inhibition ofjunction assembly incultured epithelial...

Vol. 8, 451-462, Apr11 1997 Cell Growth & Differentiation 451

Inhibition of Junction Assembly in Cultured Epithelial Cells byHepatocyte Growth Factor/Scatter Factor Is Concomitantwith Increased Stability and Altered Phosphorylationof the Soluble Junctional Molecules1

Manijeh Pasdar,2 Zhi Li, Marcello Marreli,Bao T. Nguyen, Morag Park, and Karen Wong

Department of Anatomy and Cell Biology, University of Alberta,Edmonton, Alberta, T6G 2H7 Canada [M. Pas., Z. L, M. M., B. T. N.,K. W.], and Molecular Oncology Group, Royal Victoria Hospital, andDepartments of Oncology, Medicine, and Biochemistry, McGillUniversity, Montreal, Quebec, H3A 1A1 Canada [M. Par.]

AbstractHepatocyte growth factor/scatter factor (HGF/SF) is amesenchymally derived glycoprotein with a strong

scattering effect on epithelial cells. A receptor tyrosinekinase encoded by the met proto-oncogene has beenidentified as the cellular receptor for HGF/SF.Following stimulation with HGF/SF, cell scatteringoccurs concurrent with decreased cell-cell adhesionand disassembly of junctional components. In cufture,junction formation is cell-cell contact dependent and

can be regulated by modulating the Ca2�concentrations of the growth media. Decreasing theCa2� concentrations below 50 �tM causes rapiddisassembly of junctions, whereas increasing the Ca2�concentrations to I .8 m�.i induces cell-cell contact andjunction assembly. Although associated withdecreased cell-cell adhesion and disassembly of thejunctional complex, HGF/SF-induced scattering occursunder high extracellular Ca2� concentrations. To gaininsight into the mechanisms of HGF/SF-inducedscattering of epithelial cells, we have studied theeffect(s) of HGF/SF on junction assembly by examiningthe solubility, stability, phosphorylation, and subcellularlocalization of the major components of the adheringjunctions, plakoglobin (Pg) and E-cadherin, in Madin-Darby canine kidney (MDCK) epithelial cells and in aMDCK cell line expressing an exogenous chimeric metreceptor (CSF-MET) that scatters in response to

colony-stimulating factor I (CSF-1). The resufts have

Received 7/24/96; revised 1 1/27/96; accepted i/i 7/97.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to mdi-cate this fact.1 This research is supported by a grant from the Medical Research Coun-cil of Canada (to M. Pas.). The University of Alberta Faculty of MedicineConfocal Laser Scanning Microscopy Facility is supported in part byfunds from the Medical Research Council and the Alberta Heritage Foun-dation for Medical Research. M. Pas. is an Alberta Heritage Foundation forMedical Research scholar. M. Par. is a senior scholar of the NationalCancer Institute of Canada.2 To whom requests for reprints should be addressed. Phone: (403) 492-3356; Fax: (403) 492-0450; E-mail: [email protected].

shown that in HGF/SF-stimulated MDCK cells, adheringjunctions were not assembled upon induction of cell-cell contact. Immunofluorescence analyses showedthat larger amounts of Pg and E-cadherin were Tritonx-100 extractable, and more significantly, theseproteins were homogeneously distributed along themembrane and were not concentrated at the areas ofcell-cell contact. Similar results were obtained for CSF-MET expressing MDCK cells in response to CSF-1 . Incontrast, none of the above effects were detected in

MDCK cells expressing a mutant CSF-MET chimeracontaining a phenylalanine substitution at tyrosine 1356in met, which fails to scatter in response to CSF-1.When compared with the unstimulated cells, theinhibition of cell adhesion promoted by HGF/SF

correlated with an increased stability of the newlysynthesized soluble E-cadherin and Pg and an alteredphosphorylation pattern of E-cadhenn, as determinedby partial proteolytic peptide mapping.

IntroductionCell dissociation and motility are fundamentally important to

normal embryogenesis and tissue remodeling as well as toinvasiveness of tumor cells (1). HGF/SF3 is a secretory gly-

coprotein of mesenchymal origin that has potent motogenic,mitogenic, and morphogenetic effects on epithelial and en-

dothelial cells (recently reviewed in Refs. 2 and 3). HGF/SF isa mitogen for hepatocytes, primary renal tubule cells, andmelanocytes in culture and has been shown to promote

cellular invasiveness and tumor progression (4-8). The cel-lular receptor for HGF/SF has been identified as the product

of the met proto-oncogene, which is a member of the RTK

family (5, 9, 10). Recent studies suggest that stimulation of

the signal transduction pathways involving HGF/SF and its

receptor met may be an important event for the control of cell

differentiation, proliferation, and morphogenesis during em-

bryonic development (1 1-1 4). The HGF/SF receptor is first

expressed in epithelial organs during early stages of theirdevelopment, whereas the ligand is expressed in the sur-rounding mesenchyme (1 3, 1 5). In vitro (depending on targetcells and culture conditions) activation of the met receptor byHGF/SF elicits different responses in epithelial cells including

cell dissociation and scattering with loss of adhesion andjunctional communication (reviewed in Refs. 8 and 16-18).

3 The abbreviations used are: HGF/SF, hepatocyte growth factor/scatterfactor; ATK, receptor tyrosine kinase; Pg, plakoglobin; Dsg, desmoglein;Dsc, desmocollin; HCM, high Ca2� medium; LCM, low Ca2� medium;CSF, colony-stimulating factor.

PM

0MET I HGFR

II C5F-1R

III MET-C5F

IV Y1356F

Fig. 1 . Schematic diagram of chimeric wild-type and mutant receptorsconstructed from cDNAs encoding the human CFS-1 and p19o��t (fordetails, see Aefs. 39 and 41). Chimeric receptors (iii and i� were con-structed by fusing the extracellular domain of CSF-i receptor (II) to thetransmembrane and intracellular domain of p1 9fP� (0. The position oftyrosine 1356 (V) and its mutation to phenylalanine (F) is indicated in thewild-type (iii) and mutant (il’) qhimeras, respectively. PM, plasma mem-brane; a and 13,subunits of the MET/HGF receptor.

452 Inhibition of Junction Assembly by HGF/SF

Cell-cell adhesion in epithelia is regulated by cadherins, afamily of Ca2�-dependent cell-adhesion molecules (19). E-

cadherin initiates the early recognition and contacts between

epithelial cells. These initial contacts are subsequently sta-bilized by the formation of the adhesive junctional compo-nents, zonula adherens and desmosomes, at the cell surface

(20, 21). The extracellular domains of adherens junctions and

desmosomes are enriched with E-cadhenn and desmosomal

cadherins, respectively. Intracellularly, each junctional com-

ponent is associated with a different cytoskeletal element. In

zonula adherens, E-cadhenn interacts with actin microfila-

ments via associations with cytoplasmic proteins, vinculin,

a-actinin, and a-, �3-, and ‘y-catenins. Recently, y-catenin has

been identified as Pg (22-24). Desmosomes are associatedwith cytokeratin intermediate filaments. In this junction, cy-

tokeratins interact with the desmosomal cadherins Dsg and

Dsc via their associations with the cytoplasmic proteins des-

moplakins and Pg (21 , 25, 26). Pg is the major protein com-ponent common to the plaques of both kinds of adhering

junctions in epithelia and is also present in nonepithelial cells

(27). Pg has been shown to interact with a number of proteins

including cadherins (E-, N-, and P-cadherins as well as des-mosomal cadherins Dsg and Dsc; reviewed in Ref. 28).

Junction assembly and disassembly in cultured epithelial

cells are regulated by reversible protein phosphorylation andcan be induced by modulating the degree of cell-cell contactvia changing the extracellular Ca2� concentrations (29-32).

Growth media Ca2� concentrations of �180 p.M (HCM) in-

duces cell-cell contact and rapid assembly of junctionalcomponents, whereas decreasing the Ca2� concentration

below 50 p.M (LCM) causes rapid disassembly of junctional

components (33-35).HGF/SF induces scattering of epithelial cells maintained in

HCM. Although the underlying mechanisms of HGF/SF-in-

duced scattering are not known, cell scattering is associatedwith decreased cell-cell adhesion and disassembly of thejunctional complexes (6, 18, 36-38). To gain further insight

into the mechanisms of HGF/SF-induced scattering of epi-

thelial cells, we have studied the effect(s) of HGF/SF on the

assembly of adhering junctions by examining the expression

and stability of Pg and E-cadherin in MDCK cells and an

MDCK cell line expressing an exogenous chimeric met re-ceptor (CSF-MET; Ref. 39). Stimulation of the receptor chi-mera with CSF-1 leads to the activation of the met kinase invivo and mediates the motogenic, invasive, and morpho-genic responses of MDCK cells (39). Here, we show thatstimulation of MDCK cells expressing CSF-MET with CSF-1or HGF/SF inhibits cell-cell contact and junction formation.Our results demonstrate that stimulation of the met receptor

inhibits the cell-cell contact-regulated assembly of junctional

components in MDCK cells concurrent with a significantincrease in the stability of soluble E-cadherin, Pg, and the

altered phosphorylation of E-cadherin.

Resufts

The CSF-MET chimeric receptor contains the five immuno-

globulin repeats derived from the extracellular domain of theCSF-1 receptor, which are believed to be necessary forligand binding (40). The met portion contains 470 amino

acids, including the cytoplasmic and transmembrane do-mains as well as 13 amino acids of the met extracellular

domain (Fig. 1 and Ref. 39). Cells expressing this receptorscatter and form branching tubules in response to CSF-1(39). In contrast, cells expressing a mutant chimeric receptorcontaining a substitution of a phenylalanine residue for tyro-

sine 1 356 (V1356F) fail to scatter, invade, and form branch-

ing tubules in response to CSF-1 (41). This noncatalytic

residue in the COOH-terminus of the met receptor forms a

multisubstrate binding site for multiple substrates, includingphospholipase C�’, phosphatidylinositol 3’-kinase, Src,

growth factor receptor binding protein 2, and Src homology

and collagen adaptor proteins (41 , 42). The Vi 356F mutationdoes not affect the exogenous kinase activity of the receptor(41) but prevents the mutant protein from associating with

and activating multiple signaling pathways.HGF/SF Treatment Inhibits Junction Assembly upon

Induction of Cell-Cell Contact. The effects of HGF/SF onadhering junction assembly was studied by ultrastructuralexamination of control MDCK and MDCK cultures express-ing chimeric receptors following the induction of cell-cell

contact in the absence and presence of HGF/SF and CSF-1(Fig. 2). Confluent cultures of unstimulated MDCK, HGF/SF-

stimulated MDCK, and CSF-1-stimulated MDCK cells ex-pressing the chimeric receptors were processed for electronmicroscopy 4 h after the induction of cell-cell contact as

described in “Materials and Methods.”Unstimulated (Fig. 2, a and e, US) or CSF-1 -stimulated

MDCK(Fig. 2c) cultures showed the presence of well-formed

MDCK MDCK MDCK CSF-MET(US) (+HGF) (+ CSF) (+CSF)

Fig. 2. Junction assembly uponinduction of cell-cell contact is in-hibited in control and CSF-MET-expressing MDCK cells stimu-lated with HGF/SF and CSF-i,respectively. Control MOCK (a-c,e) and MDCK calls expressing thewild-type CSF-MET receptor(CSF-MET; dand h)orthe mutantV1356F receptors (f and g) wereestablished in LCM for 3 days.Cultures were incubated in se-rum-free LCM for 24 h and re-mained unstimulated (US; a ande) or were stimulated with 50ng/ml of either HGF/SF (b and 1)or CFS-i (c, d, g, and h) for 8 h.Cell-cell contact was induced byreplacing the media with serum-free HCM containing the sameconcentrations of growth factors,and incubation was continued for4 h. Cells were fixed, processedfor electron microscopy as de-scribed in “Materials and Meth-ods,” and examined in a Philips4i0 transmission electron micro-scope. e and h are lower magni-fications of micrographs in a andd, respectively. Bars: 0.5 �m in a,b, c, d, f, andgand i.5�tmineand h.

MDCK(US)

Y1356F Y1356F(+HGF) (+CSF)

Cell Growth & Differentiation 453

CSF-MET(+CSF)

desmosomes, whereas MDCK cultures stimulated with

HGF/SF (Fig. 2b) exhibited wide intercellular spaces and the

absence of junctional complexes. Occasionally, areas of

closely opposed membranes were observed in these cul-

tures (Fig. 2b). Examination of the MDCK cultures expressing

the wild-type CSF-MET chimeric receptor, which were stim-

ulated with CSF-i (Fig. 2, d and h), also revealed extensive

intercellular spaces and the absence of junctional com-

plexes. Similar results were observed when these cultures

were stimulated with HGF/SF (data not shown). CSF stimu-

lation of the MDCK cultures expressing the mutant CSF-MET

chimeric receptor (Yi 35619 had no apparent effect on junc-

tional component assembly (Fig. 2g). Exposure of these cul-

tures to HGF/SF, however, inhibited junction assembly, sim-

ilar to the effects of HGF/SF on control cultures (Fig. 2t).

Together, these results suggest that stimulation of the metreceptor inhibits cell-cell contact-regulated assembly of

junctional components in MDCK cells.

Subcellular Localization of E-cadherin and Pg inCSF-1 - and HGF/SF-stimulated Cuftures. In cultured ep-

ithelial cells, the induction of cell-cell contact results in re-

distribution of junctional proteins to the periphery of cells atthe areas of cell-cell contact and assembly of junctionalcomponents. Ultrastructural examination of HGF/SF-stimu-

lated cells revealed the absence of junctional complexes

following induction of cell-cell contact. Therefore, we used

indirect immunofluorescence microscopy to determine thesubcellular localization of E-cadherin (Fig. 3) and Pg (Fig. 4).

Confluent cultures of MDCK and MDCK cells expressing the

chimeric receptors were maintained in LCM. Duplicate cul-

tures were serum starved for 24 h and stimulated with

HGF/SF or CSF-i for 8 h and then transferred into HCM

containing the appropriate growth factors for 2, 8, and 24 h.

Control cultures remained unstimulated and were transferred

into HCM for the same periods of time. Cells were eitherfixed/permeabilized with absolute methanol to determine the

distribution of total E-cadherin and Pg (Figs. 3 and 4, Total)

or were first extracted with CSK buffer (see “Materials and

Methods”) and then fixed with formaldehyde to identify the

distribution of the TX-100-insoluble pool of the proteins

(Figs. 3 and 4, Extracted). Following fixation and extraction,

cells were processed for indirect immunofluorescence using

anti-E-cadherin (Fig. 3) and anti-Pg (Fig. 4) antibodies as

described in “Materials and Methods.”

In unstimulated MDCK cultures in LCM, the staining for

total E-cadherin was primarily cytoplasmic with some stain-

ing at the periphery of the cells (Fig. 3a�). Following extrac-

tion with 05K buffer, the most of both cytoplasmic and


Time

(hour)

0

2

8

24

Total

MDCK(US)Extracted

MDCK (+HGF)Total Extracted

MDCK (+CSF)Total Extracted

Fig. 3. Stimulation of MDCK cultures with HGF/SF and CSF-MET-expressing cultures with CSF-i results in increased solubility of E-cadherin and cellscattering. Confluent cultures were established on collagen-coated coverslips in LCM for 3 days and serum starved for 24 h. Cultures remained unstimulated[MDCK (US), a1-h1] or were stimulated with either 50 ng/ml of HGF/SF [MDCK (+HGF), a2-h2] or CSF1 [CSF-MET(+CSF), a3-h3] in serum-free LCM for8 h. Cell-cell contact was induced by replacing the LCM with serum-free HCM +1- the same concentrations of growth factors, and incubation wascontinued for various time periods (0, 2, 8, and 24 h). Cells were either fixed/permeabilized with absolute methanol at -20CC (Total) or extracted with CSKbuffer and then fixed with 1.75% formaldehyde (Extracted). Cells were processed for indirect immunofluorescence and confocal microscopy with amonoclonal anti-E-cadherin antibody and FITC-conjugated anti-mouse lgG. Displayed images represent optical sections recorded from the middle of thecells. Arrowheads, cell processes. Bar, 20 �tm.

peripheral staining was removed (Fig. 3b1). Two h after the

induction of cell-cell contact in HCM, most of the E-cadherin

staining was redistributed to the areas of cell-cell contact in

both fixed (Fig. 3c1) and extracted cultures (Fig. 3d1). The

intensity of the staining for total and extracted cultures was

very similar, indicating that the majority of E-cadherin at the

lateral membrane is in a TX-i 00-insoluble form. This pattern

of E-cadherin staining persisted throughout the 24-h incu-

bation in HCM (Fig. 3, e1-h1).

In MDCK cultures stimulated with HGF/SF in LCM, E-

cadherin distribution was similar to that observed in control

cultures; however, very little or no peripheral staining wasdetectable in these cells (Fig. 3, a2 and b2). After 2 h in HCM,

E-cadherin was detected both in the cytoplasm and at the

periphery of the fixed/permeabilized cells (Fig. 3c2); however,

the most of this staining was removed following extraction of

the cultures with CSK buffer (Fig. 3d2). With further incuba-

tion in HCM in the presence of HGF/SF, cells began to

develop processes (Fig. 3g2, arrowhead) and scatter, con-

current with a decrease in the TX-i 00-insoluble E-cadherin

staining (Fig. 3, e2-h2); therefore, by 24 h, very little staining

was detected in the extracted cultures (Fig. 3h2).Analysis of the MDCK cultures expressing the chimeric

CSF-MET receptor and stimulated with CSF-1 revealed an

E-cadherin staining pattern similar to those observed for

HGF-treated control MDCK cultures (Fig. 3, a3-h3). Cells in

these stimulated cultures became motile and began to scat-ter about 2 h after incubation in HCM (Fig. 3, c3 and d3,

arrowheads). In these cultures, E-cadherin was primarily de-

tected in the soluble form throughout the 24 h in HCM (Fig.

3, e3 and g3), and very little protein was detected in the

extracted cultures (Fig. 3, f3-h3).

Examination of the total Pg distribution in unstimulated

MDCK cultures in LCM revealed homogeneous intracellular

staining (Fig. 4a�), more than one-half of which was removed

following extraction of the cells with CSK buffer (Fig. 4b1).Two h after the induction of cell-cell contact in HCM, a

significant amount of the intracellular Pg staining was redis-


came rapidly insoluble and localized to the periphery of the

Time(hour)

0

2

8

24

TotalMDCK (US)

Extracted

MDCK (+HGF)Total Extracted

MDCK (+CSF)Total Extracted

Fig. 4. HGF/SF stimulation of MDCK and CSF-i stimulation of CSF-MET-expressing cultures disrupts lateral localization of Pg and results in cellscattering. Confluent cultures were established on collagen-coated coverslips as described in the legend of Fig. 2. Cultures remained unstimulated [MDCK(US), a1-h1] or were stimulated with either 50 ng/ml of HGF/SF [MDCK (+HGF), a2-h2] or 50 ng/ml of CSF1 [CSF-MET(+CSF), a3-h3] in serum-free LCMfor 8 h. Cell-cell contact was induced by replacing the LCM with serum-free HCM +1- the same concentrations of growth factors, and incubation wascontinued for various time periods (0, 2, 8, and 24 h). Cells were either fixed/permeabilized with absolute methanol at -20CC (Total) or extracted with CSKbuffer and then fixed with 1 .75% formaldehyde (Extracted). Cells were processed for indirect immunofluorescence and confocal microscopy with polyclonalanti-Pg antibodies and rhodamine-conjugated anti-rabbit lgG. Arrowheads, cell processes. Bar, 20 �m.

tributed to the periphery of the cell (Fig. 4c1). The extraction

of these cells resulted in the removal of the majority of the

intracellular staining as well as some of the peripheral stain-

ing, leading to the appearance of a punctate, lateral mem-

brane staining (Fig. 4d1). With further incubations in HCM, Pg

staining was progressively cleared from the cytoplasm. By

8 h, the staining was mainly peripheral. On extraction, the

staining at the lateral membranes was punctate, and this

pattern of staining persisted throughout the 24-h incubation

in HCM (Fig. 4, e1-h1).

In MDCK cultures stimulated with HGF/SF in LCM then

fixed/permeabilized with methanol, Pg staining was homo-

geneous and intracellular (Fig. 4a2).The intensity of this in-

tracellular staining was significantly reduced when these

cells were extracted prior to fixation (Fig. 3b2). In contrast to

unstimulated cultures, this pattern of staining persisted forseveral hours after the induction of cell-cell contact in HCM

(Fig. 4, c2-f2). Eight h after stimulation, cells had become

motile and began to scatter. At this time, some of the Pg

staining was distributed along the periphery of the cells,

although the majority of the staining was still distributed

intracellularly in both unextracted and extracted cultures(Fig. 4, e2 and f�). This pattern of staining persisted through

the 24-h time period. By this time, cells had developed long

processes that exhibited significant staining of Pg in both

fixed and extracted cells (Fig. 4, g2-h2, arrowheads).

Examination of the MDCK cells expressing the chimeric

CSF-MET receptor and stimulated with CSF-1 in LCM re-

vealed an intracellular Pg staining pattern similar to those

observed for HGF/SF-stimulated MDCK cultures (Fig. 4, a3

and b3). Two h after stimulation in HCM, cells began to

scatter. At this time, most of the Pg staining was intracellular

and was removed from the cells following TX-100 extraction

(Fig. 4, c3 and d3). By 8 h, cells exhibited processes, and

some of the Pg staining was redistributed to the periphery of

the cells as well as to the processes (Fig. 4e3, arrowheads).

A portion of this peripheral Pg staining was resistant to

extraction (Fig. 4f3). This staining pattern of Pg remained

throughout the 24 h in HCM (Fig. 4, g3 and h3).

Together these results showed that upon induction of cell-

cell contact in unstimulated cultures, junctional proteins be-

E- cadherin

Chase(hour) 0 2 4 8 12 24

S P S PS P S P SP SP

Plakoglobin

0 2 4 8 12 24

S P S PS P S PSP SP

MDCK(US) -

MDCK(+ HCF)

- 116

97

1116

� 97

-

- -

-


CSF-MET(+CSF)

Y1356 F(+CSF)

Fig. 5. HGF/SF stimulation increases the stability of newly synthesized E-cadherin on induction of cell-cell contact. Confluent cultures of MDCK and MDCKcells expressing chimeric receptors were established in LCM for 3 days and then serum starved for 24 h. Cultures remained unstimulated [MOCK (US)] orstimulated with 50 ng/ml of either HGF [CSF-MET(+ HGF)] or CSF1 [CSF-MET(+ CSF) or Yi356F (+ CSF)] in serum-free LCM for 8 h. Cells were incubatedin methionine- and serum-free LCM +1- HGF/SF or CSF1 for 45 mm and then pulse-labeled in the same media supplemented with 125 pCi of[35S]methionine for 15 mm. Cultures were chased in iO,000-fold excess methionine in serum-free HCM +1- HGF/SF or CSF-i for 0, 2, 4, 8, 12, and 24 h.Following the chase period, cells were extracted with CSK buffer, and the soluble (5) and insoluble (P) fractions were processed for immunoprecipitationwith anti-E-cadherin and anti-Pg antibodies followed by SDS-PAGE and fluorography. Bars to the left of the panels indicate, from top to bottom, the positionof protein bands corresponding to the E-cadherin precursor, mature E-cadherin, a-catenin, �3-catenin, and Pg. respectively. Bars to the right of the panelsindicate the position of the Mr i 1 6,000 and 97,000 (1 16 and 97, respectively) molecular weight markers.

cells, as demonstrated previously. HGF/SF stimulation of

control MDCK and CSF-i stimulation of CSF-MET-express-

ing cells resulted in accumulation of an intracellular and

peripheral soluble pool of both E-cadherin and Pg. Ultra-

structural studies demonstrated that the peripherally distrib-

uted proteins were not organized into assembled junctional

components. Furthermore, CSF-1 -stimulated cells were

scattered faster than HGF/SF-stimulated cells.Effect of HGF/SF on Stability of E-Cadherin and Pg

upon Induction of Cell-Cell Contact. The results of themorphological studies led us to determine the stability of

newly synthesized Pg and E-cadherin in HGF/SF- and CSF-

1-stimulated cultures. For these studies, confluent cultures

of control MDCK and CSF-MET-expressing cells were main-

tamed in LCM. Replicate cultures were serum starved for

24 h and either remained unstimulated or were stimulated

with HGF/SF (MDCK) or CSF-1 (CSF-MET) in LCM for 8 h.

Cultures were metabolically labeled with [35S]methionine in

LCM ± HGF/SF or CSF-1 and chased for 0-24 h in HCM

containing the same concentrations of the growth factors.

Following cell fractionation, TX-100-soluble and -insoluble

fractions were processed for immunoprecipitation with anti-

E-cadherin and Pg antibodies, as described in “Materials andMethods.”

Immunoprecipitation of the soluble and insoluble fractions

of unstimulated MDCK cultures after 15 mm labeling de-

tected, only in the soluble fraction, five bands of Mr

-135,000, 120,000, 102,000, 98,000, and 83,000 corre-

sponding to E-cadherin precursor, mature E-cadherin,

a-catenin, j3-catenin, and Pg, respectively (Fig. 5, E-cad-herin, 0 S, bars from top to bottom). Two h into the chase

period in HCM, all of the E-cadherin precursor protein was

processed to the mature, Mr 1 20,000 form in association with

a- and �-catenins and Pg [Fig. 5; E-cadherin, MDCK (US), 2

SI. This complex was relatively unstable and disappeared

rapidly (t112, --2.7 h).

lmmunoprecipitation of the soluble and insoluble fractions

of control MDCK cells metabolically labeled in the presence

of HGF/SF detected the same protein complex as in the

control cultures, i.e., E-cadherin precursor, E-cadherin,

a-catenin, �3-catenin, and Pg, only in the soluble fraction [Fig.

5, MDCK (+ HGF), E-cadherin, 0 5]. In these stimulated

l-_-�� LULL wu.C.)Cl) �Q �Cl)OD u�i L1_o (0(i)� (0+ co.±� �c.)

0’- C.)

Time(minutes) � � � -�- � � � � � 60

Sp spsPsP sPsPsP sPsPsp

IlIlIJItI, IC‘C0.CwC

.�00,00a. �ttti


Fig. 6. Overall phosphorylation of E-cadherin and Pg in unstimulated (US) and HGF/SF- or CSF-1-stimulated MDCK and CSF-MET-expressing cultures.Confluent cultures of MDCK and CSF-MET cells were established in LCM for 3 days and serum starved for 24 h. Cultures were preincubated in serum- andphosphate-free LCM for 45 mm and were then labeled in LCM supplemented with 750 �Ci/ml [32P]orthophosphate and 1 m�i sodium orthovanadate fora total of 3 h. Ca2� (to the final concentration of 1.8 mM) was added to all cultures during the final hour of labeling. In replicate cultures, 50 ng/ml HGF/SFor CSF-1 were added at 20, 40, and 60 mm prior to the end of labeling. Cells were then extracted with CSK buffer, and the soluble (5) and insoluble (F)fractions were immunoprecipitated with E-cadhenn and Pg antibodies followed by SDS-6% PAGE and autoradiography. Bars to the right of panel indicatethe position of the Mr 1 16,000 and 97,000 (1 16 and 97, respectively) molecular weight markers.

cultures, however, the complex was significantly more stable(t112, -17 h).

E-cadherin immunoprecipitates from MDCK cultures ex-

pressing the chimeric CSF-MET receptor and stimulated with

CSF-1 showed the presence of a stable cadherin/catenin/Pg

complex similar to that observed in MDCK cultures stimu-

lated with HGF/SF [Fig. 5, CSF-MET (+ CSF), E-cadherin,0-24 S; t112, -18 h]. However, immunoprecipitation ofE-cadherin from CSF-MET cultures expressing the mutant

chimeric receptor Vi 356F detected an E-cadherin/

catenins/Pg complex that disappeared rapidly, with a turn-

over rate similar to the unstimulated MDCK cells [Fig. 5,

Y1356 F (+ CSF), E-cadherin, 0-24 5; t112, -‘3 h].

lmmunoprecipitation of the soluble and insoluble frac-

tions of control MDCK cultures with anti-Pg antibodies

detected an Mr 83,000 band corresponding to Pg in both

the soluble and insoluble fractions in a 60:40 ratio, respec-

tively. A protein band of M� �75,000 coimmunoprecipi-

tated with Pg; the identity of the protein band is unknown.

An E-cadherin/catenin complex was also detectable in the

soluble fractions immunoprecipitated with Pg from 0-4-h

time points. In these cultures, Pg is rapidly transferred

from the soluble (t1,2, -3 h) into an insoluble pool where

proteins remain stable (t1,2, -24 h)

In HGF/SF-stimulated MDCK cells, newly synthesized Pgwas also present in a 60:40 ratio in the soluble and insoluble

pools, respectively. However, HGF/SF stimulation prior to

and during metabolic labeling and chase increased the sta-

bility of the soluble pool by 2-fold (t1,2, -6 h) and had no

significant effect on the turnover rate of the insoluble pool of

Pg [t1,2, -25 h; Fig. 5, Plakoglobin, MDCK (+ HGF)]. In these

cultures, E-cadherin, coimmunoprecipitated with the Pg an-

tibodies, was detectable in the soluble fraction through 8 h ofthe chase period [Fig. 5, Plakoglobin, MDCK (+ HGF), 8 5].

Analysis of the turnover rate of newly synthesized soluble

and insoluble Pg from MDCK cultures expressing the wild-

type chimeric CSF-MET receptor and stimulated with CSF-1

showed similar results to those of control cultures stimulated

with HGF/SF [Fig. 5, Plakoglobin, CSF-MET (+ CSF); t1,2:

soluble, -5.5 h, and insoluble, -27 h]. However, stimulation

of the cultures expressing the mutant CSF-MET receptor

(Y1356F) with CSF-1 had no effect on the turnover rate of the

newly synthesized Pg [Fig. 5, Plakoglobin, Y1356 F (+ CSF);

t1,2: soluble, -3.5 h, and insoluble, -26 h]. These results

suggest that stimulation of cultures with HGF/SF significantlyincreases the stability of the soluble E-cadherin (>6-fold)

and, to a lesser extent, soluble Pg (-2-fold).

HGF/SF Stimulation Has No Significant Effect on theTotal Amount of Phosphorylated E-cadherin and Pg.Phosphorylation analysis was used to determine whether the

increase in the stability of the soluble E-cadherin and Pg in

HGF/SF-treated cultures could be due to differences in the

phosphorylation of these proteins in the stimulated cultures.

Control and CSF-MET-expressing MDCK cultures were es-

tablished at confluent density in LCM and serum starved as

described in “Materials and Methods.” Cultures were [32P]P,

labeled in the absence of serum for 2 h, at which time Ca2�

was added (to the final concentration of 1 .8 mr�i, HCM) to all

cultures, and labeling was continued for another hour. Inreplicate cultures, HGF/SF or CSF-1 was added at 20, 40,

and 60 mm prior to the end of labeling. Cells were extracted,

and the soluble and insoluble fractions were processed for

immunoprecipitation with anti-E-cadherin and Pg antibodies(Fig. 6).

Immunoprecipitation of the soluble and insoluble fractions

of the control and growth factor-stimulated cells with anti-

E-cadherin antibody detected, primarily in the soluble frac-tions, three major bands corresponding to E-cadherin pre-


cursor, mature E-cadherin, and �-catenin. The

coimmunoprecipitated a-catenin and Pg, although detecta-

ble, were significantly less phosphorylated (Fig. 6, E-cad-herin). Stimulation of control MDCK and CSF-MET cultureswith HGF/SF and CSF-1 , respectively, led to a slight increasein the amount of phosphorylation compared with the un-

stimulated cultures [Fig. 6, E-cadherin; compare MDCK (US)

with MDCK (+ HGF) and CSF-MET(+ CSF-1)]. There was no

difference in the amount or pattern of phosphorylation of

E-cadherin between control MDCK cells and the cells ex-

pressing the mutant chimeric CSF-MET receptor (Y1356F)

that were stimulated with CSF-1 [Fig. 6, E-cadherin; compare

MDCK (US) with Y1356 F(+ CSF)].lmmunoprecipitation of [32P]P1-labeled cultures with an-

ti-Pg antibodies detected Pg in both the soluble and insol-

uble fractions. In the soluble extracts, however, a faster-

migrating band was also detectable. The soluble Pg was

significantly more phosphorylated than the insoluble Pg, as

has been shown previously (43). HGF/SF stimulation of the

control cultures and CSF-1 stimulation of CSF-MET-ex-

pressing cultures resulted in a slight increase in the amount

of phosphorylated Pg, similar to E-cadherin [Fig. 6, Plako-

globin; compare MDCK (US) with MDCK (+ HGF) and CSF-

MET (+ CSF)}, whereas stimulation of Y1356F cultures with

CSF-1 had no effect on the phosphorylation of Pg [Fig. 6,

Plakoglobin; compare MDCK (US) with Y1356 F (+ CSF)].

In addition, we did not detect any tyrosine phosphorylation

(either by phosphoamino acid analysis or immunoblotting

with anti-phosphotyrosine antibodies) of E-cadherin or Pg in

these cultures following stimulation with the growth factors

(data not shown). Together, these results suggest that

HGF/SF stimulation of MDCK cultures increases slightly the

total amount of phosphorylated E-cadherin and Pg but doesnot affect the type of phosphorylation.

HGFISF Stimulation Alters the Phosphorylation Patternof E-cadherin but not Pg. We proceeded to further analyzethe phosphorylation patterns of E-cadherin and Pg in growth

factor-stimulated cultures. For this purpose, confluent cul-tures were labeled with [32P}P1, as described above, for a

total of 3 h and were stimulated with HGF/SF or CSF-1

during the final hour of labeling. Following extraction and cell

fractionation, the soluble and insoluble fractions were immu-noprecipitated with anti-E-cadherin and Pg antibodies andprocessed for partial proteolytic digestion with V8 protease

as described in “Materials and Methods.” We did not detect

any differences between the partial proteolytic maps of sol-

uble or insoluble Pg immunoprecipitated from unstimulated

and HGF/SF-stimulated cultures (Fig. 7, Plakoglobin). How-

ever, new phosphopeptides were identified for the E-cad-herin immunoprecipitated from HGF/SF-stimulated MDCKcultures (Fig. 7, E-cadherin, 5; compare Unstimulated with

Stimulated). These results suggest that on stimulation with

HGF/SF, new residues become phosphorylated in E-cad-

herin.

DiscussionEpithelial morphogenesis is mediated by a series of cellular

and molecular events regulated by a wide array of mesen-

chymal factors affecting cell-cell and cell-matrix interactions

E-cadherin Plakoglobin‘D0-.!

�.�

�C�

‘00

��E

�

(I)

‘00�-

�.E�C

�

V0

��E

:�U)

Fig. 7. HGF/SF stimulation changes the phosphorylation pattern of E-cadherin but not Pg in MDCK cells. Unstimulated and HGF/SF-stimulatedMDCK cultures were [�2P]orthophosphate labeled as described in thelegend to Fig. 5. Following extraction with CSK buffer, the soluble (S) andinsoluble (F) fractions of unstimulated and HGF/SF-stimulated (60 mm)cultures were immunoprecipitated with E-cadherin and Pg antibodies,separated by SDS-6% PAGE and processed for autoradiography. Usingthe autoradiogram as a template, bands corresponding to E-cadherin andPg were excised from the gel. The excised bands were loaded ontoSDS-1 5% polyacrylamide gels, digested with 7.5 mg of V8 protease/laneas described in “Materials and Methods,” electrophoresed, and pro-cessed for autoradiography. Bars to the left of the diagram indicate theposition of molecular weight markers, from top to bottom: M, 116,000,97,000, 66,000, and 45,000. Arrowheads, the appearance of new E-cadherin phosphopeptides in HGF/SF-stimulated cells.

(reviewed in Ref. 44 and references therein). HGF/SF is one

such factor. The signal mediated by HGF/SF is transmitted

into the cell via met, a RTK (reviewed in Refs. 1 and 2). The

hallmark response to HGF/SF stimulation in vitro is cell dis-

sociation and motility (scattering), both of which require de-

creased cellular adhesion associated with junction disas-

sembly. The present study was undertaken to gain further

insight into the mechanism(s) by which HGF/SF induces

scattering of epithelial cells. Previous studies were based on

determining how HGF/SF causes disassembly of junctionalcomplexes. In this study, we used MDCK cells, the standard

cell line for scattering assay, to determine how HGF/SF af-

fects adhering junction assembly at the time of induction of

cell-cell contact. Our approach was based on the previous

observations that demonstrated a role for protein phospho-rylation/dephosphorylation in junction disassembly and as-sembly (34, 35, 45-48). The effect(s) of HGF/SF on junctionassembly was studied by analyzing the solubility, phospho-rylation, subcellular distribution, and stability of two majoradhering junction proteins, Pg and E-cadherin. Pg is a com-ponent of both zonula adherens and desmosomes (49) and

has been shown to interact with a number of cadherins

including desmosomal cadherins, Dsg and Dsc (28). Pg alsomediates the interactions between cytokeratin intermediate

filaments and desmosomal plaques and, therefore, desmo-some assembly (50). In addition to its structural role in des-

mosomes, Pg plays a role in signal transduction pathways

regulating differentiation and morphogenesis (51). E-cad-


herin is the transmembrane protein of the zonula adherens

and the major cell adhesion molecule in epithelia, with well-documented roles in morphogenesis and for the close asso-

ciations of epithelial cells and, therefore, their normal func-tion (19, 52, 53).

Ultrastructural examination of MDCK cells induced to form

cell-cell contact in HCM showed that they failed to form

junctions when stimulated with HGF/SF (Fig. 2). In contrast to

unstimulated cells, these cells exhibited wide intercellular

spaces similar to unstimulated cells maintained in LCM (datanot shown; see Fig. 4 in Ref. 31), a condition that does not

allow intimate cell-cell contact required for the assembly of

junctional complexes. Immunoblot analysis of unstimulatedand stimulated cells showed no differences in the total

steady-state levels of Pg and E-cadherin (data not shown),as reported previously for E-cadherin (6, 38). Because theinsoluble form of Pg and E-cadhenn is known to be involvedin cell-cell contacts (43, 49, 54-56), we analyzed the solu-

bility properties of these proteins in HGF/SF-stimulated cul-tures. lmmunofluorescence analysis in conjunction with dif-ferential fixation/extraction showed that in stimulated cells,

unlike in control cultures, the majority of Pg and E-cadherinstaining were TX-i 00 extractable and present in both cyto-

plasmic and membrane distributions (37, 57). In addition,

under these conditions the membrane distributions of Pg

and E-cadhenn were homogeneous and not concentrated in

the areas of cell-cell contact (as observed in control cells,Figs. 3 and 4). Watabe et al. (38) reported similar distributions

for E-cadherin and desmoplakins in HGF/SF-stimulated

mouse keratinocyte cell ‘lines. The increase in the amount ofthe TX-100-extractable E-cadherin and Pg in HGF/SF-stim-

ulated cells was later confirmed by the results of the meta-

bolic labeling studies that detected a 6- and 2-fold increasein the stability of newly synthesized soluble E-cadherin and

Pg, respectively. We did not detect any significant differ-ences in the total amount of proteins between unstimulatedand stimulated cells (see above). HGF/SF stimulation has

been shown to inhibit gap junctional communication; how-ever, unlike adhering junction proteins, HGF/SF down-regu-lated the expression of the gap junctional proteins, connex-

ins, presumably because of an increase in proteindegradation (58, 59). It is noteworthy that the formation ofgap junctions is also cadherin dependent (60).

Decreased cellular adhesion has been correlated with

changes in the phosphorylation of junctional proteins. In-

creased protein tyrosine phosphorylation, deterioration ofadherens junctions, and increased invasiveness have been

observed in various cell types upon transformation (45, 47,

61-63). In addition, Pg and E-cadherin have been shown to

be more heavily phosphorylated in the nonjunctional, soluble

form (43), the form of Pg and its homologues, �-catenin andDrosophila Armadillo protein, which exhibit signaling func-

tions (64, 65). Phosphorylation of j3-catenin and Armadillo are

regulated in part by Zeste white 3 (ZW3) and GSK-3�3 serine/

threonine kinases, respectively. The activity of ZW3 is in turnregulated negatively by the Wnt/Wg (wingless) signal (66).

Wnt is a family of genes involved in signal-transducing path-

ways that regulate cell fate and morphogenesis (67). Inter-estingly, a recent study demonstrated decreased Wnt5a ex-

pression following stimulation of mammary epithelia with

HGF/SF, thereby placing Wnt5a downstream of HGF/SF sig-

naling, suggesting a link in the mechanism of cell motility/

scattering by this growth factor (68). However, thus far wehave not detected altered phosphorylation of Pg.

HGF/SF receptor is localized to the basolateral domain of

MDCK cells in a distribution resembling that of E-cadherin,

i.e., concentrated in the areas of cell-cell contact (69). This

RTK may decrease cellular adhesion by altering the phos-phorylation of thejunctional proteins. In fact, increases in the

tyrosine phosphorylation of Pg and E-cadherin have beenreported in carcinoma cells stimulated with HGF/SF and

epidermal growth factor (57). We, therefore, determinedwhether changes in the solubility of Pg and E-cadherin in

HGF/SF-stimulated cells were correlated with changes in the

phosphoiylation of the proteins. However, Pg and E-cad-herin are not tyrosine phosphorylated in MDCK cells, nor

have we detected any tyrosine phosphorylation of E-cad-

herin or catenins/Pg in HGF/SF-stimulated MDCK cells.

Nonetheless, we detected slightly increased levels of phos-

phorylation of the soluble Pg and E-cadherin in HGF/SF-

stimulated cultures, and most interestingly, the appearance

of new phosphopeptides for E-cadherin. The inability of the

vi 356F MDCK cell line to respond to the growth-factor

stimulation demonstrated the specificity of HGF/SF action

via the met receptor, because tyrosine 1356 in the COOH-

terminal domain of the HGF/SF receptor has been shown to

be essential for transducing signals for motility and morpho-

genesis (41). How HGF/SF increases the stability of the sol-uble junctional molecules is not clear. One possibility is that

HGF/SF may regulate the activity and/or function of a kinase/

phosphatase responsible for cadherin phosphorylation/de-

phosphorylation. Changes in the phosphorylation pattern ofE-cadherin may influence its interactions with cytoskeleton

and thus its insolubility, which has been shown to be re-

quired for its adhesive function (70, 71). Expression of func-

tional E-cadherin is necessary and precedes the formation of

the junctional components: tight, adherens, and gap junc-

tions and desmosomes (72). Therefore, the inhibition of junc-

tion assembly in HGF/SF-stimulated cells may be secondaryto perturbations in E-cadherin function by increasing the

stability of the soluble form of this protein. Furthermore, the

increase in stability of the soluble Pg may be due to the

stability of the E-cadhenn-associated form of this protein. Pg

has been shown to interact with E-cadherin only in the sol-

uble form (43, 73, 74). It is conceivable that the scattering

function of HGF/SF is mediated via inhibition of junction

assembly rather than induction of junction disassembly. Our

data suggest that in HGF/SF-stimulated cells, E-cadherin

remains soluble and may not be able to participate in junction

formation upon induction of cell-cell contact. In support of

this observation, scattering of MDCK cells in HCM normally

occurs 24 h after HGF/SF stimulation, suggesting that, unlike

removal of the extracellular Ca2� that induces rapid (within

minutes) disassembly of junctions scattering by HGF/SF, it

may function by inhibiting the assembly of new junctionswhile previously established ones are undergoing disassem-

bly. This notion would be consistent with the estimated turn-

4ee Inhibition of Junction Assembly by HGF/SF

over rate of various cadherins in HCM cultures (5-24 h; Refs.55 and 75).

Materials and MethodsCell Lines and Culture Conditions. Development and characterizationof the stable lines of MDCK cells expressing receptor chimeras have beendescribed previously in detail (39, 41). Briefly, receptor chimeras wereexpressed as fusion proteins consisting of the extracellular domain ofhuman CSF-1 receptor fused to the transmembrane and cytoplasmic

domains of human p190��et (39). Cells were maintained in DMEM/10%fetal bovine serum as described previously (32). Cultures were grown to

75% confluency, trypsinized, and then plated at a confluent density of2.5-3 x 1 05/cm2 in collagen-coated 35-mm petn dishes or on coverslipsin LCM (5 p.M Ca2�) for 3 days. Cell-cell contact was induced synchro-nously by replacing the LCM with HCM (1.8 m� Ca2�) as described (76).HGF/SF was produced and purified as described previously (39). Recom-binant human CSF-1 was kindly provided by Dr. Gordon Wong (GeneticsInstitute, Boston, MA). HGF/SF and recombinant human CSF were added

at 50 ng/ml to cultures that were serum starved for 24 h.Antibodies. Polyclonal antibodies to Pg were raised in rabbits and

have been characterized previously (43). Mouse monoclonal E-cadherin

antibody was a gift from Dr. Warren Gallin (University of Alberta) and wasused as ascites fluid. This antibody recognizes both the TX-i 00-solubleand -insoluble E-cadherin in immunofluorescence and immunoblothngbut only the soluble E-cadherin in immunoprecipitation.

Metabolic Labeling, Cell Fractionation, and lmmunoprecipitation.Replicates of confluent cultures of MOCK and MDCK cells expressingCSF-MET receptor were established and maintained in LCM in collagen-coated 35-mm petn dishes. Duplicate cultures were serum starved for24 h and stimulated with HGF or CSF-1 for 8 h in LCM. For tumoverstudies, replicate cuitures were preincubated in methionine-free LCM ±

CSF-1 or HGF for 45-60 mm and were then labeled in 0.5 ml of LCM ±

CSF-1 or HGF supplemented with 125 �Ci of [�5SJmethionine (1189Ci/mmol; ICN Biochemicals, Irvine, CA) for 15 mm. At the end of thelabeling period, duplicate cultures were rinsed twice with PBS and thenchased in >10,000- fold excess methionine in CSF-1 or HGF containingLCM or HCM for 0-24 h. For [�2P]P� labeling, replicate 35-mm cuituredishes were prepared in LCM, serum starved ovemight, preincubated inphosphate-free LCM ± CSF-1 or HGF for 45-60 mm, and were thenlabeled in 1 .0 ml of HCM ± CSF-1 or HGF supplemented with 750 �CVmlof [�2P]P1 (500 mCi/mI; ICN) and 1 m� sodium orthovanadate for 4 h.Alternatively, cultures were labeled in LCM supplemented with 750 �CVml

of [�2P]P� and 1 mM sodium orthovanadate for 2 h and then switched toHCM for 1 h. In replicate cultures, CSF-1 or HGF were added at 20, 40,

and 60 mm prior to the termination of labeling. On completion of labelingand/or chase, cells were extracted in the petn dish with 1 ml of CSK(cytoskeleton extraction buffer; Ref. 33) buffer [1 m� sodium orthovana-date, 300 mM sucrose, 10 m,�i 1 ,4-piperazinediethanesulfonic acid (pH

6.8), 50 mM NaCI, 3 mM MgCl2, 0.5% (v/v) Triton X-100, 1 .2 m� phenyl-methyl sulfonyl fluoride, 0.1 mg/mI DNase, and 0.1 mg/mI RNase] at 4�Cfor 10 mm. Cells were scraped from the petri dish and centrifuged at

48,000 x g for 10 mm; then the soluble and insoluble fractions were

separated. Aliquots of 250 �.d were processed for immunoprecipitationwith Pg and E-cadherin antibodies, followed by SOS-PAGE and fluorog-raphy as described previously (43).

The relative amount of radioactivity in the protein bands correspondingto Pg and E-cadhenn was determined by scanning densitometry frommultiple exposures of the resulting fluorograms, as described previously

(76). Each experiment was repeated at least five times, and data from onetypical experiment are presented.

Partial Proteolytic Peptide Mapping. Confluent cultures of unstimu-lated and growth factor-stimulated MOCK cells were [�2P]P� labeled andprocessed for immunoprecipitation with anti-Pg and anti-E-cadherin an-

tibodies as described previously. Following autoradiography, and usingthe autoradiograms as templates, bands corresponding to E-cadhenn andPg were excised from the gels and processed for partial V8 proteasedigestion (7.5 mg/lane) on a SOS-i 5% PAGE (77). On completion ofelectrophoresis, gels were processed for fluorography.

lmmunofluerescence and Electron Microscopy. MOCK and MOCKcells expressing chimenc receptors were grown in LCM on collagen-

coated coverslips as described previously (43). For indirect immunofluo-

rescence, replicate cultures were serum starved for 24 h and stimulatedwith CSF-1 or HGF for 8 h. Cell-cell contact was induced by replacing theLCM ± CSF-i or HGF/SF with HCM ± growth factors for different periodsof time (0-24 h). Coverslips were rinsed with PBS and were either flxed/permeabilized with methanol or extracted with CSK buffer, fixed in 1.75%formaldehyde, and processed for indirect immunofluorescence with theprimary antibodies, Pg and E-cadhenn, as described (78). Rabbit anti-bodies were visualized with rhodamine-conjugated goat anti-rabbit lgG,whereas the mouse monoclonal antibody was visualized with FITC-con-

jugated goat anti-mouse lgG. Secondary antibodies were purchased fromBoehnnger Mannheim Biochemicals (Indianapolis, IN). Primary and sec-ondary antibodies were diluted 1:100 with PBS. Coverslips were mountedin elvenol containing 0.2% paraphenylene diamine (pH 8.6) and viewed

with a x 100 objective using a laser scanning confocal microscope (LeicaLasertechnic, Heidelberg, Germany). Optical sections through the middleof the cells were recorded on optical discs and viewed on a silicongraphics imager (SGI; Silicon Graphics, CA).

For electron microscopy, duplicates of confluent cultures grown inLCM were serum starved for 24 h and stimulated with CSF-i or HGF/SFin LCM for 8 h. Cell-cell contact was induced by replacing the LCM withHCM ± CSF-1 or HGF/SF and incubation continued for another 4 h. Cellswere fixed in a freshly made solution containing 1.25% glutaraldehyde,1% 0504 in 50 m� 1,4-piperazinediethanesulfonic acid (pH 7.2) for 5 mmat rcom temperature, processed for electron microscopy as described(31), and examined in a Philips 410 transmission electron microscope.

AcknowledgmentsWe are grateful to Honey Chan for valuable assistance in electron micros-copy and to Ors. Ellen Shibuya and Henry Parker for critical review of themanuscript.

References1 . Rosen, E. M., and Goldberg, I. 0. Protein factors which regulate cellmotility. In Vitro Cell Dev. Biol., 25: 1079-1087, 1989.

2. Zamegar, R., and Michalopoulos, G. K. The many faces of hepatocytegrowth factor: from hepatopoiesis to hematopoiesis. J. Cell Biol., 129:1177-1180, 1995.

3. Rosen, E. M., Nigarn, S. K., and Goldberg, I. 0. Scatter factor and thec-Met receptor: a paradigm for mesenchymaVepithelial interaction. J. Cell

Biol., 127: 1783-1787, 1994.

4. Nakamura, T., Teramoto, H., and Ichihara, A. Purification and charac-terization of a growth factor from rat platelets for mature parenchymal

hepatocytes in primary culture. Proc. NatI. Acad. Sci. USA, 83: 6489-

6494, 1986.

5. Bottaro, 0. P., Rubin, J. S., Faletto, 0. L, Chan, A. M. L, Kmiecik, T.

E., Vande Woude, G. F., and Aaronson, S. A. Identification of the hepa-tocyte growth factor receptor as the c-met-proto-oncogene product. Sci-

ence (Washington DC), 251: 802-804, 1991.

6. Weidner, K. M., Behrens, J., Vandekerckhove, J., and Birchmeier, W.Scatter factor: molecular characteristics and effect on the invasiveness ofepithelial cells. J. Cell Biol., 1 1 1: 2097-2108, 1990.

7. Shibamoto, S. Hepatocyte growth factor and transforming growthfactor-�3 stimulate both cell growth and migration of human gastric ade-

nocarcinoma cells. Cell Struct. Funct., 17: 185-190, 1992.

8. Grant, 0. S., Kleinman, H. K., Goldberg, I. 0., Bhargava, M. M.,Nickoloff, B. J., Kinsella, J. L, Polverini, P., and Rosen, E. M. Scatterfactor induces blcod vessel formation in vivo. Proc. NatI. Acad. Sci. USA,90: 1937-1941, 1993.

9. Naldini, L, Vigna, E., Narsimhan, R. P., Gaudino, G., Zarnegar, R.,Michalopoulos, G. K., and Comoglio, P. M. Hepatocyte growth factor(HGF) stimulates the tyrosmne kinase activity of the receptor encoded bythe protooncogene c-MET. Oncogene, 6: 501-504, 1991.

10. Park, M., Dean, M., Kaul, K., Braun, M. J., Gond, M. A., and VandeWoude, G. Sequence of met-proto-oncogene cONA has features charac-teristic of tyrosine kinase family of growth factor receptors. Proc. NatI.Acad. Sci. USA, 84: 6379-6383, 1987.


1 1 . Defrances, M. C., Wolf, H. K., Michalopoulos, G. K., and Zamegar, R.The presence of hepatocyte growth factor in the developing rat. Devel-opment (Camb.), 1 16: 387-395, 1992.

12. Stem, C. 0., Ireland, G. W., Herrick, S. E., Gherardi, E., Gray, J.,Perryman, M., and Stoker, M. Epithelial SF and development of the chickembryonic axis. Development (Camb.), 1 10: 1271-1284, 1990.

13. Sonnenberg, E., Weidner, K. M., and Birchmeier, C. Expression of themet-receptor and its ligand, HGF-SF during mouse embryogenesis. EXS.,

65: 381-394, 1993.

14. Yang, X. M., and Park, M. Expression of the met/hepatocyte growthfactor/scatter factor receptor and its ligand during differentiation of murine

p19 embryonal carcinoma cells. 0ev. Biol., 157: 308-320, 1993.

15. Sonnenberg, E., Meyer, 0., Weidner, K. M., and Birchmeier, C. Scatterfactor/hepatocyte growth factor and its receptor, the c-met tyrosine ki-nase, can mediate a signal exchange between mesenchyme and epithelia

during mouse development. J. Cell Biol., 123: 223-235, 1993.

16. Gherardi, E., Sharp, M., Lane, K., Sirulnik, A., and Stoker, M. Hepa-tocyte growth factor/scatter factor (HGF/SF), the c-Met receptor and thebehaviour of epithelial cells. Symp. Soc. Exp. Biol., 47: 163-181, 1993.

17. Birchmeier, W., Weidner, K. M., and Behrens, J. Molecular mocha-nisms leading to loss of differentiation and gain of invasiveness in epithe-hal cells. J. Cell Sd. Suppl., 17: 159-164, 1993.

18. Nusrat, A., Parkos, C. A., Bacarra, A. E., Godowski, P. J., Deip-Archer,C., Rosen, E. M., and Madara, J. L Hepatocyte growth factor/scatterfactor effects on epithelia. Regulation of intercellular junctions in trans-formed and non-transformed cell lines, basolateral polarization of c-metreceptor in transformed and natural intestinal epithelia, and induction ofrapid wound repair in transformed model epithelium. J. Clin. Invest., 93:

2056-2065, 1994.

19. Takeichi, M. Cadherin cell adhesion receptors as morphogeneticregulator. Science (Washington DC), 252: 1451-1455, 1991.

20. Geiger, B., and Ayalon, 0. Cadherins. Annu. Rev. Cell Biol., 8: 307-332, 1992.

21 . Garrod, 0. R. Desmosomes and hemidesmosomes. Curr. Opin. CellBiol., 5: 30-40, 1993.

22. Knudsen, K. A., and Wheelock, M. J. Plakoglobmn, or an 83 kD ho-mologue distinct from �3-catenin, interacts with E-cadherin and N-cad-hem. J. Cell Biol., 118: 671-679, 1992.

23. Peifer, M., McCrea, P. 0., Green, K. J., Wieschaus, E., and Gumbiner,B. M. The vertebrate adhesive junction proteins �3-catenin, and plakoglo-bin and the Drosophila segment polarity gene armadillo form a multigenefamily with similar properties. J. Cell Biol., 118: 681-691, 1992.

24. Piepenhagen, P. A., and Nelson, W. J. Defining E-cadherin associatedprotein complexes in epithelial cells: plakoglobmn, �3-catenin and y-cateninare distinct components. J. Cell Sci., 104: 751-762, 1993.

25. Buxton, R. S., Cowin, P., Franke, W. W., Garrod, 0. R., Green, K. J.,King, I. A., Koch, P. J., Magee, A. I., Rees, 0. A., Stanley, J. R., andSteinberg, M. S. Nomenclature of the desmosomal cadherins. J. Cell Biol.,

121: 481-483, 1993.

26. Schwarz, M. A., Owaribe, K., Kartenbeck, J., and Franke, W. W.Desmosomes and hemidesmosomes: constitutive molecular compo-

nents. Annu. Rev. Cell Biol., 6: 461-491, 1990.

27. Kapprell, H-P., Cowin, P., and Franke, W. W. Biochemical character-ization of the soluble form of the junctional protein, plakoglobin, fromdifferent cell types. Eur. J. Biochem., 166: 505-517, 1987.

28. Cowin, P. Unraveling the cytoplasmic interactions of the cadherinsuperfamily. Proc. NatI. Acad. Sd. USA, 91: 10759-10761 , 1994.

29. Henning, H., and Holbrook, K. A. Calcium regulation of cell-cell con-tact and differentiation of epidermal cells in culture: an ultrastructural

study. Exp. Cell Res., 143: 127-142, 1983.

30. Citi, S. Protein kinase inhibitors prevent junctional dissociation in-

duced by low extracellular calcium in MOCK epithelial cells. J. Cell Biol.,

117: 169-178, 1994.

31 . Pasdar, M., Li, Z., and Chan, H. Oesmosome assembly and disas-

sembly are regulated by reversible protein phosphorylation in culturedepithelial cells. Cell Motil. Cytoskeleton, 30: 108-122, 1995.

32. Balada, M. S., Gonzales-Mariscal, L, Matter, K., Cerejido, M., andAnderson, J. M. Assembly of the tight junction: the role of diacylglycerol.J. Cell Biol., 123: 293-302, 1993.

33. Pasdar, M., and Nelson, W. J. Kinetics of desmosome assembly inMadin-Darby canine kidney epithelial cells: temporal and spatial regula-

tion of desmoplakin organization and stabilization upon cell-cell contact.I. Biochemical analysis. J. Cell Biol., 106: 677-685, 1988.

34. Duden, R., and Franke, W. W. Organization of desmosomal plaqueproteins in cells growing at low calcium concentrations. J. Cell Biol., 107:1049-1064, 1988.

35. Mattey, 0. L, and Garrod, 0. R. Splitting and internalization of the

desmosomes of cultured kidney epithelial cells by reduction in calciumconcentration. J. Cell Sci., 85: 95-1 1 1 , 1986.

36. Behrens, J., Weidner, K. M., Frixen, U. H., Schipper, J. H., Sachs, M.,Arakaki, N., Daikuhara, Y., and Birchmeier, W. The role of E-cadherin andscatter factor in tumor invasion and cell motility. In: I. 0. Goldberg (ed),

Cell Motility Factors, pp. 109-126. Basel, Switzerland: Birkhauser Verlag,

1991.

37. Tannapfel, A., Yasui, W., Yokozaki, H., Wittekind, C., and Tahara, E.Effect of hepatocyte growth factor on the expression of E- and P-cadherinin gastric carcinoma cell lines. Virchows Arch. A Pathol. Anat. Hist., 425:

139-144, 1994.

38. Watabe, M., Matsumoto, K., Nakamura, T., Takeichi, M. Effect of

hepatocyte growth factor on cadhermn-mediated cell-cell adhesion. CellStruct. Funct., 18: 117-124, 1993.

39. Zhu, H., Naujokas, M. A., and Park, M. Receptor chimeras indicate

that the Met tyrosmne kinase mediates the motility and morphogenic re-sponses of hepatocyte growth/scatter factor. Cell Growth & Differ., 5: 1-8,1994.

40. Williams, A, F., and Barclay, A. N. The immunoglobulin superfamily:domains for cell surface recognition. Annu. Rev. Immunol., 6: 381-405,1988.

41 . Zhu, H., Najokas, M. A., Fixman, E. 0., Torossian, K., and Park, M.Tyrosmne 1356 in the carboxyl-terminal tail of the HGF/SF receptor isessential for the transduction of signals for cell motility and morphogen-esis. J. Biol. Chem., 269: 29943-29948, 1994.

42. Ponzetto, C., Bardelli, A., Zhen, Z., Mama, F., Dallazonca, P., Gior-dano, S., Graziani, A., Panayotou, G., and Comoglio, P. M. A multifunc-

tional docking site mediates signaling and transformation by hepatocytegrowth factor/scatter factor receptor family. Cell, 77: 261-271 , 1994.

43. Pasdar, M., Li, Z., and Chlumecky, V. Plakoglobin: kinetics of syn-thesis, phosphorylation, stability, and interactions with desmoglemn andE-cadherin. Cell Motil. Cytoskeleton, 32: 258-272, 1995.

44. Gumbiner, B. M. Epithelial morphogenesis. Cell, 69: 385-387, 1992.

45. Matsuyoshi, N., Hamaguchi, M., Taniguchi, S., Nagafuchi, A., Tsukita,S., and Takeichi, M. Cadherin mediated cell-cell adhesion is perturbed byv-src tyrosine phosphorylation in metastatic fibroblasts. J. Cell Biol., 118:

703-714, 1991.

46. Hamaguchi, M., Matsuyoshi, N., Ohnishi, Y., Gotoh, B., Takeichi, M.,and Nagai, Y. p6ovsrc causes tyrosine phosphorylation and inactivation ofthe N-cadherin cell adhesion system. EMBO J., 12: 307-314, 1993.

47. Behrens, J., Vakaet, L, Fritis, R., Winterberger, E., Van Roy, F.,Mareel, M. M., and Birchmeier, W. Loss of epithelial differentiation andgain of invasiveness correlates with tyrosmne phosphorylation of theE-cadherin/f3-catenin complex in cells transformed with a temperature

sensitive v-SRC gene. J. Cell Biol., 120: 757-766, 1993.

48. Nigam, S. K., Denisenko, N., Rodriguez-Boulan, E., and Citi, S. Therole of phosphorylation in development of tight junctions in cultured renalepithelial(MOCK) cells. Biochem. Biophys. Res. Commun., 181: 548-553,1991.

49. Cowin, P., Kapprell, H. P., Franke, W. W., Tamkun, J., and Hynes, R.0. Plakoglobin: a protein common to different kinds of intercellular ad-hering junctions. Cell, 46: 1063-1073, 1986.

50. Troyanovsky, S. M., Troyanovsky, R. B., Eshkind, L G., Leube, R. E.,and Franke, W. W. Identification of amino acid sequence required forplakoglobmn binding and plaque formation. Proc. NatI. Acad. Sci. USA, 91:10790-10794, 1994.


51 . Karnovsky, A., and Klymkowsky, M. W. Anterior axis duplication inXenopus induced by the over-expression of the cadhermn-binding proteinplakoglobin. Proc. NatI. Acad. Sd. USA, 92: 4522-4526, 1994.

52. Takeichi, M. Morphogenetic roles of classic cadherins. Curr. Opin.Cell Biol., 7: 619-627, 1995.

53. Kemler, R., Gossler, A., Mansouri, A., and Vestweber, 0. The celladhesion molecule uvomorulin. In: G. M. Edelman, B. A. Cunningham, and

J. P. Theiry (eds.), Morphoregulatory Molecuies, pp. 41-56. New York:John Wiley & Sons, 1990.

54. Hirano, S., Nose, A., Hatta, K., Kawakami, A., and Takeichi, M. Cal-cium-dependent cell-cell adhesion molecules (cadherins): subclass spec-ificities and possible involvement of actin bundles. J. Cell Biol., 105:

2501-2510, 1987.

55. Shore, E. M., and Nelson, W. J. Biosynthesis of the cell adhesionmolecule uvomorulin (E-cadhermn) in Madin-Darby canine kidney epithelialcells. J. Biol. Chem., 266: 19672-19680, 1991.

56. McNeill, H., Ryan, T. A., Smith, S. J., and Nelson, W. J. Spatial andtemporal dissection of immediate and early events following cadherin-mediated epithelial cell adhesion. J. Cell Biol., 120: 1217-1226, 1993.

57. Shibamoto, S. M., Hayakawa, K., Takeuchi, T., Hori, T., Oku, N.,Miyazawa, K., Kitamura, N., Takeichi, M., and Ito, F. Tyrosine phospho-rylation of the E-cadherin/p-catenin complex in cells transformed with atemperature sensitive vSRC gene. J. Cell Biol., 120: 757-766, 1994.

58. Moorby, C. 0., Stoker, M., and Gherardi, E. HGF/SF inhibits junctional

communication. Exp. Cell Res., 219: 657-663, 1995.

59. Ikejlma, K., Watanabe, S., Kitamura, T., Hirose, M., Miyazaki, A., andSato, N. Hepatocyte growth factor inhibits intercellular communication viagap junctions in rat hepatocytes. Biochem. Biophys. Res. Commun., 214:440-446, 1995.

60. Musil, L S., Cunningham, B. A., Edelman, G. M., and Goodenough, 0.A. Differential phosphorylation of the gap junction protein connexin 43 injunctional communication-competent and deficient cell lines. J. Cell Biol.,111: 2077-2088, 1990.

61 . Sommers, C. L, Gelmann, E. P., Kemler, R., Cowin, P., and Byers, S.W. Alterations in �-catenin phosphorylation and plakoglobin expression in

breast cancer cells. Cancer Res., 54: 3544-3552, 1994.62. Volberg, T., Gieger, B., Dror, R., and Zick, Y. Modulation of intercel-lular adherens-type junctions and tyrosine phosphorylation of their com-

ponents in RSV-transformed cultured chick lens cells. Cell Regul., 2:105-120, 1991.

63. Volberg, T., Zick, Y., Oror, R., Sabanay, I., Gilon, C., Levitzki, A., andGeiger, B. The effect of tyrosine-specific protein phosphorylation on theassembly of adherens-type junctions. EMBO J., 1 1: 1733-1742, 1992.

64. Gumbiner, B. Signal transduction by j3-catenin. Curr. Opin. Cell Biol.,

7: 634-640, 1995.

65. Pelter, M., Sweeton, 0., Casey, M., and Weischaus, E. Winglesssignal and Zeste-white 3 kinase trigger opposing changes in the intracel-lular distribution ofArmadillo. Development (Camb.), 120: 369-380, 1994.

66. Peifer, M. Regulating cell proliferation: as easy as APC. Science

(Washington DC), 272: 974-975, 1996.

67. Nusse, R., and Varmus, H. E. Wnt genes. Cell, 69: 1073-1087, 1992.

68. Huguet, E. L, Smith, K., Bicknell, R., and Harris, A. L Regulation of

Wnt5a mRNA expression in human mammary epithelial cells by cell

shape, confluence, and hepatocyte growth factor. J. Biol. Chem., 270:12851-12856, 1995.

69. Crepaldi, T., Pollack, A. L, Prat, M., Zborek, A., Mostov, K., andComoglio, P. M. Targeting of the SF/HGF receptor to the basolateraldomain of polarized epithelial cells. J. Cell Biol., 125: 313-320, 1994.

70. Ozawa, M., Baribault, H., and Kemler, R. The cytoplasmic domain ofthe cell adhesion molecule uvomorulin associates with three independentproteins structurally related in different species. EMBO J., 8: 171 1-171 7,1989.

71 . Nagafuchi, A., and Takeichi, M. Cell binding function of E-cadhermn isregulated by the cytoplasmic domain. EMBO J., 7: 3679-3684, 1988.

72. Gumbiner, B. M., Stevenson, B. R., and Grimaldi, A. Role of cell

adhesion molecule uvomorulin in formation and maintenance of the epi-thelial junctional complex. J. Cell Biol., 107: 1575-1587, 1988.

73. Hinck, L, Nathke, I. S., Papkoff, J., and Nelson, W. J. Dynamics ofcadherin/catenin complex formation: novel protein interactions and path-ways of complex assembly. J. Cell Biol., 125: 1327-1340, 1994.

74. Nathke, I. S., Hinck, L, Swedlow, J. R., Papkoff, J., and Nelson, W. J.Defining interactions and distributions of cadhenn and catenin complexes

in polarized epithelial cells. J. Cell Biol., 125: 1341-1352, 1994.

75. Pasdar, M., and Nelson, W. J. Regulation of desmosome assembly in

epithelial cells: kinetics of synthesis, transport, and stabilization of des-moglemn I, a major protein of the membrane core domain. J. Cell Biol., 109:

163-177, 1989.

76. Pasdar, M. Biochemical approaches for analyzing de novo assemblyof epithelial junctional components. In: B. R. Stevenson, W. J. Gailin, and0. L Paul (eds.), Cell-Cell Interactions, A Practical Approach, pp. 203-226. Oxford: IRL Press, 1992.

77. Harlow, E., and Lane, 0. (eds.). Antibodies: A Laboratory Manual.Cold Spring Harbor, NY: Cold Spring Harbor Laboratory, 1988.

78. Pasdar, M., and Nelson, W. J. Kinetics of desmosome assembly inMadin-Darby canine kidney epithelial cells: temporal and spatial regula-tion of desmoplakin organization and stabilization upon cell-cell contact.II. Morphological analysis. J. Cell Biol., 106: 687-695, 1988.

inhibition ofjunction assembly incultured epithelial...

Documents