J. Cell Sd. 71, 267-282 (1985) 267Printed in Great Britain © The Company of Biologists Limited 1985

EFFECT OF MICROTUBULE-DESTROYING DRUGSON THE SPREADING AND SHAPE OF CULTUREDEPITHELIAL CELLS

L. V. DOMNINA1, J. A. ROVENSKY2, J. M. VASILIEV2 AND I. M.GELFAND1

'A. N. Belozersky Laboratory of Molecular Biology and Bioorganic Chemistry, MoscowState University, Moscow GSP-234, 119899 U.S.S.R.2All-Union Oncological Research Center of the Academy of Medical Sciences, Moscow,U.S.SM.

SUMMARY

The role of microtubules in the spreading of cells from the liver-derived IAR2 rat cell line wasstudied. Cells in control medium seeded on a flat isotropic glass surface rapidly spread to formdiscoid shapes. Spreading in colcemid-containing medium was disorganized and delayed; partialreversal of spreading was observed. Nevertheless, even in the presence of colcemid the cells finallyspread to discoid flattened shapes. IAR2 cells in medium without colcemid spread not to discoid butto elongated shapes under three different sets of conditions: (1) when the cells were forced to spreadon narrow strips of adhesive glass surface between two non-adhesive lipid films; (2) when the cellsspread on the poorly adhesive surface of poly(HEMA)-covered glass; (3) when the cells spread onthe usual glass surfaces in medium containing cytochalasin D. Addition of colcemid to the mediareversed the polarized spreading under the first two conditions; colcemid did not reverse theformation of the elongated cell shape acquired by the cells spreading in cytochalasin-containingmedium. Effects of microtubule-destroying drugs on the spreading of epithelial and fibroblast cellsare compared and discussed. It is suggested that microtubules are essential for the stabilization ofthe spread state of those attached cytoplasmic processes and lamellae that do not have numerous andstable cell-substratum contacts, e.g. the processes formed at the early stages of spreading or theelongated processes of polarized cells. Possibly, microtubules stabilize the non-contracted state ofthe actin cytoskeleton in these processes.

INTRODUCTION

The role of microtubules in the development and maintenance of the shape of tissuecells at interphase was studied in detail in experiments with elongated polarized cells,such as fibroblasts or neurons. It has been found that depolymerization ofmicrotubules by specific drugs prevents or reverses polarization of these cells. Inparticular, fibroblasts become unable to achieve and to maintain elongated shapes;differentiation of their edges into pseudopodially active and stable zones disappears(Vasilievef al. 1970; Vasiliev & Gelfand, 1976). Elongated cytoplasmic processes ofneuronal cells also disappear (Magendantz & Solomon, 1981). The role of themicrotubular system in the morphogenesis of the cells of the other main morphologi-cal type, epitheliocytes, is much less clear. These cells spread to discoid shapes on theusual isotropic culture substrata. Although the spread epithelial cells, like fibroblasts,

Key words: microtubules, cell shape, cell spreading.

268 L. V. Domnina, J. A. Rovensky, J. M. Vasiliev and I. M. Gelfand

contain a well-developed system of microtubules occupying almost all their cytoplasm(Bershadsky et al. 1978), microtubule-destroying drugs have not been observed tocause any striking changes in their shape (Di Pasquale, 1975; Domnina, Pletyush-kina, Vasiliev & Gelfand, 1977). The aim of experiments described in this paper wasto examine the effects of a microtubule-destroying drug, colcemid, on the spreadingof individual epithelial cells under various conditions. We have chosen IAR2 cellsderived from rat liver for our experiments because these cells have a typical epithelialmorphology that has been described in detail (Montesano, Saint Vincent, Drevon &Tomatis, 1975; Montesano et al. 1977; Bannikov, Saint Vincent & Montesano, 1980;Bannikov et al. 1982). In contrast to many other epithelial cells, which spread wellonly when they are in contact with one another (Middleton, 1976), single IAR2 cellsspread easily on glass. Therefore, in experiments with this cell line it is easy to observealterations in the morphology of individual cells in the course of spreading without thecomplicating effects of cell—cell contacts. We have examined the effect of colcemid onthe final shape of spreading IAR2 cells and on the intermediate stages of their spread-ing. The effects on cells spreading on the usual substratum were compared with thoseexerted by colcemid under special conditions, in which epithelial cells were forced toundergo polarized spreading and to acquire not discoid but elongated shapes. Al-though, in agreement with previously published results, colcemid did not cause anystriking alteration of well-spread discoid cells, this drug, under certain other con-ditions, was found to alter the morphology of epithelial cells as profoundly as that offibroblasts; this happened, for instance, at the early stages of spreading on the usualsubstratum or in certain cases of forced polarized spreading. These data give reasonto suggest that microtubules have an important common function in the spreading ofcells of different tissue types; the possible nature of this function is discussed in thelast part of this paper.

MATERIALS AND METHODS

Cell line and growth conditionsThe IAR2 epithelial cell line originally obtained from normal rat liver was used (see Introduc-

tion). The cells are not tumorigenic in syngeneic rats. The cell line was cultivated in William's Emedium (Flow Labs, Irvine, Scotland), supplemented with 10% foetal calf serum (Gibco-Biocult,Glasgow, Scotland) and lOOIU/ml monomycin. For scanning electron microscopy (SEM) andindirect immunofluorescence the celta were plated at an initial density of 2 X 104 to 5 X 104 cells/cm2

on coverslips placed in Petri dishes and were cultured at 37 CC in a humidified incubator suppliedwith 5 % CO2 in air. For phase-contrast or interference-reflection microscopy the cells were platedat an initial density of 2 X lC^/cm2 and living cells were viewed and photographed at 37 °C in glasschambers as described by Vasiliev et al. (1970).

Colcemid (Serva, FRG) at 0-05-0-2 /Jg/ml or cytochalasin D (Serva, FRG) at 1-2 jig/ml wereadded to the culture medium at the time of cell plating, or at different intervals after cell seeding.Contact interaction of the cells with substrate surfaces of different adhesiveness was studied inexperiments in which the coverslips were coated with a film of poly(2-hydroxyethyl methacrylate)(poly(HEMA); Hydron Lab., New Brunswick, N.J., U.S.A.). As shown by Folkman & Moscona(1978) the adhesiveness of plastic tissue culture can be reduced in a graded manner by applyingincreasing concentrations of poly(HEMA). We used serial dilutions (1/2000, 1/1000, 1/600 or 1/250) of the ethanol/poly(HEMA) solution for coating the coverslips. For examination of cell

Microtubule-destroying drug, effect on epithelium 269

orientation and cell elongation the cultures were plated on coverslips coated with a non-adhesive thinfilm of lecithin, prepared from lecithin solution (2mg/ml), which was a kind gift from Dr L. B.Margolis (this laboratory). Narrow linear scratches on the glass surface were made in the lecithinfilm with a sharp needle as described by Ivanova & Margolis (1973).

Scanning electron microscopyThe cells were fixed in sodium cacodylate-buffered isotonic 2% glutaraldehyde (pH 7-2). After

washing, the specimens were dehydrated in increasing concentrations of the water/acetone solutionand dried in a critical-point drier (Balzere Union, Liechtenstein). The cells were coated with gold/palladium in a cold sputter (Polaron Equipment Ltd, England) and examined in a CambridgeStereoscan S4 microscope.

Antibodies and immunqfluorescence microscopyMonospecific rabbit antibodies against chicken gizzard actin and bovine brain tubulin have been

characterized elsewhere (Bershadsky, Gelfand, Svitkina & Tint, 1980). Monoclonal antibodiesagainst vimentin, clone NT30, were kindly provided by Dr A. A. Neifakh and Dr I. S. Tint (AJ1-Union Cancer Research Center, Moscow). For indirect immunofluorescence microscopy, cultureswere washed thoroughly in phosphate-buffed saline (PBS) and fixed with 4 % formaldehyde in PBS.Before fixation the cells were extracted for 3 min with 1 % Triton X-100 in imidazole buffer, asdescribed earlier (Bershadsky et al. 1980). Observations were made with a Zeiss PhotomicroscopeIII (Carl Zeiss, FRG) equipped with epifluorescent illumination and a Planapo 40 X oil-immersionobjective.

Interference-reflection microscopy (IRM) was carried out according to Izzard & Lochner (1976)using living .cultures incu,bated at 37°C. Observation* were made with a Photomicroscope IIIequipped with accessories for interference contrast and a Planapo 40 X oil-immersion objective.

RESULTS

Spreading in control medium

Control cells spreading on glass (Fig. 1) formed the first narrow outgrowths(filopodia), seen by SEM and by phase-contrast, at 5—7 min after the first contact withthe substratum. Soon after that, at 7-10min, numerous flattened lammelipodia(ruffles) were formed around the cell body, which at this stage was still almost spherical.Formation and gradual expansion of the ring of lamellar cytoplasm from thesubstratum-attached ruffles took place during the next 20-40 min. Ruffling continuedat the peripheral edges of this ring; circular ruffles moving centripetally across theupper surface of the lamellar cytoplasm were often seen during this period. Simul-taneously with the widening of the ring of lamellar cytoplasm, gradual flattening ofthe central cell body took place. At 60 min most cells had already acquired the finaldiscoid shape with a flat upper surface covered only with short microvilli. Rufflesformed at the edges of these discoid cells were smaller than those seen at earlier stagesof spreading; 24 h later the cells retained the same shape, except that the ruffling hadstopped almost completely and the outer edges had acquired smooth circular shapes.After divisions the daughter cells remained in close contact with one another so thatcoherent small cell groups were formed. Any significant translocation of single cellsor of cell groups on the substratum was not observed. By IRM, first small dot-likefocal contacts were seen at the periphery of the cells at 10-30 min after seeding; thefully spread discoid cells had a regular ring of black or dark-grey contacts at the

270 L. V. Domnina, J. A. Rovensky, J. M. Vastliev and I. M. Gelfand

Fig. 1

Microtubule-destroying drug, effect on epithelium 271

periphery; some contacts forming this ring were dot-like, while Others were linear andoriented tangentially to the edge (Fig. 2). Circular bundles of microfilaments wererevealed by anti-actin antibody in the lamellar cytoplasm of discoid cells; straightbundles crossing the internal parts of the cytoplasm of these cells were seen onlyrarely. In the cell groups peripheral marginal belts of microfilament bundles werebetter developed near contact-free lateral edges than near internal edges (Fig. 3).Microtubules radiated from the perinuclear zones of discoid cells into their lamellarcytoplasm and often formed arcs there (Fig. 4). Peripheral parts of the microtubulesoften formed marginal bundles almost parallel to the cell edge. Intermediate filamentsrevealed by antibody against vimentin formed a loose network in the cytoplasm (Fig.5A). AS shown in our laboratory, the IAR2 line, in contrast to many other epithelialcells lines, did not have any intermediate filaments containing pre-keratin (Troyanov-sky, personal communication).

The cells spreading on narrow strips of glass between non-adhesive phospholipidfilms, at 24 h after seeding acquired highly elongated shapes with small lamellae at oneor both ends and with badly spread fusiform central bodies (Fig. 6A,B). Ruffling wasseen at the edges of the lamellae of these cells; their lateral edges were inactive. ByIRM diffuse zones of close contacts were seen at various sites on the ventral surfacesof these cells, with occasional small black zones near the edges.

The cells seeded on poly(HEMA) diluted 1:2000 or 1:1000, 24 h later had the sameshapes as the cells spread on glass (Fig. 7A) . The degree of spreading at 24 h decreasedwhen the cells were seeded on poly(HEMA) diluted 1:600, and especially onpoly(HEMA) diluted 1:250. At 24 h most cells on this last substratum had poorlyspread dome-like or elongated fusiform shapes (Fig. 7B).

Fig. 1. Consecutive phase-contrast micrographs of an IAR2 cell spreading on glass incontrol medium. Time after seeding: A, 12min; B, 23min; c, 25min; D, 28min; E,41 min; F, 1 h 15min. Bars, 20/tfn.

Fig. 2. The ring of contacts of a spread discoid cell, 24 h after seeding in control medium:A, phase-contrast; B, interference-reflection microscopy. Bars, 20/tfn.

Fig. 3. Peripheral microfilament bundles in a group of cells spread in control medium 24 hafter seeding. Indirect immunofluorescence using anti-actin antibody. Bar, 30/tfn.

Fig. 4. Microtubules in the control cell, 24 h after seeding. Indirect fluorescence usinganti-tubulin antibody. Bar, 20/tfn.

Fig. 5. Indirect immunofluorescence using antibody against vimentin; 24 h after seeding:A. Cell in control medium. Network of intermediate filaments in the cytoplasm; B, cellincubated with 0-l /ig/ml colcemid for 24 h. A ring of collapsed filaments can be seenaround the nucleus. Bare, 10 fun.

Fig. 6. A,B. Elongated IAR2 cells spread on narrow strips of glass surrounded by non-adhesive phospholipid film. Phase-contrast; 24 h after seeding. Bars, 30 ftm.

Fig. 7. Cells spread on poly(HEMA)-covered substrate, 24 h after seeding. Scanningelectron microscopy: A. Well-spread discoid cell in poly(HEMA) diluted 1:2000; B, poor-ly spread elongated cell on poly(HEMA) diluted 1:250. Bars, 15 (im.Fig. 9. Cells in control medium (A) and in colcemid-containing medium (0-1 /^g/ml) (B),1 h after seeding. Scanning electron microscopy. Notice numerous blebs at the surface ofthe colcemid-treated cell. Bare: A, 15 jim; B, 5 [an.

272 L. V. Domnina, J. A. Rovensky, jf. M. Vastltev and I. M. Gelfand

2A

Figs 2-5. For legend see p. 271

Microtubule-destroying drug, effect on epithelium 273

Figs 6—7, 9. For legend see p. 271

274 L. V. Domnina, Jf. A. Rovensky, Jf. M. Vasiliev and I. M. Gelfand

Fig. 8

Microtubule-destroying drug, effect on epithelium 275

Effects of colcemid

The cells spreading on the usual glass substrate in colcemid-containing medium(Figs 8, 9B) began to form numerous blebs at their periphery at 5-60 min after contactwith the substrate. Most blebs were 5-10 /im in diameter; large elongated balloon-likeprocesses (lobopodia) were also formed. The blebs appeared and disappeared veryquickly; some of them existed for only a few minutes. Besides the blebs, attachedlamellipodia were seen at the cell periphery at 1-1*5 h; formation of the narrowcircular ring of lamellar cytoplasm from these lamellipodia took place at 1*5-2*0 hafter seeding, that is, much later than in control cultures. At 2-4 h large blebs wereoften seen at the upper surfaces of lamellar rings. In contrast to the controls, thespreading of these cells often underwent partial reversal: some parts of the lamellarring detached themselves from the substrate and contracted. In spite of these reversalsthe average degree of cell spreading gradually increased. At 6—8 h after seeding mostcells had acquired relatively well-spread discoid shapes; further reversal of spreading

Fig. 8. Consecutive phase-contrast photographs of a cell spreading on glass in mediumcontaining 0-1/ig/ml colcemid. Time after seeding: A, 30min; B, l h l S m i n ; c,1 h30mLn;D, 1 hSOmjnjE, 4h;F, 4 h 30 min. Formation of blebs and of lobopodia (A,B,C);partial retraction of the lamellae on the left side (D) ; final flattening (E,F) . Compare withFig. 1. Bare, 20 /an.

Fig. 10. Consecutive phase-contrast photographs of a cell in control medium (A) and1-5 h after the transfer into the medium with 0-2/ig/ml of colcemid (B) show slightretraction and peripheral thickening of lamellae; thickening of central part of the cell.Bars, 20 Jan.

Fig. 11. A cell spread for 24h in colcemid-containing medium: A, phase-contrast; B,interference-reflection microscopy. Compare with Fig. 2. Notice irregular pattern andshape of peripheral contacts. Bare, 20/an.

Fig. 12 A cell spread on a narrow strip of adhesive substratum in medium with colcemid(0-2/ig/ml). Phase-contrast; 24h after seeding. Compare with Fig. 6. The cell is notelongated; numerous blebs are seen at the edges. Bar, 10/on.

Fig. 13. Branching cytoplasmatic strands of a cytochalasin-treated cell. The cell wasspread for 24 h in control medium on the usual substratum and then incubated withcytochalasin D (2/ig/ml for 2h). Phase-contrast. Bar, 20 /an.

Fig. 14. Branching cytoplasmatic strands of a cell, spread for 24 h on a narrow strip ofglass and then incubated for 2 h in medium with cytochalasin D (2 /^g/ml). Phase-contrast.Compare with Fig. 6. Bar, 25 /an.

Fig. 15. Elongated cells spread for 24 h in medium containing 2 /ig/ml of cytochalasin D.Phase-contrast. Bar, 30 /on.Fig. 16. Consecutive phase-contrast photographs of an elongated cell spread for 24 h inmedium with cytochalasin D (2 /ig/ml) before (A) and 3 h after (B) the addition ofcolcemid (0-2//g/ml) to this medium. The elongated process did not disappear. Bar, 20/an.Fig. 17. Thinned cytoplasmatic process of a cell spread for 24 h in medium withcytochalasin D (2 /^g/ml) and then transferred for another 24 h into the medium contain-ing both cytochalasin D and colcemid (0-1 /ig/ml). Phase-contrast. Bar, 20/an.Fig. 18. A cell of ellipsoid shape that was spread for 24 h in medium with cytochalasin D(2/ig/ml) and then transferred for another 24 h into control medium. Phase-contrast.Bar, 30/an.

276 L. V. Domntna, J. A. Rovensky, J. M. Vasiliev and I. M. Gelfand

Figs 10-12. For legend see p. 275

Microtubule-destroying drug, effect on epithelium 111

Figs 13-18. For legend see p. 275

278 L. V. Domnina, J. A. Rovensky, J. M. Vasiliev and I. M. Gelfand

was not observed. Discoid cells seen at 16-24 h in colcemid-containing medium oftenhad certain morphological differences from the discoid cells seen at the same times incontrol cultures: (1) the contours of their outer edges usually were not smooth butindented; they had numerous short flat extensions up to 5 [im in length and 1-5 /imin width. Occasionally blebs and lobopodia were also seen at the edge; (2) the lamellarcytoplasm was somewhat smaller in size; it was thicker near the edges than in morecentral parts; (3) the central part of the cell body was less flattened and more elevatedthan in controls.

When cells spread in control medium for 24 h were transferred into colcemid-containing medium, during the next 1—2 h they became indistinguishable from cellskept continuously in this last medium: their contours became less smooth, the areaof lamellae became somewhat decreased and their central part became thicker (Fig.10A,B). Removal of colcemid from the medium lead to restoration of normal morphol-ogy. By IRM, discoid cells in colcemid-containing medium, like control cells, had aring of contacts but the width of this ring was wider and more variable (Fig. 11).Increased variability in the width of the marginal ring of actin was also observed.Incubation of colcemid-treated cells with anti-tubulin antibodies did not reveal anymicrotubules in the cytoplasm. Most intermediate filaments revealed by anti-vimentin antibody collected around the nucleus forming brightly fluorescent rings(Fig. 5B).

Effects of vinblastin (0-05-0-2 fig/ml) on the course of spreading were similar tothose of colcemid. The cells seeded in colcemid-containing medium on narrow stripsof glass between two lipid films did not spread to form elongated shapes; at 24 h thesecells still had rounded, almost unspread, shapes with numerous blebs at the periphery(Fig. 12). When cells spread in control medium on narrow strips were transferred intothe colcemid-containing medium, rapid detachment of their elongated processes tookplace; and the cells also acquired rounded shapes. The elongated shapes were restoredafter transfer into drug-free medium. The cells on the substrate covered with 1/250poly(HEMA) also did not spread to elongated shapes in colcemid-containingmedium. At 24 h most cells remained almost spherical and could be easily washed offthe substrate.

Effects of cytochalasin D

Several types of experiments with cytochalasin D were performed.(1) Cells were spread for 24 h on glass in control medium and transferred into

medium containing cytochalasin D (CD-medium). During the next hour their discoidlamellae contracted leaving a network of branching cytoplasmic strands on the sub-stratum; this network surrounded an elevated central cell body (Fig. 13).

(2) Cells were spread for 24 h on narrow strips of glass in control medium and thentransferred into CD-medium. During the next 1-2 h two peripheral lamellae of theseelongated cells were transformed into a tree-like system of thin cylindrical processes(Fig. 14).

(3) Suspended cells were seeded on the usual glass substratum in CD-medium.During the first 2-4 h of spreading these cells formed several narrow elongated

Microtubule-destroying drug, effect on epithelium 279

processes at opposite edges of the cell body. Later these processes increased in lengthso that by 24-48 h the cells had acquired elongated dipolar shapes (Fig. 15).Flattened lamellae were not formed at the edges of these cells; by IRM, numerousdot-like dark-grey and black contacts were seen at the lower surfaces of these cells.Addition of colcemid (0-1 /ig/ml) to CD-medium 24 h after seeding did not lead tothe disappearance of elongated processes 2-4 h later, although the thickness of theseprocesses became more variable (Fig. 16). After 24 h of incubation with colcemid, theaverage thickness of these processes had decreased considerably (Fig. 17). Transferof the elongated cells from CD-medium into control medium was followed by gradualflattening and widening of the cellular processes until, 24 h after the transfer, thecells acquired ellipsoid shapes (Fig. 18); at 48 h the usual discoid shapes wererestored.

(4) Suspended cells were seeded on the usual substratum in medium containingboth CD (1-2/ig/ml) and colcemid (0-1-0-2 ̂ g/ml). These cells remained non-spread for the next 24 h and formed only small rudimentary pseudopodia at variousparts of the cell perimeter.

DISCUSSION

Isotropic and non-isotropic spreading of epithelial cells

IAR2 cells, like other epithelial cells, spread isotropically to form discoid shapes onthe usual substrata. Unlike fibroblasts, these cells do not undergo spontaneouspolarization. However, we have found that IAR2 cells can spread anisotropically toform polarized shapes in three sets of special conditions: on narrow strips of adhesivesubstrate, on poly(HEMA) substrate and on the usual substrate in cytochalasin-containing medium.

The mechanism of polarization is most obvious in the first case: cells spreading onnarrow strips are able to attach their pseudopodia only along the strip and thusundergo forced polarization.

When cells spread on a poorly adhesive substrate, such as poly(HEMA), thensimultaneous attachment of pseudopodia extending in all directions becomes improb-able; but the cells can still acquire elongated shapes, which requires successful attach-ment of pseudopodia only in two opposite directions. Studies of the effects ofcytochalasins on the spreading of fibroblasts have shown that these drugs decrease thenumber of sites on the cell edge from which pseudopodia are extended as well as thewidth of the attached pseudopodia (Bliokh et al. 1980; Domnina et al. 1982).

High efficiency of pseudopodial attachment and rapid widening of the lamellae areessential for isotropic spreading. As suggested earlier (Vasiliev, 1982), other con-ditions being equal, all the factors decreasing the efficiency of extension and attach-ment of pseudopodia should favour the transition from isotropic to anisotropic spread-ing. Results of experiments with poly(HEMA) and with cytochalasin are in goodagreement with this suggestion.

It is interesting that neoplastically transformed IAR cells, in contrast to non-transformed parents, often acquire elongated polarized shapes even on the usual

280 L. V. Domnina, J. A. Rovensky, jf. M. Vasiliev and I. M. Gelfand

culture substratum. This elongation is probably a manifestation of the decreasedefficiency of their spreading (Bannikov et al. 1982).

Effects of colcemid on isotropic and anisotmpic spreading oflAKZ cells

All the effects of colcemid on IAR2 cells observed in our experiments were inducedby relatively small concentrations of the drug, which resulted in complete dis-appearance of cytoplasmic microtubules but were not visibly toxic. These effects werecompletely reversed by the removal of the drug from the medium. Another inhibitorof the polymerization of microtubules, vinblastin, had essentially the same effects ascolcemid. These facts give us reason to think that the effects of colcemid on spreadingare specific, that is, associated with depolymerization of microtubules.

We have found that colcemid completely prevents polarization of IAR2 cells onspecial substrates, such as poly(HEMA) and adhesive strips. This drug also dis-organizes the early stages of isotropic spreading on the usual substrate. Once reached,the well-spread discoid shapes of IAR2 cells become relatively resistant to colcemid.

These effects are essentially similar to the effects of microtubule-destroying drugson fibroblasts described by Ivanova, Margolis, Vasiliev & Gelfand (1976).

The effects of colcemid on fibroblast cultures are more obvious, because in this casecells spread on the usual substrates acquire polarized shapes that are highly sensitiveto colcemid. Therefore, the differences between the control and colcemid-treatedcultures are easy to see at any time after the end of spreading. In contrast, morphologi-cal differences between drug-treated and control epithelial cultures become less strik-ing at the final stages of spreading.

To explain the observed effects of colcemid and similar drugs we suggest thatmicrotubules counteract the contractility of the actin cortex. Owing to this contractil-ity pseudopods and lamellae that had extended and attached during spreading tendto retract and to detach themselves from the substrate (see review by Harris, 1982).Possibly, microtubules, when present, diminish these effects of contractility andtherefore stabilize the shape of attached processes. This suggestion is supported bythe experiments of Magendantz & Solomon (1981).

In these experiments drug-induced depolymerization of microtubules in culturedneuroblastoma cells lead to contraction and disappearance of long axon-like processes;this contraction was prevented by simultaneous treatment of these cells withcytochalasin. The results of our experiments with IAR2 cells were essentially similar:the only variant of polarization that was not reversed rapidly by colcemid was thatobtained in the cytochalasin-containing medium, that is, under the conditions inwhich the contractility of actin cortex was impaired.

The different sensitivity of various stages and types of spreading to colcemid canalso be explained by the opposite effects of microfilaments and microtubules on thestability of attached pseudopodia. Obviously, pseudopodia can retract more easily atthe early stages of spreading when cell-substrate attachments are not yet numerous;therefore, prevention of excessive contraction of the microtubules is essential for rapidprogress of spreading through these stages. Microtubules are also essential underconditions in which a group of contacts of the attached lamellae is separated from

Microtubule-destroying drug, effect on epithelium 281

other attachments by an elongated stretched process, that is, in polarized cells. Incontrast, microtubules are less important for the prevention of retraction of lamellaeof non-polarized discoid cells, which have numerous attachments with the substratealong the whole cell perimeter. It is significant that depolymerization of microtubulescauses small retraction of lamellae even in these discoid cells.

We do not know how microtubules prevent retraction of extended pseudopodia.They can act mechanically as stiff rods counteracting the contractility of the actincortex. More complex interactions of microtubules with the actin cytoskeleton are alsopossible. For instance, microtubules can direct the transport of non-polymerizedactin towards the sites of polymerization of new microfilaments, especially to theactive edges. It has been suggested (Dunn, 1980) that an 'actin flow' is essential forpseudopodial extension. It is also well known that destruction of microtubules disor-ganizes many types of intracellular movements (see review by Schliwa, 1984). Allthese suggestions require further experimental tests.

Destruction of microtubules in IAR2 cells and in many other cell types (see reviewby Lazarides, 1980) is accompanied by the collapse of vimentin-containing inter-mediate filaments. Are the effects of colcemid on spreading of these cells mediated byalterations in the intermediate filaments? Selective collapse of intermediate filamentsinduced by the injection of specific antibody did not lead to any obvious alteration incell shape (Klymkowsky, Miller & Lane, 1983). Recently, we (Karavanova,Troyanovsky & Vasiliev, unpublished) found that colcemid reverses polarization ofelongated mouse hepatoma cells containing cytokeratin intermediate filaments;incubation of these cells with colcemid destroyed their microtubules but did not leadto the collapse of intermediate filaments. Therefore, it is unlikely that intermediatefilaments play an essential role in the effects of destruction of the microtubular systemon the shape of epithelial cells.

REFERENCES

BANNIKOV, G. A., GUELSTEIN, V. I., MONTESANO, R., T INT, I. S., TOMATIS, L., TROYANOV-SKY, S. M. & VASILIEV, J. M. (1982), Cell shape and organization of cytoskeleton and surfacefibronectin in non-tumorigenic and tumorigenic rat liver cultures. J . Cell Sci. 54, 47-67.

BANNIKOV, G. A., SAINT VINCENT, L. & MONTESANO, R. (1980). Surface proteins in normal andtransformed rat liver epithelial cells in culture. Br.jf. Cancer 42, 596-609.

BERSHADSKY, A. D., GELFAND, V. I., SVITKINA, T. M. & T I N T , I. S. (1980). Destruction ofmicrofilament bundles in mouse embryo fibroblasts treated with inhibitors of energy metabolism.Expl Cell Res. 127, 421-429.

BERSHADSKY, A. D., T I N T , I. S., GELFAND, V. I., ROSENBLAT, V. A., VASILIEV, J. M. &GELFAND, I. M. (1978). Microtubular system in cultured mouse epithelial cells. Cell Biol. Int.Rep. 2, 345-351.

BLIOKH, Z. L., DOMNINA, L. V., IVANOVA, O. Y., PLETJUSHKINA, O. Y., SVITKINA, T. M.,SMOLJANINOV, V. V., VASILIEV, J. M. & GELFAND, I. M. (1980). Spreading of fibroblasts inmedium containing cytochalasin B. Proc. natn. Acad. Sci. U.SA. 77, 5919-5922.

Di PASQUALE, A. (1975). Locomotory activity of epithelial cells in culture. Expl Cell Res. 94,191-215.

DOMNINA, L. V., GELFAND, V. I., IVANOVA, O. Y., LEONOVA, E. V., PLETJUSHKINA, O. Y.,VASILIEV, J. M. & GELFAND, I. M. (1982). Effect of small doses of cytochalasin on fibroblasts:preferential changes of active edges and focal contacts. Proc. Natn. Acad. Sci. U.SA. 79,7754-7757.

282 L. V. Domnina, jf. A. Rovensky, jf. M. Vasiliev and I. M. Gelfand

DOMNINA, L. V., IVANOVA, 0 . Y., MARGOLIS, L. B., OLSHEVSKAYA, L. V., VASILIEV, J. M.,

GELFAND, I. M. (1972). Defective formation of the lamellar cytoplasm by neoplastic fibroblasts.Pmc. natn. Acad. Sci. U.SA. 69, 248-252.

DOMNINA, L. V., PLETJUSHKINA, 0 . Y., VASILIEV, J. M. & GELFAND, I. M. (1977). Effect of

antitubulins on the redistribution of cross-linked receptors on the surface of fibroblasts andepithelial cells. Pmc. natn. Acad. Set. U.SA. 74, 2865-2868.

DUNN, G. A. (1980). Mechanisms of fibroblast locomotion. In Cell Adhesion and Motility, BSCBSyntp. 3 (ed. A. S. G. Curtis & J. D. Pitts), pp. 409-23. Cambridge University Press.

FOLKMAN, J. & MOSCONA, A. (1978). Role of cell shape in growth control. Nature, Lend. 273,345-349.

HARRIS, A. K. (1982). Traction and its relation to contraction in tissue cell locomotion. In CellBehaviour, A Tribute to Michael Abercrombie (ed. R. Bellaris, A. Curtis & G. Dunn), pp.109-134. Cambridge University Press.

IVANOVA, O. Y. & MARGOLIS, L. B. (1973). The use of phospholipid membranes for preparationof cell cultures of given shape. Nature, Land. 242, 200—201.

IVANOVA, O. Y., MARGOLIS, L. B., VASILIEV, J. M. & GELFAND, I. M. (1976). Effect of colcemidon the spreading of fibroblasts in culture. Expl Cell Res. 101, 207-219.

IZZARD, C. S. & LOCHNER, L. R. (1976). Cell-to-substrate contacts in living fibroblasts; aninterference reflection study with an evaluation of the technique. J . Cell Sci. 21, 129-159.

KLYMKOWSKY, M. W., MILLER, R. H. & LANE, E. B. (1983). Morphology, behaviour and interac-tion of cultured epithelial cells after the antibody-induced disruption of keratin filament organiza-tion. J. CellBiol. 96, 494-509.

LAZARIDES, E. (1980). Intermediate filaments as mechanical integators of cellular space. Nature,Land. 283, 249-256.

MAGENDANTZ, M. & SOLOMON, F. (1981). Cytochalasin separates microtubules disassembly fromloss of assymetric morphology..7. CellBiol. 89, 157-161.

MIDDLETON, C. A. (1976). Contact-induced spreading is a new phenomenon depending oncell-cell contact. Nature, Land. 259, 311-313.

MONTESANO, R., DREVON, C , KUROKI, T., SAINT VINCENT, L., HANDLEMAN, S., SANFORD,

K. K., D E FEO, D. & WEINSTEIN, I. B. (1977). Test for malignant transformation of rat livercells in culture: cytology, growth in soft agar and production of plasminogen activator, jf. natn.Cancer Inst. 59, 1651-1658.

MONTESANO, R., SAINT VINCENT, L., DREVON, C. & TOMATIS, L. (1975). Production of

epithelial and mesenchymal tumors with rat liver cells transformed in vitro. Int. J. Cancer 16,550-558.

SCHLIWA, M. (1984). Mechanisms of intracellular organelle transport. In Cell and Muscle Motility(ed. by J. W. Shay), vol. 5, pp. 1-82, 403-406. New York: Plenum.

VASILIEV, J. M. (1982). Spreading and locomotion of tissue cells: factors controlling thedistribution of pseudopodia. Phil. Trans. R. Soc. Land. B, 299, 159-167.

VASILIEV, J. M. & GELFAND, I. M. (1976). Effect of colcemid on morphogenetic processes andlocomotion of fibroblasts. In Cell Motility (ed. by R. Goldman, T. Pollard & J. Rosenbaum) ColdSpring Harbor Conf. Cell Proliferation, vol. 3, pp. 279-304. New York: Cold Spring HarborLaboratory.

VASILIEV, J. M., GELFAND, I. M., DOMNINA, L. V., IVANOVA, O. Y., KOMM, S. G. & OLSHEV-

SKAYA, L. V. (1970). Effect of colcemid on the locomotory behaviour of fibroblasts. J . Embryol.exp. Morph. 24, 625-640.

(Received 3 July 1984-Accepted 16 October 1984)