Role of morphogenetic factors in metastasis of mammary carcinoma cells

Silke Meiners, Volker Brinkmann, Helga Naundorf and Walter Birchmeier

Max DelbruÈck Center for Molecular Medicine, Robert-RoÈssle-Str. 10, D-13122 Berlin, Germany

We have analysed the role of the morphogenetic factorshepatocyte growth factor/scatter factor (HGF), neur-egulin and E-cadherin in the process of metastasis andmorphogenesis of mammary carcinoma cells. ThecDNAs for HGF, neuregulin and E-cadherin were stablyexpressed in anaplastic human MDA MB 435 carcinomacells. The altered cells were then injected into themammary fat pads of nude mice, where they formtumors which can spontaneously metastasize to the lungs.We found that expression of HGF or neuregulinpromoted metastasis whereas expression of the celladhesion molecule E-cadherin was inhibitory. Moreover,expression of E-cadherin reconstituted the ability of thecells to form morphogenetic structures in matrigelcultures in response to HGF. These data demonstratethat HGF and neuregulin, which control branching orlobulo-alveolar morphogenesis of normal breast epithe-lium, do have metastasis-promoting e�ects on breastcarcinoma cells. Moreover, our results suggest that thedi�erential activities of the two factors can be explainedby the degree of epithelial di�erentiation: induction ofmorphogenesis requires an intact epithelial adhesion anddi�erentiation system, whereas induction of metastasis isobserved when the cells have lost their epithelialcharacteristics.

Keywords: HGF; neuregulin; E-cadherin; metastasis;morphogenesis; epithelial-mesenchymal transition

Introduction

Metastasis of carcinomas is a multistep processinvolving angiogenesis and invasion of local mesench-yme, intravasation and survival in lymph and bloodvessels and extravasation and growth at secondary sites(Folkman et al., 1989; Nicolson, 1989; Fidler and Ellis,1994). In the recent years, a variety of molecularcomponents have been identi®ed which promote orsuppress metastasis by di�erent means. For instance,invasion of metastatic cells involves uncontrolledaction of matrix-degrading proteases and theirinhibitors (Mignatti and Rifkin, 1993; Stetler-Steven-son et al., 1993; Powell and Matrisian, 1996). More-over, deregulated cell adhesion allows detachment ofmetastatic cells from their old partners and reattach-ment to new ones. Among the widespread members ofthe families of cell adhesion molecules, certaincadherins, integrins, immunoglobulins, selectins, andhyaluronic acid receptors as well as their ligands areimplicated in the spread of metastatic cells (Behrens etal., 1989; Lehmann et al., 1989; Giancotti and

Ruoslahti, 1990; Albelda et al., 1990; Chan et al.,1991; Hynes and Lander, 1992; Birchmeier andBehrens, 1994; GuÈ nthert, 1996). Lastly, growth andmotility factors play an essential role in migrationprocesses at various levels of the metastatic cascade.These factors act in autocrine or paracrine mannersthrough speci®c receptors which mediate di�erentsignals, e.g. through tyrosine phosphorylation (Liottaand Schi�mann, 1988; Rosen et al., 1994; Silletti andRaz, 1996; Boyer et al., 1996).Here we have analysed the role of the growth and

di�erentiation factors HGF and neuregulin and of thecell adhesion molecule E-cadherin in metastasis ofbreast carcinomas. E-cadherin mediates adhesion ofepithelial cells through adherens junctions. Theextracellular part of E-cadherin is responsible forhomophilic interaction with neighboring cells, whilethe cytoplasmic part interconnects the actin cytoskele-ton of epithelial cells via catenins (Kemler, 1993;Tsukita et al., 1993; HuÈ lsken et al., 1994). Armadillo-type catenins have recently been shown to interact withtranscription factors of the LEF-1/TCF-family andmediate signals of the wnt/wingless pathway to the cellnucleus (Behrens et al., 1996; Molenaar et al., 1996).Expression of E-cadherin and the associated cateninswas found to be down-regulated in various carcinomacell lines resulting in increased invasiveness of the cellsin vitro (Behrens et al., 1989; Frixen et al., 1991;Vleminckx et al., 1991; Oyama et al., 1994; Kawanishiet al., 1995). Similarly, downregulation of expression ormutation of the genes that encode proteins of the E-cadherin/catenin complex have been shown to occur invarious human carcinomas such as head and neck(Schipper et al., 1991), breast (Moll et al., 1993;Gamallo et al., 1993; Berx et al., 1995; Zschiesche etal., 1997), colon (Dorudi et al., 1993), bladder(Bringuier et al., 1993; Otto et al., 1994), prostate(Umbas et al., 1994) or stomach carcinomas (Mayer etal., 1993; Becker et al., 1994). In several of thesestudies, the loss of function of components of the E-cadherin/catenin complex has been correlated withreduced survival and poor prognosis (Birchmeier andBehrens, 1994 for review).HGF, which is identical to scatter factor, is

produced mainly by mesenchymal cells and inducesdissociation of various epithelial cells in cell cultureand invasion in vitro (Stoker et al., 1987; Nakamura etal., 1989; Weidner et al., 1991; Rosen et al., 1994). Thereceptor of HGF has been identi®ed as the c-mettyrosine kinase which is expressed in epithelial cells(Bottaro et al., 1991; Naldini et al., 1991; Sonnenberget al., 1993). HGF and c-met play essential roles indevelopment: ablation of the two genes in mice lead toidentical phenotypes, i.e. embryonic lethality due todisturbed placenta and liver development and loss ofmigration of muscle precursor cells (Bladt et al., 1995).The roles of HGF and c-met in tumor formation has

Correspondence: W BirchmeierReceived 25 June 1997; revised 8 August 1997; accepted 8 August1997

Oncogene (1998) 16, 9 ± 20 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00

been extensively studied: concomitant expression of c-met and HGF in tumor cell lines results in enhancedtumorigenicity in nude mice; in mesenchymal cellsmetastasis is promoted (Kanda et al., 1993; Bellusci etal., 1994; Rong et al., 1994; Je�ers et al., 1996).Aberrant expression of HGF and c-met are found invarious tumors of mesenchymal origin such as osteo-sarcomas, sarcomas of patients with Li ±Fraumenisyndrome, Kaposi sarcomas, melanomas and leuke-mias (Natali et al., 1993; Naidu et al., 1994; Rong etal., 1995; Ferracini et al., 1995). Similarly, over-expression of c-met and/or HGF was also found indi�erent human carcinomas such as bladder, gastric,pancreatic, prostatic and thyroid carcinomas (Ebert etal., 1994; Humphrey et al., 1995; Di Renzo et al., 1995;Joseph et al., 1995). In breast carcinoma patients,overexpression of HGF correlates with poor prognosis(Yamashita et al., 1994). Transgenic mice whichoverexpress HGF under the control of a metallothio-nein promoter develop various tumors of mesenchymaland epithelial origin (Takayama et al., 1997).Neuregulin, also known as NDF or heregulin, is the

ligand of the tyrosine kinase receptors c-erbB3 and c-erbB4 and can activate c-erbB2 through heterodimer-ization (cf. Carraway and Cantley, 1994 for a review).A potential role of neuregulin and its receptors intumor progression has been implicated in a variety ofinvestigations: in transgenic mice, overexpression ofneuregulin lead to abnormal development of themammary glands and oncogenic activation of c-erbB2resulted in the formation of metastatic mammarycarcinomas (Guy et al., 1992; Krane and Leder,1996). Enhanced expression of c-erbB-2 is found in avariety of human tumors, e.g. in breast, ovary,stomach, salivary gland and pancreas carcinomas(Slamon et al., 1989; Park et al., 1989; Bacus et al.,1994; Press et al., 1994). In approximately 25% ofprimary breast and ovarian carcinomas, overexpressionof c-erbB2 correlates with poor prognosis (Hynes andStern, 1994). Similarly, overexpression of c-erbB3 andc-erbB4 have been correlated with metastasis and poorclinical outcome (Lemoine et al., 1992; Kraus et al.,1993; Rajkumar and Gullick, 1994).A variety of data has shown that HGF and

neuregulin and their receptors are not only involvedin tumor progression but also in epithelial morphogen-esis. HGF can activate intrinsic morphogeneticprograms of di�erent epithelial cells in three-dimen-sional cultures, e.g. branching morphogenesis of kidneyand mammary cells and crypt formation of intestinalcells (Montesano et al., 1991; Berdichevsky et al., 1994;Brinkmann et al., 1995; Soriano et al., 1995; Niemannet al., 1997). Neuregulin promotes di�erentiation ofmammary epithelial cells in cell culture and theformation of alveolar-like structures in matrigel (Peleset al., 1992; Niemann et al., 1997). Moreover, HGFand neuregulin stimulate branching and lobulo-alveolarmorphogenesis of mammary glands in organ culture,respectively, and both factors are expressed at theappropriate times during development (Yang et al.,1995). Thus, the question arises how the same factorsare able to promote invasive and metastasis on the onehand and morphogenesis and di�erentiation on theother.In the present study, we have investigated the role of

HGF, neuregulin and E-cadherin in the process of

spontaneous metastasis of breast carcinoma cells. ThecDNAs of the three molecules were stably expressed inhuman MDA MB 435 mammary carcinoma cells. Thetransfectants were injected into the mammary fat padsof nude mice where they form primary tumors. Tumorcells then spontaneously metastasize to the lungs, thusre¯ecting the course of metastatic spread in breastcarcinoma patients. This model system thus allowsboth growth of breast carcinoma cells as a primarytumor at the anatomically correct (orthotopic) site andformation of spontaneous metastases controlled by theorgan environment of the host (Fidler, 1990; Price etal., 1990; Leone et al., 1993). Therefore, it provides avaluable tool for the analysis of molecular componentsinvolved in metastasis of mammary carcinomas. Wefound that expression of HGF and neuregulin in theMDA MB 435 mammary carcinoma cells resulted inpromotion of metastasis, whereas expression of E-cadherin lead to suppression of metastasis. We alsoshow that expression of E-cadherin partly reconstitutedthe ability of MDA MB 435 cells to form morphoge-netic structures in response to HGF. These resultsindicate that the growth and di�erentiation factorsHGF and neuregulin are able to promote morphogen-esis or the formation of metastasis depending on thedegree of di�erentiation of epithelial cells.

Results

Characterization of human mammary carcinoma cellsexpressing HGF and neuregulin

The human mammary carcinoma cell line MDA MB435, subline S1, was stably transfected with expressionvectors containing HGF or neuregulin cDNA se-quences (for details see Materials and methods). TheS1 subline was previously isolated in our laboratory(Frixen et al., 1991) and was here found to be non-metastatic when injected into the mammary fat pad ofnude mice (of Materials and methods and below).HGF expressing MDA MB 435 S1 cell clones were ®rstidenti®ed by examining the supernatants of thetransfectants for scattering activity on MDCK cells(Stoker et al., 1987; Weidner et al., 1991). Representa-tive cell clones expressed HGF mRNA in amountsapproximately one order higher than endogenousexpression in MRC-5 ®broblasts, as shown by North-ern blotting (Figure 1a). Parental S1 cells and neo-transfectants did not express HGF. Expression levelscorrelated well with the secreted HGF activity, e.g. cellclone HGF 1 produced 60 ng/ml, MRC-5 ®broblasts6 ng/ml (cf. also Weidner et al., 1991). Neuregulinexpressing MDA MB 435 S1 cell clones were ®rstanalysed on the basis of speci®c mRNA expression.Representative cell clones showed a prominent signal at4.5 kb (close to 28S) in Northern blot analysis (Figure1b). Neo-controls were negative for neuregulin mRNA;CID-9 mammary epithelial cells (cf. Schmidhauser etal., 1990) expressed multiple endogenous transcripts.The MDA MB 435 S1 cell clones NRG 7 and NRG 33secreted approximately 60 ng/ml neuregulin as deter-mined in a bioassay (Plowman et al., 1993), CID-9 cellsalso produced 60 ng/ml.Both HGF and neuregulin expression resulted in a

pronounced alteration of morphology in the trans-

Metastasis of mammary carcinomasS Meiners et al

10

fected MDA MB 435 S1 cells (Figure 2): while thecontrol cells were moderately ®broblastoid with looselyassociated cell clusters, HGF expression resulted in anextremely spindle-shaped morphology and total loss ofintercellular adhesion. When grown at a high densitycontact inhibition was lost and the cells grew multi-layered. Expression of neuregulin lead to a loosening ofcell clusters and the appearance of ®lopodia. Thesecombined data demonstrate that overexpression ofHGF and neuregulin in MDA MB 435 S1 cells resultsin severe morphological changes which are apparently

due to the formation of autocrine loops with therespective receptors.

Induction of spontaneous metastasis by HGF andneuregulin expression in mammary carcinoma cells

The HGF and neuregulin expressing MDA MB 435 S1cells were injected into the mammary fat pad of nudemice, and metastasis to the lungs was examined (cf.Materials and methods). We observed that expressionof both HGF and neuregulin induced strong growth of

28S —

18S —

HGF

actin

neo

HG

F 1

HG

F 4

MR

C-5

a

28S —

18S —

b

NRG

actin

neo

NR

G 7

NR

G 3

3

CID

-9

28S —

18S —

HGF

actin

HG

F 1

cell

cult

ure

HG

F tu

mo

r

HG

F m

etas

tasi

s

neo

tu

mo

r

c

Figure 1 Northern blot analysis of human mammary carcinoma cells (a,b) and tumors (c) showing expression of transfected HGFand neuregulin. HGF1 and HGF4 represent RNA of HGF expressing cell clones, NRG 7 and NRG 33 represent RNA of neuregulinexpressing clones, and neo are corresponding neo-transfected control clones. MRC-5 and CID-9 represent RNA extracts of human®broblasts and mouse mammary epithelial cells which express endogenous HGF and neuregulin, respectively. HGF tumor and HGFmetastasis represent RNA from a HGF expressing primary tumor and corresponding lung metastasis (from animals injected withHGF1 transfectants), neo tumor represents RNA of a corresponding neo-transfected control tumor. The blots in a and c wereprobed with the 2.2 kb XbaI fragment of human HGF cDNA (Hartmann et al., 1992), the blot in b with full length mouse b2neuregulin cDNA (Meyer and Birchmeier, 1994). Blots were also analysed with a b-actin probe. Arrows mark exogenouslyexpressed HGF and neuregulin transcripts in transfectants, arrowheads mark endogenous transcripts in control cells. The reducedtranscript size of exogenously expressed HGF in comparison to endogenous HGF transcripts in MRC-5 cells is due to a shortened3' untranslated region (cf. Weidner et al., 1991)

Figure 2 Morphology of transfected MDA MB 435 S cells. (a) Is the control (neo) cell clone, (b) are HGF and (c) neuregulintransfectants. HGF expressing cell clones appear extremely spindle-shaped, neuregulin expressing cell clones show reduced cell-cell-contacts and develop multiple ®lopodia. Bar, 100 mm

Metastasis of mammary carcinomasS Meiners et al

11

the primary tumor (Table 1A). For instance, tumors of15 ± 20 mm size were formed after 8 weeks with 56105

HGF or neuregulin expressing cells/animal. Tumorswere not visible with control cells at this time butreached a similar size after 24 weeks. Remarkably,metastasis to the lungs was observed in all animalsinjected with HGF and neuregulin transfected cells butnot in the controls, when analysed at the same size ofthe primary tumor (i.e. at a diameter of 19 ± 25 mm,Table 1A). The number of visible lung metastasesinduced by the transfectants ranged from 20/animal tostages where the lungs were completely overgrown bytumor cells. Histological analysis demonstrated theabsence of micro-metastases in lungs of animalsinjected with control MDA MB 435 S1 cells.In a second set of experiments, we compensated for

the di�erent growth rates of the primary tumors byinjecting increased numbers of control cells. Further-more, we ruled out possible clonal e�ects by usingpools of transfectants. No lung metastases were seeneven when a 10-fold higher number of control cellswere injected (56106 cells/animal; Table 1B). Pools ofHGF and neuregulin transfectants were both meta-static (Table 1B, cf. also legend). A second neuregulintransfectant, NRG 57, also produced lung metastaseswhen only 105 cells/animal were injected (Table 1B).We also examined MDA MB 435 S1 transfectantswhich expressed an alternatively spliced variant ofHGF, the two kringle variant (2KV; Hartmann et al.,1992). The HGF-2KV transfectants did not inducemetastasis to the lungs (Table 1B). We then measuredthe expression of HGF in primary tumors of HGFtransfectants and in corresponding metastases byNorthern blotting. HGF mRNA was clearly detectedin the primary tumor and in the lung metastases(Figure 1c); the slightly reduced expression level incomparison to cells in culture might be due to necrosisin the primary tumor and to contaminating lungtissues.

Characterization of E-cadherin expressing humanmammary carcinoma cells

Parental MDA MB 435 cells were transfected with anexpression plasmid containing the E-cadherin cDNAsequence (for details see Materials and methods). Incontrast to the S1 subline used above, the parentalMDA MB 435 cells were highly metastatic wheninjected into the mammary fat pad. We found that

representative cell clones Cad 2 and Cad 4 expressed E-cadherin mRNA at levels similar to epithelial cellsfrom salivary glands, CSG 120/7 (Figure 3a). Westernblot analysis showed production of mature E-cadherinat a molecular weight of 124 kD (Figure 3b; cf. alsoFrixen et al., 1991). Parental MDA MB 435 cells andneo-controls did not express endogenous E-cadherin.Expression of E-cadherin in MDA MB 435 cells

resulted in a shift towards a more epitheloidmorphology (Figure 4a,b; cf. also Frixen et al., 1991).E-cadherin was found to be located at cell-to-cellboundaries by immuno¯uorescence analysis (Figure4f). Moreover, expression of E-cadherin was accom-panied by the redistribution of b-catenin from thecytoplasm and cell nucleus to the plasma membrane(Figure 4g,h), and the reorganization of the actincytoskeleton along the cell borders (Figure 4c,d). Thesedata indicate that exogenous expression of E-cadherinin MDA MB 435 mammary carcinoma cells results inthe formation of a membrane-associated cadherin/catenin complex which provides a link to the actincytoskeleton and allows functional cell-to-cell adhe-sion.

Suppression of spontaneous metastasis by E-cadherinexpression in human mammary carcinoma cells

When E-cadherin-transfected MDA MB 435 cells (105

Cad 4 cells/animal) were injected into the mammary fatpads of nude mice, signi®cant inhibition of growth ofthe primary tumor was observed (Table 2A). Remark-ably, in mice with comparable sizes of the primarytumor expression of E-cadherin resulted in a strongsuppression of metastatic spread to the lungs (Table2A). We again compensated for the di�erences in thegrowth rates of the primary tumors by injecting 10-foldhigher numbers of a single E-cadherin cell clone (Cad2) and a pool of transfectants (Cad pool). E-cadherinexpression again strongly reduced metastasis incomparison to control cells (Table 2B).We also examined the expression of E-cadherin in

the primary tumors and in the few corresponding lungmetastases by immuno¯uorescence analysis. E-cadherinexpression was detected throughout the primarytumors (Figure 5b) but was completely absent in themacroscopically and microscopically visible lungmetastases (Figure 5e). Cytokeratin was used as amarker to indicate epithelial and tumor cells (Figure5c,f).

Table 1 Induction of metastasis by HGF and NRG expression in human mammary carcinoma cells

Expression of No. of injected Tumor size Final tumor Lung metastases/ Lung metastasesTypes of cells factor (ng/ml)a cells/mouse after 8 weeks (mm) size (mm) mouse (mean number)

A) MDA MB 435 S1HGF 1 transfectantNRG 7 transfectant

06060

56105

56105

56105

019+3.315+1.5

19+9.5b

19+3.3c

25+4.6d

0/47/79/9

022+13>30e

B) MDA MB 435 SI(neo-pool)f

HGF poolf

NRG poolf

NRG 57 transfectantHGF-2KV transfectant

0

20302030

56106

56105

56105

105

56105

11+6.7

17+2.34+10

8+0.7

23+4.1g

20+2.4d

20+2.5g

19+3.6b

22+2.3g

0/8

9/94/84/81/9

0

21+9.414+4.4ndnd

aFactors were measured in bioassays as described (Weidner et al., 1991; Plowman et al., 1993). bAnimals were sacri®ced after 24 weeks.c Animals were sacri®ced after 8 weeks. dAnimals were sacri®ced after 10 weeks. e Lungs were overgrown with metastases and therefore countingof single metastases was not possible. f Pools of transfectants were prepared by mixing identical numbers of single cell clones shortly beforeinjection of the cells into the mammary fat pad. gAnimals were sacri®ced after 17 weeks

Metastasis of mammary carcinomasS Meiners et al

12

Major e�orts were undertaken to obtain doubly-transfected MDA MB 435 cells expressing growthfactors and E-cadherin. We did this by starting withboth HGF and neuregulin or E-cadherin expressingcells and transfected these with the second cDNA usingthe selection marker hygromycin. We were unable toobtain MDA MB 435 cells stably expressing both typesof molecules.

Reconstitution of the morphogenetic potential ofmammary carcinoma cells towards HGF by expressionof E-cadherin

A variety of epithelial cell types undergo morphoge-netic programs in response to HGF or neuregulin: forinstance, HGF induces the formation of branchingtubules of kidney epithelial cells in three-dimensionalculture (Montesano et al., 1991; Weidner et al., 1993),and mammary epithelial cells from tubules and alveolarstructures in the presence of HGF and neuregulin,respectively (Soriano et al., 1995; Yang et al., 1995;Niemann et al., 1997). This property is lost inanaplastic epithelial cells which do not express E-cadherin such as MDA MB 435 mammary carcinomacells (cf. Brinkmann et al., 1995).Here we examined MDA MB 435 cells which were

transfected with E-cadherin cDNA for their ability toform morphogenetic structures in three-dimensionalculture in the presence of HGF. Indeed, HGF inducedthe outgrowth of branched cellular cords fromcompact aggregates (Figure 6Ab, 6Ba; cf. quantita-tion in the Figure legend). Actin staining revealed thatthe cells in these rod-like structures made tightcontacts to neighboring cells resulting in cuboidalcell shapes (Figure 6Bc, see enlarged in d). In

contrast, HGF induced cell scattering of clusteredneo-controls (Figure 6Ad, 6Bi). These cells showedreduced contact areas and an elongated morphologywith long protrusions (Figure 6Bi,k,l). Expression ofE-cadherin in the transfected cells were con®rmed byimmuno¯uorescence analysis (Figure 6Bb,f). Neo-controls were negative (Figure 6Bj,n). The morphoge-netic e�ect of HGF in E-cadherin transfectants isdistinct from HGF-induced proliferation, which is alsoobserved in control MDA MB 435 cells (Figure 6Ad,6Bi). The combined results indicate that induction ofcell-cell adhesion and epithelial di�erentiation by E-cadherin suppresses metastasis and reconstitutes themorphogenetic potential of the mammary carcinomacells.

Discussion

In the present study we have examined human MDAMB 435 mammary carcinoma cells, which weretransfected with the cDNAs for HGF, neuregulin orE-cadherin, for their ability to form spontaneous lungmetastases from primary tumors in the mammary fatpad. We found that both HGF and neuregulin stronglypromoted lung metastasis, while metastasis wasabrogated by E-cadherin. Furthermore, expression ofE-cadherin in MDA MB 435 cells partly reconstitutedtheir morphogenetic capacity, i.e. cord-like structureswere induced by HGF in three-dimensional gels. Thesedata suggest that the di�erential response of epithelial(carcinoma) cells towards HGF and neuregulin dependson the degree of cellular di�erentiation: morphogenesisrequires an intact epithelial adhesion and di�erentiationsystem such as a functional E-cadherin/catenin com-

a b

E -cadherin

— 28S

— 18S

E -cadherin

actin

neo

Cad

2

Cad

4

CS

G 1

20/7

CS

G 1

20/7

neo

Cad

2

Cad

4

— 120kD

— 87kD

— 48kD

Figure 3 Northern blot (a) and Western blot (b) analysis showing exogenous expression of E-cadherin in MDA MB 435 cell clones.Cad 2 and Cad 4 represent RNA of representative E-cadherin expressing cell clones, neo are corresponding neo-transfected controlclones. CSG 120/7 is an epitheloid mouse salivary gland cell line which expresses E-cadherin endogenously. The Northern blot wasprobed with the 2.5 kb SacI fragment of mouse E-cadherin cDNA (Frixen et al., 1991) and with a b-actin probe. The Western blotwas developed using DECMA-1 monoclonal antibody against E-cadherin (Vestweber and Kemler, 1985). Arrows indicateexpression of E-cadherin mRNA (a) or protein (b). Molecular weight markers are given on the right

Metastasis of mammary carcinomasS Meiners et al

13

plex, whereas metastasis is observed in cells which haveundergone epithelial-mesenchymal transitions.In order to analyse the role of morphogenetic and

di�erentiation factors in metastasis, we have chosen an

animal model which comprises several steps of themetastatic cascade and re¯ects aspects of metastasis ofhuman breast carcinomas. Transfected human MDAMB 435 mammary carcinoma cells were transplantedinto orthotopic sites, the mammary fat pads of nudemice, which resulted in the formation of primarytumors and spontaneous metastases to the lungs. Wefound that both HGF and neuregulin stronglypromoted metastasis of the MDA MB 435 mammarycarcinoma cells. Our data are thus in line with variousclinical studies which correlate expression of HGF andc-met with poor prognosis and metastasis (cf.Introduction) and with experiments that show metas-tasis of c-met transfected tumor cells of mesenchymalorigin (Rong et al., 1994, Je�ers et al., 1996).Moreover, HGF has recently been described as amigratory factor in development: In c-met and HGF -/-mice, migration of myogenic precursor cells into thelimb anlagen is abolished (Bladt et al., 1995).Inappropriate expression of HGF in transgenic miceresults in abnormal skeletal muscle formation andmelanosis in the nervous system, indicating that HGFinduces migration of myogenic or neural crest-derivedcells to ectopic sites (Takayama et al., 1996). Aninvolvement of c-erbB2 in metastasis has beensuggested by a variety of clinical data (cf. Introduc-tion), and has also been demonstrated in transgenicmice overexpressing c-erbB2 under the control of theMMTV-promoter (Guy et al., 1992). Neuregulin maymediate metastasis of MDA MB 435 cells througherbB3 or erbB4 mediated activation of c-erbB2. BothHGF and neuregulin also had a stimulating e�ect onthe growth of primary tumors. Increased proliferationis consistent with the known mitogenic activities of thetwo factors and with the fact that their receptors, c-metand the c-erbBs, are familiar proto-oncogenes (Park etal., 1986; Bargmann et al., 1986). However, themitogenic e�ect cannot explain induction of metastasissince control tumors of the same size did notmetastasize. Note that we have excluded faster growthof the primary tumor as a parameter in metastasis bydi�erent means (e.g. by injecting increased numbers ofcontrol cells) and that the absence of micro-metastasesin the controls was shown by histological analysis.In contrast, expression of E-cadherin in the human

mammary carcinoma cells suppressed formation ofspontaneous metastases from orthotopic sites. More-over, the few lung metastases obtained with E-cadherin transfected cells had actually lost expressionof E-cadherin. This is consistent with the role of theE-cadherin/catenin system as a suppressor of invasionand metastasis, as suggested by a variety ofexperimental and clinical data (cf. Introduction).Expression of E-cadherin also lead to a reduction of

Figure 4 Phase contrast (b) and immuno¯uorescence staining(d,f,h) of E-cadherin expressing MDA MB 435 mammarycarcinoma cells. (a) and (c,g,e) are corresponding neo transfectedcontrols. Expression of E-cadherin results in an epitheloidmorphology of MDA MB 435 cells as shown by phase-contrastmicrographs and immuno¯uorescence staining for actin (cf. c,d)and E-cadherin (cf. f). In E-cadherin expressing cells b-cateninis located along the cell membranes indicating a functionaladhesion complex (cf. g,h). Exogenous expression of E-cadherinwas detected using DECMA-1 monoclonal antibody (whichis directed against mouse E-cadherin), the absence of endogenous(human) E-cadherin in control cells (e) was con®rmed usingthe 6F9 monoclonal antibody (27). Bar in a,b=100 mm, inc ± e=50 mm

Table 2 Suppression of metastasis by E-cadherin expression in human mammary carcinoma cells

No. of injected Tumor size Final tumor size Lung metastases/ Lung metastasesTypes of cells cells/mouse after 8 weeks (mm) size (mm) mouse (mean number)

A) MD MB 435Cad 4 transfectant

105

10521+5.84+0.5

21+5.8a

10+3.9b6/60/4

14+110

B) MDA MB 435 (pool)Cad 2 transfectantCad pool

105

106

106

16+312+1.510+1.2

22+3.7c

20+2.4d

18+1.6d

9/92/103/7

>30e

2.8+6.26.8+9.9

aAnimals were sacri®ced after 8 weeks. bAnimals were sacri®ced after 15 weeks. c Animals were sacri®ced after 10 weeks. dAnimals weresacri®ced after 14 weeks. e Lungs were overgrown by metastases and counting of single metastasis was therefore not possible

Metastasis of mammary carcinomasS Meiners et al

14

the growth of the primary tumors in comparison tocontrols. When we compensated for the growthdi�erences, we found that metastasis was stillmarkedly inhibited by expression of E-cadherin. Thesuppressive e�ect on cellular proliferation of an intactE-cadherin/catenin system has also been observed byothers (Navarro et al., 1991; Vleminckx et al., 1991;Watabe et al., 1994). Recent experimental evidencehas indicated that the E-cadherin/catenin systemfunctions not only in cell-cell adhesion but also insignal transduction to the cell nucleus. b-Catenin,besides being a component of adherens junctions, isalso a member of the wnt/wingless signaling pathwayand can directly associate with the tumor suppressorgene product APC or transcription factors of theLEF-1/TCF family (Su et al., 1993; Rubinfeld et al.,1993; HuÈ lsken et al., 1994; Gumbiner, 1995; Millerand Moon, 1996; Behrens et al., 1996; Molenaar etal., 1996). The in¯uence of the E-cadherin/catenincomplex on metastasis might thus be a consequence ofhomophilic cell adhesion on the one hand and ofmodulated signal transduction through the wnt/wingless pathway on the other. This is in line withthe observation that expression of E-cadherin inducedthe redistribution of b-catenin from the nucleus andcytoplasm to the plasma membrane.Various recent investigations have shown that HGF

and neuregulin are strong morphogenetic factors formammary epithelial cells (Berdichevsky et al., 1994;Soriano et al., 1995; Brinkmann et al., 1995; Niemannet al., 1997). For instance, HGF induces tubularoutgrowth and branching of mammary epitheliaduring puberty whereas neuregulin promotes lobulo-

alveolar morphogenesis during pregnancy (Yang et al.,1995; Niemann et al., 1997). On a ®rst glance,morphogenesis and metastasis are entirely oppositebiological activities. However, morphogenesis involvescomplex rearrangement of tissues which requirescontinuous cell movement and the breaking andremaking of cell-cell contacts. HGF and its receptor,the tyrosine kinase c-met, can mediate dissociation(scattering) of epithelial cells in culture, but also allowsrearrangement of cells into organoid structures inthree-dimensional matrices (Stoker et al., 1987;Montesano et al., 1991; Brinkmann et al., 1995).Interestingly, morphogenetic activity of c-met couldbe transferred onto a scattering receptor, trkA (thenerve growth factor receptor), by fusing to it the C-terminal substrate binding site of c-met (Sachs et al.,1996). This indicates that morphogenesis is a `moredemanding activity' than scattering. It has recentlybeen shown that the novel multiadaptor protein Gab1transduces the morphogenetic activity of c-met(Weidner et al., 1996; Niemann et al., 1997); whetherGab-1 also transmits signals important for metastasisremains to be examined.Here we also present evidence for the requirement of

a functional E-cadherin/catenin complex in morpho-genesis. We found that E-cadherin (i) suppressesmetastasis of mammary carcinoma cells and (ii) partlyreconstitutes the morphogenetic potential of these cells.We previously screened a variety of cell lines forcellular responses towards HGF and found that *50%of the non-transformed epithelial and 95% of thecarcinoma cells did not respond by morphogenesis(Brinkmann et al., 1995). Many of the non-responding

Figure 5 Cryosections of a primary tumor (a ± c) and a lung metastasis (d ± f) from mice injected with pools of E-cadherintransfected MDA MB 435 cells (cf. Table 2B). Sequential sections were stained with hematoxylin/eosin (a,d; overview) and double-stained for immuno¯uorescence with anti-E-cadherin (b,e) and anti-cytokeratin antibodies (c,f). The squares in a and d mark thearea enlarged for immuno¯uorescence analysis. The selected area in d ± f shows a metastasis (m) bordering normal lung tissue (1).Note, that the metastasis stains for cytokeratin but is negative for E-cadherin

Metastasis of mammary carcinomasS Meiners et al

15

cell lines did not express E-cadherin, such as thehuman mammary carcinoma cells MDA MB 435,indicating that these cells had lost speci®c epithelial celladhesion and thus had undergone permanent epithelial-mesenchymal transition. We hypothesize here that it isthe absence of a functional E-cadherin/catenin adhe-sion system which allows these cells to generatemetastasis instead of morphogenesis. We thereforeexamined the E-cadherin transfected MDA MB 435cells for their potential to undergo morphogenesis. Infact, the transfectants responded to HGF with theformation of cord-like structures when seeded ontomatrigel. This suggests that the degree of differentia-tion of epithelial cells determines the di�erentialresponse towards HGF, and that one of the keyelements of epithelial di�erentiation is a functional E-cadherin/catenin complex. An essential role of E-cadherin in morphogenesis of epithelia has also beendemonstrated in E-cadherin -/- mice: these mice fail toform trophectodermal epithelia and are unable toimplant resulting in death at day 3.5 of embryonaldevelopment (Larue et al., 1994; Riethmacher et al.,1995). Moreover, disturbance of E-cadherin by speci®c

antibodies in mouse mammary glands lead todisruption of ductal structures (Daniel et al., 1995),and expression of a dominant-negative cadherinmutant in the intestine of transgenic mice results insevere defects of crypt formation (Hermiston andGordon, 1995a,b).Our model of the di�erential activities of HGF and

neuregulin in morphogenesis and metastasis is illu-strated in Figure 7: well di�erentiated epithelial cellsrespond towards the two factors with the formation ofmulticellular morphogenetic structures (seen on theleft), which involves in case of HGF a transientepithelial-mesenchymal transition (EMT), i.e. cellscattering. In contrast, carcinoma cells which haveundergone a permanent epithelial-mesenchymal transi-tion during tumor progression (cf. also Reichmann,1994; Boyer et al., 1996) respond to the same factorswith increased cell motility of single tumor cells leadingto metastasis (seen on the right). In case of the humanmammary carcinoma cell line MDA MB 435, thedi�erentiated state and the morphogenetic potentialcould be reconstituted in part by exogenous expressionof E-cadherin.

A

Metastasis of mammary carcinomasS Meiners et al

16

Materials and methods

Cell culture and cell transfection

The metastatic variant of human MDA MB 435 mammarycarcinoma cells was obtained from the American TypeCulture Collection (ATCC, HTB 129; USA), the non-metastatic variant MDA MB 435 S1 was previously subcloned in our laboratory (Frixen et al., 1991). Bothvariants are negative for the expression of HGF,neuregulin and E-cadherin, but do express moderateamounts of c-met, c-erbB2 and 4, as determined byNorthern blotting. The MDA MB 435 cells are clearly ofepithelial origin since they express various cytokeratins butalso vimentin (Leone et al., 1993; S Meiners, unpublishedobservations). The two variants of MDA MB 435 cells arestable in their metastastic properties; the molecular reasonfor their di�erent metastatic behaviour is unknown. Cells

were grown in Dulbecco's modi®ed Eagle medium(DMEM) and 10% heat inactivated fetal calf serum andwere transfected by Ca2+- phosphate coprecipitation withmouse E-cadherin cDNA (pBATEM2; Nagafuchi et al.,1987), human HGF cDNA (in the pBAT expressionplasmid; Hartmann et al., 1992) and the cDNA of mouseb2 neuregulin (2.2 kb XbaI cDNA-fragment in pBAT;Meyer and Birchmeier, 1994) together with a plasmid forG418-resistance (pSV2neo). E-cadherin expressing cellswere initially screened by immuno¯uorescence, HGFexpression was determined by testing the supernatants inthe MDCK colony dissociation assay (Stoker et al., 1987;Weidner et al., 1991), and neuregulin expression wasanalysed by Northern blotting. Expression of E-cadherinand HGF in representative cell clones was veri®ed byNorthern blot analysis. Neuregulin was found to besecreted and biologically active, as determined in amorphology assay on mammary epithelial cells (Plowman

Figure 6 (A) Morphogenesis assay with E-cadherin transfected MDA MB 435 cells (a,b) and neo-cells (c,d) using matrigel cultures. HGFwas added at a concentration of 50 U/ml (b,d). In (a,c), cells are without addition of HGF. The morphogenetic response after HGF-additionwas quantitated by counting branched versus non-branched structures (300 aggregates were counted): the branching factor was 1.4 in E-cadherin expressing cells (b) and 0.3 in control cells (d). (B) Staining of HGF-induced morphogenetic structures in E-cadherin-transfectedMDA MB435 cells (a ± d) for E-cadherin (b,f,j,n) and actin (c,g,k,o). (d,h,l,p) are enlargements of the structures indicated by a square in theprevious colunm. (e ± h) show E-cadherin-transfected cells in the absence of HGF. (i ± l) and (m± p) are neo-control cells in the presence andabsence of HGF, respectively. Conditions of incubations are as in (A). Bars in A and Bc,g,k,o=100 mm, in Bd,h,l,p=50 mm

B

Metastasis of mammary carcinomasS Meiners et al

17

et al., 1993). For Northern blotting, total RNA (10 mg) waselectrophoresed in 1% agarose-formaldehyde gels, trans-ferred to Hybond-N+ membranes (Amersham Interna-tional, UK) and hybridized to 32P-labeled cDNA probesfollowed by auto radiography.

Immunological methods

E-cadherin expression in stably transfected MDA MB 435cells was determined by Western blotting: cells wereextracted in bu�er containing 140 mM NaCl, 4.7 mM

KCl, 0.7 mM MgSO4, 1.2 mM CaCl2, 1% Triton-X-100,1 mM phenyl methyl sulfonyl ¯uorid, 10 mM HEPES,pH 7.4, (Behrens et al., 1989). Equal amounts of proteinwere separated by SDS gel electrophoresis on 7.5%polyacrylamide gels, transferred to nitrocellulose mem-brane (Millipore Corporation, USA), probed with theDECMA-1 hybridoma supernatant (anti-mouse E-cadher-in-antibody, Vestweber and Kemler, 1985), and speci®cantigens were detected using the enhanced chemolumi-nescence system (ECL, Amersham International). Forimmuno¯uorescence detection of E-cadherin, b-cateninand actin, cells were seeded onto glass slides, ®xed in 4%paraformaldehyde for 10 min at room temperature,permeabilized with 0.5% Triton-X 100 and incubatedwith DECMA-1 antibody (2 mg/ml) and polyclonal rabbit

antiserum against b-catenin (1 : 400 for 1 h at 378C. Afterwashing, cells were incubated with Cy5-conjugated goatanti-rabbit, Cy3-conjugated goat anti-rat antibody or Cy3-conjugated goat anti-mouse, respectively (1 : 100, Dianova,Germany) and FITC-conjugated phalloidin (1 : 100, Sigma-Aldrich, Germany) for 1 h at 378C and embedded inMowiol (Calbiochem, USA). For double-immuno¯uor-escence detection of E-cadherin and cytokeratins intissues, 16 mm cryosections were ®xed in acetone for10 min at room temperature, incubated with DECMA-1and anti-pan cytokeratin antibody (C-2562; Sigma-Aldrich,Germany), and detected using CY3-conjugated goat anti-rat and CY2-conjugated goat anti-mouse antibody(Biotrend, Germany). The sections were analysed using aLeica TCS confocal microscope.

Orthotopic injection of MDA MB 435 cells into the mammaryfat pad of nude mice

Female athymic nude mice (NMRI-nu), 6 ± 8 weeks old,were obtained from Bomholdgard Breeding and ResearchCenter, Denmark, and housed under speci®c pathogen-freeconditions. Mice were anesthetized with Radenarkon(40 mg/kg; Arzneimittelwerke Dresden, Germany). A5 mm incision was then made in the skin in order toexpose the mammary fat pad, and 30 ml cell suspension in

Figure 7 Schematic representation of the di�erential e�ects of factors in morphogenesis and metastasis. On the left, epithelial cellswith intrinsic morphogenetic potential are induced by HGF to undergo transient epithelial-mesenchymal transitions (EMT) allowingrearrangement of the cells into morphogenetic structures such as tubules or alveolae. On the right, anaplastic epithelial cells whichhave undergone permanent EMT, respond to morphogenetic factors by metastasis. Functional E-cadherin is an importantdeterminant which di�erentiates between morphogenesis and metastasis, since metastatic cells could be reverted by exogenous E-cadherin expression into morphogenetically active cells

Metastasis of mammary carcinomasS Meiners et al

18

PBS was injected into the fat pad. The skin incisions wereclosed with wound clips which were removed 1 week later.Primary tumor growth was measured once weekly using acaliper. Mice were sacri®ced when moribund or when theprimary tumor had reached an approximate size of 20 mmin diameter. At autopsy, organs were examined for grossmetastases, and macroscopic metastases in the lungs werecounted. Primary tumor, anxillary lymph nodes, the lung,and any suspected metastatic lesion were para�n-embedded for histological analysis or shock-frozen forexpression analysis. The formation of micro metastases inthe lung and in the lymph nodes was determined by serialsectioning of the para�n-embedded organs and subsequenthistological analysis.

Morphogenesis assay

Glass coverslips (Nunc, Denmark) were coated with 20 mlmatrigel (basement membrane from Engelbreth-Holm-

Swarm tumor; Serva, Germany), and cells were seeded ata density of 16104/well and cultured in DMEM containing10% fetal calf serum and hormones (bovine prolactin andinsulin, 5 mg/ml, Sigma; hydrocortisone, 1 mg/ml, Merck,Germany) as described (Barcellos-Ho� et al., 1989;Niemann et al., 1997). On the fourth day, medium wasreplaced by medium containing hormones and HGF at aconcentration of 50 U/ml (Weidner et al., 1993). Mediumwas changed daily, and after 7 days of culture photo-micrographs of the aggregates were taken with a Zeiss lightmicroscope equipped with Nomarski interference optics.

AcknowledgementsWe thank Drs C Birchmeier and J Behrens for helpfuldiscussions and critically reading the manuscript, BBuÈ ttner for excellent technical assistance with nude miceand the Deutsche Krebshilfe for ®nancial support.

References

Albelda SM, Oliver PD, Romer LH and Buck CA. (1990). J.Cell Biol., 110, 1227 ± 1237.

Bacus SS, Zelnick CR, Plowman G and Yarden Y. (1994).Am. J. Clin. Pathol., 102, S13 ± 24.

Barcellos-Ho� MH, Aggeler J, Ram TG and Bissell MJ.(1989). Development, 105, 223 ± 235.

Bargmann CI, Hung MC and Weinberg RA. (1986). Cell, 45,649 ± 657.

Becker KF, Atkinson MJ, Reich U, Becker I, Nekarda H,Siewert JR and Ho¯er H. (1994). Cancer Res., 54, 3845 ±3852.

Behrens J, von Kries JP, Kuhl M, Bruhn L, Wedlich D,Grosschedl R and Birchmeier W. (1996). Nature, 382,638 ± 642.

Behrens J, Mareel MM, Van Roy FM and Birchmeier W.(1989). J. Cell Biol., 108, 2435 ± 2447.

Bellusci S, Moens G, Gaudino G, Comoglio P, Nakamura T,Thiery JP and Jouanneau J. (1994). Oncogene, 9, 1091 ±1099.

Berdichevsky F, Alford D, D'Souza B and TaylorPapadimitriou J. (1994). J. Cell Sci., 107, 3557 ± 3568.

Berx G, Cleton Jansen AM, Nollet F, de Leeuw WJ, van deVijver M, Cornelisse C and van Roy F. (1995). EMBO J.,14, 6107 ± 6115.

Birchmeier W and Behrens J. (1994). Biochim. Biophys. Acta,1198, 11 ± 26.

Bladt F, Riethmacher D, Isenmann S, Aguzzi A andBirchmeier C. (1995). Nature, 376, 768 ± 771.

Bottaro DP, Rubin JS, Faletto DL, Chan AM, Kmiecik TE,Vande Woude GF and Aaronson SA. (1991). Science, 251,802 ± 804.

Boyer B, Valles AM and Thiery JP. (1996). Curr. Top.Microbiol. Immunol., 213, 179 ± 194.

Bringuier PP, Umbas R, Schaafsma HE, Karthaus HF,Debruyne FM and Schalken JA. (1993). Cancer Res., 53,3241 ± 3245.

Brinkmann V, Foroutan H, Sachs M, Weidner KM andBirchmeier W. (1995). J. Cell Biol., 131, 1573 ± 1586.

Carraway KL, 3. and Cantley LC. (1994). Cell, 78, 5 ± 8.Chan BM, Matsuura N, Takada Y, Zetter BR and HemlerME. (1991). Science, 251, 1600 ± 1602.

Daniel CW, Strickland P and Friedmann Y. (1995). Dev.Biol., 169, 511 ± 519.

Di Renzo MF, Poulsom R, Olivero M, Comoglio PM andLemoine NR. (1995). Cancer Res., 55, 1129 ± 1138.

Dorudi S, She�eld JP, Poulsom R, Northover JM and HartIR. (1993). Am. J. Pathol., 142, 981 ± 986.

Ebert M, Yokoyama M, Friess H, Buchler MW and Korc M.(1994). Cancer Res., 54, 5775 ± 5778.

Ferracini R, Di Renzo MF, Scotlandi K, Baldini N, OliveroM, Lollini P, Cremona O, Campanacci M and ComoglioPM. (1995). Oncogene, 10, 739 ± 749.

Fidler IJ and Ellis LM. (1994). Cell, 79, 185 ± 188.Fidler IJ. (1990). Cancer Res., 50, 6130 ± 6138.Folkman J, Watson K, Ingber D and Hanahan D. (1989).

Nature, 339, 58 ± 61.Frixen UH, Behrens J, Sachs M, Eberle G, Voss B, Warda A,Lochner D and Brichmeier W. (1991). J. Cell Biol., 113,173 ± 185.

Gamallo C, Palacios J, Suarez A, Pizarro A, Navarro P,Quintanilla M and Cano A. (1993). Am. J. Pathol., 142,987 ± 993.

Giancotti FG and Ruoslahti E. (1990). Cell, 60, 849 ± 859.Gumbiner BM. (1995). Curr. Opin. Cell Biol., 7, 634 ± 640.Gunthert U. (1996). Curr. Top. Microbiol. Immunol., 213,271 ± 285.

Guy CT, Webster MA, Schaller M, Parsons TJ, Cardi� RDand Muller WJ. (1992). Proc. Natl. Acad. Sci. USA, 89,10578 ± 10582.

Hartmann G, Naldini L, Weidner KM, Sachs M, Vigna E,Comoglio PM and Birchmeier W. (1992). Proc. Natl.Acad. Sci. USA, 89, 11574 ± 11578.

Hermiston ML and Gordon JI. (1995a). Science, 270, 1203 ±1207.

Hermiston ML and Gordon JI. (1995b). J. Cell Biol., 129,489 ± 506.

Hulsken J, Birchmeier W and Behrens J. (1994). J. Cell Biol.,127, 2061 ± 2069.

Humphrey PA, Zhu X, Zarnegar R, Swanson PE, Ratli� TL,Vollmer RT and Day ML. (1995). Am. J. Pathol., 147,386 ± 396.

Hynes NE and Stern DF. (1994). Biochim. Biophys. Acta,1198, 165 ± 184.

Hynes RO and Lander AD. (1992). Cell, 68, 303 ± 322.Je�ers M, Rong S, Anver M and Vande Woude GF. (1996).

Oncogene, 13, 853 ± 856.Joseph A, Weiss GH, Jin L, Fuchs A, Chowdhury S,O'Shaugnessy P, Goldberg ID and Rosen EM. (1995). J.Natl. Cancer Inst., 87, 372 ± 377.

Kanda H, Tajima H, Lee GH, Nomura K, Ohtake K,Matsumoto K, Nakamura T and Kitagawa T. (1993).Oncogene, 8, 3047 ± 3053.

Kawanishi J, Kato J, Sasaki K, Fujii S, Watanabe N andNiitsu Y. (1995). Mol. Cell Biol., 15, 1175 ± 1181.

Kemler R. (1993). Trends Genet., 9, 317 ± 321.Krane IM and Leder P. (1996). Oncogene, 12, 1781 ± 1788.Kraus MH, Fedi P, Starks V, Muraro R and Aaronson SA.(1993). Proc. Natl. Acad. Sci. USA, 90, 2900 ± 2904.

Metastasis of mammary carcinomasS Meiners et al

19

Larue L, Ohsugi M, Hirchenhain J and Kemler R. (1994).Proc. Natl. Acad. Sci. USA, 91, 8263 ± 8267.

Lehmann JM, Riethmuller G and Johnson JP. (1989). Proc.Natl. Acad. Sci. USA, 86, 9891 ± 9895.

Lemoine NR, Barnes DM, Hollywood DP, Hughes CM,Smith P, Dublin E, Prigent SA, Gullick WJ and Hurst HC.(1992). Br. J. Cancer, 66, 1116 ± 1121.

Leone A, Flatow U, VanHoutte K and Steeg PS. (1993).Oncogene, 8, 2325 ± 2333.

Liotta LA and Schi�mann E. (1988). Cancer Surv., 7, 631 ±652.

Mayer B, Johnson JP, Leitl F, Jauch KW, Heiss MM,Schildberg FW, Birchmeier W and Funke I. (1993).Cancer Res., 53, 1690 ± 1695.

Meyer D and Birchmeier C. (1994). Proc. Natl. Acad. Sci.USA, 91, 1064 ± 1068.

Mignatti P and Rifkin DB. (1993). Physiol. Rev., 73, 161 ±195.

Miller JR and Moon RT. (1996). Genes and Development, 10,2527 ± 2539.

Molenaar M, van de Wetering, M Oosterwegel M, PetersonMaduro J, Godsave S, Korinek V, Roose J, Destree O andClevers H. (1996). Cell, 86, 391 ± 399.

Moll R, Mitze M, Frixen UH and Birchmeier W. (1993). Am.J. Pathol., 43, 1731 ± 1742.

Montesano R, Matsumoto K, Nakamura T and Orci L.(1991). Cell, 67, 901 ± 908.

Naidu YM, Rosen EM, Zitnick R, Goldberg I, Park M,Naujokas M, Polverini PJ and Nickolo� BJ. (1994). Proc.Natl. Acad. Sci. USA, 91, 5281 ± 5285.

Nagafuchi A, Shirayoshi Y, Okazaki K, Yasuda K andTakeichi M. (1987). Nature, 329, 341 ± 343.

Nakamura T, Nishizawa T, Hagiya M, Seki T, ShimonishiM, Sugimura A, Tashiro K and Shimizu S. (1989). Nature,342, 440 ± 443.

Naldini L, Vigna E, Narsimhan RP, Gaudino G, ZarnegarR, Michalopoulos GK and Comoglio PM. (1991).Oncogene, 6, 501 ± 504.

Natali PG, Nicotra MR, Di Renzo MF, Prat M, Bigotti A,Cavaliere R and Comoglio PM. (1993). Br. J. Cancer, 68,746 ± 750.

Navarro P, Gomez M, Pizarro A, Gamallo C, Quintanilla Mand Cano A. (1991). J. Cell Biol. , 115, 517 ± 533.

Nicolson GL. (1989). Curr. Opin. Cell Biol., 1, 1009 ± 1019.Niemann C, Brinkmann V, Spitzer E, Hartmann G, Sachs M,Naundorf H and Birchmeier W. (1997). J. Cell Biol.,submitted.

Otto T, Birchmeier W, Schmidt U, Hinke A, Schipper J,Rubben H and Raz A. (1994). Cancer Res., 54, 3120 ±3123.

Oyama T, Kanai Y, Ochiai A, Akimoto S, Oda T,Yanagihara K, Nagafuchi A, Tsukita S, Shibamoto S,Ito F, Takeichi M, Matsuda H and Hirohashi S. (1994).Cancer Res., 54, 6282 ± 6287.

Park JB, Rhim JS, Park SC, Kimm SW and Kraus MH.(1989). Cancer Res., 49, 6605 ± 6609.

Park M, Dean M, Cooper CS, Schmidt M, O'Brien SJ, BlairDG and Vande Woude GF. (1986). Cell, 45, 895 ± 904.

Peles E, Bacus SS, Koshi RA, Lu HS, Wen D, Ogden SG,Levy RB and Yarden Y. (1992). Cell, 69, 205 ± 216.

Plowman GD, Culouscou JM, Whitney GS, Green JM,Carlton GW, Foy L, Neubauer MG and Shoyab M.(1993). Proc. Natl. Acad. Sci. USA, 90, 1746 ± 1750.

Powell WC and Matrisian LM. (1996). Curr. Top. Microbiol.Immunol., 213, 1 ± 21.

Press MF, Pike MC, Hung G, Zhou JY, Ma Y, George J,Dietz Band J, James W, Slamon DJ, Batsakis JG et al.(1994). Cancer Res., 54, 5675 ± 5682.

Price JE, Polyzos A, Zhang RD and Daniels LM. (1990).Cancer Res., 50, 717 ± 721.

Rajkumar T and Gullick WJ. (1994). Breast Cancer Res.Treat., 29, 3 ± 9.

Reichmann E. (1994). Semin. Cancer Biol., 5, 157 ± 165.Riethmacher D, Brinkmann V and Birchmeier C. (1995).

Proc. Natl. Acad. Sci. USA, 92, 855 ± 859.Rong S, Donehower LA, Hansen MF, Strong L, Tainsky M,Je�ers M, Resau JH, Hudson E, Tsarfaty I and VandeWoude GF. (1995). Cancer Res., 55, 1963 ± 1970.

Rong S, Segal S, Anver M, Resau JH and Vande Woude GF.(1994). Proc. Natl. Acad. Sci. USA, 91, 4731 ± 4735.

Rosen EM, Nigam SK and Goldberg ID. (1994). J. CellBiol., 127, 1783 ± 1787.

Rubinfeld B, Souza B, Albert I, Muller O, Chamberlain SH,Masiarz FR, Munemitsu S and Polakis P. (1993). Science,262, 1731 ± 1734.

Sachs M, Weidner KM, Brinkmann V, Walther I, ObermeierA, Ullrich A and Birchmeier W. (1996). J. Cell Biol., 133,1095 ± 1107.

Schipper JH, Frixen UH, Behrens J, Unger A, Jahnke K andBirchmeier W. (1991). Cancer Res., 51, 6328 ± 6337.

Schmidhauser C, Bissell MJ, Myers CA and Casperson GF.(1990). Proc. Natl. Acad. Sci. USA, 87, 9118 ± 9122.

Silletti S and Raz A. (1996). Curr. Top. Microbiol. Immunol.,213, 137 ± 1701.

Slamon DJ, Godolphin W, Jones LA, Holt JA, Wong SG,Keith DE, Levin WJ, Stuart SG, Udove J, Ullrich A et al.(1989). Science, 244, 707 ± 712.

Sonnenberg E, Meyer D, Weidner KM and Birchmeier C.(1993). J. Cell Biol., 123, 223 ± 235.

Soriano JV, Pepper MS, Nakamura T, Orci L andMontesano R. (1995). J. Cell Sci., 108, 413 ± 430.

Stetler Stevenson WG, Aznavoorian S and Liotta LA.(1993). Annu. Rev. Cell Biol., 9, 541 ± 573.

Stoker M, Gherardi E, Perryman M and Gray J. (1987).Nature, 327, 239 ± 242.

Su LK, Volgelstein B and Kinzler KW. (1993). Science, 262,1734 ± 1737.

Takayama H, Larochelle J, Sharp R, Otsuka T, Kriebel P,Anver M, Aaronson SA and Merlino G. (1997). Proc.Natl. Acad. Sci. USA, 94, 701 ± 706.

Takayama H, La Rochelle WJ, Anver M, Bockman DE andMerlin G. (1996). Proc. Natl. Acad. Sci. USA, 93, 5866 ±5871.

Tsukita S, Itoh M, Nagafuchi A and Yonemura S. (1993). J.Cell Biol., 123, 1049 ± 1053.

Umbas R, Isaacs WB, Bringuier PP, Schaafsma HE,Karthaus HF, Oosterhof GO, Debruyne FM andSchalken JA. (1994). Cancer Res., 54, 3929 ± 3933.

Vestweber D and Kemler R. (1985). EMBO J., 4, 3393 ±3398.

Vleninckx K, Vakaet L, Jr Mareel M, Fiers W and van RoyF. (1991). Cell, 66, 107 ± 119.

Watabe M, Nagafuchi A, Tsukita S and Takeichi M. (1994).J. Cell Biol., 127, 247 ± 256.

Weidner KM, DiCesare S, Sachs M, Brinkmann V, Behrens Jand Birchmeier W. (1996). Nature, 384, 173 ± 176.

Weidner KM, Sachs M and Birchmeier W. (1993). J. CellBiol., 121, 145 ± 154.

Weidner KM, Arakaki N, Hartmann G, Vandekerckhove J,Weingart S, Rieder H, Fonatsch C, Tsubouchi H, HishidaT, Daikuhara Y and Birchmeier W. (1991). Proc. Natl.Acad. Sci. USA, 88, 7001 ± 7005.

Yamashita J, Ogawa M, Yamashita S, Nomura K,Kuramoto M, Saishoji T and Shin S. (1994). CancerRes., 54, 1630 ± 1633.

Yang Y, Spitzer E, Meyer D, Sachs M, Niemann C,Hartmann G, Weidner KM, Birchmeier C and BirchmeierW. (1995). J. Cell Biol., 131, 215 ± 226.

Zschiesche W, SchoÈ nborn I, Behrens J, Herrenknecht K,Hartveit F, Lilleng P and Birchmeier W. (1997). Anti-cancer Research, 17, 1 ± 7.

Metastasis of mammary carcinomasS Meiners et al

20