Dynamics of E-cadherin and gamma-catenin complexes during dedifferentiation of polarized MDCK cells

Kidney International, Vol. 56 (1999), pp. 910–921

CELL BIOLOGY – IMMUNOLOGY – PATHOLOGY

Dynamics of E-cadherin and g-catenin complexes duringdedifferentiation of polarized MDCK cells

DANIEL F. BALKOVETZ and VIJAYA SAMBANDAM

Department of Veterans Affairs, Medical Center, and Departments of Medicine and Cell Biology, Nephrology Research andTraining Center, University of Alabama at Birmingham, Birmingham, Alabama, USA

Dynamics of E-cadherin and g-catenin complexes during de- including kidney [1]. E-cadherin–mediated epithelial cell-differentiation of polarized MDCK cells. cell adhesion is critical for the development of polarized

Background. E-cadherin mediated cell-cell adhesion and he- apical/basolateral membrane domains [2–4]. Down-regu-patocyte growth factor (HGF) are important for renal epitheliallation of cadherin expression and/or function is associatedmorphogenesis. We previously showed that HGF dedifferenti-with the malignant transformation and invasiveness ofates previously well polarized Madin-Darby canine kidney

(MDCK) cell monolayers grown on filters. The regulation of tumor cells [1]. Dedifferentiation of polarized renal epi-E-cadherin during epithelial dedifferentiation is not known. thelia is an early event during complex processes suchWe hypothesized that E-cadherin mediated cell-cell adhesion as malignant transformation or epithelial regenerationis modulated during HGF induced dedifferentiation of MDCK

following injury. The regulation of E-cadherin mediatedcell monolayers.cell-cell adhesion during dedifferentiation of polarizedMethods. We analyzed E-cadherin/g-catenin interaction and

distribution during epithelial dedifferentiation in vitro using a epithelia is not well understood. Defining mechanismsmodel of polarized MDCK cell monolayers treated with HGF. that influence E-cadherin mediated cell-cell adhesion is

Results. Surface immunoprecipitation experiments showed a key step towards understanding these complex cellularthat HGF increased the amount of cell surface E-cadherinprocesses.associated with g-catenin. Biochemical and morphological ex-

Cytosolic proteins called catenins bind noncovalentlyamination of the TX-100 solubility of junctional E-cadherinand g-catenin in control and HGF treated cells showed an to the cytoplasmic domain of cadherins [5, 6]. Linkageincrease in solubility of only E-cadherin during loss of cell of E-cadherin/catenin complexes to the actin cytoskele-polarity. Metabolic labeling of control and HGF treated cells ton by catenins is thought to be necessary for full cad-showed that HGF stimulated the synthetic rate of E-cadherin

herin function [7, 8]. In addition to their involvement withand g-catenin molecules. Inulin flux across MDCK cell mono-cadherin molecules at the adherens junction, cateninslayers increases with HGF treatment.

Conclusion. These data provide evidence for both the disso- also function in cell signaling [9].ciation of E-cadherin molecules from the actin cytoskeleton Hepatocyte growth factor (HGF) is a polypeptideand an increase in the total number of E-cadherin/g-catenin growth factor with multiple effects on epithelial cells andcomplexes on the cell surface during HGF-induced dedifferen-

tissues, including mitogenesis, cell motility, morphogene-tiation of polarized renal epithelium. These data support thehypothesis that E-cadherin function is inhibited by a mecha- sis, and the development and regeneration of organs [10].nism of detachment from the actin based cytoskeleton during In the kidney HGF appears to play a role in complexHGF induced dedifferentiation of polarized renal epithelia. morphogenetic processes such as development, regenera-

tion following injury and transformation to carcinoma [11].HGF has also been postulated to be a mediator in the

Cadherins are a family of transmembrane glycoproteins pathogenesis of glomerulonephritis [12]. In vitro modelsthat function in cell-cell adhesion via homophilic, Ca21– employed for the characterization of HGF-induced renaldependent interactions between adjacent cells. E-cadherin epithelial morphogenetic events use the MDCK epithe-is important in the morphogenesis of epithelial tissues lial cell line [13]. In the first model, MDCK cells are

grown on impermeant supports as small colonies at alow density. When exposed to HGF these cells dissociateKey words: hepatocyte growth factor, cell adhesion, epithelial morpho-

genesis, renal cell polarity, c-met, injury. and scatter away from the colonies (hence the synonymfor HGF, scatter factor) [14, 15]. In the second model,Received for publication August 25, 1998MDCK cells are grown as hollow cysts in type I collagenand in revised form March 10, 1999

Accepted for publication March 24, 1999 gels. When exposed to HGF they form complex branchingtubules extending out from the cysts [16, 17]. Our labora- 1999 by the International Society of Nephrology

910

Balkovetz and Sambandam: E-cadherin in epithelial dedifferentiation 911

tory has recently standardized a new model for character- USA) [20]. Mouse monoclonal antibodies against E-cadherin (intracellular epitope) and g-catenin were ob-izing HGF-induced morphogenetic events in epithelial

cells using Madin-Darby canine kidney (MDCK) cells tained from Transduction Laboratories (Catalog Num-bers C37020 and C26220, respectively; Lexington, KY,cultured on permeable filter supports. Under these con-

ditions MDCK cells form well polarized monolayers with USA). Secondary antibody for immunofluorescence wasgoat antimouse-FITC from Jackson ImmunoResearchapical and basolateral membrane domains separated by

a tight junction [2]. Using this model, we observed that Laboratories (West Grove, PA, USA). Recombinant hu-man HGF was generously provided by R. Schwall (Gen-HGF decreases the polarity of MDCK cells grown on

filters and leads to morphological changes that are con- entech, South San Francisco, CA, USA). [3H]-inulin waskindly provided by Dr. Paul Sanders (University of Ala-sistent with dedifferentiation [18]. Furthermore, HGF-

treatment of filter-grown MDCK cells also abrogates bama at Birmingham, Birmingham, AL, USA).contact inhibition of mitosis resulting in cell proliferation

Cell surface immunoprecipitation of E-cadherin[19]. During HGF induced dedifferentiation, E-cadherinassociation with immunoprecipitated b-catenin increased Cell surface immunoprecipitation was performed as

previously described [18]. Cells were cultured on Trans-along with the synthetic rates of b-catenin [18]. The pres-ent study focuses on the dynamics of E-cadherin molecules well filters and cooled to 48C in MEM/BSA and exposed

to rr1 conditioned media from the basolateral compart-and their interaction with g-catenin during HGF induceddedifferentiation of polarized MDCK cell monolayers. ment for one hour. After this incubation, cells were

quickly washed three times at 48C with PBS1 and lyzedMorphogenetic changes within sheets of well polarizedepithelial cells are required for complex biological pro- in RIPA buffer as described in the cell lysate preparation

and immunoprecipitation protocol (see below).cesses including epithelial organ regeneration and trans-formation of epithelia into carcinoma. During both of

Cell lysate preparation and immunoprecipitationthese processes there is a loss of epithelial cell polarity.Because of the importance of E-cadherin in formation MDCK cells grown on 24-mm Transwells with or with-

out HGF treatment were rinsed twice with Dulbecco’sof epithelial polarity and its suspected role in morpho-genesis, we have now studied the effect of the HGF phosphate-buffered saline containing Mg21 and Ca21

(PBS1) at 48C. Cells were solubilized in either 0.8 ml ofon E-cadherin localization and interaction with the actincytoskeleton during dedifferentiation of polarized MDCK 20 mm Tris-HCl pH 7.4, 150 mm NaCl, 0.1% SDS, 1%

TX-100, 1% DOC, 5 mm EDTA (RIPA buffer) or 50cells grown on permeable filter supports. The data dem-onstrate that during the loss of epithelial cell polarity, mm NaCl, 10 mm Pipes, pH 6.8, 3 mm MgCl2, 0.5% Triton

X-100, 300 mm sucrose (CSK buffer) each containingthere is an increase in surface expression of E-cadherin/g-catenin complexes and the TX-100 solubility of junc- inhibitors of proteases (2 mm PMSF, 50 mg/ml pepstatin,

50 mg/ml chymostatin and 10 mg/ml antipain) for 15 min-tional E-cadherin is also increased. The latter finding issupportive of a mechanism by which E-cadherin dissoci- utes at 48C on a rocking platform. The cells were scraped

from the filter with a rubber policeman and sedimentedates from g-catenin and the actin cytoskeleton. The datareinforce the importance of E-cadherin interaction with in a 48C microfuge. The soluble and pellet fractions were

collected. The pellet fraction was solubilized as previouslythe actin cytoskeleton in the maintenance of cell polarityand implicate the modulation of these interactions during described [21]. Protein concentration of cell lysates was

determined by Pierce BCA kit. Immunoprecipitationsdedifferentiation of renal epithelial monolayers.were performed as previously described [18, 21]. Briefly,cell lysates were rotated 1.0 hours at 48C with E-cadherin

METHODSor g-catenin mAb (Transduction Laboratories). Immuno-

Cells and reagents complexes were collected with affinity purified rabbitantimouse IgG (Jackson ImmunoResearch Labs) cou-MDCK type II cells were grown in modified Eagle’s

medium (MEM) containing Earle’s balanced salt solution pled to Protein-A Sepharose beads. Immunoprecipitatebeads were washed three times with RIPA buffer, onesupplemented with 5% fetal bovine serum, 100 U/ml

penicillin, 100 mg/ml streptomycin, and 0.25 mg/ml am- time with 140 mm NaCl, 20 mm TEA-Cl, pH 8.6, 5 mmEDTA, pH 8.0, 0.1% Trasylol, and 0.02% NaN3 (FWB)photericin B in 5% CO2/95% air at 378C. Cells grown

on Transwells (Costar, Cambridge, MA, USA) were and samples were eluted by boiling in Laemmli buffercontaining 100 mm dithiothreitol.seeded at confluency. Cell monolayers were used for

experiments after three to four days of culture with dailyElectrophoresis and blottingmedia change. Hybridoma cells secreting mouse antican-

ine renal tubular E-cadherin mAb (rr1), which recog- Immunoprecipitate or cell lysate protein was electro-phoresed on 8% SDS-PAGE minigels and transferrednizes an extracellular epitope, were purchased from De-

velopmental Studies Hybridoma Bank (Iowa City, IA, to Immobilon P filters (Millipore Corp., Bedford, MA,

Balkovetz and Sambandam: E-cadherin in epithelial dedifferentiation912

USA). Filter samples were blocked for one hour at room rescence microscope configured with both an Argon Ion(5 mW, 488 nm) and a Krypton Ion (10 mW, 568 nm)temperature with PBS2, 5% milk, and 0.1% Tween-20

(block solution) and probed with mAb anti-E-cadherin laser.(Transduction Laboratories) diluted 1:1000 in block so-

Metabolic labelinglution or mAb g-catenin (Transduction Laboratories) di-luted 1:2000 in block solution. The filters were washed five Filter-grown cells were starved in MEM lacking cysteine

and methionine supplemented with 5% fetal bovine se-times for five minutes each with PBS2, 0.1% Tween-20(wash solution) and probed with goat-antimouse HRP rum for 20 minutes, then pulse-labeled for 15 minutes in

the same medium supplemented with [35S] cysteine anddiluted 1:30,000 in block solution for one hour. Filterswere washed five times for five minutes each with wash methionine (5 mCi/ml Tran35Slabel) as previously de-

scribed, except all lysates prepared with CSK buffer [18].solution. All filters were visualized on Amersham ECLfilm with an enhanced chemiluminescence kit (ECL, Am- For pulse-chase analysis experiments, pulse-labeled cells

were washed twice with fresh medium containing 10,000-ersham, UK). Autoradiographs were scanned and savedas Adobe Photoshop files with a Microtek Scanmaker II. fold excess cold methionine and cysteine, and chased in

the same medium with and without HGF (100 ng/ml) forFixation, extraction and fluorescent labeling of cells up to 24 hours. 35S incorporation into immunoprecipitated

proteins was analyzed by sodium dodecyl sulfate–poly-MDCK cells grown on 6-mm Transwell filters werefixed with paraformaldehye (PFA). Triton soluble frac- acrylamide gel electrophoresis (SDS-PAGE) and quanti-

tated with a Bio-Rad Molecular Imaging Screen-CS.tions in filter-grown cell monolayers were extracted byincubating the filters in CSK buffer with protease inhibi-

Inulin diffusion measurementstors for 20 minutes at 48C [4]. Extracted and unextractedcells were fixed with ice-cold 4% PFA in PBS1 for 20 To quantitatively measure the functional integrity of

the monolayer we measured apical to basolateral [3H]-minutes. After washing filters three times with PBS1,the cells were quenched with 75 mm NH4Cl and 20 mm inulin leakage across cell cultures grown on 12 mm

Costar Transwells. MEM/bovine serum albumin (BSA)glycine, pH 8.0 with KOH (quench solution) for 10 min-utes at room temperature. Filters were washed one time containing 1.25 3 105 cpm of [3H]-inulin was placed in

the apical compartment and 1 ml of MEM/BSA in thewith PBS1 and permeabilized with PBS1, 0.7% fish skingelatin, and 0.025% saponin (PFS) for 15 minutes at basal well. Cultures were maintained at 378C and aliquots

were collected, 20 ml from the apical side and 40 ml from378C. Filters were labeled with E-cadherin or g-cateninmAb diluted in PFS 1:500 for one hour at 378C. Filters the basal side at 1, 2, 4 and 8 hours after addition of [3H]-

inulin. Aliquots were counted in a liquid scintillationwere then washed four times for five minutes each withPFS at room temperature and then labeled with the FITC counter.conjugated secondary antibody diluted 1:100 in PFS for

Statisticsone hour at 378C. Filters were rinsed four times for fiveminutes each with PFS, one time with PBS1, two times Data are presented as means 6 sd. Each experiment

was performed at least three times. The paired two-with PBS1 containing 0.1% TX-100, and one time withPBS1. Cells were postfixed in 4% PFA for 15 minutes sample for means t-test was used to determine the proba-

bility (P-value) that the sample means are equal. A P 5at room temperature. Filters were cut from the supportwith a scalpel and mounted in Vectashieldt Mounting 0.05 was considered to be significant. Statistical analyses

were carried out with Microsoft EXCEL version 5.0aMedium (Burlingame, CA, USA).The distribution of filamentous actin was revealed with (Microsoft Corporation).

TRITC-labeled phalloidin (Sigma Cat. No. P-1951).Briefly, filter grown cells were fixed for 10 minutes in

RESULTS4% PFA in PBS2. After rinsing once with PBS1, cells

HGF increases cell surface E-cadherin as determinedwere incubated for one minute in 0.2% Triton X-100 inby surface immunoprecipitationPBS2. Cells were then incubated with 2 mm TRITC-

phalloidin for 30 minutes at 378C. Filters were washed We have previously reported that E-cadherin expres-sion on the apical surface of filter-grown MDCK cellsthree times with PBS1 and mounted as described earlier

in this article. increases with HGF treatment and correlates with dedif-ferentiation of the polarized phenotype [18]. Surface bio-

Confocal laser scanning microscopy analysis of tinylation determination of basolateral surface E-cad-fluorescently labeled cells herin levels also increased following a 24 hour exposure

to HGF. We therefore explored the nature of the in-The FITC and TRITC labeled samples were analyzedby an Olympus Fluoview Confocal Laser Scanning Mi- creased levels of surface E-cadherin during HGF induced

dedifferentiation of polarized MDCK cell monolayers.croscope System mounted on a B-MAX50 upright fluo-


expected, and suggested that this population of E-cad-herin was not functional.

g-Catenin binds to classical cadherins within adherensjunction complexes in polarized epithelia. However, therole of g-catenin in cadherin-mediated adhesion is not wellunderstood. We therefore examined the association ofg-catenin with surface E-cadherin during HGF-induceddedifferentiation of polarized MDCK cell monolayers.Western blot of surface E-cadherin complexes with ananti-g-catenin antibody demonstrated a parallel increasein molecules of g-catenin associated with cell surfaceE-cadherin during the dedifferentiation process (Fig. 1).At 72 hours, the amount was 296 6 22% of control(P 5 0.00045, N 5 4). Thus, while surface E-cadherinincreases during dedifferentiation, a greater percentageof the pool of surface E-cadherin is associated withg-catenin.

HGF increases total g-catenin, E-cadherin andg-catenin associated with E-cadherin in totalIPA buffer cell lysates

In order to gain insight into the molecular nature ofthe higher levels of cell surface E-cadherin/g-catenincomplexes during dedifferentiation, we measured theabsolute amounts of E-cadherin, g-catenin and their re-ciprocal interactions by immunoprecipitation of E-cad-herin/g-catenin complexes from RIPA buffer whole celllysates of control MDCK cell monolayers and mono-

Fig. 1. Analysis of hepatocyte growth factor (HGF) effect on surface layers treated with HGF for 24 hours. The results showedexpression of E-cadherin and the interaction of g-catenin with cell surfacethat g-catenin (g-catenin blot, Fig. 2A) levels in the cellE-cadherin by surface immunoprecipitation. MDCK cell monolayers

were treated with HGF (100 ng/ml) for the indicated incubation period. lysates increases with HGF treatment (at 24 hr, theSurface immunoprecipitation of E-cadherin was carried out as described amount was 184 6 14.5% of control; P 5 0.0014, N 5 4).in the Methods section. Immunoprecipitates were separated by SDS-

E-cadherin association with immunoprecipitated g-cate-PAGE, transferred to nitrocellulose, and the presence of E-cadherinor g-catenin was determined by Western blot. nin complexes increased by a similar magnitude (at 24

hr, the amount was 193 6 31% of control; P 5 0.0059,N 5 7). Additionally, these data showed that a higherpercentage of the total E-cadherin pool is associated

Figure 1 shows that increasing exposure periods to HGF with the g-catenin (g-catenin blot, Fig. 2B; at 24 hr, theled to higher levels of cell surface E-cadherin, which amount was 204 6 21% of control; P 5 0.0089, N 5 3).could be immunoprecipitated from the cell surface with Because quantitative E-cadherin immunoprecipitationa monoclonal antibody that recognizes an extracellular was not carried out in these experiments because of eco-epitope of E-cadherin (rr1) [20]. These results show an nomical considerations (that is, cost of the anti-E-cadherinincrease in the amounts of E-cadherin that can be surface antibody), these data were not informative regardingimmunoprecipitated following HGF treatment periods the effect of HGF on the total amounts of E-cadherin.extending to 72 hours (at 72 hr, the amount was 185 6 However, Western blot analyses of equal amounts of33% of control; P 5 0.00009, N 5 5). Because HGF RIPA lysate protein from control and HGF treatedincreases accessibility through tight junctions as well as monolayers showed an increase in both total E-cadherinmislocalizes E-cadherin above tight junctions [18], this (at 24 hr, the amount was 140 6 23% of control; P 5technique allows for the immunoprecipitation of the en- 0.047, N 5 3) and g-catenin (at 24 hr, the amount wastire pool of cell surface E-cadherin. Functional E-cad- 206 6 16% of control; P 5 0.0041, N 5 3; Fig. 3). Collec-herin expression is necessary for the full expression of tively, the immunoprecipitation/Western blot data dem-an epithelial polarized phenotype [22]. The observation onstrated that, during HGF-induced epithelial dediffer-that cell surface E-cadherin increases during the loss of entiation, the amounts of total g-catenin and g-catenin/

E-cadherin complexes increased in the RIPA solublepolarity of MDCK epithelial cell monolayers was not


Fig. 2. Effect of HGF on the association ofE-cadherin with g-catenin in MDCK cells byreciprocal immunoprecipitation. TriplicateMDCK cell monolayers grown on filters weretreated with HGF (100 ng/ml) for a 24-hourincubation period and lyzed in RIPA buffer.(A) Total g-catenin was immunoprecipitatedfrom 100 mg of cell lysate with 0.5 mg of mousemonoclonal g-catenin antibody. (B) A popula-tion of the total E-cadherin pool was immuno-precipitated with 0.5 mg of mouse monoclonalE-cadherin antibody. Immunoprecipitates wereseparated by SDS-PAGE, transferred to ni-trocellulose and blotted with monoclonal anti-bodies directed against both E-cadherin andg-catenin as described in the Methods section.

Fig. 3. Effect of HGF on the amounts ofE-cadherin and g-catenin. Triplicate MDCKcell monolayers grown on filters with and with-out treatment with HGF (100 ng/ml) for a 24-hour incubation period and lyzed in RIPAbuffer. Equal amounts cell lysate protein wereseparated by SDS-PAGE, transferred to ni-trocellulose and blotted with monoclonal anti-bodies directed against both E-cadherin andg-catenin as described in the Methods section.

fraction. Additionally, a higher percentage of the E-cad- based lysis buffer (CSK). The interaction of E-cadherinand catenins with the actin cytoskeleton correlates withherin pool is associated with g-catenin.an increase in resistance to extraction with nonionic de-

HGF increases the levels of E-cadherin and g-catenin tergents such as TX-100 [7, 21]. Because the insolublein the TX-100 soluble pool fraction of E-cadherin is known to be involved in ad-

herens junctions [23–25], we analyzed the solubility prop-To investigate the interaction of E-cadherin and g-cate-nin with the actin cytoskeleton during dedifferentiation, erties of E-cadherin in HGF-stimulated MDCK cell

monolayers. Biochemical analysis of E-cadherin andwe examined the solubility of these molecules in TX-100


Fig. 4. Effect of HGF on the amounts ofE-cadherin and g-catenin in the TX-100 solublefraction. Triplicate MDCK cell monolayersgrown on filters with and without treatmentwith HGF (100 ng/ml) for a 24-hour incuba-tion period and lyzed in CSK buffer. Equalamounts of cell lysate protein were separatedby SDS-PAGE, transferred to nitrocelluloseand blotted with monoclonal antibodies di-rected against both E-cadherin and g-cateninas described in the Methods section.

g-catenin in CSK buffer (TX-100 detergent lysis buffer) described in the Methods section (Fig. 5). Images werecollected under identical confocal microscope settingssoluble fractions by Western blot showed that HGFwith the focal plane set at the middle of the lateral aspecttreatment for 24 hours caused an increase of bothof the cell junctional complexes. In control MDCK cellg-catenin (at 24 hr, the amount was 346 6 76% of control;monolayers, the staining for both E-cadherin and g-cate-P 5 0.0086, N 5 4) and E-cadherin (at 24 hr, the amountnin was primarily peripheral, with some staining in thewas 238 6 47% of control; P 5 0.0012, N 5 6) levels incytoplasm. Following extraction with CSK buffer of con-the TX-100 soluble fraction (Fig. 4). In the insoluble pooltrol cell monolayers, the staining of the peripheral E-cad-(blots not shown), levels of E-cadherin were decreasedherin and g-catenin was very similar to the unextractedfollowing HGF treatment (at 24 hr, the amount was 85 6control monolayers. These results indicate that the ma-5% of control; P 5 0.0075). However, no change wasjority of the E-cadherin and g-catenin at the peripherydetected in the amounts of g-catenin in the insolubleof the control cells is in the adherens junction associatedpool of control and HGF treated cells (at 24 hr, thewith the actin cytoskeleton. In MDCK cells stimulatedamount was 102 6 13% of control; P 5 0.778, N 5 4).with HGF, E-cadherin and g-catenin distribution wasIn cultured epithelial cells, functional adhesion junc-also peripheral in the unextracted cells; additionally,

tional molecules localize to the periphery of cells at thequalitatively higher levels of cytosolic E-cadherin were

regions of cell-cell contact and are not extractable with apparent as compared to control (the morphology of theTX-100 containing buffers [26]. We used immunofluo- cells was also different from control cells as previouslyrescence microscopy to localize the TX-100 soluble and described [18]). Following extraction of the HGF treatedinsoluble fractions of E-cadherin and g-catenin in control cells with CSK buffer, most of the peripheral staining ofand HGF-treated MDCK cell cultures. Cells were either E-cadherin, but not g-catenin, was removed. These datafixed/permeabilized with paraformaldehyde to deter- provided qualitative morphological evidence that duringmine the distribution of total E-cadherin and g-catenin HGF induced dedifferentiation of polarized epithelia theor were first extracted with TX-100 containing CSK peripheral localized, TX-100 insoluble, populations ofbuffer, and then fixed with paraformaldehyde to identify E-cadherin become TX-100 soluble (suggesting dissocia-the distribution of the TX-100-insoluble pool of the mol- tion from the actin based cytoskeleton). In contrast,ecules. Following fixation and extraction, cells were pro- g-catenin remains TX-100 insoluble (suggesting associa-cessed for indirect confocal laser immunofluorescence tion with the actin based cytoskeleton during dedifferen-

tiation). These data are congruent with the results of theusing anti-E-cadherin and antig-catenin antibodies as


Fig. 5. Laser-based confocal immunofluores-cent analysis of the effect of HGF on the TX-100solubility of E-cadherin and g-catenin. MDCKcell monolayers grown on filters with and with-out exposure to HGF (100 ng/ml) treatmentfor 24 hours were labeled with a monoclonalantibody directed against E-cadherin org-catenin as described in the Methods section.Cells were fixed with 4% paraformaldehyde(Total) or extracted with CSK buffer and thenfixed with 4% paraformaldehyde (Extracted)as described in the Methods section. All im-ages were collected with 3100 objective usingidentical exposure times. The focal plane wasset at the middle of the cell junctional com-plexes (bar, 50 mm).

Western blot analysis of E-cadherin and g-catenin in the ton by confocal microscopy of cells labeled with theTRITC-phalloidin (Fig. 6). Phalloidin is a fungal metabo-TX-100 soluble and insoluble pools. One explanation for

the increase in E-cadherin TX-100 solubility by HGF lite that specifically binds actin subunits. The imageswere collected under identical conditions in the sametreatment would be the disruption of the actin-based

cytoskeleton. However, this explanation seems unlikely mid-junctional focal plane and show that both controland HGF-treated cell monolayers have intact subcorticalbecause junctional g-catenin does not become TX-100

soluble with HGF treatment. We also examined the ef- actin cytoskeletal structure. Thus, HGF does not lead todisassembly of the subcortical actin cytoskeleton.fect of HGF treatment on the subcortical actin cytoskele-


Fig. 6. Laser-based confocal immunofluorescent analysis of the effect of HGF on the structure of the subcortical actin cytoskeleton. Actin waslabeled with TRITC-phalloidin. All images were collected with 340 objective using identical exposure times. The focal plane was set on the middleof the cell junctional complexes (bar, 50 mm).

Effect of HGF on the synthetic rates of g-catenin and of increased levels of E-cadherin/g-catenin complexes inE-cadherin in polarized MDCK cell monolayers the TX-100 soluble pool is that during HGF-induced

dedifferentiation the degradation of TX-100 solubleBecause the previous experiments provided evidenceg-catenin and E-cadherin molecules are inhibited. Tothat HGF increased the amount of E-cadherin/g-catenintest this hypothesis, we performed a pulse-chase analysiscomplexes and the amounts of TX-100 soluble E-cad-of E-cadherin and g-catenin molecules in the TX-100herin and g-catenin in MDCK cells, we examined thesoluble pool of control and HGF-treated MDCK celleffect of HGF treatment on the biosynthetic rates ofmonolayers. Using this technique, no difference in theg-catenin and E-cadherin. Polarized monolayers ofapparent half-lives of both E-cadherin and g-cateninMDCK cells were treated with HGF for 1, 3, 6, 12, orcould be detected (data not shown).24 hours, pulse labeled with [35S] cysteine and methionine

for 15 minutes, and then cell lysates were immediatelyHGF treatment of MDCK cell monolayers growncollected. Equal amounts of CSK buffer solubilized pro-on filters increases diffusion of inulin acrosstein were subjected to immunoprecipitation by E-cad-the monolayerherin or g-catenin mAb. Amounts of newly synthesized

If the E-cadherin function is inhibited during HGFE-cadherin and g-catenin in the immunoprecipitatesinduced dedifferentiation, then tight junction integritywere quantitated by PhosphorImager analysis. The re-would be secondarily inhibited. The assembly and main-sults showed that HGF treatment of polarized MDCKtenance of epithelial tight junction function requires thecells increased the synthetic rates of both precursor andpresence of E-cadherin–mediated cell-cell adhesion [20].mature E-cadherin as well as g-catenin molecules (Fig. 7).We examined the effects of HGF on the diffusion ofAfter 24 hours, the apparent synthetic rates of immature[3H] inulin across monolayers of MDCK cells grown onE-cadherin and g-catenin were 184 6 28% of controlfilters as a surrogate marker of cell-cell junction integrity.(P 5 0.0012, N 5 4) and 153 6 16% of control (P 5Figure 8 shows that following 24 and 48 hours of HGF0.0002, N 5 4), respectively. In both cases the effect ontreatment there is an increase in the rate of diffusion ofprotein synthesis was clearly detectable after six hoursinulin from the apical compartment to the basolateralof HGF treatment. This observation suggests that HGFcompartment. These findings are consistent with lossis stimulating the rate of E-cadherin/g-catenin complexof cell-cell junction integrity through a mechanism thatformation in the TX-100 soluble pool, in part, throughrequires inactivation of E-cadherin–mediated cell-cellincreasing E-cadherin and g-catenin synthetic rates dur-adhesion. Alternatively, HGF may directly modify com-ing HGF induced dedifferentiation of polarized epi-ponents of the tight junction complex including ZO-1,thelia.

Another factor that might account for the observation ZO-2, and/or occludin. This effect of HGF on cell-cell


Fig. 7. Effect of HGF on synthetic rates ofE-cadherin and g-catenin found in E-cadherinand g-catenin immunoprecipitates. MDCKcells were grown on filters and treated withHGF (100 ng/ml) for the indicated incubationperiods. Cells were then metabolically labeledwith Translabel as described in the Methodssection. Cells were lyzed in CSK buffer andE-cadherin and g-catenin containing com-plexes were immunoprecipitated from equalamounts of cell protein with 0.5 mg of mousemonoclonal E-cadherin or g-catenin antibody.Immunoprecipitates were separated by SDS-PAGE and transferred to nitrocellulose. Theidentity of each band was confirmed by West-ern blot analysis with specific antibodies. 35Sincorporation into immunoprecipitated proteinswas analyzed by SDS/PAGE and quantitatedwith a Bio-Rad Molecular Imaging Screen-CS.

[16], and dedifferentiation [18]. Decreased cell-cell adhe-sion and junctional disassembly are intuitively requiredfor all three of these processes. Nusrat et al demonstratedthat HGF attenuates transepithelial resistance in non-transformed epithelial cell lines, and speculated thatHGF loosened intercellular junctions between cells soas to facilitate separation, spreading, and migration ofepithelial cells during physiologic processes such aswound healing [27]. Our study provides evidence thatdissociation of E-cadherin from the cytoskeleton andsecondary inactivation is part of the HGF induced signal-ing cascade during epithelial morphology changes in vitro.

In this study, we have demonstrated for the first time,to our knowledge, the dissociation of E-cadherin mole-cules from the actin cytoskeleton based on increasedTX-100 solubility and the stimulation of TX-100 solubleE-cadherin/g-catenin complex formation during HGF in-duced dedifferentiation of renal epithelia. The evidencefor this observation can be summarized as follows: (1)

Fig. 8. Effect of HGF on inulin diffusion across MDCK cell mono- cell surface immunoprecipitation of E-cadherin/g-cateninlayers. MDCK cells were grown on filters and treated with HGF (100

complexes increases during dedifferentiation; (2) immu-ng/ml) in medium or fresh medium alone for (j) control, (h) 24 hoursHGF, ( ) 48 hours HGF, or ( ) Transwell filter/no cells. For the noprecipitation and Western blot analysis of whole celldiffusion assay the apical medium of each filter was replaced with [3H]- lysates demonstrate increased amounts of total g-catenininulin-containing medium and the basal medium was replaced with

and g-catenin associated with E-cadherin; (3) TX-100medium 6 HGF. Aliquots were collected after 0, 1, 2, 4, and 8 hours.Each bar represents mean 6 sd (N 5 3). solubility of E-cadherin, but not g-catenin, increases by

Western blot and immunofluorescent microscopy; and(4) metabolic labeling and experiments show that HGFstimulated synthetic rates of E-cadherin and g-catenin.junctional integrity has previously been demonstrated inCollectively, these data support the hypothesis that duringT84 cells in which there was a decrease in transepithelialepithelial dedifferentiation E-cadherin is inactivated byresistance with HGF treatment [27].a mechanism of dissociation from the actin based cyto-skeleton. During this process levels of g-catenin associ-

DISCUSSION ated with the actin cytoskeleton are not altered, whilelevels of g-catenin in the TX-100 soluble pool are in-Hepatocyte growth factor (HGF) appears to be a ma-creased by HGF. Additionally, the synthetic rates of bothjor factor in the signaling of renal epithelial morphologyE-cadherin and g-catenin are increased. Based on thesechanges [11]. The in vitro epithelial phenotypic changes

elicited by HGF include scattering [14], tubulogenesis insights, we propose a model of E-cadherin inactivation


Fig. 9. Conceptual model of dissociation of E-cadherin molecules from the actin cytoskeleton and increase of free cytoplasmic (TX-100 soluble)E-cadherin/g-catenin molecules during HGF induced epithelial dedifferentiation. The model is subdivided into three phases (A-C) of HGF-induceddedifferentiation. (Phase A) Steady-state situation of well polarized MDCK cell monolayers in which E-cadherin and associated g-catenin areassociated with the actin cytoskeleton. During this phase, the rates of E-cadherin synthesis, degradation, and incorporation into the junctionalcomplex are equal. Following HGF stimulation (phase B), there is post-translational modification of junctional g-catenin molecules (*) leading todissociation of E-cadherin from the junctional complex. At this point there is also an increase in E-cadherin/g-catenin synthetic rates and complexformation in the CSK soluble pool. Ultimately, this results in an increase in cell surface E-cadherin/g-catenin complexes with the majority ofE-cadherin molecules not associated with the actin cytoskeleton and without function (phase C).

by dissociation from the actin cytoskeleton (Fig. 9). This be essential for the initial contact formation betweenepithelial cells and the formation [29] and maintenanceleads to an increase in E-cadherin/g-catenin complexes

not associated with the actin cytoskeleton. This observa- of epithelia during development [30, 31]. g-Catenin (pla-koglobin) acts as a tumor suppressor by augmentingtion suggests that during HGF-induced dedifferentia-

tion, E-cadherin molecules develop a higher affinity for N-cadherin function [32]. The present data provide evi-dence that g-catenin interaction with E-cadherin at theg-catenin molecules as compared to their affinity for

other proteins prior to HGF stimulation. Candidate mol- adherens junction is disrupted. These findings are consis-tent with the hypothesis that HGF treatment results inecules include b-catenin and pp120CAS. These issues are

currently under investigation in our laboratory. post-translational modification of junctional g-catenin,which leads to the dissociation of E-cadherin from theg-Catenin is a member of the armadillo family of pro-

teins (also including b-catenin and p120CAS in vertebrates), junction and its inactivation. The inactivation of E-cad-herin may be important in the dedifferentiation processwhich bind to the cytoplasmic domain of classical cadher-

ins. Catenin mediated linkage of cadherins to the actin that is fundamental for carcinomatous transformationand tumorgenicity. In the absence of HGF, E-cadherincytoskeleton is thought to be essential for the adhesion

function of cadherins [28]. However, the role of g-catenin remains associated with the cytoskeleton via g-catenin(and presumably b-catenin) and the cells remain polar-containing E-cadherin complexes in the formation and

maintenance of adherens junctions is not well defined. ized (differentiated). A popular candidate for the post-translational modification is g-catenin phosphorylationThe dynamics of the E-cadherin pool interacting with

g-catenin is not known. g-Catenin does not appear to (model in Fig. 9 and later in this article).


The mechanism of the increased levels of total E-cad- ACKNOWLEDGMENTSherin and g-catenin appears to be due to synthesis (E-cad- This study was supported in part by grants from the Polycystic

Kidney Research Foundation, the National Kidney Foundation, andherin and g-catenin) and release from the actin cytoskele-the Medical Research Service of the Department of Veterans Affairs.ton (E-cadherin). The metabolic labeling experimentsD.F.B. is a recipient of a Veterans Affairs Career Development Award.

demonstrate that with increasing exposures to HGF, the A preliminary version of this work appeared in abstract form (J AmSoc Nephrol 7:308, 1996). We are grateful to Dr. David G. Warnockrelative rates of formation of mature E-cadherin andfor his advice, support, and encouragement, and to Mr. Robert B.g-catenin increase (Fig. 7). Another consideration is effectPace for providing excellent technical assistance. We also thank Dr.

of HGF on the stability (degradation rates) of E-cad- Marianne L. Egan for thoughtful comments and discussion.herin. However, under the conditions of the present study,

Reprint requests to Daniel F. Balkovetz, M.D., Departments of Medi-we have been unable to detect a change in the half-life cine and Cell Biology, Nephrology Research Training Center, The Uni-of E-cadherin and g-catenin molecules with HGF treat- versity of Alabama at Birmingham, 668 Lyons-Harrison Research

Building, 701 South 19th Street, Birmingham, Alabama 35294–0007,ment (data not shown). We have also previously shownUSA.that HGF stimulates b-catenin synthesis in MDCK cell E-mail: [email protected]

monolayers [18]. Thus, HGF stimulates synthetic ratesof E-cadherin, b-catenin and g-catenin in filter-grown REFERENCESMDCK cell monolayers. This finding is of interest in

1. Takeichi M: Morphogenetic roles of classic cadherins. (Review)light of the recent observations that cytosolic levels of Curr Opin Cell Biol 7:619–627, 1995

2. Rodriguez-Boulan E, Nelson WJ: Morphogenesis of the polar-these molecules increase during MDCK cell mitosis [33]ized epithelial cell phenotype. Science 245:718–725, 1989and that HGF abrogates contact inhibition of mitosis in

3. Nelson WJ, Wilson R, Wollner D, Mays R, McNeill H, SiemersMDCK cell monolayers [19]. Our laboratory is presently K: Regulation of epithelial cell polarity: A view from the cell

surface. (Review). Cold Spring Harbor Symposia on Quantitativeinvestigating the role of adherens junction molecules inBiol 57:621–630, 1992the signaling of contact inhibition of epithelial cell mitosis.

4. Nathke IS, Hinck L, Swedlow JR, Papkoff J, Nelson WJ: Defin-The receptor for HGF is the receptor tyrosine kinase, ing interactions and distributions of cadherin and catenin com-

plexes in polarized epithelial cells. J Cell Biol 125:1341–1352, 1994c-met. c-Met is overexpressed in renal cell carcinomas5. Nagafuchi A, Takeichi M: Transmembrane control of cadherin-[34]. This observation is of interest in light of our obser- mediated cell adhesion: A 94 kDa protein functionally associated

vation that HGF dedifferentiates previously well-polar- with a specific region of the cytoplasmic domain of E-cadherin.Cell Regul 1:37–44, 1989ized renal epithelia. HGF-induced E-cadherin cytoskeletal

6. Ozawa M, Baribault H, Kemler R: The cytoplasmic domaindetachment and inactivation leading to the dedifferentia- of the cell adhesion molecule uvomorulin associates with threetion of renal epithelial monolayers provides a possible independent proteins structurally related in different species.

EMBO J 8:1711–1717, 1989mechanism by which increased c-met levels in vivo might7. Nagafuchi A, Takeichi M: Cell binding function of E-cadherin

predispose kidney epithelia to transform into renal cell is regulated by the cytoplasmic domain. EMBO J 7:3679–3684,1988carcinoma. Tyrosine phosphorylation of the catenins has

8. Ozawa M, Ringwald M, Kemler R: Uvomorulin-catenin complexbeen implicated in the dissociation of the cadherin fromformation is regulated by a specific domain in the cytoplasmic

the actin cytoskeleton and a concomitant decrease in region of the cell adhesion molecule. Proc Natl Acad Sci USA87:4246–4250, 1990intercellular adhesion [35–37]. Activation of the epider-

9. Benzeev A, Geiger B: Differential molecular interactions of betamal growth factor (EGF) and HGF receptor tyrosinecatenin and plakoglobin in adhesion, signaling and cancer. Curr

kinases in transformed intestinal epithelial cell lines in- Opin Cell Biol 10 (5):629–639, 199810. Matsumoto K, Nakamura T: Emerging multipotent aspects ofcreases tyrosine phosphorylation of g-catenin [38]. Dy-

hepatocyte growth factor. (Review) J Biochem 119:591–600, 1996namic regulation of cadherin mediated cell-cell adhesion 11. Balkovetz DF, Lipschutz JH: Hepatocyte growth factor and therequires that the changes by tyrosine phosphorylation be kidney: It is not just for the liver. Internat Rev Cytol 186:225–260,

1999reversible. There is growing evidence that the cadherin-12. Yo Y, Morishita R, Yamamoto K, Tomita N, Kida I, Hayashi S,catenin complex is associated with protein tyrosine phos- Moriguchi A, Kato S, Matsumoto K, Nakamura T, Higaki J,

phatases that may counteract the activity of tyrosine Ohihara T: Actions of hepatocyte growth factor as a local modula-tor in the kidney: Potential role in pathogenesis of renal disease.kinases in regulating adhesion [39–41]. We are currentlyKidney Int 53:50–58, 1998

investigating the phosphorylation status of E-cadherin 13. Balkovetz DF: Hepatocyte growth factor and Madin-Darby ca-nine kidney cells: In vitro models of epithelial cell movement andand associated catenins during HGF induced dedifferen-morphogenesis. Microsc Res Tech 43:456–463, 1998tiation of polarized MDCK cells. Preliminary data using

14. Stoker M, Gherardi E, Perryman M, Gray J: Scatter factor is aseveral commercially available antiphosphotyrosine an- fibroblast derived modulator of epithelial cell mobility. Nature

327:239–242, 1987tibodies have failed to demonstrate modulation of tyro-15. Weidner KM, Behrens J, Vandekerckhove J, Birchmeier W:sine phosphorylation of g-catenin in MDCK cells by Scatter factor: Molecular characteristics and effect on the invasive-

HGF. Future studies will address the phosphorylation ness of epithelial cells. J Cell Biol 111(5 Part 1):2097–2108, 199016. Montesano R, Matsumoto K, Nakamura T, Orci L: Identificationstatus of E-cadherin and associated catenins during HGF

of a fibroblast-derived epithelial morphogen as hepatocyte growthinduced dedifferentiation using metabolic labeling with factor. Cell 67:901–908, 199117. Santos OF, Moura LA, Rosen EM, Nigam SK: Modulation of[32P] orthophosphate.


HGF-induced tubulogenesis and branching by multiple phosphory- protein interactions and their implications for cadherin function.J Cell Biochem 61:514–523, 1996lation mechanisms. Dev Biol 159:535–548, 1993

29. Adams CL, Nelson WJ, Smith SJ: Quantitative analysis of cad-18. Balkovetz DF, Pollack AL, Mostov KE: Hepatocyte growthherin catenin actin reorganization during development of cell cellfactor alters the polarity of Madin Darby canine kidney cell mono-adhesion. J Cell Biol 135(6 Part 2):1899–1911, 1996layers. J Biol Chem 272:3471–3477, 1997

30. Ruiz P, Brinkmann V, Ledermann B: Targeted mutation of pla-19. Balkovetz DF: Evidence that hepatocyte growth factor abrogateskoglobin in mice reveals essential functions of desmosomes in thecontact inhibition of mitosis in Madin-Darby canine kidney cellembryonic heart. J Cell Biol 135:215–225, 1996monolayers. Life Sci 64:1393–1401, 1999 31. Bierkamp C, McLaughlin KJ, Schwarz H, Huber O, Kemler R:

20. Gumbiner B, Simons K: A functional assay for proteins involved Embryonic heart and skin defects in mice lacking plakoglobin.in establishing an epithelial occluding barrier: Identification of a Dev Biol 180:780–785, 1996uvomorulin-like polypeptide. J Cell Biol 102:457–468, 1986 32. Simcha I, Geiger B, Yehuda-Levenberg S, Salomon D, Ben-

21. Hinck L, Nathke IS, Papkoff J, Nelson WJ: Dynamics of cadherin/ Ze’ev A: Suppression of tumorigenicity by plakoglobin: An aug-catenin complex formation: Novel protein interactions and path- menting effect of N cadherin. J Cell Biol 133:199–209, 1996ways of complex assembly. J Cell Biol 125:1327–1340, 1994 33. Bauer A, Lickert H, Kemler R, Stappert J: Modification of the

22. Watabe M, Nagafuchi A, Tsukita S, Takeichi M: Induction of E cadherin catenin complex in mitotic Madin Darby canine kidneyepithelial cells. J Biol Chem 273:28314–28321, 1998polarized cell cell association and retardation of growth by activa-

34. Natali PG, Prat M, Nicotra MR, Bigotti A, Olivero M, Comog-tion of the E cadherin catenin adhesion system in a dispersedlio PM, Di Renzo MF: Overexpression of the met/HGF receptorcarcinoma line. J Cell Biol 127:247–256, 1994in renal cell carcinomas. Int J Cancer 69:212–217, 199623. Cowin P, Kapprell HP, Franke WW, Tamkun J, Hynes RO:

35. Hoschuetzky H, Aberle H, Kemler R: Beta catenin mediatesPlakoglobin: A protein common to different kinds of intercellularthe interaction of the cadherin catenin complex with epidermaladhering junctions. Cell 46:1063–1073, 1986growth factor receptor. J Cell Biol 127:1375–1380, 199424. Hirano S, Nose A, Hatta K, Kawakami A, Takeichi M: Calcium

36. Ochiai A, Akimoto S, Kanai Y, Shibata T, Oyama T, Hirohashidependent cell cell adhesion molecules (cadherins): Subclass speci- S: c-erbB-2 gene product associates with catenins in human cancerficities and possible involvement of actin bundles. J Cell Biol 105(6 cells. Biochem Biophys Res Commun 205:73–78, 1994Part 1):2501–2510, 1987 37. Kinch MS, Clark GJ, Der CJ, Burridge K: Tyrosine phosphoryla-

25. McNeill H, Ryan TA, Smith SJ, Nelson WJ: Spatial and temporal tion regulates the adhesions of ras-transformed breast epithelia. Jdissection of immediate and early events following cadherin medi- Cell Biol 130:461–471, 1995ated epithelial cell adhesion. J Cell Biol 120:1217–1226, 1993 38. Shibamoto S, Hayakawa M, Takeuchi K, Hori T, Oku N, Miya-

26. Pasdar M, Li Z, Marreli M, Nguyen BT, Park M, Wong K: zawa K, Kitamura N, Takeichi M, Ito F: Tyrosine phosphoryla-tion of beta-catenin and plakoglobin enhanced by hepatocyteInhibition of junction assembly in cultured epithelial cells by hepa-growth factor and epidermal growth factor in human carcinomatocyte growth factor/scatter factor is concomitant with increasedcells. Cell Adhes Commun 1:295–305, 1994stability and altered phosphorylation of the soluble junctional mol-

39. Brady KS, Rimm DL, Tonks NK: Receptor protein tyrosine phos-ecules. Cell Growth Differ 8:451–462, 1997phatase PTPmu associates with cadherins and catenins in vivo. J27. Nusrat A, Parkos CA, Bacarra AE, Godowski PJ, Delp AC,Cell Biol 130:977–986, 1995Rosen EM, Madara JL: Hepatocyte growth factor/scatter factor

40. Kypta RM, Su H, Reichardt LF: Association between a trans-effects on epithelia. Regulation of intercellular junctions in transmembrane protein tyrosine phosphatase and the cadherin cateninformed and nontransformed cell lines, basolateral polarization of complex. J Cell Biol 134:1519–1529, 1996c-met receptor in transformed and natural intestinal epithelia, and 41. Balsamo J, Leung T, Ernst H, Zanin MK, Hoffman S, Lilieninduction of rapid wound repair in a transformed model epithelium. J: Regulated binding of PTP1B like phosphatase to N cadherin:J Clin Invest 93:2056–2065, 1994 Control of cadherin mediated adhesion by dephosphorylation of

beta catenin. J Cell Biol 134:801–813, 199628. Aberle H, Schwartz H, Kemler R: Cadherin catenin complex

Documents

Dynamics of E-cadherin and gamma-catenin complexes during dedifferentiation of polarized MDCK cells