Heparin II Domain of Fibronectin Uses α4β1 Integrin to Control Focal Adhesion and Stress Fiber Formation, Independent of Syndecan-4*

Co-signaling events between integrins and cell surface proteoglycans play a critical role in the organization of the cytoskeleton and adhesion forces of cells. These processes, which appear to be responsible for maintaining intraocular pressure in the human eye, involve a novel cooperative co-signaling pathway between α5β1 and α4β1 integrins and are independent of heparan sulfate proteoglycans. Human trabecular meshwork cells isolated from the eye were plated on type III 7–10 repeats of fibronectin (α5β1 ligand) in the absence or presence of the heparin (Hep) II domain of fibronectin. In the absence of the Hep II domain, cells had a bipolar morphology with few focal adhesions and stress fibers. The addition of the Hep II domain increased cell spreading and the numbers of focal adhesions and stress fibers. Cell spreading and stress fiber formation were not mediated by heparan sulfate proteoglycans because treatment with chlorate, heparinase, or soluble heparin did not prevent Hep II domain-mediated cell spreading. Cell spreading and stress fiber formation were mediated by α4β1 integrin because soluble anti-α4 integrin antibodies inhibited Hep II domain-mediated cell spreading and soluble vascular cell adhesion molecule-1 (α4β1 ligand)-induced cell spreading. This is the first demonstration of the Hep II domain mediating cell spreading and stress fiber formation through α4β1 integrin. This novel pathway demonstrates a cooperative, rather than antagonistic, role between α5β1 and α4β1 integrins and suggests that interactions between the Hep II domain and α4β1 integrin could modulate the strength of cytoskeleton-mediated processes in the trabecular meshwork of the human eye.


Co-signaling events between integrins and cell surface proteoglycans play a critical role in the organization of the cytoskeleton and adhesion forces of cells.
These processes, which appear to be responsible for maintaining intraocular pressure in the human eye, involve a novel cooperative co-signaling pathway between ␣5␤1 and ␣4␤1 integrins and are independent of heparan sulfate proteoglycans. Human trabecular meshwork cells isolated from the eye were plated on type III 7-10 repeats of fibronectin (␣5␤1 ligand) in the absence or presence of the heparin (Hep) II domain of fibronectin.

In the absence of the Hep II domain, cells had a bipolar morphology with few focal adhesions and stress fibers. The addition of the Hep II domain increased cell spreading and the numbers of focal adhesions and stress fibers.
Cell spreading and stress fiber formation were not mediated by heparan sulfate proteoglycans because treatment with chlorate, heparinase, or soluble heparin did not prevent Hep II domain-mediated cell spreading. Cell spreading and stress fiber formation were mediated by ␣4␤1 integrin because soluble anti-␣4 integrin antibodies inhibited Hep II domain-mediated cell spreading and soluble vascular cell adhesion molecule-1 (␣4␤1 ligand)induced cell spreading. This is the first demonstration of the Hep II domain mediating cell spreading and stress fiber formation through ␣4␤1 integrin. This novel pathway demonstrates a cooperative, rather than antagonistic, role between ␣5␤1 and ␣4␤1 integrins and suggests that interactions between the Hep II domain and ␣4␤1 integrin could modulate the strength of cytoskeletonmediated processes in the trabecular meshwork of the human eye.
Adhesive interactions between cells and extracellular matrix molecules form highly specific yet very diverse signaling conduits that regulate a wide variety of biological processes asso-ciated with cell morphology, proliferation, differentiation, migration, and survival (1)(2)(3)(4)(5). These adhesive events are mediated mainly by the integrin family of receptors and lead to the formation of dynamic multi-molecular structures called focal adhesions, focal complexes, and fibrillar adhesions (6,7).
The formation of focal contacts often depends on co-signaling events between integrins and cell surface proteoglycans. In fibroblasts and A375-SM melanoma cells plated on fibronectin, cooperative signaling between ␣5␤1 integrin and syndecan-4, a cell surface heparan sulfate proteoglycan (HSPG), 1 is needed to promote the formation of focal adhesions and stress fibers (8,9). If fibronectin-null fibroblasts are plated on anti-␤1 integrin antibodies or the RGD cell binding domain of fibronectin, cells bind but do not assemble the actin cytoskeleton. However, if soluble antibody directed against the ectodomain of syndecan-4 is added, the cells will assemble focal adhesions and stress fibers (10). A similar scenario has been documented for ␣4␤1 integrin and cell surface chondroitin sulfate proteoglycans, supporting the idea that co-engagement of integrins and proteogylcans results in cooperative signaling (11,12). These cosignaling events are often mediated by two different but adjacent sites within fibronectin. For instance, fibroblasts plated on the RGD cell binding domain of fibronectin can adhere, but they require additional signals from the heparin (Hep) II domain of fibronectin to form focal adhesions and stress fibers (13)(14)(15).
The ability of co-signaling events to differentially mediate these processes lies in the fact that different integrins utilize different signaling mechanisms to trigger focal adhesion for-mation. For instance, signaling events mediated by ␣5␤1 integrin require a HSPG co-receptor such as syndecan and involve the activation of protein kinase C␣. In contrast, ␣4␤1 integrinmediated focal adhesion formation is independent of syndecans and protein kinase C␣ activation (8). The cytoplasmic domain of ␣4 integrin also interacts directly with the signaling adaptor protein paxillin, in contrast to ␣5␤1 integrin, which does not directly interact with paxillin (20).
In this study, we investigated the need for co-signaling between integrins and syndecans in focal adhesion and stress fiber formation using primary diploid human trabecular meshwork (HTM) cells. HTM cells are a unique cell type found in the anterior chamber of the human eye. The cytoskeletal organization and adhesive forces of these cells play a key role in maintaining intraocular pressure. Chemical agents that disrupt the cytoskeleton or the signaling pathways that maintain the actomyosin network, such as H-7, cytochalasins, and latrunculins, cause a decrease in intraocular pressure (21)(22)(23)(24)(25). Therefore, understanding the signaling pathways that regulate these processes is critical for understanding the mechanisms by which intraocular pressure is maintained. Recently, the Hep II domain of fibronectin was shown to lower intraocular pressure in a human eye organ culture system, suggesting that cell-matrix signaling events mediated by the Hep II domain may be involved in controlling intraocular pressure (26). Using a recombinant Hep II domain and the type III 7-10 repeats of fibronectin, the present studies show that limited focal adhesion and stress fiber formation can be mediated by ␣5␤1 integrin independently of a syndecan co-receptor and that ␣4␤1 integrin signaling mediated by the Hep II domain can augment focal adhesion and stress fiber formation in HTM cells. This is the first report that ␣4␤1 integrin can act as a co-receptor for ␣5␤1 integrin and that the ␣5␤1 and ␣4␤1 signaling pathways converge to enhance focal adhesion and stress fiber formation. This is also the first time that the specific activation of ␣4␤1 integrin by the Hep II domain of fibronectin has been demonstrated. This dual signaling through ␣5␤1 and ␣4␤1 integrins may serve to control the adhesive strength and contractility of HTM cells and may provide a mechanism by which these cells can regulate intraocular pressure.
Preparation of Recombinant Proteins-Recombinant HepII domain (type III 12-14 repeats of fibronectin) and the type III 7-10 repeats of fibronectin ( Fig. 1) were made as described previously (26,29). Recombinant type III 4 -5 repeats of fibronectin (Hep III domain) was made as described previously (12). The mutated Hep II domain (Hep II/RK) was prepared using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions by mutating the residues Arg 9 and Lys 25 in the major heparin binding site in the type III 13 repeat of the Hep II domain (15,30,31) to serines using the following oligonucleotides: Arg 9 3 Ser, 5Ј-CCACCAAGAAGG-GCTTCTGTGACAGATGCTACTG-3Ј (forward) and 5Ј-CAGTAGCATC-TGTCACAGAAGCCCTTCTTGGTGG-3Ј (reverse); and Lys 25 3 Ser, 5Ј-CATTAGCTGGAGAACCTCGACTGAGACGATCACTG-3Ј (forward) and 5Ј-CAGTGATCGTCTCAGTCGAGGTTCTCCAGCTAATG-3Ј (reverse). The changed nucleotides are underlined. Hep II/RK, which was in the bacterial expression vector pGEX 4T1, was then expressed as a glutathione S-transferase fusion protein and purified as described previously (26).
Spreading Assay-HTM cells serum-starved for 24 h were harvested with 0.05% trypsin and 0.25 mM EDTA and resuspended in low-glucose Dulbecco's modified Eagle's medium containing 2 mM L-glutamine, 2.5 g/ml amphoteracin B, 25 g/ml gentamicin, and 25 g/ml cyclohexi-mide. Cells were then plated (3-5 ϫ 10 4 cells/well) in the presence of cycloheximide onto 12-mm round glass coverslips (Bellco Glass, Inc., Vineland, NJ). Coverslips were pre-coated for 1 h at 37°C with 47 nM (1.7 g/ml) type III 7-10 repeats of fibronectin diluted in PBS (137 mM NaCl, 2.7 mM KCl, 8.1 mM Na 2 HPO 4 , and 1.5 mM KH 2 PO 4 ). The Hep II domain was added to cells just before plating at 472 nM (12.75 g/ml), and cells were incubated for 3 h at 37°C. In some cases, coverslips were co-coated with molar equivalents (47 or 472 nM) of the Hep II domain or the type III 4 -5 repeats of fibronectin and 47 nM of the type III 7-10 repeats of fibronectin. At times, the Rho kinase inhibitor Y-27632 (Tocris Cookson, Inc., Ellisville, MO) or the serine/threonine kinase inhibitor H-7 (Sigma-Aldrich) was added 1 h after plating the cells.
To determine the involvement of HSPGs in Hep II domain-mediated cell spreading, cells were pretreated with 30 mM sodium chlorate for 24 h in sulfate-free media and 15% sulfate-free serum, followed by an additional 24 h in the same medium without serum as described previously (32,33). As a control, cells were incubated with 10 mM sodium sulfate or 30 mM sodium chlorate and 10 mM sodium sulfate. In other experiments, soluble heparin (50 g/ml) was added as an inhibitor at the time of plating, or cells were treated with heparitinase and heparinase (0.0024 IU/ml; ICN Biochemicals, Inc., Irvine, CA). In these latter experiments, cells were incubated with these enzymes for 4 h before plating, with fresh enzyme added after 2 h and throughout the spreading assay (34).
To determine whether ␣4␤1 integrin was involved in cell spreading, an Immunofluorescence Microscopy-HTM cells were washed with 50 mM MES at pH 6.0, permeabilized for 2 min with 0.5% Triton X-100 in 50 mM MES, and fixed for 30 min with 4% paraformaldehyde in PBS, pH 7.4. Cells were blocked for 1 h with 1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) (BSA/PBS). Blocked cells were incubated with anti-vinculin antibodies (Sigma-Aldrich) diluted 1:3000 in 0.1% BSA/PBS for 1 h and then incubated simultaneously with Alexa 546-conjugated goat anti-mouse secondary antibody (Ab) (4 g/ml; Molecular Probes, Eugene, OR) and Alexa 488-conjugated phalloidin (0.67 unit/ml; Molecular Probes) in 0.1% BSA/PBS for 1 h. Coverslips were mounted onto slides using Immu-mount (Shandon Lipshaw, Pittsburgh, PA). To visualize cell surface heparan sulfate, cells were washed with PBS, fixed, blocked as described above, and incubated with mouse (IgM) antibody 10E4, which detects an epitope present in most heparan sulfates (Seikagaku America, Inc., East Falmouth, MA) (35). Cultures were then labeled with a rabbit anti-mouse IgM (Zymed Laboratories, Inc., San Francisco, CA) followed by an Alexa 546-conjugated goat anti-rabbit secondary Ab. All labeling was done for 1 h in 0.1% BSA/ PBS, and coverslips were mounted as described above. Cell images were acquired using a Zeiss AxioCam HRm camera (Thornwood, NY) mounted on a Zeiss Axiophan 2 Imaging fluorescence microscope together with AxioVision version 3.1 software. In some experiments, the extent of cell spreading was determined by measuring cell width and length on the computer screen using the AxioVision software. Measurements of cells were made from 8 -12 different fields of view per coverslip (n ϭ 40). Only cells with clear borders were measured. The ratio of cell width to length was compared with control cell ratios from four different experiments (n ϭ 160; Fig. 3) or two different experiments (n ϭ 80; Fig.  8). In some cases, the number of cells positive for stress fibers was also counted. Cell counts were made from 8 -10 different fields of view per coverslip (n ϭ 50). The number of stress fiber-positive cells was compared with control cells from two different experiments for analysis (n ϭ 100).
F-actin Detection-HTM cells (3 ϫ 10 5 cells/well) were plated onto 22-mm square glass coverslips pre-coated with the type III 7-10 repeats (47 nM) in the presence or absence of the Hep II domain (472 nM). After 3 h, the coverslips were transferred to new wells, washed with PBS, and fixed for 10 min with 4% paraformaldehyde/PBS followed by permeabilization for 10 min with 0.5% Triton X-100/PBS. Cells were incubated for 30 min at 37°C with Alexa 488-conjugated phalloidin (0.67 unit/ml) in 0.1% BSA/PBS. Cells were washed, and bound phalloidin was released by incubating the cells for 2 h at room temperature in humidified chambers with 2 ml of methanol under mild agitation (36). The entire sample from each well was collected, and phalloidin levels were detected using the FluoroMax-3 spectrofluorometer (Jobin Yvon Horiba, Edison, NJ) and Data Max v.2.2 software with excitation/emission wavelengths of 495/519 nm.
Immunoblotting-HTM cells were plated onto 10-cm tissue culture plates in the presence or absence of the Hep II domain (472 nM). All plates were pre-coated with 236 nM of the type III 7-10 repeats (8.5 g/ml) and blocked with 1% heat-denatured BSA/PBS (85°C for 10 min) for 1 h at room temperature. After 0.5, 1, 2, or 3 h, cells were washed with PBS and lysed for 10 min at 4°C with 15 mM CHAPS in 10 mM HEPES, pH 7.4, containing 150 mM NaCl, 1 mM EDTA, 1 mM sodium orthovanadate, 2.5 mM sodium pyrophosphate, 10 mM NaF, and 10 g/ml of pepstatin, leupeptin, and aprotinin. Lysate (10 g) from the control and treated cells was separated on an 8% SDS-PAGE and transferred to Immobilon-P (Millipore, Billerica, MA). The membrane was blocked in 3% BSA/TBS and incubated with either focal adhesion kinase (FAK) polyclonal antibody (pAb; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or phosphospecific FAK pY397 pAb (Upstate Group, Inc.) in 1% BSA/TBS/0.1% Triton X-100 for 1 h. Membranes were then washed with TBS/0.1% Triton X-100 and incubated for 1 h with horseradish peroxidase-conjugated goat anti-rabbit Ab (Santa Cruz Biotechnology, Inc.). Bound antibody was detected with the ECL Plus Western blotting detection kit (Amersham Biosciences). The area of the bands from three different experiments was measured using Scion Image software (Scion Corp., Frederick, MD) and averaged together to determine fold induction in FAK pY397 phosphorylation.
Cell Adhesion Assay-Serum-starved HTM cells were plated in the presence of 10 g/ml adhesion blocking ␤1, ␤3, ␣5, and ␣v integrin Abs or control mouse IgG (Sigma) into 96-well plates. Wells had been pre-coated with 47 nM of the type III 7-10 repeats for 1 h at 37°C, blocked with 2% fatty acid-free BSA/PBS for 1 h at room temperature, and washed before plating the cells. Unbound cells were removed by washing with PBS. Bound cells were fixed for 20 min with 4% paraformaldehyde/PBS and stained overnight with 0.5% toluidine blue in 4% paraformaldehyde/PBS. Bound dye was redissolved in 2% SDS and detected at 600 nm using a microplate reader as described previously (37). Rat ␤1 integrin Ab m13 and rat ␣5 integrin Ab m16 were both kindly provided by Dr. Steve Akiyama (National Institutes of Health, Research Triangle Park, NC). The ␤3 integrin monoclonal antibody (mAb) 1 and the ␣v integrin mAb M9 were purchased from BD Biosciences and Chemicon International, Inc. (Temecula, CA), respectively.
Data Analysis-All comparisons were made as a percentage of the control. Data are presented as mean Ϯ S.E. Statistical comparisons were done using Student's t test.

The Hep II Domain Induces Stress Fiber Formation and Cell
Spreading-To determine whether focal adhesion and stress fiber formation in HTM cells involved co-signaling, HTM cells were plated on the type III 7-10 repeats of fibronectin containing the RGD integrin binding site in the absence or presence of soluble Hep II domain (Fig. 1) for 3 h. As shown in Fig. 2, AϪC, when HTM cells were plated on the type III 7-10 repeats in the absence of the Hep II domain, cells were able to attach and spread. Morphologically, however, the cells exhibited a bipolar morphology with few stress fibers and focal adhesions, indicating that they were not completely spread. If soluble Hep II domain was added to the media, cell spreading was enhanced, as was stress fiber formation and the number and intensity of focal adhesions (Fig. 2, EϪG). Comparable results are seen when cells were plated on intact fibronectin or when soluble fibronectin was added to cells plated on the type III 7-10 repeats (data not shown). As shown in Fig. 3A, the Hep II domain significantly increased the number of stress fiber-positive cells by 56% over control cells. Measurements of cell width and length verified the immunofluorescent images and showed that the Hep II domain significantly enhanced cell spreading by 76% compared with control cells (Fig. 3B). Concomitantly with the increase in spreading, measurements of F-actin levels  indicated that polymerized stress fiber levels increased by 51% over controls in the presence of soluble Hep II domain (Fig. 3C). In addition, within 0.5 h, the soluble Hep II domain induced a 2.15-fold increase (p Ͻ 0.02) in the phosphorylation of FAK at the Tyr 397 site compared with cells plated on the type III 7-10 repeats alone, demonstrating that the Hep II domain also enhanced focal adhesion formation (Fig. 3D). FAK pY397 phosphorylation remained elevated (although not significantly) for the duration of the experiment (3 h) compared with untreated cells. At 3 h, a similar Hep II domain-mediated increase in FAK phosphorylation was seen when HTM cells were plated on intact fibronectin (data not shown).
Stress fiber formation in either the absence or presence of the Hep II domain was dependent on the activation of Rho kinase. Treatment of HTM cells plated on the type III 7-10 repeats with the Rho kinase inhibitor Y-27632 further decreased the already low numbers of focal adhesions and stress fibers in control cells and induced the formation of broad lamellipodia (Fig. 2D). In Hep II domain treated cells, Y-27632 dramatically decreased focal adhesions and stress fibers (Fig.  2H). These cells were more elongated and formed lamellipodia, such that they closely resembled Y-27632-treated control cells plated on the type III 7-10 repeats. Unexpectedly, H-7, a known inhibitor of serine/threonine kinase that was previously shown to block signaling via the Hep II domain (13,40), did not block stress fiber formation. Furthermore, the protein kinase C inhibitors calphostin C and bisindolylmaleimide I were equally ineffective (data not shown). Collectively, the data showed that HTM cells plated on the type III 7-10 repeats display limited focal adhesion and stress fiber formation in the absence of any co-signaling from the Hep II domain and that the Hep II domain significantly increases cell spreading, focal adhesion formation, and stress fiber formation in a Rho kinase-dependent, protein kinase C-independent pathway.
HSPGs Are Not Involved in Cell Spreading-Previous studies have indicated that the Hep II domain, which contains several HSPG binding sites (14,30,31), uses syndecan-4 to mediate complete spreading of fibroblasts on the type III 7-10 repeats of fibronectin (10,(13)(14)(15). To determine whether HSPGs are involved in Hep II domain-mediated spreading of HTM cells, FACS analysis was first performed using antibodies against the ectodomains of all four syndecans to determine the syndecan expression profile of HTM cells. As shown in Fig. 4, HTM cells only express syndecan-1 at the cell surface. Syndecans-2, -3, and -4 were not found at the cell surface. Expression of syndecan-4 was further examined using immunofluorescence microscopy experiments. This study showed that a few HTM cells express syndecan-4 at the cell surface, and the level of expression was very low. In most of the cells, syndecan-4 was found intracellularly, mainly in the Golgi and endoplasmic reticulum (data not shown). The significance of this is unknown, but such low levels of cell surface syndecan-4 suggest that it is unavailable as a co-receptor in HTM cells. These results suggest that syndecans are not involved in the formation of focal adhesions in HTM cells because syndecan-1 has yet to be implicated in focal adhesion and stress fiber formation.
To rule out the possibility that the Hep II domain was interacting with syndecan-1 or -4 to control focal adhesion and stress fiber formation, cells were pre-treated with heparitinase and heparinase for 4 h before plating and throughout the cell spreading assay. Previous studies have indicated that removal of heparin sulfate moieties from the core protein of syndecan will abolish interactions between the Hep II domain and syndecans. As shown in Fig. 5, C and D, heparitinase and heparinase treatment had no inhibitory effect on Hep II domainmediated cell spreading of HTM cells plated on the type III 7-10 repeats. To verify that heparitinase and heparinase treatment successfully removed heparan sulfate moieties, cells were labeled with the monoclonal 10E4 antibody that detects an epitope present in most heparan sulfates. Enzymatic treatment abolished 10E4 antibody labeling (Fig. 5, E and F), indicating the cleavage of heparan sulfate. Enzymatic treatment alone had no effect on cell morphology (Fig. 5B). In addition, soluble heparin had no inhibitory effect on Hep II domain-mediated spreading, nor did chlorate treatment, which blocks sulfation of heparan sulfate glycosaminoglycan chains (41) (data not shown). As a final approach, a Hep II/RK domain that had an arginine and lysine residue mutated to serine in the heparin binding cationic cradle in the type III 13 repeat (30) was tested for its ability to mediate cell spreading. This Hep II/RK domain, which was no longer able to bind heparin (data not shown), mediated cell spreading to the same extent as intact Hep II domain (Fig. 7E). Taken together, the data support the idea that HSPGs are not necessary for Hep II domain-mediated cell spreading of HTM cells plated on the type III 7-10 repeats.
Hep II Domain Mediates Cell Spreading via ␣4␤1 Integrin-In addition to having HSPG binding sites, the Hep II domain also has a potential ␣4␤1 integrin binding site, IDAPS, in the type III 14 repeat (42,43). As shown in Fig. 6, FACS analysis indicated that HTM cells express ␣4␤1 integrin on their cell surface along with ␣2, ␣3, ␣5, ␤1, ␣v␤3, ␣v␤5, and ␣5␤1 integrins. To determine whether ␣4␤1 integrin could mediate cell spreading on the type III 7-10 repeats, HTM cells were incubated with soluble extracellular domain of VCAM-1, which is a known ␣4␤1 integrin ligand (44). Similar to the Hep II and Hep II/RK domains (Fig. 7, C and E), soluble VCAM-1 induced spreading and increased stress fiber formation in cells plated on the type III 7-10 repeats of fibronectin (Fig. 7G), indicating that HTM cell spreading on type III 7-10 repeats is ␣4␤1 integrin-dependent. To determine whether all ␣4␤1 integrin binding sites found in fibronectin induce cell spreading, cells plated on the type III 7-10 repeats were also incubated with soluble molar equivalents (472 nM) of the type III 4 -5 repeats (12,45) and the IIICS domain (46,47). These studies showed that whereas the IIICS domain was equally effective as the Hep II domain in promoting cell spreading, the type III 4 -5 repeats were less effective than the Hep II domain in promoting cell spreading (data not shown).
To determine whether the Hep II domain was using ␣4␤1 integrin to mediate cell spreading, soluble ␣4 integrin blocking antibodies were added at the time of cell plating in the absence or presence of the Hep II domain, the Hep II/RK domain, or the soluble VCAM-1 extracellular domain. As shown in Fig. 7, the anti-␣4 integrin antibodies blocked cell spreading in the presence of these ligands. These cells had fewer stress fibers and were more elongated than cells treated with soluble peptide alone (Fig. 7, D, F, and H). Soluble ␣4 integrin blocking antibody alone (Fig. 7B) or control non-immune IgG had no effect on cell spreading (data not shown), suggesting that cell spreading mediated by the Hep II domain involved ␣4␤1 integrin. Comparison of the cell width versus length ratios verified these images and showed that the Hep II domain, Hep II/RK domain, and VCAM-1 increased cell spreading by 80%, 86%, and 77%, respectively, over control cells (Fig. 8), and the addition of soluble ␣4 integrin blocking antibody significantly reduced this increased cell spreading to 14%, 5%, and 2% of control cell spreading, respectively.
Interestingly, all the ␣4␤1 ligands except for VCAM-1 had to be presented as a soluble ligand in order for cell spreading to be induced. Co-coating coverslips with the type III 7-10 repeats and the Hep II domain or the III4 -5 repeats did not induce cell spreading above that observed with coverslips coated with the III 7-10 repeats alone (data not shown). Co-coating experiments were not done with the IIICS domain.
To determine which integrin(s) the HTM cells were using to attach to the type III 7-10 repeats, cell adhesion assays were performed in the presence of integrin blocking antibodies. As shown in Fig. 9, the HTM cells adhered to the type III 7-10 repeats via ␣5␤1 integrin because cell adhesion was inhibited with ␣5 and ␤1 integrin blocking antibodies by 85% and 92%, respectively (p Ͻ 0.001), whereas ␣v and ␤3 integrin blocking antibodies had no inhibitory effect on cell adhesion. Thus, the partial cell spreading observed on the type III 7-10 repeats was due to interactions with ␣5␤1 integrin. Cell spreading was never observed in HTM cells plated on the Hep II domain (data not shown). Thus, cell spreading in the presence of the Hep II domain was the result of ␣5␤1/␣4␤1 co-signaling. DISCUSSION In this study, we showed that cooperative signaling between ␣5␤1 and ␣4␤1 integrins regulated cell spreading in trabecular meshwork cells and that neither ␣5␤1 nor ␣4␤1 integrin signaling alone was sufficient to mediate complete cell spreading. Activation of this signaling pathway occurred in the absence of syndecan-4 and utilized the Hep II domain as an ␣4␤1 integrin ligand. This signaling pathway is in contrast to previously reported studies in fibroblasts in which ␣5␤1 integrin-mediated cell spreading required syndecan-4 as a co-receptor and the Hep II domain was used as a syndecan-4 ligand. This is a novel signaling mechanism to control focal adhesion and stress fiber formation, and it indicates that different cell types utilize different receptor signaling pathways to control similar biological processes.
This novel integrin signaling pathway suggests that ␣5␤1 and ␣4␤1 integrin signaling pathways can converge to enhance cell spreading. In the presence of only the type III 7-10 repeats, ␣5␤1 integrin-mediated cell attachment resulted in only partial cell spreading, as evident by the formation of few focal adhesions and stress fibers. Concomitant with this partial ac-tivation of cell spreading, low levels of FAK phosphorylation were observed. Full cell spreading in these trabecular meshwork cells required additional signaling from ␣4␤1 integrin, which resulted in a higher level of FAK phosphorylation and increased stress fiber formation. An increase in the intensity of focal adhesions was also observed, suggesting that additional signaling complexes may have been recruited to the sites of cell attachment as a result of the ␣5␤1/␣4␤1 integrin co-signaling.
The ability of co-signaling between ␣5␤1 and ␣4␤1 integrins to control cell spreading is in contrast to previously reported studies in fibroblasts and A375-SM melanoma cells (8,13). In those studies, cell spreading was mediated by an ␣5␤1 integrin/ syndecan pathway, whereas in the HTM cells, it is mediated by an ␣5␤1/␣4␤1 integrin pathway. In addition, cell spreading in fibroblasts and the A375-SM cells was protein kinase C-dependent, whereas in HTM cells, it was independent of protein kinase C activation. Thus, HTM cells appear to use a different co-signaling pathway to activate cell spreading, focal adhesion, and stress fiber formation compared with that typically seen in fibroblasts and A375-SM cells, even though all the cells are using the same fibronectin domains. The difference in these pathways may simply reflect differences in the expression of  9. HTM cells attach to the type III 7-10 repeats via ␣5␤1 integrin. The ␣5 and ␤1 integrin blocking antibodies significantly inhibited cell attachment to the type III 7-10 repeats (p Ͻ 0.001). The ␣v and ␤3 integrin blocking antibodies had no effect. HTM cells were plated on the type III 7-10 repeats for 1 h in the absence or presence of integrin blocking antibodies or control. Bound cells were colorimetrically quantified using toluidine blue as described under "Experimental Procedures." Data are the mean percentage of control Ϯ S.E. the receptors on the cell surface because proliferating HTM cells in culture appear to express low levels of syndecan-4 on their cell surface. However, it could also reflect the functional activity of the Hep II domain. In HTM cells, the Hep II domain interacted with ␣4␤1 integrin, whereas in fibroblasts and A375-SM cells, it interacted with HSPGs, presumably syndecan-4. We have recently shown that the ability of the Hep II domain to utilize HSPGs to govern cell spreading is dependent on the splice pattern of the IIICS domain, which is adjacent to the Hep II domain (48). Because many of the other studies have used proteolytic fragments of the Hep II domain and not a recombinant fragment as we have, the splice pattern of the IIICS domain in the proteolytic fragment could be affecting which receptor binds the Hep II domain.
The co-signaling observed between ␣5␤1 and ␣4␤1 integrins in HTM cells indicates that the ␣5␤1/␣4␤1 integrin co-signaling response is an inherent feature of the cell type rather than a specific ␣4␤1 ligand. In SKMEL-178 melanoma cells, cosignaling between ␣5␤1 and ␣4␤1 integrins inhibited stress fiber formation, rather than activating it, and it was inhibited by a cryptic ␣4␤1 integrin binding site in either the type III 4 -5 repeats or IIICS domains of fibronectin (16). The ␣5␤1/␣4␤1 integrin co-signaling in HTM cells, on the other hand, activated cell spreading, focal adhesion formation, and stress fiber formation, and ␣4␤1 integrin signaling could be activated by the Hep II domain, the type III 4 -5 repeats, or the IIICS domain of fibronectin. These different responses to ␣5␤1/␣4␤1 integrin co-signaling must arise from the activation of different signaling pathways. In SKMEL-178 melanoma cells, ␣4␤1 integrin acts as an antagonist to ␣5␤1 integrin by interfering with the Rho activation pathway. In HTM cells, ␣4␤1 integrin ligation with the Hep II domain induced cell spreading, presumably by activating Rho kinase. The factors responsible for the differential activation of these signaling pathways are unknown, but they could be dependent on the activation state of the integrins as well as the expression of co-receptors such as chondroitin sulfate, which have been shown to enhance ␣4␤1 integrin signaling in melanoma cells (49 -51). Whether HTM cells express chondroitin sulfate proteoglycans at the cell surface is not known.
Unlike the studies with SKMEL-178 melanoma cells (16), this co-signaling response with ␣4␤1 integrins in HTM cells appears to be dependent on the conformational state and perhaps the affinity of the ligand. When any of the lower affinity ␣4␤1 ligands (Hep II, type III 4 -5 repeats) were co-coated with the III 7-10 repeats, the enhanced cell spreading was not observed. Presumably, absorption of these domains to the coverslip induced a conformational change that precluded the ␣4␤1 integrins from interacting with them. This may be inherent in the size of the domain because the larger VCAM-1 was still able to induce cell spreading when absorbed to the coverslip together with the type III 7-10 repeats, or it may be due to the affinity of the ␣4␤1 ligand. VCAM-1 has a much higher affinity for ␣4␤1 integrins than any of the other ligands (52), and recent studies have indicated that signaling via ␣4␤1 integrin is dependent on the activation state of the integrin due to the affinity of the ligand (53). This would have significant relevance in vivo, especially during physiological processes such as cell migration and invasion, wound healing (54), and intraocular homeostasis (55,56), where proteolytic fragments of fibronectin are normally produced as a result of these physiological events. In addition, other soluble ligands present in the serum that interact with ␣4␤1 integrin, such as VCAM-1 (57), may be present in aqueous humor and thus able to participate in intraocular homeostasis as well.
The ␣5␤1/␣4␤1 integrin co-signaling pathway in HTM cells exhibits a close similarity to the ␣4␤1 integrin/protein kinase C-independent signaling pathway reported previously in A375-SM melanoma cells (8). This suggests that in HTM cells, the ␣5␤1/␣4␤1 integrin co-signaling pathway may serve a similar function as the ␣4␤1 integrin signaling pathway in A375-SM cells. One such function may be to regulate a rapid reorganization of the actin cytoskeleton and adhesion contacts. Both of these processes have been shown to regulate migration and homeostasis of intraocular pressure (58,59).
Whether the co-signaling in HTM cells is just a matter of co-clustering two integrins into a single complex or the two different integrin signaling pathways converge is unclear. The addition of soluble Hep II domain induced HTM cell spreading, which suggests that the two integrin pathways converged. In addition, soluble Hep II domain also enhanced cell spreading on intact fibronectin, where presumably the Hep II domain and RGD cell binding domain would be able to easily co-cluster the two integrins, thereby negating the need for a second signal from the soluble Hep II domain.
This convergence leads to FAK phosphorylation because FAK phosphorylation increased when ␣4␤1 integrin was activated. The observation that ␣5␤1/␣4␤1 integrin co-signaling acts cooperatively to regulate FAK phosphorylation supports a previous study that showed that co-signaling events involving the cell binding and the Hep II domains of fibronectin acted cooperatively to regulate FAK phosphorylation (60). However, in that instance, the second signal appeared to have occurred via an interaction between the Hep II domain and syndecan-4.
Signaling via ␣4␤1 integrin also differed in these cells by the fact that the Hep II domain acted as an ␣4␤1 integrin ligand. To the best of our knowledge, this is the first time that the Hep II domain has been shown to activate ␣4␤1 integrin signaling mechanisms. Whether it is doing so via a direct interaction with ␣4␤1 integrin or indirectly via another cell surface receptor is unknown. A putative ␣4␤1 integrin binding site (IDAPS) has been identified within this domain (42), and in vitro assays have shown that the Hep II domain could bind ␣4␤1 integrin (42,43). Thus, it is plausible that in the absence of syndecan-4, the Hep II domain could bind and activate ␣4␤1 integrin.
The cooperative signaling of ␣5␤1 and ␣4␤1 integrins in trabecular meshwork cells seems counterintuitive compared with the traditional roles of ␣5␤1 and ␣4␤1 integrins. The ␣4␤1 integrin is normally thought to play a role in mediating cell migration by weakening cell contacts, and ␣5␤1 integrin mediates cell attachment. It addition, the need for two ␤1 integrins to promote adhesion would seem redundant. However, dual signaling via these two integrins may serve to control the adhesive strength and hence contractility of the HTM cells. Contractility of the actin cytoskeleton plays an important role in mediating the movement of aqueous humor through the anterior chamber of the human eye (21)(22)(23)(24)(25). At low levels of adhesiveness, weakly attached cells may not generate sufficient force to promote contraction of the trabecular meshwork and enhance movement of aqueous humor. At high levels of adhesiveness, strongly attached cells would generate too much contractility force, and it would be difficult to regulate changes in the movement of aqueous humor through the eye. Thus, a dual signaling mechanism of ␣5␤1 and ␣4␤1 integrins that could regulate the adhesiveness of the cell contacts could serve to adjust the contractility forces generated by the trabecular cells. Such a signaling mechanism could be used to control the flow rate of aqueous humor outflow through the trabecular meshwork in response to pressure changes in the anterior chamber.