Vol. 2, 365-3 72, August 1991 cell Growth & Differentiation 365

Cell Adhesion Molecule Uvomorulin Expression in HumanBreast Cancer Cell Lines: Relationship to Morphologyand Invasive Capacities

Connie L Sommers, Erik W. Thompson, Jeffrey A. Torn,Rolf Kemlen, Edward P. Gelmann,and Stephen W. Byers’

Vincent T. Lombardi Cancer Research Center [C. L. S., E. W. 1., J. A. T.,

E. P. G., S. W. B.] and Department ofAnatomy and Cell Biology [C. L. S.,E. W. T., E. P. G., S. W. B.], Georgetown University, Washington, DC20007, and Max-Planck-Institut f#{252}rImmunobiologie, Stubeweg 51,

D-7800 Freiberg, Federal Republic of Germany [R. K.]

AbstractLoss of cell-cell adhesion in carcinoma cells may be animportant step in the acquisition of an invasive,metastatic phenotype. We have examined theexpression of the epithelial-specific cell adhesionmolecule uvomorulin (E-cadhenin, cell-CAM 120/80,L-CAM) in human breast cancer cell lines. We find thatfibroblastoid, highly invasive, vimentin-expressingbreast cancer cell lines do not express uvomorulin. Ofthe more epithelial-appearing, less invasive, keratin-expressing breast cancer cell lines, some expressuvomorulin, and some do not. We examined themorphologies of the cell lines in the reconstitutedbasement membrane matrix Matnigel and measured theability of the cells to traverse a Matnigel-coated filteras in vitro models for detachment of carcinoma cellsfrom neighboring cells and invasion through basementmembrane into surrounding tissue. Colonies ofuvomorulin-positive cells have a characteristic fusedappearance in Matrigel, whereas uvomorulin-negativecells appear detached. Cells which are uvomorulinnegative and vimentin positive have a stellatemorphology in Matnigel. We show that uvomorulin isresponsible for the fused colony morphology inMatnigel since treatment of uvomorulin-positive MCF-7cells with an antibody to uvomorulin caused the cells todetach from one another but did not induceinvasiveness in these cells, as measured by their abilityto cross a Matrigel-coated polycarbonate filter in amodified Boyden chamber assay. Two uvomorulin-negative, vimentin-negative cell lines are also nothighly invasive as measured by this assay. We suggestthat loss of uvomorulin-mediated cell-cell adhesionmay be one of many changes involved in theprogression of a carcinoma cell to an invasivephenotype.

IntroductionCAMs2 may play a role in the invasion and metastasis of

carcinoma cells. Metastasis represents a critical event inthe course of tumor progression, usually leading to tumorlethality in cancer patients. Invasion and metastasis oftumor cells into their surrounding connective tissue andto distant sites is a multistep process. Carcinoma cellsmust first detach from neighboring tumor cells, invadethrough the basement membrane, and migrate into sur-rounding tissue. The invading cells may then enter thevascular or lymphatic circulation. Finally, the cells maythen attach to the wall of a capillary at a second site andpass through the basement membrane into a target or-gan, where growth resumes. CAMs may play a role bothin the detachment of tumor cells from a primary tumorand in reimplantation at a distant site. Other factors thatare likely to be involved in invasion and metastasis arecollagenases and other proteases, motility factors, andsubstrate adhesion molecules (1). In this study, we ex-amine the expression of the CAM uvomorulin in breastcancer cell lines and propose that loss of uvomorulin-mediated cell-cell adhesion may be one event requiredfor initial local invasion potentially leading to metastasisin these cells.

Uvomorulin (E-cadherin, cell-CAM 120/80, L-CAM) isa 120 kilodalton cell surface glycoprotein involved incalcium-dependent cell-cell adhesion (see Refs. 2-5 forreviews). It is expressed early in mouse development andis involved in the Compaction of preimplantation em-bryos. In the adult, it is expressed specifically in epithelialcells. Disruption of uvomorulin function in nontu-morigenic MDCK epithelial cells by specific antibodytreatment resulted in a change from a cobblestone,epithelioid appearance to a more elongated, fibroblasticmorphology when the cells were plated on plastic. Asmall percentage of the antibody-treated cells also ac-quired the ability to invade collagen gels and embryonalheart tissue (6).

To investigate a possible correlation between CAMexpression and invasiveness in human breast cancer, weanalyzed 1 7 human breast cancer cell lines for theexpression of uvomorulin. We have previously describedthe morphology of these cell lines, which correlated withexpression of the intermediate filament protein vimentin(7). The cell lines which appeared fibroblast-like ex-

pressed vimentin, whereas the epithelioid cell lines ex-pressed only keratins. There is increasing evidence thatvimentin may be a marker for poor prognosis in breastcancer(8-10), and vimentin-expressing breast cancer celllines are more invasive as measured by in vitro assays.3Here, we find that fibroblastoid, invasive, vimentin-ex-pressing breast cancer cell lines do not express uvomo-

Received 12/26/90.I To whom requests for reprints should be addressed, at the Departmentof Anatomy and Cell Biology, Georgetown University, Washington, DC20007.

2 The abbreviations used are: CAM, cell adhesion molecule; MDCK cells,

Madin-Darby canine kidney cells; cDNA, complementary DNA; IMEM,

Iscove’s minimal essential medium; SSC, standard saline citrate.3 E. W. Thompson et a!., submitted for publication.

366 Uvomorulin Expression in Breast Cancer Cell Lines

rulin. Another subset of keratin-expressing (vimentin-negative), less invasive cell lines also does not expressuvomorulin. In general, these cell lines are not moreinvasive than other vimentin-negative, uvomorulin-posi-tive breast cancer cell lines, as measured by an in vitroassay in which cells must penetrate a basement mem-brane matrix-coated, porous, polycarbonate filter. Theseresults indicate that although loss of uvomorulin-me-diated cell-cell adhesion may facilitate invasion and me-tastasis, absence of uvomorulin expression does not, byitself, lead to an invasive phenotype as detected by thein vitro assay described here.

Results

Seventeen breast cancer cell lines were analyzed foruvomorulin expression by immunofluorescence micro-scopy and Northern blot analysis. Cell lines positive foruvomorulin expression, such as MCF-7, T47D, and MDA-MB-i 75-7 (Fig. 1 , a-c) showed bright staining at cell-cellborders by immunofluorescenCe microscopy similar tothat previously described for MCF-7 (1 1) and other epi-thelial cells such as MDCK cells (6). Negative cell linesshowed no specific cell-cell border staining (Fig. 1, d-f).A summary of uvomorulin expression in the 1 7 cell linesexamined is presented in Table 1.

Northern blot analysis was performed to investigatewhether uvomorulin mRNA was present in the breastcancer cell lines that were negative by immuno-fluorescence microscopy. As shown in Fig. 2a, a band ofapproximately 5 kilobases was seen in three breast can-cer cell lines that were positive by immunofluorescencemicroscopy (MCF-7, BT474, and MDA-MB-361 ). No uvo-morulin mRNA could be detected in cell lines MDA-MB-231 and HS578T, even when MDA-MB-231 polyadeny-lated RNA was used (Fig. 2b). To show the presence ofintact RNA in the MDA-MB-231 and HS578T lanes in Fig.2a, the blot was subsequently hybridized to a vimentincDNA probe. Both MDA-MB-231 and HS578T expressthe intermediate filament protein vimentin as previouslydescribed (7), whereas MCF-7, BT474, and MDA-MB-361 do not express vimentin. It is interesting to note thatthe MCF-7ADR line, selected for resistance to adriamycin(12), has lost expression of uvomorulin and acquiredvimentin expression.

We investigated the morphology of the breast cancercell lines plated in the reconstituted basement membranematrix Matrigel, since potentially invasive carcinoma cellsare in contact with basement membrane in vivo. Theuvomorulin-positive, vimentin-negative cells form large,well-circumscribed colonies of cells in Matrigel in whichindividual cell borders are not distinguishable, and cellsappear fused (Fig. 3, a and b). Uvomorulin-negative,vimentin-negative cells form clusters of spherical cells inMatrigel in which cell borders are easily distinguishable(Fig. 3, c and d). The uvomorulin-negative, vimentin-positive cells have a stellate morphology in Matrigel (Fig.3, e and 1).

To investigate whether the phenotype of fused cob-nies versus colonies composed of separated cells wasindeed dependent on uvomorulin expression, uvomo-rulin-positive MCF-7 cells were treated with gp84 rabbitpolyclonal antiserum to uvomorulin and DECMA-1 ratmonoclonal antibody to uvomorubin, both of which havebeen shown to block cell-cell interaction in mouse preim-

plantation embryos (13, 14). As illustrated in Fig. 4, b ande, treatment of MCF-7 cells with either the gp84 or theDECMA-i antibody caused the cells to detach from eachother in the Matrigel outgrowth assay, giving a morphol-ogy similar to that seen in uvomorulin-negative, vimentin-negative cells. This effect was seen whether the cellswere treated with antibody at the time of plating orwhether antibody was added the day after plating. Cellsbegan to detach within 4 h after antibody treatment. Inaddition, this effect was reversible since, after removalofgp84 and DECMA-1 antibodies by changing the media,the cells exhibited the fused, uvomorulin-positive phe-notype seen in untreated cells (Fig. 4c). As negativecontrols, MCF-7 cells were treated with similar concen-trations of normal rabbit serum (Fig. 4d) or a rat mono-clonal antibody to CD44/hyaluronate receptor (15, 16)(Fig. 41). CD44/hyaluronate receptor is an actin-assoCi-ated cell adhesion molecule involved in lymphocytehoming and in interactions with extracellular hyaluronate(16). After treatment with normal rabbit serum or CD44/hyaluronate receptor antibody, MCF-7 colonies in Matri-gel remained fused, demonstrating the specific involve-ment of uvomorulin in this phenotype.

As an in vitro measure of invasive capacities, we testedthe abilities of some of the breast cancer cell lines tocross a Matrigel-coated, porous, polycarbonate filter in amodified Boyden chamber assay. Three uvomorulin-pos-itive, vimentin-negative cell lines were tested, as weretwo uvomorulin-negative, vimentin-negative and twouvomorulin-negative, vimentin-positive lines (Table 2).The vimentin-negative lines were less invasive than thevimentin-positive lines, as previously reported.3 Of thevi mentin-negative cell lines, the uvomorulin-negativelines CAMA1 and SK-BR-3 were not more invasive thanuvomorulin-positive cells as measured by this assay.Since these cell lines were not more invasive, we alsotested the effect on invasiveness of disrupting uvomoru-bin function in uvomorulin-positive MCF-7 cells. In aseparate experiment, MCF-7 cells were treated with thesame concentration of DECMA-1 antibody to uvomorulinthat produced an effect in the Matrigel outgrowth assayand were assayed in the modified Boyden chamberchemoinvasion assay. The antibody did not increase in-vasiveness in MCF-7 cells; in fact, invasiveness was some-what decreased. A similar effect was observed in theuvomorulin-positive cell lines T47D and BT474 (data notshown). As can be observed by Comparing Tables 2 and3, some interassay variation was observed; therefore, thecells listed in Table 2 were assayed at one time, as werethe cells listed in Table 3. Treatment of MCF-7 cells withthe anti-CD44/hyaluronate receptor antibody had noeffect on invasion (Table 3). Therefore, in this assaysystem, loss of uvomorulin function did not, by itself,lead to increased invasiveness.

Discussion

A significant role for CAMs in malignant progression issuggested by several recent studies. A gene with homol-ogy to neural CAM, DCC, is a putative tumor suppressorgene deleted in 70% of coborectal cancers (1 7). This genemay represent a novel CAM whose loss contributes tomalignant progression in coborectal cancer. Transforma-tion of B-lymphoblastoid cells by the myc oncogeneresulted in a specific down-regulation of the integrin

d

e

Cell Growth & Differentiation 367

fig. 1. Uvomorulin expression as detected by immunofluorescence microscopy. Immunofluorescence microscopy was performed as described in“Materials and Methods” usingthe rabbit polyclonal antibody against uvomorulin, gp84. a, MCF-7; h, T47D; c, MDA-MB-175-7; d, HS578T; e, MDA-MB-

231;t, BT549.

, Vimentin mRNA as detected by Northern blot analysis.

b Uvomorulin expression as detected by Northern blot analysis (mRNA)

or immunofluorescence microscopy (IF), as described in ‘Materials and

Methods.�( Spherical for about 10 days, then stellate.

aH

(0 (Y)

c’_) Nth co I-

r� � � �ii. N- < < U�)

0 �: o 0 �i)� co � �

b

Hci,Nth

00

,

#{149}#{149}..- vim

Fig. 2. Northern blot analysis of

uvomorulin expression. TwentyILg of total cellular RNA (a) or 10

Mg of polyadenylated selectedRNA (b) were analyzed byNorthern blot analysis. In a, the

- uvo blot was hybridized to a mouse

uvomorulin (uvo) cDNA probe

(27) and subsequently to a hu-man vimentin (vim) cDNA probe(7). In b, the mouse uvomorulin

(uvo) cDNA probe was used.

368 Uvomorulin Expression in Breast Cancer Cell Lines

Tab!e 1 Uvomorulin expression in human breast can cer cell lines

Cell line VimentinUvomorulin”

mRNA IF

Morphology.in Matrigel

MDA-MB-361 - +++ + FusedMDA-MB-175-7 - +++ + FusedBT483 - +++ + Fused

T47D - ++ + FusedZR-75-B - ++ + Fused

BT474 - ++ + FusedMCF-7 - ++ + FusedMDA-MB-468 - + - FusedSK-BR-3MDA-MB-134

CAMA1MDA-MB-453

MDA-MB-231

-

-

-

-

+

- -

- -

- -

- -

- -

SphericalSpherical

SphericalSpherical

StellateMDA-MB-436 4- - - StellateBT549 + - - Stellate

HS578T + - - StellateMDF-7ADR + - - SteIIate�

CAM lymphocyte function-associated antigen 1 and adecrease in cell-cell adhesion (18). We believe that thesestudies and others point to the involvement of decreased

homotypic cell-cell adhesion in tumor progression. Lossof cell-cell adhesion may occur by loss of expression, asobserved in our study, or by reduction of expression,unstable expression, or mutation.

We have characterized breast cancer cell lines thathave varying degrees of invasive potential for uvomorulin

expression. We have observed three groups of cells with

the following phenotypes: (1) uvomorulin positive/vi-

!.#{149}

mentin negative, (2) uvomorulin negative/vimentin neg-ative, and (3) uvomorulin negative/vimentin positive. Al-though loss of uvomorulin-mediated cell-cell adhesionmay allow cells to detach from one another, an eventwhich is likely to be required for local invasion, somebreast cancer cell lines that did not express uvomorulinwere not highly invasive, as measured by the in vitroassay described here. Clearly, additional changes otherthan uvomorulin loss are involved in the progression toa more invasive phenotype. One important event in theacquisition of invasive properties seems to involve achange in morphology accompanied by the onset ofvimentin expression as represented by the uvomorulin-positive/vimentin-positive breast cancer cell lines, whichwere highly invasive. Other changes that are necessarilyinvolved in invasion through basement membrane areincreases in collagenase and other protease activitiesinvolved in breaking down basement membrane mole-cubes (19). This progression may be analogous to epithe-lial-mesenchymal transitions in embryonic development,in which keratin-expressing, uvomorulin-positive epithe-hal cells lose their surface polarity and give rise to mi-grating, uvomorulin-negative, vimentin-expressing mes-enchymal cells. These cells can later reorganize intoepithelia or differentiate into connective tissue deriva-tives (20). If primary carcinoma cells can undergo such atransition, their ability to metastasize could be greatlyenhanced.

A correlation between uvomorulin expression andmetastatic ability has been observed in the murine ovar-ian carcinoma cell line OV2944 (2 1 ). A subline of OV2944that was highly metastatic in nude mice had variable butlow levels of uvomorulin compared to two weakly met-astatic sublines, which consistently expressed high levelsof uvomorulin. These data also suggest that a loss orinstability of uvomorulin expression correlates with in-creased metastatic ability. A more direct test of the

1�%

I

�N

� ..

-� � � � � .

�. ,, �r � \�, .� .,4 � .‘�

Fig. 3. Cell morphology in Ma-

trigel. Cell lines were plated ona thin layer of the basementmembrane matrix Matrigel and

were incubated for 5-10 days as

described in “Materials andMethods.” a, ZR-75-B; b, BT474;c, CAMA1; d, SK-BR-3; e, MDA-

MB-436; 1, HS578T.

:.

‘-p -‘�‘�5

:‘� ,� -�

... ,p, �

. �

�‘ �

� �Y;�

. . , -.‘--

4�dr’,. -�4:/�’k.4#{231}c,#.?�

a/-,�, , , #{149},.. ,.,

$_,_.,� . r..�,.

‘4

p

q.

,�

.wf

-�..

“

‘. � #{149}.�

SF:. � � -

. . -s.’�.

,� � .‘

�� .

�1�

‘I

::� 4i

Cell Growth & Ditk’rentiation 369

involvement of uvomorulin in invasion was performed

by Behrens et a!. (6) in which nontumorigenic caninekidney epithelial MDCK cells were treated with antibod-

ies to the canine form of uvomorulin. The antibody-

treated cells became more fibroblast-like in appearance,

and approximately 1% of the cells invaded chick em-

bryonal heart tissue and collagen gels. In addition, MDCK

cells infected with Harvey or Moloney sarcoma viruses,

which have fibroblastic morphologies and are invasive,

no longer express uvomorulin (6). The fibroblast-like,

invasive MDCK cells may be considered analogous

to the invasive, uvomorulin negative/vimentin positive

breast cancer cell lines described here. However, al-

though the antibody-treated cells in our experiments

detached from one another, they did not exhibit a fibro-

blast-like morphology and did not increase invasion

a b

. .‘,

� .�

. �

, �

‘�“b

C

p #{149}

Ti

‘-. f

,

,

p

I

..i.

9,..

p

I

. ,fL

fig. 4. Antibody treatment of MCF.7 cells in Matrigel. Cells were plated in Matrigel as described for Fig. 3 and were subsequently subjected to thefollowing treatments: a, control; h, gp84; c, gp84 treatment for 1 day followed by a medium change and incubation for 4 additional days; d, normal rabbitserum; e, DECMA. 1 ; I, rat monoclonal antibody to CD44/hyaluronate receptor.

, �

370 Uvomorulin Expression in Breast Cancer Cell Lines

#{149}�., 1..’�,

‘�,, . , .11�..

I,,: i� .

through a Matrigel-coated filter in the modified Boyden

chamber assay. This may indicate a difference in assays

or a difference in the ability of human breast cancer cells

to go through an epithelial-mesenchymal transition com-

pared to the nontumorigenic canine kidney epithelial

cells. One of the many differences between MDCK cells

and human breast cancer cells such as MCF-7 is that

MDCK cells express vimentin. Future experiments aimed

at expressing uvomorulin in invasive human breast can-

cer cells may help elucidate a role for uvomorulin in

invasion and metastasis.

‘(4� �

Our experiments also suggest that immunohistochem-ical analysis of uvomorulin levels in tumor tissues maybe worthwhile to determine possible diagnostic or prog-

nostic value. A study of 44 human lung carcinomas and5 metastases did not show decreased levels of uvomo-rulin expression (22). Another analysis of over 60 humantumors (including 8 breast tumors) did not show anabsence of uvomorulin expression in any of the speci-

mens, although the authors stated that the pattern of

expression sometimes became nonpolarized, and, ingeneral, the levels of uvomorulin expression in tumorsappeared to be lower than in normal tissue, although this

Cell Growth & Differentiation 371

Tab!e 2 Invasion a nd chem otaxis of selected breast cancer cell lines

Cell line Uvoa Vimb Invasionc Chemotaxisd

MCF-7BT474

T47D

+++

-

-

-

426±1822648±1671

1143±730

1749±2251940±366

3552± 1745

CAMA1

SK-BR-3-

-

-

-

95 ± 24239±89

294 ± 59288 ±80

MDA-MB-231

HS578T

-

-

++

7417±548

6504 ± 3578529± 1080

6679 ± 854

a Expression of uvomorulin.b Expression of vimentin.C Invasive activity as described in “Materials and Methods,” expressed as

mean number of cells to traverse a Matrigel-coated filter/field ± SD; 1field = 5.94 mm2.

d Chemotactic activity as described in “Materials and Methods,” ex-pressed as mean number of cells to traverse a collagen IV-coated filter/field ± SD; 1 field = 5.94 mm2.

observation was not quantified (23). Alterations in uvo-morulin-mediated cell adhesion caused by point muta-tions or deletions of one copy of the gene would prob-ably not be observed by immunohistochemical analysis.Disruption of uvomorulin contacts with intracellularconnecting proteins [catenins (24, 25)] could also affectcell-cell adhesion but would not show altered levels ofuvomorulin protein. Although the invasive cell lines ex-amined in this study have completely lost expression ofthe uvomorulin gene, it is possible that, in vivo, otheralterations could disrupt cell-cell adhesion and facilitateinvasion. We suggest that additional studies, examiningtumor specimens immunohistochemically at invasivefronts and examining the uvomorulin gene at the molec-ular level, are warranted to study the role of uvomorulinin the progression of breast cancer and other epithelialmalignancies.

Materials and Methods

Cell Lines. The cell lines listed in Table 1 were obtainedfrom the American Type Culture Collection (Rockville,MD) with the following exceptions: MCF-7, T47D, ZR-75-1, HS578T, and MDA-MB-231 were obtained fromthe laboratory of Dr. Marc E. Lippman (Lombardi CancerCenter, Washington, DC). CAMA1 cells were obtainedfrom Dr. Benjamin Leung (University of MN). All cell lineswere passaged in IMEM containing 10% fetal bovineserum (Biofluids, Rockville, MD).

Immunofluorescence Microscopy. Cells were grownon glass coverslips to near confluence. After fixation inmethanol, the cells were incubated with 1:100 dilutionof gp84 rabbit antiserum directed against mouse uvo-morulin (1 3). The secondary antibody used was fluo-rescein-conjugated anti-rabbit lgG (H&L; HyCboneLaboratories, Inc., Logan, UT) at 2.5 zg/ml. Samples werevisualized by epi-illumination on a Zeiss photomiCro-scope Ill.

Northern Blot Analysis. RNA was isolated using theguanidinium isothiocyanate/cesium chloride centrifuga-tion method as described (26). Polyadenylated RNA wasselected by oligo(dT) column chromatography accordingto the manufacturer’s instructions (Boehringer Mann-heim, Indianapolis, IN). Twenty �g oftotal RNA or 10 �tgof polyadenylated RNA were electrophoresed through a1.1% 2.2 M formaldehyde-containing agarose gel. After

Tab!e 3 Effect of uvomorulin antibody treatment on invasion andchemotaxis of MCF-7 cells

Cell line and treat-ment

.Invasion

. I,Chemotaxis

MCF-7control 175±93 417± 131

McF-7 DECMA-1 88 ± 13 285 ± 70MCF-7anti-CD44 152±61 619± 149

MDA-MB-231 control 4913 ± 750 3497 ± 663

a Invasive activity as described in #Materials and Methods,” expressed as

mean number of cells to traverse a Matrigel-coated filter/field ± SD; 1field = 5.94 mm2.

b Chemotactic activity as described in “Materials and Methods,” ex-pressed as mean number of cells to traverse a collagen IV-coated filter/field ± SD; 1 field = 5.94 mm2.

transfer to nitrocellubose (Schleicher and Schuell, Keene,NH) by capillary action, the blots were hybridized torandom primer-labeled mouse uvomorulin probe F20a(27) or to a human vimentin cDNA probe described

previously (7). The filters were hybridized at 37#{176}Cin 3x

SSC/50% formamide and were washed at 60#{176}Cin O.5xSSC/O.i% sodium dodecyl sulfate (ix SSC = 0.15 M

NaCI-0.015 M sodium citrate).Matnigel Outgrowth Assay. One x iO� cells were

plated in one well of a 48-well cluster dish (Costar,Cambridge, MA) previously coated with a thin layer ofMatrigel (kindly provided by Hynda K. Kleinman, NIH).After a period of incubation of 5-10 days (unless other-wise noted), the cells were photographed using Hoffmanoptics on a Leitz Labovert photomicroscope. For anti-

body treatment studies, cells were plated in the presenceof 1:100 dilutions of antibody, or antibody was added tothe cells the next day. Cells were then photographed 24-48 h later, unless otherwise noted. , a ratmonocbonal antibody to uvomorulin, has been shown todisrupt adhesive functions in mouse preimplantation em-bryos and in MDCK cells (14). DECMA-i ascites was

obtained from Dr. I. Damjanov (Jefferson Medical Cob-lege, Philadelphia, PA). The rat monocbonal antibody toCD44/hyaluronate receptor (1 5, 16) was obtained fromDr. C. B. Underhill (Georgetown University, Washington,DC).

Matnigel bnvasion Assay. The Boyden chamber che-moinvasion assay was performed as described previously(28, 29). Matrigel (25 zg) was dried onto polycarbonate

filters (12-jzm pore, polyvinyl pyrrolidone-free; Nucleo-

pore, Pleasonton, CA) and then reconstituted at 37#{176}Ctogive an even layer over the filter surface. Fibroblast-conditioned medium, obtained by incubating confluentNIH-3T3 cells for 24 h with IMEM supplemented with L-

ascorbic acid (GIBCO, Grand Island, NY; 50 �g/ml), wasused as the chemoattractant itt the bottom chamber.Cells were harvested by trypsinization, washed in IMEMcontaining 5% fetal bovine serum, and added to the topchamber (300,000 cells/chamber). Antibody (DECMA-ior rat monocbonal antibody to CD44/ hyaluronate recep-

tor) was included in the upper chamber at 1 :1 00 dilutionwhere appropriate. Chambers were incubated for 6 h.The cells which had traversed the Matrigel and spreadon the lower surface of the filter were stained with Diff-Quik (American Scientific Products, McGaw Park, IL) andquantitated electronically using the Zeiss IBAS 2000 im-age analysis system with Kontron Aiag processor inter-

372 Uvomorulin Expression in Breast Cancer Cell Lines

faced with a Zeiss Axiophot microscope equipped withan automated stage.

Chemotaxis Assay. Chemotaxis assays were performedas described for the chemoinvasion studies with the

single exception that the filter surfaces were coated with5 �Lg of collagen IV instead of Matrigel. Chemotaxis assayswere performed in parallel to the chemoinvasion assaysusing the same cells and conditioned media and werequantitated similarly.

AcknowledgmentsThe authors would like to thank Dr. Ramon Colomer and Dr. Marc E.Lippman for helpful discussions and lulia Nguyen for technical assistance.

References1 , Nicolson, G. L., and Hug, V. Breast cancer growth and metastasis. M.

D. Anderson Oncol. Case Rep. Rev., 5; 1 - 1 1 , 1990.

2. Albelda, S. M., and Buck, C. A. Integrins and other cell adhesionmolecules. FASEB J,, 4: 2868-2880, 1990.

3. Takeichi, M. The cadherins: cell-cell adhesion molecules controllinganimal morphogenesis. Development (Camb.), 102: 639-655, 1988.

4. Edelman, G. M. Cell adhesion molecules. Science (Washington DC),219: 450-457, 1983.

5. Kemler, R., Ozawa, M., and Ringwald, M. Calcium-dependent celladhesion molecules. Current Opinion Cell Biol., 1: 892-897, 1989.

6. Behrens, J., Mareel, M. M., van Roy, F. M., and Birchmeier, W.Dissecting tumor cell invasion: epithelial cells acquire invasive propertiesafter the loss of uvomorulin-mediated cell-cell adhesion. J. Cell Biol., 108:2435-2447, 1989.

7. Sommers, C. L., Walker-Jones, D., Heckford, S. E., Worland, P.,Valverius, E., Clark, R., McCormick, F., and Gelmann, E. P. Vimentinrather than keratin expression in some hormone-independent breastcancer cell lines and in oncogene-transformed mammary epithelial cells.Cancer Res., 49: 4258-4263, 1989.

8. Raymond, W. A., and Leong, A. S-Y. Vimentin-a new prognosticparameter in breast carcinoma? J. Pathol., 158: 107-1 14, 1989.

9. Domagala, W., Lasota, J., Bartkowiak, J., Weber, K., and Osborn, M.

Vimentin is preferentially expressed in human breast carcinomas withlow estrogen receptor and high Ki-67 growth fraction. Am. J. Pathol., 136:219-227, 1990.

10. Cattoretti, G., Andreola, S., Clemente, C,, D’Amato, L., and Rilke, F.vimentin and p53 expression in epidermal growth factor receptor-posi-tive, oestrogen receptor-negative breast carcinomas. Br. I. Cancer, 57:353-357, 1988.

11. Damsky, C. H., Richa, J., Solter, D., Knudsen, K., and Buck, C. A.

Identification and purification of a cell surface glycoprotein mediatingintercellular adhesion in embryonic and adult tissues. Cell, 34: 455-466,

1983.

12. Batist, G., Tulpule, A., Sinha, B. K., Katki, A. G., Myers, C. E., andCowan, K. H. Overexpression of a novel anionic glutathione transferasein multidrug-resistant human breast cancer cells. J. Biol. Chem., 261:

15544-15549, 1986.

13. Vestweber, D., and Kemler, R. Rabbit antiserum against a purifiedsurface glycoprotein decompacts mouse preimplantation embryos andreacts with specific adult tissues. Exp. Cell Res., 152: 169-178, 1984.

14. Vestweber, D., and Kemler, R. Identification of a putative cell adhe-sion domain of uvomorulin. EMBO J., 4: 3393-3398, 1985.

15. Miyake, K., Medina, K. L., Hayashi, S. I., Ono, S., Hamaoka, T., andKincade, P. W. Monoclonal antibodies to Pgp-1/CD44 block lympho-hemopoiesis in long term bone marrow cultures. J. Exp. Med., 171: 477-488, 1990.

16. Culty, M., Miyake, K., Kincade, P. W., Sikorski, E., Butcher, E. C.,and Underhill, C. The hyaluronate receptor is a member ofthe CD44 (H-CAM) family of cell surface glycoproteins. J. Cell Biol., 1 1 1: 2765-2774,1990.

17. Fearon, E. R., Cho, K. R., Nigro, J. M., Kern, S. E., Simons, J. W.,Ruppert, J. M., Hamilton, S. R., and Vogelstein, B. Identification of achromosome 18q gene that is altered in colorectal cancers. Science(Washington DC), 24: 49-56, 1990.

18. lnghirami, G., Grignani, F., Sternas, L., Lombardi, L., Knowles, D. M.,and DaIla-Favera, R. Down-regulation of LFA-1 adhesion receptors by c-myc oncogene in human B lymphoblastoid cells. Science (WashingtonDC), 250: 682-686, 1990.

19. Liotta, L. A., and Kohn, E. Cancer invasion and metastasis. JAMA,263: 1123-1126, 1990.

20. Thiery, I. P., Boyer, B., Tucker, G., Gavrilovic, J., and Valles, A. M.Adhesion mechanisms in embryogenesis and in cancer invasion andmetastasis. Ciba Found. Sympos., 141: 48-74, 1988.

21. Hashimoto, M., Niwa, 0., Nitta, Y., Takeichi, M., and Yokoro, K.Unstable expression of E-cadherin adhesion molecules in metastaticovarian tumor cells. Jpn. J. Cancer Res., 80: 459-463, 1989.

22. Shimoyama, V., Hirohashi, S., Hirano, S., Noguchi, M., Shimosato,Y,, Takeichi, M., and Abe, 0. Cadherin cell-adhesion molecules in humanepithelial tissues and carcinomas. Cancer Res., 49: 2128-2133, 1989.

23. Eidelman, S., Damsky, C. H., Wheelock, M. J., and Damjanov, I.Expression of the cell-cell adhesion glycoprotein cell-CAM 120/80 innormal human tissuesand tUmors. Am. J. Pathol., 135: 101-110, 1989.

24. Ozawa, M., Baribault, H., and Kemler, R. The cytoplasmic domain

of the cell adhesion molecule uvomorulin associates with three inde-pendent proteins structurally related in different species. EMBO J., 8:1711-1717, 1989.

25. Nagafuchi, A., and Takeichi, M. Transmembrane control of cadherin-mediated cell adhesion: a 94 kDa protein functionally associated with aspecific region of the cytoplasmic domain of E-cadherin. Cell Regul., 1:137-44, 1989.

26. David, L. G., Dibner, M. G., and Battey, J. F. Basic Methods inMolecular Biology. pp. 130-135. New York: Elsevier Science PublishingCo., Inc., 1986.

27. Schuh, R., Vestweber, D., Riede, I., Ringwald, M., Rosenberg, U. B.,

JackIe, H., and Kemler, R. Molecular cloning of the mouse cell adhesionmolecule uvomorulin: cDNA contains a B1 -related sequence. Proc. NatI.Acad. Sci. USA, 83: 1364-1368, 1986.

28. Albini, A., Iwamoto, V., Kleinman, H. K., Martin, G. R., Aaronson, S.A., Kozlowski, J. M., and McEwan, R. N. A rapid in vitro assay forquantitating the invasive potential of tumor cells. Cancer Res., 47: 3239-3245, 1987.

29. Thompson, E. W., Shima, T. B., Reich, R., Graf, J., Albini, A., Martin,G. R., Dickson, R. B., and Lippman, M. E. Differential regulation of growthand invasiveness of MCF-7 breast cancer cells by antiestrogens. CancerRes., 48: 6764-6768, 1988.