The Polarized Effect of Intracellular Calcium on the Renal Epithelial Sodium Channel Occurs as a Result of Subcellular Calcium Signaling Domains Maintained by Mitochondria

,

The renal epithelial sodium channel (ENaC) 2 is a known contributor to the development of salt-sensitive hypertension, par-ticularly in African Americans (1)(2)(3). Its regulation, therefore, has been the subject of much research over recent years. The distal portion of the renal tubule functions to fine-tune sodium reabsorption to regulate plasma sodium concentration. This is done transcellularly by movement of sodium through ENaC channels on the apical surface of the cell, followed by basolateral movement through the Na ϩ /K ϩ -ATPase. The epithelial cells in this segment must express different proteins on the apical versus basolateral surface of the cell to allow for regulated transcellular ion movement. Polarization of the single layer of epithelial cells lining the tubule is essential for the nephron to sense changes in plasma and tubular fluid composition and regulate ENaC and other membrane proteins appropriately (4 -6). ENaC must, therefore, be regulated differently by hormones present in the blood versus the tubular fluid.
The role of intracellular calcium ([Ca 2ϩ ] i ) in ENaC regulation is beginning to emerge. P2Y2 receptors are G protein-coupled receptors located on the apical membrane in principal cells, where they function to inhibit ENaC by a Ca 2ϩ -dependent mechanism (7,8). [Ca 2ϩ ] i inhibition of ENaC is, in fact, a well known phenomenon, and several publications by different investigators have proposed the following model. G q -coupled receptors activate phospholipase C, which causes release of Ca 2ϩ through IP 3 receptors on the endoplasmic reticulum near the apical plasma membrane, where ENaC is expressed (7). Following an increase in [Ca 2ϩ ] i , PKC is activated (9), and, via a separate pathway, Ca 2ϩ binds to calmodulin. ENaC is usually held in the membrane by phosphatidylinositol phosphates, particularly PIP 2 and phosphatidylinositol 1,4,5-trisphosphate (10,11). A protein, myristoylated alanine-rich C kinase substrate (MARCKS), normally stabilizes PIP 2 to create an anchoring domain for ENaC (12). When MARCKS is bound by calmodulin or phosphorylated by PKC, this causes loss of MARCKS from the membrane, destabilization of PIP 2 , and endocytosis of ENaC. [Ca 2ϩ ] i can also activate the ubiquitin ligase Nedd4-2, causing ubiquitination and proteasomal degradation of ENaC (13).
Although increases in [Ca 2ϩ ] i in the cytosol just beneath the apical plasma membrane most likely inhibit ENaC, research from our group and others suggests that increasing [Ca 2ϩ ] i in the cytosol very near the basolateral membrane of the cell may stimulate ENaC (14,15). Basal P2X4 receptors stimulate ENaC in a Xenopus distal tubule cell line (14). P2X4 (16,17). We tested the hypothesis that mitochondria in cortical collecting duct (CCD) function in a similar manner to create regions of high and low [Ca 2ϩ ] i within the cell (Ca 2ϩ pools) and that these pools allow the same second messenger to affect the same protein (i.e. ENaC) in opposing ways, depending on the origin (apical versus basal) of the signal. This allows the cell to respond differently to hormone signals that signal via [Ca 2ϩ ] i , depending on whether the hormone is sensed in the serosal or luminal compartment.

Materials and Methods
Cells-Experiments used either the 2F3 clone of A6 Xenopus distal tubule cells or mouse mpkCCD cells from the cortical collecting duct. The type of cell used in each experiment was decided by ease of use. 2F3 cells are suited for electrophysiology because of their ability to remain at room temperature for extended periods of time, high expression of ENaC, and low expression of other channels. mpkCCD cells were used for microscopy experiments because they are easy to transfect and easily take up mitochondrial dyes. Cells were grown on permeable supports to confluency, and then cells were polarized, and tight junctions were fully developed. For 2F3 cells, medium was supplemented with aldosterone (1.5 M) to increase ENaC activity. Ionomycin, when used, was dissolved in Ͻ0.05% ethanol. Although ethanol can stimulate ENaC, it has no effect at this concentration (18).
Imaging-To visualize mitochondria, cells were loaded with MitoTracker Red (Life Technologies) for 30 min before visualization. To visualize the ER, cells were loaded with ERTracker Blue (Life Technologies) for 10 min prior to imaging. To label the membrane, cells were transfected with PLC␦-1 PH-GFP (1.5 g/well, Addgene) using the Xfect system (Clontech). pN1-Lck-GCaMP tagged with enhanced GFP was used to visualize changes in [Ca 2ϩ ] i near the plasma membrane of the cell (19). To detect changes in mitochondrial Ca 2ϩ , cells were coldloaded with Rhodamine-2/AM (Life Technologies) because cold loading has been shown previously to increase the specificity of the dye for mitochondrial versus cytosolic Ca 2ϩ (20). MpkCCD cells were incubated with Rhod2/AM for 1.5 h at 4°C, followed by incubation in media without serum for 4 h at 37°C. Cells were rinsed briefly prior to use. Cells were imaged using an Olympus FV-1000 confocal microscope. To resolve subcellular distribution of fluorescent markers, z-stacks were obtained using sequential optical slices starting at the basal membrane. All images were taken using the same parameter settings.
Single-channel Patch Clamp-Single-channel patch clamping was performed as described previously (14). Briefly, a microelectrode was filled with physiological buffer (96 mM NaCl, 0.8 mM CaCl, 0.8 mM MgCl, and 20 mM HEPES (pH 7.4)) and lowered to touch the apical or basolateral membrane of a single cell. Suction was applied gently so that the membrane remained intact and a Ͼ1 G⍀ seal was formed. All current was recorded at a holding potential of 0 mV. ENaC was identified by its characteristic channel kinetics and current-voltage relationship. ENaC probability of opening was analyzed using ClampFit software. Empty patches (those with no apparent activity) comprised 30 -50% of all patches and were excluded from the study.
Data Analysis and Statistics-To quantify the confocal microscopy experiments, ImageJ was used. The number of pixels in a given area was calculated before and after drug addition. Averages were compared using Student's t test with p Ͻ 0.05 considered significant.

[Ca 2ϩ ] i Influences Renal ENaC in a Polarized Fashion-We
sought to test the hypothesis that ENaC is affected differently by changes in [Ca 2ϩ ] i depending from which pole of the cell the signal originates. To do this, we performed single-channel patch-clamping as described previously (15), and cells were treated with either 5 M apical ionomycin (Fig. 1A) or 15 M basolateral ionomycin (Fig. 1B) to increase [Ca 2ϩ ] i . After 1-3 min of apical application of ionomycin, we observed a significant decrease in channel activity measured as probability of opening. We saw that the decrease in ENaC activity was maintained through 4 -6 min of ionomycin addition. In contrast, basolateral application of ionomycin increased ENaC activity, but only after 13-15 min. In other words, the effect of [Ca 2ϩ ] i on ENaC is dependent on the side of the cell in which the increase in [Ca 2ϩ ] i occurs. Note that the difference in time course and concentration here is likely due to limited diffusion of agents administered from the basal side of the cell (see "Discussion"). We hypothesize that this polarized effect is the result of two different Ca 2ϩ -mediated signaling pathways present in the same cell type. Beneath the apical surface of the cell, Ca 2ϩ activates pathways that inhibit ENaC, but, beneath the basal surface, Ca 2ϩ activates pathways that stimulate ENaC. Despite the relatively diffusibility of Ca 2ϩ , these two pools of Ca 2ϩ must not interact for polarization to occur. We hypothesize that mitochondrial Ca 2ϩ uptake prevents diffusion of [Ca 2ϩ ] i within principal cells to allow for these two pathways to remain separated.
Preventing Mitochondrial Ca 2ϩ Transport Destroys the Polarized Effect of [Ca 2ϩ ] i on ENaC-We next tested whether the polarized effects of [Ca 2ϩ ] i on ENaC function were dependent on mitochondrial Ca 2ϩ uptake. Mitochondrial Ca 2ϩ uptake is dependent on movement through the inner mitochondrial membrane, a process facilitated by the mitochondrial calcium uniporter (MCU) (21) Ru360 is a drug that inhibits the MCU and the voltage-dependent anion channel, a channel that may facilitate movement across the outer membrane (22,23). We first tested the effect of apical application of Ru360 alone on ENaC function and found it to have an inhibitory effect that persisted over time ( Fig. 2A). This is not surprising because inhibiting Ca 2ϩ uptake by the mitochondria that lie just beneath the apical plasma membrane likely increases apical [Ca 2ϩ ] i slightly, inhibiting the channel. We then performed experiments in which Ru360 was first added to the apical surface of the cell to inhibit mitochondrial Ca 2ϩ uptake, followed by apical (Fig. 2B) or basolateral (Fig. 2C) addition of ionomycin, while ENaC activity was monitored by patch clamp. Although Fig. 1 shows that apical and basolateral ionomycin normally have very different effects on the ENaC probability of opening, the shape of the graphs in Fig. 2, B and C (the same experiment repeated in the absence of mitochondrial calcium uptake) are the same, suggesting that the polarized effect of [Ca 2ϩ ] i on ENaC is dependent on mitochondrial calcium uptake. ATP has been shown repeatedly to inhibit ENaC when applied apically (a process that is dependent on an increase in [Ca 2ϩ ] i (24)). Fig. 2D shows that, in the absence of mitochondrial calcium uptake, apical ATP has a stimulatory effect on ENaC. Notice that, in each of the panels of Fig. 2, B-D, the shape of the response is the same. Although Ru360 alone has an inhibitory effect on ENaC that persists over time, adding any agent that increases [Ca 2ϩ ] i on the apical or basolateral side of the cell has a stimulatory (or null effect that is likely Ru360 inhibition opposed by stimulation) effect on ENaC in the presence of Ru360. In these experiments, we apply Ru360 to the apical surface of the cell. The inability of mitochondria to take up Ca 2ϩ would lead to an increase in regional [Ca 2ϩ ] i . We know from the data in Fig. 1 (17,18). We hypothesized that mitochondria may be localized in bands in the cortical collecting duct to allow for [Ca 2ϩ ] i polarization and opposing effects on ENaC. We began testing this hypothesis by using MitoTracker Red to observe mitochondrial localization in cortical collecting duct cells. We used two live cortical collecting duct cell lines: A6 cells from Xenopus CCD and mpkCCD cells from mice. The z axis view of both mpkCCD cells (Fig. 3A) and A6 cells (Fig. 3B) demonstrates that, in fact, mitochondria localize in the cytosol very close to each membrane in the cortical collecting duct. To pinpoint the location of mitochondrial bands with respect to the plasma membrane (Fig. 3C), we used the fluorescent PIP 2 indicator PLC␦-1 PH-GFP to label the apical and lateral membranes. We found that the apical mitochondrial bands are located just beneath the plasma membrane. Release of ER Ca 2ϩ through IP 3 receptor channels is known to inhibit ENaC. We hypothesized that, for the Ca 2ϩ exiting the ER through the IP 3 receptors to access ENaC and not be taken up by the mitochondria, sections of the ER would have to lie between the apical mitochondrial barrier and the apical plasma membrane. To observe ER localization in relation to mitochondria, we loaded mpkCCD cells with MitoTracker Red and ERTracker Blue to label the mitochondria and ER, respectively (Fig. 3D). We found that the ER does pass the mitochondrial barrier in spots in live cells. It is possible that this localization is necessary for G q receptor inhibition of ENaC because IP 3 receptor-mediated Ca 2ϩ release would have to occur in the cytosol between the apical membrane, where ENaC is localized, and the mitochondrial band, where Ca 2ϩ would be sequestered for the majority of the Ca 2ϩ to access the channel. Fig. 3E shows transmission electron microscopy of a mouse cortical collecting duct principal cell. Note the localization of mitochondria, as indicated by black lines highlighting the bands. Location of ϳ900 mitochondria in more than 10 different cells is quantified in Fig. 3F.
[ Fig. 3, we show that mitochondria in the CCD form bands underneath the apical and basal membranes. This morphology is similar to other epithelia, such as airway epithelia, in which mitochondrial bands slow Ca 2ϩ movement across the cell. Because morphology does not necessarily imply function, we tested whether Ca 2ϩ can move across live CCD cells from one membrane to the other (Fig. 4). To be able to clearly observe Ca 2ϩ beneath the cell membrane at the apical and basal poles, we transfected cells with a membrane-tethered Ca 2ϩ sensor, pN1-Lck-CaMP2 (19). This protein contains a membrane-tethering domain from Src and a GFP molecule that is fused to calmodulin so that an increase in [Ca 2ϩ ] i beneath the membrane enhances the fluorescence of the GFP molecule. In Fig.  4A, we applied ionomycin to the apical surface of the cell and found that [Ca 2ϩ ] i increased in the apical but not basal region of the cytosol. Conversely, Fig. 4 B shows that basal ionomycin increases [Ca 2ϩ ] i in the basal but not apical portion of the cytosol, suggesting that [Ca 2ϩ ] i is compartmentalized in mpkCCD cells. Fig. 4C shows that, in the absence of mitochondrial Ca 2ϩ uptake, [Ca 2ϩ ] i compartmentalization is lost and that apical application of ionomycin results in an increase in [Ca 2ϩ ] i in both the apical and basal portions of the cytosol.

Ca 2ϩ ] i Does Not Spread in the Cortical Collecting Duct-In
Mitochondrial Bands Take Up Ca 2ϩ following [Ca 2ϩ ] i Increases in CCD-To test whether mitochondria take up Ca 2ϩ following an increase in [Ca 2ϩ ] i , we cold-loaded mpkCCD cells with the dye Rhod2/AM, which has been shown to be specific for mitochondrial Ca 2ϩ when cells are cold-loaded (20) (Fig. 5). After apical application of ionomycin, there was an increase in mitochondrial Ca 2ϩ in the apical but not basal mitochondrial band. These data show that mitochondria in CCD cells do take up Ca 2ϩ rapidly following an increase in [Ca 2ϩ ] i at the apical pole of the cell and also confirm that there is limited Ca 2ϩ movement across the cell because there is no Ca 2ϩ uptake by the basal band following an apical increase in [Ca 2ϩ ] i . Fig. 5B shows localization of Rhod2/AM in mitochondria, and Fig. 5C shows that Ru360 prevents Ca 2ϩ uptake into mitochondria.

Discussion
The data shown here are the first to directly show that the effect of increasing [Ca 2ϩ ] i at the apical versus basolateral membrane produces a different effect on ENaC function in the kidney. Polarization of [Ca 2ϩ ] i implies that [Ca 2ϩ ] i is compartmentalized in renal epithelia and does not diffuse between compartments. We show that, in a principal cell line, apical application of the Ca 2ϩ ionophore ionomycin inhibits ENaC activity (Fig. 1). The effects of increasing apical [Ca 2ϩ ] i on renal ENaC function have been well studied. The most accepted physiological mechanism by which Ca 2ϩ inhibits ENaC occurs downstream of purinergic signaling via the P2Y2 receptor on the apical membrane (7). P2Y2 receptors signal via a G q -coupled receptor pathway to induce Ca 2ϩ release from ER stores. This release of Ca 2ϩ can act via a variety of signaling pathways, including PKC activation, to inhibit ENaC (25). This is demonstrated physiologically by the fact that mice lacking PKC have overactive ENaC and salt-sensitive hypertension (9,26). ENaC must be tethered to the apical membrane by PIP 2 or phosphatidylinositol 1,4,5-trisphosphate. Although both PIPs and ENaC are rare, they are recruited together by MARCKS. Binding of calmodulin to MARCKS or phosphorylation by PKC leads to removal of MARCKS and, therefore, ENaC from the membrane. Ca 2ϩ may also inhibit ENaC directly or activate the ubiquitin ligase Nedd4-2 to tag ENaC for proteasomal degradation (13,27). Interestingly, our data also show, for the first time, that ENaC is stimulated by basolaterally applied ionomycin (Fig. 1B). It is of note that stimulating ENaC requires a very high dose of ionomycin and takes much longer than apical ionomycin to cause an effect. This could imply that effects of basal [Ca 2ϩ ] i on ENaC are not physiologically relevant. It is likely that the delay in response and high dose of ionomycin are due to the larger volume of media on the basolateral side of the cells, small size of the pores in the polyester membranes used as a surface in our experiments, and invagination of the basal membrane which limit the diffusion of ionomycin so that a larger dose and more time are needed. Unfortunately determining whether this is the case would be difficult to test experimentally.
What is not clear from this work (or work published previously) are the signaling mechanisms dictating how basolateral increases in [Ca 2ϩ ] i work to stimulate ENaC activity. Our previous work shows that basolateral P2X4 receptor stimulation of ENaC is dependent on PI3K and Ca 2ϩ . Because Ca 2ϩ is not known to modulate PI3K directly, there must be an intermediate signaling molecule mediating the process. Src is one protein that can be regulated by Ca 2ϩ and can modulate PI3K and has been shown to regulate ENaC (28). It is possible, therefore, that src is an intermediate in P2X4 signaling or that P2X4 signaling cannot occur without cross-talk from receptor tyrosine kinases upstream of src. Future investigation of these signaling pathways will provide valuable insights into ENaC regulation.
Using two cell culture lines derived from this kidney segment, we observe that, in a polarized, live cell culture system, mitochondria form distinct bands at the apical and basal poles of the cells (Fig. 3). Bands such as these have been identified previously in other epithelial cell types: pancreatic acinar cells and airway epithelia (16,17). Interestingly, airway and renal epithelia are the primary sites of ENaC expression (29). Although the influence of airway mitochondrial bands on ENaC activity has not been addressed, it is interesting that all tissues in which banding has been observed rely heavily on ENaC for function. We also show that this banding occurs in vivo because transmission electron microscopy of mouse kidney shows banding in cortical collecting duct principal cells.
In this study, we observed the apical mitochondrial band to be situated just beneath the plasma membrane with portions of the ER jutting through the band. This particular localization is interesting in that it would allow the mitochondria to participate in Ca 2ϩ regulation near lipid raft domains. Indeed, ENaC is known to localize in such domains and proteins such as Nedd4 -2 and MARCKS, known to regulate ENaC in these FIGURE 4. Calcium localization following ionomycin treatment in mpkCCD cells. Cells were transfected with Lck-CaMP to visualize Ca 2ϩ at the plasma membrane. All images are z stacks, with the apical surface at the top of the image. A, ionomycin was added apically, and the image was taken about 1 min later. B, ionomycin was added basally, pixels were quantified at 5 and 10 min, and the image shown is at 10 min following addition. C, Ru360 was added apically followed by apical ionomycin. n Ն 3/group; *, p Ͻ 0.05 compared with baseline.
domains are influenced by [Ca 2ϩ ] i (12,30). Future experiments should be aimed at determining whether disrupting mitochondrial Ca 2ϩ uptake influences lipid raft Ca 2ϩ signaling.
Using innovative Ca 2ϩ sensors for [Ca 2ϩ ] i beneath the membrane and mitochondrial Ca 2ϩ , we show that [Ca 2ϩ ] i is taken up by mitochondria following apical application of ionomycin and that [Ca 2ϩ ] i stimulated by ionomycin does not spread throughout the cell (Fig. 4). To visualize mitochondrial Ca 2ϩ , we used the dye Rhod2/AM (Fig. 5). This dye can sense [Ca 2ϩ ] i as well but has a preference for mitochondrial Ca 2ϩ when coldloaded and seemed to localize in a band in our study (20). We were really only able to visualize the apical mitochondrial band using Rhod2/AM. This could be due to the low uptake of the dye and the faintness of the basal band compared with the apical one. For this reason, a limitation of our study is that we were unable to measure Ca 2ϩ uptake by the basal mitochondrial band. We did see the Ca 2ϩ in the apical mitochondrial band increase following apical ionomycin application, indicative of Ca 2ϩ sequestration. To visualize [Ca 2ϩ ] i , we used the membrane-tethered [Ca 2ϩ ] i sensor Lck-CaMP (19). This vector consists of an enhanced GFP molecule fused to calmodulin and the membrane domain of Src. When Ca 2ϩ increases at the membrane, the conformation of calmodulin changes so that GFP is excited. This vector helped us to see [Ca 2ϩ ] i just under the membrane and determine whether [Ca 2ϩ ] i spread following ionomycin application. We found that mitochondria sequester [Ca 2ϩ ] i following apical increases in [Ca 2ϩ ] i and that, when [Ca 2ϩ ] i is increased on one side of the cell, it does not migrate to the opposing pole.
Mitochondria transport Ca 2ϩ across the inner mitochondrial membrane via the MCU (31). Ru360, a ruthenium compound used in this study to obliterate the polarized effects of [Ca 2ϩ ] i on ENaC, inhibits MCU as well as a component of outer mitochondrial membrane Ca 2ϩ uptake (22,32). In the absence of mitochondrial Ca 2ϩ uptake, a similar biphasic response of ENaC to [Ca 2ϩ ] i was observed regardless of which pole ionomycin was added to (Fig. 2). These data suggest that mitochondrial Ca 2ϩ uptake is required for the polarized response of ENaC to [Ca 2ϩ ] i .
It is interesting that mitochondrial Ca 2ϩ uptake has been implicated in blood pressure regulation in a Chinese family that displays maternally inherited hypertension (33). The data presented here provide the groundwork for future investigations into the role of mitochondrial Ca 2ϩ uptake in ENaC regulation in salt-sensitive hypertension.
Overall, we conclude that the renal epithelial sodium channel ENaC likely exists in microdomains of Ca 2ϩ signaling and that these domains are maintained by belts of mitochondria within the renal epithelial cell. Because this is the first work suggesting the existence of mitochondrial barriers to [Ca 2ϩ ] i movement in renal epithelia, much work remains to be done to determine what role these barriers play in the regulation of other proteins or systemic blood pressure.