A Phosphoinositide 3-Kinase (PI3K)-serum- and glucocorticoid-inducible Kinase 1 (SGK1) Pathway Promotes Kv7.1 Channel Surface Expression by Inhibiting Nedd4-2 Protein*

Background: PI3K regulates the surface expression of Kv7.1. Results: Kv7.1 surface expression depends on PI3K activity in polarized epithelial cells. SGK1 is the primary downstream target of PI3K in this process, which involves inhibition of Nedd4-2-dependent endocytosis of the channel. Conclusion: The surface expression of Kv7.1 is regulated by a PI3K-SGK1-Nedd4-2-mediated pathway. Significance: This pathway could regulate Kv7.1 cell surface expression levels in epithelial cells and cardiac myocytes. Epithelial cell polarization involves several kinase signaling cascades that eventually divide the surface membrane into an apical and a basolateral part. One kinase, which is activated during the polarization process, is phosphoinositide 3-kinase (PI3K). In MDCK cells, the basolateral potassium channel Kv7.1 requires PI3K activity for surface-expression during the polarization process. Here, we demonstrate that Kv7.1 surface expression requires tonic PI3K activity as PI3K inhibition triggers endocytosis of these channels in polarized MDCK. Pharmacological inhibition of SGK1 gave similar results as PI3K inhibition, whereas overexpression of constitutively active SGK1 overruled it, suggesting that SGK1 is the primary downstream target of PI3K in this process. Furthermore, knockdown of the ubiquitin ligase Nedd4-2 overruled PI3K inhibition, whereas a Nedd4-2 interaction-deficient Kv7.1 mutant was resistant to both PI3K and SGK1 inhibition. Altogether, these data suggest that a PI3K-SGK1 pathway stabilizes Kv7.1 surface expression by inhibiting Nedd4-2-dependent endocytosis and thereby demonstrates that Nedd4-2 is a key regulator of Kv7.1 localization and turnover in epithelial cells.

The potassium channel Kv7.1 (KCNQ1, KvLQT) plays an important role in a number of tissues where it associates with the auxiliary KCNE ␤-subunits. In the heart, Kv7.1, together with KCNE1, forms the delayed rectifier potassium current I Ks , which is an important contributor to the repolarization of the cardiac action potential (1,2). Kv7.1 is also present in pancreatic ␤-cells, where it is thought to be implicated in the regulation of insulin secretion (3,4). In addition, Kv7.1 is expressed in several epithelia, where it is involved in salt and water transport (5,6). Most importantly, the channel regulates gastric acid secretion (7,8) and contributes to the release of potassium to the endolymph in the inner ear (9,10). Accordingly, Kv7.1 knock-out mice show gastric hyperplasia and are completely deaf (11). Mutations in the KCNQ1 gene are furthermore associated with long QT (LQT) 4 syndrome, an inherited form of cardiac arrhythmia that can lead to cardiac arrest (12). In its recessive form, the Jervell and Lange-Nielsen syndrome (13), the disease additionally leads to hearing loss due to disturbances in the flow of potassium in the inner ear. The mechanism underlying the LQT syndrome is reflected in a loss of Kv7.1 function, frequently originating from trafficking disorders, and hence a decrease in number of channels in the plasma membrane (14 -16). Nevertheless, the molecular and cellular mechanisms controlling the cell surface expression of Kv7.1 in cardiomyocytes and epithelial cells are still largely unknown.
We recently observed that the basolateral Kv7.1 potassium channel displays a very dynamic localization pattern during Madin-Darby canine kidney (MDCK) cell polarization controlled by a calcium switch (17). We found that initiation of MDCK cell polarization results in removal and degradation of surface-expressed Kv7.1 and subsequent accumulation of newly synthesized channels in the endoplasmic reticulum (ER).
Later in the polarization process, Kv7.1 is released from the ER, and surface expression is recovered. While the initial removal of Kv7.1 from the cell surface is mediated by the AMP-activated protein kinase and E3 ubiquitin ligase Nedd4-2 (neuronal precursor cell expressed developmentally down-regulated 4-2) (18), the subsequent recovery of Kv7.1 surface expression depends on PI3K activity (17). PI3K is an important kinase that is implicated in the control of a number of cellular processes including cell proliferation, cell survival, and epithelial cell polarization (19 -22). It has in particular received a lot of attention in relation to human cancer as the kinase is one of the most common oncogenes (reviewed in Ref. 23). PI3K is composed of a regulatory subunit and a catalytic subunit that phosphorylates phosphatidylinositol 4,5-bisphosphate into phosphatidylinositol (3,4,5)-trisphosphate. Phosphatidylinositol (3,4,5)-trisphosphate is an important signaling molecule that binds proteins via a pleckstrin homology domain, which is found in e.g. 3-phosphoinositide-dependant-kinase 1 and the Akt kinase (also denoted protein kinase B) (24,25). In polarizing MDCK cells, PI3K is activated by adherens junction assembly, resulting in Rac1-dependent changes in the actin cytoskeleton (26,27). In polarized MDCK cells, adherens junctions are enriched in phosphatidylinositol (3,4,5)-trisphosphate suggesting that PI3K remains tonically active at this subcellular location (28). Furthermore, long term inhibition of PI3K reduces MDCK cell height, suggesting that tonic PI3K activity regulates basolateral membrane formation and maintenance (19,28).
Two well described downstream targets of PI3K are the serum-and glucocorticoid-inducible kinase 1 (SGK1 (29)) and Akt (reviewed in Ref. 30). Both protein kinases have been reported to stimulate Kv7.1-KCNE1 currents in Xenopus oocytes (31,32) and inhibit the actions of Nedd4-2 (33)(34)(35), another well known regulator of Kv7.1 (36). Nedd4-2 is an E3 ubiquitin ligase that ubiquitylates target membrane proteins such as ion channels, thereby increasing the rate of their internalization and degradation (37,38). SGK1 and Akt can phosphorylate Nedd4-2, thus increasing the binding affinity to 14-3-3 proteins (39). For the epithelial sodium channel ENaC, it has been found that 14-3-3 protein binding to Nedd4-2 prevents Nedd4-2-mediated ubiquitylation and thereby increases surface expression levels of the channel (39,40). Because the interaction of Nedd4-2 with both ENaC and Kv7.1 is mediated by intrinsic sequences known as PY motifs, it is possible that the interaction of Nedd4-2 with Kv7.1 is inhibited by the same phosphorylation of Nedd4-2. Overall, these observations position SGK1 and Akt as possible downstream targets of the PI3K effect upon Kv7.1 localization during the polarization process.
In this study, we demonstrate that a PI3K pathway stabilizes cell surface expression of Kv7.1 channels in both polarizing and polarized cells. We find that PI3K primarily acts through SGK1 and demonstrate that SGK1 controls Kv7.1 localization by inhibiting Nedd4-2-dependent endocytosis of the channel. Intriguingly, the PY motif Kv7.1 mutant that cannot interact with Nedd4-2 displayed increased expression in the apical membrane, suggesting that Nedd4-2-dependent endocytosis is important to avoid apical trapping of the channel during polarization. Overall, our data suggest that a PI3K-SGK1 pathway promotes cell surface stabilization of Kv7.1 by inhibiting Nedd4-2-dependent endocytosis and places Nedd4-2 as a key regulator of Kv7.1 surface expression and turnover.

Transient and Stable Expression in MDCK Cells
MDCK (strain II) cells were grown in DMEM (Invitrogen) supplemented with 100 units/ml penicillin, 100 mg/ml streptomycin, and 10% FCS (Sigma-Aldrich; henceforth called normal calcium medium, NCM) at 37°C in a humidified atmosphere with 5% CO 2 .
Transient Transfections-MDCK cells were transfected in suspension using Lipofectamine and Plus Reagent (Invitrogen) and plated on glass coverslips (12 mm in diameter, Thermo Fischer Scientific).

Calcium Switch and Inhibitor Experiments
MDCK cells stably expressing hKv7.1, hKv7.1-YA (and when relevant, transiently expressing the ER marker DsRed2-ER) or empty cells (negative control for Western blotting) were plated on glass coverslips or in Petri dishes (for Western blotting), and the calcium switch experiment was performed as described previously (17). Briefly, the cells were allowed to attach to the coverslips for 1.5 h in full DMEM and then washed three times in a low calcium medium (calcium-free MEM, Spinner Modification (Sigma-Aldrich) supplemented with 5% dialyzed fetal bovine serum (Invitrogen), 100 units/ml penicillin, 100 mg/ml streptomycin, 2 mM L-glutamine, and 1.6 M calcium chloride) and grown for 2 days in low calcium medium until reaching confluency. The medium was then changed to NCM, which started the polarization process, and the cells were allowed to polarize for up to 24 h.
For the calcium switch inhibitor studies, NCM was added for 1 h and 45 min and then changed to NCM supplemented with either 10 M LY294002, 10 M Akt inhibitor IV, or 1 M GSK650394 to inhibit PI3K, Akt, or SGK1, respectively. The polarization process was followed for up to 24 h. For inhibitor studies performed on polarized cells, a 24-h calcium switch was performed followed by a 3-h treatment with NCM supplemented with either 10 M LY294002, 1 or 10 M Akt inhibitor IV, or 1 M GSK650394.

Immunofluorescence
Transiently or stably transfected cells grown on glass coverslips were fixed in 4% paraformaldehyde in PBS for 30 min at room temperature. Blocking and permeabilization was performed by a 30-min incubation with 0.2% fish skin gelatin in PBS supplemented with 0.1% Triton X-100 (PBST). The cells were incubated for 1 h in primary antibodies diluted in PBST. Secondary antibodies were diluted in PBST and applied for 45 min. The coverslips were mounted in Prolong Gold (Invitrogen).

Confocal Microscopy and Imaging
Laser scanning confocal microscopy was performed using either the Leica TCS SP2 system equipped with argon and helium-neon lasers or the Zeiss LSM 780 confocal system. Images were acquired using a 63ϫ water immersion objective, NA of 1.2 (Leica TCS SP2) or a 63ϫ oil immersion objective, numerical aperture of 1.4 (Zeiss LSM 780) with a pinhole size of 1 and a pixel format of 1024 ϫ 1024. Line averaging was used to reduce noise. For double-and triple-labeling experiments, sequential scanning was employed to allow the separation of signals from the individual channels.

Quantifications
Total cell fluorescence and intracellular fluorescent signals were quantified using the Zen 2010 confocal software and the submembranous actin cytoskeleton to define the localization of the plasma membrane. Specifically, the exterior and interior of plasma membrane-associated Alexa Fluor 647 phalloidin staining was used to define the regions quantified. Signals were subtracted for background fluorescence (the mean intensity of the background was obtained from an area of the nucleus as none of the proteins quantified would be expressed there.) For data analysis, signals originating from the surface membrane were obtained by subtracting the intracellular signal from the total cell fluorescent signal. The surface-associated signal was subsequently expressed as a percentage of the total cell fluorescence. n Ͼ 20 cells from three independent experiments were quantified. Data analysis was carried out using one-way analysis of variance test or t test and the Prism statistical software package. Data are presented as S.E.

Live Cell Imaging
MDCK cells stably expressing pEGFP-N2-hKv7.1 were subjected to a 24-h calcium switch before being transferred to an imaging chamber. The medium was changed from NCM to Hanks' balanced salt solution supplemented with 10 M LY294002, 10 mM HEPES, pH 7.3. The experiment was performed at 37°C. Cells were visualized by confocal laser scanning microscopy using the CSU-X1 spinning disk module on a Zeiss CellObserver (Carl Zeiss microimaging GmbH, Jena, Germany) with a 63ϫ oil immersion objective (numerical aperture of 1.2).
Time Lapse-The first z-stack was acquired 30 min after addition of Hanks' balanced salt solution supplemented with 10 M LY294002, 10 mM HEPES, pH 7.3, and thereafter, z-stacks were acquired every 3 min with an exposure time of 350 ms using the AxioVision software (version 4.8). The three-dimensional reconstruction of the obtained time lapse was created by using the Imaris software (version 7.2). Data are presented as a QuickTime movie.

Cell Lysates
MDCK cells stably expressing Kv7.1 or Kv7.1-YA were plated in 25 cm 2 flasks and subjected to a 24-h calcium switch followed by a 3-h incubation in NCM supplemented with 40 g/ml cycloheximide. Cells were collected and solubilized in solubilization buffer (50 mM Tris, pH 7.4, 10 mM NaCl, 10 mM KCl, 10 mM NaF, 1% Triton X-100 (Sigma Aldrich) and 0.5% sodium deoxycholate (Sigma Aldrich), 50 l of protease inhibitor mixture (dilution of 1:200, Sigma Aldrich), and a phosphatase inhibitor tablet (Roche Diagnostics)) for 3 h at 4°C. The samples were centrifuged at 15,000 rpm for 10 min, and the super- DECEMBER 27, 2013 • VOLUME 288 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 36843 natant was collected. The protein concentration was determined by performing Bradford assay (Bio-Rad) according to manufacturer's instructions and calibrated to 2 g/l.

Nedd4-2 Knockdown in MDCK Cells
A double-stranded 5Ј-GGGAAGAGAAGGUGGACAA-3Ј siRNA targeting canine Nedd4-2 was employed as it has previously been demonstrated to result in robust knockdown of Nedd4-2 in MDCK cells (43). Non-targeting siRNA (5Ј-CCAUCCUGAUGUCGCAAUA-3Ј) was used as a negative control. Both siRNA oligonucleotides were purchased from Eurogentec. MDCK cells stably expressing Kv7.1 were transfected with 20 nM total siRNA oligonucleotides using siLentFect TM (Bio-Rad) according to manufacturer's instructions. In addition, eGFP (eGFP-pcDNA3) was co-transfected serving as a marker for siRNA-transfected cells. The cells were plated on glass coverslips and grown for 2 days before being treated with 10 M LY294002 for 90 min prior to fixation.

cRNA Generation and Oocyte Injection
cRNA was generated from linearized pXOOM plasmids carrying the gene of interest using the Ambion T7 m-Message Machine kit (Ambion, Austin, TX) according to the manufacturer's instructions. cRNA concentrations were determined using the ND-1000 NanoDrop UV spectrophotometer, and its integrity was confirmed by gel electrophoresis. cRNAs were stored at Ϫ80°C until injection. Xenopus laevis oocytes were obtained from EcoCyte Bioscience (Castrop-Rauxel, Germany). Oocytes were kept at 19°C in Kulori solution consisting of 4 mM KCl, 90 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, pH 7.4, for a minimum 24 h before injection. Injection of 50 nl of cRNA (1 ng of all Kv7.1, SGK1, and Akt variants, and 0.2 ng of Nedd4-2/Nedd4-2 CS) was accomplished using a Nanoject microinjector from Drummond (Drummond Scientific, Broomall, PA). Injected oocytes were kept at 19°C for 2 days prior to measurements.

Electrophysiological Recordings on X. laevis Oocytes
Kv7.1 currents were recorded using a two-electrode voltageclamp amplifier (Dagan CA-1B, Minneapolis, MN). Electrodes were pulled from borosilicate glass capillaries on a horizontal patch electrode puller (DMZ universal puller, Zeitz Instruments, München, Germany) and had tip resistance between 0.5 and 2.0 megohms when filled with 2 M KCl. The recordings were done at room temperature under continuous superfusion with Kulori solution. Kv7.1 currents were analyzed by applying a standard step protocol with pulses from Ϫ60 to ϩ40 mV (5 s) in 20-mV increments from a holding potential of Ϫ80 mV. The tail currents were measured at Ϫ30 mV. The condition of each single oocyte was tested before measurements by recording membrane potentials. Only oocytes with a membrane potential more negative than Ϫ40 mV were used in the experiments. The results were analyzed by GraphPad Prism software (GraphPad Software, San Diego, CA).

Data Analysis
Data analysis and drawings were performed using IGOR software (version 4, WaveMetrics, Lake Oswego, OR) or GraphPad Prism software (version 4, GraphPad Software, San Diego, CA). All deviations of calculated mean values are given as S.E. values. Statistical significance was determined by t test or one-way analysis of variance for more than two groups of data. The quantitative analysis of Western blot experiments was done using Odyssey Imaging Software (LI-COR Biosciences, Lincoln, NE). Confocal images were adjusted by the use of either Leica Confocal Software (Leica Microsystems, Mannheim, Germany) or Zen 2009 (Carl Zeiss) and Adobe Photoshop CS5.

Kv7.1 Surface Expression Requires Continuous PI3K Activity
in Polarized MDCK Cells-Epithelial cell polarization, a process dependent on the calcium concentration in the extracellular medium, can be controlled by the calcium switch assay (44). We have previously reported that Kv7.1 is retained intracellularly during the early stages of epithelial cell polarization induced by the calcium switch but is redistributed to the basolateral cell surface in a PI3K-dependent manner at a later stage (17). As PI3K activity not only regulates the formation of the basolateral membrane but also appears to be required for its preservation (28), we speculated whether tonic PI3K activity would be required to maintain Kv7.1 expression at the basolateral plasma membrane of polarized MDCK cells. To test this, MDCK cells were subjected to a 24-h calcium switch and then incubated in NCM containing 10 M of the PI3K inhibitor LY294002 for an additional 3 h. As illustrated in Fig. 1A, LY294002 treatment resulted in a disappearance of Kv7.1 from the basolateral cell surface and an intracellular accumulation of the channel (27 h, LY294002). Kv7.1 was primarily observed in vesicular structures and in the ER, as revealed by a partial colocalization with the ER marker, DsRed-ER. LY294002-treated cells displayed a 60% decrease in Kv7.1 signal at the cell surface compared with control cells (see Fig. 8B). Thus, PI3K activity appears to be required for maintaining basolateral surface expression of Kv7.1 in polarized MDCK cells. This effect was not due to changes in the overall polarity of the cells as the localizations of the tight junction protein ZO-1, the adherens junction protein E-cadherin, and the desmosome component desmoplakin all remained unaffected by a 3-h treatment with LY294002 ( Fig. 1, C and D).
To follow the effects of PI3K inhibition upon the localization of Kv7.1 more dynamically, live-cell imaging was performed on MDCK stably expressing EGFP-tagged hKv7.1. Z-stacks were acquired every 3 min for 1 h (starting at 30 min (24 h, 30 min) after addition of LY294002, Fig. 1B). The obtained images were then presented as a three-dimensional reconstruction (see supplemental Movie 1 or Fig. 1B). The live-cell imaging experiments confirmed our observations from fixed cells as LY294002 treatment resulted in loss of the Kv7.1-GFP membrane-associ-ated signal and an accumulation of the channel in vesicular structures.
To reveal the identity of the intracellular structures in which Kv7.1 resided, polarized MDCK cells treated with PI3K inhibitor for up to 1 h were co-stained for Kv7.1 and endosomal markers. The vesicular structures containing Kv7.1 partly colocalized with markers of the early (EEA1) and late (M6PR) endosomes as well as lysosomes (LAMP2, Fig. 2). This observation suggests that PI3K inhibition initiates endocytosis and subsequent degradation of surface-expressed Kv7.1 channels.
SGK1 Inhibition Mimics the Effects of PI3K Inhibition-As mentioned above, interesting downstream partners of PI3K are the SGK1 and Akt kinases, which have been reported to up-regulate Kv7.1 currents (31,32,(45)(46)(47). To test for a possible involvement of the two kinases upon Kv7.1 localization, the effects of the inhibitors GSK650394 (SGK1) and Akt inhibitor IV (Akt) were examined. In polarized MDCK cells, addition of 1 M of GSK650394 resulted in an intracellular accumulation of Kv7.1 (Fig. 3). In contrast, 1 M Akt inhibitor IV was without effect (Fig. 3). At higher concentrations (10 M), Akt inhibitor IV also caused intracellular accumulation of Kv7.1 (Fig. 3). However, at this concentration, Akt inhibitor IV can also target SGK1 (48), thus causing uncertainty in regards to the specificity of the effect. Quantifications of the Kv7.1 fluorescent signals revealed a decrease of ϳ60% in the level of Kv7.1 at the plasma membrane when inhibiting SGK1 (see Fig. 8B). Similar to the observations in polarized cells, GSK650394 prevented accumulation of Kv7.1 at the basolateral cell surface when applied during the calcium switch (Fig. 4). These observations suggest that SGK1, similarly to PI3K, regulates the basolateral surface expression of Kv7.1 in both polarizing and polarized MDCK cells.   DECEMBER 27, 2013 • VOLUME 288 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 36845
Constitutively Active SGK1 Prevents LY294002-induced Internalization of Kv7.1-To examine whether SGK1 could be the primary mediator of the PI3K effects on Kv7.1 localization, we tested the impact of PI3K inhibition in MDCK cells overexpressing a constitutively active SGK1 mutant, SGK1-S422D. We transfected MDCK cells stably expressing Kv7.1 with this mutant and applied LY294002 for 1-3 h to the cells at the polarized state. In support of SGK1 being the primary target of PI3K, SGK1-S422D overexpression prevented LY294002-induced removal of Kv7.1 from the basolateral cell surface (Fig. 5). To further establish the important role of SGK1, we examined the localization of Kv7.1 during the calcium switch in cells co-expressing SGK1-S422D (Fig. 6). SGK1-S422D expression prevented the clearance of Kv7.1 from the cell surface, which is normally observed upon switch initiation (17). This observation supports an important role for SGK1 in regulating Kv7.1 cell surface expression.
The PI3K-SGK1 Pathway Promotes Kv7.1 Surface Expression by Inhibiting Nedd4-2-dependent Endocytosis-To further examine whether PI3K and SGK1 act on Kv7.1 through a Nedd4-2-dependent mechanism, we tested the effects of PI3K and SGK1 inhibition upon the localization of a Nedd4-2 interaction-deficient Kv7.1 mutant (Kv7.1-YA). As speculated, the localization of Kv7.1-YA was unaffected by inhibition of PI3K and SGK1 in polarized MDCK cells, suggesting that the two kinases act through Nedd4-2 (Fig. 8). Moreover, the involvement of Nedd4-2 was additionally supported by the observation that inhibition of both PI3K and SGK1 were without effect on the membrane localization of Kv7.1-YA in polarizing cells, i.e. during the calcium switch (data not shown). Interestingly, the Kv7.1-YA mutant also displayed localization in the apical membrane (side scans in Fig. 8A, 27 h control, versus Fig. 1A, 27 h), indicating that Nedd4-2-dependent endocytosis is important to avoid apical trapping of the channel upon cell polarization.
To confirm that Nedd4-2 acts downstream of PI3K-SGK1, we reduced the expression of Nedd4-2 in MDCK cells stably expressing Kv7.1 by using a previously employed Nedd4-2 siRNA oligonucleotide (43). First, we verified that the siRNA oligonucleotide indeed decreased Nedd4-2 expression levels. As illustrated in Fig. 9C, we observed a reduction in the protein level of Nedd4-2 in cells transfected with the Nedd4-2-specific siRNA compared with control cells transfected with a non targeting control siRNA. By quantifying the signal intensity of the Nedd4-2 band and the corresponding actin band, we calculated an ϳ30% reduction in Nedd4-2 expression (data not shown). As our transfection efficiencies rarely exceed 25-30%, it indicates that cells transfected with the Nedd4-2-specific siRNA have a very robust reduction in Nedd4-2 protein expression. This is in agreement with previous results using this specific Nedd4-2 siRNA in MDCK cells (43). Therefore, MDCK cells were next transfected with Nedd4-2 siRNA followed by treatment of the cells with LY294002. Nedd4-2 silencing prevented LY294002induced internalization of Kv7.1, which was not the case when non-targeting siRNA was used as a control (Fig. 9A). Quantifi- cations revealed that LY294002-treated cells transfected with Nedd4-2 siRNA had significantly higher membrane to intracellular Kv7.1 signal ratio than the cells transfected with control siRNA (Fig. 9B), suggesting that Nedd4-2 silencing protects Kv7.1 from internalization upon PI3K inhibition.
Nedd4-2 is a direct substrate of SGK1 and contains three well described SGK1 phosphorylation sites (33,35). Phosphorylation of these sites increases interaction with proteins of the 14-3-3 family, keeping Nedd4-2 in an inactive phosphorylated state (40). To determine whether LY294002 treatment of polarized MDCK cells directly affected the phosphorylation state of Nedd4-2, we examined the degree of phosphorylation on one of these key sites, Ser-468 in human Nedd4-2, after treatment. As shown in Fig. 9D, exposure to 10 M LY294002 for 2 h significantly reduced the fraction of Nedd4-2 phosphorylated at this site by 28%. This observation suggests a direct impact of the PI3K-SGK1 pathway upon Nedd4-2 phosphorylation state.

Kv7.1 Turnover Is Primarily Mediated by a Nedd4-2-related Mechanism in Polarized MDCK Cells-
The central involvement of Nedd4-2 in the regulation of Kv7.1 endocytosis suggests that this ubiquitin ligase is a key regulator of Kv7.1 cell surface expression. To examine this further, we investigated the turnover of Kv7.1 and Kv7.1-YA in polarized MDCK cells. MDCK cells stably expressing Kv7.1 or Kv7.1-YA were subjected to a 24-h calcium switch, and protein synthesis was inhibited with cycloheximide for 3 h. Total cell lysates were harvested at time points 24 and 27 h and analyzed by Western blotting. As illustrated in Fig. 10A, Kv7.1 protein levels were reduced by ϳ50% after a 3-h cycloheximide treatment, indicating a fast turnover of the channel. In contrast, Kv7.1-YA protein levels were unaffected by the 3-h treatment with cycloheximide (Fig.  10B), demonstrating that the turnover of this mutant is significantly slower. This suggests that the normal turnover of Kv7.1 is primarily mediated by a Nedd4-2-dependent mechanism. Insulin Counteracts Nedd4-2-mediated Reduction in Kv7.1 Current-As a final step, we wanted to investigate whether activation of PI3K in Xenopus oocytes could rescue the reduction in Kv7.1 current observed upon co-expressing Nedd4-2. Insulin, an activator of PI3K, has previously been demonstrated to stimulate Kv1.5 currents in an SGK1-dependent manner when the channel was expressed in oocytes (49) and was therefore used to activate the proposed pathway. Oocytes expressing Kv7.1 alone or together with Nedd4-2 were incubated in Kulori buffer with and without insulin for 6 h prior to current measurements. Insulin treatment significantly rescued (p Ͻ 0.001) the Kv7.1 current reduction caused by Nedd4-2 co-expression (Fig. 11B), thereby indicating that activation of PI3K also can lead to inhibition of Nedd4-2 in Xenopus oocytes.

DISCUSSION
Epithelial cell polarization involves a cascade of processes that ultimately divide the membrane into two distinct parts: an apical and a basolateral. To ensure vectorial transport of ions and solutes, these two membrane domains express different ion channels and transporters. In this study, the mechanisms regulating cell surface expression of the basolateral Kv7.1 channel were studied in MDCK cells both during epithelial cell polarization and in polarized cells. We previously reported that the activity of PI3K is required for recovering Kv7.1 cell surface expression during the polarization process (17). Here, we demonstrate that tonic PI3K activity stabilizes cell surface expression of Kv7.1 in polarized cells, which positions PI3K as an important regulator of Kv7.1 localization and turnover.
To uncover the molecular mechanisms underlying the PI3Kmediated promotion of Kv7.1 cell surface expression, we looked at two well described downstream targets of PI3K, the homologous kinases SGK1 and Akt. We found that pharmacological inhibition of SGK1 in polarized MDCK cells resulted in disappearance of Kv7.1 channels from the surface membrane and intracellular accumulation of the channel. We observed similar effects by using Akt inhibitor IV but only at concentrations where it can also suppress SGK1 (48), suggesting that the observed effect could be due to inhibition of SGK1. In support   SGK1 and Akt are both known inhibitors of Nedd4-2 (33,34,50), and several observations in our study support that the PI3K-SGK1 signaling pathway promotes Kv7.1 cell surface expression by inhibiting Nedd4-2. First, SGK1 can antagonize Nedd4-2-mediated down-regulation of Kv7.1 currents in Xenopus oocytes. Second, insulin, a known activator of PI3K, also antagonizes Nedd4-2-mediated reductions in Kv7.1 currents. In addition, the Nedd4-2-interaction deficient Kv7.1-YA mutant did not respond to PI3K and SGK1 inhibition and Nedd4-2 knockdown in MDCK cells prevented endocytosis of the channel in response to PI3K inhibition. Nedd4-2 inhibition by the PI3K-SGK1 pathway is further supported by the observation that the LY294002 treatment results in decreased phosphorylation of Ser-468 in Nedd4-2, a previously described SGK1 phosphorylation site, which plays a critical role in mediating interaction with the inhibitory 14-3-3 proteins (40). Overall, these observations suggest that Nedd4-2 is the primary effector of the PI3K-SGK1 pathway controlling Kv7.1 cell surface expression and positions Nedd4-2 as an important regulator of Kv7.1. Indeed, the Kv7.1-YA mutant displayed a much slower turnover than the wild-type channel, indicating that Nedd4-2 is a key regulator of the Kv7.1 degradation process. In support of this, AMP-activated protein kinase-mediated endocytosis of Kv7.1 in the beginning of the calcium switch is also Nedd4-2-dependent (18). Interestingly, AMP-activated protein kinase activation in polarized MDCK cells is without effect upon Kv7.1 surface expression, which suggests that the Nedd4-2 activation is overruled by other pathways at the more polarized state (18). The PI3K-SGK1-mediated inhibition of Nedd4-2-dependent endocytosis described in this study could explain this observation.
The PI3K-SGK1 pathway possibly promotes Kv7.1 surface expression by additional mechanisms. SGK1 was reported previously to increase the surface expression of the Kv7.1-KCNE1 channel complex by promoting RAB11-mediated channel recycling (32). In line with this observation, we also observed an increase in Kv7.1 current levels upon SGK1 co-expression. Interestingly, co-expression of SGK1 and Nedd4-2 did not increase the current level to that of the control but fully restored it to the same level as SGK1-expressing oocytes. This indicates that there are possibly two different SGK1 pathways that work simultaneously to promote the surface expression of Kv7.1. The Nedd4-2 pathway decreases internalization of surface-expressed channels, whereas the RAB11 pathway enhances forward trafficking and recycling. Interestingly, a similar two-way mechanism appears to regulate the surface expression of the epithelial sodium channel ENaC. ENaC is regulated by Nedd4-2, and SGK1 promotes ENaC surface expression by inhibiting the interaction between Nedd4-2 and ENaC. However, SGK1 also increases the surface expression of ENaC channels carrying PY motif mutations, suggesting that an additional pathway exists (51). RAB11 has recently been shown to enhance the surface expression of ENaC, which raises the possibility that the mechanism is similar to that reported for Kv7.1 (52,53).
One explanation for activating and inhibiting Nedd4-2-dependent endocytosis during polarization could be to avoid localization of Kv7.1 channels in the apical membrane after the The PI3K-SGK1-Nedd4-2 pathway described here could also regulate Kv7.1 surface expression levels under other conditions than cell polarization. As mentioned, Kv7.1 regulates acid secretion in the stomach. Interestingly, glucocorticoids have been reported to increase gastric acid secretion and up-regulate Kv7.1 by an SGK1-dependent mechanism (54). SGK1-mediated inhibition of Nedd4-2 could possibly explain this effect. Furthermore, the pathway could be important in other cell FIGURE 9. Nedd4-2 knockdown protects Kv7.1 from internalization upon PI3K inhibition. A, MDCK cells stably expressing Kv7.1 were transiently cotransfected with eGFP and siRNA targeting Nedd4-2 or a non-coding siRNA, which was used as a negative control. Polarized cells were treated with 10 M PI3K inhibitor LY294002 for 90 min. The localization of Kv7.1 was examined in transfected cells expressing eGFP (GFP) as these were expected also to be transfected with the siRNA. As demonstrated, Kv7.1 was internalized in cells expressing the non-targeting control siRNA. This internalization was not observed in cells expressing siRNA targeting Nedd-2 (Nedd4-2 siRNA). Scale bar, 10 m. B, quantification of membrane and intracellular Kv7.1 signals from cells expressing either non-targeting control siRNA or siRNA targeting Nedd4-2 (Nedd4-2 siRNA). The two bars demonstrate the ratios between membrane and intracellular signals in cells expressing the two siRNAs. The data are from four experiments with a total n Ͼ 17. The cells transfected with Nedd4-2 siRNA have a significantly higher membrane to intracellular signal ratio than the cells transfected with control siRNA; ***, p Ͻ 0.001. Bars represent means of each group Ϯ S.E. C, Western blot of MDCK cells transfected with control siRNA (lane 1) or Nedd4-2 siRNA (lane 2). A reduction in the signal intensity of the band at ϳ120 kDa, which correspond to Nedd4-2, is observed in lane 2 compared with the lane 1. Actin was used as a loading control. D, Western blots of cell lysates from polarized, non-treated control MDCK cells (control) and cells treated with 10 M LY294002 (LY294002) for 2 h. The blots were probed for total Nedd4-2 (left blot) or with Ser-468 phospho-specific Nedd4-2 antibody (right blot). The graph to the right displays band intensity quantifications from n ϭ 3 experiments. The amount of phospho-Nedd4-2 relative to the total amount of Nedd4-2 is shown. The control condition was set to 1. A significant reduction in the relative signal for phospho-Nedd4-2 was observed in response to LY294002 treatment. *, p Ͻ 0.05. Actin was used as a loading control.
types and not only limited to epithelial cell physiology. As mentioned, Kv7.1, together with KCNE1, forms the delayed rectifier potassium current I Ks , which is one of the main contributors to the repolarization of the cardiac tissue (1,2). Mutations leading to loss of channel function causes the most common type of congenital LQT syndrome, type 1 (LQT1). It results in a reduction of I Ks and thus a prolongation of action potential duration and the QT interval (see Ref. 55), which often causes dangerous arrhythmias. A recent study reported a mechanism for druginduced QT interval prolongation that involves changes in several ion currents caused by a decrease in PI3K signaling (56). Although PI3K inhibition of canine cardiac myocytes caused an increase in action potential duration, this could be reversed by intracellular infusion of PIP3. This increase in action potential duration was caused by a decrease in I Ks as well as a number of other action potential-shaping currents. Interestingly, the effects of PI3K inhibition were only observed after prolonged exposure suggesting that the effect was not due to direct channel modulation. As Nedd4-2 is also strongly expressed in heart (36), the identified PI3K-SGK1-Nedd4-2 pathway could well explain how PI3K inhibition can cause the observed decrease in I Ks .
In summary, the present study positions Nedd4-2 as a key regulator of Kv7.1 cell surface expression. In our previous work, we found that an AMP-activated protein kinase pathway inhib-  its Kv7.1 surface expression through activation of Nedd4-2 (18). Here, we report a PI3K-SGK1 mechanism with the opposing effect on Kv7.1, namely, stabilizing its surface expression, through inhibition of the ubiquitin ligase Nedd4-2.