Actin Cytoskeleton of Spread Fibroblasts Appears to Assemble at the Cell Edges

J. Cell Sd. 82, 235-248 (1986) 235 Printed in Great Britain © The Company of Biologists Limited 1986

ACTIN CYTOSKELETON OF SPREAD FIBROBLASTS APPEARS TO ASSEMBLE AT THE CELL EDGES

TATJANA M. SVITKINA, ALEXANDER A. NEYFAKH, JR Laboratory of Molecular Biology and Bioorganic Chemistry, Moscow State University, Moscow 119899, USSR AND ALEXANDER D. BERSHADSKY All-Union Cancer Research Center, Academy of Medical Sciences, Moscow 115478, USSR

SUMMARY The action of metabolic inhibitors on actin cytoskeleton of cultured quail embryo fibroblasts has been studied using electron microscopy of platinum replicas and immunofluorescence microscopy. Sodium azide as well as other inhibitors (oligomycin and dinitrophenol) caused the disassembly of all types of actin structures: actin meshwork at the cell active edges, microfilament sheath underlying the cell surface, and microfilament bundles. Studying the time- and dose-dependence of the destruction process we have found that the active edge meshwork and microfilament sheath are much more labile than microfilament bundles. After the removal of metabolic inhibitors actin cytoskeleton restoration begins at the cell edges. The first sign of this process is the formation of actin meshwork along the whole cell perimeter (l-10min of recovery). Sometimes fragments of this meshwork bend upwards forming ruffles. Later (10-20 min of recovery) the microfilament sheath appears at the cell periphery as a narrow band. The sheath seems to be formed from the edge meshwork, since ruffles in the process of transformation to sheath could be seen. During the following restoration the microfilament sheath gradually expands towards the cell centre. The last step of actin cytoskeleton restoration (60—120 min of recovery) is the formation of bundles. We suggest that the actin in spread fibroblasts .polymerizes predominantly at the cell edges and then moves centripetally, forming the microfilament sheath and bundles.

INTRODUCTION The actin cytoskeleton of non-muscle cells is a highly dynamic and complex system. In cultured fibroblasts actin microfilaments form several morphologically and biochemically different structures of which microfilament bundles are the most obvious and well characterized (see review by Byers et al. 1984). In addition, actin forms a fine meshwork at the cell's leading edges (Small et al. 1978). The third type of actin structure is the microfilament sheath underlying the upper surface of spread fibroblasts. The sheath is clearly seen in detergent-extracted cells by electron microscopy of platinum replicas (Svitkina et al. 1984). The mechanism of formation of these three cytoskeletal structures and their relationships are unclear. Since actin structures are highly dynamic, the promising approach for studying these problems is the investigation of the processes of destruction and the following restoration of the actin cytoskeleton in living cell. Cytochalasins are well-known agents causing actin destruction (Weber et al. 1976;

Key words: actin cytoskeleton, fibroblasts, metabolic inhibitors, platinum replicas. 236 T. M. Svitkina, A. A. Neyfakh, jfr and A. D. Bershadsky

Schliwa, 1982). However, cytochalasin-induced actin depolymerization is accom- panied by a dramatic change in cell shape (Sanger & Holtzer, 1972). In this study we used metabolic inhibitors, particularly NaN3, for the dis- integration of actin cytoskeleton. We have shown that metabolic inhibitors destroy microfilament bundles, while the cell morphology changes very slightly (Bershadsky etal. 1980). Here we show that all actin-containing structures of cultured quail embryo fibroblasts are sensitive to metabolic inhibitors. The restoration of actin cytoskeleton after the removal of inhibitors begins at the cell periphery. In a few minutes the actin edge meshwork is formed. Then the microfilament sheath appears near the cell edge and gradually covers the central parts of the cytoplasm. We suppose that under normal conditions actin polymerization occurs also at the cell edges and then this polymerized actin moves centripetally, forming the microfilament sheath and possibly microfilament bundles.

MATERIALS AND METHODS Cells Quail embryo fibroblast culture was obtained by trypsinization of 11- to 13-day-old embryos. Cells were cultivated in 199 medium containing 0-25 % lactalbumin hydrolysate, 5 % tryptose phosphate broth and 5% bovine serum. Secondary cultures were seeded onto 5mmX5mm coverslips at low density and incubated for 48-72h at 37°C. After washing with phosphate- buffered saline (PBS), cells were treated with sodium azide (Sigma), oligomycin (Serva) or dinitrophenol (Sigma) in Dulbecco's salt solution at 37°C. Before fixation, control and drug-treated cell cultures were extracted with 1 % Triton X-100 in buffer M (50mM-imidazole, 50mM-KCl, 0-5mM-MgCl2, OlmM-EDTA, 1 mM-EGTA, 1 mM-2-mercaptoethanol, pH6-8) containing 4% polyethyleneglycol (Mr 40 000) for 5 min at room temperature.

Immunofluorescence Triton-extracted cells were rinsed with buffer M and fixed with 4% formaldehyde in PBS for 20 min or more. Actin antibody and the procedure for cell staining have been described (Bershadsky et al. 1980). For myosin staining monospecific rabbit antibody to calf spleen myosin was used. Its specificity was tested by Western blotting. The isolation and purification of myosin, as well as affinity chromatography of immune serum, were performed by Dr F. K. Gioeva.

Electron microscopy Details of the platinum-replica technique have been described (Svitkina etal. 1984). Briefly, Triton-extracted cells were rinsed with buffer M and fixed with 2% glutaraldehyde in 0-1M- sodium cacodylate buffer (pH 7-2-7-4) for 20 min or more. Then specimens were dehydrated in graded acetones, treated with 0-1 % uranyl acetate in acetone for 20 min, washed in two changes of pure acetone and critical point dried in a Balzers device. Dried cytoskeletons were rotary shadowed with Pt-C in a BAE 080 T apparatus (Balzers) at an angle of 45 °. The thickness of the platinum layer as determined by the method of Heuser (1983) was 2nm. The replicas were strengthened with carbon shadowing at an angle of 90°. Coverslips were removed with HF. Cell residues were dissolved with 30% aqueous C1O3. Replicas were examined in a Jeol 100C or Philips EM400 electron microscope at 80 kV. Negatives were photographically reversed before printing.

ATP measurement Subconfluent cultures of fibroblasts on 35 mm Petri dishes (3X105 cells/dish) were treated as described in Fig. 9, washed with Dulbecco's salt solution and extracted with ice-cold 3 % tri- chloroacetic acid for 3 min. Samples of the extract were diluted 500-fold with 0-1 M-Tris—acetate Assembly ofactin cytoskeleton 237 buffer (pH7-7S) containing 2mM-EDTA, and frozen. ATP measurement was performed by the luciferin-luciferase assay using ATP Monitoring Reagent (LKB) and PICO ATP Luminometer (Jobin Yvon, France).

RESULTS The actin cytoskeleton of control cells In order to study the actin cytoskeleton structure of quail embryo fibroblasts we have exploited two techniques: electron microscopy of platinum replicas of detergent-extracted cells, and immunofluorescence microscopy of the cells stained with actin and myosin antibodies. Previously we used the platinum replica technique for the study of the cytoskeleton of mouse embryo fibroblasts (Svitkina et al. 1984). The structure of the cytoskeleton of quail fibroblasts is principally the same. The central part of a quail cell is covered with a dense planar sheath of microfilaments (Fig. 1). Local parallel orientation of microfilaments is a characteristic feature of the sheath (Fig. 2). The peripheral lamellar parts of the cells are not covered by the sheath and distal parts of microfilament bundles terminating in adhesion plaques can be seen there (Fig. 1). These sheath-free lamellar regions are narrow in comparison with similar regions of mouse fibroblasts cytoskeleton. At the active edges of lamellae dense planar meshworks of microfilaments are visible (Fig. 1). Immunofluorescence with actin antibody revealed three types of actin-containing structures (Fig. 3). Numerous bright straight lines crossing the cell clearly cor- respond to microfilament bundles. Bright fluorescence at the active cell edges corresponds to an active edge actin meshwork. Diffuse fluorescence all over the cell most probably represents the cortical microfilament sheath. Myosin antibody intensively stained microfilament bundles and the microfilament sheath, while active edges were unstained (Fig. 4), confirming the previous data (Hegeness et al. 1977).

Destruction of actin cytoskeleton by metabolic inhibitors The actin cytoskeleton of quail fibroblasts was dramatically destroyed after treatment of cells for 1 h with metabolic inhibitors in buffered saline without an energy source (Dulbecco's solution). Metabolic inhibitors acting by different mechanisms, such as the cytochrome oxidase inhibitor sodium azide (10—20 mM), the H+- ATPase inhibitor oligomycin (1/iM) and the oxidative phosphorylation uncoupler 2,4-dinitrophenol (0-3 mM) had similar effects. Staining of the active edge and microfilament sheath with actin antibody completely disappeared. Microfilament bundles became fragmented and their remnants became randomly distributed in the cytoplasm (Fig. 5). Myosin antibody also stained only the remnants of the bundles in drug-treated cells (Fig. 6). Electron microscopy shows the absence of active edge actin meshwork and microfilament sheath. The fragments of the microfilament bundles are interspersed with the dense network of intermediate filaments and microtubules (Figs 7, 8). Fig. 1. The periphery of control quail fibroblast cytoskeleton. Edge actin meshwork (m), microfilament bundles (b), terminating by adhesion plaques (ap), and microfilament sheath (s) are seen. Electron microscopy. Bar, 1 /an. Fig. 2. The fine structure of microfilament sheath of control fibroblast. Electron microscopy. Bar, 1 /an. Assembly of actin cytoskeleton 239

Fig. 3. Anti-actin staining of control fibroblast. Immunofluorescence microscopy. Bar, 20 /an. Fig. 4. Anti-myosin staining of control fibroblast. Immunofluorescence microscopy. Bar, 20/an. The destruction of the actin cytoskeleton was reversible: after removal of inhibitors the actin cytoskeleton was re-formed (see below). In contrast, the structure of microtubule and intermediate filament cytoskeletal systems was not altered, as proved by immunofluorescense staining with tubulin and vimentin antibodies (not shown). The metabolic inhibitors seem to act on the actin cytoskeleton by decreasing the cytoplasmic ATP pool. The quantity of ATP in cells was diminished by a factor of SO after treatment for 1 h with 20mM-NaN3 (Fig. 9). The addition of a substrate for glycolysis (30 mM-glucose) to the inhibitor-containing solution prevented the effect of NaN3 on both the actin cytoskeleton structure (not shown) and the intracellular ATP pool (Fig. 9).

Different sensitivities of actin structures to metabolic inhibitors In order to study the sensitivity of different actin structures to ATP depletion we treated the fibroblasts with NaN3 under mild conditions (1—2-5 mM for 1 h or 20 mM 240 T. M. Svitkina, A. A. Neyfakh, jfr and A. D. Bershadsky Assembly of actin cytoskeleton 241

-14_ I

'••••}

\ '•••

10-1" 5

Depletion ^^ Recovery 10-16 10 20 30 40 50 10 20 30 40 Time (min)

Fig. 9. ATP levels in NaN3-treated and recovering fibroblasts. Cells were treated with 20mM-NaN3 (•), l-25mM-NaN3 (O), 20mM-NaN3 plus 30mM-glucose (•), and inhibitor-free Dulbecco's salt solution (•). The cells were replaced in growth medium at the moment indicated by the arrow.

for 10 min). As shown in Fig. 9, after such treatment three to five times more ATP was retained in the cells than after the procedure described above (20 mM for 1 h). Immunofluorescence and electron microscopy have shown that after the mild treatment microfilament bundles remained intact while both active edge actin meshwork and microfilament sheath disappeared (Figs 10-12). In platinum replicas, microfilament bundles become visible along their whole length due to sheath destruction. They are interspersed with the dense network of intermediate filaments and microtubules (Fig. 12). The characteristic feature of the cells treated under mild conditions is the appearance of granular material in the cytoskeleton (Fig. 12), which possibly represents the remnants of the microfilament sheath. Fig. 5. Anti-actin staining of the fibroblast treated with 20mM-NaN3 for 1 h. The remnants of microfilament bundles are seen. Immunofluorescence microscopy. Bar, 20 nm.

Fig. 6. Anti-myosin staining of the fibroblast treated with 20 mM-NaN3 for 1 h. Myosin is associated with the remnants of the bundles. Immunofluorescence microscopy. Bar, 20 yxn.

Fig. 7. Part of the fibroblast treated with 20mM-NaN3 for 1 h. Note the absence of edge actin meshwork and microfilament sheath. Fragments of the bundles are retained in the cell. Electron microscopy. Bar, S/im.

Fig. 8. A fragment of microfilament bundle in the fibroblast treated with 20mM-NaN3 for 1 h. The bundle is surrounded by a dense network of intermediate filaments. Electron microscopy. Bar, 0'5/im. 212 T. M. Svttkina, A. A. Neyfakh, jfr and A. D. Bershadsky Assembly ofactin cytoskeleton 243

Thus, active edge actin meshwork and microfilament sheath were found to be more sensitive to ATP depletion than microfilament bundles.

Restoration ofactin cytoskeleton after the removal of metabolic inhibitors In order to study the process of actin cytoskeleton restoration, cells were treated with 20mM-NaN3 for lh and then returned to the growth medium. As shown in Fig. 9 the replacement caused a very rapid increase in the ATP pool. Even after 1 min of incubation in growth medium the ultrastructure of the cytoskeleton at the cell edges had changed significantly. Small fragments of fine active edge microfilament meshwork appear at the cell boundaries (Fig. 13). The recovery during 3-5 min led to the appearance of a well-formed substrate-attached microfilament meshwork at the cell edges (Fig. 14). Later (10 min) the whole cell perimeter was contoured by such meshwork. At many places meshwork was turned upwards, forming ruffles. Often the substrate-attached microfilament meshwork was found distally to the ruffles. Immunofluorescence study confirmed the electron microscopic data. Anti-actin staining apparently corresponds to microfilament meshwork that appears at the cell boundaries (not shown). Electron microscopy of cells incubated in the growth medium for 15-20 min showed that the quantity of ruffles substantially increased. At this stage cells were already partially covered with a microfilament sheath. This sheath did not cover the whole endoplasm, as in control cells, but only its peripheral parts, forming a band along the cell margin behind the ruffles and the active edge microfilament meshwork. The central parts of cell were free of sheath (Figs 15—17). In full agreement with electron microscopy, immunofluorescence methods have shown the presence of diffuse anti-actin and punctate anti-myosin staining, located predominantly near the cell edges. After incubation for 30 min in growth medium this staining expanded over the whole cell. Electron microscopy confirmed that at this stage of recovery the microfilament sheath was completely restored (data not shown). Later, at 60-120 min, the quantity of active edge actin meshwork and ruffles decreased, returning to the control level, as was seen by both electron microscopy and immunofluorescence. The cytoplasmic staining with actin and myosin antibodies also became less intense. Microfilament bundles with normal morphology appeared.

Fig. 10. Anti-actin staining of the fibroblast treated with 2mM-NaN3 for 1 h. At low concentrations of NaN3 microfilament bundles remain intact while staining of the active edge and diffuse staining of the cytoplasm disappear (compare with Fig. 3). Immuno- fluorescence microscopy. Bar, 20 fan. Fig. 11. The edge of the fibroblast treated with 1 mM-NaN3 for 1 h. Note the absence of microfilament meshwork at the edge. The bundle with adhesion plaque is retained. Electron microscopy. Bar, 1 /an. Fig. 12. The central part of the cell treated with 2mM-NaN3 for 1 h. Microfilament sheath has completely disappeared, and actin bundles surrounded by intermediate filaments have become visible. Note the appearance of granular material (arrow). Electron microscopy. Bar, 1 ^m. Fig. 13. The edge of the cell recovering after NaN3 treatment (20mM, 1 h) for 1 min. Note the beginning of edge actin meshwork formation (arrows). Electron microscopy. Bar, 0-5/un.

Fig. 14. The edge of the cell recovering after NaN3 treatment (20 mM, 1 h) for 5 min. Massive edge actin meshwork appears. Electron microscopy. Bar, 1 /im. Assembly of actin cytoskeleton 245

Thus, the restoration of the actin cytoskeleton after the enlargement of the ATP pool begins at the cell edges. At the first stage the edge actin meshwork is formed excessively. Later the microfilament sheath begins to form at the cell periphery and expands towards the central parts of the cell. Microfilament bundles are the last actin-containing structures to be restored.

DISCUSSION In this paper we describe the processes of destruction and restoration of the actin cytoskeleton in quail embryo fibroblasts. We have shown that metabolic inhibitors cause the destruction of all actin-containing structures: i.e. active edge meshwork, microfilament sheath and actin bundles. The metabolic inhibitors seem to act on the actin structures by the reduction of the ATP pool. A good correlation between the extent of cytoskeleton destruction and ATP depletion was observed. The addition of glucose to the incubation medium as substrate for glycolysis prevented ATP depletion as well as disassembly of actin structures. The absence of ATP can impair the actin polymerization process in cells: the ATP dependence of actin polymerization is known from in vitro studies (see Lai etal. 1984; Pollard, 1984). By studying the time- and the dose-dependence of the destruction process we have found that the active edge microfilament meshwork and the microfilament sheath are much more labile than microfilament bundles. This can be explained by faster turnover of their protein components. This suggestion is supported by photobleaching recovery measurements (Kreis etal. 1982), which showed that actin in microfilament bundles is much less mobile than actin at the leading edge and in the interfibrillary space of fibroblasts. The dynamic properties of the active edge meshwork and the microfilament sheath are clearly demonstrated also by our finding that these structures are formed much faster than microfilament bundles after the removal of metabolic inhibitors. We have studied the process of actin cytoskeleton restoration in detail since it should resemble the natural process of the formation of actin-containing structures. We have found that the first sign of actin cytoskeleton restoration is the appearance of the microfilament meshwork at the cell edges shortly (1 min) after the immersion of sodium-azide-treated cells in inhibitor-free culture medium. This phenomenon is characterized by several interesting features: the process of meshwork formation is very fast and intensive; the quantity of meshwork material in recovering cells significantly exceeds the normal level; actin meshwork is formed not only at previously active regions of the cell edge but along the whole cell perimeter. An interesting question is why the process of meshwork assembly is strongly restricted to the cell edges. Several mechanisms, such as specific ionic conditions, increase in nucleation sites for actin polymerization, or inactivation of some actin monomer-sequestering proteins at cell edges, can be suggested. The next event (10-20 min) in actin cytoskeleton restoration is the formation of a narrow band of microfilament sheath located along the cell periphery proximal to the 246 T. M. Svitkina, A. A. SeyfakhJrandA. D. Bershadsky Assembly ofactin cytoskeleton 247 edge meshwork. What is the mechanism of sheath formation? We suppose that it is formed from fragments of edge meshwork, which move centripetally and change its fine structure. Indeed, the microfilament sheath begins to grow only after the formation of a massive edge actin meshwork. The fragments of edge meshwork often turn upwards and one can see some of them just in the process of transformation to microfilament sheath in Fig. 17. The biochemical nature of the centripetal movement of the sheath is unclear. It should be noted that in the sheath, in contrast to edge meshwork, actin associates with myosin, and actomyosin could be the driving force for this movement. Thus, we suppose that there is a flow of polymerized actin from the cell periphery to the centre along the cell surface. The hypothesis of actin flow has been suggested (Dunn, 1980; Dembo & Harris, 1981). The centripetal movement of cell surface ruffles and arcs enriched with submembranous actin (Heath, 1983a) supports this idea. This hypothesis can also explain such phenomena as centripetal movement of adsorbed particles and cross-linked surface receptors (Abercrombie et al. 1970, 1972; Harris & Dunn, 1972; Vasiliev et al. 1976; Dembo & Harris, 1981; Heath, 19836). The last step in actin cytoskeleton restoration is the formation of microfilament bundles. Actin bundles can be represented as strips of the sheath stretched between fixed points, i.e. focal contacts. Indeed, these structures seem to have similar protein compositions (both contain myosin, in contrast to active edge meshwork) and a similar pattern of microfilament arrangement. The hypothesis of actin flow can explain the process of bundle formation by the anchoring of flowing actomyosin by the components of focal contacts and its subsequent stretching due to actomyosin contraction. The material in the actin bundles leaves the 'flow' due to the anchoring. This explains the relatively high stability of microfilament bundles that was mentioned above. Many points in the proposed scheme of actin cytoskeleton formation need to be proved and described in detail. Nevertheless, this scheme allows one to consider the actin cytoskeleton as a united dynamic system rather than a collection of independent actin-containing structures.

We are grateful to Professor Ju. M. Vasiliev for permanent encouragement and constructive criticism.

Fig. 15. Part of the cell recovering after NaN3 treatment (20mM, 1 h) for 15min. Microfilament sheath appears at the periphery of the cell while the central part of the cell remains uncovered. Arrows indicate the inner boundary of the sheath. Electron microscopy. Bar, 5 fim. Fig. 16. The boundary between growing sheath (right, top) and the uncovered central part of the recovering cell (left, bottom). Treatment as for Fig. IS. Electron microscopy. Bar, 1 /Mm. Fig. 17. Ruffles in recovering cell (treatment as for Fig. IS). Substrate-attached edge actin meshwork (m), ruffles (r), growing microfilament sheath (s) and uncovered central part of the cell (c) are indicated. Electron microscopy. Bar, 5 /un. 248 T. M. Svitkina, A. A. Neyfakh, jfr and A. D. Bershadsky

REFERENCES ABERCROMBIE, M., HEAYSMAN, J. E. M. & PEGRUM, S. M. (1970). The locomotion of fibroblasts in culture. III. Movement of particles on the dorsal surface of leading lamella. Expl Cell Res. 62, 389-398. ABERCROMBIE, M., HEAYSMAN, J. E. M. & PEGRUM, S. M. (1972). The locomotion of fibroblasts in culture. V. Surface marking with concanavalin A. Expl Cell Res. 73, 536-539. BERSHADSKY, A. D., GELFAND, V. I., SVTTKINA, T. M. & TINT, I. S. (1980). Destruction of microfilament bundles in mouse embryo fibroblasts treated with inhibitors of energy metab- olism. Expl Cell Res. 127, 421-429. BYERS, H. R., WHITE, G. E. & FUJIWARA, K. (1984). Organization and function of stress fibers in cells in vitro and in situ: a review. In Cell and Muscle Motility, vol. 5, The Cytoskeleton (ed. J. W. Shay), pp. 83-137. New York, London: Plenum Press. DEMBO, M. & HARRIS, A. K. (1981). Motion of particles adhering to the leading lamella of crawling cells. J. CellBiol. 91, 528-536. DUNN, G. A. (1980). Mechanisms of fibroblast locomotion. In Cell Adhesion and Motility, BSCB Symp. 3 (ed. A. S. G. Curtis & J. D. Pitts), pp. 409-423. Cambridge University Press. HARRIS, A. & DUNN, G. A. (1972). Centripetal transport of attached particles on both surfaces of moving fibroblasts. Expl Cell Res. 73, 519-523. HEATH, J. P. (1983a). Behaviour and structure of the leading lamella in moving fibroblasts. J. Cell Sd. 60, 331-354. HEATH, J. P. (19836). Direct evidence for microfilament-mediated capping of surface receptors on crawling fibroblasts. Nature, Land. 302, 532-534. HEGENESS, M. H., WANG, K. & SINGER, S. J. (1977). Intracellular distribution of mech- anochemical proteins in cultured fibroblasts. Proc. natn. Acad. Set. U.SA. 74, 3883-3887. HEUSER, J. E. (1983). Procedure for freeze-drying molecules adsorbed to mica flakes. J. molec. Biol. 169, 155-195. KRE1S, T. E., GEIGER, B. & SCHLESSINGER, J. (1982). Mobility of microinjected rhodamine-actin within living chicken gizzard cells determined by fluorescence photobleaching recovery. Cell 29, 835-845. LAL, A. A., BRENNER, S. L. & KORN, E. D. (1984). Preparation and polymerization of skeletal muscle ADP-actin.J. biol. Chem. 259, 13061-13065. POLLARD, T. D. (1984). Polymerization of ADP-actin. J. CellBiol. 99, 769-777. SANGER, J. W. & HOLTZER, H. (1972). Cytochalasin B: effects on cell morphology, cell adhesion, and mucopolysaccharide synthesis. Proc. natn. Acad. Set. U.SA. 69, 253-257. SCHLIWA, M. (1982). Action of cytochalasin D on cytoskeletal networks. J. CellBiol. 92, 79-91. SMALL, J. V., ISENBERG, G. & CELJS, J. E. (1978). Polarity of actin at the leading edge of cultured cells. Nature, bond. 272, 638-639. SVTTKINA, T. M., SHEVELEV, A. A., BERSHADSKY, A. D. & GELFAND, V. I. (1984). Cytoskeleton of mouse embryo fibroblasts. Electron microscopy of platinum replicas. Eur. J. Cell Biol. 34, 64-74. VASILIEV, J. M., GELFAND, I. M., DOMNTNA, L. V., DORFMAN, N. A. & PLETJUSHKINA, O. Y. (1976). Active cell edge and movements of concanavalin A receptors on the surface of epithelial and fibroblastic cells. Proc. natn. Acad. Sd. U.SA. 73, 4085-4089. WEBER, K., RATHKE, P. C, OSBORN, M. & FRANKE, W. W. (1976). Distribution of actin and tubulin in cells and in glycerinated cell models after treatment with cytochalasin B. Expl Cell Res. 102, 285-297.

(Received 9 September 1985 - Accepted 13 November 1985)