Differences in adhesion to tissue culture plastic of clonally related

transformed and control sublines from an epithelial cell strain

JOHN G. STEELE*, B. ANN DALTON, P. ANNE UNDERWOOD

Division of Biomolecular Engineering, Laboratory for Molecular Biology, Commonwealth Scientific and Industrial ResearchOrganization, PO Box 184, North Ryde, New South Wales 2113, Australia

and GARRY J. SMITH

Carcinogenesis Research Unit, School of Pathology, University of New South Wales, Kensington, New South Wales 2033, Australia

* Author for correspondence

Summary

Clonally related sublines of the NAL1A lung epi-thelial cell strain were used in a comparison of themechanism of attachment and the morphology ofcontrol and transformed epithelial cells. The initialattachment and spreading of the control cells ontissue culture plastic was shown to be dependentupon adsorption of serum vitronectin to the sub-stratum. The alphav subunit of the vitronectinreceptor was detected in both the control andtransformed cells by immunoprecipitation and im-munoblot methods. The spontaneously transformedcells differed from the control cells in that, whereasattachment to tissue culture plastic could occur bybinding to adsorbed vitronectin, the transformedcells could also become attached, with time, by avitronectin-independent mechanism. Attachment bythis vitronectin-independent reaction was inhibitedby the protein synthesis inhibitor cycloheximide andalso by the microtubule-disrupting drugs demicolce-mid and nocodazole. The morphologies of attached

control and transformed cells cultured on tissueculture plastic were disrupted by treatment withcytochalasin B, demicolcemid or nocodazole, indi-cating that the shape of these cultured epithelial cellsis dependent upon the microtubule system as well asthe actin filaments. These results show one importantdifference between the control and transformedcells, in that the transformed cells can attach to tissueculture plastic by a vitronectin-independent mechan-ism that involves new protein synthesis by the cell.Another interesting difference is that this vitronec-tin-independent attachment of the transformed cellswas sensitive to inhibition by microtubule-disruptingagents. On the other hand, the attachment of eithertransformed or control cells to fibronectin- or vitro-nectin-coated surfaces was not affected by micro-tubule-disrupting agents.

Key words: vitronectin, cell attachment, microtubules,epithelial cells, demicolcemid, cell shape.

Introduction

One difference often observed between malignant andnonmalignant cells derived from the same cell type is inthe shapes that the cells adopt in culture (see Ben-Ze'ev,1985; Vasiliev, 1985). Having observed a transformation-associated change in shape of cultured epithelial cellsderived from pulmonary alveolar epithelium, we havebeen investigating the mechanism of this change (Smith etal, 1986; Steele et al. 1988, 1990). In the NAL1A culturesystem that we have been using, nonmalignant controlcells derived from normal lung tissue have undergone aspontaneous transformation in culture to become capableof anchorage-independent growth and formation of meta-static carcinomas in mice (Smith and Lykke, 1985; Smithet al. 1986) and to resemble malignant cells cultured fromlung adenoma (Smith and Bentel, 1987). The transform-ation step was initially evident from a change in cell shapeand growth pattern; the transformed NAL1A cells grow-ing as islands of tightly associated cells that spread poorlyon tissue culture polystyrene (TCP) surface whereas the

Journal of Cell Science 100, 195-203 (1991)Printed in Great Britain © The Company of Biologists Limited 1991

control cells spread on the surface quite readily whenseeded at sparse densities (Smith et al. 1986). In previousstudies we have shown that the control NAL1A cellssynthesize and accumulate fibronectin (Fn) and laminininto a pericellular matrix, whereas these glycoproteinscould not be detected in the transformed cells (Steele et al.1988, 1990). The control cells show enhanced attachmentand spreading on Fn- or laminin-coated substrata. Incontrast, the transformed cells can attach to Fn orlaminin, but do not show enhanced spreading on thesesubstrata. In contrast to the result with transformedfibroblasts, the shape of the transformed NAL1A cellscannot be made to revert to that of the control cells byseeding on a substratum of Fn or laminin (Steele et al.1988, 1990). Therefore, although reduced expression ofextracellular matrix components is one aspect of thetransformed phenotype of NAL1A cells, there must besome additional change in the mechanism controlling theshape of the transformed cells that is also responsible forthe refractory nature of the shape of these cells to theculture substratum.

195

This paper reports the results of experiments to testwhether the difference in cell shape of the transformed andcontrol cells following attachment to TCP arises as a resultof the use by these cells of different adhesive molecules forattachment to the TCP substratum. The initial attach-ment and spreading of fibroblasts and endothelial cellsseeded on TCP is dependent upon adsorption of serumvitronectin (Vn) to TCP (Underwood and Bennett, 1989;Norris et al. 1990; Steele et al. 1991). In this study we havecompared the control and transformed NAL1A epithelialcells for their dependence upon serum Vn for the initialattachment and spreading onto TCP. The microtubule-active drugs demicolcemid and nocodazole and the actin-disrupting drug cytochalasin B have been used in a studyof the sensitivity of the attachment and shape of thetransformed and control NAL1A cells to disruption ofmicrotubules and actin filaments. The results of theexperiments show that the transformed cells have adifferent mechanism for attachment to TCP to that of thecontrol cells.

Materials and methods

Cell lines and cell cultureThe clonal cell strain C4E10 is a subclone of the C4 clone ofmurine cell strain NAL1A, which is derived from type 2pneumocytes of alveolar epithelium. C4E10 cells are notmalignant (Smith and Lykke, 1985; Smith et al. 1986). CloneC4SE9 emerged from clone C4 of NAL1A as a focus of cells withdifferent morphology and is clonally related to the C4E10 clonebut C4SE9 cells are malignant (Smith et al. 1986). The cells weregrown in CMRL1066 supplemented with 2.5/igml"1 fungizone,lOO/igml"1 kanamycin, 20mM Hepes at pH7.4 (Flow Labora-tories, Australia) and 10 % (v/v) foetal calf serum (FCS).

Cell attachment assaysTCP dishes (well surface area of 3.8 cm2) were precoated with10/jgmr1 Fn (from bovine plasma, Sigma Chemicals) or5 ;<g ml"' Vn by incubation at 37 °C for at least 1 h, then the disheswere washed twice with phosphate buffered saline (PBS)immediately before seeding of cells. Vn was purified and serumselectively depleted of Vn by passage over an affinity columnconsisting of anti-bovine Vn monoclonal antibody (mAb) A27immobilized on Sepharose 4B by the methods reported previously(Underwood and Bennett, 1989; Norris et al. 1990). Polystyrenedishes (well surface area of 3.24 cm2, from Sterilin, England) wereincubated with Fn or Vn solutions (coating concentrations asshown in Fig. 1C) at 37 °C for lh , the dishes were washed twicewith PBS and residual protein binding sites were blocked byincubation with 10 mgml"1 bovine serum albumin (BSA) for afurther hour.

For cell attachment experiments involving metabohcallylabelled cells, C4E10 and C4SE9 cell cultures that were withinone day of confluence were metabolically labelled with 0.050 mCi(1.85 MBq) ml"1 [^SJmethionine (SJ1515 from Amersham Radio-chemical Centre, UK) for 18-22 h in a culture medium of DMEMcontaining only Smgl"1 methionine (1 vol. complete DMEM plus9 vols DMEM without methionine, Flow Laboratories) and 10 %(v/v) FCS. Cells for attachment and spreading analysis weretrypsinized with 0.05 % (w/v) trypsin in PBS, detached from theculture dish, and then the trypsin was neutralized by addition ofsoybean trypsin inhibitor (Sigma Chemicals) to a concentration of0.1 % (w/v). As reported previously (Steele et al. 1988), the C4E10cells detached from the substratum after only a short treatmentwith trypsin whereas the C4SE9 cells required more extensiveincubation with trypsin and tituration. Cells were collected bygentle centrifugation and resuspended by gentle tituration inCMRL1066 medium containing fungizone, kanamycin and Hepesbuffer. Cells were seeded onto different substrata by addition towells containing an equal volume of medium containing serum or

BSA, to give a final concentration of either 10 % (v/v) FCS (intactor Vn-depleted) or 10 mg ml"1 BSA (Fig. 1C). After 90 min, 4 h, 6 hor 16 h incubation (see figure and table legends for details),culture dishes seeded with radiolabelled cells were washed fourtimes with PBS to remove unattached cells, and then theadherent cells were removed from the dish by trypsinization andsolubilization with 1 % (w/v) Nonidet P40 (NP40) and the MScontent was determined by liquid scintillation counting.

Detection of Vn receptor subunit alpha^Cultures that were within 1 day of confluence were metabolicallylabelled with 0.2 mCi (7.4 MBq) ml"1 [^SJmethionine for 16-18 hin the culture medium containing the reduced concentration ofmethionine. The cells were washed with PBS, then extracted with1% (v/v) NP40 in 50 mM Tris-HCl, pH7.6, containing lmM 1,2-di(2-aminoethoxy)ethane-ivyv',./vyV'-tetraacetic acid and 150 mMNaCl. Immunoprecipitation was conducted by preloading ofprotein A-Sepharose 4B beads (between 100 and 200 jd of a 50 %(v/v) slurry of beads from Pharmacia per incubation tube) witheither rabbit antiserum raised against the alphav subunit of thehuman vitronectin receptor isolated from human placenta (TeliosPharmaceuticals Inc., catalogue no. A109) or nonimmune rabbitserum. After incubation with shaking for 2 h at room temperatureto load the beads with antibody, the beads were washed once withPBS containing 1% (w/v) BSA, pH7.5, and then the antibody-coated beads were incubated with cell extract (usually 500 /<1) for2 h at room temperature with vigorous shaking. The beads werecollected and washed 4 times using 1 % NP40 buffer, and then theprecipitated proteins were solubilized by boiling in electrophor-esis sample buffer and electrophoresed on nonreduced 6.5 % (w/v)polyacrylamide gels.

DrugsStock solutions of demicolcemid (1 mgml"1, Sigma Chemicals) inwater and of nocodazole (lmgml"1, Sigma Chemicals) orcytochalasin B (5mgml"1, Sigma Chemicals) in dimethylsulphoxide (DMSO) were stored at -20°C. Controls in exper-iments involving nocodazole or cytochalasin B were cultured inmedium containing an equivalent concentration of DMSO alone.Cycloheximide (2.5mgml"1 in water, Sigma Chemicals) wasstored at -70°C.

Immunofluorescence microscopy distribution of tubulinand of F-actinCells used for these immunostaining experiments were grown onFn-coated glass. Cells to be stained for tubulin were cultured forone day either with or without 1/zgml"1 demicolcemid, washedwith PBS and fixed and permeabilized by incubation for 14 h at4°C with 4% (v/v) solution of 40% (w/v) formaldehyde in PBS,also containing 0.2% (w/v) Triton X-100. The fixed cells werewashed twice with PBS, incubated with 10 % (w/v) skim milk inPBS (pH 7.2) for 45 min and then incubated with rabbit antiserumraised against tubulin from chick embryo brain (1 in 12 dilutionin skim milk solution, code 65-095 from Miles-Yeda Ltd) for 1 h.The cells were washed three times with PBS containing 0.02%(w/v) Tween 20 and then incubated with 70;igml~1 fluoresceinisothiocyanate (FITC)-conjugated swine IgGs to rabbit IgGs(Dako) in PBS for 1 h in the dark. Following one wash with PBScontaining 0.02 % Tween 20 and two washes with PBS, the cellswere viewed at a wavelength of 550 nm using a x 100 objective ona Nikon Diaphot inverted microscope connected to a BioRadMR500 laser confocal microscope.

Cells to be stained for F-actin distribution were cultured for 1day and then treated with 25 /<M cytochalasin B for 4h at 37 °Cprior to fixation. Cells were fixed with formaldehyde-PBSsolution (without Triton X-100) for 15 min at room temperature.Cells were then washed with PBS, permeabilized by 15 minincubation in methanohacetone (4:1, v/v, at -20°C) and then airdried. The slides were then washed with PBS, incubated for 1 hwith 12 units ml"1 rhodamine-phalloidin (Molecular Probes Inc.,OR, USA) in the dark, washed three times with PBS, mountedwith 50% (v/v) glycerol-PBS and examined with an OlympusDMT-2 microscope at 514 nm, with an immunofluorescenceattachment.

196 J. G. Steele et al.

Results

Role of serum vitronectin in attachment of C4E10 andC4SE9 cells to TCPWe have previously reported that single C4E10 cellsattach and spread, within 90 min of seeding in culturemedium containing 10 % (v/v) serum, on TCP or on TCPcoated with Fn. The attachment of clonally related C4SE9cells to TCP is relatively slow, requiring more than 3-12 hafter seeding to be complete. The C4SE9 cells seed asclumps and grow as islands of cells (Steele et al. 1988).

To determine the role that the serum Vn may play in theinitial attachment and spreading of C4E10 and C4SE9cells on TCP, Vn was selectively removed from the serum

used in the culture medium. Fig. 1A shows the effect ofdepletion of Vn from FCS upon the attachment of C4E10cells after 90 min and 16 h of culture, and the effect uponC4E10 cell spreading after 90 min is shown in Fig. IB.Depletion of Vn from the culture medium completelyabolished the attachment and spreading of C4E10 cells toTCP during the first 90 min of culture, and continuedculture for a 16 h period did not result in cell attachment(Fig. 1A,B). The attachment and spreading of C4E10 cellsseeded on Fn-TCP in this Vn-depleted medium wasunimpaired (as compared to C4E10 cells seeded on TCP inintact medium), indicating that the Vn-depleted mediumwas not toxic to the cells. The ability of C4E10 cells toattach to Vn was directly demonstrated by coating

90 min 16 h

•8

Ia

PCSTCP

- Vn

TCP

-Vn

Fn-TCP

FCS

TCP

- Vn

TCP

-Vn

Fn-TCP

100-

80

2 60-

V 40-

0.01 0.1 1 10Coating concentration (/<gml~')

100

B 90 min 16 hlOOn

80-

60-

O 40-

20-

FCS - Vn -Vn

TCP TCP Fn-TCP

FCS - Vn -Vn

TCP TCP Fn-TCP

D 1 2 3 4

••207

Fig. 1. Involvement of serum Vn and Fn in attachment of C4E10 and C4SE9 cells to TCP. (A and B) Histograms showingproportion of C4E10 (open bars) and C4SE9 (hatched bars) cells attached and spread on TCP or Fn-TCP (as indicated) afterseeding in culture medium containing intact FCS (FCS) or FCS depleted of Vn (-Vn). A shows cell attachment (% of total cellsseeded) 90 min and 16 h after seeding, and B shows spreading of C4E10 cells 90 min, and of C4SE9 cells 16 h, after seeding.Mean±s.E.M. (C) Effect of the concentration of Vn (squares) or Fn (circles) used to precoat a polystyrene culture substratum uponthe proportion of C4E10 cells (% of total cells seeded) that become attached during 90 min incubation in serum-free medium.Mean±s.E.M. (D) Detection of cellular vitronectin receptor alphav subunit by immunoprecipitation analysis. C4E10 cells (tracks 1and 2) or C4SE9 cells (tracks 3 and 4) were biosynthetically labelled with [ S]methionine and then the proteins in a cell extractsolubilized in Nonidet P-40 were immunoprecipitated with anti-alphav antiserum (tracks 2 and 4) or preimmune serum (tracks 1and 3). Immunoprecipitated proteins were separated (without reduction) on a SDS-polyacrylaniide gel; the arrows on the rightshow the calculated Mr values (xl0~3) of key bands.

Epithelial cell adhesion and shape 197

Table 1. Effect of cycloheximide (25 fig ml 1) onattachment of C4SE9 cells to Fn-TCP, Vn-TCP or TCPover a 6h incubation in medium containing Vn-depleted

serum

Cells

C4SE9

C4SE9

C4SE9

•The-;depleted

Surface Cycloheximide

Fn-TCP

Vn-TCP+

TCP+

i cells attached when seeded in

% Cells attached(mean±s D.)

100 0*50 6±7.068 0±6.620.7±0.6

143.4±9.04.0±0.5

i medium containing

%Inhibition

49.4

69.5

97.2

Vn-serum is expressed relative to attachment to the Fn-TCP

surface in the absence of 25/(g ml"1 cycloheximide.

polystyrene with purified Vn and incubating the cells withthis substratum using serum-free culture medium(Fig. 1C). Optimal C4E10 cell attachment was obtainedusing a Vn coating concentration of O.SjUgml"1 or higher,but half-maximal attachment could be detected usingcoating concentrations as low as SOngml"1 Vn. When Fnwas used to precoat TCP, optimal attachment required Fnconcentrations of 2-10 Hgml"1 and half-maximal attach-ment required 1 /igml , but the maximum attachment ofC4E10 cells to a Fn-coated surface was the same as for aVn-coated surface (Fig. 1C). These results indicate thatC4E10 cells are very sensitive to Vn, and that attachmentof C4E10 cells to TCP is absolutely dependent uponadsorption of serum Vn to the substratum.

As reported previously (Steele et al. 1988, 1990), theattachment of C4SE9 cells to TCP during the first 90 minafter seeding in medium containing intact serum wasrelatively poor compared to that of C4E10 cells. Theattached cells were clumped and had not spread on thesubstratum. This attachment to TCP was inhibited byremoval of Vn from the serum, indicating that the limitedattachment of C4SE9 cells observed during the initial90 min of culture in medium containing intact serum mayoccur by binding to adsorbed Vn (Fig. 1A). The higherattachment of 0 ^ 9 cells to Fn-TCP as compared to TCPduring 90 min was consistent with previous results (Steeleetal. 1990). Extension of the incubation to 16 h resulted inthe C4SE9 cells seeded in Vn-depleted medium becomingattached and spread on TCP (Fig. 1A,B). The morphologyof the cells attached to TCP under these culture conditionswas the same as C4SE9 cells seeded in medium containingintact serum. The requirement for protein synthesis byC4SE9 cells to permit this Vn-independent attachment toTCP was examined by testing the effect of the inhibitorcycloheximide upon attachment of cells in Vn-depletedmedium to TCP compared to attachment to Fn-TCP(Table 1). The presence of cycloheximide inhibited the Vn-independent attachment to TCP by 97 % whereas attach-ment to the control surfaces, Fn-TCP and Vn-TCP, wasinhibited by 49 % and 70 %, respectively.

Taken together, these results show that both C4E10 andC4SE9 cells can initially attach to TCP by binding toserum Vn adsorbed to TCP. C4SE9 cells differ from theC4E10 cells, however, in that C4SE9 cells are capable ofattachment and spreading on TCP by a Vn-independentmechanism that requires cellular protein synthesis.

Expression of vitronectin receptors on C4E10 and C4SE9cellsThe C4E10 and C4SE9 cells were compared for their

content of receptors for vitronectin, by immunoprecipi-tation experiments using polyclonal antiserum againstthe alphav subunit of human vitronectin receptor (Suzukietal. 1986). Immunoprecipitated extracts of metabolicallylabelled C4E10 cells had a major band at 152xlO3Mr anda minor band at 173xlO3Mr (Fig. ID, track 2). Extracts ofC4SE9 cells had a major band at 152xlO3Mr and faintbands at 173xlO3Mr and 142xlO3Mr (track 4). Theseexperiments show that both C4E10 and C4SE9 cellssynthesize and contain the alphav subunit of Afr 152 xlO3,consistent with the mechanism of the initial attachment ofthese cells to TCP being the binding to serum Vn adsorbedto the TCP.

Sensitivity of the attachment of cultured C4SE9 cells bythe Vn-independent mechanism to disruption bycytoskeletal-active drugsThe effect of cytoskeletal inhibitors on the attachment toTCP of C4SE9 cells by the Vn-independent and Vn-dependent mechanisms was tested, using two drugs thataffect the microtubule system (demicolcemid and nocod-azole) and an inhibitor of actin filaments, cytochalasin B.The drugs were tested for effects upon the initialattachment of metabolically labelled cells to the sub-stratum when added to the culture medium at the time ofseeding. In other experiments, any effects upon cellmorphology when the inhibitors were added to established

0.1 1Demicolcemid (/igmP

0.1 _ 1Nocadazole (/igml"1)

Fig. 2. Histograms showing the effect of demicolcemid (A) andnocodazole (B) on cell attachment to TCP (% of total cellsseeded) in medium containing intact serum. Diagonally stripedbars, C4SE9 cells incubated for 24 h; and open bars, C4E10cells incubated for 4h. Mean±s.E.M.

198 J. G. Steele et al.

cultures or at the time of seeding were observed. Theeffects of the drugs upon the attachment to TCP of thecontrol C4E10 cells were also tested in these experiments.

The addition of demicolcemid or nocodazole at the timeof seeding of C4SE9 cells in medium containing intactserum caused a concentration-dependent, significant butincomplete inhibition of attachment of C4 E9 cells to TCP(Fig. 2A and B). To test whether the partial inhibition bydemicolcemid or nocodazole of attachment to TCP ofC4SE9 cells may be related to the ability of these cells toattach to TCP by a Vn-independent mechanism, theattachment experiments were also carried out usingmedium containing Vn-depleted serum. With this me-dium, 1 j<g ml"1 demicolcemid inhibited Vn-independentattachment of C4SE9 cells to TCP by 78% whereasattachment from medium containing intact serum wasinhibited by 55 % (Fig. 3 and Fig. 2A). Addition ofdemicolcemid had no effect upon the attachment of C4SE9cells from either of the media to either Fn-TCP (Fig. 3) orVn-TCP (not shown). The morphology of C4SE9 cellsseeded on TCP in Vn-depleted serum was sensitive to theaddition to the culture medium after 24 h culture oflmgml" 1 demicolcemid (Fig. 4A cf. B), which causedretraction of the spread edges of marginally positionedcells. The demicolcemid caused a partial collapse of thefine filamentous network of tubulin microtubules (Fig. 5Cand D). On a Fn-TCP substratum, the shape of C4SE9cells seeded in the same medium was unaffected bydemicolcemid (Fig. 4C,D).

There was no effect of 25J/M cytochalasin B on the

•au

U

FCS

Fn-TCP

Fig. 3. Histogram showing the effect of seeding of C4SE9 cellsin medium containing liigml"1 demicolcemid (hatched bars) or25 /IM cytochalasin B (stippled bars) or no drugs (open bars)upon attachment (% of total cells seeded) to TCP or to Fn-TCP(as so marked). The cells were seeded in medium containingintact serum (FCS) or serum selectively depleted of Vn (—Vn)and cultured for a 24h period. Mean±s.E.M.

number of C4SE9 cells that attached to TCP when seededin medium containing intact serum (86% of cells attachedas compared to controls, mean of 4 experiments) and the

Fig. 4. Effect of demicolcemid on the morphology of C4SE9 cells grown in Vn-depleted medium. Cells were grown on TCP (A,B) orFn—TCP (C,D) for 24 h then l/zgrnl"1 demicolcemid (B,D) was added and the cells were photographed after 3h further culture. Bar,50 pan.

Epithelial cell adhesion and shape 199

Fig. 5. Effect of demicolcemid and cytochalasin B on distribution of tubulin and actin in C4E10 (A,B,E,G,H) and C4SE9 (C,D,F,I,Jjcells cultured on Fn-coated glass. Tubulin (A-D) was detected by indirect immunofluorescence using an anti-tubulin antibody;E and F show staining with preimmune serum. F-actin (G-J) was detected by staining with rhodamine-phalloidin. Note: there wasno detectable immunofluorescence at 514 nm (as used for detection of rhodamine-phalloidin) in unstained cells. Cells in B and Dwere seeded in medium containing l/igml"1 demicolcemid and cultured for 1 day; cells in H and J were treated with 25 /IMcytochalasin B for 3 h before staining. Bars: A-F, 15 /on; G-J, 50 /on.

Vn-independent attachment to TCP was not inhibited byaddition of cytochalasin B to the culture medium (Fig. 3).When cytochalasin B was added to the culture medium atthe time of cell seeding, the spreading of marginallypositioned C4SE9 cells on TCP, Fn-TCP or Vn-TCPsubstrata was inhibited (not shown). Addition of 25 /.CMcytochalasin B to the culture medium for 3h causedretraction of the edges of marginally positioned C4SE9cells and collapse of the F-actin stress fibres (Fig. 5I,J).

The attachment of C4E10 cells to TCP was not inhibitedby the addition of demicolcemid or nocodazole to theculture medium at the time of seeding (Fig. 2 A and B). Theshape of C4E10 cells cultured on either TCP or Fn-TCPwas sensitive to the presence of demicolcemid in theculture medium. This sensitivity to demicolcemid wasparticularly evident when the cells were cultured in thepresence of the drug and on a Fn-glass substratum forimmunostaining of tubulin distribution (Fig. 5A vs B).There was no effect of 25 /IM cytochalasin B on the number

of C4E10 cells that attached to TCP (89 % of cells attachedas compared to controls, mean of 3 experiments). Additionof lfiM cytochalasin B to established cultures of C4E10cells caused retraction of the cell borders and rounding upof the cells during 1-3 h of treatment and disruption of theF-actin stress fibres (Fig. 5G,H). C4E10 cells that attachedto TCP or Fn-TCP in the presence of cytochalasin B didnot display the shape polarity that was seen in the controlcells (not shown).

Taken together, the results of the drug experimentsshow that the C4SE9 cells differ from the C4E10 cells inthe effect of disruption of their microtubule system oninitial cell attachment to TCP. The initial attachment ofC4SE9 cells to TCP was sensitive to disruption of themicrotubules. The effects of the inhibitors were differentfor the different mechanisms of their attachment to TCP:the initial attachment of C4SE9 cells to TCP by a Vn-independent mechanism was almost completely inhibitedby poisoning of the microtubules whereas the attachment

200 J. G. Steele et al.

of C4SE9 cells to either Fn-TCP or Vn-TCP was notinhibited by disruption of the microfilaments. The initialattachment to TCP of C4E10 cells (which must attach toTCP by a Vn-dependent mechanism) was unaffected bymicrotubule-active drugs.

Discussion

The NAL1A alveolar cell system is an attractive cellmodel for studies of the mechanism of neoplastic trans-formation of epithelial cells, as the transformed cells maybe compared to nontransformed cells derived from thesame clonal cell source. The transformed cells arose fromthe clonal control cells as the result of a spontaneousevent, initially evident as a spontaneous morphologicalchange during culture; the resultant transformed cellshave the stable phenotype of cells that attach to TCP asislands, with the marginally positioned cells showing poorspreading on the culture substratum (Smith and Lykke,1985; Smith et al. 1986; Steele et al. 1988, 1990). Inprevious studies of the altered morphology of the trans-formed C4SE9 cells and clonally related control C4E10cells, it has been shown that the C4E10 cells grown on TCPcan assemble an extracellular matrix containing Fn andlaminin, whereas the C4SE9 cells do not. This failure ofC4SE9 cells to secrete Fn and laminin is not the sole factorcausing the altered shape of these malignant cells,

however, as these cells retain a poorly spread morphology(as compared to C4E10 cells) when seeded onto Fn-,laminin- or extracellular matrix-coated substrata (Steeleet al. 1988, 1990). In this study we have examined themechanism by which the control and transformed cellsinitially attach to TCP, and have used cytoskeletallydisruptive drugs to look for differences in thecytoskeletal-cell surface-TCP substratum linkages ofthese cells.

The attachment of C4E10 cells to TCP occurs during thefirst 90 min of culture and is due to cell binding to serumVn adsorbed to the substrata. This mechanism is the sameas that demonstrated for the initial attachment to TCP offibroblasts and endothelial cells (Underwood and Bennett,1989; Norris et al. 1990; Steele et al. 1991). The selectivedepletion of Vn from the serum completely inhibited theattachment of C4E10 cells to TCP, indicating that it is Vn,rather than any other serum-adhesive protein, thatbecomes adsorbed to TCP in sufficient quantity to permitinitial attachment of C4E10 cells. It was shown previouslythat in the presence of serum concentrations greater than1 % (v/v), adsorption of Fn to TCP is inhibited and at 10 %(v/v) very little Fn adsorbs to TCP (Grinnell and Feld,1982; Underwood and Bennett, 1989). As a result of thisinhibition, the relative amounts of Fn and Vn that adsorbto TCP during a 90 min incubation with medium contain-ing 10 % (v/v) FCS are 38ngcm"2 Fn and 227 ngcm"2 Vn(J. G. Steele, G. Johnson and P. A. Underwood, unpub-

Epithelial cell adhesion and shape 201

lished data). This initial attachment of C4E10 cells toadsorbed serum Vn was not inhibited by actin- ormicrotubule-active drugs, although other studies suggestthat integrin receptors are linked at the cytoplasmic faceof the plasma membrane to the actin filaments, eitherdirectly or indirectly (Horwitz et al. 1986; Burridge et al.1988; Otey et al. 1990). Following the initial binding step,C4E10 cells spread rapidly and quite extensively on thesubstratum. This spreading of attached C4E10 cells wassensitive to disruption of either the actin filaments (bycytochalasin B) or the microtubules (by nocodazole ordemicolcemid). Similar sensitivity of the shape of culturedepithelial cells to microtubule-disruptive drugs has beenobserved with other cell lines (Domnina et al. 1985) but notwith primary cultures (Middleton et al. 1988), suggestingthat the importance of microtubules for the shape ofepithelial cells may be acquired as a result of adaptation tocell culture (see Middleton et al. 1988, for discussion).

The attachment of C43E9 cells to TCP can occur by bothVn-dependent and Vn-independent mechanisms. Theattachment of transformed C4SE9 cells to TCP occursrelatively slowly compared with that of C4E10 cells(Steele et al. 1990 and this study) and those C4SE9 cellsthat attach within the first 90 min bind to Vn adsorbed toTCP. We show here that with incubation beyond 90 minthe C4SE9 cells can attach by a Vn-independent mechan-ism. This Vn-independent mechanism is dependent uponprotein synthesis, but other biochemical details of themechanism remain to be determined. It seems likely thatthe Vn-independent mechanism requires cellular se-cretion of some adhesive molecule, as we have observedthe Vn-independent attachment reaction to BSA-blockedTCP in serum-free medium (data not shown). In demon-strating this ability of C4SE9 cells to attach in a Vn-independent reaction, we show that there is a cleardifference between the control and transformed NAL1Acells in the mechanism of attachment to TCP in thepresence of FCS. In experiments with microtubule inhibi-tors, the Vn-independent attachment of C4SE9 cells toTCP was highly sensitive to inhibition by the microtubule-active drugs demicolcemid and nocodazole whereas attach-ment to a Fn- or Vn-coated surface was not. Furthermore,whereas the shape of the C4SE9 cells attached to TCP bythe Vn-independent mechanism was quite sensitive todemicolcemid, when these cells were cultured on Fn-TCPor Vn—TCP their shape was relatively insensitive todemicolcemid. It is not known whether the mechanism ofthe sensitivity of cell attachment to inhibitors of micro-tubules is the same as that underlying the Vn-indepen-dent attachment of C4SE9 cells to TCP. One hypothesis toexplain the inhibition could be that there is a directlinkage between the microtubules and a cell surfaceadhesive molecule. Alternatively, the Vn-independentattachment reaction and the sensitivity of that reaction tomicrotubule-disrupting drugs may be due to differentmechanisms, arising simultaneously or sequentially.Furthermore, an indirect drug effect due to inhibition oftransport of newly synthesized proteins to the cell surfacecannot be excluded. The partial inhibition by demicolce-mid of attachment of C4SE9 cells to TCP from normalculture medium compared to the more complete inhibitionof attachment from Vn-depleted medium (Fig. 3) suggeststhat these two attachment mechanisms may be of similarrelative importance for attachment of C4SE9 cells to TCPduring the first 24 h of culture in normal medium(containing intact serum). It is difficult to design mean-ingful experiments to determine the relative importance

of the two attachment mechanisms for C4SE9 cells, due tothe relatively slow kinetics of attachment of C4SE9 cells,the requirement for trypsin treatment during preparationof cell suspensions and the difficulty of separating the twomechanisms of attachment of C4SE9 cells.

Studies comparing control and transformed fibroblastsfor synthesis and content of cell surface fibronectinreceptors have reported no effect of transformation ordecreased expression of specific receptor species in thetransformed cells (Plantefaber and Hynes, 1989; Ylanneand Virtanen, 1989; Akiyama et al. 1990). The C4E10 andC4SE9 cells synthesize the'Mr=152xl03 alphav subunit ofthe vitronectin receptor, a member of the integrin familyof cell surface receptors (Suzuki et al. 1986; Pytela et al.1986; Bodary and McLean, 1990). This is consistent withthe sensitive ability of C4E10 cells to attach to Vn-coatedsubstrata, and the requirement for serum Vn by these cellsin order to attach to TCP. The results of the immuno-precipitation studies with these C4E10 and C4SE9 cells donot suggest that there is a complete down-regulation ofsynthesis of the Vn receptor in the transformed cellscompared to the control, but it is not possible to obtainreliable quantitative data on cell receptor numbers by thistechnique. In further work, it will be of interest todetermine whether integrin receptors are involved inC4SE9 cell attachment to TCP by the Vn-independentreaction; however the design of these experiments needs totake into account both the technical difficulties outlinedabove and the requirement that any antisera used shouldbe cross-reactive with murine receptors (and so mono-clonal antibodies are unlikely to be useable).

Numerous studies have pointed to a 'switch off of thesynthesis of extracellular matrix components upon neo-plastic transformation of fibroblasts, and a similar effect isobserved in the NAL1A epithelial cell system (Alitalo andVaheri, 1982; Hayman et al. 1982; Ruoslahti, 1984; Steeleet al. 1988, 1990). The specific aim of this study was tocompare the mechanisms of attachment and shape controlof the transformed and control NAL1A cells on TCP, inorder to identify the mechanism that altered the mor-phology of the transformed cells (Smith and Lykke, 1985;Smith et al. 1986; Steele et al. 1990). The study shows thatthe transformed NAL1A cells have another attachmentmechanism, in addition to those described for the controlcells. The working hypothesis that arises from theseresults is that the transformed NAL1A cells synthesize acell surface attachment molecule and that the expressionof this (as yet unidentified) molecule by these transformedcells permits attachment to TCP without a requirementfor exogenous attachment factors such as serum Vn or Fn.We suggest that this Vn-independent, microtubule-relatedattachment mechanism may be responsible for the alteredshape and spreading behaviour of the transformed cellscompared to the controls. In further studies in this cellsystem we shall work towards the identification of theattachment molecule, which should make it possible todetermine the exact role that it plays in the control of theshape and growth of the transformed NAL1A cells and todetermine whether it is present in normal tissue and othercarcinoma cells.

This work was supported in part by grants from the NationalHealth and Medical Research Council of Australia (to G.J.S.), theState Cancer Council of New South Wales (to G.J.S.) and theCSIRO-University of New South Wales Research Grants Scheme(to J.G.S. and G.J.S.). We thank G. Johnson and F. Bennett forcritical review of the manuscript.

202 J. G. Steele et al.

ReferencesAKIYAMA, S. K., LARJAVA, H. AND YAMADA, K. M. (1990). Differences in

the biosynthesis and localization of the fibronectin receptor in normaland transformed cultured human cells. Cancer Res 50, 1610-1607.

ALITALO, K. AND VAHERI, A. (1982) Pericellular matrix in malignanttransformation. Adv. Cancer Res. 37, 111-158

BEN-ZE'EV, A. (1985). The cytoskeleton in cancer cells Biochim. biophys.Ada 780, 197-212.

BODARY, S. C. AND MCLEAN, J. W. (1990). The Integrm beta one subunitassociates with the vitronectin receptor alphav subunit to form anovel vitronectin receptor in a human embryonic kidney cell line. J.biol. Chem. 265, 5938-5941.

BURRIDGE, K., FATH, K., KELLY, T., NUCKOLLB, G AND TURNER, C.(1988). Focal adhesions: transmembrane junctions between theextracellular matrix and the cytoskeleton. A. Rev. Cell Biol. 4,487-526.

DOMNINA, L. V., ROVENSKY, J. A., VASIUEV, J. M AND GELFAND, I. M.(1986). Effect of microtubule-destroying drugs on the spreading andshape of cultured epithelial cells. J. Cell Sci. 74, 267-282.

GRINNELL, F AND FELD, M. K. (1982). Fibronectin adsorption onhydrophihc and hydrophobic surfaces detected by antibody bindingand analyzed during cell adhesion in serum-containing medium J.biol. Chem. 254, 4888-4893.

HAYMAN, E. G., OLDBERG, A , MARTIN, G. R. AND RUOSLAHTI, E (1982)Codistribution of heparan sulfate proteoglycan, laminin andfibronectin in the extracellular matrix of normal rat kidney cells andtheir coordinate absence in transformed cells J. Cell Biol. 94, 28-35.

HORWITZ, A., DUGCAN, K., BUCK, C , BECKERLE, M. C. AND BURRIDGE, K.(1986). Interactions of plasma membrane fibronectin receptor withtalin - a transmembrane linkage. Nature 320, 531-633

MIDDLETON, C. A., BROWN, A. F., BROWN, R. M. AND ROBERTS, D J H.(1988). The shape of cultured epithelial cells does not depend on theintegrity of their microtubules. J. Cell Sci. 91, 337-346.

NORRI8, W. D., STEELE, J. G., JOHNSON, G. AND UNDERWOOD, P. A.(1990). Serum enhancement of human endothehal cell attachment oncollagens I and FV does not require serum fibronectin or vitronectinJ. Cell Sci 95, 265-262

OTEY, C. A., PAVALKO, F. M. AND BURRIDOE, K (1990). An interactionbetween alpha-actinin and the beta-one Integrin subunit in vitro. J.Cell Biol. I l l , 721-729.

PLANTEFABBR, L C. AND HYNES, R. O. (1989). Changes in lntegnnreceptors on oncogenically transformed cells. Cell 56, 281-290.

PYTELA, R., PIERSCHBACHER, M. D., GINSBERG, M. H., PLOW, E. F ANDRUOSLAHTI, E. (1986). Platelet membrane glycoprotein Hb/IIIa:

member of a family of arg-gly-asp-speciflc adhesion receptors. Science231, 1659-1562.

RUOSLAHTI, E. (1984). Fibronectin in cell adhesion and invasion. CancerMetast Rev. 3, 43-51.

SMITH, G. J., BENNETT, F. A , STEELE, J. G. AND BKNTEL, J. M (1986).Onset of neoplastic phenotype in an epithelial cell strain from adultBALB/c mouse lung alveolus. J. natn. Cancer Inst. 76, 73-79.

SMITH, G. J. AND BENTEL, J. M. (1987) Malignant epithelial cell strainscultured from BALB/c mouse lung adenoma Cell Biol. Int. Rep. 11,111-118

SMITH, G. J. AND LYKKE, A. W. J. (1986). Characterization of aneoplastic epithelial cell strain derived by dexamethasone treatmentof cultured normal mouse type 2 pneumocytes. J. Path. 147, 165-172.

STEELE, J. G., JOHNSON, G., NORRIS, W. D. AND UNDERWOOD, P. A.(1991). Adhesion and growth of cultured human endothelial cells onperfluorosulphonate: role of vitronectin and fibronectin in cellattachment. Biomatenals (in press).

STEELE, J. G., SAVOLAINEN, T. A., DALTON, B. A., SMITH, L. A. ANDSMITH, G. J. (1990). Adhesion to laminin and expression of laminin inclonally-related transformed and control sublines from an alveolarepithelial cell strain. Cancer Res. 50, 3383-3389.

STEELE, J. G., SAVOLAINEN, T. A. AND SMITH, G. J. (1988). Expression of

fibronectin on clonally related transformed and control sublines froman epithelial cell strain and a tumor line of mouse alveolus. CancerRes. 48, 4933-4940.

SUZUKI, S., ARGRAVES, W. S., PYTELA, R., ARAI, H., KRUSIUS, T.,PIERSCHBACHER, M. D. AND RUOSLAHTI, E. (1986). cDNA and aminoacid sequences of the cell adhesion protein receptor recognizingvitronectin reveal a transmembrane domain and homologies withother adhesion protein receptors. Proc natn. Acad. Sci. U S.A. 83,8614-8618.

UNDERWOOD, P. A. AND BENNETT, F. A. (1989). A comparison of thebiological activities of the cell-adhesive proteins vitronectin andfibronectin. J. Cell Sci. 93, 641-649.

VASIUEV, J. M. (1986). Spreading of non-transformed and transformedcells Biochim. biophys. Ada 780, 21-65.

YLANNE, J AND VIRTANEN, I. (1989). The Mr 140,000 fibronectinreceptor complex in normal and virus-transformed human fibroblastsand in fibrosarcoma cells: identical localization and function. Int. J.Cancer 43, 1126-1136.

{Received 10 December 1990 - Accepted, in revised form, 12 June 1991)

Epithelial cell adhesion and shape 203