Ouabain Enhances Cell-Cell Adhesion Mediated by beta 1 Subunits of the Na ( + ) , K ( + )-ATPase in CHO Fibroblasts

Adhesion is a crucial characteristic of epithelial cells to form barriers to pathogens and toxic substances from the environment. Epithelial cells attach to each other using intercellular junctions on the lateral membrane, including tight and adherent junctions, as well as the Na+,K+-ATPase. Our group has shown that non-adherent chinese hamster ovary (CHO) cells transfected with the canine β1 subunit become adhesive, and those homotypic interactions amongst β1 subunits of the Na+,K+-ATPase occur between neighboring epithelial cells. Ouabain, a cardiotonic steroid, binds to the α subunit of the Na+,K+-ATPase, inhibits the pump activity and induces the detachment of epithelial cells when used at concentrations above 300 nM. At nanomolar non-inhibiting concentrations, ouabain affects the adhesive properties of epithelial cells by inducing the expression of cell adhesion molecules through the activation of signaling pathways associated with the α subunit. In this study, we investigated whether the adhesion between β1 subunits was also affected by ouabain. We used CHO fibroblasts stably expressing the β1 subunit of the Na+,K+-ATPase (CHO β1), and studied the effect of ouabain on cell adhesion. Aggregation assays showed that ouabain increased the adhesion between CHO β1 cells. Immunofluorescence and biotinylation assays showed that ouabain (50 nM) increases the expression of the β1 subunit of the Na+,K+-ATPase at the cell membrane. We also examined the effect of ouabain on the activation of signaling pathways in CHO β1 cells, and their subsequent effect on cell adhesion. We found that cSrc is activated by ouabain and, therefore, that it likely regulates the adhesive properties of CHO β1 cells. Collectively, our findings suggest that the β1 subunit adhesion is modulated by the expression levels of the Na+,K+-ATPase at the plasma membrane, which is regulated by ouabain.


Introduction
The Na + ,K + -ATPase or sodium pump is an ubiquitous plasma membrane transporter that creates the ionic gradients that drive the net movement of glucose, amino acids, and ions across cellular membranes [1]. The Na + ,K + -ATPase belongs to the P-type ATPase family, whose members are characterized by the transitory formation of a phosphorylated enzyme intermediate [2]. The pump structurally consists of three subunits: a catalytic α subunit, an accessory β subunit and a regulatory γ subunit. The catalytic α subunit is constituted by 10 transmembrane domains (TM), and exchanges 3 Na + ions from the cytosol for 2 K + ions from the extracellular milieu using the energy released from ATP hydrolysis [3]. The β subunit is constituted by a single TM domain, and a long glycosylated extracellular domain.
Its functions are discussed in detail below. The γ subunit is a small, single span TM protein belonging to the FXYD family, differentially expressed in tissues, which modulates the pump's function [4,5]. In mammals, there are four α subunit isoforms, three β subunit isoforms and seven FXYD members [6,7].
The β subunit of the sodium pump has different functions that depend on the type of isoform expressed (β 1, β 2 or β 3 ), and on the accompanying isoform of α subunit (α 1 -α 4 ) [8]. The main and essential function is to chaperone for the α subunit [9] contributing to the assembly and arrival of the pump to the plasma membrane [10]. In addition, the β subunit undergoes conformational changes during the catalytic cycle [11]. Different β isoforms have been associated to different K + affinities [12]. Furthermore, some studies suggest that the β 1 subunit regulates cell polarity, cell motility, epithelial to mesenchymal transition, and oncogenic transformation [13][14][15]. In epithelia, the β 1 isoform functions as a homophylic cell adhesion molecule [16,17]. Moreover, the β 2 isoform is an adhesion molecule on glia (AMOG, [18]).
Emerging evidence showed that the Na + ,K + -ATPase may have additional regulatory functions other than pumping ions across cell membranes. Ouabain and related cardiotonic steroids are highly specific Na + ,K + -ATPase ligands that bind to all catalytic α isoforms [19][20][21]. Studies from various laboratories have documented an important signaling function of the Na + ,K + -ATPase [8,22]. In epithelia, the sodium pump also acts as a membrane receptor that transduces signals in response to ouabain and related cardiotonic steroids. The binding of ouabain and cardiotonic steroids (at nM concentrations) to the sodium pump activates signaling pathways that resemble those triggered by hormone/receptor interactions, which regulate gene expression, membrane trafficking, cell adhesion, proliferation, and cell death [23][24][25][26][27][28]. Interestingly, nM concentrations of ouabain neither inhibit K + pumping nor disturb the K + balance of the cell [29]. Therefore it was proposed that low ouabain concentrations bind and activate a non-pumping population of the Na + ,K + -ATPase [30,31].
Ouabain was suggested to be a hormone when Hamlyn and Mathews demonstrated the presence of a substance in plasma similar to ouabain of vegetal origin [20,32]. Thereafter, it was demonstrated that endogenous ouabain is synthesized and secreted by the hypothalamus [33,34], and the adrenocortical gland [35][36][37]. The status of ouabain as a hormone was clearly recognized upon the discovery of increased concentrations in plasma during exercise, salty meals ingestion, and pathological conditions such as arterial hypertension and myocardial infarction [38][39][40][41][42][43]. However, its physiological role remained unknown. Work from our group has shown that ouabain binding to the Na + ,K + -ATPase modulates epithelial cell adhesion and communication [29,44,45].
Our laboratory has studied the role of the Na + ,K + -ATPase β 1 subunit in epithelia.
We demonstrated that the β 1 -subunits of Na + ,K + -ATPases on neighboring cells interact with each other in a species specific manner [46,16]. Numerous studies demonstrated that the intercellular homotypic interaction between β 1 subunits of the Na + ,K + -ATPase are important for the stability of adherent junctions (AJ) and integrity of the tight junctions (TJ) and GAP junctions [17,45,47]. Thus, β 1 -β 1 interactions between epithelial cells are critical for the integrity of intercellular junctions. Since ouabain modulates different cell-attachment complexes, we wondered whether ouabain also regulates the β 1 -β 1 mediated cell adhesion. In this work we used CHO fibroblasts, which lack the classical cell-cell adhesion complexes (TJs, AJ), and that stably express a canine β 1 subunit in the plasma membrane. This model system targets efficiently the Na + ,K + -ATPase to the membrane contributing to the cell-cell contact [46]. To determine whether β 1 -β 1 interactions are modulated by ouabain, we investigated the effect of a physiological concentration (50 nM) of ouabain on the adhesion of CHO cells overexpressing the dog β 1 subunit. This work shows that ouabain increases the amount of Na + ,K + -ATPase at the cell membrane, rendering increased cell adhesion properties mediated by β 1 -β 1 interactions. This effect is facilitated by the ouabain-dependent activation of kinases such as cSrc, ERK1/2 and AKT which contribute to enhance the adhesive properties of CHOβ 1 cells. Adherent CHO fibroblasts are well attached to the extracellular matrix and the   substrate, but establish weak cell-cell contacts, which are easily disrupted by gentle shaking or pipetting [46,52,53]. We have shown that CHO fibroblasts transfected with the canine β 1 subunit of the Na + ,K + -ATPase (CHOβ 1 ) form large cellular aggregates, due to an increase in cell-cell adhesion [46]. Moreover, we demonstrated that the epithelial β 1 subunit of Na + ,K + -ATPase is an adhesion molecule that mediates the interaction of sodium pumps on neighboring cells by establishing homotypic interactions [16]. Therefore, we hypothesized that the cell-cell adhesion observed in CHOβ 1 cells is mediated by β 1-β 1 interactions. To To confirm the hypothesis that the cell-cell adhesion observed in CHOβ 1 cells is due to β 1-β 1 interactions, we tested whether the soluble domain of the β 1 subunit would impair the formation of cellular aggregates in this cell line. We took advantage of a truncated version of the canine β 1 subunit that only expresses the soluble extracellular C-terminal domain (Sec β 1 ) [16,51]. CHOβ 1 cells were allowed to interact with supernatants obtained from CHO Secβ 1 cells containing this protein, and the formation of cellular aggregates was analyzed by light microscopy. Figure   1D shows that the presence of the soluble domain of the dog β 1 subunit (Secβ 1 ) reduced the size of the CHOβ 1 cellular aggregates. Statistical analyses confirmed that the aggregates formed by CHOβ 1 cells were significantly smaller (~50%) than control cells (Fig. 1E). Interestingly, confocal microscopy analyses showed that CHOβ 1 cells pre-incubated for 24 h with Secβ 1 presented a proliferation defect compared to control cells (Fig. 1D, lower panel). These results confirmed that cell-cell adhesion dependent in the Na + ,K + -ATPase is at least partially due to an interaction between β 1 subunits, and further showed that the cell culture model based on CHOβ1 cells is suitable for studying β 1 -β 1 interactions.
Therefore, we hypothesized that ouabain may also modulate the cell-cell upper and lower panels). Importantly, the inhibitory ouabain concentration (100 μM) prevented the cell adhesion phenotype observed in CHOβ 1 cells, to a level similar to wild type cells. Therefore, we concluded that physiological concentrations of ouabain increase cell adhesion mediated by β 1 -β 1 interactions of CHOβ 1 cells.

The interactions of β 1 -β 1 subunits are stable in vitro independently of ouabain treatment
The effect of ouabain on cell-cell interactions mediated by β 1 subunits could be explained if this hormone induces a conformational change in β 1 subunits resulting in a more adhesive molecule. Accordingly, ouabain binding to its receptor, the α subunit of the pump, should be sufficient for inducing the same effect on β 1 -β 1 interactions in vitro, out of the cellular context. Therefore, we studied whether ouabain directly affects β 1 -β 1 subunits interaction in a pool down assay. In this case, we used the canine β 1 tagged with a hexa-histidine repeat (CHOβ 1 His 6 ) immobilized on Ni + -NTA as the bait. The prey was obtained from total cellular extract of CHOβ 1 cells tagged with the yellow fluorescent protein (CHOβ-YFP). CHO wild type cells (CHO WT) were used as negative control. Cellular extracts (bait and prey) were allowed to interact in absence or presence of ouabain, and the formation of complexes interacting β 1 subunit were analyzed by Western blot. Figure 3 shows that the immobilized CHOβ 1 His 6 was capable to interact with the recombinant β 1 YFP obtained from cellular extracts. Importantly, the interaction in vitro was maintained even in the presence of ouabain (Fig. 3A). Statistical analyses demonstrated that ouabain treatment did not affect significantly the amount of interacting proteins (Fig. 3B). All eluted fractions contain α subunit, which means that the β 1 His 6 is assembled with the α subunit on the Ni + -NTA. The data suggest that the effect of ouabain on cell-adhesion does not occur directly on β 1 subunits, and it is likely dependent on additional cellular components.

Ouabain increases the expression and localization of the sodium pump at the plasma membrane
It has been shown that treatment with cardiotonic steroids (ouabain and 21-benzylidene digoxin) increases the expression level of the α subunit of sodium pump in the typical pig kidney tubular epithelium cell line LLC-PK1 and MDCK [54,55]. Therefore, we wondered whether the cell adhesion effect of ouabain induced in  ; [56]). Considering these phenotypes, we evaluated the effect of ouabain on the localization of the β 1 subunit in the monolayer of CHOβ 1 cells. Confocal microscopy analyses showed that in our model system, the β 1 subunit presents a similar distribution to epithelial cells upon ouabain treatments (Fig. 4B). The membranal localization of the sodium pump was disrupted at higher concentrations of ouabain (µM) (Fig. 4B). Importantly, the fluorescence intensity of the β 1 subunit at the plasma membrane only increased significantly upon treatment with 50 nM ouabain (Fig. 4C).
To confirm that physiological concentrations of ouabain increase the sodium pump expression at the plasma membrane, we used a surface biotinylation assay.

The β 1 -β 1 adhesion induced by ouabain depends on the activation of cSrc and AKT signaling pathway
Previous studies demonstrated that binding of cardiotonic steroids such as ouabain to the sodium pump stimulates multiple kinase cascades [57]. Therefore, the effect of ouabain on the β 1 -β 1 interaction could be explained by the potential activation of some components from the signaling machinery that is known to modulate cell junctions such as TJs, AJs and GAP junctions [29,58]. We first screened for signaling proteins of the family of tyrosine kinases that were differentially activated compering between CHOβ 1 and CHO wild type cells. Overall, CHOβ 1 cells had more active signaling proteins than wild type cells, indicating that the increased expression of Na + ,K + -ATPase in CHO fibroblast per se induces the activation of signaling cascades that are usually of low activity ( Figure 1). Then, we treated CHOβ 1 cells with ouabain and identified various kinases that were altered when treating the cells with 50 nM Ouabain ( Figure S2). Of all modified pathways detected, we focused on the potential mechanisms by which cSrc, AKT, and ERK1/2 contribute to the interaction between β 1 subunits. Therefore, we investigated whether the inhibition of those pathways would impair the effect of ouabain on cell adhesion mediated by β 1 subunits. The selected phosphorylation inhibitors were Dasatinib for cSrc, Perifosine for AKT, and U01266 for ERK1/2. From the three kinases selected, phosphorylated cSrc increased upon treatment with ouabain compared to control cells (Fig. 5A). Importantly, incubation of the CHOβ 1 cells with all the inhibitors was effective and prevented the corresponding kinase phosphorylation (Fig. 5A-C).
To test the effect of kinase inhibition in the adhesive properties of CHOβ 1 cells dependent on ouabain (50 nM), we analyzed the size of the cellular aggregates.
Representative light microscopy imaging showed that inhibition of cSrc (Fig. 5D), and AKT (Fig. 5E) impaired the formation of the ouabain-induced aggregates. No changes on the size of the cellular aggregates of CHOβ 1 cells that were inhibited for ERK1/2 were observed (Fig. 5F). Consistently, statistical analyses showed significant differences in the size of the aggregates of CHOβ 1 cells treated with either perifosine or dasatinib; no significant differences were observed for CHOβ 1 cells treated with UO126 when compared to control cells (Fig. 5G). These data suggest that ouabain binding to the sodium pump stimulates at least cSrc kinase and AKT signaling pathways, which may be involved in the regulation of β 1 -β 1 mediated cell adhesion.

pNaKtide reduces cell adhesion effect induced by ouabain
The Na + ,K + -ATPase interacts with cSrc kinase and forms a complex that serves as a receptor for ouabain to stimulate various protein kinase cascades [30]. Specifically, ouabain binding to the Na + ,K + -ATPase disrupts this interaction and results in the assembly and activation of different signaling pathways [59]. Our results thus far have shown that the adhesive properties of β 1 subunits require the cellular context to occur, since no effect of ouabain was observed in vitro (Fig. 3). Moreover, the increased phosphorylation of cSrc in CHOβ 1 cells treated with ouabain (Fig. 5A), was reversed by the cSrc kinase inhibitor, and prevented formation of cellular aggregates (Fig. 5D). Therefore, we hypothesized that the β 1 subunit adhesive properties dependent in ouabain, could be partially due to the activation of Na + ,K + -ATPase/cSrc complex. To test this hypothesis we took advantage of the pNaKtide, which is a peptide that antagonizes the effect of ouabain on cSrc [60], and assayed for cell-adhesion in the presence of ouabain and with or without the pNaKtide. Figure 6A shows a representative Western blot of the ouabain-dependent increase in the phosphorylation of cSrc (~70% higher than control cells). In the presence of 1 μM of the pNaKtide, the effect on cSrc phosphorylation was abolished, and remained at similar levels than control cells. No changes were detected in the expression of the α subunit of the sodium pump. Finally, we analyzed the effect of the antagonist in the formation of ouabain-dependent aggregates of CHOβ 1 cells. Representative light microscopy images showed that the size of the aggregates is smaller compared to those of ouabain stimulated cells (Fig. 6B). Statistical analyses showed that in the presence of the pNaKtide the size of ouabain-stimulated aggregates is similar in size than those of control cells (Fig. 6C). These results support our hypothesis that the adhesion effect of β 1 subunits stimulated by ouabain is partially regulated by cSrc activation.

Discussion
As all mammalian cells, CHO fibroblasts express the necessary amount of sodium pumps at their plasma membrane for maintaining a suitable ionic balance that would permit their viability. In our overexpression system, the assembled α 1 β 1 dimer in CHOβ 1 cells is targeted to the plasma membrane and concentrated at cell-cell contacts adopting an epithelial-like phenotype and maintain both cell-cell and cell-substrate adhesion [46]. This phenotype makes the CHOβ 1 overexpression system an ideal model to investigate cell adhesion. Using cellular and biochemical strategies our group demonstrated that β 1 subunits of neighboring cells can interact directly in a species-specific manner [16]. In the present work we show that increasing the density of the β 1 subunit of Na + -pump in cell contacts of CHO fibroblasts is sufficient for displaying a cell-adhesive phenotype. These adhesive properties are mediated by β 1 interactions which are regulated by signaling cascades conserved in establishment and maintenance of classical epithelial adhesion complexes [26,29,45].
Early works from Nelson´s group on epithelial cell polarity have shown that expression of E-cadherin in L-fibroblasts, which lack surface polarity resulted in membrane domains with apical and basolateral identities [61]. Remarkably, the C Figure 6 A B endogenous Na + -pumps in those experiments were recruited to cell-cell contacts. Our results show that CHO fibroblasts transfected with E-cadherin, a Ca 2+ -dependent cell adhesion molecule, or with the β 1 subunit, a Ca 2+ -independent adhesion molecule, display similar aggregation properties. To validate that cell-cell adhesion of CHOβ 1 cells was in fact due to β 1 -β 1 interactions, cellular aggregation was challenged with Secβ 1 , the soluble extra-cellular domain of the canine β 1 subunit. As expected, this truncated version of the β 1 prevented the ouabain induced cell-cell adhesion properties of CHOβ 1 cells. These results strongly suggest that secβ 1 competes for cell-cell adhesion mediated by β 1 subunits on neighboring cells. And that ouabain-induced cell-cell adhesion is indeed through β 1 -β 1 interactions. Remarkably, Secβ 1 blocks cell-cell aggregation between CHOβ 1 cells only when it is added short time after seeding, before the formation of a confluent monolayer. This observation indicates that Secβ 1 is not able to compete with already formed β 1 -β 1 interactions but is able to interact with non-occupied β 1 subunit and prevent cell-cell adhesion mediated by β 1 subunits. Moreover, CHO β 1 cells pre-incubated with secβ 1 do not reach confluence and are apparently not proliferating. The cells form small patches and many single cells are observed. Usually, contact naïve epithelial cells do not express the Na + -pump at the plasma membrane [62]. Importantly, MDCK cells co-cultured with CHO wild type cells do not express the Na + ,K + -ATPase at the heterotypic contacting membrane [46]. Nevertheless, some of the CHOβ 1 single cells showed a highly fluorescent signal at the plasma membrane, which corresponded to the β 1 subunit. Therefore, it is plausible that these contact naïve CHOβ 1 cells, are actually surrounded by soluble Secβ 1 molecules associated with the membrane-bound β 1 subunit and that it is mimicking cell-cell contacts. As such, we hypothesize that the apparent proliferation defect is due to a contact inhibition effect induced by Secβ 1 . These data open a new paradigm on the modulation of cell proliferation dependent on the sodium pump. Altogether, these results confirm that cell-cell adhesion of CHO fibroblasts that overexpress Na + ,K + -ATPase is mainly based on β 1 -β 1 interactions between neighboring cells.
Ouabain binding to α subunit of the epithelial Na + ,K + -ATPase has a dual effect on the pump, high concentrations of ouabain (>300 nM) triggers pump inhibition and cell detachment [44,56]. Low concentrations of ouabain increases sodium pump activity [63,64] and stimulate signaling routs in a cellular context dependent manner [65]. In epithelial MDCK cells, low ouabain concentrations (10-100 nM) modulate cell-cell contacts such as TJs, AJs and GAP Junctions [26,29,44,45]. We analyzed the effect of ouabain on β 1 -β 1 mediated cell adhesion. Our results show that CHOβ 1 cells incubated with low concentration of ouabain (50 nM) make bigger aggregates than the non-treated ones, while incubation with high concentration of ouabain (100 µM) does not enhance cell-cell adhesion. Thus we conclude that physiological concentrations of ouabain induce an increase in β 1 -β 1 interaction in CHOβ 1 cells.
Then we asked whether ouabain has a direct or indirect effect on the β 1 -β 1 interaction. Ouabain binding to the α subunit could induce a localized conformational change that would make the β 1 subunit more adhesive; which then could be considered as a direct effect. For instance, fluorometry assays identified three positions on the β 1 subunit, that indicate that α and β subunits move toward each other during conformational transition, and produce a conformal rearrangement [66,67]. Additionally, ouabain binding produce conformational rearrangements in α subunit of the sodium pump [68]. To further support this idea, our in vitro experiments showed that although the α subunit, is present during the immobilization on Ni-NTA and pool down assays, ouabain treatment does not modify β 1 -β 1 interaction. As such, our results suggest that ouabain does not modify the adhesion between β 1 subunits directly and that it requires additional cellular components to fulfil this function. In this regard, binding of ouabain to α subunit could activate signaling cascades that would up-regulate the amount of Na + ,K + -ATPase exposed to the extracellular space and thus increase indirectly cell-cell adhesion mediated by β 1 subunits. During last years it was established that the Na + ,K + -ATPase is not just a pump [31,69,70]. Experimental evidence strongly indicate that epithelial cells contain at least two populations of Na + ,K + -ATPases: A pumping and a signaling (non-pumping) one [23,59,70]; which could also be the case for transfected CHOβ 1 fibroblasts. Overexpression of β 1 subunit up regulates the expression of α subunit [31,71]. Both are assembled in endoplasmic reticulum and targeted to the plasma membrane [72]. Consistent with this idea, confocal microscopy and surface biotinylation analyses performed in this study show that ouabain stimulates the expression and delivery of the pump to the plasma membrane. Apparently, the signaling population is increased because we observed that even without ouabain the basal expression of cSrc, AKT, and ERK1/2 is increased in comparison to non-transfected CHO cells (Sup 1). Moreover, physiological concentrations of ouabain stimulated the activation of cSrc; furthermore, the pNaKtide which only binds to cSrc associated with Na + ,K + -ATPase, successfully reverted the effect of ouabain on cSrc activation and cell-cell adhesion. In addition, AKT also seems to regulate the ouabain-dependent cell-cell adhesion observed in CHOβ 1 fibroblasts, since the specific inhibitor of AKT abolished the β 1 -mediated cell-cell adhesion. This is not surprising as it has been shown that AKT trans-activates IP3K which forms part of the signaling complex in association with Na + ,K + -ATPase [73,74]. On the other hand, the inhibition of ERK signaling cascade had a minor effect in the ouabain-dependent cellular aggregation. This might be partially explained by the fact that ERK1/2 is not the only signaling effector activated by cSrc [75,76]. Moreover, our initial screening revealed various signaling proteins that were not studied here and that are probably involved such as TrkB, EphB4 and PDGFR. Future studies will be directed to understand the participation of these signaling molecules in the β 1 -β 1 adhesion phenotype.
All together our results suggest that overexpression of β 1 subunit in CHO cells increases the population of Na + ,K + -ATPases at the plasma membrane. This pool of pumps is most probably involved in mediating β 1 -β 1 interactions between neighboring cells. Whether these proteins have an active role in ion transport remains to be elucidated. However, a plausible model would involve the activation of cSrc and AKT upon stimulation of CHOβ 1 cells with physiological concentrations of ouabain. These signaling cascades would result in the overexpression of adherent pumps targeted to cell contacts at the plasma membrane. Nonetheless, an open question that would have to be addressed in the future is if CHOβ 1 cells form signalosomes, as epithelial cells do and if β 1 -β 1 interaction occurs between specific pools (pumping and non-pumping pool) or not.
For experiments including secβ 1 the CHOβ 1 cells were seeded in 24 wells plates and after 90 minutes the monolayer was washed three times with PBS and a serum free medium with or without secβ 1 was added. The next day the medium was replaced with a medium containing ouabain and secβ 1. The kinase inhibitors used were against cSrc (Dasatinib 100nM), AKT (Perifosine 10μM), ERK1/2 (U0126, 10μM) (0952, 14240 and 9903, respectively, Cell Signaling Technologies) and pNaKtide (1μM, kind gift of Dr. Z. Xie, Marshall Institute for Interdisciplinary Research). Inhibitors were added 1 h before ouabain treatment.

Immunofluorescence and confocal microscopy analyses
CHO and MDCK cells grown on coverslips at the indicated conditions in the figures, were immunostained as previously described [46]. Briefly, the cells were washed with PBS, fixed and permeabilized with ice-cold methanol for 5 min. After washing with PBS, the cells were blocked with 3% bovine serum albumin for 1 h, followed by 1 h incubation with a mouse primary antibody against the β 1 subunit of the Na + ,K + -ATPase (1:50 dilution) at room temperature. Cells were washed 10 times quickly with PBS, and incubated with a goat anti-mouse Alexa 488 secondary antibody (1:500 dilution) for 30 min at room temperature. After washing, the cells were mounted on glass slides with FluoroGuard antifade reagent (170-3140, BioRad). Samples were imaged using a confocal microscope (Leica TCS SP8), and visualized using the Leica Lite software. The relative fluorescence intensity was quantified using Fiji 1.0 software.

Cell surface Biotinylation
Confluent monolayers of CHOβ 1 cells were depleted of serum for 24 h and then incubated with or without 50 nM ouabain for 24 h. Then, the monolayers were washed 3 times with PBS, and incubated with 1 mg/ml of EZ-Link Sulfo-NHS-SS-Biotin (21331, Thermo scientific) for 30 min. Subsequently the monolayers were washed 3 times with PBS containing 100 mM glycine to quench the excess of Biotin, followed by a final wash with PBS containing 1% Triton X-100 and protease inhibitors. After 30 min, the cells were scraped, and the cell lysate collected into a 1.5 ml microcentrifuge tubes. The extract was homogenized by passing it 10 times through insulin syringe and centrifuged for 10 min at 9,000 x g at 4 ºC. The supernatant was recovered, and the protein content was measured using the BCA protein assay method. The biotinylated extracts were incubated overnight at 4 ºC with 100 μl of streptavidin-agarose suspension (S1638, Sigma). The next day, bead-adherent complexes were washed 5 times with PBS, and finally the proteins were eluted in 2X Laemmli buffer and boiling for 5 min.

Cell adhesion assay (Dispase assay)
Confluent Monolayers of CHOβ 1 cells seeded on 24 well plates were depleted of serum for 24 h, and then incubated with or without ouabain (50nM) for 24 h. Then the monolayers were washed with ice-cold PBS, and detached from the plates by incubation with PBS without Ca 2+ supplemented with 0.6 U/ml of Dispase I (D4818, Sigma) for 35 min at 37 ºC. Subsequently the Dispase solution was carefully removed using a 200 µl pipette tip, and replaced by 100 μl of PBS. The cells were then mechanically stressed by pipetting up and down 5 times using a 200 µl pipette. The resulting aggregates were visualized by light microscopy using the 10 X and 20 X objectives (Axiovert 200M Fluorescence/Live Cell Imaging, Carl Zeiss). Three independent biological replicates were imaged using the AxioVision 4.8 sofware. The number of aggregates was counted using the Fiji 1.0 software cell counter (aggregates <200μm 2 were excluded from the quantification).

Pull Down assay (PD)
Total extract of CHO cells expressing the β 1 His 6 construct were immobilized with nickel-nitrilotriacetic acid beads (Ni + -NTA, His Trap FF column; GE Healthcare) equilibrated with 10 ml of RIPA buffer containing protease inhibitors. Total protein extracts (500 µg) were loaded into the resin, and allowed to interact for 3 h at 4°C with gentle shaking. Then the unbound protein was washed as indicated by the manufacturer, and the total extract of CHOβ 1 YFP cells incubated in the presence or absence of ouabain (50 nM and 100 µM) were loaded as a prey. Samples were incubated overnight at 4°C, and washed with PBS supplemented with 10 and 20 mM imidazole. Interacting proteins were eluted in PBS containing 500 mM imidazole and were analyzed by western blot.

Statistical analysis
GraphPad Prism version 7.00 software was used for all statistical analyses. The data are presented, as the mean ± SEM or SD as indicated in the figures. Statistical significance was determined using ANOVA one way and Kruskal Wallis test, and t-test for two conditions. P ≤ 0.05 was considered significant.