Phosphatidylinositol 4,5-Bisphosphate (PIP2) Stimulates Epithelial Sodium Channel Activity in A6 Cells*

Phosphatidylinositol 4,5-bisphosphate (PIP2) is a membrane lipid found in all eukaryotic cells, which regulates many important cellular processes, including ion channel activity. In this study, we used inside-out patch clamp technique, immunoprecipitation, and Western blot analysis to investigate the effect of PIP2 on epithelial sodium channel activity in A6 cells. A6 cells were cultured in media supplemented with 1.5 μm aldosterone. Single sodium channel activity in excised, inside-out patches was increased by perfusion of the bath solution with 30 μm PIP2 plus 100 μm GTP (NP o = 1.34 ± 0.14) compared with the paired control (NP o = 0.09 ± 0.02). However, neither 30 μmPIP2 (NP o = 0.11 ± 0.02) nor 100 μm GTP (NP o = 0.10 ± 0.02) alone stimulated the sodium channels. The PIP2-stimulated channel activity was abolished by application of 10 nm G protein βγ subunits (NP o = 0.14 ± 0.05). However, 10 nm Gαi-3 + 30 μmPIP2 increased both NP o andP o. The stimulating effect of 10 nmGαi-3 + 30 μm PIP2 is similar to that of 30 μm PIP2 plus 100 μm GTP. Immunoprecipitation and Western blot analysis show that both Giα-3 and PIP2 bind β and γ epithelial Na+ channels (ENaC), but not α ENaC. These results indicate that PIP2 increases ENaC activity by direct interaction with β or γ xENaC in the presence of Gαi-3.

Amiloride-sensitive epithelial Na ϩ channels in the distal nephron play a critical role in regulation of sodium transport across renal epithelial cells and thus play a central role in maintenance of salt balance and normal blood pressure. A6 cells, derived from the distal nephron of Xenopus kidney, are a convenient system for in vitro study of ENaC 1 regulation. A6 cells express ENaC with high Na ϩ selectivity (P Na /P K Ͼ 20), high amiloride sensitivity (K i ϭ 0.1), low conductance (4 -5pS), and long open and closed times of the order of seconds (1). ENaC in A6 cells have all the hormonal regulation and pharmacological characteristics expected for ENaC in mammalian cortical collecting duct principal cells (2). The regulation of A6 cell ENaC has been extensively studied over the past 15 years.
However, the direct regulation of ENaC activity by membrane lipid has, to our knowledge, never before been demonstrated.
Phosphatidylinositol 4,5-bisphosphate (PIP 2 ) is a trace, but ubiquitous, component of eukaryotic membrane phospholipid. In the traditional pathway of phosphoinositide metabolism, phosphatidylinositol is phosphorylated to produce phosphatidylinositol 4,5-bisphosphate by phosphatidylinositol 4-kinase and phosphatidylinositol 5-kinase. PIP 2 can be further phosphorylated to produce phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ) by phosphatidylinositol (PI) 3-kinase (3). Or alternatively, PIP 2 can be hydrolyzed by phospholipase C to produce 1,4,5-inositol trisphosphate (IP 3 ) and diacylglycerol (DAG). For a long time, it was assumed that the sole purpose of PIP 2 was to act as precursor of the two intracellular second messengers, IP 3 and DAG, in response to phospholipase C activation. IP 3 and DAG are known to mobilize calcium from the endoplasmic reticulum and to activate protein kinase C, respectively. However, recent evidence has suggested that this lipid has additional functions including regulation of ion channels. For example, PIP 2 can modify inwardly rectified K ϩ channels, G protein-gated inwardly rectified K ϩ channels (4), and ATPsensitive K ϩ channels (K ATP ) (5-7). Besides various types of K ϩ channels, a Na ϩ -gated nonselective cation channel in lobster olfactory receptor neurons is also activated by PIP 2 (8). Since a growing number of ion channels appear to be regulated by PIP 2 , it is possible that PIP 2 is generally important in maintaining ion channel activity (8).
Recent studies using short circuit measurements demonstrated that phosphatidylinositol 3-kinase (PI 3-kinase) is present in A6 cells and that this enzyme is required for regulation of ENaC by insulin (6), aldosterone, and vasopressin (9 -12). The products of PI 3-kinase include phosphatidylinositol 3-phosphate, phosphatidylinositol 3,4-bisphosphate, and phosphatidylinositol 3,4,5-triphosphate (13). These studies indicate that phosphoinositides are involved in the regulation of ENaC activity, and these membrane lipids may be common effector molecules for hormonal regulation. Nevertheless, the short circuit current measurements in these studies could not answer the following questions. 1) Do phosphoinositides regulate ENaC or other membrane transporters? 2) If phosphoinositides regulate ENaC, do they directly interact with ENaC or interact through some other closely associated intermediate signaling molecule? 3) Which of the phosphoinositides alter ENaC activity? 4) How do the lipids modify single channel activity?
In the present work, we examined the PIP 2 regulation of single ENaC activity in A6 cells with the patch clamp technique and identified PIP 2 binding to ENaC subunits by coimmunoprecipitation and Western blot analysis.

MATERIALS AND METHODS
A6 Cells Preparation-A6 cells were purchased from American type Culture collection (Manassas, VA) in the 68th passage. All experiments were performed on passages 71-80 with no discernable variation among different passages. The cells were maintained in plastic tissue flasks (Corning, Corning, NY) at 27°C in a humidified incubator with 4% CO 2 in air. The culture medium was a mixture of Coon's medium F-12 (3 parts) and Leibovitz's medium L-15 (7 parts) supplemented with 10% fetal bovine serum (Invitrogen) and 1.5 M aldosterone for amphibian cells with 103 mM NaCl, 25 mM NaHCO 3 , pH 7.4. For patch clamp experiments, the cells were subcultured on permeable, collagen-coated aluminosilicate supports (Nunc Corp.) attached to the bottoms of plastic rings.
Patch Recording-A plastic ring containing an A6 cell monolayer was mounted in a recording chamber on an inverted microscope (Nikon, Tokyo, Japan). Both the apical side and basolateral side of the monolayer were bathed in amphibian saline solution containing (in mM): 96 NaCl, 3.4 KCl, 0.8 CaCl 2 , 0.8 MgCl 2 , and 10 HEPES at pH 7.4 (titrated with 1 N NaOH). A perfusion setup was used to exchange the apical bath solution to expose A6 cells to PIP 2 (Roche Molecular Biochemicals), GTP (Sigma), G protein ␣ subunit (G i␣3 ) (Calbiochem), or G protein ␤␥ subunit (a generous gift from Dr. David Clapham, Harvard University). For these experiments, we used inside-out patch clamp methods following standard procedure. Patch pipettes were fabricated from TW 150 glass (World Precision, New Haven, CT) and fire-polished to produce tip resistance of 5-10 M⍀ when filled with amphibian saline solution. Inside-out patches were obtained from cell-attached patches following exchange of the apical bath solution from amphibian saline to high potassium solution containing (in mM): 3 NaCl, 85.4 KCl, 4 CaCl 2 , 1 MgCl 2 , 10 HEPES, and 5 EGTA at pH 7.4 (titrated with 1 N KOH). Experiments were performed at room temperature (22-23°C). Single channel events from inside-out patches were measured with an Axopatch 200 amplifier, low pass filtered at 5 KHz, recorded on a digital video recorder (Sony, Tokyo, Japan), and then digitized at 500 Hz using a scientific Solutions A/D converter and pentium computer equipped with Axotape software (Axon Inc., Foster City, CA). The convention for applied voltage to the apical membrane patch (V pipette ) represents the voltage deflection from the patch potential (i.e. the resting membrane potential for cell-attached patches and 0 mV for inside-out excised patches). Inward Na ϩ current (pipette to cell) is represented as downward transitions in single channel records.
Data Analysis-The channel events were analyzed using pClamp 6.02 (Axon Inc.). This program is designed to determine valid channel openings and closings based on 50% threshold crossing method. When multiple channel events are observed in each patch, the total number of functional channels (N) in the patch was determined by observing the number of peaks detected on all point amplitude histograms. NP o , the product of the number of channels and the open probability, or the open probability (P o ), itself was used to measure the channel activity within a patch. The NP o was calculated by the relative area under all points amplitude histograms and expressed as follows.
In the above equation, A is the area under Gaussian curve, N is the total number of functional channels in a patch, i is the number of channels, P o is the open probability of an individual channel in a patch. The assumptions made are that the channels within a patch are identical and open and close independently.
Immunoprecipitation and Western Blot-Lysates of A6 cells are incubated with PIP 2 (Roche Molecular Biochemicals) for 30 min in the presence of G␣ i-3 and then immunoprecipitated using an anti-PIP 2 antibody (Assay Designs) coupled to protein A/G-Sepharose beads. The immunoprecipitates are resolved by 7.5% SDS-PAGE, transferred to Immobilon paper, and immunoblotted with anti-␣, anti-␤ ENaC, anti-␥ ENaC, or anti-Gi␣ i-3 (Calbiochem) antibodies, respectively. The xENaC subunit specific antibodies were developed by our research center and have been extensively characterized in previous reports (14 -20). Antigen-antibody complexes are determined using the Amersham ECL chemiluminescence detection system. The rationale for this experimental design is that all of the PIP 2 binding proteins, including ENaC, are immunoprecipitated by anti-PIP 2 antibody. If ENaC and G␣ i-3 co-immunoprecipitate with PIP 2 , they could be detected with anti-ENaC or anti-G␣ i-3 antibody, respectively. To examine the expression of PI 5-kinase in A6 cells, the A6 cell lysates were resolved in SDS-PAGE and detected with antibody against ␣ subunit of PI 5-kinase.
Statistics-After analysis using pClamp, the data are presented as mean Ϯ S.D. Paired t tests or repeated measures ANOVA were used as appropriate to compare experimental groups. Results are considered significant when p Ͻ 0.05. Fig. 1 (upper panel) illustrates representative single channel records from a cell-free, inside-out patch. The control group has a low level of activity which resulted from the general loss of activity after the switch from cell-attached patch mode to inside-out, excised patch mode. Neither PIP 2 nor GTP alone altered the channel activity, but the combination of these two agents dramatically increased the number of active channels and open probability. Fig. 1 (lower panel) is a summary of single channel activity expressed as NP o (left) and P o (right). This figure shows that PIP 2 or GTP alone did not change NP o or P o , while PIP 2 ϩ GTP increased both the apparent number of active channels (N) and open probability (P o ). The requirement for GTP suggests that a G protein is involved in the PIP 2 -mediated regulation of Na ϩ channel activity. This idea is supported by the observation that stimulation of ENaC activity by PIP 2 ϩ GTP was abolished by addition of G protein ␤␥ subunit protein in the bath solution (cytoplasmic side of membrane) ( Table I). The G protein, which has been previously described as associated with the apical membranes of A6 cells, is G i-3 (21). Application of the activated ␣ subunit of G i-3 protein (G␣ i-3 ) by itself to the cytosolic surface of excised patches does not increase single Na ϩ channel activity expressed as NP o or P o ; however, application of G␣ i-3 ϩ PIP 2 strongly activated Na ϩ channels (Table II). The stimulatiing effect of G␣ i-3 ϩ PIP 2 is comparable with that of PIP 2 ϩ GTP (as shown in Fig. 1). The results in Tables I and II imply that the presence of G protein ␣ subunit (G␣ i-3 ), but not ␤␥ subunit, is required for the effect of PIP 2 on Na ϩ channel activity. On the other hand, the presence of ␤␥ subunit may prevent the effect of PIP 2 by binding the endogenous ␣ subunit in the excised patches. Unlike PIP 2 , PIP 3 ϩ GTP did not change Na ϩ channel activity (data not shown), suggesting that PIP 2 -induced Na ϩ channel activity is not mediated by one of its metabolic products, PIP 3 . In addition, application of the PIP 2 isomer, phosphatidylinositol 3,4 bisphosphate, also had no effect on channel activity (data not shown).

RESULTS
Three homologous subunits of ENaC have been isolated from A6 cells (␣ xENaC, ␤ xENaC, and ␥ xENaC). We examined the PIP 2 binding subunits using immunoprecipitation and Western blot analysis. Fig. 2 shows that anti-PIP 2 antibody was unable to precipitate ␣ENaC, but was able to precipitate ␤ and ␥ subunits. This result suggested that ␤ xENaC, and ␥ xENaC, but not ␣ xENaC, are the PIP 2 binding ENaC subunits in A6 cells. Since PIP 2 is only about 1100 daltons and does not stay associated with ENaC under the high detergent conditions of the gels, we could not perform the reverse experiment in which we immunoprecipitated with anti-ENaC antibodies and detected PIP 2 .
The Western blots in Fig. 2 suggest that ␤ xENaC and ␥ xENaC, but not ␣ xENaC, are responsible for PIP 2 binding. Yu et al. (22) compared the putative PIP 2 binding sites in gelsolin with that of other PIP 2 binding proteins. They found two consensus sequences associated with PIP 2 binding in gelsolin. Each of the two contains a stretch of basic amino acid sequence. One of the consensus sequences, KXXXKXKK, is also found in the N-terminal domain of ␥ xENaC and is highly conserved among ␥ rENaC, ␥ hENaC, and ␥ mENaC (Fig. 3, lower panel). The other consensus sequence is K(R)XXXXKXKK(RR). A similar sequence, RXXXXXXRR, is present in all the N-terminal regions of ␤ xENaC, ␤ rENaC, ␤ hENaC, and ␤ mENaC (Fig. 3,  upper panel). Interestingly, a similar consensus sequence, KXXXXKKR, was identified in the PIP 2 binding domain of several types of inwardly rectified K ϩ channels (4). Since both ␤ and ␥ subunits contain potential PIP 2 binding sequences, while ␣ ENaC does not, it is not surprising that PIP 2 precipitated ␤ xENaC and ␥ xENaC. Therefore, PIP 2 may bind the N-terminal domains of ␤ xENaC and ␥ xENaC and, thereby, modulate the Na ϩ channel activity in A6 cells.
We also wondered whether G␣ i-3 would co-immunoprecipitate with anti-PIP 2 , since there are indications in the literature that G␣ i-3 is associated with ENaC subunits (23). For this experiment we used a commercially available antibody (Calbiochem), which is purportedly specific for G␣ i-3 . Fig. 3 shows that the antibody is not as specific as advertised; however, to determine specificity, we used the original antigenic decameric peptide (KNNLKECGLY) to determine which bands were effectively competed. In fact, there is a band at the correct molecular weight for G␣ i-3 , suggesting that the G protein subunit also co-immunoprecipitated with PIP 2 . There was, however, another higher molecular weight band that was also competed by the antigenic peptide. A BLAST search for the antigenic peptide revealed, as expected, mostly G␣ i-3 ; however, one protein, Nedd4 (neural precursor cell-expressed developmentally downregulated-protein) contained an epitope with seven amino acids identical to the antigenic sequence. The size of the higher molecular weight protein is consistent with Nedd4; and in other work, we and others have demonstrated that Nedd4 is present in A6 cells and that it does co-immunoprecipitate with ␤ or ␥ ENaC (19,24). Therefore, we suspect that the higher molecular weight band is Nedd4.
If PIP 2 were actually a physiological regulator of ENaC, then the enzyme responsible for the production of PIP 2 , PIP 5-kinase, would necessarily have to be present in A6 cells. Fig. 5 demonstrates that PIP 5-kinase is, indeed, present in A6 cells, and therefore, regulation of PIP 5-kinase could alter ENaC activity.

DISCUSSION
Our single channel recordings using inside-out patch clamp methods demonstrated that PIP 2 , a trace membrane lipid, could dramatically alter the activity of ENaC in A6 cells. PIP 2 activated ENaC by increasing the single channel open probability and the apparent number of active channels. However, since ENaC open probability is extremely low under untreated conditions, we might have been unable to observe two or more channels opening simultaneously, a necessary criterion to determine the number of channels within a patch. That we failed to observe simultaneous openings does not mean that there was only one active channel under untreated conditions, since when open probability is very low, different open events may be due to different channel proteins with a very low probability of simultaneous openings. Therefore, it is hard to determine the true number of active channels under untreated conditions. Therefore, we cannot be sure PIP 2 increased the true number of active channels, although it certainly increased the single channel open probability.
The metabolic products of 4,5-PIP 2 include IP 3 , DAG, and PIP 3 . IP 3 stimulates calcium release from endoplasmic reticu-TABLE I G protein ␤␥ subunit protein inhibits PIP 2 /GTP-induced channel activity Patches were excised from A6 cells (see "Materials and Methods") and the initial activity measured for 10 min after which 30 M PIP 2 ϩ 100 M GTP was perfused onto the cytosolic surface of the patch. As expected this treatment substantially increased channel activity. After recording for 10 additional minutes, the cytosolic solution was replaced with one containing 30 M PIP 2 ϩ 100 M GTP ϩ 10 nM ␤␥ G protein subunit. This treatment reduced activity close to the level of untreated patches. The mean values (ϮS.D.) for five experiments are given below.  plus G protein ␣ subunit protein stimulates channel activity Patches were excised from A6 cells (see "Materials and Methods") and the initial activity measured for 10 min after which 10 nM GTP␥Sactivated G␣ iϪ3 protein was perfused onto the cytosolic surface of the patch. This treatment produced little change in channel activity. After recording for 10 additional minutes, the cytosolic solution was replaced with one containing 30 M PIP 2 ϩ 10 nM G␣ iϪ3 . This treatment increased activity as much as application of PIP 2 ϩ GTP. The mean values (ϮS.D.) for five experiments are given below. lum, but it is apparently not responsible for PIP 2 stimulation of ENaC activity, since endoplasmic reticulum is not present in inside-out patches. DAG also cannot be responsible for the activation, since DAG activates protein kinase C that inhibits ENaC (25,26). 3,4,5-PIP 3 is also not responsible, since direct addition of PIP 3 did not alter ENaC activity in A6 cells. Moreover, phosphatidylinositol 3,4-bisphosphate, a membrane lipid structurally similar to 4,5-PIP 2 , did not activate ENaC activity (data not shown). These results suggest that only 4,5-PIP 2 and not its metabolic products, stimulate ENaC activity. This conclusion is different from the reports that activation of PI 3-kinase and production of PIP 3 can stimulate Na ϩ transport in Ussing chambers (9,11). There are three possible explanations for this apparent discrepancy. First, PIP 3 stimulates membrane transporters other than Na ϩ channels, since short circuit current could reflect not only Na ϩ transport but also K ϩ or Cl Ϫ transport. Second, PIP 3 might stimulate Na ϩ channel activity through an intermediate intracellular signaling molecule, which would be missing in inside-out patch clamp experiments. Third, PIP 3 is increased, but is broken down by a lipid 3-phosphatase to 4,5-PIP 2 at the apical membrane where its increase is difficult to measure against a large cellular background of other inositol lipids.
Western blot analysis suggests that ␤ xENaC and ␥ xENaC, but not ␣ xENaC, are responsible for PIP 2 binding (Fig. 2). In fact, both ␤ and ␥ subunits contain consensus PIP 2 binding sequences that are conserved across subunits from several species (Fig. 3). Therefore, PIP 2 may bind the N-terminal domains of ␤ xENaC and ␥ xENaC and thereby modulate the Na ϩ channel activity in A6 cells. Further investigation with mutagensis and PIP 2 binding assay will identify the PIP 2 sites in xENaC subunits. Since the crystal structure of ENaC is still unknown, the molecular interaction between PIP 2 and xENaC and subsequent regulation of ENaC activity is not clear, but allosteric regulation is always a possible explanation.
Despite the questionable quality of the commercial antibody, Fig. 4 shows that, besides ␤ and ␥ ENaC, anti-PIP 2 antibody can also co-immunoprecipitate G␣ i-3 . However, it is unlikely that PIP 2 directly binds G␣ i-3 , because unlike ␤ and ␥ ENaC, G␣ i-3 has no obvious PIP 2 binding sites. A reasonable explanation for the result in Fig. 4 is that both PIP 2 and G␣ i-3 bind ␤ or ␥ ENaC (and are consequently all immunoprecipitated together). This is consistent with the previous finding that G␣ i-3 is associated with the ENaC channel protein complex (23,27) and with the observation that PIP 2 directly binds the K ϩ channel protein in a G protein-coupled inward rectifier K ϩ channel (4).

FIG. 2. Binding of PIP 2 to ENaC subunits.
Cell lysates were incubated with PIP 2 and G␣ i-3 . The lysates were immunoprecipitated with anti-PIP 2 antibody (except for the upper right panel, which was immunoprecipitated with ␣ ENaC antibody) and the immunoprecipitates resolved on gels and blots detected with subunit-specific ENaC antibodies. The left upper panel shows that ␣ ENaC is not co-immunoprecipitated with anti-PIP 2 , the antibody can easily detect ␣ ENaC when the precipitating antibody is anti-␣ ENaC (upper right panel). The left lower panel and right lower panel show that anti-PIP 2 can immunoprecipitate ␤ and ␥ EnaC. In each case the specificity of the antibody is confirmed by competition with the original antigenic peptide, which blocks all antibody binding. Since our patch clamp studies indicate that the presence of G␣ i-3 is indispensable for PIP 2 activation of ENaC activity, PIP 2 and G␣ i-3 must cooperate to interact and regulate ENaC. Recent reports regarding K ϩ channels show that PIP 2 is responsible for the gating of the channels (4, 28), while ␤␥ subunits of G protein activate G protein-gated inwardly rectified K ϩ channels by stabilizing the interaction of PIP 2 and C-terminal region (4). We will examine the roles of PIP 2 and G␣ i-3 in regulation of ENaC activity.
PIP 5-kinase is a key enzyme for production of PIP 2 . The presence of PIP 5-kinase in A6 cells (Fig. 5) suggests that PIP 2 is produced in A6 cells. While the present excised patch clamp study has shown that the exogenously applied PIP 2 activated ENaC activity, further investigation will clarify whether PIP 5-kinase is responsible for the endogenous PIP 2 production and subsequent ENaC activation in intact A6 cells.
In summary, the present patch clamp study suggested that PIP 2 increases Na ϩ channel activity in A6 cells by direct interaction with ENaC protein in the presence of G protein. The PIP 2 -induced Na ϩ channel activity is not mediated by one of its metabolic products, PIP 3 , or other intracellular lipid metabolites. PIP 2 regulated ENaC activity possibly by interaction between PIP 2 and N-terminal domains of ␤ xENaC and ␥ xENaC. FIG. 4. Anti-PIP 2 co-immunoprecipitates G␣ i-3 . A commercially available antibody (Calbiochem) was used to detect G␣ i-3 in the anti-PIP 2 immunoprecipitate. Since the antibody stained several bands, we used the antigenic peptide to compete for specific binding. Two bands were reduced after treatment with peptide. The band slightly above 40 KDa is the correct molecular mass for G␣ i-3 . The higher molecular mass band is likely Nedd4, since Nedd4 has an amino acid sequence that has 7 of 10 amino acids in common with the antigenic peptide (and which, other than G proteins, is the only protein with such identity to the antigenic peptide).