Morphogenetic response of cultured normal and transformed fibroblasts, and epitheliocytes, to a cylindrical substratum surface Possible role for the actin filament bundle pattern

Mammalian tissue cells have the ability to respond to the geometry of a chemically isotropic substratum surface in culture. Depending on the geometrical configuration and its quantitative parameters the cell’s response can vary (Curtis and Clark, 1990). On substrata with parallel fine ridges or grooves cells can show changes in shape, become aligned (Ohara and Buck, 1979; McCartney and Buck, 1981; Buck, 1982; Dunn, 1982; Brunette, 1986a,b; Dunn and Brown, 1986; Clark et al., 1990, 1991), and migrate from sufficiently deep grooves (Rovensky et al., 1971; Rovensky and Slavnaya, 1974; Slavnaya and Rovensky, 1975; Vesely et al., 1981). On cylindrical substrata with a high degree of surface curvature fibroblasts have been shown to elongate and to align (Curtis and Varde, 1964; Rovensky et al., 1971; Dunn and Heath, 1976; Fisher and Tickle, 1981). According to the hypothesis of Dunn and Heath (Dunn and Heath, 1976; Dunn, 1991) a single mechanism may be the basis of the cell’s response to the geometry of the substratum surface. As these authors proposed, actin microfilament bundles and their associated focal contacts cannot be formed in circumstances in which the actin bundles would be bent. Hence the efficiency with which the cell can exert traction and can move on a non-planar substratum surface is considerably limited in definite directions. This would lead to cell shape changes, to cell alignment and to changes in the direction of cell locomotion. In the light of these speculations it is interesting to study the response of cells with various patterns of actin microfilament bundles to the geometrical configuration of the substratum surface. For example, it is possible to investigate the response of normal fibroblasts at early and late stages of spreading: at the early (radial) stage a fibroblast contains a circular bundle of actin microfilaments; at a later stage (polarization) this bundle disappears and new straight bundles oriented perpendicularly to the active cell edges are formed (Vasiliev, 1985). Circumferential actin bundles at the cell periphery similar to those of fibroblasts at the radial stage of spreading are also characteristic of single spread epitheliocytes of some primary cultures or of ‘normal’ lines (Connoly et al., 1981; Bannikov et al., 1982). On the other hand, in some sorts of transformed epitheliocytes the circular actin bundles disappear and straight bundles can be formed (Brinkley et al., 1980; Bannikov et al., 1982). In this paper we have used quantitative analyses of cell shape and cell alignment to study the response of normal and transformed fibroblasts, and epitheliocytes with various patterns of actin filament bundles, to a cylindrical substratum surface.

Primary cultures of mouse embryo fibroblasts (MEF) and also some cell lines were used in the experiments. Epithelial cell lines of the IAR series were originally derived from rat liver, the previous history of those cell lines has been described in detail (Montesano et al., 1975, 1977; Kuroki et al., 1977). We used two cell lines of the IAR series: IAR 20 and IAR 6-1. Cells of the IAR 20 line in sparse cultures display discoid epithelial shapes, with a wide ring of lamellar cytoplasm on the cell periphery; in dense cultures coherent sheets of the cells are formed (Guelstein, 1981; Bannikov et al., 1982). The line is non-tumorigenic. Cells of the IAR 6-1 line are polygonal, triangular, spindle-like or semispherical, with reduced lamellar cytoplasm; in dense cultures the cells display multi-layer growth (Guelstein, 1981; Bannikov et al., 1982). The line is tumorigenic. The epithelial cell line FBT was originally derived from foetal bovine trachea (Machatkova and Pospisil, 1975). The cell morphology and the growth pattern of FBT line are similar to those of the IAR 20 line described above (Svitkina and Kaverina, 1989). Fibroblastic cell line L was originally derived from mouse subcutaneous connective tissue explants treated in vitro with a carcinogen 20-methyl cholanthren (Earle, 1943). L cells have polygonal, triangular or spindle-like shapes, with reduced lamellar cytoplasm. The cells are irregularly tumorigenic. The culture medium consisted of L-15 medium (Flow, UK) (for IAR 20 or IAR 6-1 cells) or 199 medium (for MEF and also for L or FBT cells) supplemented with 10% foetal calf serum (GIBCO Biocult, Scotland) and 100 units/ml monomycin. The cells were resuspended in fresh culture medium and plated at an initial density of 10×103 to 20×103 cells/cm2 on flat or cylindrical substrata (see below) placed in a tissue culture multiwell plate with an area of approximately 2 cm²per well (Linbro, Flow, Scotland). In experiments to test the effect of cytochalasin B (Sigma) on IAR 6-1 cells the agent was added to the culture medium (final concentration of the drug 2 μg/ml) on IAR 6-1 cell seeding. Control cells were cultivated in the same medium with 0.02% dimethyl sulfoxide (DMSO) used as solvent for cytochalasin B. In some experiments, after 24 hours cultivation, cytochalasin-containing medium was replaced by fresh medium (cell washing). The cells were cultured for 2-48 hours at 37°C in a humidified incubator supplied with 5% CO₂in air.

As substrata with cylindrical surfaces we used fibre-optic light conduit-fused quartz cylindrical fibres with diameter 50 μm (used as the substrata for MEF) or 24-26 μm (substrata for cell lines used). These fibres were prepared in the Institute of Glassfibers and Glass-plastics (Moscow, Russia). Coverslips were covered with a layer of melted agarose; after the agarose acquired a semi-solid state 2-4 mm fragments of the fibres were placed on its sticky surface. After 18-24 hours the agarose had hardened and the fibres were firmly attached. Since an agarose surface is not adhesive for cells, the only substratum surfaces the cells were attached to were the surfaces of quartz cylindrical fibres. As control substrata with flat surfaces we use glass coverslips. The surfaces of glass coverslips and of fused quartz fibres used here are not significantly different in their chemical composition. In our previous experiments we could not show any differences in the morphological or temporal characteristics of the spreading of mouse fibroblasts on the flat surfaces of fused quartz plates or glass cover-slips.

The cell cultures were washed in warm Hanks’ balanced salt solution and fixed with 2% glutaraldehyde in 0.1 M sodium cacodylate buffer with 0.1 M sucrose. The cells were dehydrated in a graded series of acetone and critical point dried from liquid CO₂in the CPD apparatus (Balzers Union, Liechtenstein). The specimens were coated with gold/palladium in a cold sputter (Polaron, UK) and examined in a PSEM-501 microscope (Philips, Netherlands). The stage tilt angle used for the photoregistration of the cells on flat or cylindrical substrata was 0°.

To visualize actin cytoskeleton in the cells on flat substrata the cultures were stained with TRITC-labeled phalloidin (at a concentration 0.1 μg/ml). For fluorescence microscopy, the cell cultures were washed in phosphate-buffered saline (PBS) and fixed with 4% formaldehyde in PBS. After fixation the cells were extracted for 10 minutes with 1% Triton X-100 in PBS and incubated in TRITC-labeled phalloidin (Sigma, USA) for 30 minutes at room temperature. Observations were made with a microscope (Aristoplan, Leitz, Germany) equipped with epifluorescent illumination and a Fluotar 50-X water-immersion objective.

Scanning electron micrographs in which the contours of isolated cells on the visible halves of cylindrical substrata were continuous were selected for quantitative analysis. Selection of cells on flat (control) substrata was random. The cell outlines and the edges of the cylindrical substrata were traced by projecting the negatives onto paper with a photographic enlarger. The magnifications of the scanninng electron microscope ranged from ×350 to ×1400, and the magnifications provided by the enlarger, while drawing, ranged from ×3 to ×6. To quantify the effects of a curvature of the substratum surface on the morphology of the cells we used: (a) a measure of cell size: cell area; (b) two measures of cell shape: dispersion and elongation; (c) a measure of cell alignment: cell orientation (α°). Cell area and dispersion were evaluated by computer morphometric methods, which have previously been fully described (Dunn and Brown, 1986; Brown et al., 1989). Cell elongation and orientation were calculated by the method described below (Fig. 1).

The projections of the edges of the cylindrical substratum (CE) and the axis of the cylinder (a line running at equal distances from both CEs) and also the projection of the cell outline were drawn (Fig. 1A). Polygonal approximation of the cell outline as the most protruding points (a, b, c,…) (from 5 to 12 for each cell) and their projections (a′, b′, c′,…) on the axis of the cylinder were marked. Then the surface of the visible half of the cylinder was ‘unrolled’ and transformed into a flat surface; the distance between the edges of the unrolled cylinder was equal to half of the distance between two CE(s), multiplied by p. On the axis of the unrolled cylinder A′, B′, C′,… points were transferred (Fig. 1B). On the basis of the a′a, b′b, c′c… distances the lengths of A′A, B′B, C′C,… arcs were calculated and corresponding lines were drawn on the unrolled cylinder. Successive joining of A, B, C,… points on the unrolled cylindrical surface results in a new unrolled cell contour. Both minimal (W) and maximal (L) distances between two parallel lines tangent to the unrolled cellular contour but not transecting it are measured. Ratio L/Wis a measure of elongation of the cell; L/W≥1. An angle (α°) formed by the axis of the unrolled cylinder and the parallel lines at minimal (W) distance between them is a measure of orientation of the cell according to the axis of the cylinder; 0°≤α°≤90°. The following numbers of isolated cells were measured: 65-113 MEF (2 hours); 56-65 MEF (8 hours), 25 MEF (24 hours), 29-64 L, 25-38 FBT, 38-41 IAR 20, 139-220 IAR 6-1 and 55 IAR 6-1 treated with cytohalasin B. The significance of the differences was determined using Student’s t-test. For all statistical comparisons the significance level was taken to be P<0.05.

The cell types studied had various patterns of actin microfilament bundles on flat substrata. MEF cultured for 2 hours and single IAR 20 cells contained circular actin bundles located along the periphery of the disc-like cells (Fig. 2A, D). MEF cultured for 8 or 24 hours, IAR 6-1 and single FBT cells had predominantly straight actin bundles, which were longer in IAR 6-1 cells and comparatively short in FBT (Fig. 2B, E, F). L cells were practically devoid of actin bundles (Fig. 2C).

Areas of MEF cultured for 2 or 8 hours on cylindrical substrata were considerably decreased (correspondingly 3.6-and 5.7-fold) compared with those on flat substrata; after 24 hours of cultivation there was no significant difference between the area of MEF on cylindrical and those on flat substrata. The area of L-cells cultured on cylindrical substrata was decreased (1.6-fold). No significant difference between areas of FBT, IAR 20 or IAR 6-1 cells cultured on cylindrical and those on flat substrata were found. Treatment of IAR 6-1 cells with cytocha-lasin B decreased the cell areas on flat (1.7 times less) as well as on cylindrical (2.7 times less) substrata. Substitution of cytochalasin-containing medium for fresh culture medium (cell washing) restored IAR 6-1 cell area on flat substrata to the value for untreated cells; IAR 6-1 cell area on cylindrical substrata 24 hours after cell washing remained decreased (2.2 times less) compared with the area of untreated cells (not shown).

Dispersion of MEF cultured for 2 hours on cylindrical substrata was somewhat increased but after 8 or 24 hours of cultivation there was no significant difference between the dispersion of the cells on cylindrical and those on flat substrata. On cylindrical substrata dispersions of FBT and IAR 20 cells were considerably increased (5-fold and 3.4-fold, respectively) but dispersions of L and IAR 6-1 cells were decreased (2.3- and 1.3-fold, respectively) compared with the cell dispersion values on flat substrata. Cytochalasin B induced a considerable decrease in dispersion of IAR 6-1 cells on flat (2.6 times less) but not on cylindrical substrata. Cell washing restored dispersion of IAR 6-1 cells cultured with cytochalasin B on both flat and cylindrical substrata to the dispersion values of the untreated cells (not shown).

There is no significant difference between the Elg of MEF cultured for 2 hours on cylindrical and that on flat substrata (Figs 3C, 4A, A′). After 8 or 24 hours of cultivation the Elg(s) of MEF on cylindrical substrata were increased (1.3 and 1.6-fold, respectively) compared with those on flat substrata (Figs 3C, 4B, B′). Elg(s) of epitheliocytes cultured on cylindrical substrata were increased; this increase was the most considerable (2.1-fold) in FBT and IAR 6-1 cells (Figs 3C, 4E, E′, F, F′) but comparatively weak (1.3-fold) in IAR 20 cells (Figs 3C, 4D, D′). There is no significant difference between the Elg of L cells on cylindrical and flat substrata (Figs 3C, 4C, C′). Cytochalasin B induced a considerable decrease in the Elg of IAR 6-1 cells on cylindrical (2.5 times less) but not on flat substrata (Fig. 3C). Cell washing restored the Elg of IAR 6-1 cells cultured with cytochalasin B on cylindrical substrata to the value of untreated cells: the Elg of IAR 6-1 cells on flat substrata after cell washing was maintained slightly decreased compared with the Elg of untreated cells (not shown).

The orientation (along the cylinder axis) of MEF cultured on cylindrical substrata was the weakest after 2 hours of cultivation, it was considerably increased (α° value is reduced twofold) after 8 hours and rather diminished after 24 hours (Figs 3D, 4A′, B′). In all cell lines studied the most considerable orientation was in FBT and IAR 6-1 cells (Figs 3D, 4E′, F′) and the weakest was in IAR 20 cells (Figs 3D, 4D′). Treatment of IAR 6-1 cells with cytochalasin B caused profound deterioration in cell orientation (α° becomes 3.9-fold) (Fig. 3D). Cell washing restored the orientation of IAR 6-1 cultured with cytochalasin B, though incompletely: 24 hours after substitution of the drug-containing medium the difference between α° values of treated and untreated cells was significant (not shown).

Results of morphometric analysis of cells cultured on flat or cylindrical substrata showed that cylindrical surfaces with a high degree of curvature (12-13 or 25 μm radii) affected cell size, shape and alignment. This influence of surface geometry was different in the various cell types studied. A cylindrical substratum surface with a sufficiently high degree of curvature is apparently less optimal for cell adhesion compared to a flat surface and can favour retraction of the cells. Normal fibroblasts (MEF) spread on cylindrical substrata more slowly than on flat ones: areas of MEF on cylindrical surfaces were considerably decreased for a period up to 8 hours after cell seeding, and after only 24 hours there was no significant difference between MEF areas on flat and cylindrical substrata. A cylindrical surface is likely to hamper successful extention of lamellar cytoplasm accompanied by the formation of firm cell-substratum contacts during fibroblast spreading. It is very likely that a cylindrical substratum surface promotes retraction of the cells reducing their ability for making effective contact with the usual flat substratum. Cells of transformed fibroblastic lines (and, in particular, transformed fibroblasts of L line) belong to the above-mentioned category of cells (Vasiliev and Gelfand, 1977, 1981; Vasiliev, 1985); in contrast to other cell types studied in our experiments, L cells are practically devoid of actin microfilament bundles and do not form very many focal contacts. As we have shown, L cell areas were significantly decreased on cylindrical surfaces. The supposition that a cylindrical substratum surface can favour retraction of cells, reducing their ability to make firm cell-substratum contacts, is confirmed by our data on the response of IAR 6-1 cells to cytochalasin B. We did not investigate the effect of a low dose (2 μg/ml) of cytochalasin B on the actin cytoskeleton of IAR 6-1 cells. Previous experiments with mouse fibroblasts have shown that the partial or complete disappearance of microfilament bundles and the disorganization of the cortical layer of microfilaments were the manifestations of alterations in the actin cytoskeleton caused by low doses of cytochalasin B or cytochalasin D (Domnina et al., 1982). Cytochalasin (destroys microfilament bundles and their associated focal contacts) apparently caused partial detachment and subsequent retraction of IAR 6-1 cells (decrease in the cell area); the retraction was more strongly pronounced in the cells on cylindrical substrata, and there it was maintained for 24 hours after cell washing. On flat surfaces IAR 6-1 cells treated with cytochalasin regained their initial cell areas after 24 hours of cultivation in fresh medium. Dispersions of FBT and IAR 20 cells and also MEF at early stages of spreading were increased whereas those of IAR 6-1 and especially L cells were decreased on cylindrical substrata in comparison with dispersions of the same cell types on flat substrata. The increase in dispersion corresponded to the appearance of jagged cell margins and to multiple very short thin outgrowths from the cells on cylindrical substrata; these ‘jags’ and outgrowths were seen more rarely in cells on flat substrata. Further work is needed to find out the cause of the increasing effect of a cylindrical surface on dispersion of cells displaying discoid but not fibroblast-like shape. Cultured fibroblasts were reported long ago to respond to a cylindrical substratum surface by elongation and orientation (Curtis and Varde, 1964). It was shown by morphometric methods that not only fibroblast bodies but their nuclei as well elongate and orient themselves on cylindrical substrata (Margolis et al., 1975; Samoilov at al., 1975, 1978; Dunn and Heath, 1976; Vesely et al., 1981). Dunn and Heath (Dunn and Heath, 1976; Dunn, 1982, 1991) postulated an important role for the actin cytoskeleton in the reaction of elongation and orientation of cells cultured on cylindrical substrata. These authors proposed that on cylindrical surfaces with curvature radii of 100 μm or less bundles of actin microfilaments could not become assembled in a bent state, and hence the efficiency with which the cell could exert traction on the cylindrical substratum was reduced in the case of highly convex curvature. As a result the cell shape would be markedly polarized, elongating along the cylindrical substratum. The mechanism described above is helpful in explaining our data on increasing elongation and orientation in mouse embryo fibroblasts cultured on cylindrical substrata for 8-24 hours. Normal fibroblasts at these times of culturing exhibit a well developed system of straight actin microfilament bundles (Vasiliev and Gelfand, 1981), which apparently prevents cell bending and ‘forces’ the cell to elongate along the cylindrical substratum. The same mechanism seems to account for the absence of significant changes in elongation of transformed fibroblasts of the L line on cylindrical substrata: this cell line is practically devoid of actin microfilament bundles. Of interest are the data from the analysis of elongation and orientation of epitheliocytes on cylindrical substrata. They have not been studied previously. An increase in elongation proved most significant in FBT and IAR 6-1 cells. Cell orientation on cylindrical substrata was the most marked in FBT and IAR 6-1 cells. The IAR 20 cell reaction to cylindrical substrata proved significantly weaker.

While examining these data it appears reasonable to compare the peculiarities of actin cytoskeleton organization in the epitheliocytes studied. They all contained actin microfilament bundles; however, the pattern of these bundles appears different in various cell lines. In isolated IAR 20 cells actin bundles have a circular shape, as a rule: the bundle is located along the periphery of the disk-like cell (Bannikov et al., 1982; Svitkina and Kaverina, 1989). In cells of the ‘normal’ epithelial line FBT and the fully transformed epithelial line IAR 6-1, circular actin bundles occur much less frequently: the cells of these lines contain predominantly straight bundles in great numbers, of considerable length (IAR 6-1), or comparatively short (FBT) (Bannikov et al., 1982; Svitkina and Kaverina, 1989). Thus, the reaction to a cylindrical substratum proved most marked in epitheliocytes containing mainly straight bundles of actin microfilaments, it was true of both ‘normal’ FBT cells and fully transformed IAR 6-1 cells; the elongation and orientation on a cylindrical substratum was weaker in epitheliocytes with mainly circular actin bundles (IAR 20). The circular location of microfilament bundles seems to inhibit the capacity for elongation on cylindrical substrata not only in epithelial cells. According to our analysis the elongation of mouse embryo fibroblasts cultured on cylindrical substrata for 2 hours does not reveal any significant change as compared to that on flat substrata; the MEF orientation after 2 hours of cultivation was the weakest. The indicated times of cultivation correspond to the radial stage of normal fibroblast spreading when the bundles of actin microfilaments have a circular location along the periphery of the concentrically developing lamellar cytoplasm (Vasiliev and Gelfand, 1981). The high capacity for mouse embryo fibroblast bending in the radial stage of spreading is particularly demonstrable on cylindrical substrata of a very small radius (6-13 μm): the cell in the form of a thin circular plate completely ‘wraps up’ the cylinder (Rovensky, unpublished data). Consequently, at different stages of spreading the normal fibroblast reaction to a cylindrical substratum varies significantly: it ranges from a very weak reaction of elongation and orientation at the radial stage (circular actin bundle) to a marked reaction at the subsequent polarization stage (radial bundles perpendicular to the active edge of the cell). The data obtained indicate that the reaction of cultured cells to the geometry of the substratum surface and, in particular, to the cylindrical shape of the surface is determined not only by the presence or absence of actin microfilament bundles but by their pattern in the cell. The cell will be relatively resistant to bending, responding to the shape of the substratum surface by elongation and orientation when it contains sufficiently straight actin bundles (embryo fibroblasts at the polarization stage of spreading, some types of transformed fibroblastic or epithelial cultures). In the case of a predominantly circular pattern of actin bundles (embryo fibroblasts at the radial stage of spreading, single spread epitheliocytes of some primary cultures or of ‘normal’ lines) as well as in case of insufficient or absent actin bundles (cells of a number of fully transformed fibroblastic and epithelial lines) the cell becomes less resistant to bending, and the guidance of the cells becomes weaker. The mechanisms providing the increased ability of cells with a circular pattern of actin microfilament bundles to bend need more study. In the circular pattern there is no direct connection between each bundle and the contact, in the straight pattern the bundles are attached directly to focal contacts at their ends (Vasiliev, 1987). It is possible that this factor could cause the increased ability of cells with a circular pattern of actin bundles to bend.

We are deeply grateful to Prof. J. M. Vasiliev for his advices and criticism. We thank Drs A. F. Brown (MRC Human Genetics Unit, Western General Hospital, Edinburgh, Scotland) and G. A. Dunn (Randall Institute, MRC, Muscle and Cell Motility Unit, London, UK) for useful discussions, and Mrs A. Kernaghan (Randall Institute, London, UK) for digitizing the cell outlines. We are grateful to Drs T. M. Svitkina for her valuable help with fluorescence microscopy, N. I. Demidova for her kind help with the preparation of the manuscript and D. V. Samoilov (Cancer Research Centre, Moscow, Russia) for technical assistance. The work of V.I.S. was supported in part by a grant from ‘Universities of Russia’.