Stimulation of extracellular matrix remodeling by the first type III repeat in fibronectin

The fibronectin matrix contains cryptic sites which are thought to modulate cellular biological responses. One of these sites, located in fibronectin's first type III repeat (III1c), influences signaling pathways that are relevant to cytoskeletal organization and cell cycle progression. The purpose of this study was to identify possible mechanisms responsible for the effects of III1c on cell behavior. Recombinant peptides representing various type III repeats of fibronectin were compared for their effects on fibronectin matrix organization and activation of intracellular signaling pathways. III1c and III13 but not III11c or III10 bound to monolayers of human skin fibroblasts in a dose- and time-dependent manner and were localized to the extracellular matrix. Binding of III13, but not III1c, to matrix was sensitive to heparitinase, suggesting that the association of III1c with the matrix was not dependent on heparan sulfate proteoglycans. Quantitative and morphological assessment indicated that, in contrast to previously published reports, the binding of III1c to cell layers did not result in the loss or disruption of matrix fibronectin. Binding of III1c but not III13 to the extracellular matrix did result in the loss of a conformationally sensitive epitope present within the EDA type III module of cellular fibronectin. III1c-induced loss of the EDA epitope did not require the presence of cells, occurred within 1 hour and was associated with the activation of p38 mitogen-activated protein kinase (MAPK) followed by the formation of filopodia. Maximal phosphorylation of p38 MAPK occurred within 1 hour, whereas cytoskeletal changes did not appear until 12 hours later. These findings are consistent with a model in which the binding of III1c to the extracellular matrix results in a conformational remodeling of the fibronectin matrix, which has both short- and long-term effects on cell physiology.


Introduction
Cells in tissues and in culture assemble an extensive extracellular fibronectin matrix that appears to undergo continual remodeling both in vivo (Deno et al., 1983;Rebres et al., 1995;Thompson et al., 1989) and in vitro (Baneyx et al., 2002;Ohashi et al., 1999). The matrix fibrils are linked to transmembrane integrin receptors which transduce structural information present in the matrix back into the cell. The association of a multidimensional extracellular matrix (ECM) with the surrounding cell layer is thought to have a major role in regulating cell adhesion, migration, cytoskeletal organization, growth, apoptosis and morphogenesis (Assoian and Zhu, 1997;Giancotti and Ruoslahti, 1999;Howe et al., 1998). The findings that cellular adhesion to fibronectin can stimulate cell signaling pathways (Clark and Brugge, 1995;Schwartz and Ingber, 1994) and that the organization of a fibrillar fibronectin matrix is dynamic (Baneyx et al., 2002;Ohashi et al., 1999) suggest that structural remodeling of the fibronectin matrix may directly influence cell behavior. Recent studies have indicated that fibronectin generates distinct cellular responses that are dependent on its assembly into matrix fibrils. For example, the organization and assembly of a fibronectin matrix has been shown to control adhesiondependent cell growth (Bourdoulous et al., 1998;Christopher et al., 1999;Sechler and Schwarzbauer, 1998;Sottile et al., 1998), enhance cell contractility (Hocking et al., 2000), regulate the stability of cell-matrix adhesion sites  and influence embryonic development (Darribere et al., 2000). Alternatively, polymerization of a fibronectin matrix may indirectly affect cell behavior by regulating the ECM composition. The deposition of collagen, thrombospondin and fibrinogen into the ECM have all been shown to be regulated in part by the assembly of a fibronectin matrix (McDonald et al., 1982;Pereira et al., 2002;Sottile and Hocking, 2002;Velling et al., 2002). These data compliment evidence indicating that an in vitro produced multimeric form of fibronectin that resembles matrix fibrils, termed superfibronectin, is functionally distinct from soluble fibronectin (Morla et al., 1994;Okada et al., 1997).
Superfibronectin is produced when a 76 amino acid recombinant fragment derived from the first type III module of fibronectin, termed III1c, is incubated with plasma fibronectin. Homophilic binding between III1c and protomeric fibronectin triggers additional homophilic binding events resulting in the polymerization of soluble fibronectin into insoluble multimers, reminiscent of matrix fibronectin (Morla and Ruoslahti, 1992). Superfibronectin has been shown to increase cell adhesiveness, decrease migration, increase viral infectivity and reduce tumor 4663 The fibronectin matrix contains cryptic sites which are thought to modulate cellular biological responses. One of these sites, located in fibronectin's first type III repeat (III1c), influences signaling pathways that are relevant to cytoskeletal organization and cell cycle progression. The purpose of this study was to identify possible mechanisms responsible for the effects of III1c on cell behavior. Recombinant peptides representing various type III repeats of fibronectin were compared for their effects on fibronectin matrix organization and activation of intracellular signaling pathways. III1c and III13 but not III11c or III 10 bound to monolayers of human skin fibroblasts in a dose-and time-dependent manner and were localized to the extracellular matrix. Binding of III 13, but not III1c, to matrix was sensitive to heparitinase, suggesting that the association of III1c with the matrix was not dependent on heparan sulfate proteoglycans. Quantitative and morphological assessment indicated that, in contrast to previously published reports, the binding of III1c to cell layers did not result in the loss or disruption of matrix fibronectin. Binding of III1c but not III13 to the extracellular matrix did result in the loss of a conformationally sensitive epitope present within the EDA type III module of cellular fibronectin. III1c-induced loss of the EDA epitope did not require the presence of cells, occurred within 1 hour and was associated with the activation of p38 mitogen-activated protein kinase (MAPK) followed by the formation of filopodia. Maximal phosphorylation of p38 MAPK occurred within 1 hour, whereas cytoskeletal changes did not appear until 12 hours later. These findings are consistent with a model in which the binding of III1c to the extracellular matrix results in a conformational remodeling of the fibronectin matrix, which has both short-and long-term effects on cell physiology.
growth, angiogenesis and metastasis (Greco et al., 2002;Morla et al., 1994;Pasqualini et al., 1996;Yi and Ruoslahti, 2001). In vivo, the III1c fragment has been shown to inhibit tumor growth and metastasis (Pasqualini et al., 1996;Yi and Ruoslahti, 2001). Addition of III1c to cultured cells affects signaling pathways, leading to cell cycle progression and cytoskeletal organization (Bourdoulous et al., 1998;Mercurius and Morla, 1998). In particular, III1c has been shown to stimulate the phosphorylation and nuclear localization of p38 MAP kinase (Aguirre-Ghiso et al., 2001;Bourdoulous et al., 1998). The p38 enzyme is a component of one mitogenactivated protein kinase (MAPK) signaling cascade known to regulate diverse cellular outputs in response to various extracellular stimuli (for a review, see Ono and Han, 2000). Previous studies have shown that p38 plays a role in the inhibition of cell cycle progression (Aguirre-Ghiso et al., 2003;Molnar et al., 1997) and alterations in the actin cytoskeleton (Guay et al., 1997;Lee et al., 2001;Rousseau et al., 1997). The mechanism underlying the effect of III1c on cell behavior is largely unknown as the published reports on the effects of III1c on matrix fibronectin have been contradictory. III1c has been reported to promote fibronectin matrix assembly, to inhibit fibronectin matrix assembly, and to cause disassembly and loss of pre-established fibronectin matrix (Bourdoulous et al., 1998;Morla et al., 1994;Morla and Ruoslahti, 1992). Nevertheless, these data provide further evidence that the physical state of the fibronectin matrix plays a crucial role in the regulation of cell behavior.
The assembly of the fibronectin matrix depends on the cellular regulation of sequential homophilic binding events among individual receptor-bound fibronectin molecules. Microaggregates of fibronectin are then remodeled through a contractile process that culminates in the polymerization of a detergent-insoluble disulfide-stabilized fibrillar matrix (reviewed by Schwarzbauer and Sechler, 1999). The fibronectin in the matrix comprises various isoforms that arise from differential splicing of gene transcripts (Schwarzbauer, 1991). Alternative splicing occurs in two type III repeats, termed EDA and EDB, and accounts for a significant portion of the diversity within fibronectin molecules (Mardon et al., 1987;Schwarzbauer et al., 1987). The functions of these alternatively spliced regions are not well understood. The EDA and EDB type III repeats are expressed in very low levels in the adult, but are upregulated at various times during development, wound healing and tumor progression (Coito et al., 1997;Ffrench-Constant and Hynes, 1989;George et al., 2000;Jarnagin et al., 1998;Peters et al., 1986;Scarpino et al., 1999). One of these repeats, the EDA domain which is found between the 11th and 12th type III repeats, has been shown to affect several cellular processes including adhesion (Manabe et al., 1997), differentiation (Gehris et al., 1997) and gene expression (Saito et al., 1999). Recently, the α4β1 and α9β1 integrins have been identified as receptors for the EDA segment (Liao et al., 2002).
The present study was undertaken to examine the interaction of III1c with fibroblast cell layers. Quantitative and morphological assessment indicated that III1c did not cause loss or disruption of matrix fibronectin. Our data indicate that III1c associates with fibronectin fibrils in a dose-and timedependent manner. The association of III1c with the ECM resulted in the loss of a conformationally sensitive epitope within the EDA module of matrix fibronectin. The effect of III1c on fibronectin fibril conformation occurred within 1 hour and was accompanied by the activation of p38 MAPK. These findings are consistent with a model in which the binding of III1c to the ECM results in a rapid conformational remodeling of the pre-existing fibronectin matrix, and suggest that p38 MAPK pathways are intracellular targets of matrix reorganization.

Reagents
Unless otherwise indicated, chemical reagents were obtained from Sigma Chemical Co. (St Louis, MO). Anti-phospho-p38 (Thr-180/Tyr-182) polyclonal antibody and anti-p38 polyclonal antibody were obtained from Cell Signaling Tech (Beverly, MA). Four polyclonal antibodies to human plasma fibronectin were used in this study: two directed towards the entire molecule (Curtis et al., 1995;Kowalczyk et al., 1990); one to the 27 kDa Fib-1/Hep-1 fragment (P. J. M.-L. and T. S. Panetti, unpublished); and one specific for the 40 kDa gelatin binding fragment (P. J. M.-L. and Panetti, unpublished). Seven fibronectin domain-specific monoclonal antibodies were also used in the study: FDB3 antibody recognizes the VTHPGY sequence within the CS domain of fibronectin (Zheng et al., 1994); FDH2 is directed toward the Hep-2 domain of fibronectin; and IST-9 and 5C11F3 recognize the EDA fibronectin type III repeat (Carnemolla et al., 1987) (L.V.D.W. et al., unpublished); Clone 568 to the eighth type III repeat of fibronectin was from Maine Biotechnology Services (Portland, ME); IST-5 to the fifth fibronectin type III repeat was purchased from Chemicon (Temecula, CA); and Clone 10, purchased from Transduction Laboratories (Lexington, KY), recognizes the second type III repeat in fibronectin (P.J.M.-L. and R.M.K., unpublished). A tetra-His monoclonal antibody (Qiagen, Valencia, CA) was used to monitor recombinant proteins. Heparitinase I was purchased from Seikagaku America (Falmouth, MA). Human plasma fibronectin was purified from a fibronectin and fibrinogen-rich byproduct of Factor VIII production via ion exchange chromatography as previously described (Mosher and Johnson, 1983) and further isolated by sequential purification over gelatin and heparin affinity chromatography (Engvall and Ruoslahti, 1977).

Cell culture
Human foreskin fibroblasts (A1-Fs) were from Lynn Allen-Hoffman (University of Wisconsin, Madison, WI). The cells were grown in Dulbecco's Modified Eagle's medium (DMEM; Gibco Laboratories, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Hyclone Labs, Logan, UT) at 37°C in an atmosphere of 7.5% CO2. Fibroblasts in passages 6-12 were used in these studies. In most experiments, cells were plated in the presence of complete medium and cultured for three days to approximately 90% confluency. Cell monolayers were then washed and treated with various fibronectin modules in DMEM containing 5% fibronectin-depleted FBS. Gelatinagarose was used to eliminate fibronectin from FBS; the resulting serum showed no detectable fibronectin by western blot analysis with anti-fibronectin polyclonal antibody.

Recombinant fibronectin modules
All recombinant fibronectin Type III modules were generated through polymerase chain reaction amplification of the human fibronectin cDNA clone pFH100 as previously described (Hocking et al., 1994). The primers used to generate fragments from the first (III1c) and eleventh (III11c) Type III repeats of fibronectin were the same as has been previously described (Morla et al., 1994). III10 sense primer (5′-CCGGATCCGTTTCTGATGTTCCGAGGGAC) was synthesized with a BamHI site (shown in bold) and antisense primer (5′-CCCAA-GCTTCTATGTTCGGTAATTAATGGAAATTGG) containing a HindIII site (shown in bold) and stop codon (underlined bases) after the last base in the sequence to be applied. III13 sense primer (5′-CCC-GCATGCGAGGATCCAATGTCAGCCCACCAAGAAGG) and antisense primer (5′-CCCAGATCTGGATCCAGTGGAGGCG-TCGATGACC) were both synthesized with BamHI sites at their 5′ end. These primers amplify DNA encoding fibronectin amino acids Val-1538 to Thr-1631 for III10; Asn-1904 to Thr-1992 for III13; Asn-631 to Pro-705 for III1c; and Leu-1654 to Thr-1721 for III11c according to the numbering method of Dufour et al. (Dufour et al., 1991). Following restriction enzyme digestion, III10 was cloned into vector pQE-30 (Qiagen) and III13, III1c, III11c were cloned into vector pQE-70 (Qiagen). The resulting plasmids were sequenced to ensure that no base changes had been introduced and to confirm that the PCR amplified DNA was cloned in frame with the bacterial 6xHis tag DNA. DNA was transformed into M15 bacteria using standard procedures (Sambrook et al., 1989). Fusion proteins were purified by affinity chromatography with metal-chelating nitrilotriacetic acidagarose (Ni-NTA) according to the manufacturer's instructions (Qiagen). Recombinant proteins were further purified by passing Ni-NTA eluates through a Sephadex G-25 column followed by cation exchange chromatography using CM agarose resin (Pharmacia). The purity of the protein was ensured with the appearance of a single band by SDS-PAGE (Laemmli, 1970). The concentrations of the proteins were determined by BCA assay with BSA as standard (Pierce, Rockford, IL). Purified stock proteins were stored at -80°C until use.

Quantitative assessment of matrix fibronectin
Fibronectin was iodinated with 125 I-Na (DuPont NEN) by using the chloramine T method (McKeown-Longo and Mosher, 1984). The iodination reaction was stopped with the addition of free tyrosine, and nonincorporated iodine was removed by chromatography on a sephadex G-25 column (Pharmacia). The trace amount of free iodine was further removed by extensive dialysis against PBS and protein was stored at -80°C until use. The integrity of the labeled protein was assessed by gel electrophoresis and autoradiography. Specific activity of radiolabeled fibronectin was from 4.0 to 9.0×10 5 Ci/mol. To obtain radiolabeled fibronectin matrices, fibroblast monolayers were grown for three days in the presence of 2×10 6 cpm/ml 125 I-fibronectin. After removal of 125 I-fibronectin, the cells were treated with various recombinant fibronectin modules. The level of matrix associated fibronectin was determined by measuring deoxycholate (DOC)insoluble material as described previously (McKeown-Longo and Mosher, 1983). Briefly, extractions were done at 4°C in a 0.02 M Tris (pH 8.3) buffer containing 2.0 mM PMSF, 2.0 mM EDTA, 2.0 mM ethylmalmeimide, and 2.0 mM iodoacetic acid and 0.5% DOC. DOC soluble and insoluble fractions were separated by centrifugation. The deoxycholate insoluble matrix fibronectin pellet was then solubilized in SDS-PAGE sample buffer and analyzed by SDS-PAGE with a 4% polyacrylamide gel under both reducing and nonreducing conditions. The radiolabeled matrix fibronectin was visualized by autoradiography.

Fluorescence microscopy
For direct visualization of the fibronectin matrix, fibronectin was derivitized with Alexa Fluor-488 according to manufacturer's protocol (Molecular Probes, Eugene, OR). Briefly, fibronectin was diluted to 0.5 mg/ml in 0.05 M sodium borate buffer, pH 9.0 and directly conjugated to Alexa Fluor-488 using N,N-dimethyl formamide. Derivitized protein was separated from free fluorescein by chromatography on a Sephadex G-25 column. The labeling ratio was determined by the ratio of A 495nm to A 280nm . Alexa Fluor-488 conjugated fibronectin was added to the culture medium 2 hours post cell seeding. Cells were then grown in the presence of derivitized fibronectin for three days. Following treatments with fibronectin modules, cell layers were washed with serum-free media, fixed for 15 minutes in 3.0% paraformaldehyde, permeabilized in 0.3% Triton X-100 for 3 minutes, and blocked with 3.0% bovine serum albumin (BSA) in PBS at room temperature for 2 hours. To visualize fibronectin, the cells were further stained with polyclonal antibodies or monoclonal antibodies directed against various epitopes in the fibronectin molecule. Specific antibodies are described in the figure legends. The cell layers were examined using an Olympus BMX-60 microscope equipped with a cooled CCD sensi-camera (Cooke, Auburn Hills, MI). Images were acquired using Slidebook software (Intelligent Imaging Innovation, Denver, CO).

Binding of recombinant protein to fibroblast monolayers and to purified fibronectin
The monolayers of human skin fibroblasts were incubated with various amounts of His-tagged recombinant protein for different times in 96-well plates. The wells were washed, fixed with 3.0% paraformaldehyde, and blocked with 1.0% BSA in PBS for 2 hours. Bound His-tagged proteins were detected by incubation of cell layers with a mouse anti-tetra His mAb diluted 1:1000 in 0.1% BSA/PBS for 1 hour. Cell layers were washed three times with PBS and incubated with secondary horseradish peroxidase (HRP) conjugated goat anti-mouse IgG (BioMol) for 1 hour. Color was developed by incubating wells with fresh substrate solution (1 µl/ml hydrogen peroxide, 0.5 mg/ml O-Phenylenediamine in 0.1 M citrate buffer, pH 5). The reaction was stopped with 2.0 N sulfuric acid and extent of color development was determined by measuring A 490nm . The wells in the absence of recombinant protein were used as blank control. In some experiments, cell layers were pretreated for two hours with increasing amounts of heparitinase I before III1c treatment. The removal of glycosaminoglycans from heparan sulfate was monitored by ELISA using the 10E4 mAb (Seikagaku America, Falmouth, MA) whose epitope includes N-sulfated glucosamine residues which are critical for the reactivity of the antibody. Alternatively, cell layers were incubated overnight with 35 S-sodium sulfate to label proteoglycans. Cells were then treated with increasing amounts of the enzyme and the loss of heparitinase-sensitive material from the cell layer was quantified by β-scintillation.
Monitoring the expression of fibronectin epitopes in the extracellular matrix fibronectin For titration of fibronectin mAbs, fibroblast cell layers were incubated 2 hours at 37°C with serial dilutions of fibronectin mAbs (8.0-0.0078 µg/ml) diluted in 5.0% fibronectin-depleted FBS/DMEM. Unbound mAb was removed by washing wells with 0.1% BSA/DMEM. Cells were fixed with 3.0% paraformaldehyde, blocked with 1.0% BSA/PBS for 2 hours and incubated with secondary HRP conjugated goat anti-mouse IgG for 1 hour at room temperature. Following extensive washing with PBS, the extent of bound mAb was determined, as described above. Equilibrium binding was determined following nonlinear regression analysis using a single site saturation binding curve equation. These titration curves were used to select the concentration of mAb chosen to assay fibronectin epitope levels in the presence of recombinant type III modules.
Solid phase binding assay III1c binding to purified plasma and cellular fibronectin was measured using solid phase binding assays. Microtiter wells were coated with 20 µg/ml of either plasma or cellular fibronectin for 2 hours at 37°C, washed with PBS and blocked with 1.0% BSA/PBS for 1 hour at room temperature. Increasing concentrations of III1c were added to wells for 1 hour. Unbound protein was removed by washing wells with PBS, and bound recombinant protein was determined using the anti-Tetra His mAb as described above. To determine the effect of III1c on the binding of anti-EDA antibodies to fibronectin, wells were coated with 20 µg/ml of cellular fibronectin and blocked with BSA, as described above. Wells were then incubated with increasing concentrations of III1c for 1 hour. Following treatments, wells were washed with 0.1% BSA/PBS, blocked with 1.0% BSA/PBS and the binding on EDAspecific mAbs, IST-9 (2 µg/ml) or 5C11F3 (8 µg/ml) was determined by ELISA, as described above.

Immunoblot analysis
After treatment, cell layers were washed three times with ice-cold PBS containing 1 mM PMSF, 1 mM Na3VO4, and 10 mM NaF before solubilization in lysis buffer [20 mM Tris-HCl, pH 7.4, 1% Triton X-100, 0.5% Nonidet P-40, 0.1 M NaCl, 40 mM NaF, 30 mM Na4P2O7, 2 mM EGTA, 1 mM Na3VO4, 0.5 mM PMSF, and one tablet of complete protease inhibitor (Roche Diagnostics) per 10 ml]. After incubation on ice for 30 minutes, cell lysates were centrifuged at 20,000 g for 20 minutes at 4°C, and the insoluble pellets were discarded. The protein concentration of the lysate was determined by BCA. Cell lysates containing equal amounts of protein were boiled for 6 minutes with Laemmli sample buffer, separated by 10% SDSpolyacrylamide gel, transferred to nitrocellulose membranes (Schleicher and Schuell Bioscience) and processed for western blot using an antibody against phosphorylated p38 MAPK. Proteins were detected by using an enhanced chemiluminescence reagent (Amersham Biosciences). After stripping, the membranes were then reprobed with an antibody to total p38 MAPK.

Kinetics of III1c binding to cell layers
Earlier studies have shown that III1c has effects on several cellular parameters including cell motility and growth (Bourdoulous et al., 1998;Mercurius and Morla, 1998;Morla et al., 1994). In vivo experiments have indicated that III1c may also affect tumor growth by inhibiting angiogenesis (Yi and Ruoslahti, 2001); however, the molecular mechanisms underlying these phenomena remain largely unknown. We began studies to evaluate the interaction of III1c with monolayers of human skin fibroblasts. To determine the kinetics of III1c binding to fibroblast cell layers, cells were treated with various recombinant fibronectin Type III modules for increasing amounts of time. Bound fibronectin modules were detected using a mAb directed to the His tag present on the recombinant proteins. Fig. 1A shows the timecourse of binding of various Type III repeats to cell layers over a 2 hour interval. The binding of III1c to cell layers occurred rapidly and reached an apparent steady-state within 1 hour (Fig. 1A). The 13th Type III module (III13), which contains a major heparinbinding site on fibronectin, displayed similar kinetics of binding to cell layers (Fig. 1A). By contrast, both the 10th Type III repeat of fibronectin (III10), containing the RGD integrin Journal of Cell Science 116 (22) Fig. 1. Kinetics and localization of III1c binding to fibroblast monolayers. Human foreskin fibroblasts were grown to ~90% confluence over three days in complete medium. (A) The growth medium was removed and replaced with fresh medium containing 5% fibronectin-depleted serum and 20 µM various His-tagged fibronectin modules for the indicated times: bound modules were detected by an ELISA using an antibody against the His tag. (B) Cell layers were incubated with 20 µM of III1c for the indicated times and bound III1c was detected by indirect immunofluorescence using an anti-His-tag (HIS) antibody. III1c was then visualized using a secondary antibody containing Alexa Fluor-594. Cells were costained for fibronectin (FN) using a polyclonal antibody against the entire molecule and a secondary antibody tagged with Alexa Fluor 488. Overlays of both images are shown in yellow. (C) Increasing concentrations of FN modules were incubated with the cell layers for 2 hours. Cell layers were washed and fixed, and bound protein was measured as described in A. In A and C, each point represents the average of triplicate samples. Error bars indicate standard error of the means. The data represents one of four experiments performed. binding motif, and III11c, a truncated fragment from the 11th Type III repeat with no known biological activity, exhibited little binding to cell layers (Fig. 1A). To localize III1c within the cell layers, cell layers were fixed and stained for both fibronectin and III1c at various times after incubation with III1c. As shown in Fig. 1B, III1c was detected in cell layers within 5 minutes and was colocalized with fibronectin.
We next examined the dose-dependent binding of recombinant proteins to fibroblast cell layers. Cell layers were incubated with increasing concentrations of recombinant fibronectin modules for 2 hours. Wells were then washed and the extent of bound protein was determined. The results are shown in Fig. 1C. Binding of III1c to cell layers occurred over all the concentrations tested (Fig. 1C). Both III1c and III13 exhibited dose-dependent binding to the cell layers (Fig. 1C), suggesting the existence of specific binding sites for III13 and III1c in the cell layer. By contrast, binding of III10 and III11c to cell layers occurred at only the highest concentrations (Fig.  1C).

Localization of III1c in the cell layer
Earlier reports have indicated that the III1 module of fibronectin as well as the III1c peptide exhibit heparin binding activity (Litvinovich et al., 1992;Morla and Ruoslahti, 1992). To evaluate whether the III1c module associated with heparan sulfate containing proteoglycans present in the cell layers, fibroblast monolayers were pretreated with increasing amounts of heparitinase I for 2 hours to remove heparan sulfate from the cell layer. Cell layers were then washed and incubated for an additional 2 hours with either the III1c or III13 peptide in the continued presence of heparitinase I. In agreement with Fig. 1, both fragments were found to bind to untreated cell layers. However, pretreatment of cells with increasing amounts of heparitinase I resulted in a dose-dependent decrease in the binding of III13 to cell layers (Fig. 2), which is consistent with the III13 module being a ligand for heparan sulfate proteoglycans within the cell layer (Bloom et al., 1999;Tumova et al., 2000;Yoneda et al., 1995). By contrast, heparitinase I treatment had no effect on the binding of III1c to the cell layers (Fig. 2). These data indicate that the association of III1c with cell layers is not dependent on binding to heparan sulfate proteoglycans. To assess the extent of loss of heparan sulfate from the matrix, cell layers treated with the indicated amounts of heparitinase were incubated with an antibody directed against heparan sulfate (Fig. 2 insert). These results indicate that maximum loss of heparan occurred at doses of 0.01 units/ml. At this dose, 90% of heparin sulfate was removed from the matrix.
To determine whether heparitinase altered the localization of the III1c and III13 peptides within the monolayer, cells were fixed and double-stained for fibronectin and recombinant fibronectin Type III repeats by indirect immunofluorescence. As shown in Fig. 3, III13 and III1c colocalized with matrix fibronectin, suggesting that the fragments bound to extracellular matrix. Consistent with results obtained in Fig. 2, pretreatment of cells with heparitinase I resulted in a complete loss of staining for III13 but had no effect on the localization of bound III1c. These data indicate that III13 but not III1c binds to heparan sulfate proteoglycans present in the extracellular matrix. As III1c is known to bind to fibronectin (Morla et al., 1994;Morla and Ruoslahti, 1992), these data suggest that the most probable binding partner for III1c in the cell layer is matrix fibronectin.

Effect of III1c on fibronectin matrices
A previous study has suggested that the addition of III1c to cells results in the disassembly and subsequent loss of preestablished fibronectin matrix (Bourdoulous et al., 1998). To monitor changes in the levels of matrix fibronectin in the presence of III1c, fibroblasts were grown in the presence of 125 I-fibronectin to allow for the assembly of a radiolabeled matrix. Cell layers containing radiolabeled fibronectin were incubated with recombinant peptides for 16 hours and then extracted with 0.5% DOC to isolate detergent-insoluble matrix. Matrix fibronectin was obtained from the detergent-insoluble fraction of the cell layer and assessed by autoradiography. As shown in Fig. 4, incubation of cell layers with III1c did not result in any loss of fibronectin compared with control cell layers. In addition, there was no evidence of fragmentation of matrix fibronectin as incubation of cell layers with III1c for 16 hours resulted in no increase in either TCA soluble radioactivity or in fragments in the cell medium detected by autoradiography (data not shown showed that the fibronectin recovered from the DOC insoluble pool was largely in the form of high molecular weight aggregates (Fig. 4). Treatment with reducing agents converted the aggregates to fibronectin monomers, indicating that the aggregates were disulfide stabilized. Taken together, the above results indicate that III1c does not result in either loss or fragmentation of matrix fibronectin.
To determine whether III1c perturbed the overall morphology of the fibronectin matrix, cells were incubated with Alexa Fluor-488 conjugated plasma fibronectin to allow for direct visualization of matrix fibronectin by fluorescent microscopy. Total fibronectin was visualized using a monoclonal antibody FDB3, which recognizes both endogenous and Alexa Fluor-488 conjugated plasma fibronectin. Endogenous cellular fibronectin was also visualized using IST-9, which recognizes a conformationally sensitive epitope found only in cell-derived fibronectin (Liao et al., 1999). After a 20 hour incubation with recombinant fibronectin modules, cells were fixed and stained, and fibronectin matrix was visualized by both direct fluorescence microscopy and indirect immunofluorescence microscopy. As shown in Fig. 5, treatment of cells with III1c resulted in no apparent change in the organization of the Alexa Fluor-488 labeled fibronectin matrix, which remained fibrillar. Similar results were observed when the entire matrix was stained with the fibronectin mAb FDB3 (Fig. 5). However, when the matrix was stained using the fibronectin mAb IST-9 to the EDA domain of fibronectin, III1c treatment resulted in a complete loss of matrix staining, suggesting that the binding of III1c to matrix fibronectin results in the loss of the IST-9 epitope. To determine whether III1c affected other epitopes in matrix fibronectin, several additional monoclonal and polyclonal antibodies were compared for their ability to stain fibronectin fibrils after III1c treatment. These antibodies and their localization within the fibronectin molecule are shown in Fig. 6. Data from the experiments indicated that III1c had no effect on the binding of 11 different antibodies directed at epitopes along the length of the fibronectin molecule (data not shown).
To characterize the effect of increasing doses of III1c on the fibronectin matrix, four anti-fibronectin mAbs to Type III modules were compared for their ability to bind matrix in the presence of III1c. To perform this experiment, antibody titration curves were performed using several different antibodies (data not shown) and four mAbs (IST-9, 5C11F3, IST-5 and clone 568) were chosen for further analysis. These antibodies were chosen because they displayed similar binding affinity for matrix fibronectin Journal of Cell Science 116 (22) Fig. 3. Effect of heparitinase on the localization of III1c in the cell layer. Human fibroblasts were seeded onto glass coverslips and grown as previously described. Wells were then incubated with or without 0.01 U/ml heparitinase I for 2 hours followed by the addition of 20 µM III1c or III13 for 2 additional hours. Cells were then washed, fixed, permeabilized and dual stained for fibronectin (FN) using a polyclonal antibody and the recombinant fragments using a monoclonal antibody against the His-tag (HIS) as indicated. Fields shown represent one experiment repeated three times. (IST-9, IST-5 and clone 568) or because the mAb epitope was located in the same region as IST-9 (5C11F3) (L.V. D.W. et al., unpublished). The epitopes for these mAbs have previously been mapped to the EDA (IST-9 and 5C11F3), III 5 (IST-5) or III8 (Clone 568) modules of fibronectin. On the basis of the data obtained from the binding curves for each of these antibodies (data not shown), mAb concentrations located within the linear range of binding were used in the experiments shown in Fig. 7.
The results shown in Fig. 7 indicate that addition of III1c to cell layers resulted in a dose-dependent decrease in IST-9 mAb binding. Inhibition of IST-9 binding was seen at ~1.0 µM III1c, with a maximal inhibition occurring at 20 µM III1c (Fig. 7A). III13 and III11c had no effect on the binding of IST-9 to matrix fibronectin. III1c was much less effective in preventing the binding of another anti-EDA mAb, 5C11F3, to matrix fibronectin (Fig. 7B), indicating that the decrease in IST-9 binding following III1c treatment was not due to a loss of EDA containing fibronectin in the matrix. The strong inhibitory effect of III1c on IST-9 binding was not seen with other fibronectin mAbs, anti-III8 (Fig. 7C) and anti-III5 (Fig. 7D), where III1c was able to inhibit mAb binding by less than 20%. III13 and III11c, which served as control modules, had no effect on the binding of any of the mAbs to the cell layer.
We also examined the kinetics of III1c induced loss of the IST-9 epitope. Fibroblast cell monolayers were incubated with recombinant fibronectin modules for various amounts of time, washed, fixed and assessed for IST-9 binding by ELISA. Initial loss of IST-9 epitope was seen within 6 minutes, with a maximum effect obtained 1 hour following the addition of III1c to cell layers (Fig. 8). The III11c fragment, which served as a control, had no effect on IST-9 binding. The results from this experiment indicate that treatment of fibroblast cell layers with III1c results in a rapid loss of the IST-9 epitope.
Binding of III 1c to fibronectins Diversity among fibronectin molecules arises, in part, from the alternative splicing of the EDA and EDB Type III repeats. The EDA Type III module is present in fibronectin derived from a variety of cell types, but is notably absent in plasma fibronectin, synthesized by hepatocytes (Umezawa et al., 1985). To test whether the effects of III1c on matrix fibronectin resulted from the preferential binding of III1c to the EDA module, III1c was compared for its ability to bind cellular (EDA + ) vs plasma (EDA -) fibronectin. Our results show that III1c bound equally well to both forms of fibronectin in a dosedependent manner (Fig. 9), suggesting that the III1c fragment does not bind preferentially to the EDA region of fibronectin. However, binding of III1c to purified cellular (EDA + ) fibronectin resulted in a dose-dependent decrease in the binding of IST-9 to fibronectin (Fig. 10). The binding of another monoclonal antibody 5C11F3, which also recognizes the EDA module, was not affected by III1c (Fig. 10). Binding of mAbs to other regions of fibronectin were also unaffected by III1c (data not shown). These results suggest that the loss of IST-9 binding results from the direct binding of III1c to fibronectin and does not require the presence of cells. These Fibroblasts were cultured on glass coverslips in the presence of fibronectin conjugated with Alexa Fluor-488 (10 µg/ml), indicated as AF-FN, for three days. Wells were washed and incubated with 20 µM III1c or carrier buffer as control for 20 hours. Cells were washed, fixed, permeabilized and stained for fibronectin using the designated mAbs. Alexa Fluor-488 labeled fibronectin was visualized and total matrix fibronectin was detected by indirect immunofluorescence using either an antibody against the CS domain (FDB3) or an antibody against the EDA domain (IST-9). The figure represents one experiment repeated three times.  Fig. 6. Map of antibody binding sites on fibronectin.
Incubation of cell layers with 20 µM of III1c resulted in the loss of IST-9 staining for fibronectin, but had no effect on the ability of any of the remaining ten antibodies to stain matrix fibronectin.
results also indicate that loss of IST-9 epitope does not result from the binding of the III1c to the EDA module.
As the IST-9 epitope has been reported to be conformationally sensitive, the data are most consistent with a model in which loss of IST-9 epitope results from the binding of III1c to fibronectin causing a change in EDA conformation.
Effect of III1c on p38 activation Previous reports have indicated that III1c treatment of cell layers results in the activation of Cdc42 and p38 MAPK, resulting in changes in cell cycle and cytoskeletal organization (Bourdoulous et al., 1998;Mercurius and Morla, 1998). These effects were reported to occur over several hours as a result of III1cmediated loss of the established fibronectin matrix. Similar to the findings of Bourdoulous et al. (Bourdoulous et al., 1998), we also found that III1c caused an increase in filopodia, but this effect on the cytoskeleton was not seen until 12-14 hours of III1c treatment (data not shown). By contrast, III1c effects on p38 activation occurred within minutes and closely paralleled the loss of IST-9 epitope. Western blot analysis indicated that the phosphorylation of p38 in response to III1c treatment occurred within 30 minutes (Fig. 11). III1c-induced activation of p38 peaked at 1 hour and remained elevated for up to 2 hours. Addition of the III13 (Fig. 11) or III11c (not shown) module had no effect on p38 phosphorylation levels, suggesting that the effects of p38 were specific to the III1c peptide. The kinetics of p38 activation were slower but more sustained than those seen using agonists whose receptors are known to activate p38. As shown in Fig.  11, activation of p38 by EGF peaked by 10 minutes and returned to baseline within 20 minutes. Similar kinetics of activation of p38 were seen using fibroblast growth factor (FGF), tumor-necrosis factor (TNF), plateletderived growth factor (PDGF), and interleukin-1 (IL-1) (data not shown), indicating that activation of p38 by growth factors or cytokines was significantly faster and less sustained than the III1c mediated effects on p38. These results are consistent with a model whereby III1c works indirectly through changes in matrix conformation to activate p38.

Discussion
In the present study, we show that addition of the III1c peptide, a fragment derived from the first Type III repeat in fibronectin, to fibroblast monolayers results in a rapid conformational remodeling of the fibronectin matrix which is accompanied by an activation of p38 MAPK. Remodeling of the fibronectin matrix in response to III1c occurs rapidly, does not require the presence of cells and is not accompanied by changes in the levels of fibronectin matrix. Previous studies have indicated that the III1 region of fibronectin participates in homophilic binding interactions among fibronectin molecules, which can regulate the assembly of the fibronectin matrix (Chernousov et al., 1987;Chernousov et al., 1991;Hocking et al., 1994;Hocking et al., 1996;Morla et al., 1994;Morla and Ruoslahti, 1992). Fibroblast monolayers, grown as described previously, were incubated with 20 µM III1c or III11c for increasing times. The wells treated with carrier buffer were set as the 100% control. After incubation with fibronectin fragments, cell layers were washed, fixed and processed for ELISA using the monoclonal antibody IST-9 (2 µg/ml). Data points are the mean of triplicate wells. The graph depicts data from a representative experiment performed in triplicate. polymerization of fibronectin matrix by cultured cells have been conflicting. Addition of low amounts of III1c to cells has been shown to enhance fibronectin deposition into matrix, whereas larger amounts of the III1c peptide have been reported to result in the disassembly of preformed fibronectin matrix (Bourdoulous et al., 1998;Morla et al., 1994). Changes in the level of fibronectin matrix have been credited with mediating the biological effects of III1c on cell behavior, including cell growth and motility (Bourdoulous et al., 1998;Mercurius and Morla, 1998;Morla et al., 1994). The molecular basis for the biological effects of III1c is not well understood. Various binding partners for the III1c fragment of fibronectin that have been reported previously include fibronectin itself (Morla et al., 1994;Morla and Ruoslahti, 1992), heparan sulfate proteoglycans and the α5β1 integrin (Mercurius and Morla, 2001). In our experiments, III1c binding was insensitive to heparitinase and colocalized with fibronectin, suggesting that in fibroblast monolayers, the primary binding site for III1c is matrix fibronectin. Potential binding sites for III1c within the fibronectin molecule have been identified. Solid phase binding assays as well as affinity chromatography have shown that the III1 module of fibronectin can bind to the amino terminal Type I repeats as well as several of the Type III modules (Aguirre et al., 1994;Hocking et al., 1994). Binding of III1 to III10, III7, and III15 was enhanced following denaturation, suggesting the existence of multiple cryptic binding sites for III1 within the fibronectin molecule (Hocking et al., 1996;Ingham et al., 1997).
Using several monoclonal antibodies to various regions along the length of the fibronectin molecule, we were able to show that IST-9, an antibody recognizing a conformationally sensitive epitope within the EDA region of matrix fibronectin (Liao et al., 1999), was no longer able to recognize the fibronectin matrix within 1 hour of treatment with III1c. Loss of the IST-9 epitope occurred within minutes and was complete within 1 hour, exactly paralleling the kinetics of III1c binding to cell layers. Our results also indicate that the effect of III1c on IST-9 binding to fibronectin occurred in the absence of cells, suggesting that III1c binds directly to fibronectin and alters its conformation. As we were unable to detect any preferential binding of III1c to either EDA containing fibronectin or to recombinant EDA modules (R.M.K., L.V.D.W. and P.J.M.-L., unpublished), it appears unlikely that the loss of IST-9 epitope resulted from steric hindrance due to III1c binding to EDA. This conclusion is supported by the observation that binding of another monoclonal antibody 5C11F3, which is similar to IST-9, recognizes epitopes within the C' loop of the EDA molecule (L.V.D.W., unpublished) is unaffected by III1c treatment.
Consistent with previous studies (Bourdoulous et al., 1998), binding of III1c to cell layers was associated with activation of p38 and with the reorganization of actin into filopodia. We found that the kinetics of these biological effects were quite distinct. The kinetics of p38 activation clearly paralleled the loss  Fig. 9. Binding of III1c to EDA + and EDAfibronectins. Triplicate wells were coated with 20 µg/ml of either (EDA + ) cellular fibronectin or (EDA -) plasma fibronectin, blocked with 1.0% BSA and incubated with increasing amounts of III1c for 1 hour. Bound III1c was measured by ELISA using a monoclonal antibody to the 6xHis tag. Fig. 10. Effect of III1c on IST-9 binding to EDA + fibronectin. Wells were coated with 20 µg/ml of cellular fibronectin (EDA + ) and then blocked with 1.0% BSA. Triplicate wells were incubated with increasing amounts of III1c for 1 hour. Following treatments, the binding of EDA specific mAbs, IST-9 (2 µg/ml), or 5C11F3 (8 µg/ml) was determined by ELISA. The graph depicts data from a representative experiment performed in triplicate.  Fig. 11. Effect of III1c on p38 MAPK activation. Fibroblast monolayers, grown 3 days in complete medium, were washed and rendered quiescent by incubation for 36 hours in DMEM containing 0.1% BSA. Monolayers were then incubated with III1c (20 µM), III13 (20 µM) or EGF (25 ng/ml) in 0.1% BSA/DMEM for the indicated times. Cell lysates were electrophoresed and immunoblotted using an antibody against phosphorylated p38 (p-p38). The membranes were then stripped and reprobed with an antibody against p38 to ensure equal loading (p38). Blots shown are from one experiment performed three times. of IST-9 epitope. III1c-induced loss of IST-9 epitope was nearly complete by 30 minutes, whereas p38 activation was first seen at 30 minutes and complete by 60 minutes. This suggests that the addition of III1c to cells causes a rapid conformational change within the fibronectin matrix that contributes to the activation of p38. We also found that III1c caused an increase in filopodia formation as reported previously (Bourdoulous et al., 1998). However, unlike p38, the effect of III1c on actin organization occurred more slowly and was not seen until 12-14 hours after the addition of III1c to the cells. These findings suggest that treatment of cells with III1c can have both shortand long-term effects on cell physiology.
Previous studies have suggested that the effects of III1c on cell behavior are caused by an III1c-mediated disassembly of the preformed fibronectin matrix (Bourdoulous et al., 1998). Although our studies were able to reproduce the previously reported biological effects of III1c (i.e. p38 activation and cytoskeletal reorganization), we were unable to show any loss of fibronectin from the matrix of III1c-treated cells. Treatment of cells with as much as 40 µM III1c for as long as 24 hours did not result in any loss of fibronectin matrix. As discussed below, the discrepancy between our results and those of Boudoulous et al. (Boudoulous et al., 1998) most probably reflects the choice of antibody used to stain the matrix in the earlier study. Immunofluorescent staining of matrix fibronectin indicated that III1c had no obvious effects on the amount or overall organization of matrix. However, the screening of several monoclonal and polyclonal antibodies to fibronectin revealed that one antibody IST-9, which is known to recognize a conformationally sensitive region within the EDA domain of fibronectin, was unable to bind fibronectin in the presence of III1c. Binding of other antibodies including a second antibody to the EDA Type III module, which is not conformationally sensitive, were only marginally affected by III1c. These data suggest that the effects of III1c on matrix fibronectin are more subtle than previously thought and that small changes within subdomains of Type III modules of fibronectin may have profound effects on cell behavior. How the loss of the IST-9 epitope might affect the activation of p38 is not clear. Recently, the EDA has been shown to contain binding sites for both the α9β1 and the α4β1 integrin receptors (Liao et al., 2002). As human fibroblasts are known to have α4β1 receptors, it is possible that the effect of III1c on cell signaling pathways may result from changes in the association of α4β1 with the EDA module present in the matrix. Alternatively, III1c-induced conformational changes in the EDA region may affect the α5β1 integrin association with the III9-III10 region in matrix fibronectin as the presence of EDA within the context of intact fibronectin has been shown to effect exposure of the integrin binding site (Johnson et al., 1999).
Taken together, these data indicate that the binding of III1c to fibronectin induces a change in the conformation of the cellular (EDA containing) fibronectin present in the extracellular matrix. Synthesis of EDA fibronectin in adult tissues occurs in response to injury or under a variety of pathological conditions. The EDA form of fibronectin is synthesized during wound healing and is required for the TGFβ conversion of fibroblasts to myofibroblasts which occurs during wound closure and as part of the tumor/stromal reaction (Ronnov-Jessen and Petersen, 1993;Serini et al., 1998). Therefore, in vivo, tissues undergoing active remodeling would be expected to be targets of III1c. In addition to regulating cell growth and motility, the III1c may be an important modulator of fibroblast differentiation and might be a useful reagent to regulate scarring and/or tumor stromal reactions.