Interaction with the Na,K-ATPase and Tissue Distribution of FXYD5 (Related to Ion Channel)*

FXYD5 (related to ion channel, dysadherin) is a member of the FXYD family of single span type I membrane proteins. Five members of this group have been shown to interact with the Na,K-ATPase and to modulate its properties. However, FXYD5 is structurally different from other family members and has been suggested to play a role in regulating E-cadherin and promoting metastasis (Ino, Y., Gotoh, M., Sakamoto, M., Tsukagoshi, K., and Hirohashi, S. (2002) Proc. Natl. Acad. Sci. U. S. A. 99, 365–370). The goal of this study was to determine whether FXYD5 can modulate the Na,K-ATPase activity, establish its cellular and tissue distribution, and characterize its biochemical properties. Anti-FXYD5 antibodies detected a 24-kDa polypeptide that was preferentially expressed in kidney, intestine, spleen, and lung. In kidney, FXYD5 resides in the basolateral membrane of the connecting tubule, the collecting tubule, and the intercalated cells of the collecting duct. However, there is also labeling of the apical membrane in long thin limb of Henle's loop. FXYD5 was effectively immunoprecipitated by antibodies to the α subunit of Na,K-ATPase and the anti-FXYD5 antibody immunoprecipitates α. Co-expressing FXYD5 with the α1 and β1 subunits of the Na,K-ATPase in Xenopus oocytes elicited a more than 2-fold increase in pump activity, measured either as ouabain-blockable outward current or as ouabain-sensitive 86Rb+ uptake. Thus, as found with other FXYD proteins, FXYD5 interacts with the Na,K-ATPase and modulates its properties.

Work in several laboratories led to the identification of a family of proteins, named after the common motif FXYD (1). Five members of this group have been shown to interact with the Na,K-ATPase and elicit different effects on its kinetics. These are as follows: FXYD1 (phospholemman, PLM), 3 (2); FXYD2 (the ␥ subunit of Na,K-ATPase, ␥) (3); FXYD3 (Mat-8) (4); FXYD4 (corticosteroid hormone-induced factor, CHIF) (5); and FXYD7 (6). In addition, a PLM-like protein from shark rectal gland has been characterized (7,8). The remaining two family members FXYD5 (related to ion channel, dysadherin) and FXYD6 have not yet been analyzed for possible interactions with the Na,K-ATPase. The working hypothesis is that all family members modulate the pump kinetics in vivo and function as tissue-specific modulators of the Na,K-ATPase (9 -11). However, other functions for FXYD proteins have also been suggested (12)(13)(14)(15)(16).
FXYD proteins are type I membrane proteins with an extracellular N terminus (sometimes including a signal peptide), a single transmembrane domain, and an intracellular C terminus. With the exception of FXYD5, the extracellular domain is shorter than 40 amino acids, including a cleavable signal peptide. In the case of FXYD5, the extracellular domain is long, Ͼ140 amino acids. On the other hand, FXYD5 has the shortest intracellular C-terminal segment of only 15 amino acids. FXYD5 has been cloned as a tissue-specific and developmentally regulated gene induced by the oncoprotein E2a-Pbx1 and termed "related to ion channel" (17). Independently, human FXYD5 was identified as a cancer-associated cell membrane protein, which down-regulates E-cadherin and promotes metastasis, and was termed dysadherin (14). 4 Dysadherin has been reported to be expressed in the plasma membrane of several types of carcinoma cells but not in nontumor cells (18 -20). The apparent molecular mass of the protein detected in these studies is 50 -55 kDa, much higher than its calculated molecular mass (Ͻ20 kDa). This observation has been suggested to reflect an extensive O-glycosylation (14,21).
This study reports the functional expression, some biochemical characterization, and the tissue and cellular distribution of FXYD5. Experiments utilizing specific antibodies demonstrate that FXYD5 can be found in normal cells and that it is particularly abundant in epithelial cells in kidney, intestine, and lung. Co-immunoprecipitation assays show that FXYD5 specifically interacts with the ␣ subunit of Na, K-ATPase, and functional assays demonstrate effects on the pump kinetics. Thus, as found for other FXYD proteins, FXYD5 is a tissuespecific modulator of the Na,K-ATPase. Part of these data have been reported in abstract form (22).

EXPERIMENTAL PROCEDURES
cDNA Clones-A mouse FXYD5 EST clone (accession number W98807) was sequenced from both ends. It contains the whole open reading frame, and its deduced amino acid sequence is identical to those previously reported (1,17). Hemagglutinin A (HA)-tagged constructs (HA-FXYD5) were prepared either by replacing amino acids 2-18 (the putative signal peptide) with an HA epitope or by inserting the HA epitope before the signal peptide (between the first and second amino acid). For expression in Escherichia coli, the FXYD5 coding region lacking the signal peptide was cloned into pGEX-4T1vector downstream and in-frame with glutathione S-transferase (GST). For expression in Xenopus oocytes, FXYD5, HA-FXYD5, rat ␣1 subunit of the Na,K-ATPase, and pig ␤1 subunit of the enzyme were subcloned between 5Ј and 3Ј sequences of Xenopus ␤-globin in pGEM or pBluescript-derived vectors (23). All constructs were verified by sequencing.
Antibodies-Rabbit polyclonal anti-FXYD5 antibodies have been raised against two peptides coupled to keyhole limpet hemocyanin through N-terminal cysteines. One is CRQLSQFCLNRHR corresponding to the last 13 amino acids at the intracellular C terminus. The other is CEATGSQTAAQT corresponding to amino acids 65-75 in the extracellular N-terminal segment. For immunocytochemistry, we have used antibodies that were purified by affinity chromatography using the immunizing peptides coupled to the HiTrap N-hydroxysuccinimideactivated HP column (Amersham Biosciences). Polyclonal antibodies to the C-terminal sequences of CHIF and ␥ were described previously (24). A monoclonal antibody to a sequence near the N terminus of the ␣1 subunit of Na,K-ATPase (6H) was kindly provided by Dr. M. J. Caplan, Yale University School of Medicine. A polyclonal antibody to the C-terminal sequence KETTY of the ␣ subunit of Na,K-ATPase was obtained from Dr. J. Kyte, University of California, San Diego. A monoclonal anti-HA antibody was purchased from Roche Applied Science. Intercalated cells were identified by a rabbit polyclonal antibody raised against a peptide corresponding to the C-terminal sequence of the 56-kDa subunit of vacuolar H-ATPase. 5 Tissue and Cell Fractionation and Immunoblotting-Mice (ICR) were sacrificed by cervical dislocation, and various organs were excised and rinsed in ice-cold HSE buffer (250 mM sucrose, 25 mM histidine, 1 mM EDTA, pH 7.2, and a mixture of protease inhibitors (1 mM PMSF, 10 mg/ml aprotinin, 10 mg/ml leupeptin, 2 mg/ml pepstatin A)). Tissue was homogenized using a Polytron homogenizer (Kinematica Switzerland, four times with 6-s pulses at setting 10). Tissue homogenates were sedimented for 15 min at 4000 ϫ g at 4°C. The supernatants were further centrifuged for 90 min at 20,000 ϫ g at 4°C. The supernatants (cytosols) were saved, and the pellets (membranes) were suspended in HSE buffer ϩ protease inhibitors. Protein content was determined by the method of Lowry. FXYD5 was found to be very sensitive to proteolytic digestion, especially from intestinal membranes. Therefore, it was important to re-add protease inhibitors to the membrane resuspension medium.
Xenopus oocytes were injected with 3 ng of cRNA encoding FXYD5 or HA-FXYD5. Three days later, oocytes were gently homogenized in a buffer containing 20 mM Tris, pH 7.6, 50 mM NaCl, 10 mM MgCl 2 , and protease inhibitors mixture (1 mM PMSF, 20 mg/ml leupeptin, and 20 mg/ml pepstatin A). The suspension was centrifuged through a 20% sucrose cushion, and the supernatant (cytosol) was collected. The pellet was then incubated for 30 min at 4°C in a lysis buffer containing 10 mM Tris-HCl, pH 8, 140 mM NaCl, 1% Triton X-100, protease inhibitors (1 mM PMSF, 20 mg/ml leupeptin, 20 mg/ml pepstatin A), and 100 mM iodoacetamide. The sample was centrifuged to separate the solubilized membrane proteins from the insoluble yolk and pigment.
Co-immunoprecipitation of FXYD5 and Na,K-ATPase-Membranes were first suspended and incubated for 30 min at room temperature in a buffer containing 25 mM imidazole, 1 mM EDTA, pH 7.5, with and without 10 mM RbCl ϩ 5 mM ouabain. The suspension was then chilled to 0°C, and C 12 E 10 was added to a final concentration of 1 mg/ml. The detergent-solubilized membranes were centrifuged for 30 min at 50,000 ϫ g; the supernatant collected, and RbCl was added to a final concentration of 100 mM. No significant amount of FXYD5 was detected in the pellet, indicating full solubilization in 1 mg/ml C 12 E 10 . For co-immunoprecipitation, aliquots of the supernatant were rotated at 4°C either overnight with 6H antibody covalently linked to protein A-Sepharose beads or for 4 h with purified anti-FXYD5 antibody and then overnight with protein A beads. Beads were sedimented for 5 min   at 10,000 ϫ g and washed three times in solubilization buffer containing 1 mg/ml C 12 E 10 , 25 mM imidazole, 1 mM EDTA, pH 7.5, with and without 100 mM RbCl ϩ 5 mM ouabain. They were resuspended in 0.1 M glycine, pH 2, for cross-linked 6H antibody or in SDS sample buffer for anti-FXYD5, and centrifuged (1 min at 3000 ϫ g). The supernatants and an aliquot of the total detergent-solubilized membranes were dissolved in SDS sample buffer, resolved on Tricine gel, and transferred to a PVDF membrane. The membrane was cut to high, medium, and low molecular weight regions that were probed with antibodies to ␣, FXYD5, and CHIF or ␥, respectively.
Na,K-ATPase Activity Measurements-Xenopus oocytes were injected with 50-nl aliquots containing 3 ng of rat ␣1 cRNA, 3 ng of pig ␤1 cRNA, and either 3 ng of FXYD5 cRNA or diluent. The pump activity was measured 3 days later either by recording ouabain-blockable and K ϩ -induced outward currents (25) or by measuring ouabain-sensitive 86 Rb ϩ uptake (26). In both cases measurements were done in the presence of 10 M ouabain, which fully blocks the endogenous Xenopus pump activity but not the expressed Na,K-ATPase. Oocytes were first loaded with Na ϩ for 2 h in a K ϩ -free medium composed of the following: 80 mM sodium gluconate, 0.82 mM MgCl 2 , 0.41 mM CaCl 2 , 10 mM N-methyl-D-glucamine/HEPES, pH 7.4, 5 mM BaCl 2 , and 10 mM tetraethylammonium chloride. For current measurements, oocytes were suspended in a Na ϩ -free medium containing 140 mM sucrose ϩ 10 M ouabain instead of sodium gluconate. Currents were recorded at a holding potential of Ϫ50 mV before and after the addition of 0.02, 0.1, 0.5, and 5 mM potassium gluconate. As a final step, 2 mM ouabain was added, and the ouabain-inhibited currents were recorded. For 86 Rb ϩ uptake measurements, Na ϩ -loaded oocytes were divided into two groups of 7-8 oocytes and incubated for an additional 15 min at 25°C in a solution containing 90 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM BaCl 2 , 10 mM HEPES, pH 7.4, and either 10 M or 2 mM ouabain. 86 Rb ϩ uptake was then initiated by the addition of 5 mM KCl ϩ 5 Ci/ml 86 RbCl. The uptake was stopped 12 min later by a 15-fold dilution and five washings in an ice-cold incubation solution containing 5 mM nonradioactive RbCl. Oocytes were then counted individually for 86 Rb ϩ content.
Immunofluorescence Microscopy of Mouse Tissue-For confocal microscopy mouse kidneys were fixed for 4 h at 4°C in 4% periodate/ lysine/paraformaldehyde, pH 6.2. They were embedded in paraffin blocks, and 2-m sections were cut with a microtome. De-paraffinized sections were preincubated in phosphate-buffered saline containing 0.1% skim milk, 0.05% saponin, and 0.2% fish gelatin. They were then incubated for 1 h at room temperature or overnight at 4°C with either purified polyclonal anti-FXYD5 antibody (1:50), a monoclonal anti-␣ antibody (6H, 1:100), or a biotinylated polyclonal antibody raise against the 56-kDa subunit of the vacuolar H-ATPase (1:300). The primary antibodies were detected with goat anti-rabbit IgG labeled with Alexa Fluor 488 (green) or with goat anti-mouse IgG labeled with Alexa Fluor 546 (red). Anti-H-ATPase was detected with fluorescein isothiocyanate-labeled streptavidin (1:200; green). For double labeling of FXYD5 and ␣, the labeling and detection were done with mixtures of antibodies. In double labeling for H-ATPase and FXYD5, the FXYD5 antibody was detected with goat anti-rabbit IgG labeled with Alexa Fluor 546 (red). The labeled antibodies were observed with a Leica TCS SL confocal laser-scanning microscope (Leica, Mannheim, Germany). . Confocal immunolocalization of FXYD5 in kidney cortex. A, cortical section exposed to anti-N antibody that had been preincubated with immunizing peptide. B, the same area imaged with differential-interference contrast. There is no labeling in collecting ducts (C) or surrounding proximal tubules. Weak autofluorescence in erythrocyte as in glomerulus (G). C and D, cortical section double-labeled for FXYD5 (green) and the ␣ subunit of Na,K-ATPase (red). C, FXYD5 antibodies label the basolateral surfaces of cells in connecting tubule/initial collecting duct but not other segments or glomerulus (G). D, double labeling with anti-␣ and anti-FXYD5 antibodies. Only the ␣ subunit of Na,K-ATPase (red) is expressed in distal (D) and proximal (P) convoluted tubules, but co-localization with FXYD5 (yellow/orange) is seen in connecting tubule and collecting duct cells. Bar, 20 m. NOVEMBER 11, 2005 • VOLUME 280 • NUMBER 45

Tissue Distribution and Biochemical Characterization of FXYD5
Electron Microscopy-For immunoelectron microscopy, mouse kidneys were fixed for 4 h at 4°C in 4% paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 7.2. Small tissue blocks were trimmed from the immersion-fixed kidneys from all kidney zones, cryo-protected with 2.3 M sucrose, mounted on holders, and frozen in liquid nitrogen. Immunoelectron microscopy was performed on cryosections prepared with a Leica Reichert Ultracut S cryo-ultramicrotome (Leica, Vienna, Austria). The ultrathin cryosections were first preincubated in phosphate-buffered saline containing 50 mM glycine and 1% bovine serum albumin or 0.1% skim milk powder. Antigen retrieval was carried out with Tris/ EGTA buffer, pH 9.0, for 60 min at 60°C. The sections were then incubated overnight at 4°C with polyclonal anti-N FXYD5 antibodies (1:50) and were visualized with goat anti-rabbit IgG conjugated to 10 nm colloidal gold particles (GAR.EM1O, BioCell Research Laboratories, Cardiff, UK) diluted 1:50 in phosphate-buffered saline with 0.1% skim milk powder and 5 mg/ml polyethylene glycol. The sections were stained with 0.3% uranyl acetate in 1.8% methylcellulose for 10 min before examination in an FEI Morgagni 208 electron microscope. The following immunolabeling controls were used: 1) substitution of the primary antibody with nonimmune rabbit IgG; 2) preincubation of the antibodies with FXYD5 peptide (ϳ50-fold molar excess); and 3) incubation without the use of primary antibody. All controls showed absence of labeling.

RESULTS
Two rabbit polyclonal antibodies have been raised against the C-and N-terminal peptides of FXYD5 and are termed anti-C and anti-N, respectively. In order to verify that the antibodies react specifically with FXYD5, we have tested them on Xenopus oocytes injected with FXYD5 and HA-FXYD5 cRNAs. The anti-C antibody recognized an ϳ24-kDa polypeptide present in oocytes injected with FXYD5 or HA-FXYD5 cRNA but not in noninjected oocytes (Fig. 1A). In oocytes injected with HA-FXYD5, a similar 24-kDa polypeptide was detected by the anti-HA and anti-C antibodies. A 24-kDa protein was recognized by the anti-C antibody also in kidney microsomes (Fig. 1B) as well as other tissue (see below). On the other hand, the anti-N antibody gave no signal in oocytes membranes (not shown) and faintly labeled a 24-kDa protein in kidney microsomes (Fig. 1B). Therefore, we concluded that both antibodies interact specifically with FXYD5, which migrates as a 24-kDa protein, but the anti-C antibody is more suitable for Western blot characterization. The anti-N antibody did, however, give better signals in staining tissue slices for FXYD5 using confocal and electron microscopy (see below).
FXYD5 has two hydrophobic regions of ϳ20 amino acids, suggested to be a signal peptide and transmembrane domain, respectively. Signal peptide prediction, done according to Ref. 27, indicates that the initial hydrophobic sequence is indeed likely to be a cleavable signal peptide, with a cleavage site between amino acids 21 and 22. The HA-FXYD5 protein detected in Fig. 1A lacks the putative signal peptide and has instead nine amino acids corresponding to the HA epitope. The fact that this protein runs somewhat higher in the gel than the one expressed by unmodified FXYD5 cRNA (the two rightmost lanes in Fig. 1A) indicates that the signal peptide is indeed cleaved in the native protein. Also, modification of the N terminus seemed to impair trafficking of the protein to the plasma membrane, because HA-FXYD5 was recovered in both the cytosolic and membrane fractions, whereas FXYD5 was detected only in membranes (Fig. 1A). In order to examine this issue further, we studied expression of a construct in which the HA epitope was inserted upstream of the signal peptide, between the first and second amino acids. The protein expressed by this construct was well recognized by the anti-C antibody but not by the anti HA antibody, confirming cleavage of the signal peptide (Fig. 1C).
The molecular weight of FXYD5 detected in Xenopus oocytes and native tissue was significantly higher than the value calculated from the sequence (19.4 and 17.2 kDa with and without the signal peptide, respectively). This discrepancy may reflect abnormal electrophoretic mobility or glycosylation of the matured protein. However, the mobility of FXYD5 was not affected by treatment with peptide N-glycosidase or O-glycosidase (data not shown), suggesting that an abnormal electrophoretic mobility is the more likely possibility. The mobility of recom- binant protein formed between GST and FXYD5 also appears abnormal because it migrated as a 50-kDa polypeptide, even though its calculated molecular mass is 44 kDa (data not shown). This fusion protein was expressed in E. coli, so it cannot be glycosylated. The 24-kDa protein detected in this study appears to be quite different in size from the 50 -55-kDa glycoprotein reported to correspond to the human ortholog, dysadherin, and was suggested to be extensively O-glycosylated (14,21). Possible reasons for this discrepancy are discussed below.
The anti-C antibody has been used to study the tissue distribution of FXYD5 by Western blotting. FXYD5 was particularly abundant in intestine, spleen, lung, and kidney but not in brain, liver, muscle, and heart ( Fig. 2A). The Ͼ40-kDa polypeptide detected in brain was seen also with preimmune serum and hence is likely to be an unrelated protein. The segmental distribution of FXYD5 in the kidney and intestine is illustrated in Fig. 2, B and C. In intestine, expression was particularly high in the duodenum and weak in the jejunum, ileum, and proximal and distal colon. Also, no FXYD5 was detected in the stomach (not shown). In kidney, labeling was high in the cortex and much lower in medulla and papilla (Fig. 2C).
Characterization of the expression of FXYD5 along the nephron was done by confocal fluorescence microscopy of mouse kidney tissue. No specific signal could be obtained with the anti-C antibody. This is presumably due to the fact that the C terminus of FXYD5 is only 15 amino acids long and is not sufficiently exposed to the antibody in intact membranes. On the other hand, the anti-N antibody bound to some nephron segments (Figs. 3 and 4). Specificity of its binding was demonstrated by the competing effect of the immunizing peptide (Fig. 3A). Labeling of cortical sections with anti-N antibody demonstrated that FXYD5 is expressed on the basolateral surfaces of cells in connecting tubules and initial collecting ducts (Fig. 3C). Double labeling with this antibody (Fig.  3D, green) and a monoclonal antibody to the ␣1 subunit of Na,K-ATPase (red) shows co-localization of both proteins in the connecting tubules and cortical collecting duct, but a lack of FXYD5 in the distal and proximal convoluted tubules or glomeruli.
In the inner stripe of outer medulla (ISOM), some cells in the collecting ducts showed distinct basolateral labeling for FXYD5 (Fig. 4, A and  B), whereas all cells in adjacent thick ascending limbs of Henle (thick ascending limb) were unlabeled. To identify the ISOM cells labeled by FXYD5, we carried out double labeling of ISOM section for FXYD5 and for the 56-kDa subunit of H-ATPase, an apical marker of intercalated cells (28). Such double labeling, depicted in Fig. 4C, demonstrates that the cells labeled basolaterally for FXYD5 (red) also expressed the 56-kDa protein on their apical surface (green), and hence are intercalated cells. The same expression pattern persisted in the first section of the collecting duct in inner medulla (IMCD1), but toward the tip of the papilla there was no labeling for FXYD5 in the collecting duct (IMCD3). Most surprisingly, in this segment cells in the thin limb of Henle showed labeling, and this label was confined to the apical surface (Fig. 4, D and  E). Because this apical labeling was seen only adjacent to IMCD3, we assume that it is specific to long loops. Blood vessels did not express FXYD5 in any part of the kidney.
Further characterization of the cellular localization of FXYD5 in kidney was done by electron microscopy. Immunogold labeling of sections from ISOM and the first section of inner medulla demonstrated expression of FXYD5 in intercalated cells of the collecting duct but not in principal cells. The gold particles were only associated with the basolateral cell membrane (Fig. 5).
Next, we have assessed the possibility that, similar to other FXYD proteins, FXYD5 interacts with the ␣␤ complex of the Na,K-ATPase. Accordingly, kidney cortex microsomes were solubilized in 1 mg/ml C 12 E 10 . This detergent was found before to preserve FXYD/␣␤ interactions, in particular in the presence of Rb ϩ and ouabain (5). The solubilized proteins were immunoprecipitated by either anti-FXYD5 or anti-␣ antibodies, and the presence of various proteins in the immunopellets was detected in Western blots. Fig. 6 depicts a typical experiment showing effective immunoprecipitation of FXYD5 by the anti-␣ antibody. Efficiency of the co-immunoprecipitation can be estimated from the ratio of the bands in the total and immunopellet lanes. This analysis indicates that the fraction of FXYD5 immunoprecipitated by the anti-␣ antibody is comparable or even higher than that of CHIF or ␥. Unlike CHIF and ␥, the co-immunoprecipitation of FXYD5 was independent of the presence of Rb ϩ and ouabain, suggesting a particularly strong association between FXYD5 and the ␣␤ complex, which is not precluded or diminished by the non-native pump structure produced by solubilization of the protein with C 12 E 10 without the stabilizing ligands. Co-immunoprecipitation of FXYD5 and ␣ has also been demonstrated using anti-FXYD5 antibody. Immunoprecipitation of FXYD5 with the anti-N antibody also co-immunoprecipitated ␣ (Fig. 7). The fraction of ␣ immunoprecipitated by the anti-FXYD5 antibody was low, presumably  NOVEMBER 11, 2005 • VOLUME 280 • NUMBER 45 due to the fact that only a small fraction of the cells in the cortex express FXYD5. The immunopellet did not contain CHIF or ␥, even though more than one FXYD protein may be present in the same kidney cell. Thus, mixed complexes containing two FXYD proteins simultaneously associated with ␣␤ are not detected, as observed before with CHIF and ␥ (5).

Tissue Distribution and Biochemical Characterization of FXYD5
Finally, we examined functional effects of FXYD5 in Na ϩ -loaded Xenopus oocytes expressing rat ␣1 and pig ␤1. At a holding potential of Ϫ50 mV and 5 mM external K ϩ , a small outward current that was inhibited by 2 mM (but not by 10 M) ouabain could be detected. This current was Ͼ2-fold higher in FXYD5-injected oocytes, indicating a substantial FXYD5-induced activation of the Na,K-ATPase (Fig. 8A). The ouabaininsensitive negative current was not significantly affected by the expression of FXYD5, suggesting that FXYD5 does not affect other conductive pathways under these conditions. Measurements of the K ϩ activation curve show again the effect of FXYD5 on V max and also an increase in K1 ⁄ 2 for external K ϩ (Fig. 8B). However, because at the relevant concentrations of external K ϩ the pump-mediated currents were very small, determination of the K1 ⁄ 2 for the K ϩ activation by this technique is sub-ject to a large experimental error. The same oocytes used to measure the K ϩ activation curve in Fig. 8B were lysed and assayed by Western blotting for the expression of ␣1, ␤1, and FXYD5 (Fig. 9A). The data clearly show that the expression of FXYD5 markedly lowers the level of ␣1 and ␤1 protein expressed. In addition, FXYD5 appeared to decrease the level of ␤-glycosylation as evident from the lower position of the maximally glycosylated form and the relative intensities of the core-glycosylated and fully glycosylated species. Assuming that the ␣1 and ␤1 abundances detected in Fig. 9A also reflect differences in the cell surface expression of these proteins, the functional effect of FXYD5 seen in Fig. 8 may in fact be 2-3-fold higher, when adjusted for the pump expression level.
Effects of FXYD5 on pump activity were further confirmed by measuring the initial rate of ouabain-sensitive 86 Rb ϩ uptake into oocytes. These measurements, summarized in Fig. 9B, show that the expression of FXYD5 increases the ouabain-blockable 86 Rb ϩ uptake by Ͼ2-fold without altering the ouabain-insensitive component.

DISCUSSION
The current study provides a biochemical characterization of FXYD5, documents its tissue distribution, and demonstrates a functional interaction with the Na,K-ATPase. The data show that FXYD5 is an epithelial protein that is specifically expressed in kidney cortex, intestinal duodenum, and lung. It interacts with the Na,K-ATPase both functionally and structurally. Thus, as found with other FXYD proteins, FXYD5 is a tissue-specific regulator or modulator of the pump. Its presence in the apical surface of thin ascending limb cells is surprising. Such apical location has also been recently reported for Mat-8 (4) and suggests additional functions for these FXYD proteins. However, one cannot exclude the possibility that the apical epitope in epithelial cells is a different, unrelated protein.
Western blots of membranes from transfected cells and native tissue showed that FXYD5 is expressed as a protein with an apparent molecular mass of 24 kDa. This protein is much smaller than the 50 -55-kDa glycoprotein reported to be human FXYD5 (dysadherin) (14). Because O-glycosylation was suggested as an explanation for the high apparent molecular mass in human FXYD5, it is possible that the above discrepancy reflects different degrees of glycosylation in the human as opposed to mouse protein or metastatic as opposed to normal cells. The . Co-immunoprecipitation of FXYD5 and the Na,K-ATPase. Kidney cortex membranes were solubilized in 1 mg/ml C 10 E 12 in the presence of 10 mM Rb ϩ ϩ 1 mM ouabain (Rb/oub.) or an equal amount of Tris-HCl (Tris). A, the detergent-solubilized proteins were immunoprecipitated with a monoclonal antibody to the N terminus of the ␣ subunit of Na,K-ATPase (6H), covalently attached to Sepharose A beads. In another sample marked as con, beads without 6H antibody were used. 5% of the solubilized membranes (Total 5%) and the whole immunopellet (IP) were resolved electrophoretically and transferred to PVDF membranes. The membranes were cut into high, medium, and low molecular weight regions that were probed with antibodies to C-terminal sequences of ␣ (KETTY), FXYD5, and CHIF or ␥, respectively. B, co-immunoprecipitation and probing for FXYD and ␣ was done as in A. Each sample was divided into two equal portions that were probed either with primary (anti-KETTY and anti-FXYD) and secondary antibodies (Ab IϩII) or secondary antibody alone (Ab II). FIGURE 7. FXYD5 does not co-precipitate with other FXYD proteins. C 12 E 10 -solubilized kidney membranes were immunoprecipitated (IP) with anti-N FXYD5 antibody, and the immunopellets were probed for the presence of ␣, FXYD5, CHIF, and ␥ using antibodies to the C-terminal sequences of these proteins. Rb/oub., Rb ϩ ϩ ouabain.
observed molecular mass of mouse FXYD5 (24 kDa) is, however, higher by 6 kDa than the calculated value (17.2 kDa), which may reflect some glycosylation. The primary sequence of FXYD5 predicts a number of O-glycosylation sites but no N-glycosylation sites. However, two enzymes that should remove sugar moieties had no effect on the electrophoretic mobility of the protein. Moreover, the apparent molecular mass of GST-FXYD5 expressed in E. coli is also higher by 6 kDa than the calculated value (50 versus 44 kDa). Therefore, it can be concluded that in normal cells FXYD5 is not significantly glycosylated and has an abnormally low electrophoretic mobility. This low mobility may be due either to an abnormally low amount of bound SDS or to its high isoelectric point (pI ϭ 9.12). It is also interesting to note that the Western blots of Fig. 2 did not detect a lower molecular mass species as predicted in Ref. 29. Expression of HA-FXYD5 constructs in which the tag was placed either before or instead of the putative signal peptide has shown that, like CHIF (30) but unlike Mat-8 (4), the signal peptide is cleaved in the mature protein. Because FXYD5 was expressed in oocytes without ␣␤ in these experiments, it appears that the two pump subunits are not required for its full processing and trafficking to the plasma membrane.
We have demonstrated that FXYD5 is expressed mainly in epithelial tissue such as kidney, intestine, and lung. In kidney, the expression level appears to be highest in the cortex with reduced labeling in the medulla and papilla. In the intestine, FXYD5 was present mainly in the duodenum. Previous studies have suggested that FXYD proteins form 1:1 complexes with the ␣ and ␤ subunits of the pump (11). If this is the case also for FXYD5, its distribution along the intestine may reflect either expression in a specific population of cells that are abundant in duodenum but not in distal colon or a decrease in the ␣␤ abundance along the intestinal tract. Fig. 2B clearly shows a marked change in the FXYD5:␣ ratio along the intestine, indicating that the first possibility is the more likely one.
A more detailed characterization of the distribution of FXYD5 along the nephron was provided by confocal fluorescence microscopy. These studies have localized FXYD5 to the basolateral membrane of the connecting tubule, the collecting tubule, and the intercalated cells of the collecting duct. This pattern of expression is different from that of CHIF and ␥, the other two FXYD proteins found to be expressed in kidney. CHIF is expressed only in the principal cells of the cortical, outer, and inner medullary collecting duct (31,32). ␥a and ␥b, on the other hand, are detected mainly in the thick ascending limb of Henle's loop distal convoluted tubule as well as in the proximal tubule (33,34). However, FIGURE 8. Effects of FXYD5 on the Na,K-ATPase activity in Xenopus oocytes. Pump currents were measured in Xenopus oocytes expressing ␣1␤1 Ϯ FXYD5 as described under "Experimental Procedures." A, ouabain-blockable and ouabain-insensitive currents at Ϫ50 mV and 5 mM external K ϩ . Means Ϯ S.E. of 25 (␣1␤1) and 14 (␣1␤1ϩFXYD5) oocytes from three different frogs are depicted. Two-tailed t test indicates significant difference in the ouabain-blockable (p Ͻ 0.01) but not ouabain-insensitive (p Ͼ 0.2) currents. B, K ϩ activation curve in oocytes expressing ␣1␤1 Ϯ FXYD5 at zero external Na ϩ . Data from 6 to 10 oocytes are averaged. The continuous line is the best fit to the equation V ϭ V max /(1 ϩ K1 ⁄2 /C n ), where V is the current at an external K ϩ concentration of C, and n the Hill coefficient. The best fit values were as follows: ␣1␤1 Ϫ V max ϭ 120.5 nA, K1 ⁄2 ϭ 34.6 M; ␣1␤1 ϩ FXYD5 V max ϭ 182.2 nA, K1 ⁄2 ϭ 133.5 M. In neither case did n differ significantly from 1.  Fig. 8B were individually lysed in 1% Triton X-100, resolved electrophoretically, and probed with anti-␣1 (6H), anti-␤, and anti-FXYD5 (anti-C terminus). cg and fg label the core-glycosylated and fully glycosylated ␤ species, respectively. B, 86 Rb ϩ uptake was measured in oocytes injected with ␣1␤1 Ϯ FXYD5 as described under "Experimental Procedures." Data were combined from three frogs and normalized to the mean uptake without ouabain in oocytes injected with ␣1␤1 (100%). Mean Ϯ S.E. of 25 (Ϫouabain) or 17 (ϩouabain) oocytes are depicted. some expression of ␥a is seen also in inner medullary collecting duct in the same cells that express CHIF, and the ␥a splice variant was also found in the intercalated cells of the inner stripe of the outer medulla, cells that also express FXYD5 (32). Thus, although the expression pattern of ␥, CHIF, and FXYD5 along the nephron are different, some cells should contain two of these FXYD proteins.
The possibility that FXYD5 is associated with the pump was demonstrated by co-immunoprecipitation. The amount of ␣ co-immunoprecipitated with FXYD5 is similar or even higher than that seen before for CHIF, ␥, and PLM (2,24,35). Unlike CHIF and ␥ (5), but similar to PLM (2), co-immunoprecipitation was efficient also in the absence of Rb ϩ and ouabain that stabilize the native pump structure. It was also found that the anti-FXYD5 antibody precipitates ␣ but not CHIF or ␥ (Fig. 7). Similarly the anti-CHIF antibody does not precipitate ␥ and vice versa (5). Thus, even in cells expressing both FXYD5 and ␥, or FXYD5 and CHIF, ␣ does not seem to interact simultaneously with more than one FXYD protein.
Finally, we have studied the effects of FXYD5 on the Na,K-ATPase activity. In the oocyte expression system FXYD5 was found to increase the pump activity measured either as ouabain-blockable and K ϩ -induced outward current or as ouabain-inhibitable 86 Rb ϩ uptake. Assuming that loading the cells increased intracellular Na ϩ to a saturating concentration, the increase in pump current and 86 Rb ϩ uptake at 5 mM external K ϩ reflects an FXYD5-induced increase in V max . Thus, at least one of the functional effects of FXYD5 is to increase maximal pump activity. Other potential effects such as modulating apparent affinities of Na ϩ , K ϩ , and ATP have not yet been examined in detail. It was also observed that the abundance of ␣ and ␤ is lower by 2-3-fold and that ␤ is less glycosylated in oocytes that express FXYD5. This effect may simply reflect saturation of the translation machinery by injecting oocytes with FXYD5 cRNA on top of ␣ and ␤. However, it may also indicate an additional role for this FXYD protein in the translational regulation of ␣ and/or ␤. Normalization of the functional data for differences in the ␣ abundance, with and without FXYD5, would at least double the observed effect on V max . Such normalization, however, assumes that cell surface expression of the pump is proportional to its total abundance detected by Western blots of whole cell lysates.
Five other FXYD proteins have been shown to modulate various pump properties (for review see Refs. 9 -11). Of these, ␥, PLM, and the PLM-like protein from shark rectal gland were shown to affect V max values (8, 36 -38). It was also reported that the PLM-induced increase in pump activity depends on its phosphorylation (39,40). With the exception of the heart, the tissue distributions of FXYD5 and PLM are nonoverlapping. However, even if two FXYD proteins with similar functional actions are present in the same heart cells, they might be subject to different regulatory mechanisms, which will make their functional effects nonredundant.
In summary, this study provides the first biochemical and functional characterization of FXYD5 and demonstrates that, as found for other FXYD proteins, it interacts with the Na,K-ATPase and modulates its activity. The atypically long extracellular domain of this FXYD protein may physically link the Na,K-ATPase to the extracellular matrix, the ectodomain of another membrane protein, or some other factor in the interstitial fluid.