing the formation of a truncated protein, missing the last 101 aa of the COOH-terminal domain. To characterize the expression and function of WT NKCC2 and Y998X proteins in renal epithelial cell lines, we transiently transfected NH 2 -terminally tagged WT and truncated NKCC2 into cultured renal cells. We showed previously that NH 2 -terminally tagging NKCC2 does not affect co-transporter trafficking and function (23).
In the first series of experiments, we verified whether Myc-NKCC2 mutant protein was functional. To address this, Na ϩ -K ϩ (NH 4 ϩ )-2Cl Ϫ transport activity was assessed by estimating the rate of intracellular acidification caused by entry into the cells of NH 4 ϩ via this transport mechanism after abrupt application of NH 4 Cl to the cells (23,24). As illustrated in Fig. 1B, the NH 4 ϩ -induced initial rate of pH i recovery (dpH i /dt) was 3-fold faster in OKP cells expressing WT NKCC2 than in mock control cells. In contrast, the NH 4 ϩ -induced initial rate of pH i recovery in cells transfected with Myc-Y998X was not significantly different from that of mock cells, indicating that the Y998X mutant did not express functional co-transporter protein.
To evaluate whether the lack of transport activity was simply caused by a dramatic reduction in total cellular protein and/or a lack of cell surface expression, we first examined cells by immunostaining ( Fig. 2A). Immunocytochemistry analysis revealed that WT NKCC2 staining co-localized with biotinylated cell surface proteins, indicating correct targeting of NKCC2 to the cell surface. In contrast, cells transfected with Y998X displayed an immunofluorescence staining pattern that is more restricted to a perinuclear ER-like distribution. Indeed, these cells did not show any colocalization with biotinylated cell-surface proteins, indicating that Y998X mutant was not expressed at the cell surface. Similar results were obtained when WT NKCC2 and Y998X proteins were NH 2 -terminally tagged with EGFP instead of Myc (Fig.  2B). Taken together, these data strongly suggest that the lack of transport activity associated with Y998X results from the absence of the protein from the cell surface.
To corroborate this observation, we performed cell surface biotinylation experiments. Surface membrane proteins were biotinylated by reaction with sulfo-NHS-SS-biotin and isolated by precipitation with streptavidin-bound agarose. Myc-NKCC2 and Y998X proteins were then identified by immunoblot using anti-Myc antibody (Fig. 2C). As expected, in total cell lysates, immunoblot analysis of cells transfected with WT NKCC2 revealed two protein bands: a lower band of FIGURE 1. Effect of NKCC2 COOH-terminal removal on NKCC2 transport activity. A, predicted topology of NKCC2 protein with two potential glycosylation sites indicated by arrows. The COOH terminus of NKCC2 harbors a large number of Bartter syndrome mutations. The three tested mutations (Y998X, N984fs, and D918fs) in the present study are marked in rounds. In the present work, we mainly focused on the effect of Y998X, the most distal mutation described to date in the NKCC2 COOH terminus. B, measurement of Na-K-2Cl co-transport activity in OKP cells expressing NKCC2 or Y998X proteins, mean initial rate of pH i recovery (dpH i /dt) from NH 4 ϩ -induced alkaline load. Each bar represents the mean Ϯ S.E. rates of cell pH recovery (dpH i /dt, pH units/min) under different experimental conditions (mock cells, OKP cells expressing Myc-NKCC2, and OKP cells transfected with Myc-Y998X). a, p Ͻ 0.05 versus WT NKCC2; b, p ϭ not significant versus Y998X; c, p Ͻ 0.01 versus Y998X (mock, n ϭ 8; WT NKCC2, n ϭ 8. Y998X, n ϭ 8). FIGURE 2. Y998X protein is not expressed at the cell surface. A, immunolocalization studies of wild-type and mutated NKCC2 proteins. Membrane proteins of confluent OKP cells were biotinylated at 4°C with the biotinylation reagent sulfo-NHS-SS-biotin. Then the monolayers were fixed and stained for Myc-tagged (A) or EGFP-tagged (B) wild-type and mutant NKCC2 fusion proteins (Cy-2, EGFP) and cell surface biotin (avidin-Cy3). The stained specimens were evaluated by confocal microscopy. Optical sections (xy) at the cell surface are depicted for the Cy2 or EGFP channel (green), Cy3 channel (red), and a merged channel. C, confluent cells were biotinylated at 4°C with sulfo-NHS-SS-biotin. Biotinylated proteins were recovered from cell extracts by precipitation with NeutrAvidin-agarose. NKCC2 proteins on the cell surface (S) were detected by Western blotting (WB) with Myc antibody. An aliquot of the total cell extract from each sample was also run on a parallel SDS-gel and subjected to Western blot for total NKCC2 expression (T). 120 kDa and an upper band of 160 kDa. The 120 kDa band represents the core-glycosylated immature form of the co-transporter protein located in the ER, and the 160 kDa band represents the complex-glycosylated mature form of the cotransporter protein located in the plasma membrane (23). As can be seen in Fig. 2C, the biotinylated protein fraction contained only the complex-type glycosylated fraction of WT Myc-NKCC2, confirming that only this form of the co-transporter is able to reach the cell surface (23). The removal of the last 101 aa (Y998X) of the COOH-terminal tail of NKCC2 produced a truncated recombinant protein that migrated with an apparent molecular mass of ϳ110 kDa, which is entirely consistent with the theoretical molecular weight of 108.58 kDa, calculated from the predicted amino acid sequence of the protein. Interestingly, this was associated with the complete abolition of the mature complex-glycosylated band, suggesting that Y998X failed to exit the ER and transit through the Golgi complex. As a consequence, by contrast to the readily detected biotinylated wild-type NKCC2, Y998X protein was virtually undetectable. A long exposure was indeed required to detect only a faint band of biotinylated Y998X protein in some experiments, indicating that this mutant is located almost exclusively intracellularly, in agreement with the fluorescence microscopy shown in Fig. 2, A and B.
Proteasome-dependent Degradation of the Core-glycosylated NKCC2-Based upon the above findings, we hypothesized that the Y998X mutation results in ER retention and degradation of NKCC2 and thus the absence of the mature and functional form of the co-transporter. This mechanism involves activation of the proteasome proteolysis pathway by the retained proteins in the ER and/or rerouting from the Golgi complex downstream vesicles systems to lysosomes (28,29). Accordingly, treatment with proteasome and lysosome inhibitors might provide important insights into possible mechanisms of mutated NKCC2 trafficking and degradation. To investigate the possible involvement of proteasomal and/or lysosomal degradation pathways, cells were treated overnight with 10 M MG132 or 100 M leupeptin, and their lysates were subjected to Western blot analysis. As shown in Fig. 3, A and B, exposure to MG132 significantly increased the protein levels of the immature forms of both WT NKCC2 and Y998X without an apparent increase in their mature forms, an effect consistent with inhibition of protein degradation associ- ated with a pre-Golgi compartment, presumably the ER. The latter notion was supported by the observation that the ER-resident protein, OS-9 (30,31), is also up-regulated under the same experimental conditions (Fig. 3C). In contrast to MG132, lysosome inhibition by leupeptin was without action on the immature form of NKCC2 and Y998X proteins (Fig. 3, D and E). Most importantly, leupeptin had no notable effect on the mature form of the co-transporter, failing therefore to reverse the loss of complex-glycosylated Y998X (Fig. 3, D and E). The same results were obtained using 100 M chloroquine (data not shown). It is worth noting that under the same circumstances, leupeptin and chloroquine were able to increase the expression of caveolin-1 (Fig. 3F), a protein known to be lysosomally degraded (32,33), ruling out the possibility that the failure to reverse the loss of the mature Y998X protein was secondary to a defect in the treatment used to inhibit lysosomal function. Altogether, these results provide additional evidence that the core-glycosylated form of the Y998X mutant is largely trapped in the ER, leading therefore to the absence of the complex-glycosylated form of the co-transporter.
Localization of the Y998X Mutant in the ER-To ensure that the immunoreactive band of Y998X is indeed the immature species produced in the ER, we evaluated its sensitivity to glycosidase digest. Peptide:N-glycosidase F removes all Asn-linked glycans, whereas endoglycosidase H specifically recognizes only high mannose-type carbohydrate backbones characteristic of partially glycosylated proteins resident in the proximal secretory pathway (ER to medial Golgi). After exit from the ER, proteins proceed through the secretory pathway, and most glycoproteins acquire endoglycosidase H resistance in the medial Golgi. Therefore, insensitivity to endoglycosidase H can reflect the maturation state of glycoproteins. Consistent with a mature protein, the complex-glycosylated form of wild-type NKCC2 was sensitive to peptide:N-glycosidase F but not to endoglycosidase H (Fig. 4A). In contrast, the Y998X mutant was sensitive to both endoglycosidase H and peptide:Nglycosidase F, since both produced a shift in the apparent molecular mass of the protein (Fig. 4A), indicating that Y998X has only a high mannose N-linked glycan of the type acquired in the ER and thus is not terminally glycosylated. Together with immunolocalization and biotinylation experiments, these data suggest that Y998X was prevented from advancing beyond the ER into the secretory pathway.
To further confirm that the Y998X mutant was retained in the ER, the subcellular localization of wild-type NKCC2 and mutant Y998X were compared with the distribution of an ER marker, the thiol-disulfide oxidoreductase PDI. As shown in Fig. 4B, WT NKCC2 displays punctuate staining distributed throughout the cell as expected for heterologous expression systems. Accordingly, besides its cell surface expression, a fraction of the co-transporter was located inside the cells, presumably representing transport intermediates en route to the cell surface. Hence, for WT NKCC2, cell surface expression surrounded the PDI signal. In contrast, an exclusively perinuclear pattern of staining was observed for Y998X, with marked colocalization with PDI (Fig. 4B). Similar results were obtained using the ER marker calnexin (Fig. 4C). Thus, these data strengthened the conclusion drawn from data described above indicating that the Y998X mutant is retained in the ER as a core-glycosylated protein.
NKCC2 COOH-terminal Removal Impairs Maturation of the Co-transporter-ER retention of membrane proteins is often associated with their dislocation into the cytosol and subsequent degradation, resulting in an overall high turnover rate of the retained proteins (28). To directly compare the stability and the maturation of Y998X with WT NKCC2, we traced the sorting delivery of newly synthesized proteins from the ER to Golgi compartments, using a pulse-chase analysis. In these experiments, cells transfected with Myc-tagged wildtype NKCC2 or Y998X were labeled for 1 h with [ 35 S]methionine/cysteine and chased with unlabeled methionine and cysteine for different intervals between 1 and 4 h. As illustrated in Fig. 4D, the wild-type NKCC2 protein was initially synthesized as the 120-kDa immature form before being gradually converted to the 160-kDa mature form. In kinetic analysis, the immature form of wild-type NKCC2 protein showed a progressive decrease with an estimated half-life of 90 min (Fig. 4E). The decrease represents the conversion of the immature to the mature form as well as the degradation of the immature form of NKCC2 protein. By contrast, Y998X protein, synthesized as the 110-kDa immature form, was unable to acquire complex sugars at any of the indicated time intervals and thus was not converted to the mature form during the chase. Consequently, the immature form of Y998X persisted with a half-life longer than that of wild-type protein. In sum, this pulse-chase analysis indicated that Y998X mutation results in a maturation and trafficking defect of the co-transporter, leading, therefore, to a lack of cell surface expression and activity.

Influence of COOH-terminal Truncations on the Steady State
Amounts of Immature and Mature NKCC2-The results described above clearly demonstrate that the last 101 aa of the COOH terminus is an important determinant of the co-transporter protein maturation. Interestingly, sequence analysis revealed that this region is enriched in adjacent hydrophobic residues. For several membrane proteins, these motifs have been reported to function as ER export sorting signals and/or cell surface targeting (34 -37). To examine a potential role of these motifs in NKCC2 trafficking, we first generated serial truncation mutants by introducing stop codons at various positions upstream of Y998X mutation, therefore progressively eliminating one or more of these motifs (Fig. 5A). 1039, 1065, and 1083 indicate the position at which a stop codon was introduced to generate mutants lacking the last 56, 30, and 12 amino acids, respectively. As illustrated in Fig. 5B, removal of the last 30 COOH-terminal aa (C⌬56 and C⌬30) prevented maturation of complex-glycosylated protein. In contrast, when only the last 12 aa were deleted (C⌬12), the co-transporter was able to acquire the complex-glycosylated form. Thus, these results indicate that the residues between 1065 and 1083 are critical for ER exit and therefore for co-transporter maturation and cell surface expression.
In light of the above findings, we examined the sequence of amino acids between 1065 and 1083 for an ER export signal. The first thing that became apparent was the presence of several dileucine-like motifs in this region. Hence, we performed further truncation analysis within this region to delimit the sequence important for ER export of NKCC2 (Fig. 5C). Cell surface biotinylation followed by immunoblot analysis not only confirmed that C⌬12 is able to acquire the mature form but also showed that this mutant is able to reach the cell surface (Fig.  5D), suggesting that the final 12 aa residues of NKCC2 are apparently not crucially involved in NKCC2 maturation and cell surface targeting. In contrast to C⌬12, mutants missing the last 15 (C⌬15) residues were still lacking complex oligosaccharides (Fig. 5D), indicating that despite the presence of an IL motif located at positions 1072 and 1073, these mutants failed to exit the ER. This suggests that the 1072 IL 1073 motif is not critical for NKCC2 ER export. Of note, C⌬12 and C⌬15 share the presence of a COOH-terminal valine. This configuration promotes ER export for several full and truncated membrane proteins (14,16,38,39). However, unlike C⌬12, C⌬15 was not able to convert to mature form, strongly suggesting that the COOH-terminal valine is very unlikely to be the sole discriminating factor for C⌬12 export from the ER. Consequently, since removal of the last 15 but not 12 aa of NKCC2 COOH terminus results in the absence of the mature form, we focused upon the carboxyl-terminal amino acid residues 1081-1083 (LLV) as a key sequence determinant of NKCC2 export from the endoplasmic reticulum.
The LLV Motif Is Required for ER Export of NKCC2-To determine whether truncation of this region impaired ER exit due to a loss of specific sequence, rather than a simple shortening of the NKCC2 protein, the specific role of each amino acid was analyzed by site-directed mutagenesis in full-length protein of the co-transporter. To this end, we mutated in the WT NKCC2 protein, individually or in com-bination, these residues to alanines (Fig. 5E). Our results indicated that, although individual mutation of Leu 1081 , Leu 1082 , and Val 1083 had no obvious effect on NKCC2 export from the ER (data not shown), double mutation of 1081 LL 1082 or 1082 LV 1083 significantly inhibited the co-transporter transport from the ER. As can be seen in Fig. 5F, the core-glycosylated form of WT NKCC2 corresponded to only 10 -20% of total immunoreactivity of WT NKCC2. In contrast, NKCC2 mutants were predominantly present in a core-glycosylated form (more than 50% of total protein immunoreactivity). Consequently, the relative amounts of mature mutant proteins 1081 AA 1082 and 1082 AA 1083 were significantly lower than that of the wildtype (Fig. 5F), suggesting an inefficient delivery of these mutants to Golgi compartments. Interestingly, in contrast to 1081 LL 1082 and 1082 LV 1083 , double mutation of IL had no effect on NKCC2 maturation (Fig. 5F). Most importantly, triple mutation of 1081 LLV 1083 completely abolished NKCC2 maturation, suggesting that this mutant failed to exit the ER. Collectively, these data strongly suggest that the residues 1081 LLV 1083 are crucial for NKCC2 exit from the ER.
To further document this, pulsechase experiments were performed to determine the kinetic effects of mutations of LLV residues on nascent chain maturation or degradation. As observed above, WT NKCC2 after the pulse appeared as a single band of about 120 kDa, which was then progressively converted to a more slowly migrating band, representing the full-length NKCC2 protein (Fig. 6A). As illustrated in Fig. 6A, although the 1081 LLV 1083 mutant was synthesized as the 120-kDa immature form at a level similar to that of wild-type NKCC2, no mature band was detected at any of the indicated time intervals during the chase. This led to an overall stabilization of 1081 LLV 1083 mutants relative to WT NKCC2. This absence of mature NKCC2 following 1081 LLV 1083 mutations to AAA was also observed by immunofluorescence microscopy, where no cell surface staining could be detected (Fig. 6B). Indeed, the 1081 LLV 1083 mutant localized predominantly to a reticular network. This reticular pattern overlapped with labeling for PDI (Fig. 6C), indicating that the 1081 LLV 1083 mutant was predominantly localized to the ER, which is consistent with the presence of primarily the core-glycosylated immature band seen in immunoblots. Taken altogether, these data support the localization of 1081 LLV 1083 mutants to the ER and indicate that the 1081 LLV 1083 mutation to AAA results in a defect in ER exit.
The Impact of the LLV Motif upon Maturation of NKCC2 Is Independent of the Expression System-The potential significance of our findings is that the 1081 LLV 1083 residues found within the NKCC2 COOH terminus could be very important sorting determinants for trafficking to plasma membranes of TAL cells. Thus, naturally occurring mutations, depriving NKCC2 of this motif, would lead to inappropriate sorting, which not only would disrupt vectorial solute transport in TAL cells but also could form the molecular basis of BS1. To further support this hypothesis, we also tested the effect of two other different mutations previously identified in the NKCC2 COOH terminus of BS1 patients, N984fs and D918fs (6,27), and studied their effect on NKCC2 expression. N984fs and D918fs mutations produce COOH-terminally truncated NKCC2 proteins missing the last 108 and 198 aa of NKCC2, respectively, and thus lacking the 1081 LLV 1083 motif. As anticipated, similar to Y998X, D918fs and N984fs mutations clearly interfered with membrane trafficking of the NKCC2 protein. Indeed, in contrast to WT NKCC2, all mutants were retained in the cyto-plasm, where they exhibited excellent co-localization with the ER marker, PDI (data not shown). Moreover, because the ER quality control mechanism and the capacity of ER export machinery may be cell-dependent, we also sought to corroborate our finding by conducting the experiments in a second renal cell line, the HEK cells. As illustrated in Fig. 7A, COOHterminally truncated NKCC2 (Y998X) and 1081 LLV 1083 mutated proteins were not expressed at the cell surface and were localized to the ER, as indicated by co-localization with the ER marker PDI. To determine if the ER localization of NKCC2 mutants reflected a defect in maturation, we again analyzed the biosynthetic processing of WT NKCC2 and mutants expressed in HEK cells by metabolic pulse-chase and immunoprecipitation. As shown in Fig. 7B, in contrast to WT NKCC2, the conversion to the mature form was not detectable during the chase period, indicating a failure in ER exit of the mutated co-transporter, indicating therefore that the absence of cell surface expression is secondary to a defect in the co-transporter maturation. Taken in concert, these data demonstrate that NKCC2 mutants behave similarly regardless of the expression system, providing additional evidence that naturally occurring mutations depriving NKCC2 its distal COOH-terminal tail and/or interfering with the 1081 LLV 1083 motif result in an ER exit defect of the co-transporter.
LLV Motif Is Conserved in the Carboxyl-terminal Tails of Cation-Chloride Co-transporters-Because no ER exit motif sequence had previously been reported for CCCs, we conducted a sequence data bank search to see if the 1081 LLV 1083 FIGURE 6. Substitution of LLV motif in NKCC2 COOH terminus results in ER retention. A, pulse-chase assay to compare maturation of LLV 1083 /AAA mutant with WT NKCC2. Transfected OKP cells were metabolically labeled for 1 h before chasing for the time indicated in the figure. Proteins were isolated by immunoprecipitation with anti-Myc antibodies and run on SDS-polyacrylamide gels and revealed by autoradiography. B, representative confocal images from anti-Myc surface-stained cells expressing WT NKCC2 or Y998X proteins. Wild-type NKCC2 was strongly expressed on the cell surface, illustrated by its colocalization with biotinylated proteins. In contrast, LLV 1083 / AAA mutants did not shown any colocalization with biotinylated membrane proteins, indicating a lack of cell surface expression. C, colocalization of Myctagged LLV 1083 /AAA mutant proteins with the ER-resident protein, PDI. OKP cells expressing mutated NKCC2 proteins were fixed, permeabilized, and stained for Myc (Cy2; green) and PDI (Texas Red). Yellow, overlap between the Myc tag of NKCC2 protein (green) and PDI (red). Yellow (merged image), co-localization of PDI and NKCC2 proteins. B, pulsechase analysis. HEK cells expressing the studied proteins were labeled for 1 h with radioactive methionine and cysteine and chased for the indicated time intervals. Lysates were subjected to immunoprecipitation, followed by SDS-PAGE and autoradiography. In contrast to WT NKCC2, the conversion to the mature form was not detectable during the chase period, indicating a failure in ER exit of Y998X and 1081 LLV 1083 /AAA proteins. motif is conserved within this protein family. As shown in Fig. 8, the result revealed that this trihydrophobic motif is highly conserved among all members of the CCC family, including the CCC-interacting protein, CIPI (40). Because this region is extremely well conserved not only across species but also between isoforms, it appears to be an excellent candidate as a universal regulatory motif for CCC family members. To further support this hypothesis, we studied the role of this motif in NCC expression. To this end, mouse NCC was NH 2 -terminally tagged with Myc and transiently expressed in HEK cells. Similar to NKCC2, converting the trihydrophobic motif (LLI) to AAA in the context of the full-length NCC abolished the formation of complex-glycosylated form of NCC (Fig. 9A). As a consequence, in contrast to WT NCC, NCC mutants were retained in the cytoplasm (Fig. 9B). Taken in concert, these data further strengthen the conclusion that the trihydrophobic motif may function as a common ER export signal for the cation-chloride co-transporters.

DISCUSSION
A large number of BS1 mutations have been identified in the COOH terminus of NKCC2, demonstrating the importance of this domain in co-transporter function. In this study, we showed that BS1 mutations depriving NKCC2 of its distal COOH-terminal region result in the absence of cell surface expression and thus transport activity of the co-transporter. We verified that the lack of surface expression was not caused by poor expression or increased rates of degradation of mutant co-transporters. Using serial truncations of the NKCC2 COOH terminus, followed by site-directed mutagenesis, we identified an ER exit motif composed of highly conserved residues, 1081 LLV 1083 , required for maturation and subsequent surface expression of NKCC2. The high conservation of this trihydrophobic motif among all members of the cation-chloride co-transporter family lends further support to the role of this regulatory locus in vivo.
Studies by Starremans et al. (5) were the first to analyze the functional consequences of BS1 mutations selected from different regions in the NKCC2 sequence, using the heterologous system in X. laevis oocytes. In contrast to our study conducted in renal cells, the authors showed that the expression levels of all mutants, when expressed in Xenopus oocytes, were significantly lower when compared with wild type. In addition, they reported that all mutants, including those unable to acquire the mature form, were correctly routed to the plasma membrane but remained nonfunctional. Thus, it was concluded that BS1 mutations impair the function of the NKCC2 because of a defect in the functional properties and not the trafficking of the co-transporter proteins. The heterologous expression system in X. laevis oocytes has been shown to be an excellent tool for a robust expression and reproducible expression of NKCC2 proteins. However, it has also been shown for several mutated proteins that, whereas they are completely retained in the ER when expressed in mammalian cells (41)(42)(43), such ER-retained proteins are often transported to the plasma membrane of oocytes to some extent. For example, the ER-retained cystic fibrosis transmembrane conductance regulator lacking Phe-508 (CFTR-⌬F508) confers chloride conductance in oocytes, but it does not exit the ER in mammalian cells (44 -46). Similar observations have been made for mutants of the HERG voltagegated potassium channel that cause hereditary long QT syn- drome (43). The expression of mutant proteins in oocyte plasma membranes has been attributed to a reduced level of ER-associated retention and/or degradation. Indeed, the Xenopus oocyte is more permissive to folding mutants, due to its lower incubation temperature (43,45). Therefore, to experimentally mimic the in vivo situation more closely, the present study was conducted in mammalian heterologous expression systems. Moreover, because the ER quality control mechanism and the capacity of ER export machinery may depend on the cell type, the study was conducted in two different renal cell lines.
Although it cannot be excluded that a defect in the functional properties of the co-transporter may contribute to BS1, our findings clearly show that C-terminally truncated NKCC2 proteins are not expressed at the cell surface of renal cells, because they are retained in the ER.
There have been several reports indicating that COOH-terminal residues are important for the correct targeting of several transmembrane proteins (12)(13)(14). Therefore, we concentrated our attention in this study on NKCC2 COOH terminus and took advantage of the existence of natural mutants altering the COOH-terminal tail of the co-transporter. We selected in this study Y998X, the most distal BS1 mutation described to date in the NKCC2 COOH terminus. We first showed that, in contrast to wild-type NKCC2, cells overexpressing mutated proteins did not express a functional co-transporter. In principle, there are multiple mechanisms by which mutations could abolish NKCC2 activity, including impaired protein synthesis, disturbed protein processing and routing, and accelerated protein retrieval or degradation. Western blot analysis and pulse-chase experiments revealed that all proteins were synthesized at similar levels when compared with wild-type NKCC2, clearly indicating that a defect in protein synthesis is not responsible for the lack of transport activity. Likewise, truncation of the distal region of NKCC2 COOH terminus did not lead to reduced stability or increased NKCC2 degradation. By contrast, pulse-chase analysis revealed that, unlike WT NKCC2, mutated NKCC2 proteins were not able to convert to the complex-glycosylated form and were sensitive to endoglycosidase H deglycosylation. Moreover, cell treatment with proteasome or lysosome inhibitors failed to restore the loss of complex-glycosylated forms, further eliminating the possibility that mutant co-transporters were processed by the Golgi apparatus and that the absence of the mature form was a result of protein degradation. Finally, co-immunolocalization revealed that mutants lacking the COOH termini are unable to exit from the ER, as indicated by extensive colocalization with the ER marker PDI. Taken together, these studies indicate that the COOH-terminal tail of NKCC2 governs co-transporter ER exit and thus cell surface expression. More importantly, they strongly suggested that NKCC2 surface expression is governed by a signal motif present within the last 101 aa of the NKCC2 COOH terminus.
Progressive truncations and mutagenesis studies in the COOH terminus of transmembrane proteins have led to the identification of several COOH-terminal export signals that regulate ER exit. These signals include adjacent bulky hydrophobic or aromatic residues (FF, FY, LL, IL, or YYM), diacidic motifs ((D/E)X(D/E)), dibasic motifs ((R/K)X(R/K)), a combi-nation of motifs, or multiple cooperating signals (as for ERGIC-53) in the cytoplasmic domains of proteins (14,18,34,47). These ER exit motifs are decoded by physically interacting with the components of COP II vesicles (48,49). In the present study, serial truncations proposed an ER exit signal to be located in the region corresponding to amino acids 1065-1083 of the NKCC2 COOH terminus, which was narrowed down to a trihydrophobic motif, LLV, at positions 1081-1083. The 1081 LLV 1083 motif we now uncovered in NKCC2 may belong to the family of adjacent bulky hydrophobic ER export signals. Without these three hydrophobic residues, NKCC2 does not exit the ER. Indeed, naturally occurring COOH truncations depriving NKCC2 of these three residues or simply converting them to alanines trap the co-transporter within the ER. Thus, this uncovered motif in the NKCC2 COOH terminus may provide an interactive site mediating the interaction of the co-transporters with selective proteins to facilitate their transport from the ER to the cell surface. Future studies should uncover the mechanism by which this motif may contribute or regulate possible interactions between NKCC2 and COPII that may direct traffic in the early secretory pathway. The motif may also be involved in interactions with putative escort proteins, as reported for other membrane proteins, such as RAMP for the calcitonin receptorlike receptor (50) and RanBP2 for opsin (51). The LLV motif may also provide a docking site for machinery proteins involved in co-transporter folding to the state competent for transport out of the ER. However, this is unlikely, since mutations in the distal region of the carboxyl terminus of NKCC2 did not lead to reduced stability or enhanced degradation of the co-transporter. Therefore, the trafficking defects we characterized using mutated and/or truncated forms of the co-transporter are unlikely to be a result of compromised folding in the ER. Most importantly, regardless of the action of this COOH-terminal motif, the results presented here indicate that the LLV motif is a major determinant of surface NKCC2 expression.
A very recent study conducted by Carmosino et al. (52) reported that the region corresponding to amino acids 807-884 contains the sorting signal that directs apical trafficking of the NKCC2 protein. The results of the present work showed evidence that this region is not sufficient for NKCC2 exit from the ER. Indeed, despite the presence of the region containing the apical sorting motif, Y998X mutants were not expressed at the cell surface, because they failed to exit the ER. Collectively, these findings suggest that during intracellular transport of NKCC2, recognition of the ER export motif precedes that of the apical targeting determinant. In conclusion, the cytoplasmic tail of NKCC2 controls the number of molecules transported to the cell surface independently of their polarized sorting. As a consequence, a COOH-terminal deletion of NKCC2 would result in a dramatic reduction of cell surface expression due to a failure of specific ER export. Hence, the cytoplasmic tail of NKCC2 contains at least two distinct major motifs; the LLV motif controls ER export and cell surface expression levels, whereas another independent motif present in the region corresponding to amino acids 807-884 determines its polarized surface expression in epithelial cells. Moreover, besides these motifs, other signals in the NKCC2 COOH terminus may also contribute to the traf-ficking and/or folding of the co-transporter. Alternatively, residues LLV may be part of a larger domain within the COOH terminus that is involved in maturation of the cotransporter. We are currently exploring these possibilities.
In conclusion, our results indicate that the NKCC2 COOH terminus is required for the generation of a functional Na-K-Cl co-transporter. In addition, they suggest that naturally occurring mutations resulting in truncations or amino acid substitution interfering with the LLV motif in the NKCC2 COOH terminus have similar consequences by impeding ER export and subsequent targeting of the co-transporter to the cell surface. Importantly, this LLV motif is highly conserved among all members of the CCC family and thus may provide a common molecular mechanism underlying their ER export. These findings raise the possibility that loss of this conserved motif may contribute to disease phenotypes in renal tubulopathies related not only to NKCC2 but also to other members of the CCC co-transporter family. Further studies are under way in our laboratory, including identification of the protein(s), among NKCC2 binding partners previously cloned using yeast twohybrid analysis (23), specifically interacting with this motif. These studies should allow us to elucidate the mechanisms involved in ER export of CCCs in general and in particular of NKCC2. The exact characterization and identification of the molecular mechanisms of the motif-facilitated ER export may open new avenues for the development of therapeutic strategies targeting co-transporter transport from the ER to the cell surface.