Bestrophin 1 Promotes Epithelial-to-mesenchymal Transition of Renal Collecting Duct Cells

Bestrophin 1 (Best1) controls intracellular Ca 2 (cid:1) concentration, induces Ca 2 (cid:1) -activated Cl - conductance, and increases proliferation of colon carcinoma cells. Here, we show that expression of Best1 in mouse renal collecting duct (CD) cells causes i ) an increase in cell proliferation, ii ) a loss of amiloride-sensitive Na (cid:1) absorption, iii ) induction of Ca 2 (cid:1) -dependent Cl - conductance (CaCC), and iv ) epithelial-to-mesen-chymal transition. During conditions of high proliferation or when we exposed CD cells to serum or TGF– (cid:1) 1, we observed upregulation of Best1, increased CaCC, redistribution of the epithelial-to-mes-enchymal transition marker (cid:1) -catenin, and upregulation of vimentin. In contrast, suppression of Best1 by RNAi inhibited proliferation, reduced CaCC, and downregulated markers of EMT. CaCC and expression of Best1 were independent of the cell cycle but clearly correlated to cell proliferation and cell density. During renal inflammation in LPS-treated mice or after unilateral ureteral obstruction, we observed transient upregulation of Best1. These data indicate that repression of cell proliferation, CaCC, and expression of Best1 occurs during mesenchymal-to-epithelial transition once CD cells polarize and terminally differentiate. These results may suggest a role for Best1 in renal fibrosis and tissue repair.

Principal cells of the renal collecting duct (CD) reabsorb Na ϩ through epithelial Na ϩ channels (ENaC), while stimulation of luminal purinergic receptors enhances intracellular Ca 2ϩ and inhibits ENaC by means of hydrolysis of phosphatidylinositol bisphosphate (P1P2). [1][2][3][4] In intact micro-perfused CDs, no evidence was found for a luminal Ca 2ϩ -dependent Clconductance (CaCC). 2,5 In contrast, CaCC was readily detectable in M1 mouse CD cells and primary cultured renal epithelial cells. 6,7 This raises the question as to what degree available CD culture models represent native renal tubular function. Discrepancies between properties in intact tubules and isolated epithelial cells are well recognized and may be caused by epithelial-to-mesenchymal transition (EMT) occurring during primary cell culture.
Transition from an epithelial to a mesenchymal phenotype during cell culture of epithelial cells is well known and has been examined previously. 8 Also during development, both EMT and the reverse process, namely mesenchymal-to-epithelial transition (MET) are fundamental processes. EMT contributes to degeneration of mature epithelial structures and to generation of fibroblasts associated with accumulation of extracellular matrix during renal inflammation (reviewed in 9 ). EMT also takes place during morphogenesis, wound healing and tissue repair, and is observed during tumor progression. The most important stimulus for EMT is TGF-␤. 9,10 In fact, TGF-␤ is able to induce a genetic program of cell plasticity that involves key pathways and regulators of epithelial dedifferentiation, cytoskeletal reorganization, and proliferation. 11 We recently detected an EMT-like process in T 84 colonic carcinoma cells, which induced enhanced proliferative activity of this cell line. 12 Notably, ion channels, such as the cell cycle regulated hEag1 potassium channel as well as the putative Ca 2ϩ -dependent Clchannel bestrophin 1 (Best1), were upregulated during EMT. Both channels supported proliferation in fast growing T 84 cells. 12 Best1 was shown to form a Ca 2ϩ -activated Clchannel but, surprisingly, also regulates intracellular Ca 2ϩ signaling. [13][14][15] We identified Best1 as a component important for the activation of endogenous Ca 2ϩ -activated Clchannels in airway cells and other epithelial tissues. 16 -18 A novel family of Ca 2ϩ activated Clchannels (TMEM16) has been identified recently, with biophysical properties strikingly similar to endogenous CaCC. 19 -21 Strong evidence was presented that TMEM16A is, in fact, the endogenous Ca 2ϩ -activated Clchannel, and studies are underway to examine the role of Best1 in activation of TMEM16A. 19 -21 Here, we asked whether expression of Best1 is upregulated during dedifferentiation of collecting duct cells, and if Best1 induces proliferation, EMT, and enhanced Ca 2ϩ -activated Clconductance.

Coincidence of CaCC and EMT in CD Cells
M1 CD cells grown on permeable supports for 3 d in the presence of 10% bovine serum formed tight monolayers (R te ϭ 228 Ϯ 23 ⍀cm 2 ; n ϭ 37), when mounted in a perfused micro Ussing chamber, and expressed amiloride sensitive Na ϩ channels (ENaC), as indicated by amiloride (10 M)-induced positive voltage deflection ( Figure 1A). Application of luminal ATP (100 M) activated a luminal Ca 2ϩ dependent Clconductance (CaCC; negative voltage deflection) and inhibited ENaC. When M1 cells were grown for 12 d in the absence of serum, they formed higher resistance monolayers (R te ϭ 874 Ϯ 56 ⍀cm 2 ; n ϭ 28) and showed augmented amiloride sensitive short circuit currents (I sc-Amil ). Under these conditions, ATP no longer activated CaCC but only induced a transient K ϩ secretion, as indicated by positive voltage deflections ( Figure 1, B through D). As these properties are similar to those of native CD, 2 M1 cells may undergo MET upon polarization of filter membranes. In fact, expression of the marker for EMT, vimentin, was reduced in polarized cultures, while E-cadherin expression was enhanced (Figure 1, E and F). In contrast, acute application of 1% serum (FBS) to polarized M1 cells inhibited I sc-Amil and enabled ATP to activate CaCC ( Figure 2). Thus, proliferation and loss of polarization of collecting duct cells correlates with upregulation of CaCC and probably with EMT.

Induction of EMT Enhances Expression of Best1 and CaCC
TGF-␤1 has been described as a major factor for EMT. 9 We incubated highly polarized M1 cells with TGF-␤1 (5 ng/ml, 5 h) and observed a remarkable decrease in amiloride sensitive Na ϩ absorption and induction of Ca 2ϩ -dependent Clsecretion, similar to the treatment with FBS ( Figure 3, A through D). Inhibition of ENaC by TGF-␤1 was slightly attenuated by 15.3 Ϯ 2.1% (n ϭ 5) upon treatment with siRNA for Best1. Notably, incubation with TGF-␤1 for 6 h upregulated expression of Best1 in CD cells (Figure 3E). In contrast, other cytokines like IL1-␤ (100 ng/ml), IL-8 (10 ng/ml), IFN-␥ (100 ng/ml), and TNF-␣ (100 ng/ml) did not activate CaCC but abolished amiloride sensitive transport as reported earlier (data not shown). 22 Moreover, activation of CaCC by TGF-␤1 was not affected by inhibition of PI3K (Ly294002; 10 M), MAPK kinase or Erk1,2 (SB203840; 25M, U0126; 5 M), or interfering with the NO pathway (SNAP; 30 M, L-NAME; 100 M) (data not shown). Redistribution and cytosolic location of ␤-catenin is a hallmark of EMT and is often observed during tumor invasion. 25 Redistribution of ␤-catenin to the cytosol was also observed in proliferative M1 cells (70% density) or after treat- BASIC RESEARCH www.jasn.org ment with FBS or TGF-␤1 ( Figure 3F). Moreover, exposure to FBS induced expression of the EMT marker vimentin but decreased E-cadherin expression (data not shown). Thus, TGF-␤1 promotes EMT, increased Best1-expression, and activation of CaCC.
Glucocorticoids such as dexamethasone are known to induce expression of epithelial Na ϩ channels and have been reported to reduce cell proliferation. 26 We compared transport properties of M1 cells that had been grown in the presence of low  Figure 4F). I sc -ATP in dexamethasone-treated cells was due to a Ba 2ϩ -sensitive K ϩ conductance, while I sc -ATP in the absence of dexamethsone was due to a DIDS-sensitive Clconductance ( Figure 4, G and H). Western blot analysis of Best1 clearly indicated downregulation of Best1-expression by dexamethasone ( Figure 4I).

CaCC and Best1 Are Expressed Only in Highly Proliferating CD Cells
Best1 has been shown previously to induce both proliferation and CaCC in colonic carcinoma cells. 12 We found pronounced Best1expression in cytosolic compartments of highly proliferative M1 cells at a cell density of approximately 70%, which was almost completely downregulated in confluent monolayers ( Figure 5A). Expression of Best1 reached a peak during the log phase of proliferation but was turned off after cells reached confluence ( Figure  5B). Patch clamp experiments indicated largely reduced activation of CaCC at high cell density, indicating a clear correlation  between cell density, Best1 expression, and CaCC ( Figure 5, C and D). Notably, the reduced ATP response in confluent monolayers was not due to a loss of P2Y 2 receptors (Supplement 1A).

Bestrophins Support Proliferation of M1 Cells
To further examine whether Best1 affects proliferation of CD cells, we used RNAi to suppress expression of Best1 and the paralog Best2, respectively ( Figure 6A). Cell proliferation was assessed by BrdU incorporation and cell counting, and was suppressed by approximately 40% with either Best1-RNAi or Best2-RNAi ( Figure 6, A and C). Notably, expression of the EMT marker vimentin was reduced in Best1-siRNA treated cells ( Figure 6B). No evidence was found for induction of apoptosis by RNAi, as indicated by a lack of caspase-3 cleavage ( Figure 6D). Activation of CaCC by ATP (I sc-ATP ) was attenuated in RNAi-treated cells, as demonstrated in Ussing chamber recordings (Figure 7, A and B). This was further confirmed by reduced iodide-quenching of the fluorescence signal from the halide sensitive dye YFP, in Best1-RNAi-and Best2-RNAitreated cells (Figure 7, C and D). In additional experiments, overexpression of Best1 increased cell proliferation by 18 Ϯ   (n ϭ 3). Taken together, bestrophins support proliferation of CD cells and are essential for generating a Ca 2ϩ activated Clconductance. Notably, common CaCC (and Best1) inhibitors, such as DIDS, also inhibited cell proliferation by approximately 30%, supporting the role of Best1 and CaCC for cell proliferation (Supplement 1B).
Since expression of Best1 was largely dependent on cell proliferation, we asked whether expression of Best1 was cell cycledependent. Yet synchronization of the cells into early G1, G1/S, and M-phase, 12 did not reveal any major changes in Best1-expression or CaCC during cell cycling (Supplement 2).
Staining of Best1 during early G1, G1/S, and M-phase detected a perinuclear staining during eG1, which changed into a cytosolic staining during G1/S (Supplement 3A). This finding was supported by western blotting, which showed an increase of Best1 in the perinuclear fraction of the cell lysate during eG1 (Supplement 3B). Expression of Best1 is Controlled by Cell Density and a Soluble Factor and Upregulated during Tissue Repair As pointed out, expression of Best1 is downregulated in dense and polarized monolayers. Notably, M1 cells at the edge of a cell island demonstrated particularly strong staining for Best1 ( Figure 8A). ␤-catenin staining was typically broad, but M1 cells changed to a membrane restricted ␤-catenin expression when grown in a media that was substituted by 50% media from confluent monolayers ( Figure  8B). Under these conditions both cell proliferation and Best1-expression were largely reduced ( Figure 8C). In contrast, when grown in a medium that was frequently refreshed (2 times per day) Best1-expression was no longer downregulated when reaching confluence and proliferation was enhanced ( Figure  8D).
These results suggest that EMT is induced by soluble factors such as TGF-␤1, while MET occurs during polarization possibly engaging autocrine mechanisms. EMT in renal epithelial cells could be important for the replacement of aged cells and during posttraumatic tissue repair. In fact, in a scratch assay using densely grown M1 cells, we detected upregulation of Best1 in the proliferating cells facing the tissue defect ( Figure 9A). Moreover intraperitoneal injection of LPS, which is known to induce renal inflammation and inhibition of the epithelial Na ϩ channel ENaC, 22,27 caused upregulation of tubular Best1 expression and increase in Best1-mRNA as detected by real-time RT-PCR ( Figure 9, B and C). Western blot analysis confirmed a transient increase in renal Best1-expression in LPS-treated animals, while expression of other ion channels such as CFTR did not change (Supplement 4A). In another model for renal damage using unilateral ureter obstruction (UUO), we also found significant upregulation of Best1 by 65 Ϯ 8.2%, as demonstrated by semiquantitative RT-PCR (n ϭ 5 animals for each series) ( Figure 9D). These data indicate that Best1 is upregulated during EMT transition. Since EMT is considered a contributor of renal fibrosis, attenuating expression or function of Best1 could be of therapeutic potential.

DISCUSSION
Here, we supply evidence that expression of Best1 in mouse renal collecting duct (CD) cells increases cell proliferation, causes Ca 2ϩ dependent Clconductance and EMT, but is inversely correlated to amiloride-sensitive Na ϩ absorption. The data indicate that collecting duct cells retain their ability to express Ca 2ϩ activated Cl -channels. Thus, in acutely isolated intramedullary CD cells, and various CD cell lines such as IMCD-K2, IMCD-K3, M1, and MD-CKII, nucleotide-sensitive CaCC were identified. 6,28 -30 This is in contrast to data obtained from isolated perfused CD tubules that are regarded as the gold standard for epithelial transport studies. In isolated perfused CD tubules no evidence was found for a significant role of CaCC in either the luminal or basolateral membrane. 2,5,31,32 The present results suggest that CD cells in culture undergo a rapid dedifferentiation program that can be characterized as EMT. Depending on the growth conditions, M1 cells transform either into rapidly proliferating nonpolarized cells or into a highly polarized nonproliferating epithelium. Thus, M1 cells undergo MET when grown under polarized conditions in a serum-free medium substituted with dexamethasone.
Notably, glucocorticoids and aldosterone-induced transcription factors such as ATF3 impinge directly on the activity of transport proteins and cellular differentiation. [33][34][35][36] Interestingly, a related transcription factor, ATF2, was shown to be activated through TGF-␤. 37 In the present study, we found that dexamethasone is able to antagonize the effects of pro-inflammatory cytokines, such as TGF-␤1. EMT is a process whereby renal tubular cells lose their epithelial characteristics and gain a fibroblast like phenotype. This process is induced by TGF-␤ mediated activation of the transcription factor E2A and has a potential role in renal fibrosis. 38 However, as cellular responses to TGF-␤ are variable, TGF-␤ may cause either growth arrest or EMT, depending on the cell type and context of stimulation.
TGF-␤1 can signal through NFB to suppress Na ϩ transport in principal cells of the CD by downregulating the serum and glucocorticoid-regulated kinase SGK1, apart from other mechanisms for inhibition of ENaC in renal and airway epithelial cells. 27,39,40 Notably, membrane expression of TGF-␤ strongly depends on cell polarization, which may explain why in the present study only cells at the edge of a monolayer expressed Best1 and proliferated, since they are able to expose their basolateral TGF-␤ receptors to serum factors. 41 Similar observations were done in a colonic carcinoma cell line (T 84 ), that changed from a polarized growth pattern to fast growing cells with augmented intracellular Ca 2ϩ signaling. 12 Thereby, ER-localized bestrophins may enhance cell proliferation and couple receptor mediated intracellular Ca 2ϩ signaling to membrane localized Ca 2ϩ -dependent ion channels such as TMEM16A. 18 -21 As Ca 2ϩ activated Clchannels and probably Best1 expression are common features of epithelial cells undergoing dedifferentiation during primary cell culture, 42 the present results suggest a novel role for both during EMT and probably embryonic development (Supplement 4B).

Animals, Cell Culture, cDNAs, siRNAs, and Transfection
Mice (C57B6) received NaCl (control) or LPS (Escherichia coli, serotype 0111:B4; Sigma, St. Louis, MO) (10 mg/kg intraperitoneal), and were killed 6, 12, and 24 h (n ϭ 5 per group) after injection. 22 Mice were treated Mack (University of Regensburg, Germany) as described previously. 23 M1 cells (mouse collecting duct cells) were grown at 37°C in 5% CO 2 as described previously. 6 M1 cells were grown on permeable supports or  proliferation and Best1-expression in cells grown in media that was frequently (twice per day) replaced (n ϭ 2 for each series). plastic dishes to highly polarized monolayers in the absence or presence of serum (FBS) or TGF-␤1. To verify EMT, cells were stained for ␤-catenin and semiquantitative RT-PCR was performed to examine expression of vimentin and E-cadherin, using ␤-actin as an internal standard. We purchased three different batches of duplexes of 25-nucleotide RNAi from Invitrogen (Karlsruhe, Germany). The sense strands of the RNAi used to silence the Best1 gene were 5Ј-UCCAGUACAUGUCACGUUC-CAUGGG-3Ј, and for Best2 gene were 5Ј-UUUGCUCCGA AACA-CAUUCAGCUCC-3Ј, scrambled sequence RNAi ds-oligomer, not homologous to any known gene (BLOCK-ITTM Fluorescence Oligo) served as control. We transfected M1 cells using Lipofectamine™2000 (Invitrogen).
Immunohistochemistry M1 cells were fixed for 2 h with 4% paraformaldehyde in 0.1 M cacodylate buffer, pH 7.4. Sections were incubated overnight at 4°C with rabbit anti-mouse Best1 antibodies diluted 1:10,000 in Tris buffer containing Triton X-100 (0.8%) and goat serum to prevent nonspecific binding. Sections were incubated with horseradish peroxidase linked goat anti-rabbit secondary antibodies (Amersham Pharmacia Biotech, Freiburg, Germany). The avidin biotin peroxidase complex (ABC) technique was used to visualize the labeling with 3,3-diaminobenzidine. The ABC technique involves application of a biotin-labeled secondary antibody followed by the addition of the ABC. 24 Sections were counterstained with MayerЈs hematoxylin. Alternatively a fluorescence-labeled secondary antibody was used (diluted 1:1,000; Acris Antibodies, Hiddenhausen, Germany).

Patch Clamping, Ussing Chamber, and Iodide Quenching
We performed patch-clamp experiments in the fast whole-cell configuration as described recently. 12 In intervals, membrane voltages (V c ) were clamped in steps of 10 mV from Ϫ50 to ϩ 50 mV relative to resting potential. We calculated the membrane conductance G m from the measured current (I) and V c values according to Ohm's law. M1 cells grown to confluence on permeable supports (Millipore, Schwalbach, Germany) were mounted into a perfused micro Ussing chamber. The luminal and basolateral surfaces of the epithelium were perfused continuously with buffer solution at a rate of 5 to 10 ml/min (chamber volume 2 ml). We carried out all experiments at 37°C under open circuit conditions. We determined transepithelial resistance (Rte) by applying short (1-s) current pulses (⌬I ϭ 0.5 A) and continuously recorded the corresponding changes in transepithelial voltage (Vte) and basal Vte. Values for the Vte were referred to the serosal side of the epithelium. We calculated Rte according to Ohms law (Rte ϭ ⌬Vte/⌬I) and the equivalent short-circuit current (Isc) according to Ohms law from Vte and Rte (Isc ϭ Vte / Rte). We induced YFPI152L fluorescence by excitation (wave length 500 nm) using an poly-chromatic illumination system for microscopic fluorescence measurement (VisiChrome; Visitron Systems, Puchheim, Germany) and light emission was measured at 535 Ϯ 15 nm with a photomultiplier detector (SF, Zeiss, Munich, Germany). We induced quenching of YFPI152L fluorescence via Iinflux by replacing 20 mM extracellular Clwith I -.

Proliferation Assays and Statistics
We seeded M1 cells at a density of 5000 cells/0.35 cm 2 on 96-well plates (Sarstedt, Nuembrecht, Germany). We quantified cell proliferation by measuring 5-bromo-2Ј-deoxyuridine (BrdU) incorporation and performed cell counting as described recently. 12 To measure the effect of RNAi treatment on cell proliferation, we seeded M1 cells at a density of 10,000 cells/0.35 cm 2 in 96-well plates and treated wells with the siRNA. We used BLOCK-IT TM Fluorescence Oligos as controls. After 48 h, we quantified cell proliferation by measuring BrdU incorporation and by cell number. All compounds used were of the highest available grade of purity and were from Sigma (Taufkirchen, Germany) or Calbiochem (Darmstadt, Germany). The t test (for paired or unpaired samples as appropriate) was used for statistical analysis. P Ͻ 0.05 was accepted as significant.