Ouabain modulates ciliogenesis in epithelial cellsIsabel Larrea, Aida Castilloa, Catalina Flores-Maldonadoa, Ruben G. Contrerasa, Ivan Galvanb, Jesus Muñoz-Estradaa,and Marcelino Cereijidoa,1

aDepartment of Physiology, Biophysics and Neurosciences, and bCentral Laboratories, Center for Research and Advanced Studies of the National PolytechnicInstitute, Mexico City, DF 07300, Mexico

Edited* by Lutz Birnbaumer, National Institute of Environmental Health Sciences, Research Triangle Park, NC, and approved October 27, 2011 (received forreview February 24, 2011)

The exchange of substances between higher organisms and theenvironment occurs across transporting epithelia whose basicfeatures are tight junctions (TJs) that seal the intercellular space,and polarity, which enables cells to transport substances vecto-rially. In a previous study, we demonstrated that 10 nM ouabainmodulates TJs, and we now show that it controls polarity as well.We gauge polarity through the development of a cilium at theapical domain of Madin-Darby canine kidney cells (MDCK, epithe-lial dog kidney). Ouabain accelerates ciliogenesis in an ERK1/2-dependent manner. Claudin-2, a molecule responsible for the Na+

and H2O permeability of the TJs, is also present at the cilium, as itcolocalizes and coprecipitates with acetylated α-tubulin. Ouabainmodulates claudin-2 localization at the cilium through ERK1/2.Comparing wild-type and ouabain-resistant MDCK cells, we showthat ouabain acts through Na+,K+-ATPase. Taken together, ourprevious and present results support the possibility that ouabainconstitutes a hormone that modulates the transporting epithelialphenotype, thereby playing a crucial role in metazoan life.

E-cadherin | occludin | cell adhesion | cardiotonic steroids

The high affinity and specificity of Na+,K+-ATPase for theplant-derived inhibitor ouabain suggested the possibility that

there might exist endogenous analogs. Hamlyn et al. (1) dem-onstrated that there is a substance in plasma that cannot yet bedistinguished from ouabain, even by electrospray ionization-massspectrometry, 1H-NMR, and liquid chromatography (2–4). Fur-thermore, the observation that this endogenous ouabain and itsanalogs increase during exercise (5), salty meals (6), and path-ological conditions [such as arterial hypertension, eclampsia (7),and myocardial infarction (8)] raised the possibility that it mayfunction as a hormone (9). This theory prompted efforts to un-ravel its physiological role (10). The present work stems from ourprevious observations that toxic levels of ouabain invariably af-fect cell-cell and cell-substrate adhesion molecules in Madin-Darby canine kidney (MDCK) cells (11), suggesting that low-doseouabain may be able to modulate cell contacts without causingirreversible damage (12). In keeping with such possibility, we havepreviously shown that ouabain promotes cell-cell contacts as wellas contact-dependent phenomena, such as increases in cell com-munication, the expression of connexin-32 (13), and the molec-ular structure and hermeticity of the tight junction (TJ). Thisprocess occurs by regulating the specific expression and distri-bution of claudin-1, -2, and -4 (12) during differentiation towardthe so-called “epithelial transporting phenotype.”Pursuing our exploration of the effect of low-dose ouabain on

cell contacts, we now study whether 10 nM ouabain can modu-late polarity, one of the two basic cell features of the trans-epithelial transporting phenotype (14, 15). We gauge polaritythrough the expression of a single cilium at the center of theapical domain, a process that relays on the polarized delivery ofmany proteins to this apical compartment (16, 17). At 10 nM,ouabain causes none of its well-known toxic effects; it does notinhibit the unidirectional transport of K+ by Na+,K+-ATPasenor distorts the K+ balance of the cell, and it does not cause celldeath (12, 18).

ResultsOuabain Accelerates Ciliogenesis. MDCK cells display prociliasome 12 h after reaching confluence (Fig. 1A). Procilia pro-gressively lengthen until they become mature cilia. At the thirdday almost all cells have a cilium (Fig. 1B). Ouabain increasesthe length of the cilium (Fig. 1C) but not its thickness (Fig. 1D).We also followed ciliogenesis by staining the cells with an anti-body against acetylated α-tubulin and counting the number ofcells at stages without cilium (Fig. 1E), with procilium (Fig. 1F),and with a mature cilium (Fig. 1G). Fig. 1H summarizes theprogressive increase in the percentage of prociliated and ciliatedcells that tend to reach 100% (open circles). Ouabain acceleratesthe kinetics of procilia and cilia formation (Fig. 1H, red filledcircles), an effect that reaches a 400% increase at 24 h.

Role of Homo- and Heterotypic Cell-Cell Contacts. To study the roleof cell-cell contacts on the effect played by ouabain on cilio-genesis, we plated cells at confluency, in mixed populations of 1%MDCK and 99% NRK cells (epithelial rat kidney). Under thiscondition, most single MDCK cells (Fig. 2A, red) surrounded byNRK cells do not exhibit a cilium (Fig. 2 A and C). Nevertheless,ouabain is able to stimulate cilliogenesis even in these MDCKcells totally surrounded by NRK cells (Fig. 2 B and C).In keeping with our previous observations (19, 20) that the TJ

is a promiscuous structure that can be established by epithelialcells from different organs and even from different animal spe-cies, we found that monolayers of mixed MDCK (Fig. 2, red) andNRK cells develop a transepithelial electrical resistance (TER).This parameter is proportional to the TER in the monolayer ofa single-cell type and their proportion in the mixture (Fig. 2D).Mixed cells also express ZO-2 at the TJs (Fig. 2E). Nagafuchiet al. (21) showed that the expression of E-cadherin requires thatthe two neighboring cells belong to the same animal species(homotypic contact). In accordance with this observation, Fig. 2Fshows that monolayers of pure MDCK (Fig. 2F, red), as well assmall groups of MDCK cells (Fig. 2F, red), express E-cadherin(Fig. 2 F and G, green) at homotypic cell-cell borders. However,this molecule is not observed at heterotypic MDCK/NRK con-tacts (Fig. 2G). Accordingly, a quiescent single MDCK cell (Fig.2H) expresses no E-cadherin at its borders. We conclude that TJsealing and ouabain stimulation of ciliogenesis occur in cells thatestablish homo- and heterotypic borders, even if these express noE-cadherin at the plasma membrane.

Ouabain and Na+,K+-ATPase. Ouabain promotes ciliogenesis whenadded from the basolateral (Fig. 3 A, B, and J, first and secondcolumns) but not from the apical side (Fig. 3C), in accordance

Author contributions: I.L., A.C., C.F.-M., R.G.C., andM.C. designed research; I.L., A.C., C.F.-M.,R.G.C., I.G., and J.M. performed research; I.L., A.C., C.F.-M., R.G.C., and M.C. analyzed data;and I.L., R.G.C., and M.C. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1To whom correspondence should be addressed. E-mail: Cereijido@fisio.cinvestav.mx.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1102617108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1102617108 PNAS | December 20, 2011 | vol. 108 | no. 51 | 20591–20596

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

July

24,

202

1

with the fact that the Na+,K+-ATPase located at the intercel-lular cell borders is only accessible from the basal side (20, 22–24). This finding suggests that this enzyme may act as receptorfor low concentrations of ouabain acting on cilliogenesis. Thispossibility was further tested using ouabain-resistant R-MDCKcells, a stable line produced by Soderberg et al. (25) with very lowaffinity for ouabain. Procilia and cilia in R-MDCK cells are ru-dimentary (Fig. 3 D and E), and their length (Fig. 3F) andthickness (Fig. 3G), as well as the number of ciliated and pro-ciliated cells (Fig. 3 H, I, and J. third and fourth columns), arenot stimulated by ouabain, as observed by scanning electronmicroscopy and immunodetection of acetylated α-tubulin.Zampar et al. (26) have clearly demonstrated that acetylatedα-tubulin binds and inhibits Na+,K+-ATPase in cells of rat brainand COS (monkey kidney fibroblasts), and Menco et al. (27)have found this enzyme in the cilium of olfactory receptor cells.In MDCK cells instead, this enzyme only occupies its well-knownspecific position at lateral homotypic cell borders (20, 23, 24),and does not colocalize with ciliar acetylated α-tubulin (Fig. 3K).Small groups of MDCK cells expressing red fluorescent proteinand surrounded by NRK cells only express Na+,K+-ATPase athomotypic MDCK-MDCK cell borders (Fig. 3L, arrow). Whena single MDCK cell is completely surrounded by NRK cells,Na+,K+-ATPase cannot be observed at the plasma membrane(Fig. 3M).

Cell Signaling. Na+,K+-ATPase forms a receptor complex by as-sociating with signaling molecules, such as c-Src and IP3-re-ceptor, which in turn may activate other proteins, such as ERK1/2. These associations may be modulated by ouabain (28, 29). In

Fig. 1. Ouabain, at a concentration of 10 nM, accelerates ciliogenesis.Scanning electron micrographs of a cilium (between arrows) in a monolayerthat has been confluent for 12 h (A), or 3 d (B) under control conditions. Cilialength (C) and thickness (D) as measured in scanning electron microscopymicrographs of wild-type MDCK cells, under control (white bars) or 24 h ofouabain treatment (red bars); ***P < 0.001. Monolayers of MDCK cellsstained with antiacetylated α-tubulin at zero (E), 24 h (F), or 72 h (G) inconfluence. Ciliogenesis as observed in confluent monolayers under control(open circles) and ouabain (red circles) treatment as a function of time (H).

Fig. 2. Ciliogenesis in proliferation-arrested MDCK cells does not dependon homotypic cell contacts. Stable red color MDCK cells were produced bytransfection of red fluorescent protein. To produce MDCK cells with arrestedproliferation, we plated them at confluence in a 1/99 ratio with NRK cells.Forty-two hours later cells were fixed and stained with the indicated anti-bodies (green) and to detect nucleus with DAPI (blue, NUC). The figureshows a single MDCK completely surrounded by NRK cells, under controlcondition (A), and treated with 10 nM ouabain for the last 30 h (B). Thecilium is stained with a primary antibody against antiacetylated α-tubulin (AcTub) and a fluoresceinated secondary one (green). Statistical analysis ofsingle MDCK cells, either under control (open circles) or ouabain treatmentconditions (red circles) (C). Establishment of tight junctions as revealed bythe value of TER (D) in monolayers of pure MDCK cells (first column), pureouabain-resistant NRK cells (second column), and a 50/50 mixture of both celltypes (third column) cultured for 48 h; ***P < 0.001. In mixed monolayers ofMDCK (red) and NRK cells, the cytoplasmic protein ZO-2 is detected by im-munofluorescence (green) (E). Immunodetection of E-cadherin (green) ina monolayer of pure MDCK cells (3 d old) (F), or in mixed monolayers(48 h old) showing a small group (G) or a single (H) MDCK cell surrounded byNRK cells. (Scale bars, 10 μm.)

20592 | www.pnas.org/cgi/doi/10.1073/pnas.1102617108 Larre et al.

Dow

nloa

ded

by g

uest

on

July

24,

202

1

previous work, we observed that 10 nM ouabain increases thecell content of claudin-1 and -4 through ERK1/2, and c-Srcparticipates in the regulation of claudin-1 but not -4 (12). Thisfinding suggests that in principle ERK1/2 may play a significantrole in ouabain-modulated ciliogenesis. Fig. 4A shows that undercontrol conditions, there are only a few ciliated cells, as observedwith an antiacetylated α-tubulin antibody, and 10 nM ouabainenhances this number (Fig. 4 B and E, first two columns).Twenty-five micromolars of PD98059 (PD), a known inhibitor ofERK1/2, is able to promote ciliogenesis by itself (Fig. 4 C and E,third column), yet it clearly prevents a full effect of ouabain (Fig.4 D and E, fourth column). These results indicate that ouabainregulates the localization of acetylated α-tubulin at the ciliumalthough ERK1/2.

Claudin-2 as a Ciliary Protein. Larre et al. (12) showed that 10 nMouabain modulates the amount and distribution pattern ofclaudin-1, -2, and -4. The staining of claudin-2 and -1 in green(Fig. 5 A, D, G, and J) and acetylated α-tubulin in red (Fig. 5 B,E, H, and K) indicates that claudin-2 is present at the cilium.Claudin-2 and acetylated α-tubulin are observed to colocalize intransversal optical section (Fig. 5 D, E, and F), as well as x/yprojections (Fig. 5 G, H, and I). Colocalization occurs along thewhole cilium; this is specific for claudin-2, as claudin-4 cannot beobserved (Fig. 5 J, K, and L). Claudin-4 and occludin cannot beobserved at the cilium either (Fig. S1). Ouabain increases ciliaryclaudin-2 (Fig. 5M, first two columns), and PD inhibits this effect(Fig. 5M, fourth vs. third columns), indicating that ouabain

induces the delivery of claudin-2 to the cilium through ERK1/2.Ouabain does not modify the content of acetylated α-tubulin(Fig. 5N, open squares), but it does increase the synthesis ofclaudin-2, as first observed by Larre et al. (12) (Fig. 5N, open

Fig. 3. Role of Na+,K+-ATPase in the effect of ouabain.Monolayers of wild MDCK cells, under control (A) and 48-htreatment with ouabain (B and C), added from the baso-lateral (A and B) or the apical side (C). Ouabain-resistantMDCK cells (MDCK-R) display very short cilia (D) that do notbecome larger (E and F) nor thicker (E and G), nor increasethe number of ciliated cells (H vs. I and J) in response to 2 dof incubation with ouabain. Cilia were stained with an an-tibody against acetylated α-tubulin. Statistical analysis ofciliogenesis in wild-type MDCK cells under control (J, firstcolumn) and ouabain (J, second column). Third and fourthcolumns indicate that ouabain-resistant MDCK cells do notincrease the number of ciliated cells when treated withouabain for 48 h (J); ***P < 0.001. In a 4-d old monolayer ofpure MDCK cells (the last 3 d at confluence), Na+,K+-ATPase(green) is localized at all cell borders, and it was never ob-served at the cilium (red) (K). In 1.5-d mixed monolayersof MDCK (red) with NRK (unstained), the enzyme is onlyobserved at homotypic MDCK-MDCK borders (L, arrow).In keeping with this observation, a singleMDCK cell does notexpress the protein because all its borders contact NRK cells(M). (Scale bars, 20 μm, except for D and E, which are 1 μm.)

Fig. 4. Ouabain induces ciliogenesis through ERK1/2. Cilia stained with anantibody against acetylated α-tubulin in a control (A) and a monolayertreated with ouabain for 48 h (B). Results corresponding to experimentssimilar to A and B, but in the presence of PD (25 μM), an inhibitor of ERK1/2(C and D). Statistical analysis of the percentage of ciliated cells under control,ouabain-stimulated, control with PD, and ouabain treatment conditions inthe presence of PD (columns 1–4) (E); ***P < 0.001.

Larre et al. PNAS | December 20, 2011 | vol. 108 | no. 51 | 20593

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

July

24,

202

1

circles). Fig. 5N (gray circles) shows that this last effect does notrequire the activation of ERK1/2. Claudin-2 is associated toacetylated α-tubulin, as both coimmunoprecipitate with acety-lated α-tubulin antibodies (Fig. 5O, lane 2). Ouabain increasesthe extent of this association (Fig. 5O, lane 3). This associationdoes not depend on ERK1/2, as PD does not block it (Fig. 5O,lanes 4 and 5). Interestingly, PD by itself elicits a similar effect asouabain. This coimmunoprecipitation of acetylated α-tubulinand claudin-2 further demonstrates that this TJ molecule is alsoa ciliary protein. Ciliary claudin-2 is not necessary for ciliogenesis,as different clones of MDCK cells that do not express this proteindevelop normal cilia (Fig. 5 P and Q).

DiscussionAt 10 nM, ouabain does not produce its well-known toxic effects,such as the inhibition of K+ and Na+ pumping, cell detachmentfrom its neighbors, and substrate or cell death (12, 13, 18, 20, 30,31). Nevertheless, at this concentration, oubain modulates thesealing of TJs by changing the cell content and localization ofdifferent claudins through specific mechanisms. In the present

work, we also observe that 10 nM ouabain accelerates ciliogenesis,as well as increases the length of the cilium. Ouabain acceleratesthe expression of acetylated α-tubulin at the cilium, yet it doesnot change its cell content. This molecule also increases the totalcell content of claudin-2, as well as its localization at the cilium,and promotes the association between these proteins. Ciliumdisassembly operates usually at a constant rate in different spe-cies and conditions (16), therefore ouabain may accelerate thepolarized traffic to provide building blocks to the growing cilium,as well as the anterograde transport.We demonstrated that ouabain requires the lateral ouabain-

sensitive Na+,K+-ATPase to modulate the sealing of TJs (32).To test if Na+,K+-ATPase is the receptor occupied by ouabainto accelerate ciliogenesis, we used resistant (R)-MDCK cells(25), whose resistance to ouabain results from a substitution ofa cysteine by tyrosine or a phenylalanine in the first trans-membrane segment of Na+,K+-ATPase (33). R-MDCK cells donot respond to low concentrations of ouabain, a result thatstrengthens the possibility that Na+,K+-ATPase is the receptor.As in other cellular processes, such as cell proliferation (34),

Fig. 5. Claudin-2, besides of being a typical TJ protein, is also a ciliary protein that is affected by ouabain via ERK1/2 signaling. Immunofluorescence picturesof monolayers incubated with ouabain for 48 h to stain claudin-2 (A, D, and G, green) and -1 (J, green) and acetylated α-tubulin (B, E, H, and K, red). C, F, I, andL correspond to the superposition of the corresponding images. D, E, and F are lateral views of monolayer showing claudin-2 at the cilium. (G, H, and I) x/yprojections showing that claudin-2 and acetylated α-tubulin colocalize along the whole cilium at the apical domain. Statistical analysis of the percentage ofcells with ciliar claudin-2 observed in images of immunofluorescent cells costained with antibodies against this protein and acetylated α-tubulin, undercontrol, ouabain-stimulated, control with PD, and ouabain treatment in the presence of PD (columns 1–4) (M); ***P < 0.001. Total cell content of acetylatedα-tubulin (N, open squares) and claudin-2 (N, open circles). The correspondent effect of PD is shown in gray circles for claudin-2 and gray squares for acet-ylated α-tubulin. To gain information on the association between acetylated α-tubulin and claudin-2, we coprecipitated the first and blotted for the second(O). Lane 1 corresponds to total homogenate (Total Ext), and the lanes 2–5 depict immunoprecipitation with an antibody against acetylated α-tubulin (IP:AcTub), blotted with an antibody against the same acetylated α-tubulin (WB: AcTub) or claudin-2 (WB: claudin-2). Ouabain effect is shown in the lane 3(ouabain). Lane 4 shows the effect of PD. Interestingly, this inhibitor increases by itself the association. The effect of ouabain does not depend on ERK1/2 asPD does not block the effect (lane 5). Claudin-2 (green) in a MDCK clone that does not express this protein in the cilia, identified as usual with fluorescent redlabel, in a x/y (P) and z/x image (Q). Ciliary claudin-2 is not required for the development of the cilium. (Magnification: A–F and J –L, 400×; G and H, 1,200×; Pand Q, 600×.) (Scale bar in J, 10μm.)

20594 | www.pnas.org/cgi/doi/10.1073/pnas.1102617108 Larre et al.

Dow

nloa

ded

by g

uest

on

July

24,

202

1

death (35), and TJ sealing (12), ouabain modulates the ciliarylocalization of acetylated α-tubulin although ERK1/2.Chrystallographic analysis shows that there is a single ouabain

binding site per α-subunit (36), supporting the possibility thatdespite being extremely different, toxic and hormone-like effectsare triggered by binding to the same site of in the α-subunit. Inclassic pharmacology, “intrinsic efficacy” is viewed as a constantfor each ligand at a given receptor, irrespective of where thereceptor is expressed. This concept is somewhat insufficient toprovide a useful theoretic framework for the multitude of ligand-receptors known today. For this reason, Urban et al. (37) pro-posed that ligands induce unique ligand-receptor conformationsthat frequently result in the differential activation of signal-transduction pathways, and coined the term “functional selec-tivity.” In this respect, the present work provides evidence thatthe interaction of ouabain with Na+,K+-ATPase at the samereceptor site, results in several drastically different cell respon-ses. The situation is further complicated by several additionalfacts: (i) the expression of Na+,K+-ATPase at the cell mem-brane depends on the interaction of two receptors present inneighboring cells that anchor each other in such a position (22,23); (ii) the overall effect (e.g., vectorial Na+ transport) requiresthat at least two cells collaborate (e.g., in the making of a TJ);and (iii) the variety of responses is further compounded by thefact that the enzyme can be assembled by at least four differentα- and three different β-isoforms, which have different sensitiv-ities to ouabain. The reasons why a given isoform predominatesin certain tissues remain to be elucidated.The exposure of Na+-K+-ATPase at a given cell border

depends on the homotypic interaction between two identicalβ-subunits of Na+-K+-ATPases (22, 23). Failure to observe theexpression of this enzyme at heterotypic borders may not bebecause of its absence, but because of the fact that it only dwellsbriefly at the plasma membrane. In this respect, we have pre-viously observed that subconfluent cells and confluent MDCKcells incubated in media without Ca2+ express up to a third ofthe Na+-K+-ATPases population at the membrane, according tothe 3H-ouabain binding method (38). We have also made a sim-ilar observation with the Shaker ion channel molecule in which wehad deleted the retention domain: the deletion does not impair thecorrect polarized expression of the channel, but results in transientdwelling (39). However, a brief exposure of Na+-K+-ATPase to 10nM ouabain is sufficient to trigger ciliogenesis in single MDCKcells completely surrounded by NRK cells.We observed that claudin-2, a typical molecular component of

the TJ, localizes also at the cilium, and that this expression isspecific, as neither claudin-1 nor -4, nor occludin were detected(Fig. 5 and Fig. S1). These results show that the expression ofclaudins at the cilium is not restricted to the human retinalpigment epithelium, as previously suggested (40). Nevertheless,here we show that ciliary claudin-2 is not necessary for cilio-genesis. Septin forms a diffusion barrier that restricts the passageof large molecules toward and away from the cilium (41), witha limit around 10 kDa (17). This barrier may prevent ciliaryclaudin-2 from diffusing toward the rest of the apical domain. Yuet al. (42) and Muto et al. (43) observed that claudin-2 has theability to form pores in the TJs where the side-chain carboxylgroup of aspartate-65 forms a site that specifically favors thepassage of Na+ ions. However, given that the cilium does notseparate the two liquid compartments, it is highly unlikely thatciliary claudin-2 would play a role in permeation. However, ifclaudin-2 contains a Na+-sensitive site, it may possibly bind tothis ion in a concentration-dependent manner and act as a sensorof Na+ concentration in the lumen of the nephron. This sensingrole of cilia would be consistent with its participation in olfactionand photoreception. Furthermore, cilia are connected to a seriesof other mechanisms, such as signaling routes (44). In this

respect, we observed that the effect of ouabain is also mediatedby ERK1/2.We have previously showed that 10 nM ouabain modulates

cell contacts, such as TJs (12), and communicating junctions(13). Here we demonstrate that 10 nM ouabain also stimulatescell polarity, the other fundamental feature of the transportingepithelial phenotype.

Materials and MethodsAntibodies and Chemicals. Antibodies against: claudin-1(Cat. no. 51–9000), -2(Cat. no 51-6100), -4 (Cat. no 32-9400), ZO-2 (Cat. no 71-1400), and occludin(Cat. no 71-1500), as well as secondary HRP-anti mouse (Cat. no 61-6620),HRP-anti rabbit (Cat. no 62-6120), FITC-anti rabbit (Cat. no F2765), TRITC-antimouse (Cat. no 62-6514), and FITC-goat anti-rat (Cat. no 62-9511) wereobtained from Invitrogen. Antiacetylated tubulin was obtained from Sigma(T-7451) and α-subunit Na+,K+-ATPase from Thermo Scientific (MA3-929).Ouabain was obtained from Sigma (O-3125), and inhibitor PD98059 wasfrom Calbiochem-Novabiochem.

Cell Culture, Chemicals, and Antibodies. MDCK-II cells (canine renal; AmericanType Culture Collection, CCL-34) were grown at 36.5 °C in a 5% CO2 atmo-sphere in DMEM (GIBCO-Invitrogen) supplemented with penicillin-strepto-mycin 10,000 U·μg·mL (In Vitro) and 10% FBS (GIBCO-Invitrogen). Thismedium will be referred to as CDMEM. Cells were harvested with trypsin-EDTA (In Vitro) and plated on glass coverslips contained in 24-well multid-ishes (Costar 3524) and other supports specified below for each experiment.The medium were cultured at a ∼70% saturating density, maintained forone day in CDMEM, followed by serum starvation (24 h, 1% FCS in CDMEM)and then treatment with or without 10 nM ouabain. NRK cells (rat kidney;American Type Culture Collection CRL 1571) were cultured in the samemanner. Monolayers were exposed to PD 1 h before starting the challengewith ouabain.

Ciliogenesis in Mixed Monolayers. To obtain small groups (1–5 cells) of MDCKcells, suspensions of MDCK cells expressing the red fluorescent protein (seebelow) were mixed at a ratio of 1:99 with nonstained NRK cells and plated toconfluence. These cells were allowed 30 min to attach, and then the mediumwas changed to medium containing 10% FCS. This medium was replaced12 h later by medium with 1% CDMEM with or without 10-nM ouabain andcells fixed at different times and stained as described below.

Immunoblot. Total cell extracts frommonolayers in inserts were washed threetimes with ice-cold PBS with Ca2+ and then incubated at 4 °C for 10 min withlysis buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate,0.1% SDS, and 50 mM Tris pH 7.5) for protein extraction. Then cells werecentrifuged 10 min at 17,000 × g. The supernatant was recovered and thetotal protein content was measured by the BCA assay (Pierce), subsequentlyboiled in Laemmli sample buffer (Bio-Rad; Cat. no. 1610737) and resolved bySDS-PAGE and transferred to PVDF sheets (Hybond-P; Amersham Bio-sciences). These sheets were blocked overnight with 5% BSA. Specific bandswere detected with specific antibodies and chemiluminiscence (ECL andHiperfilm; Amersham). Resolved bands were analyzed with the softwareKodak 1D 3.5.4 (Eastman Kodak) and data were processed with GraphPadPrism 4 (GraphPad Software).

Immunoprecipation. After 48 h of 10 nM ouabain treatment, 100 cm2 mon-olayers were washed with PBS containing 0.1 mM phenylmethylsulfonylfluoride (PMSF) and scraped into 1,500 μL of RIPA buffer (20 mM Tris pH 7.5,2 mM EGTA, 5 mM EDTA, 30 mM NaF, 40 mMβ-glycerophosphate, 20 mMsodium orthovanadate, 3 mM benzamidine; 0.5% Nonidet Nonidet P-40,and a protease inhibitor mixture from Roche) and lisated with an insulinesyringe. Samples were centrifuged at 20,000 × g, 4 °C for 10 min, the super-natants were collected and 5.5 mg of protein were used for the immuno-precipitation. This assay was performed with the exacta Cruz Kit (Cat. no.sc-45053; Santa Cruz Biotechnology) following the recommendations ofthe manufacturer.

Immunofluorescence. Monolayers on coverslips were washed three timeswith ice-cold PBS fixed and permeabilized with methanol for 8 min at −20 °C,washed three times with PBS, blocked 1 h with 0.5% BSA, and incubatedfor 1 h at 37 °C with a specific primary antibody, followed by washes asabove, and an incubation with a second antibody against the primaryantibody. Monolayers were then rinsed six times with PBS, incubated witha FITC- or TRITC-labeled secondary antibody according to the animal

Larre et al. PNAS | December 20, 2011 | vol. 108 | no. 51 | 20595

CELL

BIOLO

GY

Dow

nloa

ded

by g

uest

on

July

24,

202

1

species used (1 h at room temperature), rinsed as indicated before,mounted in Fluorguard (Bio-Rad), and examined by confocal microscopy(SP2 Leica Microsystems). Captured images were processed with ImageJ(National Institutes of Health) and figures constructed with GIMP (GNUimage manipulation program).

Scanning Electron Microscopy. After 24 h of 10 nM ouabain treatment, cellswere fixed with 2.5% glutaraldehyde in saline buffer (100 mM KCl, 10 mMCaCl2, 3.5 mM MgCl2, and 10 mM Hepes pH 7.4) for 1 h at 37 °C and post-fixed with 1% osmium tetroxide in PBS for 1 h at room temperature.Monolayers were gently washed three times with PBS, dehydrated in in-creasing concentrations of ethanol (from 50% to 100%), critical-point driedusing a Sandri-780A apparatus (Tousimis), gold-coated with a Desk II Goldsputter-etch unit (Denton Vacuum Inc.), and examined with a Jeol JSM-6510LV scanning electron microscope.

Tansepithelial Electrical Resistance. Cells were grown on Transwell permeablesupports as described by Larre et al. (13).

Cell Transfection to Obtain Permanently Red-Stained MDCK Cell. MDCK-II (3 ×105 cells/mL, that affords roughly 80–90% confluence) were seeded in 2 mLof growing medium in 35-mm diameter dish 1 d before transfection. Cellswere prewashed with Opti-MEM I Reduced Serum Medium (GIBCO; 31985–062). Lipofectamine 2000 (Invitrogen) was complexed with unmodifiedpDsRed2-N1 expression vectors (BD Biosciences Clontech). Plasmid at reagent:DNA ratio of 1.56 μL:3 μg. Complexes were prepared by mixing lipofect-amine 2000, 6 μL with 250 μL of Opti- MEM I, followed by the addition of

plasmid DNA. The mixture was incubated for 5 min at room temperatureafter the addition of the transfection reagent, and another 30 min afteraddition of DNA. Lipofectamine 2000 complexes with DNA were added ina volume of 0.5 mL per dish. Cells were incubated at 37 °C for 4 h and thetransfection complexes were removed. All transfectants were maintained inantibiotic-free complete medium. The stable cell line was acquired understress condition with antibiotic G418. The fluorescent cells were viewed witha confocal and Epi-fl microscopy. pDsRed2-N1 was used to express a DsRed2protein in the MDCK cell line as a transfection marker (red-MDCK). pDsRed2-N1 was amplified and then purified by Wizard Plus Maxipreps DNA Purifi-cation System (Promega; A7270).

Statistical Analyses. Statistical analyses were performed with GraphPadPrism 4. Results are expressed as the mean ± SE. Statistical significancewas estimated with a one-way ANOVA followed by a Bonferroni’s mul-tiple comparison test (*P < 0.05, **P < 0.01, ***P < 0.001). All experi-ments were repeated at least three times, and the data are presented asmean ± SE.

ACKNOWLEDGMENTS. We thank Dr. A. Arias for his expert input; E. del Oso,Y. de Lorenz, J. Soriano, E. Estrada, and E. Méndez for technical and admin-istrative aid; S. González (Electron Microscopy Unit, Centro de Investigacióny de Estudios Avanzados del Instituto Politécnico Nacional) for her compe-tent assistance in studies involving scanning electron microscopy; andR. Bonilla for plasmid manipulations. This work was supported by the Con-sejo Nacional de Ciencia y Technología and the Instituto de Ciencia y Tecno-logía del DF (México City Research Council).

1. Hamlyn JM, et al. (1991) Identification and characterization of a ouabain-like com-pound from human plasma. Proc Natl Acad Sci USA 88:6259–6263.

2. Kawamura A, et al. (1999) On the structure of endogenous ouabain. Proc Natl AcadSci USA 96:6654–6659.

3. Komiyama Y, et al. (2001) Identification of endogenous ouabain in culture superna-tant of PC12 cells. J Hypertens 19:229–236.

4. Schneider R, et al. (1998) Bovine adrenals contain, in addition to ouabain, a secondinhibitor of the sodium pump. J Biol Chem 273:784–792.

5. Bauer N, et al. (2005) Ouabain-like compound changes rapidly on physical exercise inhumans and dogs: Effects of beta-blockade and angiotensin-converting enzyme in-hibition. Hypertension 45:1024–1028.

6. Manunta P, Hamilton BP, Hamlyn JM (2006) Salt intake and depletion increase cir-culating levels of endogenous ouabain in normal men. Am J Physiol Regul IntegrComp Physiol 290:R553–R559.

7. Hamlyn JM, et al. (1982) A circulating inhibitor of (Na+ + K+)ATPase associated withessential hypertension. Nature 300:650–652.

8. Bagrov AYa, et al. (1991) Endogenous plasma Na,K-ATPase inhibitory activity anddigoxin like immunoreactivity after acute myocardial infarction. Cardiovasc Res 25:371–377.

9. Schoner W (2002) Endogenous cardiac glycosides, a new class of steroid hormones.Eur J Biochem 269:2440–2448.

10. Schoner W, Scheiner-Bobis G (2007) Endogenous and exogenous cardiac glycosides:Their roles in hypertension, salt metabolism, and cell growth. Am J Physiol Cell Physiol293:C509–C536.

11. Contreras RG, et al. (2004) Ouabain binding to Na+,K+-ATPase relaxes cell attachmentand sends a specific signal (NACos) to the nucleus. J Membr Biol 198:147–158.

12. Larre I, et al. (2010) Ouabain modulates epithelial cell tight junction. Proc Natl AcadSci USA 107:11387–11392.

13. Larre I, et al. (2006) Contacts and cooperation between cells depend on the hormoneouabain. Proc Natl Acad Sci USA 103:10911–10916.

14. Cereijido M, Contreras RG, Shoshani L (2004) Cell adhesion, polarity, and epithelia inthe dawn of metazoans. Physiol Rev 84:1229–1262.

15. Cereijido M, Contreras RG, Shoshani L, Flores-Benitez D, Larre I (2008) Tight junctionand polarity interaction in the transporting epithelial phenotype. Biochim BiophysActa 1778:770–793.

16. Ishikawa H, Marshall WF (2011) Ciliogenesis: Building the cell’s antenna. Nat Rev MolCell Biol 12:222–234.

17. Smith EF, Rohatgi R (2011) Cilia 2010: The surprise organelle of the decade. Sci Signal4:mr1.

18. Pchejetski D, et al. (2003) Inhibition of Na+,K+-ATPase by ouabain triggers epithelialcell death independently of inversion of the [Na+]i/[K+]i ratio. Biochem Biophys ResCommun 301:735–744.

19. González-Mariscal L, Chávez de Ramirez B, Lázaro A, Cereijido M (1989) Establish-ment of tight junctions between cells from different animal species and differentsealing capacities. J Membr Biol 107:43–56.

20. Contreras RG, et al. (1995) A novel type of cell-cell cooperation between epithelialcells. J Membr Biol 145:305–310.

21. Nagafuchi A, Shirayoshi Y, Okazaki K, Yasuda K, Takeichi M (1987) Transformation ofcell adhesion properties by exogenously introduced E-cadherin cDNA. Nature 329:341–343.

22. Padilla-Benavides T, et al. (2010) The polarized distribution of Na+,K+-ATPase: Role ofthe interaction between beta subunits. Mol Biol Cell 21:2217–2225.

23. Shoshani L, et al. (2005) The polarized expression of Na+,K+-ATPase in epithelia de-pends on the association between beta-subunits located in neighboring cells.Mol BiolCell 16:1071–1081.

24. Hammerton RW, et al. (1991) Mechanism for regulating cell surface distribution ofNa+,K(+)-ATPase in polarized epithelial cells. Science 254:847–850.

25. Soderberg K, Rossi B, Lazdunski M, Louvard D (1983) Characterization of ouabain-resistant mutants of a canine kidney cell line, MDCK. J Biol Chem 258:12300–12307.

26. Zampar GG, et al. (2009) Acetylated tubulin associates with the fifth cytoplasmicdomain of Na(+)/K(+)-ATPase: Possible anchorage site of microtubules to the plasmamembrane. Biochem J 422:129–137.

27. Menco BP, et al. (1998) Ultrastructural localization of amiloride-sensitive sodiumchannels and Na+,K(+)-ATPase in the rat’s olfactory epithelial surface. Chem Senses23:137–149.

28. Aperia A (2007) New roles for an old enzyme: Na,K-ATPase emerges as an interestingdrug target. J Intern Med 261:44–52.

29. Xie Z (2003) Molecular mechanisms of Na/K-ATPase-mediated signal transduction.Ann N Y Acad Sci 986:497–503.

30. Contreras RG, Shoshani L, Flores-Maldonado C, Lázaro A, Cereijido M (1999) Re-lationship between Na(+),K(+)-ATPase and cell attachment. J Cell Sci 112:4223–4232.

31. Akimova OA, et al. (2005) Cardiotonic steroids differentially affect intracellular Na+and [Na+]i/[K+]i-independent signaling in C7-MDCK cells. J Biol Chem 280:832–839.

32. Larre I, Cereijido M (2010) Na,K-ATPase is the putative membrane receptor of hor-mone ouabain. Commun Integr Biol 3:625–628.

33. Canessa CM, Horisberger JD, Louvard D, Rossier BC (1992) Mutation of a cysteine inthe first transmembrane segment of Na,K-ATPase alpha subunit confers ouabainresistance. EMBO J 11:1681–1687.

34. Tian J, et al. (2009) Changes in sodium pump expression dictate the effects of ouabainon cell growth. J Biol Chem 284:14921–14929.

35. Aizman O, Uhlén P, Lal M, Brismar H, Aperia A (2001) Ouabain, a steroid hormonethat signals with slow calcium oscillations. Proc Natl Acad Sci USA 98:13420–13424.

36. Ogawa H, Shinoda T, Cornelius F, Toyoshima C (2009) Crystal structure of the sodium-potassium pump (Na+,K+-ATPase) with bound potassium and ouabain. Proc Natl AcadSci USA 106:13742–13747.

37. Urban JD, et al. (2007) Functional selectivity and classical concepts of quantitativepharmacology. J Pharmacol Exp Ther 320:1–13.

38. Shoshani L, Contreras RG (2001) Biogenesis of Epithelial Polarity and the TightJunctions. Tight Junctions, eds Cereijido M, Anderson JM (CRC Press, Boca Raton, FL).

39. Moreno J, Cruz-Vera LR, García-Villegas MR, Cereijido M (2002) Polarized expressionof Shaker channels in epithelial cells. J Membr Biol 190:175–187.

40. Nishiyama K, Sakaguchi H, Hu JG, Bok D, Hollyfield JG (2002) Claudin localization incilia of the retinal pigment epithelium. Anat Rec 267:196–203.

41. Hu Q, et al. (2010) A septin diffusion barrier at the base of the primary ciliummaintains ciliary membrane protein distribution. Science 329:436–439.

42. Yu AS, Kanzawa SA, Usorov A, Lantinga-van Leeuwen IS, Peters DJ (2008) Tightjunction composition is altered in the epithelium of polycystic kidneys. J Pathol 216:120–128.

43. Muto S, et al. (2010) Claudin-2-deficient mice are defective in the leaky and cation-selective paracellular permeability properties of renal proximal tubules. Proc NatlAcad Sci USA 107:8011–8016.

44. Gerdes JM, Davis EE, Katsanis N (2009) The vertebrate primary cilium in development,homeostasis, and disease. Cell 137:32–45.

20596 | www.pnas.org/cgi/doi/10.1073/pnas.1102617108 Larre et al.

Dow

nloa

ded

by g

uest

on

July

24,

202

1