FXYD5 (dysadherin or RIC) is a member of the FXYD family of single-span transmembrane proteins associated with the Na+-K+-ATPase. Several studies have demonstrated enhanced expression of FXYD5 during metastasis and effects on cell adhesion and motility. The current study examines effects of FXYD5 on the paracellular permeability in the mouse kidney collecting duct cell line M1. Expressing FXYD5 in these cells leads to a large decrease in amiloride-insensitive transepithelial electrical resistance as well as increased permeability to 4-kDa dextran. Impairment of cell-cell contact was also demonstrated by staining cells for the tight and adherence junction markers zonula occludens-1 and β-catenin, respectively. This is further supported by large expansions of the interstitial spaces, visualized in electron microscope images. Expressing FXYD5 in M1 cells resulted in a decrease in N-glycosylation of β1 Na+-K+-ATPase, while silencing it in H1299 cells had an opposite effect. This may provide a mechanism for the above effects, since normal glycosylation of β1 plays an important role in cell-cell contact formation (Vagin O, Tokhtaeva E, Sachs G. J Biol Chem 281: 39573–39587, 2006).
- FXYD proteins
- tight junction
- adherence junction
the na+-k+-atpase (the Na+ pump) uses ATP to actively pump three Na+ ions out of the cell in exchange for two K+ ions flowing into the cell. The pump is composed of a catalytic α- and regulatory β-subunit. There are several isoforms of both α (α1–4) and β (β1–3), which are expressed in a tissue- and developmental-dependent fashion (5). Many studies have provided evidence that in addition to this “classic” role, the Na+-K+-ATPase has both structural and signaling functions (26, 28, 41). One such mechanism involves the role of β Na+-K+-ATPase in cell adhesion and cell-cell contact (11, 34, 40). It has been demonstrated that interactions between β1-subunits in neighboring Madin-Darby canine kidney (MDCK) cells participate in cell-cell contacts and affect paracellular permeability (40, 41). This interaction is mediated by the carbohydrate moieties on β1 and is largely impaired by inhibiting its N-glycosylation.
In addition to its α- and β-subunit, the Na+-K+-ATPase complex often contains a third subunit, which is a member of the FXYD family. FXYD proteins are short (<100 amino acids with the exception of FXYD5) single-span transmembrane proteins characterized by the invariant motif Phe-Xxx-Tyr-Asp in their extracellular domain. All members of this group were found to specifically associate with the Na+-K+-ATPase and modulate its kinetic properties (9, 10, 37). They are therefore thought to act as tissue-specific regulators or auxiliary subunits of the pump, whose role is to adjust its kinetic properties to specific requirements of the cell type or the physiological state under which they are expressed, without affecting it elsewhere (9).
FXYD5 (also termed dysadherin, or RIC) is a rather unique member of the FXYD family. Like other FXYD proteins, it specifically interacts with the pump and was shown to increase its Vmax (18, 19, 22). However, other studies also reported effects on cell adhesion, E-cadherin abundance, cell motility, and actin organization (13, 21, 22, 31, 38). FXYD5 is overexpressed in various tumors, and its expression level correlates with high metastasis and poor prognosis (for a review, see Ref. 24). A role for FXYD5 in cell adhesion and cell-cell contact is also supported by the fact that unlike all other FXYD proteins, FXYD5 has a relatively long extracellular domain (i.e., 145 vs. ∼30 amino acids in other FXYD proteins), and it was suggested to be heavily glycosylated. In native rodent tissues, FXYD5 runs as a ∼20-kDa polypeptide compatible with its calculated molecular weight (18, 19). In tumors and tumor-derived human cell lines, an ∼50- to 55-kDa polypeptide was reported, and the difference was suggested to reflect excessive O-glycosylation of the extracellular domain (13, 38).
The current study examines effects of FXYD5 on the rodent kidney collecting duct cell line M1. We found that expressing FXYD5 in these cells results in a large increase in the paracellular permeability measured as amiloride-insensitive transepithelial resistance (TER) and permeation of 4-kDa dextran. Expression of FXYD5 in M1 cells alters the cellular distribution of the tight and adherence markers zonula occludens-1 (ZO-1) and β-catenin and causes large dilations of the interstitial spaces. The expression of FXYD5 was also associated with a decrease in the glycosylation of β1, suggesting a possible mechanism for its effects on cell-cell contacts.
MATERIALS AND METHODS
Expression and silencing of FXYD5 in cultured cells.
M1 cells were purchased from the American Type Culture Collection and cultured in a 1:1 mixture of DMEM and F12 media supplemented with 5% fetal calf serum, 5 μM dexamethasone, and penicillin and streptomycin. Cells were transfected with mouse FXYD5 cDNA in which amino acids 103–106 were replaced by a hemagglutinin A (HA) epitope, subcloned into the pIRES-EGFP vector. This protein region is not conserved among species and therefore assumed to have no essential function. Transfection was done using jetPEI reagent (PolyPlus Transfection) according to the manufacturer's instructions. Positive clones were isolated by FACS and assayed for expression of FXYD5 using an anti-HA antibody. To silence the transfected cDNA, an FXYD5-expressing cell clone was further transfected with MISSION short hairpin (sh) RNA Plasmid DNA (clone ID NM_008761.2–515s1c1, Sigma-Aldrich), and positive clones were selected using puromycin.
Cell proliferation was quantified by reduction of 2,3-bis (2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino) carbonyl]-2H-tetrazolium hydroxide (XTT) using a cell proliferation assay kit (Biological Industries, Beit Haemek, Israel).
H1299 cells were cultured in RPMI 1640 supplemented with 10% fetal calf serum (Biological Industries) and penicillin and streptomycin. Silencing of endogenous FXYD5 was done using the pSUPER RNAi system (OligoEngine,) and the following oligonucleotides: forward-GATCCCCCACCATCGTTGGCCTGATTCTTTCAAGAGAAGAATCAGGCCAACGATGGTGTTTTTGGAAA and reverse-AGCTTTTCCAAAAACACCATCGTTGGCCTGATTCTTCTCTTGAAAGAATCAGGCCAACGATGGTGGGG. Cells were transfected using jetPEI DNA transfection reagent, and clones were selected using G-418 and assayed for the silencing of FXYD5 by Western blotting with a monoclonal anti-FXYD5 antibody and RT-PCR (19). RNA was isolated from H1299 cells using an RNeasy kit (Qiagen) and reverse transcribed from the poly A+ tail using a Super-Script II Reverse Transcriptase kit (Invitrogen). FXYD5 was then amplified using the forward and reverse primers 5′-ATGTCGCCCTCTGGTCGCCT-3′ and 5′-CTGCAATGATTCCGGCATAA-3′, respectively.
Analysis of total and surface expressed membrane proteins, coimmunoprecipitation, and Western blotting.
M1 cells were seeded in 24-mm 0.4-μm PET membrane bottom cups (Corning) and cultivated in BioWhittaker PC-1 medium (Lonza) plus 5% fetal calf serum and penicillin and streptomycin. Under these conditions, cells formed polarized epithelium with a TER of >1 kΩ·cm2. Confluent monolayers were surface biotinylated by 10-min incubation at 4°C with 1.5 mg/ml EZ-Link Sulfo-NHS-SS-Biotin (Pierce) in PBS without calcium or magnesium. The reaction was stopped by two washings plus a 20-min incubation with 100 mM glycine in PBS, and cells were lysed by rocking for 1 h at 4°C in RIPA buffer (20 mM Tris·HCl, pH 7.4, 2 mM EDTA, 2 mM EGTA, 1% Triton X-100, 0.1% SDS, 1 mM PMSF, 20 mg/ml leupeptin, and 20 mg/ml pepstatin A). The lysed cells were scraped off the plate, transferred to microtubes, and dispersed by repetitive pipetting and vortexing. Cell debris was removed by centrifugation at 5,000 g for 5 min, 2–10% of the volume was taken as “total protein” sample, and the rest (∼700 μl) was incubated overnight at 4°C with a 100-μl streptavidin agarose resin slurry (Pierce). The agarose beads were then washed two times in 50 mM Tris, pH 7.4, 50 mM NaCl, 5 mM EDTA, two times in 100 mM Tris, pH 7.4, 500 mM LiCl, and three times in 10 mM Tris, pH 7.4. The streptavidin-bound proteins were eluted by incubation with SDS sample buffer (cell surface fraction), and total and cell surface proteins were resolved electrophoretically on 7.5% acrylamide Tris tricine gels.
To quantify β1-Na+-K+-ATPase, preparations were deglycosylated by the following protocol. Whole cell lysates and proteins bound to streptavidin agarose were suspended in glycoprotein denaturing buffer (0.5% SDS, 40 mM DTT) and denaturated by 10-min incubation at 95°C followed by chilling on ice. This procedure simultaneously eluted proteins bound to the streptavidin agarose. The denatured total proteins and eluted proteins were diluted 1:2 in a 2× solution of 50 mM sodium phosphate, pH 7.5, 1% NP-40, and incubated for 1 h at 37°C with PNGase F (1 μl, 500,000 U/ml for 20 ug protein, New England BioLabs). In some experiments, the amount of β glycosylation was assessed in membrane microsomes prepared as described (14).
Coimmunoprecipitation of FXYD5 and α1-Na+-K+-ATPase from M1 cell lysates was measured under conditions that were found to preserve FXYD-pump interactions (8, 18). Confluent 100-mm plates of wild-type and HA-FXYD5-transfected M1 cortical collecting duct cells were rinsed twice with cold Ca2+- and Mg2+-free PBS; 2 ml of 1 mg/ml C12E10 in 25 mM imidazole, 5 mM Tris, pH 7.6, 100 mM RbCl, 5 mM ouabain, 1 mM PMSF, and a protease inhibitor cocktail (8340, dilute 1:100, Sigma) were added. Plates were rocked 1 h at 4°C. Cells were scraped off with a rubber scraper and transferred to 2-ml microtubes. Samples were centrifuged 5,000 g for 5 min at 4°C, and the supernatants containing solubilized proteins were used for coimmunoprecipitation. For immunoprecipitation of FXYD5 by anti-α1-Na+-K+-ATPase, detergent-solubilized proteins were incubated for 4 h with a monoclonal anti-α1 antibody (6H, 1:100) under swirling. Protein A beads were added and after an additional overnight incubation, the beads were precipitated and washed three times in the C12E10 buffer. Immunoprecipitated proteins were eluted with 1% Triton X-100. This detergent dissociates the FXYD5-Na+-K+-ATPase complex but not the association of the antibody with protein A (19). Hence contamination of the immunopellet with the antibody light chain which runs very close to FXYD5 is prevented. Coimmunoprecipitation of α1-Na+-K+-ATPase by anti-FXYD5 was done using anti-HA antibody attached to agarose beads (HA Tag IP/Co-IP Kit 26180, Pierce). Eight hundred microliters of detergent-solubilized proteins were mixed with 40 μl anti-HA-agarose in a plugged spin column and rotated for 16 h at 4°C. The spin column was unplugged and centrifuged at 16,000 g for 10 s, and the eluant was removed. The agarose pellet was suspended and washed six times with 0.5 ml TBS+0.05% Tween 20 (TBS-T). The spin column was replugged, and the agarose pellet was suspended in 60 μl 2× nonreducing 1% SDS sample buffer. This was heated on a heat block at 95°C for 10 min, and then the column was centrifuged at 16,000 g for 10 s to elute the immunoprecipitated proteins. The eluted proteins were mixed with 1/10 volume 1 M DTT and were resolved on polyacrylamide Tris tricine gels together with a sample of denatured whole cell lysate. Gel-resolved proteins were transferred to polyvinylidene difluoride membranes using CAPS buffer plus 10% methanol (15 V for 120 min). The blots were blocked with 5% milk (1 h, room temperature) and cut into low- and high-molecular weight segments. These were incubated overnight at 4°C with various antibodies, washed, and then incubated for 1 h at room temperature with horseradish peroxidase (HRP)-coupled goat anti-rabbit or goat anti-mouse IgG (1:10,000). ECL substrate (1.25 mM luminol, 0.198 mM cumaric acid, and 0.0094% hydrogen peroxide) was added, and luminescence was visualized on X-ray film and quantified using an ImageQuant LAS 4000 mini chemiluminescent reader (General Electric). All luminescence values to be compared were within the linear concentration range.
Fluorescent and electron microscopy.
For visualization of cell junction markers, M1 cells were cultured on coverslips and fixed with 3% paraformaldehyde+0.5% Triton X-100. They were incubated for 1 h at room temperature with polyclonal antibodies to either β-catenin or ZO-1 (1:500) followed by 3 ×5-min washing in PBS and a 30-min incubation at room temperature with Cy3-coupled goat anti-rabbit/mouse antibody (1:500, Jackson Laboratories). Samples were washed three times in PBS, covered with mounting medium (Immuno-mount, Thermo Scientific), and visualized by confocal microscopy (Olympus).
For electron microscopy, confluent monolayers cultivated on porous supports were fixed with 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4), washed in the same buffer, and postfixed with 1% osmium tetroxide. After en bloc staining with 2% uranylacetate in water for 1 h at room temperature, cells were dehydrated in graded ethanol solutions and embedded in Epon 812. Ultrathin sections (70- to 90-nm thickness) were prepared with Ultramicrotome Leica UCT (Leica). They were analyzed with a 120-kV SPIRIT Transmission Electron Microscope (FEI, Eidhoven) and digitized with an EAGLE (FEI) CCD camera using TIA (FEI) software.
Transepithelial permeability measurements.
Approximately 500,000 cells were seeded in 12-mm 0.4-μm PCF filter bottom cups (Millicell, Millipore). Transepithelial electrical potential and resistance were measured daily using an EVOMX voltohmmeter and chopstick STX2 electrodes (World Precision Instruments). Transcellular contributions to the current and resistance were determined by the addition of 10 μM amiloride to the apical compartment. To assess for ion selectivity via the paracellular pathway, TER was measured in the presence of amiloride in media containing 140 mM NaCl or Na-gluconate or N-methyl-d-glucamine (NMDG)-Cl.
Macromolecule permeation was measured in confluent monolayers manifesting FXYD5-dependent differences in resistance. At time 0, mixtures of rhodamine B (1 mg/ml) and either 4- or 70-kDa FITC-dextran (1 mg/ml) were added to the upper compartment. Aliquots (100 μl) were removed from the lower compartment at 30-min intervals and replaced by an equal volume of fresh medium. The samples' fluorescence was measured (rhodamine B: 520-nm excitation, 590-nm emission; FITC: 485-nm excitation, 544-nm emission) and corrected for dilutions in the lower compartments. Data are expressed as percentage of the equilibrium value determined in parallel using filters with no cells.
The following antibodies have been used for Western blotting at the indicated dilutions: a monoclonal antibody to the N terminus of α1-Na+-K+-ATPase (6H, 1:2,000, kindly provided by Dr. M. J. Caplan, Yale University School of Medicine); a polyclonal anti-β1-Na+-K+-ATPase (32) (1:10,000); a monoclonal anti-HA (1:2,000, Santa Cruz Biotechnology); a monoclonal 36/E-cadherin antibody (1:5,000, BD Biosiences); a monoclonal anti-β-tubulin (1:40,000, Sigma-Aldrich); a polyclonal anti-β-catenin (1:500, Sigma-Aldrich); a monoclonal anti occludin (1:1,000, Invitrogen 33–1500); and a monoclonal anti-FXYD5 (1:500) (19).
2-Acetamido-2-deoxy-α-d-galactopyranoside, FITC-dextran, rhodamine B, and amiloride HCl were purchased from Sigma-Aldrich.
Data are expressed as means ± SE or means ± SD as indicated in the figure legends. Statistical significance was determined by an unpaired Student's t-test.
Effects of FXYD5 on cell-cell contact were studied in the mouse kidney cortical collecting duct line M1 (36). This line was selected because of its epithelial nature and the fact that FXYD5 is natively expressed in the kidney collecting duct (18). However, no significant amounts of FXYD5 mRNA or protein were detected in these cells. Cells also were devoid of FXYD1, -2, and -4, known to be expressed in kidney (data not shown). M1 cells were stably transfected with mouse HA-tagged FXYD5. The expressed protein migrates as an ∼24-kDa polypeptide with occasional higher bands, in agreement with previous data from native rodent tissue and HA-FXYD5 expressed in Xenopus laevis oocytes (18, 19) (Fig. 1A). It is considerably lower than the 50–55 kDa reported for FXYD5 in human tumors and tumor-derived cell lines, which was suggested to reflect extensive O-glycosylation of this protein (13, 31, 38). FXYD5 could be coimmunoprecipitated from HA-FXYD5-transfected M1 cells by anti-α1-Na+-K+-ATPase (Fig. 1B, top). In agreement, α1-Na+-K+-ATPase was immunoprecipitated by anti-HA from HA-FXYD5-transfected but not from nontransfected cells (Fig. 1B, middle). Thus the association between FXYD5 and the Na+-K+-ATPase observed before in native epithelia is preserved in M1 cells (18). Also in agreement with previous oocyte data (19), expressing FXYD5 in M1 cells led to a decrease in the apparent molecular weight of β1-Na+-K+-ATPase, indicating less N-glycosylation of the pump subunit (Fig. 1C). Since this effect could be functionally important, we have confirmed it in a different cell system. H1299 is a human lung carcinoma cell line which natively expresses FXYD5. Silencing the corresponding mRNA led to an increase in the apparent molecular weight of β1, confirming the inhibitory effect of FXYD5 on β glycosylation (Fig. 1, D and E). Intriguingly, in H1299 cells FXYD5 migrated as an ∼40-kDa polypeptide, much heavier than the 24-kDa band obtained in transfected M1 cells. Possible reasons for this difference and the discrepancy with the previously reported 50- to 55-kDa polypeptide are discussed in the discussion.
The data in Fig. 1 seem to indicate that the expression FXYD5 also causes a decrease in the abundance of α1-Na+-K+-ATPase (cf. differences between wild-type and clone 4C5 in Fig. 1A and the difference between wild-type and shF5 clone in Fig. 1D). To further explore this issue, confluent cell monolayers from matched cultures that do or do not express FXYD5 were surface biotinylated and total and streptavidin-bound α1- and β1-Na+-K+-ATPase were quantified. To quantify β1, cell lysates were deglycosylated by incubation with PNGase. Indeed, a substantial downregulation of α1-Na+-K+-ATPase in FXYD5-transfected M1 cells was apparent, and the expression of FXYD5 lowered the total abundance of this protein by >40% (Fig. 2, Table 1). Surface expression of α1, however, was much less affected by the transfection, and the difference in the streptavidin pulled down fraction of α was hardly significant. For β, neither total nor plasma membrane protein was significantly affected by the expression of FXYD5.
Previously, it was reported that expression of FXYD5 is associated with downregulation of E-cadherin in some, but not all cell lines and tumors (3, 13, 30). We have therefore also examined the effect of FXYD5 on E-cadherin in M1 cells. No effect of FXYD5 on total or plasma membrane abundance of E-cadherin was detected (Fig. 2). Also, no physical association between FXYD5 and E-cadherin was noted (Fig. 1B, bottom). Thus in these cells the effect of FXYD5 on cell adhesion is not secondary to the downregulation of E-cadherin.
M1 cells are known to form polarized high-resistant epithelium expressing the amiloride-blockable Na+ channel ENaC in the apical membrane and Na+-K+-ATPase in the basolateral pole (36). Possible effects of FXYD5 on the formation of tight junctions (TJ) were measured by recording the time-dependent increase in TER in cells grown on permeable supports. Wild-type cultures formed monolayers with TER of >1 kΩ·cm2, typical of tight epithelium (Fig. 3A). These cells were also characterized by a considerable transmembrane potential, giving rise to short-circuit currents (Isc) of ∼25 μA/cm2 (Fig. 3B). Adding 10 μM amiloride to the luminal compartment abolished Isc and further increased TER to >2.0 kΩ·cm2. Under these conditions, TER is dominated by paracellular permeability. In the FXYD5-transfected cells, much lower TER values were recorded and no response to amiloride was seen. Thus FXYD5 appears to inhibit formation of TJs. To verify that the low TER values are indeed secondary to the expression of FXYD5, we performed partial silencing of FXYD5 in a cell clone that manifests the reduced TER. The decreased expression of FXYD5 led indeed to a partial reversal of its effect, resulting in a significant increase in TER and Isc compared with the FXYD5-expressing cells (Fig. 3).
Next, we assessed whether the FXYD5-induced decrease in TER reflects the activation of a particular conducting pathway. Accordingly, TER values were also measured following substitution of NaCl by either NMDG-Cl or Na-gluconate (Fig. 4). The ionic substitutions had only minor effects on TER values and clearly did not abolish the FXYD5-dependent decrease in TER. Thus this decrease is not likely to reflect induction of a Na+- or Cl−-conducting transporter.
In principle, the above observation may be secondary to effects of FXYD5 on the rate of cell growth and inability to form confluent monolayers during the 1-wk period monitored. To exclude such a possibility, we have determined the rate of cell proliferation in wild-type and FXYD5-transfected cells. No effect of FXYD5 on the rate of cell proliferation could be detected, arguing against such an option (Fig. 5A). The possibility that the FXYD5-transfected cells fail to form confluent monolayers is further excluded by the restricted permeation of large molecules as described below. Interestingly, the FXYD5-transfected cells detached from the plate by trypsinization much more quickly than the wild-type cells, confirming differences in cell-cell and/or cell-substrate contacts (Fig. 5B). Tsuiji et al. (38) reported that inhibiting O-glycosylation by incubating cells with benzyl 2-acetamido-2-deoxy-α-d-galactopyranoside (benzyl-α-GalNAc) decreases expression of FXYD5 and weakens its effects on cell adhesion and morphology. In our experiments, no effect of benzyl-α-GalNAc (4 mM present during the whole cultivation period) on TER in either wild-type or FXYD5-transfected M1 cells was observed. Analysis of proteins from these cultures on acrylamide gels showed a shift in the electrophoretic mobility of a few protein bands in the benzyl-α-GalNAc-treated sample, suggesting that the O-glycosylation inhibition does take place.
Previously, we have demonstrated that FXYD5 affects Na+-K+-ATPase kinetics and increases the pump's Vmax by about twofold (18). The next set of experiments was aimed at assessing whether the effect on paracellular TER is secondary to a higher pumping rate and presumably lower cell Na+. This was done by monitoring the paracellular resistance developed in monolayers cultivated in the continuous presence of amiloride. The rationale is that blocking Na+ entry into the cells should largely reduce cell Na+ and slow down Na+ pumping to its minimal rate ≪Vmax. Under these conditions, no difference in the Na+-K+-ATPase turnover rate is expected between cells that do or do not express FXYD5. As seen in Table 2, cultivating cells in the continuous presence of amiloride did not influence the effect of FXYD5 on TER, indicating that this effect is not secondary to the increased Vmax. We have also verified that the presence of amiloride in the culture medium did not lower surface expression of the Na+-K+-ATPase (data not shown).
The effect of FXYD5 on paracellular permeability was further confirmed by comparing transepithelial fluxes of three cell-impermeable fluorescent compounds: rhodamine B (molecular weight 536), 4-kDa FITC-dextran, and 70-kDa FITC-dextran. All three compounds permeated empty filters equally quickly and reached equilibrium within 2 h. They permeated cell monolayers at size-dependent rates, and their uptake was linear for at least 4 h (Fig. 6A). Comparing fluxes across wild-type and FXYD5-transfected cell monolayers demonstrated substantial but not dramatic differences; i.e., the FXYD5-expressing monolayers are about twofold more permeable to 4- and 70-kDa dextran than wild-type cultures. However, they still provide an effective barrier for large macromolecules and discriminate well among rhodamine B, 4- kDa, and 70-kDa dextran (Fig. 6B).
The effects of FXYD5 on cell-cell junctions were further assessed by staining confluent cell monolayers for the TJ and adherence junction (AJ) markers ZO-1 and β-catenin, respectively. In both cases, marked differences were noted between the wild-type and FXYD5-expressing cells. In the nontransfected culture, both ZO-1 and β-catenin were confined to the plasma membrane and nicely outlined cell-cell contacts, characteristic of mature TJ and AJ (Figs. 7 and 8). In the FXYD5-expressing cells on the other hand, ZO-1 was mostly intracellular, while β-catenin, although still confined to the plasma membrane, appeared perpendicular to the membrane plane. This type of organization reflects a state of nascent nonmature AJ (puncta) (17). Occludin, another transmembrane protein associated with the TJ complex, was affected as well (39). Its total and cell surface abundance were lower in the FXYD5-transfected cells by 70 and 34%, respectively (Fig. 9). In addition, in phase-contrast microscopy, the FXYD5-expressing cells were more spread shaped than the nontransfected cells (Fig. 7).
Finally, we have examined the appearance of cell-cell junctions by electron microscopy. High-magnification images shown in Fig. 10 demonstrate that expressing FXYD5 evokes large expansions of interstitial spaces just under the TJ (asterisks in Fig. 10). Interestingly, similar dilations were reported in small intestine following glucose-induced permeation of TJ and have been associated with higher intercellular water and solute permeability (20, 25). In addition, TJ in FXYD5-expressing cells appeared somewhat wider and had less electron-dense material. Thus both staining junctional markers and electron microscopic images demonstrate the effects of FXYD5 on the structure of cell-cell junctions, which presumably lead to higher paracellular permeability.
The current study examines the effects of FXYD5 on cell-cell contacts in a tight epithelium cell line. It was found that expressing FXYD5 in M1 cells decreases paracellular electrical resistance and increases its permeability to macromolecules. Cell-cell junctions of the transfected cells are wider, and impairment of TJ and AJ is also manifested by the cellular distribution or abundance of ZO-1, occludin, and β-catenin. The decrease in TER, increased permeability to dextran, and redistribution of ZO-1 demonstrate that TJ are affected, while the electron microscopy data and β-catenin staining suggest impairment of AJ. These observations, however, do not necessarily mean that both structures are independently affected by FXYD5. It is well established that cell-cell contacts are initiated by forming AJ, and in their absence TJ do not form properly (12). Thus AJ could in principle be the only site of FXYD5 action.
While we cannot exclude Na+-K+-ATPase-independent effects of FXYD5, the simplest assumption is that the phenomena reported in this study are secondary to its interaction with the Na+-K+-ATPase. However, this is not likely due to its effect on pump kinetics, since the same increase in paracellular permeability was apparent at very different pumping rates. A likely mechanism is that impairment of cell-cell contacts is due to FXYD5's effect on the structure of the Na+-K+-ATPase and possibly on the decrease in β glycosylation. Several studies have provided evidence that β-Na+-K+-ATPase functions as a cell adhesion molecule (11, 34, 40). Such a role is also supported by the finding that the structure of the β1 ecto domain resembles an immunoglobulin-like fold, typical of cell adhesion molecules (1). It has been demonstrated that interactions between β1-subunits in neighboring MDCK cells participate in cell-cell contacts and affect paracellular permeability (40, 41). This interaction is mediated by the carbohydrate moieties on β1 and is largely impaired by inhibiting N-glycosylation of β. Thus a decrease in β glycosylation may account for the FXYD5-induced impairment of cell-cell contacts characterized in the current study. The decrease in glycosylation may be due to specific interactions between the ecto domains of β and FXYD5, which limit β accessibility to glycosylation in the ER or Golgi. Specific interactions between the FXYD motif of FXYD10 and β1 are apparent from the three-dimensional structure of the shark Na+-K+-ATPase and likely to be relevant to other FXYD proteins as well (33). Also, we found that surface biotinylation of FXYD5 with N-hydroxysuccinimide ester derivatives is very inefficient even though the extracellular domain has seven lysine residues. This could be due to a strong interaction of FXYD5 with other proteins and limited accessibility of the extracellular lysines to biotinylation. A rather intriguing observation is the significant downregulation of α1 by the expression of FXYD5 depicted in Fig. 2 and Table 1. A similar observation has been made before in X. laevis oocytes injected with FXYD5 cRNA (18). It may reflect some interference by FXYD5 of the α-β association and thereby α destabilization. However, since surface expression of the Na+-K+-ATPase was hardly affected by the decrease in α1 expression, this effect is unlikely to contribute to the decrease in cell-cell contacts.
Other mechanisms by which interactions between FXYD5 and the Na+-K+-ATPase affect paracellular permeability through modified cytoskeletal organization are possible as well. Rajasekaran and coworkers (2, 16, 27) have reported that binding of the p85 subunit of phosphatidylinositol 3-kinase (PI3K) to the cytoplasmic C tail of α-Na+-K+-ATPase activates PI3K and produces phosphatidylinositol 3,4,5-trisphosphate (PIP3). Annexin II interacts with PIP3 and the pump's β-subunit to recruit Rac1 and promote lamellipodia formation. Independently, the Na+-K+-ATPase was shown to associate with protein phosphatase-2A (PP2A), and its inhibition resulted in hyperphosphorylation of occludin and paracellular permeabilization (29). Thus association of FXYD5 with the Na+-K+-ATPase may influence a number of cellular processes, affecting cytoskeletal organization and cell-cell contacts, independently of its enzymatic activity.
FXYD5 has been originally cloned as an mRNA induced in NIH 3T3 cells by the oncoprotein E2a-Pbx1 (7). The human ortholog was identified as a cancer-associated membrane protein whose expression inhibits E-cadherin and promotes metastasis (13). In a number of clinical studies, a correlation was established between its abundance and the progression and survival chances of various human cancers (for a review, see Ref. 24). Weakening of cell-cell contacts is certainly in agreement with promotion of metastasis and the above clinical correlations. Interestingly, effects on paracellular resistance were also reported for FXYD3 (Mat-8), another family member overexpressed in cancer cells. Silencing endogenous FXYD3 in Caco-2 cells lowered their transepithelial resistance by twofold (4). While this effect is in the opposite direction of those described in the current study, it may reflect a more general role of FXYD proteins in cell-cell contact.
In previously reported cancer-related studies, FXYD5 migrated as a 50- to 55-kDa polypeptide, much heavier than the <20 kDa predicted by its amino acid sequence. Since the ecto domain of FXYD5 is enriched in serines, threonines, and prolines, and protein expression was downregulated by benzyl-α-GalNAc, it was suggested that FXYD5 in tumor cells is highly O-glycosylated (13, 38). In transfected M1 cells, however, FXYD5 migrated as an ∼24-kDa polypeptide, close to the calculated molecular weight. Taken together with the lack of effect of benzyl-α-GalNAc on TER, the data suggest that excessive O-glycosylation of FXYD5 may be characteristic of the metastatic state but is not essential for the weakening of cell-cell contacts. Interestingly, in H1299 cells a much larger polypeptide of ∼40 kDa was apparent. A similar difference in size between the human and mouse proteins has also been reported (21). It cannot be accounted for by sequence differences and may reflect a difference in O-glycosylation of the human and mouse protein or a difference between cells derived from normal (M1) vs. tumor (H1299) tissues.
Another difference from previous data is the fact that we were unable to detect FXYD5-dependent downregulation of either total or surface expressed E-cadherin as reported (13). However, a second E-cadherin-independent mechanism by which FXYD5 may affect invasion and metastasis has been described (23). This mechanism involves upregulation of the chemokine ligand CCL2, which in turn exerts autocrine and paracrine tumor-promoting effects. Interestingly, CCL2 was found to induce disassembly of the TJ complex in the blood-brain barrier by triggering caveolae-dependent internalization of transmembrane TJ proteins (35).
Since FXYD5 is expressed not only during metastasis but also in a variety of normal epithelia (18, 19), one may wonder whether the observed effects on TJ and AJ serve a particular physiological role. One possibility is that this is an integral component of “leaky” low-resistance epithelia. Support for this notion is the fact that in the gastrointestinal tract FXYD5 is relatively abundant in the low-resistance small intestine (duodenum, jejunum, and ileum) and much less in tight epithelium like the distal colon (18). However, FXYD5 was also detected in tight epithelia such as the kidney collecting duct and lung. However, in the collecting duct it is expressed only in intercalated cells and not in the Na+-transporting principal cells, while its cellular distribution along the airway epithelia has not yet been determined (18).
Another option is that FXYD5 participates in a mechanism responsible for a transient permeation of the paracellular pathway. At least two such mechanisms have been described. One is enhancement of paracellular permeability of small intestine to increase nutrient absorption, triggered by high luminal glucose or alanine (39). This process is characterized by large dilations of the interstitial space, very similar to those shown in Fig. 10 (20, 25). A second option is the cytokines induced permeabilization of TJ to allow egression of leukocytes from the interstitial space to the lumen in intestinal, pulmonary, and renal epithelia (6, 39). Finally, FXYD5 may be involved in epithelial remodeling and recovery from injury (15). This option is supported by the fact that its expression also induces the epithelial-mesenchymal transition marker vimentin (21). Obviously, all putative roles will also involve regulation of the expression or activity of FXYD5. Testing these options and elucidating the physiological role of FXYD5 await further studies.
This study was supported by a research grant from the Israel Science Foundation to H. Garty.
No conflicts of interest, financial or otherwise, are declared by the authors.
I. L. and H. G. provided conception and design of research; I. L. and C. A. performed experiments; I. L, C. A, and H. G. analyzed data; I. L, C. A, and H. G. interpreted results of experiments; I. L. and C. A. prepared figures; C. A. and H. G. edited and revised manuscript; H. G. drafted manuscript; H. G. approved final version of manuscript.
The authors thank Prof. Benjamin Geiger from the Department of Molecular Cell Biology, The Weizmann Institute of Science, for useful discussions and Dr. Vera Shinder from the Moskowitz Center, The Weizmann Institute of Science, for electron microscope images. H. Garty is the incumbent of the Hella and Derrick Kleeman Chair of Biochemistry.
- Copyright © 2011 the American Physiological Society