Cellular invasion into matrix beads: localization of ... - Semantic Scholar

Cellular invasion into matrix beads: localization of ft integrins and fibronectin to the invadopodia

SUSETTE C. MUELLER and WEN-TIEN CHEN* Department of Anatomy and Cell Biology, Georgetown University School of Medicine, 3900 Reservoir Rd, N.W., Washington, DC 20007, USA * Author for correspondence

Summary

We have examined the contribution of adhesion mechanisms to cell invasiveness by growing chicken embryo fibroblasts (CEF) or Rous sarcoma virustransformed cells (RSVCEF) on fibronectin-coated crosslinked gelatin beads (FN-beads). RSVCEF attached more readily and spread more rapidly on FNbeads than CEF, suggesting an increase in the adhesion-related motility of the transformed cells. In addition, RSVCEF invaded the FN-beads, but CEF did not, by extending specialized cell surface protrusions called invadopodia at sites of cell invasion. FN removal by RSVCEF cultured on prelabeled fluorescent FN-beads (FL-FN) was evident at sites of invadopodia, and internalized FL-FN occurred in vacuoles near the ventral membrane of cells at sites of FN removal. The precise distribution of FN and integrins in cells invading FN-beads was determined by immunofluorescence and immunoelectron microscopy of frozen thin-sections. In both CEF and RSVCEF, ft integrins and FN occupied separate intracellular compartments during the early stage of spreading on FN-beads. Later, ft integrins were

largely localized at the ventral cell surface of both CEF and RSVCEF. Polyclonal anti-integrin antibody recognizing ft and several a chains, however, labeled both ventral and dorsal cell surfaces. During invasion by RSVCEF, ft integrins were concentrated at extended invadopodia and also colocalized with internalized FL-FN material in phagocytic vesicles. Furthermore, secreted FN was deposited by RSVCEF at the base of invadopodia colocalizing with ft integrin. Both FL-FN matrix removal and formation of the invadopodia were found to be resistant to treatment with GRGDS at concentrations that inhibit the interaction between cells and FNbeads. Thus, the localization of ft integrins to the plasma membrane contacting immobilized FN results in an extremely tight cellular adherence to the matrix bead, that stabilizes invadopodia and also mediates endocytic clearance of degraded FN-matrix material.

Introduction

1987; Ruoslahti and Pierschbacher, 1987). Earlier studies suggest that the FN receptor family of integrins is involved in cell migration through a process of motilitydriven adhesion. FN promotes cell migration in culture (Ali and Hynes, 1977; Rovasio et al. 1983) and is present in the embryo associated with many different cell migrations (Critchley et al. 1979; Kurkinen et al. 1979). The avian TN receptor complex', a class of molecules that probably includes more than one member of the ft integrin family, was also shown to be important for embryonic cell migration (Bronner-Fraser, 1985; Chen et al. 1985a, 1986a; Duband et al. 1986; Darribere et al. 1988; Jaffredo et al. 1988; Hynes et al. 1989). Recently, ft integrins have also been implicated in human tumor cell migration (Akiyama et al. 1990; Yamada et al. 1990). However, it is less clear at present what roles the integrins and FN play in cell invasion. In vitro invasion studies using polyclonal antibodies against integrins appear to implicate some integrin(s) in one or more aspects of the invasion process (Guirguis et al. 1987; Kramer et al. 1989). Similarly, adhesion inhibitory synthetic peptides such as GRGDS, when added along with cells, produce striking competitive inhibition in in

The ability of cells to invade the extracellular matrix (ECM) represents an important component in the response of animals to injury and pathogen infection and during cancer metastasis. In cancer, this process involves cell motility, adhesion, and degradation as demonstrated in a number of different tumor cell types (Liotta, 1986). Similar aspects of adhesion and motility also play an important role in cell migration during development (Trinkaus, 1987). It therefore seems likely that cell invasion might share many of the molecular mechanisms used in migration, in coordination with localized proteolytic activity on the cell surface of highly invasive cells (Chen et al. 1984; Chen, 1989). Although it has been shown that cell adhesion molecules are centrally important in cell migration, it is not clear precisely how these adhesion systems might contribute to the invasion process. Intensive study of integrin receptors has elucidated many adhesion systems having binding affinities for adhesive proteins including fibronectin (FN), laminin and collagen (see recent reviews by Akiyama et al. 1990; Buck and Horwitz, 1987; Hemler, 1988; Hynes, 1987; Juliano, Journal of Cell Science 99, 213-225 (1991) Printed in Great Britain © The Company of Biologists Limited 1991

Key words: fibronectin, integrin, invasion.

213

vitro invasion assays using amniotic basement membranes (Savagner et al. 1986; Gehlsen et al. 1988) and in experimental animal metastasis models (Humphries et al. 1986). It is possible that the transmembrane interaction of integrins with FN and the cytoskeleton can initiate motility-related adhesion events. The organization of integrins in transformed cells was found to be altered to a pattern similar to that found in highly motile cells, a diffuse surface distribution (Chen et al. 19866; Roman et al. 1989; Akiyama et al. 1990). Highly motile cells bind to immobilized FN and other ECM components by localization of Pi integrins and talin at sites of membrane-ECM contacts, which might account for the motility-driven adhesion necessary for cell migration and invasion (Mueller et al. 1989). To examine in greater detail the parallel between adhesive events required for migration and those required for invasion, we use a recently developed crosslinked matrix bead system (Mueller et al. 1989) as a model for studying the early events of cellular invasion into the ECM. This in vitro invasion assay provides a model system that can differentiate at the cellular level between adhesion, migration and degradation. Second, this method provides an assessment of the relative contributions of adhesion molecules, their membrane receptors, and surface-associated degradative activity to cell invasion. Finally, this method allows a precise definition of cell surface structures responsible for invasion. Cell surface extensions are commonly observed in membranes free of contact with the ECM such as ruffles, microspikes and lamellipodia (Trinkaus, 1984), as well as in the membranes of transformed or tumor cells contacting ECM components (Chen et al. 1984; Guirguis et al. 1987; Kramer et al. 1989; Chen, 1989). These cytoskeletal-linked membrane extensions are the basis for cellular movement. Here, we discovered that cell surface extensions called invadopodia are induced upon contact of transformed cells, but not CEF cells, with FN-beads (fibronectin-coated crosslinked gelatin beads). We also found that ft integrincontaining heterodimers are localized at ventral membranes and in the invadopodia, but are not concentrated in membrane extensions free of contact with the ECM. Furthermore, ft integrin heterodimers are involved in the clearance of degraded matrix material via endocytic processes and in the stabilization of invadopodial extensions by adhesive interactions with cellularly deposited FN. We propose that motility-driven invasion involves integrin-mediated adhesive events in combination with localized cell surface degradative activity, and that both are required for the expression of cell invasiveness. Materials and methods Cell culture on crosslinked gelatin beads Chicken embryonic fibroblasts (CEF) and CEF transformed by Rous sarcoma virus (RSVCEF) were cultured as described (Chen et al. 1984; Olden and Yamada, 1977). Crosslinked gelatin beads were prepared as previously described (Mueller et al. 1989) with the modification that larger beads were selected by discarding a 1 g, 5 min pellet, and collecting beads sedimenting during a 30 s spin on setting no. 2 of the Beckman microfuge in a 1.5 ml microfuge tube. A 50 //I bead pellet obtained after glutaraldehyde fixation as previously described (Mueller et al. 1989) was incubated with 50 (>.*

Fig. 7. Distribution of ft integrin and FN in individual invadopodia. The 0.1-jon frozen sections of RSVCEF or CEF cultured with FN-beads for 1 day were labeled with rat mAb ES238 and rabbit anti-FN followed by 5 nm gold anti-rat and 10 nm gold anti-rabbit secondary antibodies. (A) Integrin (5 nm gold) is found in the plasma membrane of the invadopodia (open arrows and the lower right inset), near where FN (10 nm gold) is deposited on the bead surface. The invadopodial extensions have tubular profiles. Vacuoles (Vac), labeled by anti-ft integrin, are associated with the base of the invadopodia. Integrin labeling colocalizes with FN label on the invadopodia near its base (open arrows, and inset right) but is also found at the tips of invadopodia where FN label is less apparent (arrow, and inset left). The tip of an invadopodium intensely labeled for ft integrin was caught in cross-section (arrow, and inset left). (B) The ventral surface of CEF is smooth and the integrin labeling (5nm gold) colocalizes with a continuous layer of FN (10 nm gold). (C) Where the ventral surface of a RSVCEF is smooth, the ft integrin labeling (5 nm gold) colocalizes with a continuous layer of labeling for FN (10 nm gold), ft integrin labeling (5nm gold) of the ventral membrane colocalizes with FN label in CEF (arrow in B) and RSVCEF (arrows in C). Bars, 200 nm.

Cell invasion into matrix beads

221

t

Bead

8A

B

Pm

Vac

Pm

y

Vac

Vac Bead Fig. 8. Distribution of ft integrin and FN in invading transformed cells. The 0.1-/an frozen sections of RSVCEF cultured overnight with FN-beads were labeled with rat mAb ES238 and rabbit anti-FN followed by 5 nm gold anti-rat and 10 nm gold anti-rabbit secondary antibodies (A,B)- Alternatively, sections were single-labeled with polyclonal anti-140K integrin antibody followed by 10 nm gold anti-rabbit secondary antibody (C,D). A and C are higher magnification views of the areas indicated by brackets in B and D. (A,B) FN at the surface of the bead is detected adjacent to the invadopodia. FN and ft integrin appear to colocalize in the vesicles (Vac) and the plasma membrane (Pm) near the invadopodia (A). (C,D) Vacuoles labeled with anti-integrin (Vac) are associated with the base of invadopodia. A higher magnification view of two invadopodia that are labeled with anti-140K is shown in the inset to D. Bars (A,C), 200 nm; (B,D), 500 nm.

Alternatively, since immunoblots from different cell fractionation protocols stain identically for ft integrin using either ES238 or anti-140K (Mueller and Chen, data not shown), the difference in staining observed on frozen thin-sections may be attributed to differences in protein conformation, i.e. after SDS denaturation, differences between ES238 and anti-140K labeling are no longer apparent (except the recognition of a subunits by anti140K). Experiments are in progress to isolate ventral membranes from dorsal membranes so as to compare their composition biochemically with that of known markers. Invadopodial membranes stain positively for ft integrin, with intense labeling in some areas (Fig. 7A, 5 nm gold labeling). The tip of an invadopodium appears to have been included in the section (Fig. 7 A, arrow and inset) so that ft integrin labeling is seen over the en face view of the 222

S. C. Mueller and W.-T. Chen

invadopodium. Colocalization of FN and integrin is often seen at the bases of some of the invadopodia extending into the bead (Figs 7A, 8A, open arrows). Thus, strong adhesion sites containing ft integrin—FN complexes encircle the base of the invadopodia and anchor it despite nearby degradative activity. Since we showed previously that the initial attachment of cells on FN-beads was sensitive to both the FN cellbinding peptide GRGDS and the adhesion inhibiting, antireceptor mAb JG22E (Mueller et al. 1989), we studied the inhibition of invasion processes of cells already attached on FN-beads by the GRGDS peptide, and by exogenous proteases that cleave FN. Both CEF and RSVCEF were detached from the tissue culture dish and from each other in the presence of 0 . 2 - l m g m r 1 GRGDS or 2.5mgml" 1 trypsin (Chen et al. 19856). However, after initial contact

with FN-beads (30min after co-culture), both normal and transformed cells remained attached to FN-beads in the presence of l m g m l " 1 GRGDS overnight. Cells grown on FN-gelatin substratum were resistant to detachment by 2.5mgml~ 1 trypsin. Furthermore, immunofluorescent labeling of cells demonstrated no differences in ft integrin distribution on ventral membrane profiles as well as the appearance of invadopodia in the presence of up to l.Omgmr 1 GRGDS (data not shown). These results suggest that pre-existing strong adhesion sites formed between ft integrin complex and FN preclude the effect of inhibitors on the activity of invadopodia. Discussion Previously, we showed that ft integrin-containing receptors (including the FN receptor) and the cytoskeletal protein talin aggregate at the membrane of CEF or RSVCEF in contact with FN-beads within 10 min (Mueller et al. 1989). In this paper, we examine changes of integrin distribution in the cell as a consequence of the attachment of CEF or RSVCEF to FN-beads at the level of light and electron microscopy using immunomicroscopy of frozen thin-sections. We found that both RSVCEF and CEF make similar adhesive interactions with immobilized FN-beads. However, RSVCEF, but not CEF, express the invasive phenotype and invade FL-FN-beads. The FN-bead method allows us to identify precisely the invadopodia, the membrane structures that are responsible for the initial stage of invasion. Our data show that ft integrincontaining heterodimers are important for the adhesion of invadopodial membranes to FN that is deposited during invasion. Integrin-containing membranes also mediate the clearance of degraded matrix material via endocytic processes. It remains unclear at present how invadopodia are induced after contact of transformed cells with FN-beads, and how ft integrin-containing heterodimers are localized to the invadopodia. Since the invadopodia were only present in transformed cells, the invasion process may depend upon the activity of the oncogene product, pp60v"8rc tyrosine kinase. We found that RSVCEF attached more readily and were consistently more spread on FN-beads than CEF (Figs 1-2). It is possible that the tyrosine phosphorylation of cellular proteins may increase invasion-related motility in the transformed cells to extend invadopodia following the mobilization of cytoskeletal proteins such as vinculin and talin to the plasma membrane at sites of invasion (Chen et al. 19866; Burn et al. 1988; Mueller et al. 1989). RSVCEF degrade exogenous FN at localized sites underneath the transformed cell (Chen et al. 1984; Chen, 1989). The data from this study suggest that RSVCEF internalize exogenous FL-FN matrices by a process closely coupled to proteolytic degradation, since CEF do not degrade FL-FN-beads and do not internalize FL-FN (Fig. 5). The internalized FL-FN is found in integrincontaining vesicles and vacuoles, and integrin labeling is intense on invadopodial membranes (Figs 4-5). At the electron-microscopic level (Figs 7-8), vacuoles appear at the base of individual invadopodia, and both invadopodia and associated vacuoles often label strongly for integrin. These vacuoles represent either integrin in transit to invadopodia or integrin from the invadopodia internalized via phagocytosis. It is likely that integrin and the cell surface proteases responsible for FN degradation are

localized in close physical proximity, probably in the membranes of invadopodia. Binding of integrins to FN would facilitate degradation of FN and its associated ECM by membrane-associated proteases (Chen and Chen, 1987; Aoyama and Chen, 1990) and result in its subsequent endocytosis. The possibility of protease-integrin interaction during invasion is intriguing, particularly since it has been shown that proteolytic fragments of laminin and FN, and an antibody against integrin, stimulate cellular production of collagenases (Turpeenniemi-Hujanen et al. 1986; Werb et al. 1989). Invasive cells appear to utilize concomitantly extracellular degradation and the phagocytic machinery to clear paths for invasion. A common critical stage of cell migration and invasion appears to be motility-driven adhesion that involves the direct adhesive interaction between cells and the ECM. In this study, we have used an anti-ft integrin antibody (Mueller et al. 1988) that recognizes possibly more than one FN receptor, as well as heterodimers that bind collagen and laminin (Hynes et al. 1989; Ogle et al. 1989). The ft integrins are present in extending invadopodia and are concentrated at the base of invadopodia but are absent from membrane extensions free of contact with the FNbeads (compare Figs 4 and 6 with 2 and 5), suggesting an active role for ft-containing heterodimers in adhesion of individual invadopodia to FN substrata, ft integrincontaining heterodimers are translocated to the ventral cell surface of CEF and RSVCEF following adhesion and spreading on FN-beads. During invasion, localized FN secretion may be coupled with integrin-mediated adhesion at invadopodia. RSVCEF secrete FN that is retained on newly exposed gelatin surfaces along the invasion path (Fig. 6). Subsequently, the modified surface of the gelatin bead would allow invadopodia to re-form strong integrinmediated adhesions to the substratum. The staining patterns that we have observed may be evidence of a highly dynamic process involving waves of FN secretion and integrin-mediated adhesion, followed by ECM degradation and integrin-mediated phagocytosis that results in the observed non-uniform distribution of integrin in the membranes of invadopodia (Figs 7-8). In addition to FN receptors, collagen binding integrins might also mediate invadopodial membrane adhesion to the matrix beads directly, bypassing FN, and contribute to the pattern of localized ft integrin staining of the invasive ventral membrane. Consistently, inhibitory GRGDS peptides have an apparent effect on adhesion of migratory cells to the ECM as shown in several in vitro invasion assays using GRGDS peptides and monoclonal antibodies against human ft integrin (Akiyama et al. 1989, 1990; Yamada et al. 1990). Similarly, adhesion-inhibitory synthetic peptides have been shown to inhibit the penetration of embryonic cells and tumor cells through amniotic basement membranes (Savagner et al. 1986; Gehlsen et al. 1988) and block metastasis in an experimental model (Humphries et al. 1986). The lack of effect of GRGDS on RSVCEF invasion into FN-beads compared to the effects seen in the above mentioned studies may be due to the extensive crosslinking between integrin localized in the contact membrane and the high local concentrations of FN immobilized on the bead surface (derived from both exogenous and endogenous FN) that maintains the stability of invadopodial membranes. It may also differ because of an increased requirement for degradation relative to migration for invasion into FN-beads. Alternatively, other non-RGD binding receptors participate during RSVCEF invasion into matrix beads. As shown for Cell invasion into matrix beads

223

mammalian integrins, & heterodimers that are insensitive to RGDS peptides include those containing a-chains 1, 2, 3 and 4 (Wayner and Carter, 1987; Wayner et al. 1988; Kramer and Marks, 1989). It seems likely that the combination of a tightly adherent membrane possessing protrusive activity, localized degradative activity and phagocytosis of released matrix material results in the formation of invasive channels into the matrix bead. This process is related to motility-driven adhesion and we propose that it be referred to as motility-driven invasion. This distinguishes this localized type of invasion observed in RSVCEF from other types that might depend on secreted, diffuse proteolytic activity. Motility-driven. invasion probably contributes significantly to the efficiency by which cells invade through ECM barriers and might be a general feature of highly invasive cell types. We thank Maozheng Dai and Haining Dai for ultrathin frozen sectioning, Yunyun Yeh and Robert Delsite for cell culture, and Tom Kelly for critical review of the manuscript. This investigation was supported by USPHS grant number R01 CA-39077 and R01 HL-33711 to W.-T.C, and an AHA Grant-in-Aid to S.C.M.

References AKIYAMA, S. K., NAGATA, K. AND YAMAHA, K. M. (1990). Cell surface

receptors for extracellular matrix proteins. Biochim. biophys. Acta 1031, 91-100. AKIYAMA, S. K., YAMADA, S. S., CHEN, W.-T. AND YAMADA, K. M. (1989).

Analysis of fibronectin receptor function with monoclonal antibodies: roles in cell adhesion, migration, matrix assembly, and cytoskeletal organization. J. Cell Biol. 109, 863-875. ALI, I. U. AND HYNES, K. O. (1977). Effects of cytochalasin B and colchicine on attachment of a major surface protein of fibroblasts. Biochim. biophys. Acta ill, 16-21. AOYAMA, A. AND CHBN, W.-T. (1990). A 170-kDa membrane-bound protease is associated with the expression of invasiveness by human malignant melanoma cells. Proc. natn Acad. Sci. U.S-A. 87, 8296-8300. BBONNXB-FBASER, M. E. (1985). Alterations in neural crest migration by a monoclonal antibody that affects cell adhesion. J. Cell Biol. 101, 610-617. BOCK, C A. AND HOBWITZ, A. F. (1987). Cell surface receptors for extracellular matrix molecules. A. Rev. Cell Biol. 3, 179-205. BURN, P., KUPFER, A. C. AND SINGER, S. J. (1988). Dynamic membrane-

cytoskeletal interactions, specific association of integrin and talin arises in vivo after phorbol ester treatment of peripheral blood lymphocytes. Proc. natn. Acad. Sci. U.S-A. 86, 497-501. CHEN, J. M. AND CHBN, W.-T. (1987). Fibronectin-degrading proteases from the membranes of transformed cells. Cell 48, 193-203. CHEN, W.-T. (1989). Proteolytic activity of specialized surface protrusions formed at rosette contact sites of transformed cells. J. exp. Zool. 251, 167-186. CHKN, W.-T., CHEN, J.-M AND MUELLER, S. C. (1986a). Coupled

expression and colocalization of 140K cell adhesion molecules fibronectin and laminin during morphogenesis and cytodifferentiation of chick lung cells. J. Cell Biol. 103, 1073-1090. CHKN, W. T., GREVB, J. M., GOTTLIEB, D. I. AND SINGER, S J. (1985a).

The immunocytochemical localization of 140 Kd cell adhesion molecules in cultured chicken fibroblasts, and in chicken smooth muscle and intestinal epithelial tissues. J Histochem. Cytochem. 33, 576-586 CHEN, W.-T., HASEGAWA, E., HASEOAWA, T., WEINSTOCK, C. AND

YAMADA, K. M. (19856). Development of cell surface linkage complexes in cultured fibroblasts. J. Cell Biol. 100, 1103-1114. CHEN, W.-T., OLDEN, K., BERNARD, B. A. AND CHU, F.-F. (1984).

Expression of transformation-associated proteased) that degrade flbronectin at cell contact sites. J. Cell Biol. 98, 1546-1555. CHEN, W.-T., WANG, J., HASEOAWA, T., YAMADA, S. S. AND YAMADA, K.

M. (19866). Regulation of fibronectin receptor distribution by transformation, exogenous fibronectin, and synthetic peptides. J. Cell Biol. 103, 1649-1662. CRITCHLEY, D. R., ENGLAND, M. A., WAXBLY, J. AND HYNES, R. O. (1979).

Distribution of fibronectin in the ectoderm of gastrulating chick embryos. Nature 280, 498-500. DARRIBERB, T., YAMADA, K. M , JOHNSON, K. E. AND BOUCAUT, J. C.

224

S. C. Mueller and W.-T. Chen

(1988). The 140-kDa fibronectin receptor complex is required for mesodermal cell adhesion during gastrulation in the amphibian Pleurodeles waltlii. Devi Biol. 128, 182-194. DUBAND, J. L., ROCHEH, S., CHEN, W.-T., YAMADA, K. M. AND THIERY, J.

P. (1986). Cell adhesion and migration in the early vertebrate embryo: location and possible role of the putative fibronectin receptor complex. J. Cell Biol. 102, 160-178. GEHLSEN, K. R., ARGRAVES, W. S., PIEBSCHBACHBR, M. D. AND

RUOSLAHTI, E (1988). Inhibition of in vitro tumor cell invasion by Arg-Gly-Asp-containing synthetic peptides. J. Cell Biol. 106, 925-930. GUIRGUIS, R., MARGULTSS, I., TARABOLETTI, G., SCHIFFMANN, E. AND

LIOTTA, L. (1987). Cytokine-induced pseudopodial protrusion is coupled to tumour cell migration. Nature 329, 261-263. HEMLBR, M. E. (1988). Adhesive protein receptors on hematopoietic cells. Immun. Today 9, 109-113. HUMPHRIES, M. J., OLDEN, K. AND YAMADA, K. M. (1986). A synthetic

peptide from flbronectin inhibits experimental metastasis of munne melanoma cells. Science 233, 467-470. HYNES, R. O. (1987). Integrins: a family of cell surface receptors. Cell 48, 649-564. HYNES, R. O., MARCANTONIO, E. E., STEPP, M. A., URRY, L. A. AND YEE,

G. H. (1988). Integrin heterodimer and receptor complexity in avian and mammalian cells. J. Cell Biol. 109, 409-420. JAFFREDO, T., HoRwrrz, A. F., BUCK, C. A., RONG, P. M. AND DIETEBLEN-

LIEVRE, F. (1988). Myoblast migration specifically inhibited in the chick embryo by grafted CSAT hybridoma cells secreting an antiintegrin antibody. Development 103, 431-446. JULLANO, R. L. (1987). Membrane receptors for extracellular matrix macromolecules: relationship to cell adhesion and tumor metastasis. Biochim. biophys. Acta 907, 261-278. KRAMER, R. H. AND MARKS, N. (1989). Identification of integrin collagen receptors on human melanoma cells.