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

By immunofluorescence and interference reflection microscopy (IRM) we found that F-actin and a group of cytoskeletal proteins involved in microfilament-membrane interaction, including vinculin, alpha-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 alpha-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 fibroblasts 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 fibroblasts and in monocyte-derived cells may reflect a common ability to migrate across anatomical boundaries.     

Proteolytic activity of specialized surface protrusions formed at rosette contact sites of transformed cells

Surface protrusions at the leading edge of a moving cell that make contact with the surrounding extracellular matrix (ECM) are its main motor for locomotion and invasion. Chicken embryonic fibroblasts transformed by Rous sarcoma virus (RSV-CEF) form specialized membrane rosette-shaped contact sites on planar substrata as shown by interference reflection microscopy (IRM). Such activity is lacking in normal cells. These rosette contacts are more labile than other adhesion sites, such as focal and close contacts. Ultrastructural studies demonstrate that rosettes are sites at which membrane protrusions from the ventral cell surface contact the substratum. These protrusions are filled with meshworks of microfilaments and contain the pp60src oncogene product, actin, vinculin, and alpha-actinin. However, unlike focal contacts, at the rosettes these proteins interact to extend a highly motile membrane. Rosettes have the biological activity of degrading ECM components, as demonstrated by (1) local degradation of fibronectin substrata at sites of rosette contacts, but not focal and close contacts; (2) localization of putative antiprotease antibody at sites of rosette contacts, but not at focal an close contacts; and (3) local disruption of fibronectin matrix at sites of protrusive activity seen by transmission electron microscopy (TEM). In addition, formation of the rosette contact is insensitive to the ionophore monensin, and to inhibitors of proteolytic enzymes, while local fibronectin degradation at rosette contacts is inhibited by inhibitors of metalloproteases, 1,10-phenanthroline and NP-20. I consider these membrane protrusions of the rosette contacts in RSV-transformed cells specialized structural entities--invadopodia--that are involved in the local degradation of the ECM.     

Adhesion of L1210 cells to sulfonated styrene copolymer surfaces: imaging of F-actin AND alpha-actinin

The static adhesion of living L1210 cells to sulfonated copolymer surfaces of different sulfonic group content and the actin cytoskeleton organization in the adhering cells were studied. The strength of the cell-substratum interaction was estimated by determining the relative number of cells remaining adherent despite experiencing a shearing force equal to 1.25 x 10(-11) N caused by the laminar flow of the medium. The cell-substratum interaction took place in a medium with or without serum. The distribution of F-actin and alpha-actinin in the adhering cells was determined in sequences of fluorescent images of cell optical slices with the use of a computer method of cell image analysis. It was shown that the surface sulfonic groups affect not only the rate and strength of cell-substratum adhesion but also the F-actin and alpha-actinin distribution (in the cell regions near the substratum surface) in cells adhering in the medium containing serum. These proteins, concentrated in the tips of microvilli, were observed as dots. The distinctness (discernibleness) and sizes of these dots depend on the surface content of sulfonic groups. F-actin is located at the periphery of the cells in cells adhering in the medium without serum and alpha-actinin is concentrated in small dots at the periphery and in the central part of the cells.     

Unphosphorylated gelsolin is localized in regions of cell-substratum contact or attachment in Rous sarcoma virus-transformed rat cells

Regions associated with cell-substratum contact or attachment in Rous sarcoma virus (RSV)-transformed rat fibroblasts (RR1022 cells) were identified by reflection-interference microscopy. Electron microscopy of such regions revealed the presence of discrete membrane-associated structures composed of a paracrystalline lattice of hexagons and pentagons to which actin filaments appear to be attached. Staining of actin by biotin-labeled heavy meromyosin showed that transformed cells, unlike normal fibroblasts, lack prominent actin fibers, and that, instead, much of the fluorescence is concentrated in loci corresponding to locations of transient association between the cell and the substratum. In stationary cells, such loci were found in rosette formation, predominantly in the region beneath the nucleus. In cells engaged in active movement, such as during migration into a wound, the actin-containing spots were concentrated in the region of the leading edge. A similar pattern of staining was observed with antibody to gelsolin, a 91,000-dalton Ca2+-dependent actin filament-shortening protein. Since the action of gelsolin on actin is reversible and dependent on physiologically relevant changes in calcium concentration, the localization of gelsolin, together with actin-bundling proteins such as alpha-actinin, in the regions containing many small microfilament bundles on the ventral side of cytoplasm suggests that gelsolin may be a component of the mechanism for the disassembly and assembly of actin during the dissolution and reformation of structures for cell-substratum contact during cell locomotion. Regulation of gelsolin activity was not dependent on protein phosphorylation, as shown by lack of 32P-incorporation into gelsolin in either transformed or normal fibroblasts.     

Rous sarcoma virus-transformed fibroblasts adhere primarily at discrete protrusions of the ventral membrane called podosomes

Rous sarcoma virus-transformed BHK cells (RSV/B4-BHK) adhere to a fibronectin-coated substratum primarily at specific dot-shaped sites. Such sites contain actin and vinculin and represent close contacts with the substratum as revealed by interference reflection microscopy. Only a few adhesion plaques and actin filament bundles can be detected in these cells as compared to untransformed parental fibroblasts. In thin sections examined with transmission electron microscopy (TEM) these adhesion sites correspond to short protrusions of the ventral cell surface that contact the substratum at their apical portion. These structures, which may represent cellular feet, are therefore called podosomes. By screening a number of different transformed fibroblasts plated on a fibronectin-coated substratum we find that podosomes are common to mammalian and avian cell lines transformed either by Rous sarcoma virus (RSV) or by Fujinami avian sarcoma virus (FSV), whose oncogenes encode specific tyrosine kinases. Using antibodies reacting with phosphotyrosine in immunofluorescence experiments, we show that phosphotyrosine-containing molecules are concentrated in podosomes. Podosomes are not detected in fibroblasts transformed by other retroviruses (Snyder-Theilen sarcoma virus, Abelson leukemia virus and Kirsten sarcoma virus) or by DNA tumor viruses (polyoma, SV40), indicating that podosome-mediated adhesion in transformed fibroblasts is related to the peculiar properties of some oncoproteins and possibly to their tropism for adhesion systems. Podosomes and adhesion plaques, although similar in cytoskeletal protein composition, have different mechanisms and kinetics of formation. Assembly of podosomes, in fact (i) does not require fetal calf serum (FCS) in the adhesion medium, that is necessary for the organization of adhesion plaques; (ii) does not require protein synthesis; and (iii) is insensitive to the ionophore monensin, that prevents adhesion plaque formation. Moreover, during attachment to fibronectin-coated dishes, podosomes appear in the initial phase (60 min) of attachment, while adhesion plaques require a minimum of 180 min. In conclusion podosomes of RSV- and FSV-transformed fibroblasts represent a phenotypic variant of adhesion structures.     

Actomyosin organization in stationary and migrating sheets of cultured human endothelial cells

We used immunofluorescence microscopy to study the organization of actin, myosin and vinculin in confluent endothelial cells and in cells migrating into an experimental wound and interference reflection microscopy to assess the cell-substratum adhesion pattern in these cells. In confluent stationary endothelial cell monolayers actin showed a distinct cell-to-cell organization. Myosin, on the other hand, was diffusely distributed and was clearly absent from cell peripheries. Vinculin was confined as linear arrays to cell-cell contact areas. Interference reflection microscopy revealed areas of close and distant adhesion but no focal adhesion sites in these cultures. Twelve hours after experimental wounding a distinct zone of advancing cells was seen at the wound edge. These cells showed a spreadout morphology and, in contrast to stationary cells, had a stress fibre-type organization of both actin and myosin. Vinculin was in the migrating cells seen as plaques at the ventral cell surface. In interference reflection microscopy numerous focal adhesions were seen. The results indicate that the actomyosin system forms the structural basis for monolayer organization of endothelial cells and responds by reorganization upon cell migration.     

Amoeboid locomotion of Acanthamoeba castellanii with special reference to cell-substratum interactions

The amoeboid locomotion of Acanthamoeba castellanii has been studied by observation of individual cells moving on a planar glass substratum. Cell-substratum interactions involved in traction have been observed by reflexion interference microscopy. A variable part of the ventral surface of A. castellanii formed a protean platform, the 'associated contact', from which filopodia were subtended; these established stable, focal adhesions (approximately 0.4 micron diameter) on the substratum beneath. Surprisingly, acanthopodia, a prominent feature of this protozoon, did not play an obvious role in traction. The dimensions of the cell-substratum gap in the associated contact could be modulated by the concentration of ambient electrolyte. Dilution of electrolyte from 50 mM-KC1 to 2mM resulted in (i) an increase in the cell-substratum gap, (ii) a marked decrease in cell motility, (iii) reduced cell adhesion to glass.     

Immunoelectron microscopic studies of the sites of cell-substratum and cell-cell contacts in cultured fibroblasts

Our object was to obtain information about the molecular structures present at cell-substratum and cell-cell contact sites formed by cultured fibroblasts. We have carried out double immunoelectron- microscopic labeling experiments on ultrathin frozen sections cut through such contact sites to determine the absolute and relative dispositions of the three proteins fibronectin, vinculin, and alpha- actinin with respect to these sites. (a) Three types of cell-substratum and cell-cell contact sites familiar from plastic sections could also be discriminated in the frozen sections by morphological criteria alone, i.e., the gap distances between the two surfaces, and the presence of submembranous densities. These types were: (i) focal adhesions (FA); (ii) close contacts (CC); and (iii) extracellular matrix contacts (ECM). This morphological typing of the contact sites allowed us to recognize and assign distinctive immunolabeling patterns for the three proteins to each type of site on the frozen sections. (b) FA sites were immunolabeled intracellularly for vinculin and alpha- actinin, with vinculin labeling situated closer to the membrane than alpha-actinin. Fibronectin was not labeled in the narrow gap between the cell surface and the substratum, or between two cells, at FA sites. Control experiments showed that this could not be ascribed to inaccessibility of the FA narrow gap to the immunolabeling reagents but indicated an absence or severe depletion of fibronectin from these sites. (c) CC sites were labeled intracellularly for alpha-actinin but not vinculin and were labeled extracellularly for fibronectin. (d) ECM sites were characterized by large separations (often greater than 100 nm) between the cell and substratum or between two cells, which were connected by long cables of extracellular matrix components, including fibronectin. In late (24-36 h) cultures, ECM contacts predominated over the other types. ECM sites appeared to be of two kinds, one labeled intracellularly for both alpha-actinin and vinculin, the other for alpha-actinin alone. (e) From these and other results, a coherent but tentative scheme is proposed for the molecular ultrastructure of these contacts sites, and specific functional roles are suggested for fibronectin, vinculin, and alpha-actinin in cell adhesion and in the linkage of intracellular microfilaments to membranes at the different types of contact sites.     

Adhesion structures and their cytoskeleton-membrane interactions at podosomes of osteoclasts in culture

The organization of the cytoskeleton in the podosomes of osteoclasts was studied by use of cell shearing, rotary replication, and fluorescence cytochemical techniques. After shearing, clathrin plaques and particles associated with the cytoskeleton were left behind on the exposed cytoplasmic side of the membrane. The cytoskeleton of the podosomes was characterized by two types of actin filaments: relatively long filaments in the portion surrounding the podosome core, and highly branched short filaments in the core. Individual actin filaments radiating from the podosomes interacted with several membrane particles along the length of the filaments. Many lateral contacts with the membrane surface by the particles were made along the length of individual actin filaments. The polarity of actin filaments in podosomes became oriented such that their barbed ends were directed toward the core of podosomes. The actin cytoskeletons terminated or branched at the podosomes, where the membrane tightly adhered to the substratum. Microtubules were not usually present in the podosome structures; however, certain microtubules appeared to be morphologically in direct contact with the podosome core. Most of the larger clathrin plaques consisted of flat sheets of clathrin lattices that interconnected neighboring clathrin lattices to form an extensive clathrin area. However, the small deeply invaginated clathrin plaques and the podosomal cytoskeleton were located close together. Thus, the clathrin plaques on the ventral membrane of osteoclasts might be involved in both cell adhesion and the formation of receptor-ligand complexes, i.e., endocytosis. This work was supported by the following grants to T.A.: Grants-in-Aid for Scientific Research (C) (18592020) from the Ministry of Education, Science, and Culture of Japan and the Miyata Research Fund of Asahi University.     

Raf kinase inhibitor protein positively regulates cell-substratum adhesion while negatively regulating cell-cell adhesion

Raf kinase inhibitor protein (RKIP) regulates a number of cellular processes, including cell migration. Exploring the role of RKIP in cell adhesion, we found that overexpression of RKIP in Madin-Darby canine kidney (MDCK) epithelial cells increases adhesion to the substratum, while decreasing adhesion of the cells to one another. The level of the adherens junction protein E-cadherin declines profoundly, and there is loss of normal localization of the tight junction protein ZO-1, while expression of the cell-substratum adhesion protein beta1 integrin dramatically increases. The cells also display increased adhesion and spreading on multiple substrata, including collagen, gelatin, fibronectin and laminin. In three-dimensional culture, RKIP overexpression leads to marked cell elongation and extension of long membrane protrusions into the surrounding matrix, and the cells do not form hollow cysts. RKIP-overexpressing cells generate considerably more contractile traction force than do control cells. In contrast, RNA interference-based silencing of RKIP expression results in decreased cell-substratum adhesion in both MDCK and MCF7 human breast adenocarcinoma cells. Treatment of MDCK and MCF7 cells with locostatin, a direct inhibitor of RKIP and cell migration, also reduces cell-substratum adhesion. Silencing of RKIP expression in MCF7 cells leads to a reduction in the rate of wound closure in a scratch-wound assay, although not as pronounced as that previously reported for RKIP-knockdown MDCK cells. These results suggest that RKIP has important roles in the regulation of cell adhesion, positively controlling cell-substratum adhesion while negatively controlling cell-cell adhesion, and underscore the complex functions of RKIP in cell physiology.     

Adhesion of cells to protein carpets: do cells' feet have to be black?

In most physiological situations, cell contact with a substratum is mediated by proteins of extracellular matrix. Therefore, an increasing number of cell-substratum adhesion studies employ substrata covered with one or more proteins of extracellular matrix. To visualize the most adhesive cell structures, focal contacts and focal adhesions, the interference reflection microscopy has been widely used. It has been generally accepted that these strongly adhesive structures can be seen as black streaks in interference reflection microscopy. Calculations are presented herein, which although simplified, suggest that when cells are plated on protein-covered substrata, their focal contacts may not always appear black in interference reflection microscopy.     

Effect of cell-substratum interaction on hemicyst formation by MDCK cells

On impermeable substrata MDCK cells, a cell line derived from normal dog kidney, forms a confluent monolayer that is studded with numerous hemicysts. Previous studies with this cell line suggest that thes hemicysts develop as a result of active fluid accumulation between cell sheet and substratum. However, the formation of hemicysts as a multifocal phenomenon is still unexplained. The results presented here show that the hemicysts are not only expressions of active transport of solutes and water, but also of cell-substratum interaction. The increase in number and size of the hemicyst produced by dbcAMP may be explained by a decrease in the adhesive strength to substrata produced by this compound. Moreover, when the strength of the cell-substratum adhesion was increased the number of hemicysts was reduced or abolished. On the contrary, when this strength was reduced, larger hemicysts occurred, covering practically all the area available for growth. Results from cinematographic time lapse studies, showing that 90% of the area of the monolayer is able to produce hemicysts, also suggest that hemicyst formation as a multifocal phenomenon is more an expression of local variations in cell-substratum interaction than of regional changes in transepithelial active transport.     

Effect of cell-substratum interaction on hemicyst formation by MDCK cells

Summary On impermeable substrate MDCK cells, a cell line derived from normal dog kidney, forms a confluent monolayer that is studded with numerous hemicysts. Previous studies with this cell line suggest that these hemicysts develop as a result of active fluid accumulation between cell sheet and substratum. However, the formation of hemicysts as a multifocal phenomenon is still unexplained. The results presented here show that the hemicysts are not only expressions of active transport of solutes and water, but also of cell-substratum interaction. The increase in number and size of the hemicyst produced by dbcAMP may be explained by a decrease in the adhesive strength to substrata produced by this compound. Moreover, when the strength of the cell-substratum adhesion was increased the number of hemicysts was reduced or abolished. On the contrary, when this strength was reduced, larger hemicysts occurred, covering practically all the area available for growth. Results from cinematographic time lapse studies, showing that 90% of the area of the monolayer is able to produce hemicysts, also suggest that hemicyst formation as a multifocal phenomenon is more an expression of local variations in cell-substratum interaction than of regional changes in transepithelial active transport.     

Ultrastructural localization of plasma membrane-associated urokinase- type plasminogen activator at focal contacts

We have recently shown that urokinase-type plasminogen activator (u-PA) and plasminogen activator inhibitor type 1 are both found extracellularly beneath cultured human skin fibroblasts and HT-1080 sarcoma cells, but in distinct localizations. Here, the ultrastructural distribution of u-PA was studied using immunoferritin electron microscopy. In HT-1080 cells, u-PA on the extracellular aspect of the plasma membrane was detected at sites of direct contact of the cell with the growth substratum beneath all parts of the ventral cell surface. The ferritin-labeled adhesion plaques, which were enriched in submembraneous microfilaments, were frequently seen at the leading lamellae of the cells as well as in lamellipodia and microspikes. Besides the cell-substratum adhesion plaques, ferritin label was detected at cell-cell contact sites. Double-label immunofluorescence showed a striking colocalization of u-PA and vinculin in both HT-1080 cells and WI-38 lung fibroblasts, which is consistent with u-PA being a focal contact component. The u-PA-containing focal contacts of WI-38 cells had no direct codistribution with fibronectin fibrils. In WI-38 cells made stationary by cultivation in a medium containing 0.5% FCS, vinculin plaques became highly elongated and more centrally located, whereas u-PA immunolabel disappeared from such focal adhesions. These findings show that plasma membrane-associated u-PA is an intrinsic component of focal contacts, where, we propose, it enables directional proteolysis for cell migration and invasion.     

The Role of Cell Contraction and Adhesion in Dictyostelium Motility

The crawling motion of Dictyostelium discoideum on substrata involves a number of coordinated events including cell contractions and cell protrusions. The mechanical forces exerted on the substratum during these contractions have recently been quantified using traction force experiments. Based on the results from these experiments, we present a biomechanical model of the contraction phase of Dictyostelium discoideum motility with an emphasis on the adhesive properties of the cell-substratum contact. Our model assumes that the cell contracts at a constant rate and is bound to the substratum by adhesive bridges that are modeled as elastic springs. These bridges are established at a spatially uniform rate while detachment occurs at a spatially varying, load-dependent rate. Using Monte Carlo simulations and assuming a rigid substratum, we find that the cell speed depends only weakly on the detachment kinetics of the cell-substratum interface, in agreement with experimental data. By varying the parameters that control the adhesive and contractile properties of the cell, we are able to make testable predictions. We also extend our model to include a flexible substrate and show that our model is able to produce substratum deformations and force patterns that are quantitatively and qualitatively in agreement with experimental data.     

Multiple labeling of cellular constituents by combining surface reflection interference and fluorescence microscopy

In this paper the technique for visualizing cytoskeleton in detergent-extracted cultured cells by surface reflection interference (SRI) microscopy after staining with the protein dye Coomassie Brilliant Blue (SRI-CooB technique) is used in conjunction with fluorescently-labeled antibodies or with other fluorescent probes to detect a number of constituents in the same cultured cell. Because SRI-CooB technique preferentially visualizes microfilament bundles along the ventral aspect of cells adhering to a glass substratum, we feel that this simple and rapid technique has great potential in studies of cell-substratum adhesiveness and of adhesion-related cytoskeletal organization.     

Tumor promoter and fibronectin induce actin stress fibers and focal adhesion sites in spreading human erythroleukemia (HEL) cells

The effects of plasma fibronectin (pFn) and the tumor promoter 12-0-tetradecanoyl-phorbol-13-acetate (TPA) on adhesion and cytoskeletal organization of human erythroleukemia (HEL) cells were studied. HEL cells, that normally grow in suspension, attached rapidly on pFn-coated growth substratum and some cells showed spreading. Upon exposure to TPA most of the cells adhered and showed some degree of spreading also when plated on plastic. The spread cells showed mostly peripheral accumulations of F-actin in addition to actin fibers seen in some of the cells. When the cells were plated in the presence of TPA on pFn or on pFn-fragments, containing the cell binding site, all the cells adhered rapidly, spread extensively, organized prominent F-actin stress fibers and typical ventral plaques of vinculin and alpha-actinin. Both proteins were revealed also in the suspended cells by Western blot analysis. When plated on substratum coated with other pFn-fragments or laminin, the HEL cells did not adhere or spread. Both adhesion on pFn as well as formation of stress fibers in the presence of TPA could be prevented by the synthetic peptide Arg-Gly-Asp-Ser (RGDS). HEL cells were also able to organize typical ventral fibrillar arrays of Fn. Immunostaining and metabolic labeling experiments showed that the cells did not contain or synthesize Fn, indicating that the plaques were formed from exogenous pFn by the cells. The results suggest that Fn and TPA synergistically induce the organization of the actomyosin system in HEL cells by promoting the formation of prominent actin stress fibers and focal adhesion sites.     

Locomotion and adhesion of polymorphonuclear leukocytes. Effects of the supporting substratum

Contact angle measurements have been used to correlate surface hydrophobicity of a supporting substratum with adhesion and locomotion of polymorphonuclear leukocytes. The binding of human serum albumin, a well-known chemokinetic substance, to hydrophilic glass slides gave rise to hydrophobic surfaces with adhesive properties conductive to cell polarization, thus allowing cell locomotion. Parallel contact angle and cell adhesion measurements suggested that albumin modified the cell-substratum interaction by increasing the van der Waals forces of attraction and reducing the electrostatic forces. By allowing cells to adhere to a hydrophobic surface (siliconized glass), it was found that protein could be omitted from in vitro test systems for leukocyte locomotion. It is suggested that quantitatively equal cell adhesion values may, depending on the type of attraction forces working in adhesion to the substratum, result in different locomotion patterns.     

The melanoma proteoglycan: restricted expression on microspikes, a specific microdomain of the cell surface

A cell surface chondroitin sulfate proteoglycan associated with human melanomas and defined by mAb's F24.47 and 48.7 has been characterized biochemically and localized by indirect immunogold electron microscopy. These antibodies recognize distinct epitopes on the intact proteoglycan. In addition, mAb 48.7 also recognizes an epitope on a 250,000-D glycoprotein and is therefore similar to antibody 9.2.27 (described by Bumol, T.F., and R.A. Reisfeld, 1982, Proc. Natl. Acad. Sci. USA., 79:1245-1249). Furthermore, it was shown that the glycosaminoglycan chains released by alkaline borohydride treatment of the proteoglycan recognized by mAb 48.7 had a size of approximately 60,000 D. Since the intact proteoglycan was estimated to be 420,000 D, there are probably three chondroitin sulfate chains attached to the 250,000-D core glycoprotein. Furthermore, an oligosaccharide fraction containing 42% of the 3H activity (glucosamine as precursor) was isolated. Immunolocalization studies using whole-mount electron microscopy revealed that the chondroitin sulfate proteoglycan was present almost exclusively on microspikes, a microdomain of the melanoma cell surface. These processes were present as 1-2-micron structures on the upper cell surface and as longer (up to 20 micron) structures at the cell periphery. Peripheral microspikes were involved in the initial interactions between adjacent cells and formed complex footpads that made contact with the substratum. Immunogold-labeled cells were also thin sectioned and the specific localization of the chondroitin sulfate proteoglycan antigen was quantitated. The data confirmed the results of whole-mount microscopy and demonstrated a statistically significant association of the antigen with the microspike processes as compared with other areas of the cell surface. By using two different mAb's (48.7 and F24.47) that recognize epitopes on either the core glycoprotein or the intact proteoglycan, respectively, we have demonstrated that both molecules have the same restricted distribution at the cell surface. The specific localization of the antigen to microspikes at the cell surface suggests it may play a role in cell-cell contact and cell-substratum adhesion, which could be important in the metastatic process.