Rous sarcoma virus-transformed fibroblasts and cells of monocytic origin display a peculiar dot-like organization of cytoskeletal proteins involved in microfilament-membrane interactions

Experimental

Cell Research

169 (1987) 202-214

Rous Sarcoma Virus-Transformed of Monocytic Origin Display Organization of Cytoskeletal in Microfilament-Membrane PIER

CARLO MARCHISIO, ‘* * DANIELA ALBERTA ZAMBONIN-ZALLONE2

Fibroblasts and Cells a Peculiar Dot-like Proteins Involved Interactions CIRILLO,’ and GUIDO

ANNA TETI,* TARONE’

‘Department of Biomedical Sciences and Oncology, Section of Histology, School of Medicine, University of Torino, 101.26 Torino, Italy and ‘Department of Human Anatomy, School of Medicine, University of Bari, Bari, Italy.

By immunofluorescence and interference reflection microscopy (IRM) we found that Factin and a group of cytoskeletal proteins involved in microfilament-membrane interaction, including vinculin, a-actinin, fimbrin and talin, are specifically organized in discrete dot-like structures corresponding to cell-substratum contact sites in both monocytes and monocyte-derived cells such as macrophages and osteoclasts. These proteins have a precise topological distribution; vinculin and talin form a doughnut-like ring, while actin, fimbrin and a-actinin are organized in dots matching the rings. An identical dot-like organization of F-actin and associated cytoskeletal proteins was also detected in malignant fibroblasts transformed by Rous Sarcoma virus (RSV) but not in the corresponding untransformed cells in culture. It is concluded that RSV transformation induces tibroblasts to express a cytoskeletal organization and a pattern of adhesion that are normally found in cells of monocytic origin. We propose that the occurrence of this cytoskeletal organization in RSV-transformed tibroblasts and in monocyte-derived cells may reflect a common ability to migrate across anatomical boundaries. 0 1987 Academic PKSS, IN.

The major structure formed by F-actin in vitro in normal non-transformed fibroblasts is represented by stress fibres often spanning the whole width of the cell (e.g. [16]). The ventrally located stress fibres end at the cytoplasmic face of the plasma membrane at sites coincident with areas of tight contact with the substratum denoted as adhesion plaques [12, 181. In transformed libroblasts, stress fibres mostly disappear and actin distribution is profoundly altered (e.g. [3]). At the same time, adhesion plaques also appear to be greatly reduced in number and size [38]. In RSV-transformed cells F-actin is predominantly concentrated in dots corresponding to cell-substratum close contact sites [6, 29, 351. Within these sites, Factin is associated with cytoskeletal proteins involved in membrane-cytoskeleton interactions, such as vinculin, a-actinin and fimbrin [7, 10, 19, 331. We have denoted these peculiar cell-substratum contact sites as ‘podosomes’ on the basis * To whom offprint requests should be sent. Address: Dipartimento di Scienze Biomediche Oncologia, Sezione di Istologia, Corso Massimo d’Azeglio 52, I-10126 Torino, Italy.

e

Copyright @I 1987 by Academic Press, Inc. All rights of reproduction in any form reserved 0014-4827187 $03.00

Cytoskeleton

and adhesion

in monocytes

and transformed

fibroblasts

203

of their foot-like appearance at electron microscopy [35]. It has also been shown that they contain tyrosine-phosphorylated proteins [35] which are probably mostly represented by the src gene product ~~60”~“” [29]. Independently, dot-like F-actin-rich adhesion structures have been found in monocytes [6, 7, 461 as well as in osteoclasts [24], specialized cells involved in bone resorption and derived from the fusion of blood monocytes [2, 151. These adhesion sites are likely to be similar to podosomes of transformed cells as they also contain cytoskeletal proteins involved in the membrane connection and bundling of microtilaments, like fimbrin, a-actinin and vinculin [24]. In the paper, we present a comparative study of the distribution of a number of cytoskeletal proteins, like F-actin, a-actinin, vinculin, talin and fimbrin, in RSVtransformed fibroblasts and in non-transformed cells of monocytic origin. The investigation has been extended also to gelsolin, a Ca2+-dependent regulatory protein of F-actin polymerization, reported in similar dot-like cell-substratum contact sites of RSV-transformed fibroblasts [39]. The peculiar pattern of organization of these cytoskeletal proteins in dot-shaped structures occurring in RSVtransformed fibroblasts was also found in chicken peripheral blood monocytes, peritoneal macrophages and bone matrix-derived osteoclasts. Our conclusion is that RSV-transformed fibroblasts share similar cytoskeletal architecture and patterns of adhesion with cells of monocytic origin. Such a feature may be related to a common mechanism for recognizing, interacting with and penetrating the extracellular matrix texture that is apparently shared by monocytic and transformed cells. MATERIALS

AND

METHODS

Cells Primary chicken embryo fibroblasts transformed in vitro by the Schmidt-Ruppin D strain of Rous sarcoma virus (SRD-CEF) were a gift of Dr F. Tatb, Rome. RSV/B4-BHK are baby hamster kidney (BHK) libroblasts transformed by the Bryan high titre strain of Rous sarcoma virus. Osteoclasts were isolated from the medullary bone of laying hens kept on a low-calcium diet [44]. Monocytes were isolated from the buffy coat of hen peripheral blood as previously described [45]. Human monocytes were prepared, with slight modifications of the same method, from buffy coat obtained from a local blood bank. Macrophages were obtained from peritoneal perfusion of either White Leghorn newlyhatched chicks or Swiss mice and cultured as previously reported. All non-human cells were cultured at 37°C in a water-saturated atmosphere of 95 % air-5 % CO* in Dulbecco’s modified Eagle medium (DMEM) containing antibiotics and fungizone and supplemented with 10% fetal bovine serum (FBS). For culturing human cells, DMEM was replaced by RPM1 1640 under the same culture conditions. For immunofluorescence experiments, cells were either replated on plain glass coverslips or on glass coverslips coated with purified plasma fibronectin (lOug/ml) in 150 mM sodium chloride, 10 mM sodium phosphate buffer pH 7.4 (PBS) for 60 min at room temperature.

Antibodies Anti-chicken gizzard vinculin serum was obtained through the courtesy of Dr B Geiger. Rehovot, Israel [13]. This antibody has previously been found to react with avian vinculin both in immunofluorescence microscopy and in Western blotting [24]. A monoclonal mouse antibody against chicken gizzard a-actinin (code no. 353) was purchased from Amersham International, UK, and used at 1 : 200 dilution. This antibody has also been previously employed on various cell types (e.g. 1241). Two Exp Cell

Res

169 (1987)

204

Marchisio

et al.

different batches of antibodies to chicken brush border limbrin were obtained respectively from Dr A. Bretscher, Cornell University, Ithaca, N.Y. and Drs M. Osbom and K. Webern, Max Planck Institute for Biophysical Chemistry, Gottingen, FRG. Fimbrin serum was used at 1 : 100 dilution and affinity-purified antibodies at 50 Kg/ml. These fimbrin antibodies have been extensively characterized (e.g. [4, 71). Anti-chicken gizzard talin was obtained through the courtesy of Dr K. Burridge, Chapel Hill, N.C., and used at 1 : 100 dilution. An extensive immunochemical characterization of this antibody has been published (e.g. [5, 481). A mouse monoclonal antibody against purified human platelet gelsolin [47] was provided by Dr J. Bryan, Houston, Tex., and used as undiluted hybridoma supematant. This antibody was found to cross-react with avian gelsolin. In some experiments the localization of gelsolin was studied with an affinity-purified rabbit antibody raised against chick embryo brain gelsolin [27] and kindly provided by Dr T. Petrucci, Rome, Italy. Both gelsolin antibodies gave essentially the same immunofluorescence pattern. Fluorescein- or rhodamine-tagged rabbit anti-mouse or swine anti-rabbit sera were purchased from DAKO, Denmark and used at 1: 50 dilution.

Immunojluorescence Procedures For immunofluorescence microscopy, coverslip-attached cells were treated essentially as described elsewhere [35]. Briefly, cells were fixed in PBS, pH 7.6, containing 3 % formaldehyde (from paraformaldehyde) and 2% sucrose for 5 min at room temperature and permeabilized in HEPES-Triton buffer (HEPES 20 mM, pH 7.4, sucrose 300 mM, NaCl SO mM, MgCl* 3 mM and Triton X-100 0.5 %) 5 min at 0°C and finally rinsed in PBS. Coverslips were incubated with primary antibodies for 30 min at 37”C, rinsed in PBS and further incubated with the fluorescent second antibody. For simultaneously detecting F-actin [43], 2.5 ug/ml fluorescein- or rhodamine-tagged phalloidin (F-PHD or R-PHD, kindly donated by Professor Th. Wieland, Max Planck Institute for Experimental Medicine, Heidelberg) was added in the second antibody incubation step. In controls, the primary antibody was routinely replaced with either mouse or rabbit preimmune IgGs. Stained preparations were mounted in 50% PBS-glycerol and viewed and photographed as reported [35].

RESULTS Monocytes, Macrophages, Osteoclasts and RSV-Transformed Fibroblasts Display Similar Organization of F-actin

The intracellular organization of F-actin was investigated by fluorescence microscopy using rhodamine-labelled phalloidin (R-PHD) that binds specifically to the polymerized form of actin [43]. In RSV-transformed mammalian fibroblasts (e.g. in BHK-B4 cells, fig. 1 A) Factin is mostly concentrated in dots, whereas filamentous structures are almost completely absent. As shown in fig. 1 C, this is also the case with primary chick embryo fibroblasts transformed by RSV (SRD-CEF). The pattern of distribution is somewhat different, since F-actin dots are fewer and distributed either along the cell edge or scattered over the whole ventral surface. Interference reflection microscopy (IRM), a technique which allows one to evaluate the distance separating the ventral surface of the cell from the substratum [18], was used to analyse cell-substratum interaction in these cells. When observed by IRM, actin-containing dots of RSV-transformed cells were seen to correspond to areas of close contact with the substratum in the form of grey spots (fig. 1 B, D, at arrowheads). These spots are often encircled by a white ring indicating that the ventral membrane surrounding a central contact area is removed from the substratum (see also below). Exp

Cell Res

169 (1987)

Cytoskeleton

and adhesion

in monocytes

and transformed

fibroblasts

205

Fig. 1. R-PHD fluorescence staining of F-actin (A, C, E, G, I, .Z) and IRM patterns (B, D, F, H, K, L) of RSV-transformed tibroblasts (BHK-B4, A and B; SRD-CEF, C and D), mouse peritoneal macrophages (E and F) and chicken blood monocytes (G-K). Individual actin-containing dots (A, C; e.g. at arrowheads) correspond to IRM grey dots (B, D; e.g. at arrowheads) usually surrounded by a clear ring. F-actin dots in mouse peritoneal macrophages appear in clusters (~9 that correspond to IRM grey areas in which discrete dots cannot be resolved (F). In chicken blood monocytes (G) actin dots are mostly located in a paramarginal band and correspond to grey adhesion sites (ZZ). The boxed areas are enlarged in Z-L and show that individual dots may exhibit in IRM either a central dot (at arrowheads) or a central ring (at arrows). Such patterns may represent two different modes of adhesion shown by these ventral structures. Bar, (A-H) 5 urn, (Z-L) 1 urn. Exp Cell

Res 169 (1987)

206

Marchisio

et al.

The investigation has since been extended to cells of monocytic origin on the basis of previously reported data indicating the presence of F-actin dots in these cells [6, 7, 24, 461. In fact, peripheral blood monocytes and peritoneal macrophages prepared from chicken and seeded in vitro did not display the prominent stress fibres typical of most cultured cells. Instead, F-actin was concentrated in dots located preferentially in the paramarginal belt (fig. 1 G). This distribution is also typically shown by other monocyte-derived cells such as osteoclasts (see figs. 3, 4). In mouse peritoneal macrophages, F-actin dots were often present in discrete clusters located both at the leading and at the trailing edge of cells undergoing translational movements (fig. 1 E). Occasionally, dots were observed in compact clusters lying in the cell centre: this pattern was observed mainly in macrophages which appeared to be stationary. Such polymorphic distribution of F-actin dots in mouse macrophages was quite similar to that shown by human blood monocytes in vitro and by human peritoneal macrophages (not shown). IRM analysis of monocytic cells showed that dots correspond to spotty close contacts identical with those described in RSV-transformed cells (fig. 1 H). In mouse macrophages, because of their prominent location in dense clusters, Factin dots usually corresponded to diffuse grey areas in which discrete spots could seldom be discerned (fig. 1 F). However, in flattened monocytes displaying individual discrete dots (fig. 1 G, H, e.g. at boxed areas enlarged in I-L) IRM recording shows that at least two patterns of contact may be considered. One type (indicated by arrows in fig. 1 K, L) has a central grey dot surrounded by a white ring and a further external grey ring. In a second type (fig. 1 K, at arrowheads) the central dot is replaced by a small ring conferring the appearance of a concentric circle target to the whole structure. The tight association of F-actin dots to the adhesion surface could be proved by removing cell bodies by a gentle stream of buffer before fixation. As previously reported [35] this procedure allows the visualization of the footprints of such structures as discrete R-PHD-positive dots firmly attached to the substratum (not shown). Observations in TEM of perpendicular sections of spread blood monocytes suggested that F-actin-containing dots may correspond to short protrusions of the substratum-facing ventral membrane (not shown). Such protrusions are nearly identical with those reported in osteoclasts [241 and in RSV-transformed cells E351. Fig. 2. Immunofluorescence localization of vinculin (B, F), fimbrin (0. J) and a-actinin (M in SRDCEF (B, D), chicken monocytes (F, H) and chicken peritoneal macrophages (J). The left vertical row on the plate shows the same cells stained for F-actin with F-PHD. Arrowheads indicate the correspondence of different cytoskeletal proteins to actin dots. Note that vinculin is organized in rings surrounding (A, B) individual dots and that a-actinin is also periodically associated with microtilament bundles. Bar, 5 urn. Exp Cell

Res 169 (1987)

Cytoskeleton

14-878333

and adhesion

in monocytes

201

and transformedfibroblasts

Exp Cell

Res

169 (1987)

208

Marchisio

et al.

Location of Cytoskeletal Interactions

Proteins

Involved in Membrane-Microfilament

The location of a-actinin, vinculin and fimbrin was investigated using specific antibodies and was compared with the position of F-actin dots simultaneously stained with F-PHD. In RSV-transformed chick embryo fibroblasts these proteins were found to be organized in a peculiar fashion. Vinculin was organized in a rosette-like ring surrounding the F-actin dot (fig. 2A, B) while fimbrin location was superimposable on that of F-actin (fig. 2 C, D) and a-actinin was distributed throughout the whole structure (not shown). Almost identical was the distribution of vinculin (fig. 2 E, F), a-actinin (fig. 2 G, H) and fimbrin (fig. 21, J> in blood monocytes. The above described organization of cytoskeletal proteins is completely superimposable on that described in similar adhesion structures of osteoclasts [24] such as to suggest that they represent identical structures. In the fibroblasts occasionally contaminating the preparation of chick peritoneal macrophages, vinculin, fimbrin and a-actinin were mostly associated with stress fibre endings as described in other cell types [4, 13, 401. We have extended our study to talin, a 215 kD cytoskeletal protein whose association with cell-substratum adhesion structures is superimposable on that of vinculin [5]. In osteoclasts (fig. 3A,B) as well as in monocytes (fig. 3 C,D), macrophages (fig. 3 E, F) and RSV-transformed chicken fibroblasts (fig. 3 G, H> immunofluorescence employing talin antibodies (kindly provided by Dr K. Burridge, Chapel Hill, N.C.) produced a morphological pattern very similar to that obtained with vinculin antibodies (see fig. 2). A minor difference, which may indicate non-identical co-distribution, is that talin antibodies tended to produce a slightly more diffuse staining in the areas where actin dots are concentrated. This could not be appreciated with vinculin antibodies but the significance of this discrepancy awaits higher resolution data. Location of Gelsolin Fibroblasts

in Monocyte-Derived

Cells and in RSV-Transformed

In this investigation we employed a monoclonal antibody raised against human platelet gelsolin (kindly provided by Dr J. Bryan, Houston, Tex.) which was found to recognize chicken gelsolin. In view of the possibility that such mouse monoclonal antibody is not strictly specific to cytoplasmic gelsolin, some control preparations were stained with an affinity-purified rabbit antibody against chick embryo brain gelsolin ([27], kindly provided by Dr T. Petrucci, Rome). Both antibodies gave essentially identical results. In double-fluorochrome staining experiments with F-PHD, gelsolin was colocalized precisely with F-actin and fimbrin in osteoclasts (fig. 4A, B), monocytes (fig. 4 C, D), macrophages (fig. 4 E, F) and RSV-transformed chicken fibroblasts. The latter finding is consistent with the report of Wang et al. [39] that gelsolin is Exp

Cell

Res 169 (1987)

Cytoskeleton

and adhesion

in monocytes

and transformed

fibroblasts

209

3. Immunofluorescence localization of talin in hen osteoclasts (B), chicken monocytes (D), chicken peritoneal macrophages (F) and SRD-CEF (H). The left row on the plate shows the same cells stained with F-PHD. In all cases talin is structured in rings corresponding to actin dots (e.g. at arrows). Bar, 10 pm.

Fig.

present in dot-like adhesion sites of RSV-transformed cells. Gelsolin was never detected in association with the rings in which a-actinin, vinculin and talin surround the F-actin core. As for the cytoskeletal proteins described above, the distribution of gelsolin was identical in RSV-transformed fibroblasts and in monocyte-derived cells. In cells showing ruffles and other polymorphous periphExp Cell

Res 169 (1987)

210

Marchisio

et al.

Fig. 4. Immunofluorescence localization of gelsolin in hen osteoclasts (B), chicken blood monocytes (0) and chicken peritoneal macrophages (fl. The left row shows the same cells stained for F-actin. The matching location of F-actin and gelsolin in the same dots is indicated by arrowheads. Bar, 5 pm.

era1 protrusions,

gelsolin was found in these structures as reported elsewhere [36,

371.

It should be noted that gelsolin was specifically associated with dot-shaped adhesion sites but not with typical adhesion plaques of untransformed tibroblasts (fig. 5A,B). Talin, however, which represents a specific constituent of adhesion plaques [5, 141, was indeed concentrated at such sites (fig. 5 C, 0). This finding lends further support to the idea that dot-like structures represent adhesion sites structurally and biochemically different from adhesion plaques [35]. DISCUSSION In this paper we provide a description of dot-like structures containing a core of F-actin which represent the major feature of the microfilamentous cytoskeleExp

Cell

Res 169 (1987)

Cytoskeleton

and adhesion

in monocytes

and transformed

fibroblasts

211

Fig. 5. Immunofluorescence localization of gelsolin (A) and talin (C) in chicken fibroblasts viewed also in IRM (E and D). Gelsolin does not appear to be concentrated in IRM black adhesion plaques while the correspondence between talin streaks and adhesion plaques is clearly apparent (see arrows). Bar, 10 pm.

ton in both RSV-transformed fibroblasts and in monocytes, macrophages and osteoclasts. These dot-like structures also contain a group of cytoskeletal proteins involved in actin-membrane interaction such as vinculin, a-actinin, talin and fimbrin. In each cell type these structural proteins display a precise topological distribution that is profoundly different from that previously shown in normal avian [24] and mammalian fibroblasts [35]. In fact, whereas a-actinin and fimbrin co-distribute with F-actin, vinculin and talin (and also some a-actinin) form a ring coincident with the actin dot. These highly organized complexes of cytoskeletal proteins are localized at the ventral plasma membrane precisely at areas of cell-substratum adhesion, as documented by IRM. Dot-shaped actin-containing adhesion structures are distinct from adhesion plaques, judging by morphology, type of contact with the substratum, and Exp Cell

Res 169 (1987)

212

Marchisio

et al.

distribution of cytoskeletal proteins. A number of distinctive features can be listed: (1) In both types of adhesion site a high concentration of polymerized actin is present. ,In actin dots, however, there is no obvious correspondence to the endings of individual microfilament bundles. (2) Whereas in actin dots vinculin and talin form a ring coincident with actin, in adhesion plaques these proteins form a hoof between the membrane and the microfilament bundle ending. (3) The type of contact with substratum also differs, as suggested by IRM analysis. Adhesion plaques are arrowhead-shaped areas about l-2 urn in size, generating an intensely dark IRM image indicative of a tight apposition to the substratum. Conversely, actin dots are round-shaped with a diameter of 0.2-0.4 urn. The corresponding IRM image is usually grey and is not as dark as that typical of adhesion plaques. This indicates a greater separation distance from the substratum. The contact with the substratum and the IRM pattern of these structures was rather variable. In fact the IRM grey area, which in most cases corresponded to a dot, sometimes appeared as a grey ring with a clear central spot. Since the distribution of cytoskeletal proteins seemed to be identical in both cases, these patterns probably represent two alternative types of contact that may be in a dynamic equilibrium. In other words, as shown in osteoclasts [24], these structures may appear as simple contacting protrusions, or as more complex adhering suckers. (4) A further distinctive feature is the presence of gelsolin in actin dots, but not in adhesion plaques. This finding may account for the lack of long microfilament bundles being generated from dot-shaped sites. Gelsolin is in fact a protein which, in the presence of uM Ca*+, nucleates G-actin assembly and fragments F-actin polymers by capping their ‘barbed end’ [37], thus preventing further microtilament elongation. How actin dots are formed is a matter for speculation. Certainly their formation occurs with mechanisms and biochemical requirements different from those of adhesion plaques [35]. It has also been reported that actin dots occur in fibroblasts upon transformation with src [6, 7, 10, 29, 351, fps [35] and abl [30] oncogenes, but not after transformation with a number of other agents [35]. The above three oncogenes code for tyrosine kinases which are associated with the inner face of the plasma membrane, or directly with cell substratum contact sites [ll, 21, 26, 29, 32, 411. Moreover accumulation of proteins phosphorylated at tyrosine residues in actin dots was demonstrated in RSV-transformed tibroblasts [35] as well in their residual adhesion plaques [25]. The hypothesis was therefore raised that tyrosine kinase activity might play a role in inducing the dot organization of actin and cytoskeletal proteins. If this hypothesis is correct it should also apply to in the case of actin dots of monocytes and derived cells. Interestingly it has recently been shown [31] that monocytes and macrophages express specifically high levels of the c-fps oncogene-coded protein, a known tyrosine kinase Exp Cell

Res 169 (1987)

Cytoskeleton

and adhesion

in monocytes

and transformed

Jibroblasts

213

activity. This finding is consistent with our hypothesis, but it is obviously not sufficient to prove it. Mechanisms involved in the formation of actin dots are likely to be complex and probably involve a number of other factors. Gelsolin might be one such factor, since it is specifically associated with actin dots and not with adhesion plaques. Moreover, since actin dots correspond to substratum contact sites, a specific pattern of organization of extracellular matrix components might also contribute to dot formation via a transmembrane mechanism. Finally, a further question that should be answered is why RSV-transformed cells acquire a feature which is expressed in cells of monocytic origin. In other words, what are the functional features of monocytes that are adopted by tibroblasts upon transformation and are not expressed in the untransformed state? A common feature of both monocyte-derived cells and tumour cells is the peculiar ability to travel across anatomical boundaries like basement membranes. This phenomenon occurs normally in monocytes and is enhanced in the course of inflammation reactions; it is also thought to be the basis of extravasation during metastatic diffusion and invasive growth. In both cases, this property requires peculiar motile behaviour and the ability to interact with and degrade extracellular matrix components [23, 34,421. In this respect it is worth mentioning that both monocyte-derived cells and transformed fibroblasts are characterized by their considerable motility [l] and by the ability to synthesize and secrete high levels of specific proteolytic enzymes such as plasminogen activator [22] and collagenases 123, 281 which are involved in the degradation of extracellular material. It has been shown that the actin dots of RSV-transformed cells correspond to sites of localized proteolytic activity and degradation of the extracellular matrix [8, 91. Thus it is tempting to speculate that this characteristic cytoskeletal organization is related to a highly motile behaviour and to the ability of recognizing and degrading matrix components. We would like to thank Drs A. Bretscher, K. But-ridge, B. Geiger, T. Petrucci and T. Wieland for generously providing reagents for this study. The technical assistance of MS P. Rossino is gratefully acknowledged. This research was supported by Minister0 della Pubblica Istruzione and Consiglio Nazionale delle Ricerche (CNR) to P. C. M., G. T. and A. Z. 2. Financial support came also from Progetto Finalizzato “Oncologia” to P. C. M. and G. T. (CT 84.00658.44 and 85.02231.44).

REFERENCES 1. Bennett, W E & Cohn, Z, J exp med 123 (1966) 145. 2. Bonucci, E, Clin orthop 158 (1981) 252. 3. Boschek, C B, Jockusch, B M, Friis, R R, Back, R, Grundmann, E & Bauer, H, Cell 24 (1981) 175. 4. Bretscher, A & Weber, K, J cell biol 86 (1980) 335. 5. Burridge, K & Connell, L, Cell motility 3 (1983) 405. 6. Carley, W W, Barak, S L & Webb, W W, J cell biol 90 (1981) 797. 7. Carley, W W, Bretscher, A & Webb, W W, Eur j cell biol 39 (1985) 313. 8. Chen, W T, Olden, K, Bernard, B A & Fong Chu, F, J cell biol98 (1984) 1546 9. Chen, W T, Chen, J M, Parson, S J & Parsons, J T, Nature 316 (1985) 156. Exp

Cell

Res 169 (1987)

214 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31; 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48.

Marchisio

et al.

David-Pfeuty, T & Singer, S J, Proc natl acad sci US 77 (1980) 6687. Feldman, R A, Wang, E & Hanafusa, H, J virol 45 (1983) 782. Geiger, B, Cold Spring Harbor symp quant biol 46 (1981) 671. Geiger, B, Tokuyasu, A, Dutton, H & Singer, S .I, Proc natl acad sci US 77 (1980) 4127. Geiger, B, Volk, T & Volberg, T, J cell biol 101 (1985) 1523. Gothlin, G & Ericsson, J L E, Clin orthop 120 (1976) 201. Henderson, D & Weber, K, Exp cell res 124 (1979) 301. Hovi, T, Lehto, V P & Virtanen, I, Exp cell res 159 (1985) 305. Izzard, C S & Lochner, L R, J cell sci 21 (1976) 129. Lazarides, E & Burridge, K, Cell 6 (1975) 289. Lehto, V P, Hovi, T, Vartio, T, Badley, R A & Virtanen, I, Lab invest 47 (1982) 391. Levinson, A, Courtneidge, S & Bishop, J M, Proc natl acad sci US 78 (1981) 1624. Liotta, L A, Goldfarb, R H, Brundage, R, Siegal, G P, Terranova, V & Garbisa, S, Cancer res 41 (1981) 4629. Liotta, L A, Thorgeirsson, U P & Garbisa, S, Cant met rev 1 (1982) 277. Marchisio, P C, Cirillo, D, Naldini, L, Primavera, M V, Teti, A & Zambonin-Zallone, A, J cell biol 99 (1984) 1696. Marchisio, P C, Di Renzo, M F & Comoglio, P M, Exp cell res 154 (1984) 112. Nigg, E A, Sefton, B M, Hunter, T, Walter, G & Singer, S J, Proc natl acad sci US 79 (1982) 5322. Petrucci, T C, Thomas, C & Bray, D, J neurochem 40 (1983) 1507. Riley, W B & Peacock, E E jr, Proc sot exp biol med 124 (1977) 207. Rohrschneider, L R, Proc natl acad sci US 77 (1980) 3514. Rohrschneider, L R & Najita, L M, J virol 51 (1984) 547. Samarut, J, Mathey-Prevot, B & Hanafusa, H, Mol cell biol 5 (1985) 1067. Sefton, B M, Trowbridge, I S & Cooper, J A, Cell 31 (1982) 465. Shriver, K & Rohrschneider, L R, J cell biol 89 (1981) 525. Strauli, P & Haemmerli, G, Cant met rev 3 (1984) 127. Tarone, G, Cirillo, D, Giancotti, F G, Comoglio, P M & Marchisio, P C, Exp cell res 159 (1985) 141. Yin, H L, Albrecht, J H & Fahonm, A, J cell biol 91 (1981) 901. Yin, H L, Hartwig, J H, Maruiama, K & Stossel, T P, J biol them 256 (1981) 9693. Wang, E & Goldberg, A R, Proc natl acad sci US 73 (1976) 4065. Wang, E, Yin, H E, Krenger, G J, Caliguri, L A & Tamm, I, J cell biol 98 (1984) 761. Wehland, J, Osbom, M & Weber, K, J cell sci 37 (1979) 257. Witte, 0 N, Rosenberg, N & Baltimore, D, J virol 31 (1979) 776. Woolley, D E, Cant met rev 3 (1984) 361. Wulf, E, Deboben, A, Bautz, F A, Faulstich, H & Wieland, T, Proc natl acad sci US 76 (1979) 4498. Zambonin-Zallone, A, Teti, A & Primavera, M V, Anat embryo1 165 (1982) 405. -J cell sci 66 (1984) 335. Zambonin-Zallone, A, Teti, A, Primavera, M V, Naldini, L & Marchisio, P C, J anat 137 (1983) 57. Hwo, S & Bryan, J, J cell biol 102 (1986) 227. Burridge, K & Connell, L, J cell biol 97 (1983) 359.

Received June 25, 1986 Revised version received September 9, 1986

Exp Cell

Res

169 (1987)

Printed

in Sweden