Phosphatidylinositol 4-Phosphate 5-Kinase Reduces Cell Surface Expression of the Epithelial Sodium Channel (ENaC) in Cultured Collecting Duct Cells*

Ubiquitination of ENaC subunits has been shown to negatively regulate the cell surface expression of ENaC channels. We have previously demonstrated that epsin links ubiquitinated ENaC to clathrin adaptors for clathrin-mediated endocytosis. Epsin is thought to directly modify the curvature of membranes upon binding to phosphatidylinositol 4,5-bisphosphate (PIP2) where it recruits clathrin and stimulates lattice assembly. Murine phosphatidylinositol 4-phosphate 5-kinase α (PI5KIα) has been shown to enhance endocytosis in a PIP2-dependent manner. We tested the hypothesis that PI5KIα-mediated PIP2 production would negatively regulate ENaC current by enhancing epsin-mediated endocytosis of the channel. Expression of PI5KIα decreased ENaC currents in Xenopus oocytes by 80%, entirely because of a decrease in cell surface ENaC levels. Catalytically inactive mutants of PI5Kα had no effect on ENaC activity. Expression of the PIP2 binding region of epsin increased ENaC current in oocytes, an effect completely reversed by co-expression of PI5KIα. Overexpression of epsin reduced amiloride-sensitive current in CCD cells. Overexpression of PI5KIα enhanced membrane PIP2 levels and reduced apical surface expression of ENaC in CCD cells, down-regulating amiloride-sensitive current. Knockdown of PI5KIα with isoform-specific siRNA resulted in a 4-fold enhancement of ENaC activity. PI5KIα localized exclusively to the apical plasma membrane domain when overexpressed in mouse CCD cells, consistent for a role in regulating PIP2 production at the apical plasma membrane. We conclude that membrane turnover events regulating ENaC surface expression and activity in oocytes and CCD cells can be regulated by PI5KIα.

The epithelial sodium channel (ENaC) 4 localizes to the apical plasma membrane of epithelial cells where it functions to regulate Na ϩ transport. ENaC activity is critical to the physiological maintenance of Na ϩ homeostasis, volume regulation, and thus, systemic blood pressure (1,2). Because of its central role in responding to extreme changes in Na ϩ intake, ENaC transport rates are tightly regulated and defects in the channel or factors regulating the channel result in inherited forms of hypertension and hypotension. Key modulating factors in controlling epithelial Na ϩ transport are regulation of the mechanisms that facilitate ENaC surface expression or P o . ENaC surface expression is regulated by a variety of hormonal and non-hormonal factors, including mineralocorticoids and insulin as well as osmotic changes (2)(3)(4)(5). The balance of membrane delivery and retrieval events are coordinated to regulate the number of channels available at the apical membrane for Na ϩ reabsorption (6,7). Indeed, alterations in the removal of ENaC channel from the apical plasma membrane appear to be the underlying cause of Liddle's Syndrome (8). Clearly, identifying the factors that control insertion, retrieval, and recycling of membrane channels will aid in our understanding of the complex action of hormones, physiological and pathological conditions that are associated with altered channel expression or activity.
It is widely accepted that ENaC surface expression is negatively regulated by ubiquitination (9 -13). Elegant studies have revealed that the ubiquitin ligase Nedd4-2 binds to a consensus PPXY motif in the C terminus of ENaC subunits and mediates the ubiquitination of lysine residues on the N terminus of the ␣and ␥-subunits of ENaC in MDCK cells (14,15). Ubiquitination of membrane proteins serves as a signal for retrieval of these proteins from the cell surface into the endosomal/lysosomal pathway (16). Recently, we demonstrated that ENaC binds to epsin, which facilitates the internalization of ubiquitinated ENaC from the apical plasma membrane of epithelial cells via clathrin-mediated endocytosis (17). Epsin is thought to directly modify the curvature of membranes upon binding to PIP 2 where it recruits clathrin and stimulates lattice assembly (18,19). The rate of clathrin-mediated endocytosis appears to be influenced by PIP 2 metabolism in a variety of ways. PIP 2 has been shown to increase the affinity of AP-2 complexes for the internalization signals of a variety of cargo molecules (20). AP-2, which, like epsin, functions to promote clathrin-coated vesicle assembly, also contains PIP 2 selective domains and mutations in the PIP 2 binding regions of epsin or the 2 subunit of AP-2 interfere with endocytosis (21,22). PIP 2 clearly participates in the regulation of numerous cell processes in addition to endocytosis. Proteins responsible for actin polymerization and cytoskeletal rearrangement are recruited to specific membranes via their interactions with PIP 2 (23,24). Furthermore, a diverse class of ion channels and transporters can be regulated by direct interaction with PIP 2 and/or are sensitive to dynamic changes in PIP 2 concentrations at the plasma membrane (25). Indeed, PIP 2 has been implicated in the up-regulation of ENaC activity by effects on exocytosis of channels or by direct effects on P o (26 -30).
The versatility assigned to PIP 2 suggests that there are spatially and temporally regulated pools of PIP 2 in the plasma membrane. Generation of PIP 2 at the plasma membrane can be accomplished by the three isoforms of the type I phosphatidylinositol 4-phosphate 5-kinases (PI5Ks) ␣, ␤, and ␥ (31). The mouse and human nomenclature for the ␣ and ␤ isoforms are reversed; for these studies we used the murine cDNA for PI5K␣. With regard to endocytosis, the activity of all three isoforms appears to influence the rate of internalization in various cell types. The assembly of the AP-2 clathrin adaptor complex to the plasma membrane has been shown to be influenced by the activity of both the murine ␣ and ␥ isoforms of PI5K (32,33). At the synapse, a very active area of rapid endocytosis, PI5KI␥ has been shown to be the predominant enzyme involved in PIP 2 production (34). Interestingly, PIP 2 production via murine PI5KI␤ has been shown to increase the regulated process of internalization of EGF receptors, whereas PI5KI␣ has no effect in fibroblasts (35). Most recently, the direct interaction of type I PI5Ks with AP-2 complexes has been shown to stimulate PI5K activity and the PI5KI␥ isoform has been shown to interact with AP-2 complexes and regulate endocytosis of transferrin (33,36). The majority of these studies were performed in nonpolarized cells or have examined the traffic of basolateral cargo only. It is not clearly known which isoforms are physiologically relevant to apical trafficking in polarized epithelial cells or those which contribute to the balance of ENaC turnover at the apical plasma membrane. The following experiments were designed to test whether modulation of PIP 2 levels via PI5KI␣ functions to regulate ENaC surface expression and activity in native epithelial cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Adenoviral Expression-mpkCCD c14 cells (CCD cells) were derived from the cortical collecting ducts of SV40 transformed mice and were the generous gift of Alain Vandewalle (37). CCD cells were maintained in Dulbecco's modified Eagle's medium/Ham's F12 with 2% fetal bovine serum as previously described. For adenovirus infection, cultured CCD cells were grown on 12-mm transwells (0.4-m pore; Costar, Cambridge, MA), rinsed extensively with PBS and incubated for 1 h at 37°C on 50-l drops of PBS-containing recombinant adenoviruses and 150 l of PBS/virus on the api-cal surface of the transwell, using a multiplicity of infection (m.o.i.) of 100. Recombinant adenovirus encoding the constitutive expression of the tetracycline-repressible transactivator at an m.o.i. of 50 was included in all experiments to enable doxycycline-repressible synthesis of appropriate PI5KI␣ tagged with an HA epitope (38). Following infection, cells were incubated in complete media supplemented with 20 ng/ml doxycycline for 24 h. Subsequently, cells were rinsed in complete media and incubated for 16 h in the absence of doxycycline to allow for expression of PI5KI␣ constructs. Infection efficiency was visualized by immunofluorescence with anti-HA antibodies and was estimated at greater than 90% for all viruses.
DNA and Replication-defective Recombinant Adenoviruses-The cDNA for murine PI5KI␣ was kindly provided by Chris Carpenter (Harvard Medical School). This construct was utilized for the generation and purification of replication-defective recombinant adenoviruses (AVs) encoding the tetracycline-repressible construct encoding PI5KI␣ as described (39). Site-directed mutagenesis to generate the D227A and D203A mutations was performed in the pAdtet PI5KI␣ cDNA construct by using the QuikChange TM system (Stratagene).
Whole Cell Measurements-A two-electrode voltage clamp technique was used as preciously described. Whole cell inward amiloride-sensitive currents were measured in control oocytes expressing ␣␤␥ ENaC alone or experimental oocytes expressing ␣␤␥ ENaC with appropriate PI5KI␣ constructs using a DigiData 1200 interface (Axon Instruments, Foster City, CA) and a TEV 200A voltage Clamp amplifier (Dagan Corp., Minneapolis, MN). Data acquisition and analysis were performed using pClamp 7.0 (Axon Instruments). Amiloride-sensitive currents were defined as the difference of current in the absence and the presence of 0.1 mM amiloride. Membrane potentials were clamped from Ϫ140 to ϩ60 mV in 20-mV increments with duration of 900 ms. Currents were measured at a holding potential of Ϫ100 mV, 600 ms after initiation of the clamp potential. For ROMK measurements, oocytes were bathed in a 45 mM K ϩ solution (45 mM KCl, 45 mM N-methyl-D-glucamine-Cl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, pH 7.4). Voltage sensing and current injecting microelectrodes had resistances of 0.5-1.5 M⍀ when backfilled with 3 M KCl. Once a stable membrane potential was attained, oocytes were clamped to a holding potential of Ϫ20 mV and currents were recorded during 500 ms voltage steps, ranging from Ϫ100 mV to ϩ40 mV in 20 mV increments. Data were collected using an ITC16 analog-to-digital, digital-to-analogue converter (Instrutech Corp.), filtered at 1 kHz and digitized on line at 2 kHz using Pulse software (HEKA Electronik) for later analysis. Data shown are based on the average maximum current, measured at Ϫ100 mV.
Cell Surface ENaC and ROMK Labeling-Plasma membrane expression of the external HA-tagged ROMK channel and external FLAG-tagged ENaC was measured in single oocytes following procedures outlined as previously reported (40) with slight modifications. In these studies, oocytes were fixed with 4% formaldehyde in oocytes Ringers for 15 min. at 4°C and washed four times in oocytes Ringers. To block spurious antibody binding, oocytes were then incubated for 1 h at 4°C in MBS containing 1% bovine serum albumin (BSA). Exposed HA or FLAG epitopes on the surface of intact oocytes were labeled with a rat monoclonal anti-HA antibody (0.5 g/ml, Roche 3F10, 1% BSA, 4°C, overnight) or 1 mg/ml mouse monoclonal anti-FLAG antibody (Sigma) for 1 h. Oocytes were washed with MBS containing 1% BSA, and incubated with horseradish peroxidase-coupled goat anti-rat or goat anti-mouse (1 g/ml, Jackson Laboratories, 1% BSA, 1.5 h). After extensive washing, individual oocytes were placed in chemiluminescence substrate as described (41). Luminescence from single oocytes was measured as described previously (40,41).
Cell Surface Biotinylation-The cell surface pool of ENaC was determined by surface biotinylation as described (42). Briefly, CCD cells cultured on Transwells were washed (5 min) with ice-cold PBS with agitation on ice to remove growth media. Apical domain selective biotinylation was performed in borate buffer (85 mM NaCl, 4 mM KCl, 15 mM Na 2 B 4 O 7 , pH 9) for 20 min. The basolateral surface was incubated in growth medium containing fetal bovine serum (FBS) to prevent biotinylation. The biotinylation was quenched by FBS incubation and PBS washing. Monolayers were lysed (0.4% DCA, 1% Nonidet P-40, 50 mM EGTA, 10 mM Tris-Cl, pH 7.4) at room temperature for 10 min. Protein concentration of the postnuclear supernatant was determined, and 200 g of protein was combined with a streptavidin bead slurry (Pierce Chemical Co.) and incubated overnight at 4°C. Samples were collected in 2ϫ sample buffer containing 10% ␤-mercaptoethanol and incubated for 20 min at room temperature. Samples were SDS-PAGE and Western blot analysis and visualized with enhanced chemiluminescence. Equal concentrations of cell lysate were analyzed in parallel to determine total the total cellular pool of ENaC.
Indirect Immunofluorescence-CCD cells grown on transwell filters were infected with tet-responsive adenoviruses encoding HA epitope-tagged PI5KI␣ and/or a GFP-tagged PH domain from PLC␦ and the tetracycline-repressible transactivator at a m.o.i. 100 as described above. Following infection and expression of the adenovirus, cells were rinsed in PBS and processed for immunofluorescence microscopy. Briefly, the cells were fixed with formaldehyde using a pH-shift protocol as described (43). HA epitope-tagged PI5KI␣ was visualized using monoclonal anti-HA antibody (1:500) (Covance) and AlexaFluor secondary antibodies (1:500) (Molecular Probes), ZO-1 was visu-alized using rat anti-ZO-1 hybridoma supernatant R40.76 (1:10) and AlexaFluor secondary antibody (1:500) (Molecular Probes). Images were acquired using a Leica TCL-SL confocal microscope equipped with argon and green and red helium neon lasers (Leica, Dearfield, IL). Images were taken with a 100ϫ (1.4 numerical aperture) plan apochromat oil objective and the appropriate filter combination. The images (1024 ϫ 1024 pixels) were saved in a tag-information-file-format (TIFF), and image contrast was adjusted in the Photoshop program (Adobe, Mountain View, CA).
Thin Layer Chromatography Analysis of Phospholipids-Lipid analysis was performed essentially as described in Ref. 44. Xenopus oocytes expressing appropriate cDNAs were incubated overnight with [ 32 P]orthophosphate in modified Barth's media. The following morning samples were washed 1ϫ with modified Barth's to remove unincorporated 32 P. Oocytes were homogenized in 300 l of 1 N HCl, and lipids were extracted as described below. AV-infected CCD cells grown on 6-cm dishes were starved in phosphate-free buffer for 30 min, radiolabeled for 4 h with 40 Ci/ml [ 32 P]orthophosphate (ICN). Cells were rinsed gently in ice-cold buffer and scraped into a 4:3:3 mixture of CHCl 3 /CH 3 OH/1 N HCl. The organic phase was washed with equal volumes of CH 3 OH and 1 N HCl. Aliquots were counted using a scintillation counter, and equal counts/min were spotted onto oxalate-treated Silica gel 60 TLC plates (EM Science) and developed in 1-propyl alcohol, 2 M acetic acid; H 2 O (63:4: 33). Authentic lipid standards (Avanti Polar Lipids) were included in all experiments and visualized using iodine vapor. Radiolabeled products were visualized and quantitated using a phosphorimager (Bio-Rad).
Internalization of IgA-IgA was radioiodinated using the iodine monochloride method as described (45). Endocytosis of 125 I-IgA was performed essentially as described by Apodaca et al. (46). Briefly, cells were incubated with 125 I-IgA applied either to the basolateral or apical surface on ice for 1.5 h. Cells were washed with ice-cold media and incubated in prewarmed 37°C media in a water bath for appropriate time points. After each time point, 125 I-IgA that was not internalized from the cell surface was removed by treating the cells with 25 mg/ml L-1tosylamide-2-phenylethylchloromethyl-ketone-treated trypsin. The filters were washed and removed from the insert, and the amount of cell-associated 125 I-IgA and that which was removed by washing and trypsin treatment were determined using a gamma counter (Packard Instrument, Downers Grove, IL) as described (46). An equal number of mock-infected CCD cells not expressing the pIgR were treated identically to determine nonspecific IgA uptake, and these values were subtracted from those of the pIgR-infected CCD cells. After the final time point, filters were cut out of the insert, and the amount of 125 I-IgA in all samples was determined using a gamma counter. An equal number of mock-infected CCD cells not expressing the pIgR were treated identically to determine nonspecific IgA uptake, and these values were subtracted from those of the pIgR-infected CCD cells.
Overexpression of Epsin-Mouse CCD cells were transfected with a Myc-tagged epsin construct (provided by Linton Traub, University of Pittsburgh) using Lipofectamine 2000 standard protocol. The following day, the cells were taken up by trypsin PI5KI␣-mediated Effects on ENaC treatment, pelleted with a brief low speed centrifugation, and plated on 12-mm filter inserts (Transwells) at superconfluency. Two days later, the cells were used for electrophysiological or biochemical measurements.
siRNA of PIP5K␣-Approximately 2 ϫ 10 6 of 80 -90% confluent cells were resuspended in nucleofector T solution and 2 pmol (final concentration) of siRNA specific for PI5K I␣ (Dharmacon), mixed in the cuvette and directly placed in the nucleofector device (Amaxa Biosystems). The siRNA oligonucleotides were electroporated into the cells using program K-29. An equal volume of warm culture media was added, and the cell suspension plated on a plastic 12-well plate containing 2 ml of culture media. The following day, the cells were trypsinized and plated on 12-mm filter inserts (Transwells) at superconfluency. Two days later, the cells were used for electrophysiological or biochemical study.

RESULTS
The effect of PI5KI␣-mediated PIP 2 production on ENaC activity was initially examined using the Xenopus oocyte expression system. To determine to what extent PIP 2 production was modified in oocytes expressing PI5KI␣, lipids were extracted from oocytes injected with PI5KI␣ or control cRNAs and examined by thin layer chromatography. As shown in Fig.  1A PI5KI␣ induced a 40% increase in steady state PIP 2 production in oocytes compared with control. Whole cell amiloridesensitive currents were measured in oocytes coinjected with cRNAs for ␣␤␥ ENaC and PI5KI␣. Co-expression of PI5KI␣ with ␣␤␥ ENaC reduced amiloride-sensitive current by 80% (Fig. 1B). To confirm that the effect of PI5KI␣ on ENaC surface activity is due to modulation of PIP 2 production, we examined the effect of catalytically inactive mutants PI5KI␣ D227A (47) and PI5KI␣ D203A 5 on ENaC activity in Xenopus oocytes. Co-expression of either mutant with ENaC has no effect on amiloridesensitive currents compared with control oocytes expressing ENaC subunits alone (Fig. 1B), indicating that the catalytic activity of PI5KI␣ is necessary for the down-regulation of ENaC channel activity. PIP 2 has been shown to enhance endocytic and exocytic events to control steady state levels of proteins in the plasma membrane. To determine whether the PI5KI␣-mediated reduction in ENaC activity is caused by differences in cell surface expression, plasma membrane-associated ENaC channels were examined using an enzyme-linked luminescence assay (40). The surface expression signal of externally FLAG epitopetagged ENaC was roughly 10-fold higher than the background level of oocytes expressing the wild-type ENaC lacking an external FLAG epitope (Fig. 1C, ␣␤␥ versus ␣␤ flag ␥). Co-expression of PI5KI␣ with ␣␤ flag ␥ ENaC reduced the surface signal of ENaC by nearly 6-fold (Fig. 1C) correlating with the reduction observed in amiloride-sensitive current in Fig. 1A. These data demonstrate that the reduction of ENaC activity by PI5KI␣ is caused by an effect on surface density. This negative effect of PIP 2 production on ENaC activity and surface expression is consistent with the role of PIP 2 in enhancing retrieval of ENaC from the plasma membrane.
Type I PI5Ks have been shown to produce specific pools of PIP 2 that can be utilized for distinct membrane trafficking events (34,48,49). To determine whether PI5KI␣-mediated 5 C. Carpenter, personal communication. Oocytes injected with cRNA for PI5KI␣ were 32 P-labeled overnight subject to lipid extraction. Equal counts of lipids were resolved by TLC and quantified by phosphorimage analysis. Averages of PIP 2 levels from oocytes expressing PI5KI␣ are graphed relative to the control (C ϭ 100), n ϭ 3. Panel B, PI5KI␣mediated PIP 2 production reduces amiloride-sensitive currents in oocytes expressing ENaC subunits. Xenopus oocytes were co-injected with the cRNAs for ␣␤␥ ENaC and the indicated type I PI5K. Whole cell amiloride-sensitive currents were measured 24 -46 h post-injection. Co-expression of wild-type PI5KI␣ reduced amiloride-sensitive currents by 80% while expression of catalytically inactive mutants PI5KI␣ D227A or PI5KI␣ D203A had no effect on ENaC channel activity. Results are normalized to control currents. n ϭ 18 -32; *, p Ͻ 0.001 compared with control. Panel C, PI5KI␣ reduces cell surface expression of ENaC in Xenopus oocytes. Xenopus oocytes were co-injected with external FLAG epitope-tagged ␤-ENaC with wild-type ␣ and ␥ (␣␤ flag ␥) alone or in combination with PI5KI␣. Wild-type ␣␤␥ lacking FLAG epitope-tagged ␤ (␣␤␥) is expressed as control for background. Cell surface expression of FLAGtagged ENaC was analyzed by luminometry as described under "Experimental Procedures." Co-expression of PI5KI␣ reduced the surface expression of epitope-tagged ENaC nearly 6-fold. n ϭ 25. p Ͻ 0.001 compared with ␣␤ flag ␥. PIP 2 production is important for constitutive trafficking and steady state distribution of other membrane proteins, we tested the effect of PI5KI␣ on the activity and surface expression of the renal potassium secretory channel, ROMK. Interestingly, PI5KI␣ expression enhanced ROMK activity in oocytes ( Fig. 2A). This effect was dependent upon PIP 2 production, as catalytically inactive mutants had no effect on ROMK activity ( Fig. 2A). Furthermore, PI5KI␣ induced a 50% increase in ROMK surface expression in oocytes as detected by luminometry, indicating that the up-regulation in ROMK activity is caused by increased levels of ROMK channels at the membrane (Fig. 2B). These observations suggest that the PI5KI␣-mediated effects on ENaC surface expression may be related to specific protein-protein interactions relevant to ENaC internalization.
We have previously shown that the adaptor protein epsin, which contains an N-terminal PIP 2 binding ENTH domain, functions in the retrieval of ubiquitinated ENaC from the plasma membrane (17). Co-expression of the epsin ENTH domain with ENaC subunits in Xenopus oocytes results in increased current compared with currents obtained upon expression of ENaC subunits alone. We hypothesized that expression of this domain sequesters PIP 2 at the plasma membrane, thereby inhibiting endocytosis. To test this, we examined whether enhancing levels of PIP 2 through co-expression of PI5KI␣ would block ENTH domain-mediated stimulation of ENaC current. Indeed, PI5KI␣ reversed the effect we observed with the ENTH domain alone by reducing amiloride-sensitive currents to below control levels in oocytes, (Fig. 3).
As noted above, the underlying hypothesis of this work is that ENaC subunits, ubiquitinated at the plasma membrane (11)(12)(13) is linked to clathrin adaptors by epsin, which in turn recruits PIP 2 to the complex, facilitating internalization via clathrinmediated endocytosis. Although our data in oocytes, along with data by ourselves and others (11,12,50) indicating interactions between ENaC and clathrin adaptors, tend to support this model, we next sought to examine this directly in CCD cells that endogenously express ENaC. We first examined the effects of overexpression of epsin and PI5KI␣ on ENaC activity in CCD cells. As shown in Fig. 4A, both proteins induce a substantial and virtually identical reduction in ENaC current in CCD cells. To examine the functional role of the pool of PIP 2 generated by PI5KI␣ on native ENaC channels we determined whether expression of wild-type PI5KI␣ would raise PIP 2 production in CCD membranes. Following infection, cells were labeled with [ 32 P]orthophosphate, lipids extracted, and analyzed by TLC (Fig. 4B). Overexpression of PI5KI␣ more than doubled the amount of PIP 2 produced in CCD cells compared with control infected cells. Transepithelial amiloride-sensitive currents were measured in parallel experiments to determine the effect of enhanced PIP 2 production via PI5KI␣ in CCD cells. Fig. 4A demonstrates that enhanced PIP 2 production resulted in a significant reduction in ENaC activity, consistent with the effects observed in Xenopus oocytes (Fig. 1B). Furthermore, the expression of the catalytically inactive mutants PI5KI␣ D227A of PI5K1␣ D203A had no effect on the amiloride-sensitive current in CCD cells compared with control (Fig. 4C). Next, we determined whether this effect was caused by changes in ENaC sur- Once a stable membrane potential was attained, oocytes were clamped to a holding potential of Ϫ20 mV and currents were recorded during 500-ms voltage steps, ranging from Ϫ100 mV to ϩ40 mV in 20-mV increments. Data shown are based on the average maximum current, measured at Ϫ100 mV. Panel B, cell surface expression of HA-tagged ROMK channel was measured in single oocytes as described under "Experimental Procedures." face expression. The steady state levels of ENaC expressed at the surface were determined by domain-selective cell surface biotinylation as described under "Experimental Procedures." The total cellular pool of ENaC channels was unchanged by PI5KI␣ expression in immunoblots of CCD lysates probed with anti-␣ ENaC antibodies, (Fig. 5A, top panel). However, cell surface biotinylation of polarized CCD cells demonstrate that the amount of ENaC expressed at the cell surface was reduced by 60% in cells expressing PI5KI␣ compared with control, Fig. 5A, bottom panel and Fig. 5B). These observations are consistent with the role of PIP 2 in promoting internalization events at the plasma membrane. If this is correct, then knockdown of PI5KI␣ in CCD cells should inhibit internalization, as it did in nonpolarized cells (32), and potentially result in increased ENaC activity. As shown in Fig. 6B, CCD cells express all three isoforms of Type I PI5K. To knockdown expression of the specific isoform which we had previously overexpressed, CCD cells were treated with small interfering RNA (siRNA) constructs specific for the I␣ isoform of murine PI5K. ENaC currents and abundance of PI5KI␣ were measured 2 days later. As shown in Fig. 6A, treatment of the cells resulted in substantial reduction in expression of the kinase and a 4-fold enhancement of amiloride-sensitive current.   We next examined the localization of PI5KI␣ in polarized and nonpolarized CCD cells. Cells were infected with HA epitope-tagged PI5KI␣ and processed for immunofluorescence 16 h after induction of expression. PI5KI␣ was highly enriched in the apical plasma membrane of polarized CCD cells as assessed by confocal microscopy. In xzy sections of CCD cells, PI5KI␣ (green, Fig. 7A) localizes predominantly at the apical membrane and does not stain the lateral borders, delineated by antibodies staining the junctional marker, Zo-1 (red, Fig. 7A). When CCD cells are grown subconfluently on glass coverslips, PI5KI␣ localizes predominantly at the plasma membrane (red, Fig. 7B) where it localizes with the PIP 2 -binding PH domain from PLC␦ (green, Fig. 7B) as reported in the literature (32). PI5KI␣ also can be found on intracellular structures in nonpolarized epithelial cells (Fig. 7B), consistent with previous studies (39).
The reduced pool of ENaC at the cell surface in PI5KI␣ cells prohibited experiments to determine the kinetics of internalization. Therefore, we examined the effect of PI5KI␣ on the internalization of IgA and its receptor, the polymeric immunoglobulin receptor (pIgR) in polarized CCD cells. CCD cells were co-infected with adenoviruses encoding pIgR and either PI5KI␣ or control adenoviruses. 125 I-IgA was prebound to pIgR molecules at the basolateral surface at 4°C, the cells were warmed to 37°C for the indicated times and cell associated 125 I-IgA was quantified in a gamma counter as described under "Experimental Procedures." As shown in Fig. 8A, the percent of IgA internalized from the basolateral surface was unaltered by overexpression of PI5KI␣. Similar experiments were conducted to follow the internalization of IgA from the apical domain, which has been reported in other cell lines. However, apical endocytosis of this ligand was virtually undetectable in CCD cells, and the results could not be reliably interpreted. PI5KI␣ has previously been shown to affect rates of internalization in nonpolarized cells (32), thus, we measured 125 I-IgA endocytosis in nonpolarized CCD cells infected with pIgR and PI5KI␣ or control adenoviruses. In contrast to our observations in polarized cells, IgA endocytosis was dramatically stimulated in nonpolarized cells, Fig. 8B. These results indicate that the pool of PIP 2 produced by PI5KI␣ is further compartmentalized in polarized cells, and selectively regulates internalization at the apical surface.

DISCUSSION
Here we have tested the role of PI5KI␣ on regulating the surface expression of ENaC. We show that PI5KI␣ down-regulated ENaC activity in Xenopus oocytes. The decrease in ENaC activity was dependent on the production of PIP 2 , as PI5KI␣ mutants defective in catalytic activity had no effect. This observation could be reproduced in CCD cells which natively express ENaC, indicating that the mechanism for PI5KI␣-mediated down-regulation is preserved in a polarized epithelial cell model. Furthermore, a parallel decrease in surface expression with ENaC activity is observed in oocytes and in polarized CCD cells co-expressing PI5KI␣ (Fig. 1, A and C and Fig. 4). Consistent with these observations, knockdown of PI5KI␣ by siRNA  resulted in striking increase in ENaC activity in CCD cells. Interestingly, the effects of this kinase on ENaC activity were not due to global changes in membrane internalization, as activity and expression of ROMK in oocytes was enhanced, consistent with an effect on exocytosis of this channel (Fig. 2, A  and B). These observations indicate that the influence of PI5KI␣ on ENaC activity and surface expression may be caused by the interactions required for ENaC internalization. We have previously shown that ubiquitinated ENaC interacts with epsin to promote its internalization through clathrin-coated vesicles (17). Epsin recruitment is PIP 2 -dependent; therefore, it is likely that PI5KI␣ effects are caused by modulating epsin-PIP 2 interactions through the ENTH domain in epsin. Indeed, we observed that while the expression of ENTH domain alone stimulated ENaC current in oocytes, this could be reversed by co-expressing ENTH and PI5KI␣ with ENaC subunits. These data indicate that the ENTH-PIP 2 interactions can be modulated by PI5KI␣ and that this pool of PIP 2 is important for the effects we observe on ENaC activity.
The ability of PIP 2 to recruit clathrin-accessory proteins appears to have a rate-limiting effect on constitutive endocytosis. Generation of PIP 2 at the plasma membrane is regulated by the type I PI5Ks, all three of which have documented roles in regulating clathrin-mediated endocytosis. The production of PIP 2 via PI5KI␣ increases recruitment of the clathrin adaptor complex, AP-2, to the plasma membrane with a coordinate increase in endocytosis of the transferrin receptor in nonpolarized cells, whereas the ␤ isoform has a similar effect on EGF receptor endocytosis in NR6 cells (32). Additionally, PI5KI␥ interacts directly with AP-2 in vitro and in vivo in HEK293 cells, suggesting a mechanism for selective generation of PIP 2 at sites of endocytosis. The ␥ isoform has also been shown to be responsible for the generation of PIP 2 at synaptic nerve terminals. These studies indicate that individual isoforms can be involved in selective trafficking events in specific cell types. The function and spatial organization of these individual isoforms has not been well described in polarized epithelial cells, which compartmentalize their plasma membrane into apical and basolateral domains. Recent studies have demonstrated a role for PI5KI␣ in regulating the transport of newly synthesized proteins to the apical surface of MDCK cells, while PI5KI␥ functions to regulate the endocytosis of transferrin at the basolateral surface of MDCK cells (33,39). These observations indicate a role for the compartmental organization of specific Type I PI5K isoforms in regulating PIP 2 at distinct membrane domains.
The localization of PI5KI␣ to the apical surface of CCD cells (Fig. 7A) also argues for its role in regulating membrane events at the apical domain of polarized cells. Consistent with this is our observation that IgA endocytosis from the basolateral domain is unaffected by this up-regulation in PIP 2 production via PI5KI␣ in polarized CCD cells (Fig. 8A). However, when PI5KI␣ is expressed in nonpolarized CCD cells, where it localizes uniformly to the plasma membrane, it stimulates IgA internalization, consistent with the described function of this enzyme in other nonpolarized cell models (Fig. 8B). Interestingly, recent studies have shown that PI5KI␥ localizes to basolateral membranes in polarized MDCK cells and regulates trafficking events at this membrane domain (33,51). Taken together, these stud-ies demonstrate that the spatial organization of PI5K isoforms contributes to the functional specificity of PIP 2 production in membrane trafficking. PIP 2 has been shown to influence the activity of ENaC via changes in open probability by direct interactions with the channel such that increases in PIP 2 concentration increase activity or prevent rundown in excised patch experiments (26,28,30). This points to a different type of local regulation of ENaC via PIP 2 much like that described for GIRK channels, which are down-regulated in parallel with local depletions in PIP 2 . Therefore, a local increase in PIP 2 might be expected to enhance ENaC activity, perhaps in an acute manner. While PIP 2 has previously been shown to increase ENaC channel density in the membrane, these studies utilized the murine ␤ isoform of type I PI5K (27,29), therefore, it is likely that the influence of PIP 2 on ENaC can be independently regulated by different isoforms of PI5K. Additionally, the isoforms of PI5KI are segregated in polarized epithelial cells, presumably to allow for independent modulation of PIP 2 production at different surface domains. Moreover, prolonged up-regulation of PIP 2 production could impact multiple pathways that affect channel number or cell surface residence time. Taken together, these data demonstrate that ENaC channel activity can be negatively regulated by the production of PIP 2 via PI5KI␣, likely through its effects on specific endocytic events at the apical membrane of polarized epithelial cells. Thus, PIP 2 appears to have both a positive and negative regulatory role in ENaC function in polarized renal epithelial cells, and this is maintained by spatially modulated metabolism of PIP 2 .