Ouabain Increases Gap Junctional Communication in Epithelial Cells

The high affinity and specificity of ouabain for the Na⁺-K⁺-ATPase led to search for endogenous analogs, which were subsequently identified as ouabain itself and other cardiac steroids [1,2]. The plasma levels of ouabain increase in response to muscular exercise, a high-sodium diet and pathological conditions, such as eclampsia, certain forms of arterial hypertension and myocardial infarction [2,3,4,5,6,7]. The available experimental evidence suggests that the endogenous cardiac steroids are hormones [8,9,10], raising the question of their physiological role.

High levels of ouabain (≥ 300 nM) activate the retrieval of cell-adhesion molecules that are associated with, or form part of the plasma membrane, and this process causes the cells to detach (e.g., MDCK cells) from the substrate as well as from one another [11,12]. Therefore, we hypothesized that one physiological role of ouabain is the modulation of cell contacts [9,13]. We have already shown that low, non-toxic concentrations of ouabain modulate the hermeticity and molecular composition of tight junctions (TJ) [14]. Furthermore, we found that 10 nM ouabain accelerates ciliogenesis, one of the final stages of cell polarity development [15]. Interestingly, TJ and apical/basolateral polarity (as measured by ciliogenesis) are the essential differentiating characteristics of transporting epithelia. Ouabain is capable of modulating both of these characteristics at a concentration that falls within the range that is observed in mammals [16,17] and, therefore, neither inhibits the Na⁺-K⁺-ATPase from pumping K⁺ nor distorts the cellular K⁺ balance [18,19,20].

Gap junctions are another noteworthy type of cell-cell contact that enable communication between neighboring cells and are involved in cellular processes as fundamental and diverse as proliferation, differentiation, metabolic cooperation, synchronization, cancer, metastasis and many others [21,22,23]. Accordingly, in the present study we explored whether ouabain at physiological levels observed in resting dogs and humans [16,17] exerts an effect on gap junctional communication (GJC). For this purpose, we evaluated GJC via two distinct experimental methods: 1) by micro-injecting individual cells in confluent MDCK monolayers with biotin ethylenediamine (neurobiotin, NB), a substance that can diffuse from cell to cell via gap junctions and 2) by measuring the membrane capacitance of cells coupled via GJC using whole-cell patch clamp assays [24,25]. Additionally we made silencing assays showing that connexins 32 and 43 are involved in this response.

Starter MDCK-II cell cultures (MDCK and CCL-34) were obtained from the American Type Culture Collection. The cells were grown at 36.5°C in a 5% CO₂ atmosphere in Dulbecco's modified Eagle medium (DMEM; Life Technologies, Carlsbad CA, USA) supplemented with penicillin-streptomycin 10,000 U/µg/ml (In Vitro, Acayucan, Mexico) and 10% fetal bovine serum (GIBCO). This medium is hereafter referred to as CDMEM. The cells were harvested with trypsin-EDTA (In Vitro) and seeded on glass coverslips in 24-well multi-dishes (3524; Costar Corning, NY, USA) for immunofluorescence, biotin ethylenediamine, hydrobromide (Neurobiotin, A1593, Life Technologies) injection and electrical capacitance measurements or without glass coverslips for Western blot analysis. Cells were cultured at a saturating density of ∼70%, maintained for one day in CDMEM, serum starved for 24 h in DMEM containing 1% fetal bovine serum and then treated with or without 10 nM ouabain (O-3125; Sigma-Aldrich, St Louis MO, USA). The cell monolayers were exposed to 25 µM PD98059, an inhibitor of mitogen extracellular kinase-1 (MEK-1; 513000; Merk Millipore, Darmstadt GE) that impairs the activation of ERK1/2, or 10 µM PP2, an inhibitor of c-Src kinase (529573; Merk Millipore). Actinomycin D (A9415; Sigma-Aldrich) was dissolved in DMSO (5 mg/ml) prior to use. Cycloheximide (C4859; Sigma-Aldrich) was obtained as a ready-made solution (100 mg/mL in DMSO). These inhibitors were added 1 h before the ouabain challenge was initiated.

Glass micropipettes with a tip resistance of 5-10 MΩ were backfilled with a mixture of 2% of neurobiotin (NB, MW 0.37 kD), and 1% FITC-dextran (MW 20 kD, Fine Chemical/Amersham Pharmacia) in saline solution (120 mM KCl, 5 mM NaCl, 1 mM MgCl₂, and 5 mM HEPES, pH 7.4). After the pipettes were filled, they were attached to a holder mounted to a micromanipulator (PCS-750; Burleigh Instruments) for cell impalement. Monolayers grown on glass coverslips were deposited into a glass-bottom chamber filled with PBS containing 1.8 mM Ca²⁺ at room temperature. Then, the chamber was mounted on the stage of an inverted epifluorescence-equipped microscope (Diaphot 300; Nikon) to monitor impalement and injection. The cells were randomly selected for injection, and after impalement, they were injected with a pneumatic pulse using a microinjecting device (IM300; Narishige). Effective injection was verified by the diffusion of FITC-dextran throughout the cell. Fifteen minutes after injection, the cells were fixed for 15 min using cold methanol, rinsed with PBS, and incubated overnight at 4°C in TRITC-streptavidin (Zymed, catalog number 43-4314) at a dilution of 1:200. Finally, the samples were rinsed twice and mounted using VECTASHIELD® (H-1000; Vector Laboratories, Burlingame CA, USA). Eight-bit images of the fluorescent cells were acquired at room temperature using a Zeiss M200 inverted microscope equipped with a Plan-NeoFluar 63x N.A. 1.25 objective lens, an AxioCam MRm camera and Axovision 4.8 software. The captured images were imported into FIJI Is Just ImageJ software (release 2.8, NIH, Bethesda, MD, USA) to adjust the brightness and the contrast and GIMP (release 2.8.10, NIH) to compose the figures.

GJC was estimated by recording the membrane capacitance of electrically coupled cells using the whole-cell patch-clamp technique [25,26,27]. Patch micropipettes were produced from borosilicate glass tubing (Kimax 51; Kimble) using a puller device (model p-87, Flaming/Brown micropipette puller; Sutter Instruments). The tip resistance of the micropipettes ranged from 2 to 5 MΩ after heat polishing. To generate gigaseals on the cells, the micropipettes were backfilled with saline solution (110 mM KCl, 5 nM NaCl, 5 mM EGTA-K, 1 mM MgCl₂, and 10 mM HEPES, pH 7.4) and attached via a pipette holder to a piezoelectric-driven micromanipulator (PCS-5000; Burleigh). The cells were bathed in external saline solution (115 mM NaCl, 5 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, and 10 mM glucose, pH 7.4). The pipettes were mechanically manipulated under an inverted microscope (Diaphot 300; Nikon). A capacitive current (IC) was induced using a hyperpolarizing square voltage pulse from -80 to -90 mV and was recorded at 10 KHz. The membrane capacitance was estimated offline by integrating the area of the capacitive transient beginning at the onset of the pulse.

The membrane capacitance (Cm) was calculated by dividing the integral of the IC by the amplitude of the pulse (-10 mV) according to the following equation:

where Cm is the membrane capacitance, and ΔV is the amplitude of the voltage pulse (-10 mV). The integral of the IC was calculated using the Clampfit module of the pCLAMP 8.0 software suite (Molecular Devices). The membrane surface area was estimated assuming a specific membrane capacitance of 0.01 pF/μm².

Inhibitory shRNAs were designed against dog Cnx-32 (Acc. No. XM_538073 target sequence: 5'-GCCGTCTTCACTGTCTTCAT-3' and dog Cnx-43 (Acc. No. NM_00100295, target sequence: 3'-GGGAAGAGAAACTGAACAAG-3') and cloned into the pLVX-shRNA2 vector (632179; Clontech Laboratories Inc.), which also expresses the green fluorescent marker ZsGreen1. MDCK cells were transiently transfected with the indicated plasmids using Lipofectamine® 2000 (Invitrogen); five hours after transfection, the cells were harvested and seeded at confluency on glass coverslips in 24-well dishes, maintained for one day in CDMEM and serum starved for 24h as described above.

Following incubation in 10 nM ouabain for 1 hour, dye was injected (on the same monolayer) into cells in which Cnx-32, Cnx-43 or both had been silenced (those cells expressing ZsGreen1) or not (those not expressing ZsGreen1).

Statistical analyses were performed using Prism 5 (GraphPad software). The results are expressed as the means ± standard error. Statistical significance was estimated via one-way analysis of variance (ANOVA) followed by Bonferroni's multiple comparison or Student's t-test and was denoted as follows: * P<0.05, ** P<0.005 and *** P<0.001 and n is the number of observations obtained from at least 3 independent experiments.

Cells in a confluent monolayer were randomly injected with a mixture of FITC-dextran and NB; FITC-dextran was used to verify that only one cell had been injected, and NB was used to detect the number of cells that received NB via gap junctions. In the absence of ouabain, it was observed that in most cases, both FITC-dextran and NB remained only in the injected cell (Fig. 1A) indicating that GJC is rather infrequent. In contrast, in cell monolayers treated with 10 nM ouabain for 1 h, FITC-dextran was still observed only in the injected cell, but NB diffused to several neighboring cells (Fig. 1B red), demonstrating that ouabain enhances GJC (Fig. 1C red circles). This enhancement of GJC was statistically significant from as early as 10 min after the addition of ouabain, peaked at 1 h, and remained at a level that was significantly higher than that under control conditions (Fig. 1C, open circles) for at least 24 h. Fig. 1D shows that cycloheximide, an inhibitor of protein synthesis, and actinomycin D, an inhibitor of RNA synthesis, did not suppress the enhancing effect of ouabain on GJC, suggesting that the observed enhancement of GJC by ouabain does not involve the synthesis of new protein components.

To investigate this enhancement of GJC by ouabain using an entirely completely different method, we performed whole-cell patch clamp recording to determine the effect of ouabain on the electrical coupling of neighboring MDCK cells in monolayer, as estimated from capacitance measurements. We induced capacitive current transients (IC) (Fig. 1E, above) with voltage steps from -80 to -90 mV to estimate the membrane capacitance of coupled cells (see methods) [26,27] and confirmed that 10 nM ouabain increased the area of the plasma membrane (Fig. 1E, above and below, white vs. red capacitive transients and bars), reflecting an increase in the number of electrically coupled cells. To determine whether this increase in area membrane is in effect due to enhanced GJC we tested the effect of octanol (Fig. 1E, OCT, yellow bar), an alcohol that is known to effectively uncouple gap junctions [28] and, as Fig. 1E (orange bar) shows, this treatment suppressed the observed increase in membrane surface. Therefore, these results were consistent with those obtained from observing the diffusion of NB. In summary, the two experimental approaches concur in demonstrating that 10 nM ouabain enhances GJC. Taken together, these observations strengthen our general hypothesis that ouabain modulates cell contacts [9]; as it is also in keeping with our previous observations of TJ [14] and ciliogenesis [15].

The effects of ouabain rely on the receptor function of the Na⁺-K⁺-ATPase [29,30]; in some cases, the Na⁺-K⁺-ATPase functions as a receptor regardless of the inhibition of pump activity and/or of the associated ionic imbalance [18,19,20]. Therefore, to determine whether the Na⁺-K⁺-ATPase mediates the GJC-enhancing effect of ouabain, we used R-MDCK, a mutant subclone resistant to high concentrations of ouabain (> 1 µM) [31], resulting from a substitution of cysteine by tyrosine or phenylalanine in the first transmembrane segment of the α-subunit of Na⁺,K⁺-ATPase [32]. Fig. 2A shows a control monolayer of wild-type MDCK cells (W-MDCK) injected with NB that displayed no communication between neighboring cells. The treatment of these cells with 10 nM ouabain for 24 h (Fig. 2B, and E first pair of bars) enabled NB to diffuse to several neighboring cells. Fig. 2C and 2D are analogous to the previous experiments, except that in this case, we used R-MDCK cells and detected no effect of ouabain(Fig. 2E, second pair of bars). These results indicate that the receptor by which ouabain increases GJC is the Na⁺-K⁺-ATPase.

The Na⁺-K⁺-ATPase forms a receptor complex with c-Src and IP₃-Receptor, both of which activate signaling pathways (such as ERK1/2) and stimulate Ca²⁺ waves [33,34]. In recent studies, we found that 10 nM ouabain provokes a single wave of c-Src and ERK1/2 phosphorylation that peaks within the first hour of application, coinciding with the peak of GJC. Furthermore, incubation in PP2 inhibits c-Src, and incubation in PD98059 inhibits MEK kinase at least for 8 h [12]. These signaling proteins also modulate the effect of ouabain on TJ [14] and ciliogenesis [15]. In keeping with these evidences, we now show the effect of the inhibitors PP2 and PD98059 on the GJC-enhancing effect of 10 nM ouabain after 1 h of exposure (see also Fig. 3G, white bar vs. red bar). Fig. 3C and D shows the effects of 1 h of pre-treatment with PP2 (c-Src-inhibitor) in the absence and presence of treatment with 10 nM ouabain, respectively. Pre-treatment with inhibitor PP2 in the absence of ouabain (Fig. 3C and 3G, light blue bar) increases cell-to-cell communication but completely blocks the effect of 10 nM ouabain (Fig. 3D and 3G, blue bar). Figs. 3E and F show an analogous experiment using the ERK1/2 inhibitor PD98059 prevents the effect of ouabain on GJC. Therefore, c-Src and ERK1/2 participate in the ouabain-mediated GJC enhancement at 1 h.

GJC occurs via connexons, whose primary protein components are connexins (Cnxs) [21,22,23]. Given that in previous studies using different experimental conditions and 1 µM ouabain it has been observed that ouabain modulate Cnx-32 in MDCK cells [13,35], we examined whether these two Cnxs participate in the ouabain-mediated enhancement of GJC. For this purpose, we made plasmids that contained a shRNA aimed to silence Cnx-32 or Cnx-43, along with ZsGreen1 as a reporter of expression, then we made transient transfections of MDCK cells with either one of those two plasmids as well as co-transfection of both. In order to start the dye transfer assays, we waited 48 h to insure that the analyzed connexin has sufficient time to reduce its expression in the transfected cell. In the same monolayer, we made dye transfer assays on cells expressing ZsGreen1 (i.e., contained shRNAs) or not to compare the number of communicating cells. As shown in Fig. 4 A, B, and C, dye transfer assays were performed on cells incubated for 1 h in 10 nM ouabain. Fig. 4A demonstrates that in a cell not expressing ZsGreen1 (shown by the arrow), cell-to-cell communication occurred similarly to that in Fig. 1A. Alternatively, in a cell expressing ZsGreen1 (i.e., containing an shRNA targeting Cnx-43, yellow), cell-to-cell communication was blocked (Fig. 4B). Fig. 4C (yellow) shows a cell that was transfected with a scrambled shRNA sequence, demonstrating that the passage of NB was not inhibited and that the blockade elicited by the targeted shRNA was specific, as shown in panel B (yellow). Fig. 4D contains the statistical analysis of ouabain-induced GJC in cells expressing the targeted shRNA (green bars), in cells not expressing shRNA (white bars) and in cells transfected with a scrambled shRNA (last green bar) as a negative control. As shown in the first two columns, the number of communicating cells when the cell that was injected with NB did not express shCnx32 (white bar) was compared to that when the cell that was injected with NB did express shCnx32 (green bar). This significant decrease in cell-to-cell communication demonstrates that Cnx-32 participates in the response to ouabain. The second pair of columns is analogous, but the Cnx evaluated was Cnx-43; these results indicate that Cnx-43 also participates in ouabain-induced GJC. The third pair of columns demonstrates that the co-transfection of cells with both targeted shRNAs further decreased GJC. The final pair of columns shows that this effect was effectively due to a specific silencing of the target sequence, since transfection with a scrambled sequence did not produced a significantly change. These results indicate that the enhancement of GJC by ouabain requires the presence of both Cnx-32 and Cnx-43, although the participation of other Cnx isoforms cannot be excluded.

Previous findings led us to hypothesize that one possible function of ouabain is the modulation of cell-cell contacts [13]. To explore this possibility, we first examined the effect of 10 nM ouabain on TJ and found that ouabain modifies the molecular composition and distribution pattern of specific claudins and enhances sealing of this junction (as revealed by the paracellular permeability to electrically neutral molecules and transepithelial electrical resistance, TER) [14]. Another clear demonstration of the effect of ouabain on cell-cell contacts is its modulation of ciliogenesis [15], which occurs at the culmination of the process of apical/basolateral polarization [36,37].

To further explore the hypothesis that nanomolar concentrations of ouabain modulate cell-cell contacts, we determined whether 10 nM ouabain acts on another type of cell-cell contact, the gap junction. We found that in fact, ouabain significantly enhances GJC in as early as 10 min (Fig. 1C), peaks in an hour, and lasts for at least 24 h. On these basis we decided to explore the effect of ouabain in the first hour. The observation that inhibitors of transcription and of translation do not affect the induction of GJC by ouabain, indicates that cells express a sufficient level of connexins to account for the observed enhancement of GJC.

Early studies performed with MDCK cells seeded at confluence, and explored with Lucifer Yellow as cell-cell diffusion marker and freeze fracture microscopy to observe the image of gap junctions, have shown that GJC shows an increase between the 9-15 h after plating but disappear thereafter [38], a well known phenomenon in fetal tissues. In keeping with that observation, Berthoud and coworkers [35] found that Cnx-43 expression is relatively high during proliferation, and drastically decreased after MDCK cells reach confluence. This phenomenon may not perturb the results of the present study, as our assays were performed using monolayers that had been confluent for 48 h and that displayed a TER of several hundred ohms per square centimeter.

We found that 10 nM ouabain acts on the Na⁺-K⁺-ATPase, and c-Src and ERK1/2 are necessary to modulates GJC. This modulation was detected after incubation with ouabain for just minutes, which is considerably shorter than the period of incubation required to induce modifications in TJ (hours) and is markedly shorter than the process of ciliogenesis (days), which we explored in previous studies [14,15]. Thus, ouabain regulates both TJ and GJC. However, the mechanisms by which these junctions are regulated are distinct; the inhibition of c-Src and ERK1/2 completely blocks the effect of ouabain on GJC, but its effect on TJ also depends on other signaling pathways. Our results are consistent with those of Wang et al. [33] and Zhang et al. [34], who found that Na⁺-K⁺-ATPase functions as a receptor of ouabain. Taken together, these results demonstrate that ouabain acts on this transmembrane enzyme to independently modulate various types of cell-cell contacts in the following sequence: gap junctions (within approximately 10 min), adherens junctions (8-12 h), tight junctions (24-48 h) and the cell-cell contacts that are involved in ciliogenesis (48-72 h).

When ouabain binds to the Na⁺-K⁺-ATPase in renal epithelial cells, it activates a cytosolic Ca²⁺ oscillatory wave that activates NF-κB and protects cells from serum deprivation-induced apoptosis [29]. In rats that are exposed to malnutrition and serum-starved kidney explants, ouabain activates the signaling of the calcium-dependent transcription factor NF-κB, which protects kidney development from the adverse effects of malnutrition [39,40]. It is conceivable that if cytosolic Ca²⁺ oscillatory waves diffuse across gap junctions, the enhancement of GJC by ouabain described in the present study may aid in the prevention of adverse events during kidney development.

We thank Elizabeth del Oso, E. Méndez, and J. Soriano for their technical and administrative assistance. We acknowledge Dr. K. Soderberg for donating the R-MDCK cells.

Grants: This work was supported by the CONACYT (Mexican Research Council 127329) and ICyTDF (México City Research Council). Isabel Larre was supported by a postdoctoral research fellowship from Becas Mujeres ICyTDF (2012) and financed by Cinvestav, and Adrian Romero and Jacqueline Martínez-Rendón were supported by CONACYT research fellowships.

A. Ponce and I. Larre shared first authorship