Human colonic cancer cells synthesize and adhere to ... · the spatial/developmental distribution...

1993Journal of Cell Science 111, 1993-2004 (1998)Printed in Great Britain © The Company of Biologists Limited 1998JCS9736

Human colonic cancer cells synthesize and adhere to laminin-5. Their

adhesion to laminin-5 involves multiple receptors among which is integrin

α2β1

Véronique Orian-Rousseau 1,*, Daniel Aberdam 2, Patricia Rousselle 3, Anthea Messent 4, Jelena Gavrilovic 4,Guerrino Meneguzzi 2, Michèle Kedinger 1 and Patricia Simon-Assmann 1,‡

1INSERM U.381, 3 Avenue Molière, 67200 Strasbourg, France2INSERM U.385, Faculté de Médecine, Avenue de Valombrose, 06107 Nice Cédex 2, France3Institut de Biologie et Chimie des Protéines, CNRS, 69367 Lyon Cédex 07, France4School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK*Present address: Forschungszentrum Karlsruhe, Institute of Genetics, 76021 Karlsruhe, Germany‡Author for correspondence (e-mail: [email protected])

Accepted 6 May; published on WWW 30 June 1998

In the mature gut, laminin-5 is expressed at the basal aspectof the differentiating epithelial cells. In vitro, we show thatthree more or less differentiated human colonic cancerHT29 cell lines produce and deposit laminin-5; theypredominantly synthesize and secrete the 440 kDa form oflaminin-5 that comprises the unprocessed 155 kDa γ2chain, as determined by immunoprecipitation analysis. Incontrast, the highly differentiated colon carcinoma Caco-2cells produce almost no laminin-5.

Using anti-integrin antibodies, we show that adhesion ofthe two colonic cancer cell lines to laminin-5 is mediated bymultiple integrin receptors including those for α3β1, α6β1

and α6β4 integrins like in other cell types. In addition, theimplication of integrin α2β1 in this adhesion process isdemonstrated for the first time. This has been shown by celladhesion inhibition experiments, solid phase assays andconfocal analysis. Together with previous in situobservations, these data provide a baseline knowledge forthe understanding of the regulation of laminin-5 in normaland pathological intestine.

Key words: Intestine, Epithelial cell, Basement membrane,Differentiation, Integrin

SUMMARY

a

g-ntdun

is,n

INTRODUCTION

Continual renewal of intestinal epithelium is ensured by tproliferation of stem cell populations confined to the cryregion. As cell migration proceeds from crypt base to villus tthe undifferentiated cells acquire morphological anbiochemical features of polarized cells. Among these cells, absorptive enterocytes carry the digestive enzymes locatethe apical brush border membrane. Cell adhesion to underlying basement membrane is thought to be critical in regulation of these migration/differentiation events. Thhypothesis is strengthened by the observation that changethe spatial/developmental distribution of basement membramolecules are concomitant to the morphogenetic processescompartmentalization in the adult organ (for reviews sSimon-Assmann et al., 1995, 1998). Among the basemmembrane components, laminins display a wide array biological activities such as cell adhesion, migration adifferentiation. It has been shown that a subset of epithebasement membranes contains the laminin-5 variant (Verraet al., 1987; Rousselle et al., 1991). This adhesion ligand been isolated from keratinocytes and squamous carcino

heptip,dthed atthetheiss inne

andeeentof

ndlialndohasma

cells. In these cells, laminin-5 (α3β3γ2) is initially synthesizedas a cellular heterotrimeric precursor comprising the 200 kD(α3 precursor), 155 kDa (γ2 precursor) and 140 kDa (β3)polypeptides (Marinkovich et al., 1992a; Rousselle andAumailley, 1994). The laminin α3 chain is immunologicallyrelated to a distinct 190 kDa laminin α chain interacting withthe β1 and γ1 chains to form the laminin-6 variant(Marinkovich et al., 1992b) and with β2 and γ1 chains to formthe laminin-7 variant (Champliaud et al., 1996). Laminin-5 andlaminin-1 (α1β1γ1) are expressed in the intestine and displaya differential localization according to the crypt-to-villuspositioning (for review see Simon-Assmann et al., 1998). Inparticular, laminin-5 presents a gradient of intensity increasinfrom the base to the tip of the villus (Leivo et al., 1996; OrianRousseau et al., 1996). This expression pattern is consistewith that of HD1, a protein of the hemidesmosmal plaque, anintegrin α6β4 (Simon-Assmann et al., 1994a; Orian-Rousseaet al., 1996) and suggests a possible role of laminin-5 idifferentiation and/or migration of intestinal cells.

To date, models of normal intestinal cell culture displayingthe same properties as the enterocytes of the crypt-villus axare not available for further in vitro studies. This is the reaso

1994

edg

lar

rts

.

talted

nede,alf

udyionellsd

/ml

nndn

orededsse

dA,iss)ol

innd

Mred.

tha)

al-A-lets

ith-

-ndne

V. Orian-Rousseau and others

why we used the Caco2 and HT29 cell lines that derive frohuman colonic tumors and represent interesting model systto study various aspects of intestinal cell biology (Chantretal., 1988; and for reviews see Kedinger et al., 1987; Neutra Louvard, 1989; Zweibaum et al., 1991). Caco2 cells forapical brush borders and express digestive hydrolases as as they become confluent. Their differentiation state canmodulated by exogenous laminin substrates (Vachon aBeaulieu, 1995; Basson et al., 1996). Conversely, conflucultures of HT29 cells remain mainly undifferentiated (Pinet al., 1983; Zweibaum et al., 1985). Treatment of the HTcells with anti-cancer drugs, 5-fluorouracil or methotrexagenerated HT29-Fu and HT29-MTX clones which acommitted to differentiate into various phenotypes (Lesuffleet al., 1990, 1991). Interestingly Caco2 and HT29 cell lindisplay a distinct repertoire of integrins, the cell surfareceptors that mediate interactions between cells aextracellular matrix (Basson et al., 1992; Simon-Assmannal., 1994b; Ebert, 1996).

The importance of laminin-5 in the skin has beeemphasized by the disastrous consequences of mutations constituent chains that give rise to the blistering skin diseHerlitz junctional epidermolysis bullosa (JEB; Aberdam et a1994b; Fine, 1994; Uitto et al., 1995). Although pyloric atresand various gut alterations are associated with subtypes of (Vidal et al., 1995; Brown et al., 1996; Niessen et al., 199Pulkkinen et al., 1997), modifications of laminin-5 have nevbeen analyzed in the gastrointestinal tract of these patients.analysis of the subepithelial basement membrane in intestbiopsies taken from children displaying a subtype of intractadiarrhoea with villous atrophy, revealed an abnormal lamindeposition (Goulet et al., 1995). These observations pointthe need to analyze the characteristics of laminin isoformsintestinal cells as a baseline to understand their function.

The aim of the present work was to take advantage of distinct intrinsic properties of Caco2, HT29 and clonal HT2subpopulations to: (i) define the molecular characteristics athe expression pattern of laminin-5 in intestinal epithelial cel(ii) determine the adhesion properties of intestinal epithelcells to exogeneous laminin-5, (iii) define the integrin receptoinvolved in this process. In this study, we show that the highdifferentiated Caco2 cells produce almost no laminin-5 contrast to the HT29 cells whatever their state or type differentiation. The molecular form of the intestinal laminin-differs from that of keratinocytes or squamous epithelial ceFurthermore, we demonstrate that colonic cancer cells specifically the integrin α2β1 to bind to laminin-5 in additionto the receptors described in other cell types.

MATERIALS AND METHODS

AntibodiesMouse monoclonal (mAb) and rabbit polyclonal (pAb) antibodieraised against human laminin-5 were: mAb GB3 that recogniznative laminin-5 (Verrando et al., 1987); mAb BM165 to the lamin200/165 kDa α3 chain (Rousselle et al., 1991) and 190 kDa chain laminin-6 (Marinkovich et al., 1992b); mAb K140 to laminin 140 kDβ3 chain of laminin-5 (Marinkovich et al., 1992a); pAb SE144 anSE85 directed against the laminin 155/105 kDa γ2 chain and laminin165 kDa α3 chain, respectively (Aberdam et al., 1994b). Fo

mems etandmsoon bend

entto29te,reurescend et

nin itsasel.,iaJEB6;er Theinalblein to in

the9ndls,ialrslyinof5lls.use

ses

inofad

r

sequential immunoprecipitation experiments, a rabbit pAb raisagainst mouse EHS (Engelbreth-Holm-Swarm) laminin recognizinthe three constituent chains, α1, β1 and γ1 (Simo et al., 1991) wasused.

In the cell adhesion assay, mAb 3E1 recognizing the extracelluportion of human integrin β4, mAb P4C10 to the β1 subunit and mAbP1B5 raised against the α3 subunit were purchased from Gibco(France). mAb anti-human integrin α2, P1E6 was from Telios(Pharmaceuticals, Inc.) and 6F1 was a kind gift from Dr R. Banke(Roswell Park Cancer Institute, Buffalo, NY, USA). mAb GoH3specific to the integrin α6 was from Immunotech (Marseille, France)

Colon carcinoma cell linesThe human colon adenocarcinoma cell line HT29 (HT29p, parencell line) was established by Fogh et al. (1977). HT29 cells adapto 10−5 M 5-fluorouracil (HT29-Fu) and to 10−5 M methotrexate(HT29-MTX) were obtained from Dr A. Zweibaum (INSERM Unit178, Paris, France). The characteristics of the cells have been defiin detail previously (Lesuffleur et al., 1990, 1991). Cells werroutinely grown in Dulbecco’s modified Eagle’s medium (DMEMGibco, France) supplemented with 10% heat-inactivated fetal cserum (Gibco) and 200 µg/ml gentamycin.

The human colon adenocarcinoma Caco2 cell line used in this sthas also been established by Fogh et al. (1977). Its differentiatproperties have been defined previously (Pinto et al., 1983). The cwere maintained in DMEM supplemented with 20% heat-inactivatefetal calf serum, 1% nonessential amino acids (Gibco) and 200 µggentamycin.

Immunocytochemical detection of laminin-5Detection of intracellular laminin-5 was performed on cells grown oglass coverslips, prefixed 10 minutes with 1% paraformaldehyde apermeabilized 10 minutes with 0.5% Triton X-100 before incubatiowith mAb GB3 diluted (1/50) in phosphate buffered saline (PBS) f1 hour at room temperature. After washings in PBS, cells were stainwith fluorescein-isothiocyanate-labelled sheep anti-mouse IgG dilutin PBS. Immunodetection of laminin-5 deposited on the glacoverslip was performed following removal of the cells after a onhour treatment with a solution containing 1% Triton X-100, 10 mMEDTA and 25 mM Tris-HCl, pH 7.5. The preparations were mounteunder coverslips using the Vectashield mounting medium (Biosys SCompiègne, France), observed using an Axiophot microscope (Zeand photographed using HP5 film (Asa 400, Ilford Ltd). On contrsections, the primary antibody was omitted.

Biosynthetic labelling studies and immunoprecipitationexperiments25×106 cells were seeded on a 100 mm Petri culture dish to obtaconfluent monolayers after 6 hours. Cells were then washed aradiolabelled with 100 µCi/ml trans 35S-labelTM metabolic labellingreagent containing [35S]L-methionine and [35S]L-cysteine (ICNBiomedicals, Orsay, France) in methionine-cysteine-free DMEmedium containing 2% fetal calf serum. 24 hours later, the cultumedium was collected and 2mM NEM, 1 mM PMSF were addeAfter 10 minutes centrifugation at 330 g, the medium was either storedat −80°C or used immediately.

For immunoprecipitation, samples were first preincubated wiProtein-G-Sepharose or Protein-A-Sepharose CL-4B beads (Sigmfor 1 hour at 4°C. In parallel, the monoclonal and the polyclonantibodies were incubated with Protein-G-Sepharose and ProteinSepharose beads for 1 hour at 37°C, respectively. The pelcontaining the antibodies bound to the beads were washed twice wRIPA buffer (radioimmunoprecipitation assay buffer: 10 mM TrisHCl, pH 7.4, 2 mM EDTA, 2 mM cysteine, 2 mM methionine, 250µM PMSF, 1 mM NEM, 0.5% NP-40, 0.1% SDS, 0.05% Triton X100, 0.3% sodium desoxycholate, 0.1% BSA, 150 mM NaCl) aincubated with the precleared samples 18 hours at 4°C. Immu

1995Laminin-5 and integrins in colonic cancer cells

the

es

taeee

nesved

anly

s Inr)

andl

e.

Mre

te

assted

heS-escats,heareted

hedtrstdf

5dsss

g.jor

complexes were washed 5 times with RIPA buffer and resuspendeLaemmli buffer containing β-mercaptoethanol. Pellets were boileand subjected to 6% SDS-PAGE gels. The gels were then driedexposed to Fuji Medical X-ray film (Fuji) for at least 24 hours −80°C. Gels were calibrated with 14C-labelled protein standards(Amersham) and apparent molecular masses were determinedlinear scanning with a densitometer (Shimadzu, Roucaire). For sequantitative analysis, radioactivity was measured in aliquots of esample using Ultima gold liquid scintillation cocktail (PackarInstrument Company, INC., USA) and a Tri-CARB 1 600 TR liquiscintillation analyzer (Packard, USA). When required, thradiolabelled material was pre-incubated 4 hours twice with anlaminin-1 antibodies before treatment with anti-laminin-5 antibodieFor pulse-chase experiments, attached cells were incubated 1 ho100 µCi/ml of [35S]methionine and [35S]cysteine in deficient DMEMmedium and then 0, 3, 6, 24, 48 hours in complete DMEM medifor chase. Cells were scraped into a small volume of RIPA buffground and centrifuged at 330 g for 10 minutes. Cell extracts andconditioned medium were then processed for immunoprecipitatio

Immunoblot analysisWestern analysis of cell extracts and conditioned medium wperformed at 7 days on confluent cell cultures. Cell extraction wachieved by incubating the cells 1 hour at 37°C with 0.1 M phosphbuffer, pH 7.2, containing 2 N NaBr, 2 mM PMSF, and 10 µg/mleupeptin, grinding them in 500 µl of 50 mM Tris buffer, pH 7.2containing 1% SDS, 5% β-mercaptoethanol, 10 mM EDTA, 2 mMPMSF and 10 µg/ml leupeptin. The samples were centrifugedminutes at 100,000 g and the pellet discarded. 10 µg of cleared proteextract and conditioned medium were incubated 5 minutes Laemmli buffer containing 5% β-mercaptoethanol and 2% (w/v) SDSat 100°C. Proteins were separated on 6% SDS-PAGE gels subsequently electrotransferred overnight onto nitrocellulose in mM Tris-HCl, 0.192 M glycine, pH 8.2, 20% (v/v) methanol. Aftetransfer, the nitrocellulose was saturated 1 hour at 37°C with 3% Bin buffer (10 mM Tris, pH 7.4, 0.15 M NaCl containing 1 mM EDTA0.1% (v/v) Tween-20 and 0.5% gelatin (v/v), and then incubatedhour at room temperature with specific antibodies. After reaction wthe secondary antibody, nitrocellulose sheets were treated wenhanced chemoluminescence reagent (Amersham). Prestamolecular mass markers were included in each gel.

Cell adhesion assaysMulti-well tissue culture plates (96 wells, Nunc, Naperville, USAwere coated with serial dilutions (0-40 µg/ml, 100 µl/well) of laminin5, laminin-1/nidogen complex or fibronectin by overnight adsorptiat 4°C. Laminin-5 was purified from squamous cell carcinoma cultumedium (Rousselle and Aumailley, 1994), laminin-1/nidogecomplex from the EHS tumour of the mouse (Simo et al., 1991) ahuman plasma fibronectin was kindly provided by the Centre Transfusion Sanguine (Strasbourg, France). The wells were wasonce with Ham F10 and saturated 1 hour at room temperature 1% bovine serum albumin (BSA, Vth fraction, Sigma Chimie, SainQuentin-Fallavier, France). 1-2×105 cells suspended in 0.1 ml mediumwere then incubated 2 hours and after removal of the supernatcontaining the non-adherent cells, washed twice in PBS and fixe4°C in 70% ethanol for 15 minutes. The plates were air dried forleast 30 minutes at 37°C and cells were stained 25 minutes at rtemperature with 0.1% Crystal Violet (100 µl/well) in distilled wateThe wells were rinsed briefly with distilled water and air dried. Thstain was solubilized using 10% acetic acid in distilled water (1µl/well) at room temperature. Colour yields were then measured inElisa reader (Dynatech) at 550 nm. A blank value correspondingBSA-coated wells (≤5% of maximal cell adhesion) was automaticallsubtracted. Adherent cells were photographed using a Lephotomicroscope in phase contrast mode. Each assay point performed in duplicate. The number of independent assays perfor

d ind andat

bymi-

achddeti-s.ur in

umer,

n.

asasatel,

30inin

and25rSA, 1ithith

ined

)-

onrennddehed

witht-

antsd at atoomr.e

00 an toyitzwas

med

for each antibody ranged from 2 to 4. Results were expressed asmean plus or minus one standard error. The paired Student’s t-test wasused to analyse the data for statistical significance of differencbetween means. Differences with a P value of less than 0.05 wereconsidered significant.

For cell adhesion inhibition experiments using anti-substraantibodies, wells coated with laminin-1/nidogen or laminin-5 werincubated with serial dilutions in PBS of mAb BM 165 against thtrimeric laminin-5 molecule for one hour at 37°C as indicated in thcorresponding figures. In inhibition assays with anti-integriantibodies, cell suspensions were mixed with dilutions of antibodibefore plating onto the coated wells. Each assay point was derifrom triplicate wells.

In vitro binding assay on immobilized laminin-5α2β1 integrin was purified from detergent extracts of outdated humplatelets by affinity purification on collagen-Sepharose as previousdescribed (Kern et al., 1993; Messent et al., 1998). α2β1 integrinbinding assays on immobilized laminins were performed apreviously described (Tuckwell et al., 1996; Messent et al., 1998).brief, 96-well plates (Maxisorp, Nunc) were coated with laminin-5 olaminin-1 at indicated concentrations in Tris-buffered-saline (TBSfor 1 hour at room temperature. Unbound substrate was removed wells washed 3 times with TBS and then blocked with 50 mg/mbovine serum albumin (BSA) in TBS for 1 hour at room temperaturBSA was removed and biotinylated α2β1 integrin added at 2 µg/mlin TBS containing 2 mM MgCl2 and 1 mM MnCl2 for 2 hours at roomtemperature. Wells were washed 3 times with TBS, 1 mM MgCl2 andstreptavidin-peroxidase (Amersham) added at 1:1,500 in TBS, 1 mMgCl2, 1 mg/ml BSA for 15 minutes at room temperature. Wells wewashed 3 times with TBS, 1 mM MgCl2, and bound integrinvisualised with 3,3,5,5-tetramethylbenzidine-peroxidase substramixed 1:1 with H2O2. The reaction was stopped with 2.5 M H2SO4and the OD was read at 450 nm.

Confocal microscopy analysis of α2 integrin subunitCells seeded to near-confluency and grown for 2 hours on glcoverslips, were washed once in PBS. The cells were then incuba1 hour at room temperature with the anti-α2 integrin antibody (mAbP1E6) and then with the anti-mouse secondary antibody. Tcoverslips were mounted in 0.1% paraphenylendiamine in PBglycerol and sealed with varnish. Confocal laser scanning imagcorresponding to X-Y plane serial sections were taken with a Leiconfocal microscope using a TCS4M scanner (Leica InstrumenGmbH, Nussloch, Germany) equipped with a crypton laser. Tsystem is connected to a SUN Ultrasparc Unix station using softwthat allows sections to be made in the Y-Z plane from reconstrucimages.

RESULTS

Expression of laminin-5 in colon carcinoma celllinesBased on the observation that laminin-5 is found in situ in tbasement membrane lining the differentiating/differentiatevillus enterocytes but not underlying the proliferative crypcells (Leivo et al., 1996; Orian-Rousseau et al., 1996), we ficompared laminin-5 synthesis in undifferentiated HT29p andifferentiated Caco2 cells. Immunoprecipitation analysis oconditioned medium from both cell lines using anti-laminin-mAb GB3 revealed important differences. Three major banof 165, 155, 140 kDa and faint bands with an estimated maof 100 kDa were detected in the HT29p culture medium (Fi1, lane 1). In Caco2 conditioned medium, none of the ma

1996

sureDahee

2).alnite

at

25t

aledllsDae).ednte

en

reed

ose65


Fig. 2. Immunolocalization of laminin-5 in colonic cancer cells. GB3monoclonal antibodies were used to detect the trimeric laminin-5molecule in permeabilized HT29-MTX (A) and Caco2 (B) cellscultured for 5 days. A deposition of laminin-5 in basal patches wasobserved on the ventral side of HT29-MTX cells whereas no basallabelling was detected in the case of Caco2 cells. The organizeddeposit of laminin-5 in parallel lines was more obvious when HT29-MTX cells were detached from the coverslip by a 10 mM EDTAtreatment (C). Bars, 25 µm.

Fig. 1. Immunoprecipitation of laminin-5 from HT29p (lane 1),HT29-Fu (lane 2), HT29-MTX (lane 3) and Caco2 (lane 4) cells withGB3 monoclonal antibodies. Following a 24-hour metaboliclabelling period, conditioned media were immunoprecipitated withmAb GB3, separated by reduced SDS-PAGE on a 6% acrylamidegel. The radioactive bands were revealed by fluorography. In HT29cells (lanes 1-3), three major bands of 165, 155 and 140 kDa co-precipitated; a doublet at approximately 200 kDa was also detectedin HT29-MTX cells. The 250 kDa band found in Caco2 (arrowhead),as well as minor bands under 100 kDa detected in all cell linescorrespond to unspecific material since they are found in controlimmunoprecipitation assays. Left margin indicates 14C molecularmass markers. Apparent molecular masses (in kDa) are indicated forthe main immunoprecipitated bands at the right margin.

bands was observed (Fig. 1, lane 4), yet a prolonged expoof the gel revealed that all subunits were present, the 165 kband being fainter than the other chains (not illustrated). Timmunocytochemical detection of laminin-5 confirmed thdifferential expression between HT29 and Caco2 cells (Fig.HT29 cells displayed a regular labelling of laminin-5 in baspatches; some of them were organised in several defiparallel lines (Fig. 2A) particularly obvious when cells werdetached from the substratum by EDTA treatment (Fig. 2COnly tiny spots dispersed within the cytoplasm were presenthe differentiated Caco2 cell line (Fig. 2B).

To determine whether this differential expression of lamini5 is related to the different polarisation/differentiation potentof these cell lines, we analysed the two differentiated HT2derived cell lines, HT29-FU and HT29-MTX. At confluencethe MTX-adapted cells are almost exclusively of the mucusecreting type with apical brush borders that strongly exprdifferentiation markers such as dipeptidylpeptidase IV avillin; the HT29-Fu cells form a mixed population of polarisefluid-transporting cells and of goblet cells (Lesuffleur et a1990, 1991). Immunoprecipitation with mAb GB3 of mediumconditioned by the two differentiated HT29-derived cell lineidentified the same bands as those seen in the parental cel(Fig. 1, lanes 2 and 3, versus lane 1). The intensity of the bawas significantly stronger in the case of HT29-MTX cells anan additional doublet of approximately 200 kDa was alobserved (Fig. 1, lane 3).

Immunoprecipitation experiments of conditioned mediufrom HT29-MTX cells using polyclonal antibodies SE85SE144 and mAb K140 directed, respectively, to the α3, γ2 and

e).

t in

n-ial9-,s-

essnddl.,

sl linendsd

so

m,

β3 chains (Fig. 3, lanes 1-3) revealed patterns identical to thfound with mAb GB3. Interestingly, immunoprecipitation ofconditioned medium from tongue squamous carcinoma SCCcells with the SE144 antibody identified a 105 kDa produc(Fig. 3, lane 4) which was not detected in intestinal epithelicells (Fig. 3, lane 2). Pulse-chase experiments (Fig. 4) showthat in addition to the major form secreted by the intestinal ce(165, 155 and 140 kDa bands) (Fig. 4 lanes 6-9), the 105 kpolypeptide was found in very small amounts in the culturmedium only after 48 hours of chase (Fig. 4, lane 10Furthermore, the comparative intensity of the bands showthat the intestinal cells produce a significantly smaller amouof laminin-5 than the SCC25 cells (Fig. 3, lane 4 versus lan2).

The identity of the polypeptides immunoprecipitated by thvarious anti-laminin-5 antibodies was confirmed by westerblot (Fig. 5). The antibody directed to the α3 chain (pAb SE85)specifically recognised a 165 kDa polypeptide in the cultumedium (Fig. 5A, lane 1) and HT29 cell extracts (Fig. 5B, lan5); in addition, a doublet at approximately 200 kDa was founin cell extracts. These molecular masses correspond to thdescribed for the precursor (200 kDa) and the mature (1


ies

at

e

nt

3)

he

at-

Fig. 3. Immunoprecipitation of constituent chains of laminin-5 inHT29-MTX cells. Following a 24-hour metabolic labelling period,conditioned media from HT29-MTX cells were immunoprecipitatedwith polyclonal antibodies SE85 to the α3 subunit (lane 1), SE144 tothe γ2 subunit (lane 2), and with mAb K140 to the β3 subunit (lane3). For comparison, radiolabelled culture medium from squamouscell carcinoma (SCC25) cells was immunoprecipitated with theSE144 antibodies (lane 4). The immunoprecipitated materials wereanalyzed by SDS-PAGE on a 6% acrylamide gel under reducingconditions. The radioactive bands were revealed by fluorographyand corresponded to proteins migrating at 165, 155 and 140 kDa inHT29-MTX cells (lanes 1-3) and to more strongly labelled proteinat 165-155, 140 and 105 kDa in SCC 25 cells (lane 4) as indicatethe right. In lane 4, the 230 kDa band is not yet identified.

Fig. 4.Pulse-chase experiments of the cell-associated and mediumforms of laminin-5. HT29-Fu cells were labelled in mediumcontaining [35S]methionine and [35S]cysteine for 60 minutes.Radioactive medium was removed and the cells were then cultured innon-radioactive DMEM medium. Cells and culture medium wereprocessed for immunoprecipation with the polyclonal SE144antibodies after 0, 3, 6, 24 and 48 hours of chase. Precipitatedproducts were separated by reduced SDS-PAGE on a 6% acrylamidegel visualized by fluorography. Apparent molecular masses (in kDa)of bands are indicated.

kDa) forms of laminin α3 in keratinocytes (Marinkovich et al.1992a). No bands were identified on immunoblocorresponding to the Caco2 cells (Fig. 5A, lane 2, and B, l6). In HT29 conditioned medium (Fig. 5A, lane 3) and ceextracts (Fig. 5B, lane 7), pAb SE144 reacted with a major 1kDa band and to a much lesser extent with a 105 kpolypeptide which correspond, respectively, to the precurand mature laminin γ2 chains described in keratinocyte(Marinkovich et al., 1992a) and in SCC25 cells (Fig. 3, la4). This antibody also reacted with 155 kDa and 105 kDa bain culture medium and total extracts of Caco2 cells (Fig. 5lane 4, and B, lane 8). Thus, the immunoblot aimmunoprecipitation experiments revealed that the 105 kDaγ2product was barely detectable whatever the intestinal cell considered and that no α3 chains were found in Caco2 ceextracts or medium. The identity of the 140 kDa ban

Fig. 5.Identification of medium and cell-associated forms of laminin-5 by western blotting.Proteins from conditioned medium (A) and fromcell extracts (B) of HT29-MTX (lanes 1,3), Caco2(lanes 2, 4, 6, 8), HT29-Fu (lanes 5, 7) wereseparated by reduced SDS-PAGE on 6%acrylamide gels, transferred to nitrocellulose,incubated in the polyclonal SE85 (lanes 1, 2, 5, 6)and SE144 (lanes 3, 4, 7, 8) antibodies and thenvisualized with the enhanced chemoluminescencereagent. The triplet visible on lanes 3 and 4around 250 kDa, not found in theimmunoprecipitation experiments, corresponds tounidentified bands. Apparent molecular masses(in kDa) of bands are indicated.

,ts

anell55DasorsnendsA,nd

linelld,

immunoprecipitated by the anti-α3 or -γ2 chain antibodiescould not be assessed by this technique because antibodspecific to the denaturated laminin β3 chain are not available.

Immunoprecipitation with anti-α3 chain antibody furtherdetected in HT29 cell culture medium a 220 kDa band thcould correspond to the β1 and γ1 chains of laminin-1 (Fig. 6,lane 1). Serial immunoprecipitations were performed with thanti-laminin-1 antibody (lane 2) to deprive the mediumconditioned by the cells, from laminin-1 chains. Subsequeimmunoprecipitation with mAb BM 165 revealed a drasticdecrease in the intensity of the 220 kDa band (Fig. 6, lane indicating that this material was largely due to laminin-1molecules and not attributable to laminin-5. Since a band in tposition of α3 (165 kDa) was precipitated with the anti-laminin-1 antibody (Fig 6, lane 2), it can be speculated thother α3 chain containing-laminins may be present. Laminin6 (α3β1γ1) and laminin-7 (α3β2γ1) isoforms are goodcandidates since bands migrating in the position of the β1, γ1(210, 220 kDa) and α3, β2 (190 kDa) chains can be seen.

sd on

1998

h

yasen

n9ys

ls

-

f

y

%e

.5

oo

1.

to

s


Fig. 6.Characterization of the 220 kDa high-molecular complex bsequential immunoprecipitation. Immunoprecipitation ofmetabolically labelled medium from HT29-Fu cells with mAb BM165 followed by a prolonged exposure of the gel yielded the classpattern of bands migrating at 165, 155, 140 and 105 kDa as well weaker bands at 190 and 220 kDa (lane 1). To identify this latterband, preclearing experiments were performed in which HT29-Fumedium was sequentially immunoprecipitated three times by apolyclonal anti-laminin-1 antibody (lane 2). Subsequentimmunoprecipitation with mAb BM 165 revealed that the majoritymaterial was eliminated from the 220 kDa region, correspondingtherefore to β1γ1 chains (lane 3). Further details are the same as iFig. 3.

Identification of the integrin receptors involved inadhesion of colonic cells to laminin-5The localisation of laminin-5 in the villus basement membra(Leivo et al., 1996; Orian-Rousseau et al., 1996), and observations that laminin γ2 and α3 chains might be involvedin migratory processes such as wound repair and invas(Ryan et al., 1994; Pyke et al., 1994, 1995; Lotz et al., 19led us to analyse the integrin receptors involved in the adheof HT29p and Caco2 cells to laminin-5. Cell adhesion assshowed that attachment of Caco2 and HT29 cells to increasconcentrations of exogenous laminin-5 was rather similar (F7). Their adhesion patterns on laminin-5 did not differ frothose on laminin-1/nidogen. To assess the specificity of c

Cac

0

0.2

0.4

0.6

0.8

0

Cel

l ad

hes

ion

(O

.D.

at

550

nm

)

Fig. 7.Adhesion of colonic cancer cells tovarious concentrations of substrata.Multiwell plates were coated with laminin-1(d), laminin-5 (s) or with fibronectin (m) atvarious concentrations. After saturation with1% BSA, cells were seeded and allowed toattach for 2 hours. Extent of cell adhesionwas measured with a colorimetric reaction.

nethe

ion97)sionaysingig.mell

adhesion to laminin-5, the culture wells were first coated witlaminin-5 or laminin-1/nidogen complex and then with mAbBM165 before addition of the cell suspension. The antibodhampered the adhesion of the cell lines to laminin-5 whereno effect was observed on adhesion to the laminin-1/nidogcomplex (illustrated for HT29p cells in Fig. 8).

In the literature, the putative receptors for laminin-5described so far are α3β1, α6β1 and α6β4 depending on thecell line (Carter et al., 1991; Delwel et al., 1993, 1994; Niesseet al., 1994; Rousselle and Aumailley, 1994). Caco2 and HT2cell lines are of particular interest because they displadifferent integrins on their cell surface. Caco2 cells expreshigh levels of integrin α1β1 and α2β1 and lower levels ofintegrin α3β1 and α6β1; they do not express integrin α6β4(Basson et al., 1992; Ebert, 1996). In contrast, HT29 celexpress integrin α6β4 in addition to integrins α1β1, α2β1 andα3β1 (Simon-Assmann et al., 1994b; Ebert, 1996). Todetermine the receptor(s) involved in their adhesion to laminin5, we performed inhibition assays with the anti-α2 (mAbP1E6), anti-α3 (mAb P1B5), anti-α6 (mAb GoH3), anti-β1(mAb P4C10) and anti-β4 (mAb 3E1) integrin antibodies onHT29p (Fig. 9A) and Caco2 (Fig. 9B) cells. Adhesion oHT29p cells to laminin-5 was totally inhibited by mAb P1E6.Inhibition was dose-dependent and the anti-integrin antiboddilution leading to 50% inhibition was 1/8,650. Monoclonalantibodies P1B5 and GoH3 led, respectively, to 55% and 60inhibition of HT29p cell adhesion. With antibodies against thintegrin β subunits, total inhibition was obtained with mAbP4C10 but no inhibition was observed with mAb 3E1Therefore, we can conclude that the potential laminin-receptors in HT29p cells are integrins α2β1 and α3β1.However, adhesion via the α6β4 integrin could also occurthrough the α6 subunit. In the case of Caco2 cells, adhesion tlaminin-5 was almost totally prevented by the mAb P1E6. Ninhibition was observed with mAb P1B5 while 25% inhibitionwas obtained with mAb GoH3. mAb P4C10 induced inhibitionof adhesion whereas no effect was observed with mAb 3ETherefore, the α2β1 receptor and to a lesser extent the α6β1integrin appear to play a role in the adhesion of Caco2 cells laminin-5.

Similar experiments were performed on cell suspensionseeded on laminin-1 (Fig. 9A and B). The integrin α6β4

y

icalas

of

n

o2 cells

10 20

HT29p cells

0

0.2

0.4

0.6

0 10 20

Cel

l ad

hes

ion

(O

.D.

at

550

nm

)

Concentration of substrata (µg/ml)


,;

e

rn

0

20

40

60

80

100

0 5 10Antibody concentration (µg/ml)

Cel

l ad

hes

ion

(%

of

con

tro

l)

A

Fig. 8.Effect of mAb BM165 on adhesion of HT29p cells to laminin-5. Multiwell plates were coated with laminin-5 (s) or with laminin-1 (d)at 5 and 10 µg/ml, respectively. After saturation with 1% BSA, the wells were incubated with various concentrations of mAb BM 165 for 60minutes before cell adhesion assays. Extent of cell adhesion depicted in A was measured with a colorimetric reaction and expressed as apercentage of adhesion in the absence of antibodies. The concentration of mAb BM165 leading to 50% of inhibition equals to 0.2 µg/ml. Thedata were compiled from 2 independent assays performed in duplicate. Attachment and morphology of the cells on laminin-5 (B) and onlaminin-5 in the presence of mAb BM165 (C). Bars, 50 µm.

appeared to mediate adhesion to laminin-1 in HT29 cells whlack α6β1 integrin. None of the anti-α subunits antibodiestested had a detectable effect on the adhesion of Caco2 to laminin-1 in contrast to the anti-β1 mAb P4C10; thus, theα1β1 integrin present on Caco2 cells, and on HT29 cells, mibe a candidate receptor of this ligand.

It is worth noting that a distinct cellular specificity waobserved concerning the implication of α2β1 integrin in cellbinding to laminin-5. Indeed, normal human keratinocyt

LN-5 LN-1

0

50

100

150

P1E

6

P1B

5

GoH

3

P4C

10

3E1

P1E

6

P1B

5

GoH

3

P4C

10

3E1

cell

adh

esio

n

(%

of

con

tro

l)

HT29p cellsA

**

* * 0

50

100

150

P1E

6

cell

adh

esio

n

(%

of

con

tro

l)

B

****

*

Fig. 9.Effect of anti-integrin subunit antibodies on adhesion of inte(A) and Caco2 (B) cells were seeded on 5 µg/well laminin-5 (LN-5that gave maximum inhibition of adhesion: mAb P1E6 against α2 (1/27against α3 (1/900), mAb GoH3 against α6 as well as mAb P4C10 agawere initially performed to determine the optimal concentrations othe antibody. The number of cells that attached in the absence of the average results. Asterisks indicate significant differences betw** P<0.005. (C) Representative experiment for freshly isolated norP1E6 against α2 subunit only on laminin-1.

ich

cells

ght

s

es

isolated as previously published (Rousselle and Aumailley1994) do not use this integrin for their adhesion to laminin-5yet, they use this integrin to bind to laminin-1 (Fig. 9C).

As α2β1 integrin has not yet been described as a putativreceptor to laminin-5, further experiments were done toconfirm that this integrin is indeed implicated in colonic cancecells. First, to investigate the direct interaction between integriα2β1 and laminin-5, the binding of biotinylated purified α2β1integrin to immobilised laminin-5 was measured. Fig. 10A

LN-5 LN-1

P1

B5

Go

H3

P4

C1

0

3E

1

P1E

6

P1

B5

Go

H3

P4

C1

0

3E

1

Caco2 cells

0

50

100

150

P1E

6

P1E

6cell

adh

esio

n

(%

of

con

tro

l)

LN-5 LN-1 Keratinocytes C

**

***

stinal epithelial cells and keratinocytes to laminin-5 and laminin-1. HT29p) or on 10 µg/well laminin-1(LN-1). The antibody concentrations were those00) for the HT29 cells and (1/8100) for the Caco2 cells, mAb P1B5inst β1 (1/8100), mAb 3E1 against β4 (1/25). Dose-dependent curves

f antibodies. Otherwise, no inhibition was observed whatever the dilution ofantibodies was taken as 100% adhesion. Error bars represent s.e.m. values ofeen the assay (with antibody) and the control (without antibody); *P<0.05;mal human keratinocytes showing an inhibition of cell adhesion by mAb

2000

fns

t-


A B

LN-5 LN-1

OD

450

nm

0.4

0.3

0.2

0.10.1 1 10 100

LN-5 concentration (µg/ml)

0.2

0.1

0

0.3

+EDTA +6F1 mAb

+EDTA +6F1 mAb

OD

450

nm

Fig. 10.(A) Dose response curve of biotinylated α2β1 integrin(2 µg/ml) binding to laminin-5 at indicated concentrations.(B) α2β1 integrin binding to laminin-5 (50 µg/ml) is blockedby EDTA (10 mM) and by the anti-α2 integrin antibody 6F1(5 µg/ml). A comparison is shown of α2β1 integrin binding tolaminin-1 (10 µg/ml). In this panel the background binding hasbeen subtracted from the data. In both A and B results areexpressed as the mean ± standard error for three wells.

shows that α2β1 integrin supports dose-dependent binding laminin-5. This molecular interaction was completely inhibiteby EDTA and by the anti-α2 integrin mAb 6F1 (Fig. 10B). Thisshows a specific cation-dependent interaction between α2β1 receptor and laminin-5 like for integrin-ligand binding igeneral (Calderwood et al., 1997). A second major argum

Fig. 11.Confocal analysis of α2 integrin subunitdistribution. Images correspond to X-Y (A-F) andto Y-Z (G-I) plane sections of one single cell.When the HT29p cells were seeded onto laminin-1, the α2 integrin staining was regularly found allaround the cells (illustrated for a single cell,A-C,G). When seeded on laminin-5, the cellsshowed a strong basal distribution of α2 integrinsubunit (D,H,I) while on lateral and apicalsurfaces (E,F) a lower amount of α2 integrin wasdetected. Pseudocolor image enhancement fromred to yellow indicates lower to higherfluorescence intensity. Bars, 10 µm.

ofd

thenent

in favour of α2β1 receptor binding to laminin-5 was providedby the confocal microscopy analysis. The distribution ointegrin α2 subunit, stained with mAb P1E6, was analysed oHT29p cells cultured on laminin-1 and laminin-5. Preparationthrough the X-Y section revealed a strikingly differendistribution of α2 integrin between the two substrata (Fig. 11A


itsi)e

cthell);

tion

d byes

lehea

rml.,he

ar

-6

to).- in

de

es

d;

al., Of

1

es

anan

s

the

or on

F). When seeded on laminin-1, the cells depicted a relativuniform membrane staining of α2 subunit whatever the sectionplane (illustrated for one single cell; Fig. 11A-C). On laminin5, there was a clear-cut accumulation of α2 integrin staining atthe basal pole of the cell (Fig. 11D) visualised by a segregaof intensively labelled patches. Concomitantly, there wasdecrease in the staining at the lateral and apical part of the(Fig. 11E,F). The segregation of α2 integrin on laminin-5substratum was confirmed by the confocal analysis throughY-Z section plane of cells (Fig. 11H versus G). In some casthe cells showed a well spread and flattened morpholoparallel to the α2 integrin rearrangement (Fig 11I). It is alsimportant to stress that a clear epithelial resurfacing of theα2integrin subunit at the ventral surface of the cells paralleled deposition of laminin-5 by the cells themselves after 5 daysculture (not illustrated).

DISCUSSION

In the skin, laminin-5 is known to play a major role in ceadhesion and integrity of the basement membranes in regof cell-connective tissue interactions (Rousselle et al., 199The intestine is characterised by tightly orchestrated cmovements and progressive cell differentiation events durmorphogenesis and cell renewal. Linked to these procesdifferential expression of basement membrane molecules receptors such as integrins has been described (for reviewsSimon-Assmann et al., 1995, 1998; Beaulieu, 1997). In human mature intestine, which is composed of a simepithelium, the location of laminin-5 at the basal pole epithelial cells according to an increasing gradient settled frcrypt to villus tip (Orian-Rousseau et al., 1996) would argfor a role in differentiation and/or migration.

The present study shows that the highly differentiated cocarcinoma Caco-2 cells express very low amounts of lamin5, whereas HT29p cells and two derived subclones which phenotypically related to foetal epithelial cells producsignificant amounts of the protein. These observations cancorrelated to the cellular origin of laminin-5 described recen(Orian-Rousseau et al., 1996). Indeed, detection of laminiγ2 chain with anti-mouse antibodies on hybrid chick/mouintestines pointed to a dual and developmentally-regulaexpression of this chain: synthesis by moderately differentiaepithelial cells during intestinal development, relayed by tdifferentiated mesenchymal cells at more advanced stagesfurther interest is the finding of an inverse correlation betweexpression of laminin-5 and of laminin-1. Indeed, HT29 celack laminin-αl chain in contrast to Caco-2 cells which expreand gradually accumulate laminin-1 as they underenterocytic differentiation (De Arcangelis et al., 1994; Vachoand Beaulieu, 1995). Furthermore, using an antisense Rstrategy we demonstrated the direct involvement of lamininαl chain in the basement membrane assembly and in Caccell differentiation (De Arcangelis et al., 1996). Although thmechanism of epithelial cell migration in the adult intestinremains an enigma (Heath, 1996) the hypothesis that lamin5 could be involved in cell migration rather than in cedifferentiation is supported by the following observations: (laminin-5 γ2 chain is preferentially expressed in invasive cein colon cancers (Pyke et al., 1994, 1995); (ii) the promo

ely

-

tion a cell

thees,gy

o the of

llions1).ellingses,and see

thepleofomue

lonin-aree betlyn-5setedtedhe. OfenllsssgonNA-1o-2eein-lli)llster

activity of the LAMC2 gene encoding laminin γ2 chain isupregulated by the hepatocyte growth factor, known for effect on cell motility (J. Olsen et al., unpublished data); (iilaminin-5 α3 chain gene expression is increased in thkeratinocytes at the wound sites (Ryan et al., 1994); (iv) a largecell-adhesive protein formerly called ladsin, which is in fasimilar to laminin-5, has been shown to stimulate both tchemotactic and chemokinetic migration of the rat liver celine BRL and of endothelial cells (Kikkawa et al., 1994, 1996(v) finally, a specific cleavage of laminin-5 γ2 chain by matrixmetalloprotease-2 has been shown recently to induce migraof breast epithelial cells (Giannelli et al., 1997).

Immunoprecipitation and immunoblot experiments alloweus to identify the laminin-5 molecule produced and secretedcolonic cancer cell lines. In all HT29 cell populations, thprominent form of the molecule found in the cells as well ain the media is the 440 kDa heterotrimeric molecucorresponding to the KM1 intermediate form described in tskin by Marinkovich et al. (1992a). It consists of the 165 kDα3, the 145 kDa β3 and the 155 kDa γ2 chains. Thus, in colonicepithelial cells, the 155 kDa γ2 precursor is not efficientlyprocessed into the 105 kDa form to produce the 400 kDa fofound in keratinocytes and SCC25 cells (Marinkovich et a1992a; Rousselle and Aumailley, 1994). In Caco-2 cells, tmajor chain synthesized is the γ2 chain, while α3 and β3 chainsare barely detectable; this leads to an intracellulaccumulation of the 155 γ2 chain in Caco-2 cells. Anotherinteresting feature is the possible expression of laminin(α3β1γ1) and laminin-7 (α3β2γ1) in HT29 cells. These twoisoforms have been shown to be enhanced in responsewounding during intestinal restitution (Lotz et al., 1997Complexes between laminin-5/laminin-6 and laminin5/laminin-7 have been described in the amniotic membranewhich they are covalently associated through disulphibridges (Champliaud et al., 1996).

Our present study shows that, in contrast to other cell typsuch as keratinocytes, colonic epithelial cells can use the α2β1integrin to adhere to laminin-5. This result was unexpecteindeed, the α2β1 complex, clearly identified as a collagenreceptor on MDCK cells (Saelman et al., 1995), or aslaminin-1 receptor on Clone A colon cancer cells (Lotz et a1990) has never been described as a laminin-5 receptor.particular interest is the fact that this α2β1 receptor is notimplicated in HT29 and Caco-2 cell adhesion to laminin-although in the solid phase assay α2β1 integrin is able to linklaminin-1. This observation and the finding that keratinocytdo not adhere to laminin-5 through the α2β1 receptorstrengthen the hypothesis that the ligand binding specificity cbe determined by the cellular environment. In this regard, Chand Hemler (1993) showed, by expressing the α2 cDNA in twodifferent cell types, that the α2β1 integrin expressed by thesecells displays distinct patterns of functional activity, such adifferential binding on laminin-1 or collagen. Apart from theα2β1 integrin, Caco-2 cells use the α6β1 integrin, whereasHT29 cells bind to laminin-5 through α3β1 and α6β4integrins. These data are in accordance with those of literature, showing that the α3β1 (Carter et al., 1991; Delwelet al., 1994; Rousselle and Aumailley, 1994), the α6β1 (Delwelet al., 1993; Rousselle and Aumailley, 1994), and the α6β4(Niessen et al., 1994) integrins are receptors for laminin-5. Fthese receptors, interactions with laminin-5 are dependent

2002

ce

hat

ehe

h

al

)

P.

e

rndK

d


the coil-coiled structure. Accordingly, adhesion of HT29 ceon heat-denaturated laminin-5 (in which the cell-bindinactivity carried by the carboxy-terminal domain of thmolecule was impaired) was reduced to 50% (not illustrateOn the basis of the data obtained with HT29 cells, one cpostulate that their adhesion to laminin-5 involves multimerization process (Diamond and Springer, 1994) which integrin α2β1 exerts a regulatory effect on α6β4 or vice-versa. This type of model, which presumes the existenceintegrin molecules that become apposed, is described in osystems (Karecla et al., 1994; Tozeren et al., 1994). SimilaCarter et al. (1991) have proposed that α3β1 recruits α6β4 intostable anchoring complexes in keratinocytes.

Concerning the correlation between our present data integrin binding to laminin-5 in the intestine, one can at thpoint only speculate. While laminin-5 is expressed early duridevelopment, integrin α6β4 appears only later on duringmorphogenesis (Simon-Assmann et al., 1994a) indicating involvement of the β1 integrin family for laminin-5 binding atthis period of time. During villus onset and formation, thexpression of α2β1 coincides with that of laminin-5: earlyexpression in the intestinal anlage and gradient wpronounced expression at the tips of villi (Wu and Santo1994; Perreault et al., 1995; Dieckgraefe et al., 1996; Leivoal., 1996; Orian-Rousseau et al., 1996). Thus, duridevelopment, the α2β1 integrin may serve as a receptor folaminin-5. Yet, no obvious correlation can be made betwethe locations of laminin-5 and α2β1 integrin in the adult organin which α2β1 integrin is described in the crypts (Wu anSantoro, 1994; Perreault et al., 1995), while laminin-5 located mostly along the villi (Leivo et al., 1996; OrianRousseau et al., 1996). At this stage, one can assume attachment of cells to laminin-5 in the villus region is mediatby other integrins such as α6β4 or α3β1 which are indeedfound in this area. In the human adult intestine,α6β4 integrinis also colocalized with HD1, a hemidesmosomal protewhereas the BP (bullous pemphigoid) antigens are not detesuggesting the existence of type II hemidesmosomes (OrRousseau et al., 1996). Similarly, HT29 cells only express Hand α6β4 integrin which colocalize in basal patches betweeactin stress fibers (Fontao et al., 1997). It is tempting therefto speculate that these type II hemidesmosomal structutogether with laminin-5 could be implicated mostly in cemigration, which occurs continuously in the adult organ.

Until now, binding sites of the integrins have not beemapped on laminin-5. Pfaff et al. (1994) have reported that EIXNd fragment in the cross region of laminin-1 waresponsible for binding to integrin α2β1, suggesting that theN-terminal region of the laminin α1 chain was implicated; thisregion is deleted in the laminin α3A chain (Ryan et al., 1994;Galliano et al., 1995). This shorter α3 A chain was the first tobe identified by immunological methods and cDNA clonin(Rousselle et al., 1991; Aberdam et al., 1994a). Subsequehowever, multiple α3 cDNAs were identified in human andmouse (Ryan et al., 1994; Galliano et al., 1995). RecenMiner et al. (1997) described a novel full-length α3B form thatcomprises additional domains, in particular domains V and VTherefore interaction of integrin α2β1 with this α3B isoformis possible. Alternatively, the possibility cannot be excludthat integrin α2β1 could bind to another chain of the laminin5 molecule. In accordance with this are the data of Underwo

llsged).anain

oftherrly,

andisng

the

e

ithro, etngren

dis-that

ed

in,ctedian-D1n

oreresll

nthes

gntly,

tly

I.

ed-od

et al. (1995) who identified a binding sequence on the β1subunit of laminin-1. More recently, using systematic peptidescreening, Nomizu et al. (1997) located an active sequenresponsible for interaction with integrin α2β1 on the γ1 chainin a region which is highly conserved with theγ2 chain.Definitive identification of the fragments responsible for thebinding of α2β1 integrin will be possible only when smallbiologically active fragments of laminin-5 will be available.

From the data reported here, it is reasonable to speculate tlaminin-5 could exert different physiological functions such asanchorage, cell migration or differentiation depending on thtissue considered and on the receptors used by the cells. Tdata reported herein on colonic epithelial cells together witthe former analysis of laminin-5 expression and origin in theintestinal tissue, give a basic knowledge which will contributeto the understanding of the biological regulation ordisregulation of this molecule in intestinal pathologies.

We are grateful to Prof R. E. Burgeson (Massachusetts GenerHospital, Charlestown, MA, USA) for generously providing us withthe BM 165 mAb and to Drs A. Zweibaum and T. Lesuffleur(INSERM Unit 178, Paris, France) for providing the HT29 cell lines.We also thank J. L. Vonesch (INSERM Unit 184, Strasbourg, Francefor his contribution to the localisation of integrin α2 subunit byconfocal microscopy. We thank C. Arnold and C. Leberquier forexpert technical assistance. We gratefully acknowledge Prof. Thorogood (Developmental Biology Unit, Institute of Child Health,London, UK) for critically reading the manuscript. We also thank I.Gillot, B. Lafleuriel and L. Mathern for preparation of the manuscriptand illustrations. This study was supported by INSERM, CNRS, thLigue Nationale contre le Cancer (Comité du Haut-Rhin et del’Ardèche), and by grants from the Association pour la Recherche sule Cancer (#1251, #3056 and #5003 to M. Kedinger, P. Rousselle aD. Aberdam, respectively), and the Cancer Research Campaign, U(project grant # SP2287/0/0/ to J. Gavrilovic).

REFERENCES

Aberdam, D., Galliano, M. F., Mattei, M. G., Pisani-Spadafora, A.,Ortonne, J. P. and Meneguzzi, G.(1994a). Assignment of mouse niceingenes to chromosomes 1 and 18. Mamm. Genome5, 229-233.

Aberdam, D., Galliano, M. F., Vailly, J., Pulkkinen, L., Bonifas, J.,Christiano, A. M., Tryggvason, K., Uitto, J., Epstein, E. H. Jr andOrtonne, J. P.(1994b). Herlitz’s junctional epidermolysis bullosa is linkedto mutations in the gene (LAMC2) for the gamma 2 subunit of nicein/kalinin(LAMININ-5). Nature Genet. 6, 299-304.

Basson, M. D., Modlin, I. M. and Madri, J. A. (1992). Human enterocyte(Caco-2) migration is modulated in vitro by extracellular matrixcomposition and epidermal growth factor. J. Clin. Invest. 90, 15-23.

Basson, M. D., Turowski, G. and Emenaker, N. J. (1996). Regulation ofhuman (Caco-2) intestinal epithelial cell differentiation by extracellularmatrix proteins. Exp. Cell Res. 225, 301-305.

Beaulieu, J. F. (1997). Recent work with migration/patterns of expression:cell-matrix interactions in human intestinal cell differentiation. In The Gutas a Model in Cell and Molecular Biology. Proceedings of the FalkSymposium 94 (ed. F. Halter, D. Winton and N. A. Wright), pp. 165-179.Kluwer Academic Publishers, Lancaster, UK.

Brown, T. A., Gil, S. G., Sybert, V. P., Lestringant, G. G., Tadini, G.,Caputo, R. and Carter, W. G. (1996). Defective integrin alpha 6 beta 4expression in the skin of patients with junctional epidermolysis bullosa anpyloric atresia. J. Invest. Dermatol. 107, 384-391.

Calderwood, D. A., Tuckwell, D. S., Eble, J., Kühn, K. and Humphries,M. J. (1997). The integrin α1 A-domain is a ligand binding site for collagensand laminin. J. Biol. Chem. 272, 12311-12317.

Carter, W. G., Ryan, M. C. and Gahr, P. J. (1991). Epiligrin, a new celladhesion ligand for integrin α3β1 in epithelial basement membranes. Cell65, 599-610.


h

isd

ll

of

e

gh

el

nsl

cion

al

Champliaud, M. F., Lunstrum, G. P., Rousselle, P., Nishiyama, T., Keene,D. R. and Burgeson, R. E. (1996). Human amnion contains a novel laminivariant, laminin 7, which like laminin 6, covalently associates with lamin5 to promote stable epithelial-stromal attachment. J. Cell Biol. 132, 1189-1198.

Chan, B. M. C. and Hemler, M. E. (1993). Multiple functional forms of theintegrin VLA-2 can be derived from a single α(2) cDNA clone –Interconversion of forms induced by an anti-β(1) antibody. J. Cell Biol. 120,537-543.

Chantret, I., Barbat, A., Dussaulx, E., Brattain, M. G. and Zweibaum, A.(1988). Epithelial polarity, villin expression, and enterocytic differentiatioof cultured human colon carcinoma cells: a survey of twenty cell linCancer Res. 48, 1936-1942.

De Arcangelis, A., Simo, P., Lesuffleur, T., Kedinger, M. and Simon-Assmann, P. (1994). L’expression de la laminine est corrélée à différenciation des cellules cancéreuses coliques humaines. Gastroentérol.Clin. Biol. 18, 630-637.

De Arcangelis, A., Neuville, P., Boukamel, R., Lefebvre, O., Kedinger, M.and Simon-Assmann, P. (1996). Inhibition of laminin α1-chain expressionleads to alteration of basement membrane assembly and cell differentiaJ. Cell Biol. 133, 417-430.

Delwel, G. O., Hogervorst, F., Kuikman, I., Paulsson, M., Timpl, R. andSonnenberg, A. (1993). Expression and function of the cytoplasmic varianof the integrin-α6 subunit in transfected K562 cells – Activation-dependeadhesion and interaction with isoforms of laminin. J. Biol. Chem. 268,25865-25875.

Delwel, G. O., de Melker, A. A., Hogervorst, F., Jaspars, L. H., Fles, D. L.,Kuikman, I., Lindblom, A., Paulsson, M., Timpl, R. and Sonnenberg,A. (1994). Distinct and overlapping ligand specificities of the α3A β1 andα6A β1 integrins: recognition of laminin isoforms. Mol. Biol. Cell 5, 203-215.

Diamond, M. S. and Springer, T. A. (1994). The dynamic regulation ofintegrin adhesiveness. Curr. Biol. 4, 506-517.

Dieckgraefe, B. K., Santoro, S. A. and Alpers, D. H. (1996).Immunolocalization of α-integrin subunits and extracellular matrixcomponents during human colonic organogenesis. Gastroenterology110,58-71.

Ebert, E. C. (1996). Mechanisms of colon cancer binding to substratum acells. Dig. Dis. Sci. 41, 1551-1556.

Fine, J. D. (1994). International symposium on epidermolysis bullosa. Invest.Dermatol. 103, 839-843.

Fogh, J., Fogh, J. M. and Orfeo, T. (1977). One hundred and twenty-sevecultured human tumor cell lines producing tumors in nude mice. J. Nat.Cancer Inst. 59, 221-226.

Fontao, L., Dirrig, S., Owaribe, K., Kedinger, M. and Launay, J. F. (1997).Polarized expression of HD1: relationship with the cytoskeleton in culturhuman colonic carcinoma cells. Exp. Cell Res. 231, 319-327.

Galliano, M. F., Aberdam, D., Aguzzi, A., Ortonne, J. P. and Meneguzzi,G. (1995). Cloning and complete primary structure of the mouse lamiα3 chain. Distinct expression pattern of the laminin α3A and α3B chainisoforms. J. Biol. Chem. 270, 21820-21826.

Giannelli, G., Falk-Marzillier, J., Schiraldi, O., Stetler-Stevenson, W. G.and Quaranta, V. (1997). Induction of cell migration by matrixmetalloprotease-2 cleavage of laminin-5. Science277, 225-228.

Goulet, O., Kedinger, M., Brousse, N., Cuenod, B., Colomb, V., Patey, N.,de Potter, S., Mougenot, J. F., Canioni, D. and Cerf-Bensussan, N.(1995). Intractable diarrhea of infancy with epithelial and basememembrane abnormalities. J. Pediatr. 127, 212-219.

Heath, J. P. (1996). Epithelial cell migration in the intestine. Cell. Biol. Int.20, 139-146.

Karecla, P. I., Timpl, R. and Watt, F. M. (1994). Adhesion of humanepidermal keratinocytes to laminin. Cell Adhes. Commun. 2, 309-318.

Kedinger, M., Haffen, K. and Simon-Assmann, P. (1987). Intestinal tissueand cell cultures. Differentiation36, 71-85.

Kern, A., Eble, J., Golbik, R. and Kuhn, K. (1993). Interaction of type IVcollagen with the isolated integrins α1β1and α2β1. Eur. J. Biochem. 215,151-159.

Kikkawa, Y., Umeda, M. and Miyazaki, K. (1994). Marked stimulation ofcell adhesion and motility by ladsin, a laminin- like scatter factor. J.Biochem. (Tokyo)116, 862-869.

Kikkawa, Y., Akaogi, K., Mizushima, H., Yamanaka, N., Umeda, M. andMiyazaki, K. (1996). Stimulation of endothelial cell migration in cultureby ladsin, a laminin-5-like cell adhesion protein. In Vitro Cell Dev. Biol.Anim. 32, 46-52.

nin

nes.

la

tion.

tsnt

nd

n

ed

nin

nt

Leivo, I., Tani, T., Laitinen, L., Bruns, R., Kivilaakso, E., Lehto, V. P.,Burgeson, R. E. and Virtanen, I. (1996). Anchoring complexcomponents laminin-5 and type VII collagen in intestine: association witmigrating and differentiating enterocytes. J. Histochem. Cytochem. 44,1267-1277.

Lesuffleur, T., Barbat, A., Dussaulx, E. and Zweibaum, A. (1990). Growthadaptation to methotrexate of HT-29 human colon carcinoma cells associated with their ability to differentiate into columnar absorptive anmucus-secreting cells. Cancer Res. 50, 6334-6343.

Lesuffleur, T., Kornowski, A., Luccioni, C., Muleris, M., Barbat, A.,Beaumatin, J., Dussaulx, E., Dutrillaux, B. and Zweibaum, A. (1991).Adaptation to 5-fluorouracil of the heterogeneous human colon tumor celine HT-29 results in the selection of cells committed to differentiation. Int.J. Cancer49, 721-730.

Lotz, M. M., Korzelius, C. A. and Mercurio, A. M. (1990). Human coloncarcinoma cells use multiple receptors to adhere to laminin: involvement α6β4 and α2β1 integrins. Cell Regul. 1, 249-257.

Lotz, M. M., Nusrat, A., Madara, J. L., Ezzell, R., Wewer, U. M. andMercurio, A. M. (1997). Intestinal epithelial restitution. Involvement ofspecific laminin isoforms and integrin laminin receptors in wound closurof a transformed model epithelium. Am. J. Pathol. 150, 747-760.

Marinkovich, M. P., Lunstrum, G. P. and Burgeson, R. E. (1992a). Theanchoring filament protein kalinin is synthesized and secreted as a himolecular weight precursor. J. Biol. Chem. 267, 17900-17906.

Marinkovich, M. P., Lunstrum, G. P., Keene, D. R. and Burgeson, R. E.(1992b). The dermal-epidermal junction of human skin contains a novlaminin variant. J. Cell Biol. 119, 695-703.

Messent, A. J., Tuckwell D. S., Knauper V., Humphries, M. J., MurphyG. and Gavrilovic J. (1998). Effects of collagenase-cleavage of type Icollagen on α2β1 integrin-mediated cell adhesion. J. Cell Sci.111, 1127-1135.

Miner, J. H., Patton, B. L., Lentz, S. I., Gilbert, D. J., Snider, W. D.,Jenkins, N. A., Copeland, N. G. and Sanes, J. R. (1997). The laminin αchains: expression, developmental transitions, and chromosomal locatioof α1-5, identification of heterotrimeric laminins 8-11 and cloning of a noveα3 isoform. J. Cell Biol. 137, 685-701.

Neutra, M. and Louvard, D. (1989). Differentiation of intestinal cells in vitro.In Functional Epithelial Cells in Culture(ed. K. S. Matlin and J. D.Valentich), pp. 363-398. Alan R. Liss Inc.

Niessen, C. M., Hogervorst, F., Jaspars, L. H., de Melker, A. A., Delwel,G. O., Hulsman, E. H., Kuikman, I. and Sonnenberg, A. (1994). Theα6β4 integrin is a receptor for both laminin and kalinin. Exp. Cell Res. 211,360-367.

Niessen, C. M., van Der Raaij-Helmer, L. M. H., Hulsman, E. H. M., vanDer Neut, R., Jonkman, M. F. and Sonnenberg, A. (1996). Deficiency ofthe integrin beta 4 subunit in junctional epidermolysis bullosa with pyloriatresia: consequences for hemidesmosome formation and adhesproperties. J. Cell Sci. 109, 695-706.

Nomizu, M., Kuratomi, Y., Song, S. -Y., Ponce, M. L., Hoffman, M. P.,Powell, S. K., Miyoshi, K., Otaka, A., Kleinman, H. K., Yamada, Y.(1997). Identification of cell binding sequences in mouse laminingamma-1 chain by systematic peptide screening. J. Biol. Chem. 272,32198-32205.

Orian-Rousseau, V., Aberdam, D., Fontao, L., Chevalier, L., Meneguzzi,G., Kedinger, M. and Simon-Assmann, P. (1996). Developmentalexpression of laminin-5 and HD1 in the intestine: epithelial to mesenchymshift for the laminin gamma-2 chain subunit deposition. Dev. Dynam. 206,12-23.

Perreault, N., Vachon, P. H. and Beaulieu, J. F. (1995). Appearance anddistribution of laminin A chain isoforms and integrin α2, α3, α6, β1 and β4subunits in the developing human small intestinal mucosa. Anat. Rec. 242,242-250.

Pfaff, M., Gohring, W., Brown, J. C. and Timpl, R. (1994). Binding ofpurified collagen receptors (α6β1, α2β1) and RGD-dependent integrins tolaminins and laminin fragments. Eur. J. Biochem. 225, 975-984.

Pinto, M., Robine-Léon, S., Appay, M. D., Kedinger, M., Triadou, N.,Dussaulx, E., Lacroix, B., Simon-Assmann, P., Haffen, K., Fogh, J., etal. (1983). Enterocyte-like differentiation and polarization of the humancolon carcinoma cell-line (Caco-2) in culture. Biol. Cell 47, 323-330.

Pulkkinen, L., Kimonis, V. E., Xu, Y. L., Spanou, E. N., Mclean, W. H. I.and Uitto, J. (1997). Homozygous alpha 6 integrin mutation in junctionalepidermolysis bullosa with congenital duodenal atresia. Hum. Mol. Genet.6, 669-674.

Pyke, C., Romer, J., Kallunki, P., Lund, L. R., Ralfkiaer, E., Dano, K. and

2004

t

er

n

t

-e.

.


Tryggvason, K. (1994). The gamma 2 chain of kalinin/laminin 5 ispreferentially expressed in invading malignant cells in human cancers. Am.J. Pathol. 145, 782-791.

Pyke, C., Sale, S., Ralfkiaer, E., Romer, J., Dano, K. and Tryggvason, K.(1995). Laminin-5 is a marker of invading cancer cells in some humcarcinomas and is coexpressed with the receptor for urokinase plasminoactivator in budding cancer cells in colon adenocarcinomas. Cancer Res. 55,4132-4139.

Rousselle, P., Lunstrum, G. P., Keene, D. R. and Burgeson, R. E. (1991).Kalinin – An epithelium-specific basement membrane adhesion molecthat is a component of anchoring filaments. J. Cell Biol. 114, 567-576.

Rousselle, P. and Aumailley, M. (1994). Kalinin is more efficient than lamininin promoting adhesion of primary keratinocytes and some other epithecells and has a different requirement for integrin receptors. J. Cell Biol. 125,205-214.

Ryan, M. C., Tizard, R., VanDevanter, D. R. and Carter, W. G. (1994).Cloning of the LamA3 gene encoding the α3 chain of the adhesive ligandepiligrin. Expression in wound repair. J. Biol. Chem. 269, 22779-22787.

Saelman, E., Keely, P. and Santoro, S. (1995). Loss of MDCK cell α2β1integrin expression results in reduced cyst formation, failure of hepatocgrowth factor/scatter factor- induced branching morphogenesis, aincreased apoptosis. J. Cell Sci. 108, 3531-3540.

Simo, P., Simon-Assmann, P., Bouziges, F., Leberquier, C., Kedinger, M.,Ekblom, P. and Sorokin, L. (1991). Changes in the expression of lamininduring intestinal development. Development112, 477-487.

Simon-Assmann, P., Duclos, B., Orian-Rousseau, V., Arnold, C., Mathelin,C., Engvall, E. and Kedinger, M. (1994a). Differential expression oflaminin isoforms and α6-β4 integrin subunits in the developing human anmouse intestine. Dev. Dynam. 201, 71-85.

Simon-Assmann, P., Leberquier, C., Molto, N., Uezato, T., Bouziges, F. andKedinger, M. (1994b). Adhesive properties and integrin expression profilof two colonic cancer populations differing by their spreading on laminiJ. Cell Sci. 107, 577-587.

Simon-Assmann, P., Kedinger, M., De Arcangelis, A., Orian-Rousseau, V.and Simo, P. (1995). Extracellular matrix components in intestinadevelopment. Experientia51, 883-900.

Simon-Assmann, P., Lefebvre, O., Bellissent-Waydelich, A., Olsen, J.,

angen

ule

lial

ytend

d

esn.

l

Orian-Rousseau, V. and De Arcangelis, A. (1998). The laminins: role inintestinal morphogenesis and differentiation. Ann. NY Acad. Sci. (in press).

Tozeren, A., Kleinman, H. K., Wu, S., Mercurio, A. M. and Byers, S. W.(1994). Integrin α6β4 mediates dynamic interactions with laminin. J. CellSci. 107, 3153-3163.

Tuckwell, D. S., Reid, K. B., Barnes, M. J. and Humphries, M. J. (1996). TheA domain of integrin α2 binds specifically to a range of collagens but is noin general receptor for the collagenous motif. Eur. J. Biochem. 241, 732-739.

Uitto, J., McGrath, J. A., Pulkkinen, L. and Christiano, A. M. (1995).Molecular basis of the junctional forms of epidermolysis bullosa, a disordof the cutaneous basement membrane zone. Proc. 7th Int. Symp. BasementMembranes, 257-269.

Underwood, P. A., Bennett, F. A., Kirkpatrick, A., Bean, P. A. and Moss,B. A. (1995). Evidence for the location of a binding sequence for the α2β1integrin of endothelial cells, in the β1 subunit of laminin. Biochem. J. 309,765-771.

Vachon, P. H. and Beaulieu, J. F. (1995). Extracellular heterotrimeric lamininpromotes differentiation in human enterocytes. Am. J. Physiol. 268, G857-67.

Verrando, P., Hsi, B. L., Yeh, C. J., Pisani, A., Serieys, N. and Ortonne, J.P. (1987). Monoclonal antibody GB3, a new probe for the study of humabasement membranes and hemidesmosomes. Exp. Cell Res. 170, 116-128.

Vidal, F., Aberdam, D., Miquel, C., Christiano, A. M., Pulkkinen, L., Uitto,J., Ortonne, J. P. and Meneguzzi, G. (1995). Integrin beta 4 mutationsassociated with junctional epidermolysis bullosa with pyloric atresia. NatureGenet. 10, 229-234.

Wu, J. E. and Santoro, S. A. (1994). Complex patterns of expression suggesextensive roles for the α2β1 integrin in murine development. Dev. Dynam.199, 292-314.

Zweibaum, A., Pinto, M., Chevalier, G., Dussaulx, E., Triadou, N.,Lacroix, B., Haffen, K., Brun, J. L. and Rousset, M. (1985). Enterocyticdifferentiation of a subpopulation of the human colon tumor cell line HT29 selected for growth in sugar-free medium and its inhibition by glucosJ. Cell. Physiol. 122, 21-29.

Zweibaum, A., Laburthe, M., Grasset, E. and Louvard, D. (1991). The useof cultured cell lines in studies of intestinal cell differentiation and functionIn Handbook of Physiology: The Gastrointestinal System IV (ed. R. Frizelland H. Fields), pp. 223-255. Alan R. Liss Inc. New York.

Human colonic cancer cells synthesize and adhere to ... · the spatial/developmental distribution...

Documents

Transcript of Human colonic cancer cells synthesize and adhere to ... · the spatial/developmental distribution...