Inhibition of Fibronectin Matrix Assembly by the Heparin-binding Domain of Vitronectin*

The deposition of fibronectin into the extracellular matrix is an integrin-dependent, multistep process that is tightly regulated in order to ensure controlled matrix deposition. Reduced fibronectin deposition has been associated with altered embryonic development, tumor cell invasion, and abnormal wound repair. In one of the initial steps of fibronectin matrix assembly, the amino-terminal region of fibronectin binds to cell surface receptors, termed matrix assembly sites. The present study was undertaken to investigate the role of extracellular signals in the regulation of fibronectin deposition. Our data indicate that the interaction of cells with the extracellular glycoprotein, vitronectin, specifically inhibits matrix assembly site expression and fibronectin deposition. The region of vitronectin responsible for the inhibition of fibronectin deposition was localized to the heparin-binding domain. Vitronectin’s heparin-binding domain inhibited both β1 and non-β1 integrin-dependent matrix assembly site expression and could be overcome by treatment of cells with lysophosphatidic acid, an agent that promotes actin polymerization. The interaction of cells with the heparin-binding domain of vitronectin resulted in changes in actin microfilament organization and the subcellular distribution of the actin-associated proteins α-actinin and talin. These data suggest a mechanism whereby the heparin-binding domain of vitronectin regulates the deposition of fibronectin into the extracellular matrix through alterations in the organization of the actin cytoskeleton.

The deposition of fibronectin into the extracellular matrix is an integrin-dependent, multistep process that is tightly regulated in order to ensure controlled matrix deposition. Reduced fibronectin deposition has been associated with altered embryonic development, tumor cell invasion, and abnormal wound repair. In one of the initial steps of fibronectin matrix assembly, the aminoterminal region of fibronectin binds to cell surface receptors, termed matrix assembly sites. The present study was undertaken to investigate the role of extracellular signals in the regulation of fibronectin deposition. Our data indicate that the interaction of cells with the extracellular glycoprotein, vitronectin, specifically inhibits matrix assembly site expression and fibronectin deposition. The region of vitronectin responsible for the inhibition of fibronectin deposition was localized to the heparin-binding domain. Vitronectin's heparin-binding domain inhibited both ␤ 1 and non-␤ 1 integrin-dependent matrix assembly site expression and could be overcome by treatment of cells with lysophosphatidic acid, an agent that promotes actin polymerization. The interaction of cells with the heparin-binding domain of vitronectin resulted in changes in actin microfilament organization and the subcellular distribution of the actin-associated proteins ␣-actinin and talin. These data suggest a mechanism whereby the heparin-binding domain of vitronectin regulates the deposition of fibronectin into the extracellular matrix through alterations in the organization of the actin cytoskeleton.
The deposition of fibronectin into the extracellular matrix is a dynamic, multistep process that is normally tightly regulated in order to ensure controlled matrix deposition. In certain disease states, including pulmonary fibrosis and atherosclerosis, loss of this regulation gives rise to either excess or inappropriate fibronectin deposition (1). In addition, reduced fibronectin deposition has been associated with altered embryonic development, tumor cell invasion, and abnormal wound repair (1). The mechanisms that control the rate and extent of fibronectin deposition are only partially understood.
Adherent cells polymerize an insoluble fibronectin matrix by assembling cell-or plasma-derived soluble fibronectin into insoluble fibrils (2). In one of the initial steps of matrix assembly, cell surfaces bind the amino-terminal region of fibronectin in a reversible and saturable manner (3,4). Subsequent homophilic binding interactions are thought to promote the polymerization of the fibronectin molecule into an insoluble matrix (5)(6)(7)(8)(9) and allow for the regeneration of the cell surface amino-terminal binding site (2). The binding of the amino terminus of fibronectin to cell surface receptors, termed matrix assembly sites (3), is mediated by the first five type I repeats of fibronectin (4,10). The molecule(s) that mediates the binding of the amino terminus of fibronectin to cell surfaces has not been definitively identified. It has been proposed that the III 1 module of fibronectin (8,11), large molecule mass molecules (12), and the fibronectin-binding ␣ 5 ␤ 1 integrin (13)(14)(15) interact with the amino terminus of fibronectin on cell surfaces.
Expression of cell surface matrix assembly sites may be rapidly up-and down-regulated by protein kinase C activation, cyclic AMP levels, and 1-oleoyl lysophosphatidic acid (LPA) 1 treatment (16 -18). Activators of protein kinase C, which effect both stress fiber and focal contact formation (19), enhance fibronectin matrix assembly (17). Conversely, increasing intracellular cAMP levels, which disrupt actin stress fibers and cause cell retraction (20), inhibit matrix assembly (16). In addition, studies have demonstrated loss of matrix assembly sites (21) and decreased fibronectin deposition (22) upon disruption of the actin cytoskeleton with cytochalasin D. More recent studies have defined a role for Rho-mediated contractility in the expression of matrix assembly sites and fibronectin deposition (18,23,24). Increasing Rho-stimulated contraction through treatment of cells with either LPA (18,23) or nocodazole (24) or by microinjection of recombinant, constitutively active Rho (23) results in an increase in amino-terminal fibronectin fragment binding and enhanced fibronectin deposition. Taken together, these observations indicate that matrix assembly site expression and fibronectin deposition are regulated, in part, by actin stress fiber formation and the contractile state of the cell.
Although it has been well documented that decreased fibronectin matrix assembly occurs in most transformed and tumor-derived cell lines, little is known about the extracellular signals that regulate matrix assembly activity in normal fibroblasts. Earlier studies have demonstrated that cells adherent to vitronectin exhibit decreased levels of cell surface matrix assembly sites compared with fibronectin-adherent cells (21,(25)(26)(27). These data suggest the possibility that the interaction of cells with vitronectin generates inhibitory signals that regulate matrix assembly site expression. In the present study, we demonstrate that the interaction of cells with vitronectin inhibits matrix assembly site expression and fibronectin deposition. The region of vitronectin responsible for the inhibition of fibronectin matrix assembly was localized to the heparin-binding domain. The inhibitory effect of vitronectin's heparin-binding domain was independent of the integrin receptor used to assemble the fibronectin matrix. Vitronectin-mediated inhibition of matrix assembly site expression could be overcome by treatment of cells with LPA, an agent that promotes actin polymerization (18,24). Moreover, the interaction of fibronectin-adherent cells with the heparin-binding domain of vitronectin resulted in changes in actin microfilament organization and the subcellular distribution of the actin-associated proteins, ␣-actinin, and talin. These data suggest a mechanism whereby the heparin-binding domain of vitronectin down-regulates fibronectin matrix assembly through alterations in the organization of the actin cytoskeleton.
Purification of Plasma Proteins and Fragments-Human plasma fibronectin was purified from a fibronectin-and fibrinogen-rich by-product of Factor VIII production by ion exchange chromatography on DEAEcellulose (Amersham Pharmacia Biotech) as described previously (30). The 70-kDa amino-terminal fragment of fibronectin was generated by limited digestion of intact fibronectin with cathepsin D, followed by gelatin affinity chromatography as described previously (30).
Vitronectin was purified from fibronectin-and fibrinogen-depleted human plasma by heparin-Sepharose (Amersham Pharmacia Biotech) affinity chromatography according to the method of Yatohgo et al. (31). The 40-kDa cell-binding fragment of vitronectin was generated by acid cleavage of intact vitronectin (32). Purified vitronectin was dissolved in 70% formic acid and incubated at 37°C for 46 h. The digested vitronectin was diluted 1:10 with water, lyophilized, and reconstituted in 50 mM Tris, pH 7.6. The 40-kDa cell-binding fragment of vitronectin was separated from heparin-binding fragments by exclusion on a heparin-Sepharose affinity column. Vitronectin fragments that bound to the heparin-Sepharose were eluted with 0.5 M NaCl in 25 mM phosphate buffer, pH 7.4. Both the cell-binding and heparin-binding fragments were dialyzed extensively against PBS prior to use. Purity of protein preparations was assessed by SDS-polyacrylamide gel electrophoresis, and proteins were frozen at Ϫ80°C until use.
Rat tail collagen type I was purchased from Beckton Dickinson. Laminin (purified from EHS tumors) was purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). Thrombospondin was purified from platelet releasate as described previously (72) and was a gift from Dr. Deane Mosher (University of Wisconsin, Madison, WI). Fibrinogen was purified from human plasma (33) and was a gift of Dr. John Kaplan (Albany Medical College).
Polymerase chain reaction (PCR) was used to amplify human fibronectin cDNA encoding the 12th and 13th type III modules of fibronectin (FnIII 12,13 ) (bases 5081-5623). This DNA encodes amino acids Ala-1689 through Thr-1869, which represents the major heparin-binding region of fibronectin (34). Bases are numbered from the A in the codon for the first amino acid of the mature protein (EMBL accession number X02761), and amino acids are numbered from the aminoterminal pyroglutamic acid (35). PCR was also used to amplify human vitronectin cDNA) encoding the heparin-binding domain of vitronectin (Vn HBD ) (bases 1138 -1258) (36,37). This DNA encodes amino acids from Ala-341 through Ala-380 (38,39). The sense primers for the FnIII 12,13 construct (5Ј-CCGGATCCGCTATTCCTGCACCAACTGAC) and the Vn HBD construct (5Ј-CCCGGATCCGCACCCCGCCCCTCCT-TG) were synthesized with a BamHI site (shown in boldface type) at the 5Ј-end. The antisense primers for the FnIII 12,13 construct (5Ј-CCCGA-ATTCCTATAGTGGAGGC GTCGATGACCA) and the Vn HBD construct (5Ј-CCCGAATTCCTAGGCGCGGGATGGCCGGCG) were synthesized with an EcoRi site (shown in boldface type) at the 5Ј-end. Underlined bases introduce a stop codon after the last base in the sequence to be amplified. PCR was performed according to established procedures (40), using human full-length fibronectin cDNA, pFH100 (a gift from Dr. Jean Thiery, Paris, France) or vitronectin cDNA (a gift from Dr. Erkki Ruoslahti, Burnham Institute, La Jolla, CA) as a template. Following restriction enzyme digestion, the PCR-amplified DNA was cloned into the bacterial expression vector pGEX-2T (Amersham Pharmacia Biotech) and transfected into BL21 (DE3) bacteria using standard procedures (40). PCR-amplified DNA was sequenced to confirm that no base changes had been introduced during amplification of the DNA. Fusion proteins were isolated by passing bacterial lysates over glutathioneagarose as described previously (8) and were dialyzed extensively against PBS prior to use. Both GST-FnIII 12,13 and GST-Vn HBD bound avidly to heparin-Sepharose (data not shown).
Recombinant vitronectins were produced using a baculovirus expression system. Nonmutant and mutant vitronectins were PCR-amplified using the sense primer 5Ј-GGCTACCGTTCACAACGA and the antisense primer 5Ј-GGGTCTAGACTACAGATGGCCAGGAGCTGG. The Vn RGE mutant was produced using recombinant PCR (41). The two mutagenic primers were 5Ј-GCAAGCCCCAAGTGACTCGCGGGGAG-GTGTTCACTATGCCGGAGGATGAGT (sense) and 5Ј-ACTCATCCTC-CGGCATAGTGAACACCTCCCCGCGAGTCACTTGGGGCTTGC (antisense). The mutated bases are shown in boldface type. The two outer primers used were the same as those used to amplify nonmutant vitronectin. PCR-amplified DNA was cloned into the baculovirus expression vector, PVL1392 (Pharmingen), using the restriction enzymes NotI and XbaI.
Recombinant viruses were generated according to established procedures (42,43). Viral stocks were prepared using SF21 cells grown in serum-free SF900II medium (Life Technologies, Inc.). Supernatants from infected cells were collected 72 h postinfection. To purify the recombinant vitronectins, supernatants were treated with 8 M urea for 2 h at 20°C and passed over columns of heparin-Sepharose (Amersham Pharmacia Biotech). Resins were washed with 130 mM NaCl, 4 M urea in 25 mM phosphate, pH 6.5, and bound proteins were eluted with 0.5 M NaCl, 4 M urea in 50 mM phosphate buffer, pH 8. The recombinant vitronectins were dialyzed extensively against PBS. Purity of the recombinant protein preparations was assessed by SDS-polyacrylamide gel electrophoresis, and proteins were frozen at Ϫ80°C until use.
Cell Binding Assays-The 70-kDa fibronectin fragment was iodinated with 1.0 mCi of Na 125 I (NEN Life Science Products) using chloramine T as described previously (3). Fibronectin, laminin, vitronectin, thrombospondin, anti-integrin antibodies, and FnIII 9,10 were diluted to 10 g/ml in PBS and coated onto 24-well tissue plates (Corning/Costar) for 3 h at 37°C. Thrombospondin was reduced with 20 mM dithiothreitol for 30 min at 20°C prior to use (44). Fibrinogen was coated onto culture plates at 100 g/ml in PBS. Collagen was coated onto culture plates in 0.02 N acetic acid at 50 g/ml at 4°C overnight. Protein-coated wells were washed three times with PBS before use. To minimize endogenous fibronectin levels during experimental procedures, cells were washed three times with serum-free DMEM and pretreated for 3.5 h with cycloheximide (20 g/ml) (30) in DMEM containing ITSϩ2 (Sigma) as described previously (15). Cells were seeded at 10 5 cells/well in DMEM/ ITSϩ2 with 20 g/ml cycloheximide. Following an overnight incubation, cells were washed three times with serum-free DMEM and incubated for 1 h with 0.5 ϫ 10 6 cpm/ml 125 I-labeled 70-kDa fragment (ϳ20 ng/well) in DMEM containing 0.2% bovine serum albumin (45). Cell adhesion was quantitated in parallel wells by staining with 0.5% crystal violet as described previously (15).
Immunofluorescence Microscopy-For fibronectin matrix assembly assays, non-cycloheximide-treated fibroblasts were seeded onto fibronectin-coated coverslips in 12-well cluster dishes at 10 5 cells/well in DMEM/ITSϩ2 in the absence or presence of 1.7 M GST/Vn HBD , GST/ FnIII 12,13 , or 0.75 M recombinant vitronectin and incubated overnight at 37°C. Cells were fixed and permeabilized, and fibronectin fibrils were visualized using a polyclonal anti-fibronectin antibody followed by a fluorescein isothiocyanate-conjugated goat anti-rabbit antibody (Cappel). Cells were examined using an Olympus BX60 microscope equipped with epifluorescence and photographed using a spot digital camera (Diagnostic Instruments, Sterling Heights, MI).
To assay changes in actin distribution, confluent fibroblasts were washed with serum-free media and incubated overnight in serum-free DMEM. Cells were detached with trypsin/EDTA, washed, and seeded onto fibronectin coverslips in 12-well cluster dishes at 1.5 ϫ 10 4 cells/ well in DMEM/ITSϩ2 in the absence or presence of 3.4 M of GST/ Vn HBD , GST/FnIII 12,13 , or GST. Following an overnight incubation at 37°C, actin filaments were visualized by staining fixed and permeabilized cells with fluorescein isothiocyanate-phalloidin (Molecular Probes, Inc., Eugene, OR). Vinculin was visualized using an anti-vinculin antibody (Sigma) followed by a Texas Red-conjugated goat anti-mouse secondary antibody (Cappel).

Effect of Adhesive Substrates on Matrix Assembly
Site Expression-The polymerization of a fibronectin matrix is a celldependent process that is regulated by the expression of specific matrix assembly sites on the surface of substrate-attached cells (2). These sites have been identified by radioligand binding assays using either 125 I-labeled 70-kDa or 125 I-labeled 27-kDa fragments derived from the amino terminus of fibronectin (3,4). Previous studies have demonstrated that cells adherent to vitronectin do not express matrix assembly sites (21,25,26). To examine further the effect of extracellular matrix molecules on the expression of matrix assembly sites, cycloheximide-pretreated fibroblasts were allowed to adhere and spread on tissue culture wells coated with various adhesive proteins. Matrix assembly site expression on these cells was then measured in a 1-h binding assay using 125 I-labeled 70-kDa fragment. Previous studies have identified a homophilic binding site for the amino terminus of fibronectin within the III 1 module (8,11). Therefore, fibroblasts were also seeded onto the RGD-containing III 9,10 modules of fibronectin, which mediate cell adhesion (9) but do not contain III 1 . As shown in Fig. 1A, cells adherent to vitronectin bound significantly less 125 I-labeled 70-kDa fragment than cells adherent to either laminin, fibronectin, FnIII 9,10 , collagen, thrombospondin, or fibrinogen. These results suggest that among these adhesive proteins, vitronectin is distinct in its ability to down-regulate matrix assembly sites.
Previous studies have shown that the expression of matrix assembly sites on fibroblast monolayers is rapidly up-regulated either by activation of protein kinase C with phorbol esters (47) or by increasing Rho-stimulated contractility through treatment with either LPA (18,23) or nocodazole (24). To determine whether matrix assembly site expression on vitronectin-adherent cells could be similarly increased, cycloheximide-pretreated, vitronectin-adherent cells were incubated with either PMA or LPA for 1 h, and the level of 125 I-labeled 70-kDa fragment binding was assessed. Treatment of vitronectin-adherent cells with 2 M LPA resulted in a 2-fold increase in 125 I-labeled 70-kDa fragment binding to levels similar to that observed on cells adherent to FnIII 9,10 (Fig. 1B). In contrast, treatment of vitronectin-adherent cells with PMA had no affect on 125 I-labeled 70-kDa fragment binding (Fig. 1B). These data indicate that matrix assembly site expression on vitronectinadherent cells can be up-regulated upon stimulation of Rhomediated signaling pathways.
Inhibition of 70-kDa Fragment Binding by Heparin-binding Fragments of Vitronectin-The ability of LPA to stimulate 125 Ilabeled 70-kDa fragment binding to vitronectin-adherent cells suggests the possibility that the decreased basal level of matrix assembly site expression on vitronectin-adherent cells may be due to inhibitory signals generated upon cell adhesion to vitronectin. Vitronectin is a modular glycoprotein that contains domains that mediate binding to both cells and heparin (32). Cell adhesion to vitronectin is mediated by the RGD sequence located in the amino-terminal connecting sequence (32, 48 -50), while vitronectin's heparin-binding activity has been localized to a series of basic amino acid residues in the carboxyl-terminal hemopexin II domain (32,51). To determine which regions of vitronectin are important for regulating matrix assembly site expression, intact vitronectin was subjected to limited proteolysis with formic acid. Cell-and heparin-binding fragments were then separated by heparin affinity chromatography (32). Cycloheximide-pretreated fibroblasts were seeded onto tissue culture wells coated with either the 40-kDa cell-binding fragment of vitronectin or intact fibronectin in the absence and presence of various concentrations of heparin-binding vitronectin fragments. As shown in Fig. 2A, cells adherent to the cell-binding fragment of vitronectin bound 125 I-labeled 70-kDa fragment to a similar extent as cells adherent to intact fibronectin, suggesting that ligation of vitronectin binding integrins is permissive for the expression of matrix assembly sites. The addition of increasing amounts of heparin-binding fragments of vitronectin to cells adherent to either the cell-binding fragment of vitronectin or intact fibronectin resulted in a dose-dependent decrease in 125 I-labeled 70-kDa fragment binding ( Fig. 2A). At the highest concentrations of heparin-binding fragment tested, 125 I-labeled 70-kDa fragment binding to cells decreased to that observed on cells adherent to intact vitronectin ( Fig. 2A).
Adhesion of fibroblasts to vitronectin is mediated primarily by the ␣ v ␤ 3 and ␣ v ␤ 5 integrins (52,53). To determine whether ligation of either ␣ v ␤ 3 or ␣ v ␤ 5 integrins results in the expression of matrix assembly sites, cycloheximide-pretreated cells were seeded onto tissue culture wells coated with the vitronectin integrin receptor-specific peptide, GPenGRGDSP (50), or the anti-␣ v ␤ 3 (LM609) and ␣ v ␤ 5 (P1F6) monoclonal antibodies in the absence or presence of 20 g/ml heparin-binding fragments. Similar numbers of cells adhered to all wells, as assessed by crystal violet staining (data not shown). As shown in Fig. 2B, ligation of integrin receptor ␣ v ␤ 3 or ␣ v ␤ 5 resulted in levels of 125 I-labeled 70-kDa fragment binding that were similar to the level observed on cells adherent to intact fibronectin. The addition of vitronectin heparin-binding fragments decreased 125 I-labeled 70-kDa fragment binding to cells adherent to either GPenGRGDSP or to the anti-␣ v ␤ 3 and -␣ v ␤ 5 monoclonal antibodies to that observed on cells adherent to intact vitronectin (Fig. 2B).
To further characterize the ability of the heparin-binding region of vitronectin to inhibit the cell surface expression of matrix assembly sites, intact vitronectin was preincubated with increasing concentrations of heparin prior to coating onto tissue culture wells. Following a 3-h incubation, unbound protein was removed, and wells were washed with PBS prior to the seeding of cycloheximide pretreated fibroblasts. As demonstrated in Fig. 3 leads to a down-regulation of cell surface matrix assembly site expression.

Localization of the Inhibitory Effect of Vitronectin on Matrix Assembly Site Expression to the Heparin-binding Domain-To
more precisely localize the region of vitronectin responsible for the inhibition of matrix assembly site expression, a recombinant GST fusion protein containing a 39-amino acid sequence encompassing the heparin-binding domain of vitronectin (32,51) was constructed and tested for its ability to alter 125 Ilabeled 70-kDa fragment binding. As shown in Fig. 4, the addition of increasing concentrations of the heparin-binding domain of vitronectin to cells seeded onto fibronectin-coated dishes resulted in a dose-dependent decrease in 125 I-labeled 70-kDa fragment binding. This inhibitory effect was maximal at 0.9 M Vn HBD and was specific to the heparin-binding domain of vitronectin. The addition of equal molar concentrations of either GST alone or the heparin-binding domain of fibronectin (GST/FNIII 12,13 ) (34, 54) to fibronectin-adherent cells had no effect on 125 I-labeled 70-kDa fragment binding (Fig. 4). In addition, incubation of fibronectin-adherent cells with either intact recombinant vitronectin or a recombinant vitronectin in which the RGD sequence had been mutated to RGE (rVn RGE ) resulted in inhibition of 125 I-labeled 70-kDa fragment binding (Fig. 4). This inhibitory effect was maximal at 0.35 M, providing further evidence that the inhibitory effect of vitronectin on matrix assembly site expression is not due to integrin ligation via the RGD sequence of vitronectin. Taken together, these data indicate that the interaction of cells with the heparinbinding domain of vitronectin leads to a down-regulation of matrix assembly site expression.  Fig. 4). As demonstrated in Fig. 5A, control fibroblasts elaborated an extensive fibrillar fibronectin matrix. Similar fibrillar fibronectin staining was observed on cells that had been incubated in the presence of GST/FnIII 12,13 (Fig. 5B). In contrast, treatment of cells with either the heparin-binding domain of vitronectin (Fig. 5C) or intact recombinant vitronectin (Fig. 5D) markedly reduced the assembly of a fibronectin matrix. These data demonstrate that the interaction of cells with the heparin-binding domain of vitronectin inhibits endogenous fibronectin fibril formation and suggests a role for vitronectin in the regulation of fibronectin matrix assembly.

Vitronectin's Heparin-binding Domain Inhibits Fibronectin
Inhibition of Matrix Assembly Site Expression on ␤ 1 Null Cells-Several recent studies indicate that fibronectin-binding integrins other than ␣ 5 ␤ 1 , including ␣ 3 ␤ 1 (55), "activated" ␣ v ␤ 3 (56), and "activated" ␣ IIb ␤ 3 (22), can support fibronectin matrix assembly. To determine whether the heparin-binding domain of vitronectin also down-regulates non-␤ 1 integrin-mediated matrix assembly site expression, cycloheximide-pretreated ␤ 1null (GD25) and ␤ 1 -transfected (GD25␤ 1 ) cells were seeded onto fibronectin-coated tissue culture wells in the absence or presence of either intact vitronectin or the heparin-binding domain of vitronectin. These cells, which lack ␤ 1 -containing integrin receptors due to a targeted knockout of the ␤ 1 gene (57), have been shown to assemble a fibronectin matrix through a mechanism that is thought to be dependent on the ␣ v ␤ 3 integrin (29). As shown in Fig. 6A, the addition of vitronectin to either A1F fibroblasts, ␤ 1 -null, or ␤ 1 -transfected cells resulted in a similar inhibition of 125 I-labeled 70-kDa fragment binding. Similar to the results obtained using A1F fibroblasts (Figs. 1B  and 4), the inhibitory effect of vitronectin on 125 I-labeled 70-kDa fragment binding to GD25 and GD25␤ 1 cells was mediated by the heparin-binding domain of vitronectin and was reversed by treatment of cells with LPA (Fig. 6B). These studies indicate that the interaction of cells with the heparin-binding domain of vitronectin down-regulates both ␤ 1 and non-␤ 1 integrin-mediated matrix assembly site expression.
The Heparin-binding Domain of Vitronectin Induces Actin Filament Reorganization-Several studies have correlated changes in the organization of the actin cytoskeleton with changes in 70-kDa binding site expression and fibronectin matrix deposition (18,(21)(22)(23). To determine whether the heparinbinding domain of vitronectin induced changes in the distribution of actin filaments, serum-starved, non-cycloheximidetreated fibroblasts were seeded onto fibronectin-coated coverslips in the absence or presence of the heparin-binding domains of either vitronectin or fibronectin. Cells were allowed to adhere and spread overnight, and actin was visualized by staining fixed and permeabilized cells with fluorescein isothiocyanate-conjugated phalloidin. As shown in Fig. 7, in both GST-treated cells (panel A) and GST/FnIII 12,13 -treated cells (panel E), actin filaments were organized into thickly bundled stress fibers commonly observed in fibronectin-adherent cells. In contrast, treatment of cells with the heparin-binding domain of vitronectin (3.4 M) resulted in the organization of actin into loose, broadly distributed microfilament nets (Fig. 7C). Under these conditions, 100% of the cells exhibited changes in actin organization. At lower concentrations of Vn HBD (1.7 M), approximately 50% of the cells exhibited readily visible changes in actin microfilament organization (data not shown). Vinculin, a protein that is highly concentrated at focal adhesions (58), was clearly visible under all experimental conditions (Fig. 7, B,  D, and F).
Modulation of Cytoskeletal Interactions by the Heparin-binding Domain of Vitronectin-To further characterize the vitronectin-induced changes in actin cytoskeletal organization, the subcellular distributions of actin-associated proteins from control and GST/Vn HBD -treated cells were examined. As demonstrated in Fig. 8, treatment of cells with the heparin-binding domain of vitronectin resulted in a significant decrease in the level of ␣-actinin associated with the digitonin-soluble pool as compared with levels observed in either control or GST/ FnIII 12,13 -treated cells. This decrease in soluble ␣-actinin was accompanied by an increase in the level of ␣-actinin associated with the digitonin-insoluble fraction (Fig. 8), suggesting that stimulation of cells with the heparin-binding domain of vitronectin triggers the redistribution of soluble ␣-actinin to the actin cytoskeleton. To examine the distribution of other actin-associated proteins, immunoblots were stripped and sequentially reprobed with antibodies directed against either talin or vinculin (58). In contrast with the results obtained with ␣-actinin (Fig. 8) talin detected in the soluble fraction when compared with either control or GST/FnIII 12,13 -treated cells (Fig. 8). No significant differences in the levels of vinculin staining were observed in the soluble fractions obtained from either control or GST/ FnIII 12,13 -or GST/Vn HBD -treated cells (Fig. 8). Taken together, these data indicate that the interaction of fibronectin-adherent cells with the heparin-binding domain of vitronectin stimulates the organization of actin into microfilament nets concomitant with the subcellular reorganization of specific actin-associated proteins, including ␣-actinin and talin. DISCUSSION In the present study, we demonstrate that the interaction of cells with the heparin-binding region of vitronectin inhibits fibronectin matrix assembly site expression and fibronectin deposition. The inhibitory effect of vitronectin's heparin binding domain was seen in both ␤ 1 -dependent and ␤ 1 -independent matrix assembly and could be overcome by treatment of cells with LPA, an agent known to increase actin stress fiber formation (18,24). Moreover, the interaction of fibronectin-adherent cells with the heparin-binding domain of vitronectin resulted in changes in the cytoskeletal organization of actin, as well as the actin-binding proteins ␣-actinin and talin.
Several previous studies have demonstrated that modulation of the actin cytoskeleton alters the ability of cells to assemble a fibronectin matrix (18,(21)(22)(23)(24). Our data indicate that inhibition of fibronectin matrix assembly by the heparin-binding domain of vitronectin is associated with the formation of microfilamentous nets of actin and a marked redistribution of the actin binding protein ␣-actinin to the cytoskeleton associated pool. These cytoskeletal changes were specific for the heparinbinding domain of vitronectin, since differences were not observed upon treatment of cells with the heparin-binding domain of fibronectin. Our findings are in agreement with the recent study by Zhang et al. (27), which shows that the downregulation of matrix assembly by cells adherent to vitronectin is accompanied by changes in cell morphology. Previous studies have demonstrated agonist-induced redistribution of cytoskeletal and signaling proteins in platelets (59,60) and neutrophils (61). Thus, it is possible that the redistribution of cytoskeletal components upon treatment of cells with the heparin-binding domain of vitronectin serves to localize or sequester signaling enzymes involved in regulating matrix assembly.
Recently, it has been shown that LPA up-regulates expression of matrix assembly sites as well as fibronectin polymerization through the Rho-dependent activation of cellular contraction (23,24). This finding suggests that one role for organized actin in matrix assembly is to generate the contractile forces necessary for fibronectin polymerization. Earlier studies have shown that in the absence of ␤ 1 integrins, ␤ 3 integrins can regulate matrix assembly (22,29,56). Vitronectin inhibited matrix assembly site expression on ␤ 1 null cells, suggesting that vitronectin is able to modulate fibronectin matrix assembly regulated by either ␤ 3 or ␤ 1 integrins. The ability of LPA to overcome vitronectin inhibition of matrix assembly suggests that the inhibitory signal from vitronectin may be modulating a pathway important in maintaining levels of contractility sufficient to promote matrix assembly site expression. In contrast, treatment of vitronectin-adherent cells with PMA, an agent that does not increase fibroblast contractility (62), did not result in an increase in matrix assembly site expression. Recent studies have shown that vitronectin-mediated downregulation of 70-kDa binding sites can be partially reversed by the expression of the ␤ 1 integrin and completely reversed when cells expressing ␤ 1 are treated with LPA (27). Therefore, the results from our study as well as the study by Zhang et al. (27) are consistent with a mechanism in which maximum expression of fibronectin matrix assembly sites requires both ␤ 1 integrin and appropriate levels of cellular contractility.
The effect of vitronectin's heparin-binding domain on fi- bronectin matrix assembly may be mediated by the ␣ v ␤ 5 integrin or by heparan sulfate proteoglycans. Vogel et al. (63) have demonstrated by affinity chromatography that the ␣ v ␤ 5 integrin receptor binds to a peptide derived from the heparinbinding domain of vitronectin. This non-RGD-containing peptide is composed of 12 amino acids that are also contained within the GST/Vn HBD construct. Vitronectin has also been shown to colocalize with heparan sulfate proteoglycans on endothelial cells (64). Removal of cell surface proteoglycans by treatment with heparinase (36), ␤-D-xyloside (36), or sodium chlorate (65) inhibits the binding of multimeric vitronectin to cell surfaces, suggesting that the binding of vitronectin to cells is mediated by an interaction between heparan sulfate proteoglycans and the heparin-binding domain of vitronectin.
Extrahepatic synthesis of vitronectin has been demonstrated to be spatially and temporally regulated during development and progression of certain tumors. Vitronectin synthesis is increased in association with migrating cells (66,67) and in several tumors (68 -70). Our data support the hypothesis that, in vivo, vitronectin may serve as a physiologic ligand responsible for the down-regulation of fibronectin matrix assembly sites on fibroblasts at the tumor-stromal interface. Local synthesis of vitronectin would be expected to signal a decrease in the levels of matrix assembly site expression by stromal fibroblasts. Since previous studies have demonstrated that decreased cell migration rates are associated with increased levels of polymerized fibronectin (7), dampening of matrix assembly pathways by vitronectin would provide a mechanism to facilitate tumor invasion of local connective tissue. The synthesis and deposition of vitronectin into the stroma of colorectal adenocarcinoma by stromal fibroblasts (69) may reflect an autocrine mechanism that promotes fibroblast remodeling of the stromal matrix. Recent studies have proposed that maintenance of cellular and tissue architecture may repress the tumor phenotype (71). The role of vitronectin in promoting the matrix remodeling that contributes to the disruption of local tissue architecture associated with tumor progression remains an important avenue of future investigation.