Endogenously expressed epithelial sodium channel is present in lipid rafts in A6 cells.

The epithelial sodium channel (ENaC) present in the kidney collecting duct, distal colon, and the lung is responsible for salt reabsorption and whole body volume regulation. It is composed of three homologous subunits, alpha, beta, and gamma, and mutations to these subunits can lead to the salt wasting disease pseudohypoaldosteronism type I, associated with decreased channel density at the plasma membrane or to the hypertensive disorder, Liddle's syndrome, in which channel residency time at the plasma membrane is enhanced. Regulation of ENaC trafficking and turnover is therefore critical to sodium homeostasis. In this study we examined whether ENaC is present in the cholesterol-enriched microdomains commonly called lipid rafts, in the endogenously expressing A6 cell line. We demonstrate that a fraction of alpha, beta, and gamma ENaC is present in detergent-insoluble membranes, that subunits exist in membranes that float on discontinuous sucrose density gradients, and that methyl-beta-cyclodextrin treatment causes a redistribution of ENaC subunits to higher density membranes. Furthermore, chronic aldosterone stimulation results in a shift in the membrane density of all three subunits. Biotinylation of apical membrane proteins revealed that ENaC is present in lipid rafts on the plasma membrane. In conclusion, these results show that ENaC is present in lipid rafts both intracellularly and on the cell surface. Raft association may be important for trafficking and/or function of the channel.

The epithelial sodium channel (ENaC) 1 functions at the apical membrane of epithelia in the kidney collecting duct, distal colon, and the airways and is responsible for sodium reabsorption and sodium homeostasis in mammals and amphibians (1). It is composed of three homologous subunits, ␣, ␤, and ␥, which form a heteromultimeric protein likely to be comprised of two ␣ subunits and one each of ␤ and ␥. These subunits are characterized by two membrane-spanning domains, short intracellular N and C termini, and a large ectodomain (1). In humans, mutations to this channel are known to be responsible for pseudohypoaldosteronism type I, a salt-wasting disease of infants, and Liddle's syndrome, which results in severe hypertension. In the former, mutations to ␣, ␤, and ␥ result in loss-of-function due to lower channel density at the plasma membrane (2). Conversely, in the latter, deletions or mutations to PY motifs in the C termini of ␤ or ␥ result in a gain-of-function which appears to result from increased residency time at the plasma membrane (3). Since both diseases result from altered trafficking and/or turnover of the channel it is of vital importance to understand the regulation of these processes. Recent studies from our laboratory have suggested that the trafficking and turnover of ENaC subunits may be regulated independently (4). Chronic aldosterone treatment of A6 cells, for example, dramatically increased levels of ␤ at the cell surface concomitant with increases in Na ϩ conductance, while levels of ␣ and ␥ remained unchanged. If channel subunits have distinctly different half-lives at the surface and if selective insertion and retrieval of channel subunits can regulate Na ϩ reabsorption, then biosynthesis and subunit trafficking are critically important to this process.
Membrane microdomains known as lipid rafts are enriched in cholesterol and sphingolipids and have been shown to exist as dynamic platforms important for the delivery of proteins to the apical membrane as well as for sequestering proteins in close physical proximity for functional interactions (5,6). These structures are characterized by their detergent insolubility and high buoyant density. Since mature ENaC has been described as being detergent-insoluble when expressed in COS7 and HEK293 cells (7), we hypothesized that raft localization might represent a cellular mechanism for controlling ENaC subunit density at the plasma membrane and/or ENaC subunit interactions. In addition, a number of proteins known to interact with ENaC have been localized to lipid rafts, most notably the ubiquitin ligase NEDD4 (8), which plays a role in ENaC turnover at the plasma membrane (3). Therefore we have examined whether ENaC may be present in lipid rafts in the endogenously expressing A6 cell line. Our results indicate that all three subunits are present in lipid rafts and that the membrane microenvironment that ENaC exists within is altered by exposure to aldosterone.

EXPERIMENTAL PROCEDURES
Cell Culture-A6 cells were seeded at high density and grown on 75-mm permeable supports (0.4 m pore, Costar, Cambridge, MA or Millipore, Bedford, MA) in amphibian medium (BioWhittaker, Walkersville, MD) containing 10% fetal bovine serum. Cells were maintained at 28°C in 5% CO 2 and were not used for experiments for at least 8 days. Current and resistance measurements across confluent monolayers were performed using an EVOM "chopstick" device (World Precision Instruments, Stevenage, UK). Cells were used for experiments if resistances were at least 800 ⍀ and currents of at least 1.5 A/cm 2 were recorded.
Preparation of Triton X-100-soluble and -insoluble Membranes-Confluent cells with high resistances and I sc readings were scraped from filters into TNE buffer (in mM: Tris-HCl, 25; NaCl, 150; EDTA, 5; pH 7.4) containing protease inhibitors (PIs; Complete Mini, Roche Mo-lecular Biochemicals), disrupted by suction through a 22-gauge syringe and a post-nuclear supernatant (PNS) recovered after 1500 ϫ g centrifugation for 5 min. Ice-cold Triton X-100 (final 1%) was added to the PNS for a 30 min. incubation on ice. Insoluble and soluble membranes were recovered following centrifugation (100,000 ϫ g, 60 min), and insoluble membranes in the pellet were then dissolved in Triton X-100 and incubated at 28°C for 30 min. Aliquots of soluble and insoluble membranes were added to sample buffer and boiled for 2 min prior to SDS-PAGE and Western blotting. Western blots were probed with ENaC subunit-specific antibodies as described in Refs. 9 and 10.
Isolation of Membranes by Sucrose Gradient Centrifugation-Membranes of different buoyant density were prepared essentially as described in Ref. 11. Briefly, cells were scraped into 2 ml of 500 mM sodium carbonate, pH 11, after washing three times with ice-cold PBS (all buffers and sucrose solutions contained PIs). Cell lysates and a PNS fraction were then prepared as described above. PNS membranes were left untreated or were treated with 10 mM methyl-␤-cyclodextrin for 60 min at 28°C. Samples (2 ml) were mixed with 2 ml 90% sucrose in 25 mM MES, 150 mM NaCl, pH 6.5, and placed in a centrifuge tube. The sample was then overlaid with 4 ml of 35% and 4 ml of 5% sucrose before centrifugation at 190,000 ϫ g (av) for 18 h at 4°C. Fractions (1 ml) were recovered, and protein was extracted with chloroform/methanol according to Ref. 12. Total proteins from each fraction were separated by SDS-PAGE, transferred to nitrocellulose, and probed with ENaC subunit-specific antibodies as described (4). Xenopus caveolin was detected using an N-terminal specific antibody, which was raised against human caveolin-1 and cross-reacts with mouse and rat caveolin-1 (Santa Cruz Biotechnology, Santa Cruz, CA). To confirm that the band being detected was indeed caveolin, we immunoblotted A6 cell membranes that had been separated on sucrose density gradients with another commercially available antibody with reactivity to caveolin isoforms 1, 2, and 3 (BD PharMingen, Lexington, KY). This antibody recognized a band at 22 kDa, which floated in fractions 4 and 5 of the gradient and which ran with the same relative mobility as a positive caveolin control provided by the company (data not shown), thus confirming the specificity of the Santa Cruz antibody. Flotillin 2 antibody was obtained from Santa Cruz Biotechnology.
Aldosterone Treatment-Confluent transporting monolayers of A6 cells were treated with 1 M aldosterone for 18 h at 28°C. Sucrose gradient centrifugation was then performed as described above.
Surface Biotinylation-Cells were washed three times with cold PBS and a total of 2.5 mg sulfo-NHS-SS-biotin (Pierce) in a volume of 5 ml of ice-cold borate buffer (in mM: NaCl, 85; KCl, 4; Na 2 B 4 O 7 , 15; pH 9.0) was added to the apical side of polarized A6 cells grown on 75-mm filters. Ice-cold amphibian medium was placed basally and plates were rocked gently on ice for 20 min. Biotin solution was removed, and cold amphibian medium was added apically to quench remaining biotin reagent for 5 min. Cells were washed twice with cold PBS, scraped, a PNS prepared, and then membranes were subjected to discontinuous sucrose density centrifugation as described above. Fractions collected from the gradient were incubated overnight at 4°C with streptavidinlinked agarose beads. Beads were recovered by centrifugation, washed three times with Nonidet P-40 buffer (1% Nonidet P40, 0.4% deoxycholate, 50 mM EGTA, 10 mM Tris, pH 7.4), and then sample buffer was added prior to SDS-PAGE and Western blotting.
Densitometric Quantitation of Immunoblots-Band densities were quantitated using Quantity One software from Bio-Rad.

RESULTS
When A6 cells were extracted with cold Triton X-100 a fraction of all three ENaC subunits appeared in the insoluble pellet (Fig. 1A), suggesting that ENaC might be present in lipid rafts.
To determine whether the Triton-insoluble subunits were raft localized we utilized a non-detergent, discontinuous sucrose gradient method for separating membranes of differing buoyant density. When PNS membranes from A6 cells were subjected to discontinuous sucrose density centrifugation (Fig. 1B), it could be shown that a proportion of each subunit was recovered in low density fractions 4 and 5 (␥-ENaC is shown) and that flotillin and caveolin, markers of the specific lipid raft subset known as caveolae, also colocalized to these fractions (13). This portion of the gradient corresponds to the interface between 5 and 35% sucrose. Densitometry confirmed that the distribution of ␥-ENaC and caveolae between different membrane compartments was similar (Fig. 1C). This finding dem-onstrated that ENaC subunits are found in the kinds of low density membranes that fit one of the functional definitions of cholesterol-enriched lipid rafts. To further verify that these membranes represented lipid rafts, PNS membranes were treated with 10 mM methyl-␤-cyclodextrin (CD) for 1 h at room temperature. CD is a cholesterol-sequestering drug that disrupts the hydrogen bonding interactions between cholesterol and sphingolipids which promote lipid raft formation. Fig. 2 demonstrates that CD treatment results in a significant shift of ENaC subunits to higher density membranes consistent with the loss of cholesterol and disruption of lipid raft structures. Densitometry confirmed that there was a dramatic decrease in low density ENaC and a concomitant increase in channel found at higher density (Fig. 2B). When membranes treated with CD were immunoblotted for caveolin a similar redistribution was observed (bottom two panels of Fig. 2A). Densitometry revealed that the amount of caveolin remaining in raft-like membranes after CD was dramatically reduced.
We next tested whether aldosterone would influence the membrane distribution of ENaC subunits. After 18-h exposure to aldosterone (1 M), it could be seen that there was a highly reproducible shift in the buoyant density of all three subunits (Fig. 3). ␤and ␥-ENaC were predominantly located in fraction 4 under control conditions but shifted almost entirely to fraction 5 following aldosterone exposure. For ␣-ENaC this shift was less dramatic. For all three subunits there was the appearance of a significant proportion of the total subunit pool in higher density fractions (8 -12). Interestingly, caveolin was also observed to shift from fraction 4 to fraction 5 upon aldos-

ENaC Localization in Lipid Rafts 33542
terone treatment, with the appearance of greater amounts in high density membranes.
Because a large proportion of the ENaC pool is known to reside intracellularly, we wished to examine whether ENaC subunits exist within rafts at the cell surface. Biotinylation of apical cell surface proteins and subsequent isolation using streptavidin beads revealed that ENaC was present in lipid rafts (fractions 4 and 5) in the apical plasma membrane (Fig. 4,  PM). Whole cell (WC) lysates are shown for comparison and demonstrate an approximately equivalent distribution between low and high density membranes as can be seen at the cell surface. DISCUSSION Glycosylphosphatidylinositol-anchored and acylated proteins have been found associated with lipid rafts, and it has been hypothesized that these cholesterol-rich microdomains offer a molecular scaffolding for the concentration of proteins involved in ligand binding and signal transduction as well as in protein sorting at the level of the trans-Golgi network. Proteins that preferentially localize to rafts often have long saturated acyl chains and therefore exhibit a high affinity for liquidordered domains (5,14); however, hydrophobicity alone does not seem to be sufficient for raft localization, as prenylated proteins are excluded (5,14). As a consequence of these observations, ion channels with multiple membrane-spanning domains were assumed to have little affinity for liquid-ordered membrane domains (5). However, in the last 2 years there have been a limited but growing number of reports demonstrating ion channel association with rafts. These channels include the voltage and Ca 2ϩ -activated K ϩ channel ␣ subunit (hSlo) (15); the voltage-gated K ϩ channel Kv2.1 (but not Kv4.2), which is found in non-caveolar lipid rafts (16); the voltage-gated K ϩ channel Kv1.5, which is localized to caveolae (17); the endothelial volume-regulated anion channel (18); and the voltage-gated sodium channel of cardiac myocytes (19), which also localizes to caveolae. The authors in the latter study postulated that observed increases in sodium current in response to ␤-adrenergic stimulation were occurring as the result of movement of sodium channel out of caveolae and into the sarcolemma. It now appears highly likely that specific ion channels may traffic in, or to, distinct liquid-ordered membrane domains and that their activity might be regulated by these microdomains at the cell surface.
Experiments aimed at defining the membrane microdomain association of ENaC as well as its subcellular distribution are relatively difficult to perform in endogenously expressing cells due to the low abundance of channel protein. As a consequence, a number of groups have used heterologous expression systems that are able to generate higher amounts of protein, as a means to address this question. Prince and Welsh (7) have shown that human ENaC subunits when expressed individually or concurrently in COS-7 or HEK293 cells acquire detergent insolubility at the post-endoplasmic reticulum stage of biosynthesis and that virtually all of the ENaC expressed at the cell surface is detergent-insoluble. These authors speculate that detergent insolubility may be the result of associations with caveolar or cytoskeletal proteins or alternatively be due to self-oligomerization. They did not characterize the buoyant density of their ENaC-containing membranes or attempt to alter the lipid composition of these membranes; therefore a number of interpre-FIG. 3. All three ENaC subunits shift to higher density membrane fractions upon chronic exposure to aldosterone. Polarized A6 cells grown on filters were either untreated (Ϫ) or exposed to 1 M aldosterone for 18 h (ϩ). Following this they were scraped, and a PNS was prepared as described under "Experimental Procedures." PNSs were then subjected to discontinuous sucrose density centrifugation as described. Fractions (1 ml) were collected with fraction 1 being the top of the gradient. Proteins were chloroform/methanol-precipitated, run on SDS-PAGE, and immunoblotted with antibodies to ENaC subunits or to caveolin (Cav). Data are representative of three separate experiments.
FIG. 2. ENaC Subunits shift from low density membranes to high density membranes upon treatment with cyclodextrin. A, polarized A6 cells grown on filters were scraped, and a PNS was prepared as described under "Experimental Procedures." PNSs were left untreated (Ϫ) or were treated with 10 mM cyclodextrin (ϩ) for 1 h at room temperature and then subjected to discontinuous sucrose density centrifugation as described. Fractions (1 ml) were collected with fraction 1 being the top of the gradient. Proteins were chloroform/methanolprecipitated, run on SDS-PAGE, and immunoblotted with antibodies to x-ENaC subunits or to caveolin. B, densitometric quantitation of the immunoblot analysis shown to the left in A. Data are representative of three separate experiments.

FIG. 4. ENaC is present in lipid rafts at the apical cell surface.
Apical membrane proteins of polarized transporting A6 cells were biotinylated with sulfo-NHS-SS-biotin at 4°C and then a PNS prepared and run on discontinuous sucrose density gradients as described under "Experimental Procedures." Fractions removed from the gradient were incubated with streptavidin beads and then the captured biotinylated membranes were pulled down by centrifugation, washed, and immunoblotted as described. PM, apical plasma membrane; WC, whole cell membranes.

ENaC Localization in Lipid Rafts 33543
tations, including lipid raft localization, are possible. In contrast, Hanwell et al. (20) expressed rat ENaC in MDCK cells and showed that ENaC was Triton-soluble and did not float on Optiprep density gradients, thus demonstrating that ENaC in this system was not associated with lipid rafts. To investigate further we turned to A6 cells, which endogenously express Xenopus ENaC and which show appropriate hormone responsiveness. Our results clearly demonstrate that a fraction of both intracellular and plasma membrane ENaC is associated with lipid rafts by four different biochemical criteria: 1) detergent insolubility, 2) high buoyant density after extraction in 0.5 M Na 2 CO 3 , pH 11.0, 3) colocalization with caveolin and flotillin, and 4) a dependence on cholesterol for partitioning into high buoyant density membranes. The reasons for differences in our findings from those of Hanwell et al. (20) are not clear, but may reflect the use of a heterologous expression system, overexpression, or the use of detergent versus non-detergent extraction methods for isolating cellular membranes.
Our results demonstrate that a proportion of all three ENaC subunits in the Na ϩ -transporting A6 epithelium are lipid raft localized and that this compartmentalization persists at the plasma membrane. The physiological relevance of such compartmentalization is unclear, but the observation that ENaC subunits can be shifted to membranes of higher density upon aldosterone treatment suggests strongly that their membrane domain localization is regulated by appropriate physiological stimuli. Indeed, aldosterone is known to alter membrane lipid composition through effects on phospholipid turnover and composition (21)(22)(23)(24)(25) as well as through phospholipid methylation (26).
Lipid rafts may represent a way for the cell to get ENaC to the correct cellular destination i.e. the apical membrane, at which point subunits or complete channels having been delivered appropriately may redistribute out of rafts and enter the bulk lipid phase. Alternatively, ion channel function may be regulated in part by the biophysical properties of the membrane. Rafts may even represent a means for the cell to regulate the half-life of subunits present at the cell surface. Only a fraction of channel subunits that are synthesized appear to reach plasma membrane in epithelial cells (4,20,27). The half-life of channel subunits that reach the plasma membrane is a subject of some dispute. Using an approach measuring recovery of apically biotinylated subunits two groups have estimated a half-life of hours for individual subunits in A6 cells (4,10). Using a similar approach in MDCK cells, combined with inhibition of new channel synthesis with cycloheximide, Hanwell et al. (20) described a half-life closer to 1 h (20), and Alvarez De La Rosa et al. (27) have recently estimated halflives of 12-17 min in A6 cells. Using an electrophysiologic approach, Fisher and colleagues (28) described a decline in apical channel density over a period of hours following inhibition of apical delivery by brefeldin A in A6 cells. Whatever the half-life of apical channels may be, sequestration of subunits in rafts could represent a means for the cell to maintain subunit or channel pools in inactive conformations or to target them for internalization. Many, if not most, of the proteins implicated in clathrin-independent endocytosis have been found in lipid rafts (29). Indeed, Nedd4, the ubiquitin ligase that regulates internalization of ENaC through binding to PY motifs in the C termini of the ␤ and ␥ subunits, is a raft-localized protein (30). Since a physical interaction between ENaC and Nedd4 is necessary for ubiquitination, it is tempting to speculate that raft localization of ENaC is linked to the process of ubiquitin-regulated internalization. This possibility is rendered somewhat less likely by the observation that the C2 domain of Nedd4, which apparently mediates the raft association, and apical localization of Nedd4 -1 in MDCK cells (8,30) is not required for the action of Nedd4 -2, a predominantly kidney-expressed isoform, in Xenopus oocytes (31). Whether lipid association of Nedd4 isoforms is essential to Nedd4 apical localization and function in epithelia remains to be determined. Finally, since lipid raft-localized proteins have been shown to shuttle repeatedly and rapidly between the plasma membrane and the Golgi (32), it also potentially represents a way for the cell to exert precise regulatory control over cell surface expression of ENaC and is consistent with the fact that the majority of ENaC exists intracellularly.
In conclusion, we have demonstrated that endogenously expressed ENaC subunits exist within cholesterol-enriched membrane microdomains. Further investigation will be required to define how lipid rafts influence ENaC trafficking, function, and turnover.