Apical 1 integrin in polarized MDCK cells mediates tubulocyst … · 2001-05-11 · Apical β1...

75
18Journal of Cell Science 109, 1875-1889 (1996)Printed in Great Britain © The Company of Biologists Limited 1996JCS9478
Apical β1 integrin in polarized MDCK cells mediates tubulocyst formation in

response to type I collagen overlay

Anna Zuk* and Karl S. Matlin

Renal Unit, Massachusetts General Hospital and Department of Medicine, Harvard Medical School, 149th 13th Street,Charlestown, MA 02129, USA

*Author for correspondence

A number of epithelia form tubulocysts in vitro whenoverlaid with type I collagen gel. Because collagenreceptors are generally believed to be expressed on thebasolateral domain, the mechanism by which collagenelicits this morphogenetic response from the apical surfaceis unclear. To investigate the role of β1 integrins, the majorreceptor family for collagen, in this process, we overlaidpolarized monolayers of MDCK II cells grown onpermeable supports with type I collagen gel and correlatedintegrin polarity with the polarity of other apical and baso-lateral membrane markers during tubulocyst formation.Polarized monolayers of one clone of MDCK II cells,referred to as Heidelberg MDCK, initially respond tocollagen overlay by stratifying; within 48 hours, lumenadevelop between the cell layers giving rise to tubulocysts.Tight junctions remain intact during tubulocyst formationbecause transepithelial electrical resistance does not signif-icantly change. Major alterations are observed, however,in the expression and localization of apical and basolateralmembrane markers. β1 integrins are necessary for tubulo-cyst morphogenesis because a function-blocking antibodyadministered to the apical pole of the cells completely

inhibits the formation of these structures. To determinehow apical-cell collagen interactions elicit tubulocystformation, we examined whether β1 integrins aremobilized to apical plasma membranes in response tocollagen overlay. We found that in the absence of collagen,polarized monolayers of Heidelberg MDCK cells endoge-nously express on apical plasma membranes a small poolof the β1 family, including α2β1 and α3β1. Collagenoverlay does not mobilize additional β1 integrins to apicaldomains. If β1 integrins are not already apically expressed,as in the C6 MDCK cell line (Schoenenberger et al. (1994)J. Cell Biol. 107, 527-541), β1 integrins are not directedapically and tubulocysts do not develop in response tocollagen. Thus, interaction of β1 integrins pre-existing onapical plasma membranes of polarized epithelia with typeI collagen gel is the mechanism by which apical applicationof collagen elicits the formation of tubulocysts. Depolarizedintegrins on apical plasma membranes of polarizedepithelia may be relevant to the pathogenesis of disease andinjury.

Key words: β1 Integrin, Epithelial polarity, Tubulocyst, MDCK

SUMMARY

INTRODUCTION

Epithelial cells form a selective permeability barrier separat-ing the outside world from internal compartments. Toperform this function, epithelial cells are polarized intodiscrete apical, lateral and basal domains. The apical plasmamembrane contains unique proteins and lipids not found onbasolateral plasma membranes (Simons and Fuller, 1985;Rodriguez-Boulan and Nelson, 1989). Specialized structuressuch as microvilli and cilia extend from the apical pole; theGolgi complex is positioned in the apical cytoplasm. Tightjunctions, zonula adherens and desmosomes are found onlateral borders that also contain cell adhesion molecules, suchas E-cadherin, that mediate interactions with neighboringcells. Basal plasma membranes contain integrin receptors(Ruoslahti, 1991) that interact with extracellular matrix(ECM) in the underlying basement membrane. Some

integrins may also mediate cell-cell adhesion (Larjava et al.,1990; Carter et al., 1990).

Studies on the biogenesis of epithelial polarity have focusedon the role of cell-cell and cell-substratum adhesion in estab-lishing the polarized epithelial phenotype (Nelson andVeshnock, 1986, 1987a,b; Ekblom, 1989; Nelson andHammerton, 1989; McNeil et al., 1990; Sorokin et al., 1990).In the polarized Madin-Darby canine kidney (MDCK) cell line,in vitro studies have demonstrated that interactions with thesubstratum are sufficient for establishing apical polarity(Rodriguez-Boulan et al., 1983; Vega-Salas et al., 1987). In theabsence of cell-cell contacts, MDCK cells polarize apicalplasma membrane glycoproteins in response to cell-substratumadhesion. Basolateral membrane proteins remain randomlydistributed on plasma membranes until cell-cell contacts form(Nelson and Veshnock, 1986, 1987a,b; Nelson andHammerton, 1989; McNeil et al., 1990). Studies on early

1876 A. Zuk and K. S. Matlin

kidney development also indicate that polarization requirescell-cell and cell-substratum adhesion. In kidney tubulogene-sis, the ureteric bud induces the metanephric mesenchyme tocondense by forming extensive cell-cell contacts. Conversionof the condensates into polarized tubular epithelia is latermediated by cell-substratum interactions with ECM (Sorokinet al., 1990).

A number of studies have addressed the role of cell-cellcontacts mediated by E-cadherin in epithelial polarization(Nelson and Veschnock, 1986, 1987a,b; Nelson andHammerton, 1989; McNeil et al., 1990); however, equivalentstudies on cell-substratum adhesion, particularly the role of theECM and its receptors, have been few. Cell-ECM interactionsinitiated by suspending MDCK cysts in type I collagen gel leadto membrane remodeling and polarity reversal (Wang et al.,1990; Ojakian and Schwimmer, 1994). A new basal surfacedevelops at the site of cell-collagen interaction and an apicalsurface forms at the opposite end lining an internal lumen.Further evidence that the ECM is critical in establishing thepolarized epithelial phenotype comes from studies on kidneydevelopment. Deposition of laminin A in the basementmembrane surrounding the condensed mesenchyme stimulatesthe conversion of the aggregated cells into polarized epithelia(Sorokin et al., 1990).

Epithelial interactions with ECM are mediated by theintegrins, a superfamily of cell surface receptors found onevery eucaryotic cell except erythrocytes (for reviews, seeHynes, 1992; Hemler et al., 1995). Integrins are transmem-brane glycoproteins that consist of unrelated α and β subunits.Their association into noncovalent heterodimers is necessaryfor cell surface expression and ligand binding. Most epitheliaexpress integrins belonging to the β1, β3, β4 or β5 families.The β chains form complexes with a variety of different αsubunits, yielding receptors for collagens, laminin, fibronectinand vitronectin. In addition, some αβ combinations recognizemore than one ECM ligand. For example, the α1β1 and α2β1are receptors for collagen or laminin whereas the α3β1 canrecognize collagen, laminin, or fibronectin (Hynes, 1992;Hemler et al., 1995). Receptor-ligand binding activates trans-membrane signaling, influencing cell differentiation, behaviorand gene expression (for reviews, see Damsky and Werb,1992; Juliano and Haskill, 1993; Clark and Brugge, 1995).

By mediating cell-substratum interactions, the integrinshave been implicated in the pathogenesis of disease and injury.In carcinogenesis, novel integrin-substratum interactions influ-encing epithelial differentiation and polarization (Schoenen-berger et al., 1994) have been suggested to contribute to metas-tasis (Plantefarber and Hynes, 1989; Schoenenberger andMatlin, 1991; Matlin and Caplan, 1992). New integrin-ECMinteractions are established after epithelial injury and facilitatewound healing and repair (for reviews, see Gailit and Clark,1994; Clark, 1995). In in vitro models of acute renal failure,integrin-substratum interactions are altered (Gailit et al., 1993;Goligorsky et al., 1993); the α3β1 integrin redistributes frombasal to apical cell surfaces and some integrins are lost fromepithelial plasma membranes (Gailit et al., 1993).

We have been investigating integrin-mediated cell-substra-tum interactions in epithelial cell polarization using the MDCKcell line as a model system. We have been particularly inter-ested in the role of the ECM in generating and maintainingdiscrete apical and basolateral domains. MDCK cells express

the β1, β3, and β4 integrin families, including the α2β1, α3β1,an unidentified αxβ1, α6β4 and αvβ3 (Schoenenberger et al.,1994; Ojakian and Schwimmer, 1994). The β1, β3, and β4families are present on basolateral plasma membranes. Onlythe β1 family mediates cell attachment to collagen types I andIV and laminin (Schoenenberger et al., 1994; Ojakian andSchwimmer, 1994).

To further study the role of β1 integrins in epithelial polar-ization, we took advantage of a classic observation in whichepithelia, including thyroid, mammary gland, intestine, liverand vascular endothelia, form tubulocysts when overlaid withtype I collagen gel (Chambard et al., 1981; Hall et al., 1982;Montesano, 1983; Montgomery, 1986; LeCluyse et al., 1994).Because β1 integrins are the major receptor family forcollagens (Hynes, 1992; Hemler et al., 1995) found on baso-lateral cell surfaces (Ruoslahti, 1991), the mechanism of howcollagen at apical plasma membranes elicits this morpho-genetic response was unclear. We hypothesized that apicalcell-collagen interactions would mobilize β1 integrins to apicalplasma membranes to mediate the formation of tubulocysts.Using the polarized MDCK cell line, we found to our surprisethat in the absence of collagen overlay, β1 integrins wereendogenously expressed at low levels on apical plasmamembranes in one clone of MDCK, even though other apicaland basolateral membrane markers localized exclusively totheir respective plasma membrane domains. Moreover, apicalapplication of collagen did not mobilize additional β1 integrinsto apical cell surfaces. Pre-existing apical β1 integrinsmediated tubulocyst formation because if receptor-ligandbinding was inhibited or if β1 integrins were absent from thissurface tubulocysts did not develop in response to collagen.

MATERIALS AND METHODS

Cell cultureMDCK II cells were originally cloned by Louvard (1980) for highgrowth rate and dome formation; this clone is referred to as Heidel-berg (HD) MDCK cells to identify its place of origin. HD MDCKcells have been used extensively for studies on epithelial cell polarity(Matlin et al., 1981; Matlin and Simons, 1983; for review, Simonsand Fuller, 1985). The C6 subclone of HD MDCK (MDCK pMV76.2.1.) was used for some experiments. C6 cells are stably transfectedwith the empty pMV7 expression vector (Schoenenberger et al.,1991). They are identical to the parental line with regard to ultra-structure, growth rate and transepithelial electrical resistance.

For routine culture, HD (passages 7-33) and C6 (passages 7-15)MDCK cells were grown in 75 cm2 flasks (Falcon; Becton Dickinson& Co., Boston, MA) in minimal essential medium (MEM) orDulbecco’s MEM (DME; Mediatech, Herndon, VA) containing 5%fetal bovine serum (FBS; Sigma Chemical Co., St Louis, MO orHyClone Laboratories, Inc., Logan, UT), 2 mM L-glutamine (Sigma)and 10 mM Hepes (Sigma), pH 7.4, at 37°C in 95% air, 5% CO2.Confluent cells were harvested with trypsin-EDTA (Gibco BRL, LifeTechnologies, Inc., Grand Island, NY) and subcultured at a 1:5dilution twice weekly.

For culture of cells on permeable polycarbonate supports, cellswere plated at 7.8×105 cells on Transwell filters (24 mm, 0.4 µm poresize; Corning CoStar, Cambridge, MA) pre-wetted with medium.Cultures were fed every other day for 6-8 days as well as the daybefore collagen overlay. Antibiotic-antimycotic (Gibco) was includedin the medium at a final concentration of 100 units/ml penicillin, 100µg/ml streptomycin and 0.25 µg/ml amphotericin B.

1877Apical β1 integrin in polarized MDCK cells

AntibodiesAnti-integrin antibodiesRabbit anti-human β1 integrin antibody was obtained from Dr E.Ruoslahti (La Jolla Cancer Research Foundation, La Jolla, CA). Thisantibody was generated against a 38mer peptide corresponding to thecarboxy-terminal domain (Giancotti and Ruoslahti, 1990) of thehuman β1 integrin. Under non-denaturing conditions, this antibody andother polyclonal β1 integrin antibodies (Schoenenberger et al., 1994)immunoprecipitate α2β1, α3β1, an unidentified αxβ1, and precursorβ1 from MDCK cell extracts (Schoenenberger et al., 1994). Underdenaturing conditions, this antibody and other polyclonal β1 integrinantibodies immunoprecipitate only two polypeptides corresponding tothe β1 integrin subunit and its precursor (Schoenenberger et al., 1994).

The function-blocking rat monoclonal anti-human β1 integrinantibody, AIIB2 (Hall et al., 1990), was obtained from Dr C. Damsky(University of California at San Francisco, San Francisco, CA). Thisantibody immunoprecipitates only mature β1 and its associated αsubunits in MDCK cells; it does not recognize the β1 precursor ofMDCK (A. Zuk, unpublished observation).

Antibodies to apical and basolateral membrane markersHybridomas secreting mouse monoclonal antibody 3F21D8 recog-nizing the apical membrane glycoprotein, gp135, were obtained fromDr G. Ojakian (SUNY, Brooklyn, NY; Ojakian and Schwimmer,1988). Mouse monoclonal antibody 6.23.3 was generated against dogintestinal mucosa (Balcarova-Ständer et al., 1984). This antibody rec-ognizes a basolateral membrane protein of 58 kDa, referred to as p58.Mouse monoclonal antibody rr1, generated against strain I MDCKcells, recognizes the basolateral membrane protein E-cadherin(Gumbiner and Simons, 1986). Culture supernatant and the antibody-secreting hybridoma were obtained from Dr B. Gumbiner (Sloan-Kettering, New York, NY).

Collagen overlayRat tail tendon type I collagen (Upstate Biotechnology Inc., LakePlacid, NY) in 0.02 N acetic acid was dialyzed against 1/10th Ham’sF-12 (Gibco), pH 4.0, at 0°C. Hydrated collagen gels were cast as pre-viously described (Zuk et al., 1989) by mixing on ice 7 ml of dialyzedcollagen (3 mg/ml), 1 ml 10× F-12 (101 mg/ml), 1 ml 10× sodiumbicarbonate (11.76 mg/ml; Gibco) and 1 ml FBS.

In preparation for collagen overlay, MDCK cells grown to conflu-ency on permeable supports were rinsed once with growth mediumprepared as described earlier. Medium (1.5 ml) was added to the basalcompartment; the apical compartment was overlaid with 400 µlcollagen solution after rinsing once with the same volume of collagenmixture. Within 20 minutes at 37°C, the collagen polymerized into ahydrated gel covering MDCK apical surfaces. In order to facilitatecontact of apical cell surfaces with the polymerized type I collagengel and to avoid detachment of the collagen overlay from this cellsurface, additional medium was not added to the apical compartmentof permeable supports overlaid with collagen. Control filters that werenot overlaid with collagen had 400 µl medium added to the apicalcompartment. Cultures were incubated for up to 5 days. The volumeof collagen and/or medium added to the apical or basal compartmentswas the minimal amount needed for adequate incubation of Transwellsupports for the duration of the experiment.

For some experiments, antibodies or reagents were mixed withcollagen at the time of overlay and also added basally. Cultures werefed from the basal compartment with fresh antibody or reagent 24hours later. Antibodies included AIIB2; reagents included mitomycinC (Sigma), an inhibitor of cell proliferation. Heat inactivated FBS wasused for antibody inhibition studies.

Metabolic labeling and immunoprecipitationIntegrins were radiolabeled to steady-state prior to collagen overlay.Polarized monolayers of HD MDCK cells grown on Transwell filters

for 6-8 days were incubated from the basal compartment with 100µCi/ml [35S]methionine [35S]cysteine (EXPRE35S35S, New EnglandNuclear, Boston, MA) in labeling medium (1/10 methionine-freeMEM or DME containing FBS, glutamine, Hepes, andantibiotic/antimycotic) for 16-24 hours. Apical compartmentscontained labeling medium without isotope. After steady-statelabeling, filters were prepared for collagen overlay; medium wasremoved only from the apical compartment and apical cell surfacesrinsed once with labeling medium and once with collagen solution;basal medium containing radiolabel was not removed. Apical cellsurfaces were then overlaid with 400 µl collagen solution. Filters werefurther incubated for up to 12 hours before processing for differentialsurface biotinylation (Schoenenberger et al., 1994) and immunopre-cipitation of integrins.

To examine the effects of collagen overlay on newly synthesizedintegrins, cells were pulse-labeled after collagen overlay. Pulse-chaseexperiments were done 12 hours after collagen overlay when celllayering and lumen formation are negligible (Fig. 2B). MDCK cellswere rinsed three times, 10 minutes each at 37°C with methionine-free medium containing Hepes, glutamine, and 0.35g/l sodium bicar-bonate (pulse-medium). MDCK cells were then pulsed for 10 minutesat 37°C from the basal cell surface with 100 µl/filter of 100 µCi[35S]methionine [35S]cysteine; 500 µl pulse-labeling medium withoutisotope was added apically. Radiolabeled proteins were then chasedfor up to 90 minutes at 37°C in MEM containing 10× the normal con-centration of unlabeled methionine. MDCK integrins reach the cellsurface within 60 minutes chase (G. M. Zinkl and K. S. Matlin, unpub-lished observation). After 90 minutes chase, integrins were differen-tially surface biotinylated and immunoprecipitated.

For surface biotinylation and immunoprecipitation, MDCK cells onice were rinsed three times with PBS containing 1 mM CaCl2 and 0.5mM MgCl2 (PBS+) without disturbing the collagen overlay. MDCKcells were then differentially biotinylated twice, 15 minutes each, fromthe apical or basal compartment or both with 0.5 mg/ml sulfo-NHS-biotin (Pierce Chemical Co., Rockford, IL) dissolved in TEA buffer(10 mM triethanolamine, 125 mM NaCl, 2 mM CaCl2, 0.5 mM MgCl2,pH 9.0; Gottardi and Caplan, 1992). Compartments that were notbiotinylated received only TEA buffer. Filters were then rinsed oncein serum-free medium without mixing of apical and basal washes;filters were then extensively rinsed with PBS+ before extraction.

Transwell filters were removed from their supports, cut in half andextracted in microfuge tubes (1.5 ml) containing extraction buffer (1%NP-40, 1% sodium deoxycholate (added fresh), 0.5% SDS, 10 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 10 mM glycine, 10 µg/ml aprotinin, 0.2mg/ml iodoacetamide, 1 µg/ml antipain, 17.5 µg/ml benzamidine, 1 µgpepstatin, 10 mM glycine) with rocking at 0°C for 30 minutes. Filterswere removed by centrifugation and cell extracts in the supernatant werefiltered through 0.45 µm filter units (Ultrafree-MC Filters; MilliporeCorporation, Bedford, MA) to remove cell debris. For immunoprecipi-tation, extracts were incubated overnight with rocking at 0°C withprimary antibody; Protein A Trisacryl (Pierce) was then added andsamples incubated further for a minimum of 2 hours. After washing threetimes with rocking in extraction buffer without protease inhibitors andonce with 10 mM Tris, pH 8.6, to remove salts, antigen-antibodycomplexes were released from Protein A-Trisacryl beads by heating at95°C for 3 minutes in 30 µl of 2% SDS, 20 mM Tris-HCl, pH 8.6. Asa control for 35S incorporation, 5 µl of each sample was prepared forSDS-PAGE. The remaining 25 µl were diluted tenfold with 1% TritonX-100, 20 mM Tris-HCl, pH 8.6, 150 mM NaCl. Streptavidin-agarosebeads (Pierce) were then added and samples inverted at 4°C overnight.Biotinylated integrins were released from agarose beads after washingby heating at 95°C for 3 minutes in electrophoresis sample buffer (200mM Tris-HCl, pH 8.8, 5 mM EDTA, 15% sucrose, 0.1% bromophenolblue, 20 mM dithiothreitol, and 2% SDS). Samples were alkylated with0.5 M iodoacetamide for 15 minutes at 37°C in the dark.

Immunoprecipitated polypeptides were resolved on 6% SDS-PAGE gels (Laemmli, 1970). Gels were fixed, impregnated with


Fig. 1. Autoradiogram of SDS-PAGE (6%, reduced) gels analyzingsurface expression of gp135 and E-cadherin in HD MDCK cells labeledto steady-state with [35S]methionine [35S]cysteine. Surface biotinylatedproteins were immunoprecipitated with mouse monoclonal antibodydirected against gp135 (A, arrow) or E-cadherin (B, arrow). HDMDCK cells express gp135 only on apical (lane 1), but not basolateralplasma membranes (lane 2), whereas E-cadherin (B) is found only onbasolateral (lane 4) but not apical (lane 3) cell surfaces.

EN3HANCE (E. I. DuPont deNemours & Co., Boston, MA), driedand exposed to pre-flashed KODAK X-OMAT film (Eastman KodakCo., Rochester, NY).

Light microscopyCultures were fixed in half-strength Karnovsky fixative (2.5% glu-taraldehyde, 2% paraformaldehyde in 0.1 M sodium cacodylatebuffer, pH 7.4) for 1 hour, osmicated (1% OsO4), and stained en blocwith 1% uranyl acetate, as previously described (Zuk et al., 1989).Filters were dehydrated in a graded series of ethanols, cut from thefilter holders leaving the collagen overlay intact, and infiltrated andembedded in Epon-Araldite containing 2% DMP-30.

Transverse sections (0.5-1 µm) were cut using a Reichert-JungUltracut microtome. Sections were stained with 1% toluidine blue (forlight microscopy) or 0.1% toluidine blue (for immunoperoxidasestained filters) in 1% sodium borate and photographed with T-MAXASA 100 film (light microscopy; Kodak) or Technical Pan Film ASA100 (peroxidase immunohistochemistry; Kodak).

ImmunohistochemistryApical and basolateral membrane markers were immunolocalized inwhole-mount collagen overlaid cultures using a modification of a per-oxidase-antiperoxidase technique previously described (Zuk and Hay,1994). This procedure has the advantage of localizing with high res-olution the stable peroxidase product in plastic-embedded sampleswithout the use of frozen sections (0.5-5 µm) of cells on filters and/orconfocal microscopy.

For peroxidase immunohistochemistry, an entire filter of MDCK cellsoverlaid with collagen was rinsed in PBS+ and fixed in 2%paraformaldehyde, 75 mM L-lysine, 10 mM sodium m-periodate (PLP;McLean and Nakane, 1974) for 1 hour. After further rinsing, endoge-nous peroxidase activity was quenched with 1% H2O2/PBS or 0.1%phenylhydrazine/PBS for 10 minutes; cells were rinsed with PBS andthen permeabilized with 0.1% Triton X-100/PBS for 4 minutes. Non-specific binding sites were blocked by incubating in 10% normal goatserum (NGS) for 45 minutes-1 hour. For incubation in antibodies, filterholders were removed from 6-well plates and positioned basal side downon 150 µl of antibody pipetted on Parafilm in a humidified chamber; theapical compartment was covered with an additional 150 µl antibody.Filters were incubated overnight at 4°C in primary antibody. Filterholders were returned to 6-well plates, rinsed and then incubated asdescribed above in link antibody (goat anti-mouse; Sigma) diluted 1:20in 10% NGS for 1 hour at room temperature. After further rinsing, filterswere incubated in peroxidase-antiperoxidase (PAP) mouse antibody(Sigma) diluted 1:100 in 10% NGS for 1 hour at room temperature.Filters were washed extensively and PAP complexes immunolocalizedby reacting with a freshly prepared solution of 0.05% diaminobenzidine(Sigma) dissolved in 0.01% H2O2 in 50 mM Tris-HCl, pH 7.5, for 10-15 minutes at room temperature. The immunostained filters were thenwashed and processed for light microscopy as described earlier exceptthat osmication and en bloc staining were omitted.

Transepithelial resistanceTransepithelial electrical resistance was measured in HD MDCK cellsgrown on permeable supports with or without collagen overlay usingan Epithelial Voltohmmeter (World Precision Instruments, NewHaven, CT). Warm (37°C) medium was added to the apical (1.5 ml)and basal (1 ml) compartments. Filters were allowed to equilibrate toroom temperature for 20 minutes. Electrical resistance was measuredaccording to the manufacturer’s directions. Values were calculatedafter subtracting the background contributed by blank filters plus orminus collagen overlay, as appropriate.

Thymidine incorporationCultures were labeled with [3H]thymidine (New England Nuclear) inorder to measure cellular proliferation. For these experiments, HDMDCK cells either at the time of overlay (0 hour) or at subsequent

12 hour intervals up to 48 hours were labeled from the basal com-partment with l µCi/ml [3H]thymidine in DME containing FBS,Hepes, glutamine and antibiotic-antimycotic. Twelve hours later,[3H]thymidine incorporation was assayed. Filters were rinsed on icethree times with PBS+, 4 minutes each, without disturbing thecollagen overlay. Filters were cut away from their supports withcollagen intact and proteins precipitated on ice for a minimum of 2hours in 1 ml 10% trichloroacetic acid (TCA). TCA was removed byaspiration, and samples rinsed twice with 1 ml fresh ice-cold 5% TCA.Samples were solubilized in 500 µl 0.2 M NaOH, 0.l% SDS for 1hour at 37°C and then neutralized with an equal volume of 2 M Tris,pH 6.8. Radioactivity was measured by liquid scintillation countingfrom equal aliquots of samples. Counts per minute/sample were cal-culated by subtracting background counts (filter, with or withoutcollagen). Significance (P≤0.05) was determined by the one-wayanalysis of variance and Student’s t-test.

RESULTS

HD MDCK cells overlaid with type I collagen gelform tubulocystsHD MDCK cells were cultured to confluency on Transwellfilters and then metabolically labeled to steady-state with[35S]methionine [35S]cysteine. Apical and/or basolateral com-partments were differentially cell surface biotinylated and celllysates immunoprecipitated with antibody recognizing theapical membrane glycoprotein, gp135 (mouse monoclonalantibody 3F21D8; Ojakian and Schwimmer, 1988) or the baso-lateral membrane marker, E-cadherin (mouse monoclonalantibody rr1; Gumbiner and Simons, 1986). Autoradiogramsof SDS-PAGE gels reveal that HD MDCK cells polarize apicaland basolateral membrane markers to their respective plasmamembrane domains. GP135 localizes exclusively to apicalplasma membranes (lane 1, Fig. 1) and is not expressed baso-


laterally (lane 2, Fig. 1), whereas E-cadherin is found only onbasolateral cell surfaces (lane 4, Fig. 1) and is not detectedapically (lane 3, Fig. 1).

To induce formation of tubulocysts, polarized monolayersof HD MDCK cells on Transwell filters were overlaid from theapical surface with a solution of type I collagen that polymer-izes into a hydrated gel with incubation. At the time of overlay(0 hours, Fig. 2A), HD MDCK cells are cuboidal shaped andmorphologically polarized along an apical to basal axis (alsoFigs 4A, 5A). Monolayer morphology is uniform andorganized at this time. Collagen fibrils within the polymerizedgel contact apical epithelial surfaces (Fig. 2B). Twelve hourslater (Fig. 2B), monolayer morphology is relativelyunchanged; in a few areas, however, HD MDCK cells begin tolayer (data not shown). Cell layering is more extensive at 24hours (Fig. 2C), frequently becoming 2-4 cells thick.Monolayer morphology is disrupted in these areas as theepithelium reorganizes. Changes in cell shape are noted,ranging from cuboidal to squamous. As a second cell layerforms that adheres to the overlaid collagen (Fig. 2C), cell-cellcontacts are retained within the original monolayer, theforming bilayer and regions undergoing epithelial rearrange-ment. The forming bilayer subsequently attaches to thecollagen gel and populates this surface (Fig. 2C,D,E). Lumenathen develop between cells attached to the filter and thoseattached to collagen. Morphogenesis of HD MDCK cells intoa bilayer of cells separated by a lumen, i.e. tubulocyst, iscomplete at 48 hours (Fig. 2D,E) and extends across the filter.HD MDCK cells populating the gel surface and formingtubules reacquire cuboidal cell shape (Fig. 2D).

Cell proliferation accompanies tubulocyst formation (Fig.3). Measurement of [3H]thymidine incorporation in HDMDCK cells at 12 hour intervals beginning at the time ofoverlay and extending up to 48 hours reveals no significantchange in [3H]thymidine incorporation within the first 12 hoursof culture in the absence or presence of collagen. Proliferationincreases at subsequent time points in collagen overlaidcultures. The most significant increase (P≤0.05) is measured24-36 and 36-48 hours after overlay, when MDCK cellspopulate the gel surface and form lumens. Mitomycin C (1-2µg/ml), an inhibitor of cell proliferation, prevents formation ofa complete bilayer; lumena are not well developed and tubu-locyst formation is incomplete (data not shown). This suggeststhat cell division is not required for rearrangement of the

Fig. 2. Light micrographs of transverse sections (0.5 µm) of HDMDCK cells overlaid from the apical surface with rat tail tendontype I collagen gel. Sections were stained with 1% toluidine blue. At0 hours (A), confluent HD MDCK cells grown on Transwell filtersare overlaid from the apical surface with type I collagen gel (seeMaterials and Methods). At the time of overlay, HD MDCK cells arecuboidal shaped and apically-basally polarized (also, Fig. 1). Twelvehours (B) later, monolayer morphology remains relativelyunchanged. Collagen fibrils within the gel contact apical epithelialsurfaces. At 24 hours (C), cell layering is extensive and monolayermorphology is altered in these areas. Note the heterogeneous cellshapes found throughout the re-organizing epithelium. Cells thatappear between the monolayer and collagen gel are beginning toattach to, and layer against, the collagen; free surfaces are developingbetween the cell layers. At 48 hours (D,E), a bilayer of cellsseparated by a lumen extends across the filter; lumena are welldeveloped in some areas. Bar, 10 µm.

1880 A. Zuk and K. S. Matlin IN

CO

RP

OR

AT

ION

(cp

m s

.d.)

+ -

HOURS AFTER OVERLAY

H-T

HY

MID

INE

3

0

2000

4000

6000

8000

10000

0-12 hrs 12-24 hrs 24-36 hrs 36-48 hrs

Fig. 3. [3H]Thymidine incorporation in HD MDCK cells with (blackbars) or without (gray bars) collagen overlay. Filters were incubatedfrom the basal cell surface with 1 µCi [3H]thymidine for 12 hourintervals starting at the time points indicated. No significant changein cell proliferation is measured at 0-12 hours after collagen overlaywhen values are compared to non-collagen controls. The mostsignificant increase (P≤0.05) is measured 24-36 and 36-48 hoursafter overlay.

monolayer into a bilayer but that proliferation of cells in theincipient bilayer is required for complete formation of tubulo-cysts.

Transepithelial electrical resistance does not change duringtubulocyst formation (Table 1). Transepithelial electrical resis-tance was measured at 12 hour intervals beginning at the timeof overlay (0 hour) and ending 48 hours later. No loss of elec-trical resistance was detected during tubulocyst morphogen-esis when values were compared to non-collagen overlaidcontrols.

Apical and basolateral membrane proteinsredistribute in tubulocyst formationFrom the dramatic reorganization of the MDCK cell epi-thelium during tubulocyst formation, it seemed likely thatapical and basolateral plasma membrane proteins would alsoundergo parallel rearrangements. To examine this, animmunoperoxidase technique was used to follow the localiza-tion of gp135, an apical membrane glycoprotein (Ojakian andSchwimmer, 1988), and p58, a basolateral membrane marker(Balcarova-Ständer et al., 1984). Although we detected E-cadherin on basolateral cell surfaces both biochemically (Fig.1) and by immunofluorescence microscopy (data not shown),it was difficult to consistently detect this antigen with peroxi-dase immunohistochemistry using either monoclonal or poly-

Table 1. Transepithelial electrical resistance (ohms · cm2)of HD MDCK cells

12 hours 24 hours 36 hours 48 hours

With collagen overlay 248.3±43.8 207.9±28.0 196.2±3.2 204.5±19.3No overlay control 260.4±17.0 226.8±23.3 232.8±2.0 231.0±10.6

Data represents mean ± s.e.m. of triplicate experiments, three sampleseach.

clonal antibodies. GP135 and p58 were thus localized incollagen overlaid cultures fixed at 12 hour intervals starting atthe time of overlay.

Light microscopy of PAP immunostained sections revealsthat gp135 localizes exclusively to apical plasma membranesof confluent HD MDCK cells at the time of overlay (Fig. 4A),confirming our biochemical results (Fig. 1, lane 1). However,during tubulocyst formation, gp135 expression and localiza-tion changes. During early stages of cell layering, gp135 is lostfrom plasma membranes. It is frequently absent from regionsof cell-collagen interaction (arrows, Fig. 4B,C) as attachmentsdevelop to the overlaid collagen gel. In a few areas (openarrow, Fig. 4B), however, gp135 persists at the site of cell-collagen contacts. When lumena develop (*, Fig. 4C,D) gp135immediately localizes to apical plasma membranes of the newcell layer attached to collagen and to apical cell surfaces ofMDCK cells attached to the filter. We did not detect internalvacuolar staining in tubular morphogenesis at any time pointwe examined.

Distribution of p58, a basolateral membrane marker, wasexamined next. Initially, p58 localizes only to basolateral andnot to apical cell surfaces of polarized HD MDCK cells (Fig.5A), confirming our biochemical results with the basolateralmarker, E-cadherin (Fig. 1, lane 4). During cell layering, whenexpression of the apical marker gp135 is reduced at points ofcell-collagen interaction, p58 is expressed over the entireplasma membrane at regions of cell-cell, cell-substratum, andcell-collagen interactions (Fig. 5B,C). As the epitheliumpopulates the collagen gel surface and lumena develop betweenthe cell layers, p58 reacquires a basolateral distribution and islost from forming apical (i.e. free) surfaces (Fig. 5C,D).

Interestingly, we found that localization of apical and baso-lateral membrane markers do not change during mitosis.Whatever their distribution is prior to cell division, the samedistribution is maintained during mitosis, confirming previ-ously published results (Reinsch and Karsenti, 1994). Whengp135 expression is downregulated during cell layering, itremains absent from plasma membranes during cell division(data not shown). When gp135 localizes to apical plasmamembranes with the formation of free lumenal surfaces, itretains this apical location during subsequent mitoses that areresponsible for populating the gel surface (data not shown).The basolateral protein, p58, which is randomly distributed onplasma membranes during cell layering, remains randomizedduring cell division. When basolaterally segregated in thepresence of a free surface, it persists on basolateral plasmamembranes as cells within the bilayer divide (Fig. 5D).

Overall, these observations indicate that collagen contactwith the apical surface of polarized HD MDCK cells leads totubulocyst formation within 48 hours. This reorganization isaccomplished without significant changes in transepithelialelectrical resistance but with major changes in the distributionand expression of apical and basolateral membrane proteins.

The β1 integrin family on apical surfaces of HDMDCK cells mediates tubulocyst formationBecause β1 integrins are the major receptor family forcollagen (Hynes, 1992; Hemler et al., 1995), we next deter-mined whether they initiate tubulocyst formation by polarizedHD MDCK cells overlaid with collagen gel. The function-blocking anti-β1 integrin antibody, AIIB2, which inhibits


Fig. 4. Transverse sections (0.5 µm) of HD MDCK cells cultured onTranswell filters and overlaid from the apical surface with type Icollagen gel (coll). Sections were photographed after PAPimmunolocalization of gp135 with monoclonal antibody 3F21D8.Sections were stained with 0.1% toluidine blue. At 0 hour (A),confluent monolayers of HD MDCK cells grown on permeablesupports are overlaid from the apical surface with type I collagen gel.Only the apical surface exhibits strong staining of gp135 at this time.After 12 hours of overlay (B) when the epithelium begins to layer,surface staining of gp135 decreases at regions of cell-collageninteraction. In some areas, gp135 is lost (filled arrow) whereas inothers it persists (open arrow). Some regions of the monolayercontinue to localize gp135 to apical plasma membranes. As theforming bilayer attaches to collagen (B,C), gp135 remains absent inregions of cell-collagen interactions (arrows). With the formation offree lumenal surfaces (* C,D), gp135 localizes to apical plasmamembranes of HD MDCK cells attached to the filter or collagen gel.(C,D) Represent 36 hours after overlay. Bar, 10 µm.

Fig. 5. Transverse sections (0.5 µm) of HD MDCK cells cultured onTranswell filters and overlaid from the apical surface with type Icollagen gel (coll). Sections were photographed after PAPimmunolocalization of p58, a basolateral membrane marker. Sectionswere stained with 0.1% toluidine blue. A control section (A) showsthat p58 localizes to basolateral cell surfaces of polarized HDMDCK cells. As the epithelium layers in response to collagenoverlay and attaches to it (B,C), p58 randomizes to all cell surfaces.It localizes to surfaces in contact with collagen, as well to regions ofcell-cell and/or cell-filter interactions. In the presence of a freesurface (i.e. lumen, C,D), p58 localizes basolaterally. Note thebasolateral location of p58 in the mitotic cell (D) of the bilayerpopulating the gel surface. (B) Represents 12 hours after overlay;(C,D) 36 hours after overlay. Bar, 10 µm.


MDCK cell attachment to collagens I and IV and laminin(Schoenenberger et al., 1994), was mixed with collagen priorto overlay of apical cell surfaces. We found that anti-β1integrin antibody significantly inhibits layering and lumenformation. Polarized HD MDCK cells do not respond to apicalcollagen by reorganizing into tubulocysts. In fact, monolayermorphology does not change (Fig. 6A). We frequentlyobserved during processing for light microscopy that collagengels are not attached to apical plasma membranes. In contrast,the control antibody, AD7, which recognizes a Golgi-associ-ated protein (Narula et al., 1992) has no affect on tubulocystformation (Fig. 6B).

Previous results indicated that in polarized MDCK cells β1integrins localize only to basolateral and not to apical cellsurfaces (Schoenenberger et al., 1994) and that they aretargeted directly to basolateral plasma membranes (Boll et al.,1991; G. M. Zinkl and K. S. Matlin, unpublished observation).Thus, the observation that an anti-β1 integrin antibody addedto collagen at the time of overlay inhibits tubulocyst formationraised the question of how β1 integrins that are normally baso-lateral can exert an effect at apical plasma membranes. Wehypothesized that apical application of type I collagen gelmobilizes β1 integrins to apical plasma membranes to mediatethe formation of tubulocysts. In response to collagen overlay,β1 integrins would be redistributed from basolateral to apicalcell surfaces and/or newly synthesized integrins would bedelivered to apical plasma membranes. Even though we did notdetect the β1 family on apical surfaces of polarized MDCKcells (Schoenenberger et al., 1994), we also considered the pos-sibility that low levels are endogenously expressed on apicalcell surfaces that were previously undetected.

Fig. 6. Light micrographs of transverse sections (0.5 µm) of HDMDCK cells cultured on Transwell filters and overlaid from theapical surface with type I collagen gel. Cells were incubated in thepresence of antibody for 48 hours. Sections were stained with 1%toluidine blue. The rat monoclonal anti-human β1 integrinantibody, AIIB2, which is function-blocking, inhibits layering andlumen formation of HD MDCK in response to apical collagenoverlay (A) whereas the control antibody AD7 (B) does not. In thepresence of anti-β1 antibody (A), the collagen gel does not attachto apical cell surfaces; the polarized epithelial phenotype ismaintained. Hybridoma culture supernatants were used at a l:5dilution. Bar, 10 µm.

The redistribution of β1 integrins from basolateral to apicalplasma membranes in response to collagen overlay was firstexamined. For these experiments, HD MDCK cells were meta-bolically labeled to steady-state with [35S]methionine[35S]cysteine prior to collagen overlay. The apical compart-ment was then washed, overlaid with collagen and incubatedfurther for up to 12 hours prior to differential surface biotin-ylation of apical and/or basolateral compartments andimmunoprecipitation of β1 integrins. Under these conditions,any integrin detected on cell surfaces would be synthesizedprior to collagen overlay and any redistribution from basolat-eral to apical plasma membranes would be due to apical cell-collagen interactions. The cultures were processed up to the 12hour timepoint when layering and lumen formation are sparse(see Fig. 2B).

We found that in the absence of collagen overlay, polyclonalanti-β1 integrin antibody (Giancotti and Ruoslahti, 1990)immunoprecipitates α2β1 and α3β1 integrins from basolateralplasma membranes at the 0, 6 and 12 hour timepoints undernondenaturing conditions (lanes 2, 6, 10, Fig. 7). Surprisingly,the same matrix receptors are also detected apically (lanes 1,5, 9, Fig. 7), although in amounts much less than those seenbasolaterally. Levels of the β1 family at all times are signifi-cantly greater on basolateral cell surfaces (lanes 2, 6, 10, Fig.7) than on apical cell surfaces (lanes 1, 5, 9, Fig. 7). Six and12 hours after overlay, apical collagen does not increase theamount of β1 integrins on apical plasma membranes (lanes 7,

Fig. 7. Autoradiograms of SDS-PAGE (6%, reduced) gels analyzingsurface expression of β1 integrins in HD MDCK cells labeled tosteady-state with [35S]methionine [35S]cysteine and overlaid withcollagen gel. Samples were processed at 0, 6 and 12 hours afteroverlay. Surface biotinylated integrins were immunoprecipiated withpolyclonal anti-β1 integrin antibody under non-denaturingconditions; immunoprecipitates were resolubilized and reprecipitatedwith streptavidin-agarose to capture surface biotinylated proteins.HD MDCK cells express the β1 integrin family, including α2β1 andα3β1, on basolateral (b) cell surfaces (lanes 2, 4, 6, 8, 10, 12) in theabsence (lanes 2, 6, 10) or presence (lanes 4, 8, 12) of collagenoverlay. Interestingly, HD MDCK cells express a small pool of α2β1and α3β1 on apical (a) plasma membranes with (lanes 3, 7, 11) orwithout (lanes 1, 5, 9) collagen overlay. There is no increase in levelsof apical β1 integrins with collagen overlay (compare lanes 1 vs 3;lanes 5 vs 7; lanes 9 vs 11). Identification of integrin α subunits wasbased on mobility differences observed by SDS-PAGE afterimmunoprecipitation with subunit specific antibodies(Schoenenberger et al., 1994). The band indicated by the arrowheadtentatively identifies the α1 subunit.


Fig. 8. Autoradiograms of SDS-PAGE (6%, reduced) gels analyzingpulse-chase labeling of β1 integrins in HD MDCK cells twelve hoursafter collagen overlay. Cells were pulsed for 10 minutes and chasedfor 90 minutes. HD MDCK cells were then differentially surfacebiotinylated from either the apical (a), basolateral (b) or both (ab)cell surfaces and immunoprecipitated with polyclonal anti-β1antibody under non-denaturing conditions. (A) Surface expression ofβ1 integrins. At 0 minutes chase, integrins are not detected onMDCK cell surfaces with (lane 3) or without (lane 2) collagenoverlay. After 90 minutes chase, the β1 integrin family, including theα2β1 and α3β1, are detected on basal cell surfaces in the absence orpresence of collagen (lanes 7, 9, respectively). In the absence ofcollagen (lane 6), β1 integrins, primarily the α2β1, are detected onapical cell surfaces. Collagen overlay does not mobilize additionalnewly synthesized integrins to apical plasma membranes (lane 8).Sorting of the α2β1 appears to be less stringent than the α3β1explaining its predominance on apical plasma membranes.(B) Immunoprecipitable antigen of the respective samples shown in(A). Aliquots (5 µl) were taken after immunoprecipitation with β1integrin and Protein A Trisacryl but prior to precipitation ofbiotinylated integrins with streptavidin agarose. The band indicatedby the arrowhead has been tentatively identified as the α1 integrinsubunit. The α3 integrin subunit is difficult to resolve from theprecursor to β1 (pβ1).

11, Fig. 7) when levels are compared to non-overlaid apicalcontrols (lanes 5, 9, Fig. 7). Levels of β1 integrins on basolat-eral cell surfaces are the same with (lanes 4, 8, 12, Fig. 7) orwithout (lanes 2, 6, 10, Fig. 7) collagen overlay. Thus, itappears that a small pool of the β1 integrin family is endoge-nously expressed on apical plasma membranes of HD MDCKcells and that the steady-state levels of β1 integrins on apicalsurfaces do not increase within 12 hours of collagen overlay.Because no major differences in levels of β1 integrins aredetected on apical and basolateral plasma membranes inresponse to collagen, the results further indicate that β1integrins do not redistribute from basolateral to apical cellsurfaces after collagen overlay.

These observations did not rule out the possibility thatapical application of collagen would induce increasedtargeting of newly synthesized integrins to apical cellsurfaces. To test this, we pulse labeled HD MDCK cells 12hours after collagen overlay and differentially surface biotiny-lated and immunoprecipitated β1 integrins under non-denaturing conditions after chase incubations. At 0 minuteschase, integrin is not detected on HD MDCK cell surfacesregardless of collagen overlay (lanes 2, 3, Fig. 8A). Only theprecursor to β1 integrin is synthesized at this time (lanes 1′-3′, Fig. 8B). After 90 minutes chase, all members of the β1family, including α2β1, α3β1 and the β1 precursor are made(lanes 4′-9′, Fig. 8B). As previously reported (Schoenenbergeret al., 1994), it was difficult to resolve the αxβ1. The α2β1and α3β1 are detected on basal plasma membranes in theabsence or presence of collagen (lanes 7, 9, respectively; Fig.8A) after 90 minutes chase. In agreement with our conclusionfrom steady-state labeling experiments, we found that in theabsence of collagen overlay (lane 6, Fig. 8A), the β1 family,primarily α2β1, is detected on apical cell surfaces in amountsequivalent to that seen after steady-state labeling. Moreover,collagen overlay does not stimulate delivery of additionalnewly synthesized integrin to apical plasma membranes (lane8, Fig. 8A).

The results indicate that confluent HD MDCK cells grownin transfilter culture and polarized with respect to apical andbasolateral membrane markers endogenously express lowlevels of β1 integrins on apical plasma membranes and thatthese mediate tubulocyst formation in response to collagenoverlay. β1 integrins are not mobilized to apical plasmamembranes either by redistribution from basolateral to apicalcell surfaces or by new synthesis and delivery to apical plasmamembranes.

The C6 MDCK cell line does not form tubulocystsWe previously reported that β1 integrins are not detected onapical cell surfaces in another clonal MDCK cell line knownas C6 (Schoenenberger et al., 1994). C6 MDCK cells containthe empty retroviral vector pMV7 and are identical to HDMDCK cells with respect to polarization of apical and baso-lateral membrane markers, transepithelial electrical resistanceand ultrastructure (Schoenenberger et al., 1991). They differ,however, in the polarization of β1 integrins. Steady-statelabeling of both cell lines followed by differential cell surfacebiotinylation and immunoprecipitation of β1 integrins undernon-denaturing conditions confirms that β1 integrins are notexpressed on apical plasma membranes of C6 cells (lane 1, Fig.9), whereas the α2β1 and α3β1 are detected on apical plasma

membranes of HD MDCK (lane 3, Fig. 9). The profile of β1integrins expressed on basolateral plasma membranes areidentical for both cell lines (lanes 2 and 4, Fig. 9).

The β1 integrins found on apical plasma membranes of HDMDCK cells are sufficient to mediate tubulocyst formationwithout additional receptor mobilization to the apical domain.To complement these observations, we used the C6 MDCKcell line which does not express apical β1 to determine whetherin the absence of this receptor family, cell-collagen interactions


Fig. 9. Autoradiograms of SDS-PAGE (6%, reduced) gels analyzingsurface expression of β1 integrins in C6 and HD MDCK cellslabeled to steady-state with [35S]methionine [35S]cysteine in theabsence of collagen overlay. Surface biotinylated proteins wereimmunoprecipitated with rabbit polyclonal anti-β1 integrin antibodyunder non-denaturing conditions. C6 cells do not express β1integrins on apical plasma membranes (lane 1), in contrast to HDMDCK cells that express the α2β1 and α3β1 on apical cell surfaces(lane 3). β1 integrins localize to basolateral plasma membranes ofboth C6 and HD MDCK cells (lanes 2 and 4, respectively). Theprofile of β1 expression on basolateral surfaces are identical betweenthe two cell types. The band indicated by the arrowhead has beententatively identified as the α1 subunit.

Fig. 10. Autoradiograms of SDS-PAGE (6%, reduced) gelsanalyzing pulse-chase labeling of β1 integrins in C6 MDCK cellstwelve hours after collagen overlay. Cells were pulsed for 10 minutesand chased for 90 minutes. C6 cells were then differentially surfacebiotinylated from either the apical (a), basolateral (b) or both (ab)cell surfaces and immunoprecipitated with polyclonal anti-β1integrin antibody under non-denaturing conditions. (A) Surfaceexpression of β1 integrins. At 0 minutes chase, integrins are notdetected on cell surfaces with (lane 4) or without (lane 3) collagenoverlay. After 90 minutes chase, the α2β1 and α3β1 are detected onbasal (b) cell surfaces in the absence or presence of collagen (lanes 8,10, respectively). β1 integrins are not detected on apical (a) surfacesof C6 cells (lane 7) and collagen overlay does not mobilize newlysynthesized integrins to apical domains (lane 9). Faint expression ofthe β1 subunit after 90 minutes chase is due to inefficient labelingduring the pulse owing to the pool in the ER.(B) Immunoprecipitable antigen of the respective samples shown in(A). Aliquots (5 µl) were taken after immunoprecipitation with anti-β1 integrin antibody and Protein A Trisacryl but prior toprecipitation of biotinylated proteins with streptavidin agarose. Theband indicated by the arrowhead has been tentatively identified asthe α1 integrin subunit.

depolarize β1 integrins to apical plasma membranes. Twelvehours after overlay, C6 cells were pulse-chased, differentiallysurface biotinylated and immunoprecipitated with polyclonalanti-β1 integrin antibody. Confirming our steady-state labelingdata (Fig. 9, lanes 1, 2), we found that the β1 integrin familyis expressed only on basolateral cell surfaces after a 90 minutechase (lane 8, Fig. 10A) and is not detected at appreciablelevels on apical plasma membranes (lane 7, Fig. 10A). Lowlevels are detected with prolonged exposure of autoradi-ograms; however, the signal is significantly reduced whencompared to the apical signal on HD MDCK cells (Figs 7, 8).Apical application of collagen on C6 cells does not depolarizeβ1 integrins to apical plasma membranes (lane 9, Fig. 10A).Ninety minutes after pulse-labeling, newly synthesized β1integrins are not detected on apical cell surfaces in response tocollagen overlay.

Light microscopy of C6 cells overlaid with type I collagengel further indicates that in the absence of apical β1 integrins,this cell line does not form tubulocysts. Forty-eight hours afteroverlay (Fig. 11B), tubulocysts do not develop across the entireextent of the filter as they do in overlaid HD cells (compare toFig. 2D and E); cell layering is negligible and lumena areabsent. During processing for light microscopy, we observedthat collagen gels were not attached to apical plasmamembranes of C6 cells.

These data indicate that collagen overlay does not stimulatenew synthesis or delivery of β1 integrins to apical plasmamembranes, confirming our observation using the HD MDCK

cell line. Complementing the data which establishes a role forapical β1 integrins in tubulocyst formation, experiments withC6 cells suggest that in the absence of apical β1 tubulocystsdo not develop in response to collagen.

DISCUSSION

We demonstrate that overlay of HD MDCK cells with type Icollagen gel initiates changes in organization and cell polarityleading to the formation of tubulocysts. In particular, we showthat β1 integrins pre-existing on apical plasma membranes arenecessary for tubulocyst formation. In the polarized HD


Fig. 11. Light micrographs of transverse sections (0.5 µm) of C6MDCK cells overlaid from the apical surface with type I collagengel. At 0 hours (A), confluent C6 cells grown on Transwell filters areoverlaid from the apical surface with type I collagen gel. C6 cells donot respond to collagen overlay as do HD MDCK cells (see Fig. 2).Forty-eight hours later (B), polarized monolayer morphology isunchanged. The collagen gel does not attach to apical surfaces of C6cells. Bar, 10 µm.

MDCK cell line, β1 integrins are partially depolarized to apicaland basolateral cell surfaces even though other endogenousapical and basolateral membrane markers are segregated totheir respective plasma membrane domains. If β1 integrins arenot apically expressed, as in C6 MDCK cells, tubulocysts donot develop. Moreover, collagen overlay does not mobilizeadditional β1 integrins to apical plasma membranes either byredistribution from basolateral to apical cell surfaces or by newsynthesis and delivery to apical plasma membranes. β1integrins pre-existing at low levels on apical plasmamembranes of HD MDCK cells actively bind type I collagengel. Presumably, a cascade of intracellular signaling events isactivated leading to the formation of tubulocysts.

Tubulocyst formation by HD MDCK When polarized monolayers of MDCK cells in transfilterculture are overlaid with type I collagen gel, only apical epithe-lial surfaces contact collagen. Subsequent cell-collagen inter-actions initiate epithelial rearrangement resulting in tubulocystformation, confirming previously published results (Hall et al.,1982; Schwimmer and Ojakian, 1995). In the former study,tubulocyst formation was observed only when subconfluentMDCK cells were overlaid, but not when confluent cultureswere used. Based on our results, we suspect that this may havebeen due to the lack of apically expressed integrins once thecells polarized. In the work of Schwimmer and Ojakian (1995),subconfluent MDCK cells, which expose basolateral matrixreceptors to the apical cell surface, were primarily used. Insome experiments, confluent MDCK monolayers wereemployed but the issue of overall basolateral polarity was notrigorously addressed (Schwimmer and Ojakian, 1995).

For our studies we cultured MDCK cells to confluency onpermeable supports and then overlaid polarized monolayerswith a solution of type I collagen which polymerizes into ahydrated gel with incubation. Growth of cells on permeablesupports allows studies on various aspects of epithelial cellpolarization, including localization of apical and basolateral

membrane markers, measurement of transepithelial electricalresistance and selective biochemical analyses of apical and/orbasolateral plasma membranes. The growth of MDCK cells onpermeable supports thus provided a technical advantage forstudies addressing the mechanism(s) of tubulocyst formationin response to apical overlay with ECM. Moreover, polariza-tion of apical and basolateral membrane markers and transep-ithelial electrical resistance could be monitored during theprocess.

The first step in tubulogenesis in response to apical cell-collagen interactions is cell layering. In the absence of cell pro-liferation, epithelial cells within the monolayer ‘migrate’ or‘slide’ over one another, resulting in cell layering. As cellswithin the stratified epithelium adhere to the overlaid collagen,the epithelium undergoes cell division and populates thesurface of the collagen gel. This interpretation is supported byour observation of cell division. We frequently observedmitotic images in the cell layer attached to collagen. Addi-tionally, [3H]thymidine incorporation was greatest at the timecoinciding with tubule formation. Lumena then developbetween regions of epithelia undergoing stratification and arelined by cells from the original monolayer and the newlyformed cell layer attached to collagen.

Cell-cell contacts are not lost during tubulogenesis. Althougha recently published report demonstrates a loss of transepi-thelial electrical resistance within 2 hours after collagen overlayof thyroid cell monolayers (Garbi et al., 1996), we in this studydid not measure any significant change in transepithelial resis-tance at the later time points we examined. The functionalintegrity of the tight junction thus remained intact. Cell-cellcontacts were not disrupted in regions undergoing cell layering,in the original monolayer or in the forming bilayer. Mainte-nance of cell-cell contacts has also been reported during epi-thelial polarity reversal upon transfer of MDCK cysts tocollagen gel (Wang et al., 1990). Because the tight junctionfunctions as a fence separating apical from basolateral plasmamembrane domains (van Meer and Simons, 1986), the obser-vation that tight junction integrity persists in tubulogenesisargues against the possibility that changes in polarization ofapical and basolateral membrane proteins are a consequence ofloss of cell-cell contact and diffusion of plasma membraneproteins between different domains.

While cell-cell contacts remain continuously intact, theepithelium reorganizes, having significant effects on thepolarity and expression of plasma membrane proteins. Duringcell layering, gp135, an apical membrane glycoprotein(Ojakian and Schwimmer, 1988), disappears from plasmamembranes. It is absent from regions of cell-collagen interac-tions, from points of cell-cell contact in the stratified epi-thelium, and from cell-substratum (i.e. permeable support)contacts. This observation agrees with previously publishedreports that find gp135 absent from cell surfaces of MDCKcysts during polarity reversal in collagen gels (Wang et al.1990) and in MDCK cells incubated with collagen overlays(Schwimmer and Ojakian, 1995). Unlike these studies,however, we did not detect any cytoplasmic vesicular stainingof gp135 at any time point. This finding is puzzling consider-ing that internalization of gp135 was readily observed by flu-orescence microscopy in MDCK cysts twelve hours after sus-pension in collagen gels (Wang et al., 1990), a time point thatwe also examined.


In contrast to gp135, the basolateral membrane protein, p58,randomizes during cell layering. In the absence of a freesurface, p58 distributes over the entire plasma membrane atpoints of cell-cell, cell-collagen and cell-substratum contacts.Randomization of other basolateral membrane proteins,including the Na+K+ATPase and E-cadherin, has beenobserved during epithelial polarity reversal of MDCK cystscultured in collagen gel (Wang et al., 1990).

Thus, it appears that apical cell-collagen interaction initiatesa reorientation of the apicobasal axis that leads to changes inpolarization. The cells do not reverse polarity as do suspensioncysts of MDCK cells cultured in collagen gel (Wang et al.,1990; Ojakian and Schwimmer, 1994); rather, polarization isaltered as the epithelium undergoes morphogenesis. In thepresence of both cell-cell and cell-substratum contacts, regard-less of whether the substrate is the filter or overlaid collagen,the apical membrane marker gp135 decreases in surfaceexpression during cell layering whereas the basolateralmembrane marker p58 randomizes on plasma membranes. Inorder for apical and basolateral markers to re-polarize tospecific plasma membrane domains, cell-cell and cell-substra-tum interactions alone are not sufficient; gp135 and p58localize to apical and basolateral cell surfaces, respectively,only with the creation of free lumenal surfaces coinciding withthe immediate onset of tubule formation. Thus the key in re-polarization of apical and basolateral membrane glycoproteinsin our cell culture system is the ‘asymmetry’ of contacts whichre-establishes the apicobasal axis. Because we did not observean intracellular pool, the apical appearance of gp135 is pre-sumably due to synthesis of new protein delivered to apicalplasma membranes. Whether p58 is internalized from incipientlumenal surfaces and is then delivered to basolateral plasmamembranes or whether it is newly synthesized and targeteddirectly basolaterally is a subject for further investigation.

Role of apical integrins in tubulocyst formationThe observation that epithelia from various tissues includingcell lines form tubulocysts when overlaid with type I collagengel raised the question of how collagen at apical plasmamembranes of confluent epithelia (Chambard et al., 1981;Kramer, 1985; Jackson and Jenkins, 1991; Schwimmer andOjakian, 1995) elicits this morphogenetic response. Wehypothesized that apical-ECM interactions would mobilize β1integrins to apical plasma membranes. Using radiolabeled HDMDCK cells that are confluent and polarized with respect toapical and basolateral membrane markers, we demonstratedbiochemically that β1 integrins are predominantly expressedon the basolateral domain, confirming our previously publishedresults (Schoenenberger et al., 1994). In this particular strainof MDCK cells, however, a small amount of β1 integrins werealso detected on the apical domain.

Thus, we provide the first biochemical evidence that β1integrins do indeed exist on apical plasma membranes ofotherwise polarized MDCK cells. While previous studies havedetected apical integrins on MDCK cells by immunocyto-chemical techniques (Ojakian and Schwimmer, 1994), in theabsence of biochemical data the authors of this work wereunable to conclusively determine if just α subunits or αβ1 het-erodimers were expressed apically. As these authors state, theformer possibility would be unlikely given the biochemicalevidence that only integrin heterodimers and not monomers are

transported to the plasma membrane (Heino et al., 1989; Rosaand McEver, 1989). Indeed, their observation that polarityreversal of MDCK cell cysts was inhibited by an anti-β1antibody strongly suggests that functional αβ1 integrins werepresent apically (Ojakian and Schwimmer, 1994). Apical β1integrins have also been detected by immunocytochemistry ona subpopulation of confluent thyroid epithelia where they arebelieved to initiate reorganization of the cell layer in responseto collagen overlay (Garbi et al., 1996).

We also show that β1 integrins expressed apically arenecessary for tubulocyst formation. Two pieces of evidencesupport this conclusion. First, antibody perturbation experi-ments indicate that inhibition of receptor-ligand binding atapical plasma membranes prevents the formation of tubulo-cysts. In particular, the function-blocking anti-β1 integrinantibody, AIIB2, which we have shown inhibits MDCK celladhesion to types I and IV collagen and laminin (Schoenen-berger et al., 1994) prevents not only tubulocyst formation butphysical attachment of collagen gels to apical plasmamembranes. Second, experiments with C6 MDCK cells,demonstrate that this cell line expresses little or no β1 integrinson apical plasma membranes either before or after collagenoverlay (Schoenenberger et al., 1994, and this study), and areunable to form tubulocysts. While it can be argued that theabsence of apical β1 integrins is not the only trait that has beenaltered in the C6 MDCK cell clone, these results and theantibody perturbation experiments with the HD MDCK cellsstrongly suggest that without apical β1 integrins, tubulocystsdo not form.

Contrary to our original hypothesis, our biochemical dataclearly indicate that collagen overlay of the apical surface doesnot mobilize the appearance of β1 integrins to the apicalplasma membrane domain. Through both pulse-chase experi-ments and steady-state labeling we showed that the level ofapical integrins in HD MDCK cells was not increased throughnew integrin synthesis or redistribution of pre-existingintegrins upon collagen overlay. In the C6 clone, collagenoverlay did not lead to the expression of any significantamounts of β1 integrins on the apical surface. Thus, it seemsunlikely that small amounts of integrins or other collagenreceptors that exist on apical surfaces of polarized epithelia actas ‘sensors’ of ectopic ECM proteins and signal the cell tomove additional integrins to the apical domain.

As this work was in progress, a study by Schwimmer andOjakian (1995) appeared which also examined the role ofspecific integrins in tubulocyst formation. While we agree ingeneral terms with their conclusions, we were unable toconfirm or deny their finding that α2β1 in particular wasresponsible for tubulocyst formation because the antibodycited in their study gave inconsistent results in our hands. Moreimportantly, we believe that our work extends their experi-ments by providing biochemical evidence for apical integrins,by demonstrating that these integrins are pre-existing and arenot newly synthesized or redistributed from other nonapicalpools upon collagen overlay, and by showing that the lack ofapical integrins and the inability to form tubulocysts are cor-related in the C6 MDCK cell clone.

The β1 integrins that pre-exist on apical plasma membranesactively bind fibrils within the polymerized type I collagen gel.We observed that within 12 hours of overlay, collagen gels areattached to apical plasma membranes of HD cells. This is an


interesting observation considering that apical plasmamembranes of polarized epithelia are not specialized to formcell contacts. β1 integrins are typically found on basolateralcell surfaces and an apical location is not considered to be theirnormal context. Binding of receptor to its ligand at apicalsurfaces presumably induces assembly of the signal transduc-tion machinery leading to the formation of tubulocysts. In aprocess known as outside-in signaling, receptor-ligand bindingcauses conformational changes in integrin cytoplasmicdomains that in turn activate a cascade of cytoplasmic andnuclear signals resulting in changes in gene expression, cellulardifferentiation and behavior (for review, see Hynes, 1992;Clark and Brugge, 1995; Hemler et al., 1995; Yamada andMiyamoto, 1995). Although the nature of the transductionmechanism in integrin signaling is not well understood, it islikely to include activation of kinases, small molecular massGTPases, phospholipid mediators and/or changes in cytoskele-tal linkages. Preliminary observations suggest that phosphory-lation of integrins or focal adhesion kinase is not involved (A.Zuk and K. S. Matlin, unpublished observation). In addition,preliminary studies using a number of reagents that affect intra-cellular signaling, including activators and inhibitors of proteinkinase C, protein kinase A, and heterotrimeric G proteins,suggest that the signal transduction mechanism(s) mediatingtubulocyst morphogenesis is complex, perhaps involvingmultiple pathways (A. Zuk and K. S. Matlin, unpublishedobservation).

Possible roles of apical integrins in vivoOur findings raise the issue of the functional significance ofdepolarized β1 integrins on apical plasma membranes ofpolarized epithelia. In renal ischemia, redistribution of integrinreceptors from basolateral to apical surfaces of kidney tubularepithelial cells has been suggested to contribute to acute renalfailure (Goligorsky et al., 1993). Damaged cells exfoliated intothe tubule lumen after injury are hypothesized to lose integrinpolarity and attach to apically localized integrins on epitheliaremaining attached to basement membranes, resulting intubular obstruction (Goligorsky et al., 1993). Indeed in in vitromodels of ischemic insult the α3 integrin subunit redistributesfrom a basal to apical location (Gailit et al., 1993). RGD con-taining peptides alleviate tubular obstruction indicating apossible role for RGD sensitive integrins in cell-cell attach-ments that may contribute to obstruction of the tubular lumen(Goligorsky and DiBona, 1993; Noiri et al., 1994). Althoughit has been reported that a basolateral membrane marker, theNa+K+ATPase, depolarizes after ischemia (Molitoris et al.,1988, 1992), it is unclear whether those same cells alsodepolarize integrins from basolateral to apical plasmamembranes or whether apical proteins also redistribute. Thesequestions are currently being pursued in our laboratory usingan in vivo model of ischemic insult.

Apically directed integrins in polarized simple epithelia mayalso play a role in the pathogenesis of infection. The bacterialprotein invasin is a component of the cell wall of Yersinia thatdirectly binds plasma membranes of mammalian cells prior toentry (Isberg and Leong, 1990). Invasin binds to members ofthe β1 family, including the α3β1 which our study shows isexpressed at low levels on apical plasma membranes. Inaddition, the α2β1 which is detected on the apical domain ofpolarized HD cells, has recently been shown to function as a

virus receptor, mediating surface attachment and infection byechovirus 1 (Bergelson et al., 1993), a human pathogen respon-sible for febrile illnesses and aseptic meningitis. It is unclearhow bacteria and viruses in lumenal compartments gain entryinto the cellular interior of simple polarized epithelia such asthose that line the gastrointestinal, kidney and respiratorysystems. Although apically localized integrins are not normallydetected in in situ tissue sections (Ruoslahti, 1991), our obser-vation that a polarized epithelial cell line that segregates apicaland basolateral membrane proteins to specific plasmamembrane domains yet expresses partially depolarized integrinon apical plasma membranes raises the possibility that integrinreceptors may also be expressed at low levels on apicalsurfaces of simple polarized epithelia in vivo. These in turnmay mediate bacterial and viral entry and subsequent infection.

Apical integrins may also play an important role in epithelialoncogenesis. If integrins are depolarized at an early stage of car-cinogenesis, then our results indicate that exposure of apicalsurfaces in the incipient neoplasm to ECM proteins would theninitiate cell layering, ultimately resulting in the formation of atumor. Indeed changes in integrin distribution and expressionhave been reported in carcinomas in situ and in malignant cells(Pignatelli et al., 1990, 1991; Koretz et al., 1991; Koukoulis etal., 1991; Weinel et al., 1992). Our previous results in factsuggest that oncogenic transformation and loss of integrinpolarity may go hand-in-hand. When polarized MDCK cells aretransformed by expression of the Kirsten ras oncogene, apicalantigens and basolateral β1 integrins lose polarity as themonolayer reorganizes into a multilayer (Schoenenberger et al.,1991, 1994). Other basolateral antigens do not redistribute fromareas of cell-cell contact and tight junctions remain functionaland located at the border between the free surface and the lateralplasma membranes (Schoenenberger et al., 1991). How disrup-tion of the ras signaling pathway leading to oncogenic trans-formation and the alteration in cell polarity and integrinexpression are interrelated remain unanswered but importantquestions for the future.

In conclusion, we have in this report addressed themechanism by which apical application of collagen onpolarized epithelia results in the formation of tubulocysts. Weshow that the asymmetry of contact which develops with thecreation of free lumenal surfaces during tubulocyst formationre-establishes the apicobasal axis leading to re-polarization ofapical and basolateral membrane markers. We emphasize thepoint that in one clone of MDCK cells apical and basolateralmembrane markers are polarized to their respective plasmamembrane domains yet β1 integrins which are considered tobe basolateral matrix receptors are expressed on apical cellsurfaces where they mediate the formation of tubulocysts. Ifapical β1 integrins are absent, tubulocysts do not form.Moreover, additional β1 integrins are not mobilized to apicalplasma membranes in response to collagen. The β1 integrinsexpressed apically actively bind ECM despite an apicallocation not being their normal place of function. We suggestthat previous studies documenting the ability of epithelia fromvarious tissue sources to form tubulocysts in response to typeI collagen overlay is likely due to β1 integrins pre-existing onapical plasma membranes.

We are grateful to Drs Caroline Damsky, Barry Gumbiner, GeorgeOjakian and Erkki Ruoslahti for their generous gifts of antibodies and


to Dr James Casanova and Gregory Zinkl for critically reading themanuscript and helpful discussions. This work was supported by theNational Institutes of Health (R01 DK46768, P01 DK38452, and P30DK43351).

REFERENCES

Balcarova-Ständer, J., Pfeiffer, S. E., Fuller, S. D. and Simons, K. (1984).Development of cell surface polarity in the epithelial Madin-Darby caninekidney (MDCK) cell line. EMBO J. 3, 2687-2694.

Bergelson, J. M., Chan, B. M. C., Finberg, R. W. and Hemler, M. E. (1993).The integrin VLA-2 binds echovirus 1 and extracellular matrix ligands bydifferent mechanisms. J. Clin. Invest. 92, 232-239.

Boll, W., Partin, J. S., Katz, A. I., Caplan, M. J. and Jamieson, J. D. (1991).Distinct pathways for basolateral targeting of membrane and secretoryproteins in polarized epithelial cells. Proc. Nat. Acad. Sci. USA 88, 8592-8596.

Carter, W. G., Wayner, E. A., Bouchard, T. S. and Kaur, P. (1990). The roleof integrins α2β1 and α3β1 in cell-cell and cell-substrate adhesion of humanepidermal cells. J. Cell Biol. 110, 1387-1404.

Chambard, M., Gabrion, J. and Mauchamp, J. (1981). Influence of collagengel on the orientation of epithelial cell polarity: follicle formation fromisolated thyroid cells and from preformed monolayers. J. Cell Biol. 91, 157-166.

Clark, E. A. and Brugge, J. S. (1995). Integrins and signal transductionpathways: the road taken. Science 268, 233-239.

Clark, R. A. F., editor (1996). The Molecular and Cellular Biology of WoundRepair. 2nd edn. New York: Plenum Press (in press).

Damsky, C. H. and Werb, Z. (1992). Signal transduction by integrin receptorsfor extracellular matrix: cooperative processing of extracellular information.Curr. Opin. Cell Biol. 4, 772-781.

Ekblom, P. (1989). Developmentally regulated conversion of mesenchyme toepithelium. FASEB J. 3, 2141-2150.

Gailit, J., Colflesh, D., Rabiner, I., Simone, J. and Goligorsky, M. S. (1993).Redistribution and dysfunction of integrins in cultured renal epithelial cellsexposed to oxidative stress. Am. J. Physiol. 264, F149-F157.

Gailit, J. and Clark, R. A. F. (1994). Wound repair in the context ofextracellular matrix. Curr. Opin. Cell Biol. 6, 717-725.

Garbi, C., Negri, R., Calì, G. and Nitsch, L. (1996). Collagen interaction withapically expressed β1 integrins: loss of transepithelial resistance andreorgnization of cultured thyroid cell monolayer. Eur. J. Cell Biol. 69, 64-75.

Giancotti, F. G. and Ruoslahti, E. (1990). Elevated levels of the α5β1fibronectin receptor suppress the transformed phenotype of Chinese hamsterovary cells. Cell 60, 849-859.

Goligorsky, M. S. and DiBona, G. F. (1993). Pathogenetic role of Arg-Gly-Asp recognizing integrins in acute renal failure. Proc. Nat. Acad. Sci. USA90, 5700-5704.

Goligorsky, M. S., Lieberthal, W., Racusen, L. and Simon, E. E. (1993).Integrin receptors in renal tubular epithelium: new insights intopathophysiology of acute renal failure. Am. J. Physiol. 264, F1-F8.

Gottardi, C. J. and Caplan, M. J. (1992). Cell surface biotinylation in thedetermination of epithelial membrane polarity. J. Tiss. Cult. Meth. 14, 173-180.

Gumbiner, B. and Simons, K. (1986). A functional assay for proteins involvedin establishing an epithelial occluding barrier: identification of anuvomorulin-like polypeptide. J. Cell Biol. 102, 457-468.

Hall, D. E., Reichardt, L. F., Crowley, E., Holley, B., Moezzi, H.,Sonnenberg, A. and Damsky, C. H. (1990). The α1/β1 and α6/β1 integrinheterodimers mediate cell attachment to distinct sites on laminin. J. Cell Biol.110, 2175-2184.

Hall, H. G., Farson, D. A. and Bissell, M. J. (1982). Lumen formation byepithelial cell lines in response to collagen overlay: a morphogenetic modelin culture. Proc. Nat. Acad. Sci. USA 79, 4672-4676.

Heino, J., Ignotz, R. A., Hemler, M. E., Crouse, C. and Massague, J. (1989).Regulation of cell adhesion receptors by transforming growth factor β:concomitant regulation of integrins that share a common β1 subunit. J. Biol.Chem. 264, 380-388.

Hemler, M. E., Weitzman, J. B., Pasqualini, R., Kawaguchi, S., Kassner, P.D. and Berdichevsky, F. B. (1996). Structure, biochemical properties, andbiological functions of integrin cytoplasmic domains. In The Integrins: TheBiological Problems (ed. Y. Takada). CRC Press (in press).

Hynes, R. O. (1992). Integrins: versatility, modulation, and signaling in celladhesion. Cell 69, 11-25.

Isberg, R. R. and Leong, J. M. (1990). Multiple β1 chain integrins arereceptors for invasin, a protein that promotes bacterial penetration intomammalian cells. Cell 60, 861-871.

Jackson, C. J. and Jenkins, K. L. (1991). Type I collagen fibrils promote rapidvascular tube formation upon contact with the apical side of culturedendothelium. Exp. Cell Res. 192, 319-323.

Juliano, R. L. and Haskill, S. (1993). Signal transduction from theextracellular matrix. J. Cell Biol. 120, 577-585.

Koretz, K., Schlag, P., Boumsell, L. and Moller, P. (1991). Expression ofVLA-α2, VLA-α6, and VLA-β1 chains in normal mucosa and adenomas ofthe colon, and in colon carcinomas and their liver metastases. Am. J. Pathol.138, 741-750.

Korhonen, M., Laitinen, L., Ylanne, J., Gould, V. E. and Virtanen, I.(1992). Integrins in developing, normal and malignant human kidney.Kidney Int. 41, 641-644.

Koukoulis, G. K., Virtanen, I., Korhonen, M., Laitinen, L., Quaranta, V.and Gould, V. E. (1991). Immunohistochemical localization of integrins inthe normal, hyperplastic, and neoplastic breast. Am. J. Pathol. 139, 787-799.

Kramer, R. H. (1985). Extracellular matrix interactions with the apical surfaceof vascular endothelial cells. J. Cell Sci. 76, 1-16.

Laemmli, U. K. (1970). Cleavage of structural proteins during assembly of thehead of bacteriophage T4. Nature 227, 680-685.

Larjava, H., Peltonen, J., Akiyama, S. K., Yamada, S. S., Gralnick, H. R.,Uitto, J. and Yamada, K. M. (1990). Novel function for β1 integrins inkeratinocyte cell-cell interactions. J. Cell Biol. 110, 803-815.

LeCluyse, E. L., Audus, K. L. and Hochman, J. H. (1994). Formation ofextensive canalicular networks by rat hepatocytes cultured in collagen-sandwich configuration. Am. J. Physiol. 266, C1764-C1774.

Louvard, D. (1980). Apical membrane aminopeptidase appears at sites of cell-cell contact in cultured kidney epithelial cells. Proc. Nat. Acad. Sci. USA 77,4132-4136.

Matlin, K. S., Reggio, H., Helenius, A. and Simons, K. (1981). Infectiousentry pathway of influenza virus in a canine kidney cell line. J. Cell Biol. 91,601-613.

Matlin, K. S. and Simons, K. (1983). Reduced temperature prevents transferof a membrane glycoprotein to the cell surface but does not prevent terminalglycosylation. Cell 34, 233-243.

Matlin, K. S. and Caplan, M. J. (1992). Epithelial cell structure and polarity.In The Kidney: Physiology and Pathophysiology (ed. D. W. Seldin and G.Giebisch), pp. 447-473. Raven Press, Ltd, New York.

McLean, I. W. and Nakane, P. K. (1974). Periodate-lysine-paraformaldehydefixative. A new fixative for electron microscopy. J. Histochem. Cytochem.22, 1077-1083.

McNeil, H., Ozawa, M., Kemler, R. and Nelson, W. J. (1990). Novel functionof the cell adhesion molecule uvomorulin as an inducer of cell surfacepolarity. Cell 62, 309-316.

Molitoris, B. A., Hoilien, D. A., Dahl, R. A., Ahnen, D. J., Wilson, P. D. andKim, J. (1988). Characterization of ischemia-induced loss of epithelialpolarity. J. Membr. Biol. 106, 233-242.

Molitoris, B. A., Dahl, R. and Geerdes, A. (1992). Cytoskeleton disruptionand apical redistribution of proximal tubule Na(+)-K(+)-ATPase duringischemia. Am. J. Physiol. 263, F488-F495.

Montesano, R., Orci, L. and Vassalli, P. (1983). In vitro organization ofendothelial cells into capillary-like networks promoted by collagen matrices.J. Cell Biol. 97, 1648-1654.

Montgomery, R. K. (1986). Morphogenesis in vitro of dissociated fetal ratsmall intestinal cells upon an open surface and subsequent to collagen geloverlay. Dev. Biol. 117, 64-70.

Narula, N., McMorrow, I., Plopper, G., Doherty, J., Matlin, K. S., Burke,B. and Stow, J. L. (1992). Identification of a 200kDa brefeldin-sensitiveprotein on Golgi membranes. J. Cell Biol. 117, 27-38.

Nelson, W. J. and Veshnock, P. J. (1986). Dynamics of membrane-skeleton(fodrin) organization during development of polarity in Madin-Darby caninekidney epithelial cells. J. Cell Biol. 103, 1751-1765.

Nelson, W. J. and Veshnock, P. J. (1987a). Modulation of fodrin (membraneskeleton) stability by cell-cell contact in Madin-Darby canine kidneyepithelial cells. J. Cell Biol. 104, 1527-1537.

Nelson, W. J. and Veshnock, P. J. (1987b). Ankyrin binding to Na, K-ATPaseand implications for the organization of membrane domains in polarizedcells. Nature 328, 533-536.

Nelson, W. J. and Hammerton, R. W. (1989). A membrane-cytoskeletalcomplex containing Na, K-ATPase, ankyrin and fodrin in Madin-Darby


canine kidney (MDCK) cells: implications for the biogenesis of epithelialcell polarity. J. Cell Biol. 108, 893-902.

Noiri, E., Gailit, J., Sheth, D., Magazine, H., Gurrath, M., Muller, G.,Kessler, H. and Goligorsky, M. S. (1994). Cyclic RGD peptides ameliorateischemic acute renal failure in rats. Kidney Int. 46, 1050-1058.

Ojakian, G. K. and Schwimmer, R. (1988). The polarized distribution of anapical cell surface glycoprotein is maintained by interactions with thecytoskeleton of Madin-Darby canine kidney cells. J. Cell Biol. 107, 2377-2387.

Ojakian, G. K. and Schwimmer, R. (1994). Regulation of epithelial cellsurface polarity by β1 integrins. J. Cell Sci. 107, 561-576.

Pignatelli, M. and Bodmer, W. F. (1990). Integrin cell adhesion moleculesand colorectal cancer. J. Pathol. 162, 95-97.

Pignatelli, M., Hanby, A. M. and Stamp, G. W. H. (1991). Low expression ofβ1, α2 and α3 subunits of VLA integrins in malignant mammary tumors. J.Pathol. 165, 25-32.

Plantefaber, L. C. and Hynes, R. O. (1989). Changes in integrin receptors ononcogenically transformed cells. Cell 56, 281-290.

Reinsch, S. and Karsenti, E. (1994). Orientation of spindle axis anddistribution of plasma membrane proteins during cell division in polarizedMDCK II cells. J. Cell Biol. 126, 1509-1526.

Rodriguez-Boulan, E., Paskiet, K. T. and Sabatini, D. (1983). Assembly ofenveloped viruses in MDCK cells: polarized budding from single attachedcells and from clusters of cells in suspension. J. Cell Biol. 96, 866-874.

Rodriguez-Boulan, E. and Nelson, W. J. (1989). Morphogenesis of thepolarized epithelial cell phenotype. Science 245, 718-725.

Rosa, J. P. and McEver, R. P. (1989). Processing and assembly of the integringlycoprotein IIb-IIIa in HEL cells. J. Biol. Chem. 264, 12596-603.

Ruoslahti, E. (1991). Integrins as receptors for extracellular matrix. In CellBiology of the Extracellular Matrix (ed. E. D. Hay), pp. 343-363. PlenumPress, New York.

Schoenenberger, C.-A. and Matlin, K. S. (1991). Cell polarity and epithelialoncogenesis. Trends Cell Biol. 1, 87-92.

Schoenenberger, C., Zuk, A., Kendall, D. and Matlin, K. S. (1991).Multilayering and loss of apical polarity in MDCK cells transformed withviral K-ras. J. Cell Biol. 112, 873-889.

Schoenenberger, C., Zuk, A., Zinkl, G. M., Kendall, D. and Matlin, K. S.

(1994). Integrin expression and localization in normal MDCK cells andtransformed MDCK cells lacking apical polarity. J. Cell Sci. 107, 527-541.

Schwimmer, R. and Ojakian, G. K. (1995). The α2β1 integrin regulatescollagen-mediated MDCK epithelial membrane remodeling and tubuleformation. J. Cell Sci. 108, 2487-2498.

Simons, K. and Fuller, S. D. (1985). Cell surface polarity in epithelia. Annu.Rev. Cell Biol. 1, 243-288.

Sorokin, L., Sonnenberg, A., Aumailley, M., Timpl, R. and Ekblom, P.(1990). Recognition of the laminin E8 cell-binding site by an integrinpossessing the α6 subunit is essential for epithelial polarization indeveloping kidney tubules. J. Cell Biol. 111, 1265-1274.

van Meer, G. B. and Simons, K. (1986). The function of tight junctions inmaintaining differences in lipid composition between the apical andbasolateral cell surface domains of MDCK cells. EMBO J. 5, 1455-1464.

Vega-Salas, D. E., Salas, P. J. I., Gundersen, D. and Rodriguez-Boulan, E.(1987). Formation of the apical pole of epithelial (Madin-Darby kidney)cells: polarity of an apical protein is independent of tight junctions whilesegregation of a basolateral marker requires cell-cell interactions. J. CellBiol. 104, 905-916.

Wang, A. Z., Ojakian, G. K. and Nelson, W. J. (1990). Steps in themorphogenesis of a polarized epithelium. II. Disassembly and assembly ofplasma membrane domains during reversal of epithelial cell polarity inmulticellular epithelial (MDCK) cysts. J. Cell Sci. 95, 153-165.

Weinel, R. J., Rosendahl, A., Neumann, K., Chaloupka, B., Erb, D.,Rothmund, M. and Santoso, S. (1992). Expression and function of VLA-α2, -α3, -α5, and -α6 integrin receptors in pancreatic carcinoma. Int. J.Cancer. 52, 827-833.

Yamada, K. M. and Miyamoto, S. (1995). Integrin transmembrane signalingand cytoskeletal control. Curr. Opin. Cell Biol. 7, 681-689.

Zuk, A., Matlin, K. S. and Hay, E. D. (1989). Type I collagen gel inducesMadin-Darby canine kidney cells to become fusiform in shape and loseapical-basal polarity. J. Cell Biol. 108, 903-920.

Zuk, A. and Hay, E. D. (1994). Expression of β1 integrins changes duringtransformation of avian lens epithelium to mesenchyme in collagen gels.Dev. Dynam. 201, 378-393.

(Received 13 February 1996 - Accepted 11 April 1996)

Apical 1 integrin in polarized MDCK cells mediates tubulocyst … · 2001-05-11 · Apical β1...

Documents

Transcript of Apical 1 integrin in polarized MDCK cells mediates tubulocyst … · 2001-05-11 · Apical β1...