Human Fibroblasts with Mutations in COL5A1 and COL3A1 Genes Do Not Organize Collagens and Fibronectin in the Extracellular Matrix, Down-regulate α2β1 Integrin, and Recruit αvβ3 Instead of α5β1 Integrin

Dermal fibroblasts derived from types I and IV Ehlers-Danlos syndrome (EDS) patients, carrying mutations in COL5A1 and COL3A1 genes, respectively, synthesize aberrant types V and III collagen (COLL) and show defective organization of these proteins into the extracellular matrix (ECM) and high reduction of their functional receptor, the α2β1 integrin, compared with control fibroblasts. EDS cells also show reduced levels of fibronectin (FN) in the culture medium and lack an FN fibrillar network. Finally, EDS cells prevalently organize αvβ3 integrin instead of α5β1 integrin. The αvβ3 integrin, distributed on the whole EDS cell surface, shows FN binding and assembly properties when the cells are treated with purified FN. Treatment of EDS cells with purified COLLV or COLLIII, but not with FN, restores the control phenotype (COLL+, FN+, αvβ3–, α5β1+, α2β1+). Function-blocking antibodies to COLLV, COLLIII, or α2β1 integrin induce in control fibroblasts an EDS-like phenotype (COLL–, FN–, αvβ3+, α5β1–, α2β1–). These results show that in human fibroblasts α2β1 integrin organization and function are controlled by its ligand, and that the α2β1-COLL interaction, in turn, regulates FN integrin receptor recruitment: high α2β1 integrin levels induce α5β1 integrin organization, while low α2β1 integrin levels lead to αvβ3 integrin organization.

The extracellular matrix (ECM) 1 is a complex structure formed by distinct molecular networks that interact with specific cell receptors, triggering numerous responses that play essential roles in cell behavior regulation (1). The ECM pro-vides a substrate for cell migration during embryonic development and wound healing, regulating tissue architecture and morphogenesis (2), and is also signaling variations in the cell microenvironment affecting cell proliferation, differentiation, and death (3)(4)(5)(6)(7).
FN is a dimeric glycoprotein that triggers cell adhesion, migration, cell cycle progression, and differentiation (4,7,21). FN deposition in vivo represents the initial event during fibrillogenesis of connective tissue matrices occurring during embryogenesis and wound healing (3,22,23). Human skin fibroblasts adhere in vitro through the organization of an ECM mainly composed of FN and types III, V, and VI COLLs (3,24). FN can bind to COLLs through several COLL binding sites (25), and FN binding sites in COLL molecules have been reported (26). Many data indicate the interdependence of FN and COLL network assembly; in particular, in several cell systems, the assembly of COLLs has been shown to depend on FN organization (3,27), whereas, in others, COLLs have been shown to influence FN fibrillogenesis (26,28).
FN and COLLs fibrils interact with the cells through specific plasma membrane receptors belonging to the integrin family (29,30). Integrins are ␣␤ heterodimeric transmembrane receptors with specific ligand binding potential involved in structural and regulatory functions such as linking ECM to actin cytoskeleton at focal adhesion sites and providing bidirectional transmission of signals across the plasma membrane (31,32). Cultured dermal fibroblasts express ␣ 2 ␤ 1 integrin (mediating cell adhesion to types I-VI COLLs), ␣ 1 ␤ 1 integrin (a minor COLL receptor that preferably binds to COLLIV and COL-LXIII) (33,34), ␣ 5 ␤ 1 integrin (as primary FN receptor (FNR), organized either in focal adhesions or in fibrillar adhesions) (7), and ␣ v ␤ 3 integrin (as minor FNR) (35). The ␣ v ␤ 3 integrin, highly expressed by dermal fibroblasts in cutaneous wound repair, supports their migration in the provisional matrix (36,37) and mediates adhesion to FN, vitronectin (VN), and fibrinogen (38). Another receptor binding either to FN or VN in cultured fibroblasts is the ␣ v ␤ 5 integrin (39,40).
We have previously demonstrated that all types of EDS fibroblasts (EDSI to EDSVII) do not organize the FN-ECM in vitro and express low levels of FN and of ␣ 5 ␤ 1 FNR integrin, compared with control fibroblasts (41)(42)(43). We report that EDSI and EDSIV skin fibroblasts, which carry mutations in COL5A1 and COL3A1 genes, respectively, do not organize the altered COLL molecules they encode in the ECM and show a reduced organization of ␣ 2 ␤ 1 integrin. This event is associated with the lack of the FN-ECM, the high reduction of ␣ 5 ␤ 1 FNR, and the prevalent organization of ␣ v ␤ 3 integrin. Treatment with purified COLLV or COLLIII, but not with FN, induces a control phenotype in both EDS fibroblasts, whereas perturbation experiments of COLLV and COLLIII or of the ␣ 2 ␤ 1 integrin receptor with specific antibodies induces an EDS phenotype in control fibroblasts. This evidence shows that perturbation of specific COLLs, either by mutation or by their functional blocking, leads to a cascade of alterations involving through their receptor other ECM components such as FN and its receptors.

EXPERIMENTAL PROCEDURES
Cells, Antibodies, and Reagents-Human control skin fibroblasts were established in our laboratory from an arm biopsy from an 18-yearold healthy donor. Types I and IV EDS fibroblasts strains were established and characterized for missense dominant mutations in the gene coding for the ␣ 1 chain of COLLV (44) and in the gene for the ␣ 1 chain of COLLIII (22), respectively. Two other EDS fibroblast strains were also analyzed: type I EDS (ATCC CRL 1215 JS) (43), type IV EDS (ATCC CRL 1243 CP). All cell strains, used at similar in vitro passages (4 -6), were grown in vitro at 37°C in modified Eagle's medium (Invitrogen) supplemented with 10% FBS (Invitrogen), 100 g/ml penicillin, and 100 g/ml streptomycin.
Cell Adhesion, Migration, and Proliferation-To study the adhesive properties of control and EDS fibroblasts, the cells were seeded on tissue culture wells not coated or coated overnight at 4°C with 10 g/ml human pFN (New York Blood Center Inc.), VN, or polylysine (Sigma Chemical Co.) and blocked for 60 min at 37°C by the addition of 1% BSA in PBS. 2 ϫ 10 3 cells were detached using 500 g/ml trypsin and 200 g/ml EDTA. Trypsin activity was inhibited by washing the cells with 1 mg/ml soybean trypsin inhibitor (Sigma Chemical Co.). Cells in the presence of serum-free medium were allowed to adhere in the absence and in the presence of FN for 30, 60, and 120 min at 37°C. Adhered cells were stained with crystal violet in 20% methanol and counted with a light microscope. The reported values are means Ϯ S.D. of three independent measurements.
The in vitro migration potential of control and EDS cells was ana-lyzed in Transwell 8-m filters (Corning Costar Corp. Cambridge, MA) as follows: 5 ϫ 10 4 fibroblasts, resuspended in serum-free medium, were plated onto the upper chambers and allowed to migrate for 6 h through the polycarbonate filter into the bottom wells filled with complete medium. The cells that migrated into the lower chambers were collected and counted. Each assay was performed in triplicate.
To study EDS cell proliferation, DNA synthesis was evaluated by immunofluorescence detection of BrdUrd (Roche Applied Science), according to De Petro et al. (45). 4 ϫ 10 3 cells were seeded on coverslips and grown to semiconfluence in complete medium. After washing three times with PBS, the cells were maintained in serum-free medium for 48 h to induce the non-proliferating G 0 phase. The cells were stimulated with 10% FBS-supplemented medium for 20 and 40 h at 37°C and treated with 10 M BrdUrd. After washing three times in PBS, the cells were fixed in methanol for 20 min at Ϫ20°C, blocked with 0.3% (w/v) BSA in PBS for 5 min at room temperature and incubated with mouse monoclonal anti-BrdUrd Ab (1:10 v/v in 66 mM Tris buffer, 0.66 mM MgCl 2 , 1 mM 2-mercaptoethanol) at 37°C for 30 min. After washing three times in PBS, a sheep FITC-conjugated anti-mouse secondary Ab (1:10 v/v in PBS) was added for 30 min at 37°C, and the coverslips were analyzed at a ϫ20 magnification. The number of fluorescent nuclei over all the cells was determined, and the data were expressed as percentage of labeled nuclei.
IF and FACS Analysis-The study of COLLs, FN, and their integrin receptors was performed by IF on control and EDS fibroblasts grown in modified Eagle's medium supplemented with 10% FBS. For this purpose, 1.5 ϫ 10 5 cells of each strain were seeded on 22 ϫ 22-mm glass coverslides and grown at 37°C in a 5% CO 2 incubator for 48 h. The cells were fixed in 3% paraformaldehyde and 60 mM sucrose for 5 min, permeabilized in 0.5% (v/v) Triton X-100 for 90 s, washed twice in PBS and 0.15 M glycine for 3 min, and immunoreacted with anti-FN, anti-COLLV, anti-COLLIII, or anti-COLLI polyclonal Abs (1:100 in diluting buffer containing 0.3% BSA and 0.1% NaN 3 in PBS). In parallel, the cells were reacted with mouse anti-␣ 5 ␤ 1 (clone JBS5), anti-␣ v ␤ 3 , anti-␣ 2 ␤ 1 integrin mAbs diluted in 1% BSA at a concentration of 4 g/ml, for 30 min at room temperature. All cells were washed 3ϫ 5 min in PBS and reacted with anti-rabbit IgG or anti-goat IgG and anti-mouse IgG, (10 g/ml), as secondary Abs, for 45 min at room temperature. The coverslides were mounted on glass slides in 1:1 PBS/glycerol solution; the IF signals were acquired by a CCD black and white TV camera (SensiCam-PCO Computer Optics GmbH, Germany) mounted on a Zeiss fluorescence microscope and acquired by Image Pro Plus program (Media Cybernetics, Silver Spring, MD).
Quantitative evaluation of fluorescence associated with the FN-and COLL-ECM organized by control and EDS cells, was performed as previously reported (25). The fluorescent images, with the same spatial resolution and comparable light intensity, were captured by a CCD black and white TV camera mounted on a Zeiss Axiovert 10S/H fluorescence microscope. Using the Image-Pro Plus program (Media Cybernetics, Silver Spring, MD), a task list file was developed that could be executed in semi-automatic mode on different images. Each image was processed by applying Sharpen and Rank nonlinear filters to optimize the image contrast and to remove the background noise. A binary image, visualized as a red overlay, corresponding to the fluorescence signal, was obtained by setting two thresholds in the fluorescence peak corresponding to the image gray tones visualized, with the lighter tones at the highest value of the peak and the darker ones below the slice background. The sequence of options in the task list was repeated on five images captured on each slide. The integrated optical density (IOD) related to the field area containing the signal was measured, and the cell number present in each image was counted.
The levels of integrin receptors distributed on the control, EDSI and EDSIV cell surfaces were evaluated by FACS analysis. For each strain, 5 ϫ 10 5 fibroblasts were washed twice in 0.1% BSA/PBS, centrifuged at 1500 rpm at ϩ4°C, and immunoreacted for 30 min on ice with saturating levels of anti-␣ 5 ␤ 1 (clone HA5, 10 g/ml), anti-␣ v ␤ 3 (5 g/ml), and anti-␣ 2 ␤ 1 (10 g/ml) integrin mAbs diluted in 1% BSA/PBS. After washing in cold 0.5% BSA/PBS, the cells were reacted with FITC-conjugated anti-mouse IgG and washed in 0.5% BSA/PBS. The cells were acquired with FACScan (BD PharMingen) and analyzed by CellQuest software (BD PharMingen) equipped with a 480-nm argon laser. The same cell strains that immunoreacted only with anti-mouse IgG were used as negative controls.
To study the role of FN integrin receptors in cell adhesion, control and EDS fibroblasts allowed to adhere in the absence and in the presence of 10 g/ml FN, VN, or polylysine coating, for 30, 60, and 120 min, were rinsed twice with PBS, fixed in 3% paraformaldehyde and 60 mM sucrose for 7 min, permeabilized in 0.5% (v/v) Triton X-100 for 90 s, washed twice with 0.15 M glycine/PBS, and immunoreacted with mouse anti-␣ 5 ␤ 1 and anti-␣ v ␤ 3 integrin mAbs as reported above.
In Vitro Treatment of Skin Fibroblasts with COLLs, Antibody Perturbation of COLLs, and ␣ 2 ␤ 1 Integrin Function-To test the role of COLLs on ECM assembly and on FN and COLL integrin receptor clustering, control and EDS fibroblasts were cultured for 48 h in serumfree medium in the presence of 5 g/ml human purified COLLV or COLLIII. After treatment with COLLs and immunoreaction with anti-COLLV, anti-COLLIII, and anti-FN serum, anti-␣ 5 ␤ 1 , ␣ v ␤ 3 , and ␣ 2 ␤ 1 integrin mAbs, the cells were analyzed by IF for the organization of COLLs and FN into the ECM and for ␣ 5 ␤ 1 , ␣ v ␤ 3 , and ␣ 2 ␤ 1 integrin distribution. The levels of integrin receptors organized after COLL treatment were evaluated by FACS analysis, as reported above. The effect of human purified pFN on COLL-and FN-ECM and on COLLs and FN integrin receptor organization was tested by culturing control and EDS cells in the presence of 10 g/ml pFN for 48 h.
To study the effect of COLL inhibition on ECM assembly, human control fibroblasts were cultured on glass coverslips in complete medium supplemented with 20 g/ml goat anti-COLLV and anti-COLLIII polyclonal antibodies for 4 days. Confluent cell cultures were fixed in 3% paraformaldehyde and 60 mM sucrose for 7 min, permeabilized in 0.5% (v/v) Triton X-100 for 90 s, immunoreacted with rabbit anti-FN, goat anti-COLLV, goat anti-COLLIII polyclonal antibodies, mouse anti-␣ 5 ␤ 1 , anti-␣ 2 ␤ 1 , and anti-␣ v ␤ 3 integrin mAbs, and analyzed by epifluorescence as reported above.
In parallel, human control fibroblasts were treated for 4 h with 10 g/ml anti-human ␣ 2 ␤ 1 integrin mAb recognizing the collagen binding site of the receptor and competing with the specific integrin ligand (46) and with a polyclonal anti-human-␣2 cytoplasmic tail subunit Ab.
Western Blot Analysis-The evaluation of FN in control and EDS cells was performed either in the culture medium or in deoxycholatesoluble and deoxycholate-insoluble fraction of each cell strain as follows: 3 ϫ 10 5 control, EDSI, and EDSIV cells were seeded for 2, 4, 8, 16, 24, and 48 h in complete medium. At each time complete medium was collected in the presence of protease inhibitors (25 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM aminobenzamidine). The cell layers were rinsed three times in cold PBS and treated for 10 min at 4°C in the sodium deoxycholate lysis buffer (0.5% deoxycholate in 10 mM Tris-HCl, pH 8.0, 1 mM phenylmethylsulfonyl fluoride, 2 mM EDTA) to collect the FN cytoplasmic soluble fraction. The pericellular matrices attached to the culture dishes (insoluble FN fraction) were scraped into a mixture of 4% SDS, 20% glycerol, 25 mM Tris-HCl, pH 8.0 and 0.002% bromphenol blue (47). Alternatively, the FN-ECM was washed in PBS, fixed in methanol, and immunoreacted with the polyclonal anti-FN Ab, as reported above.
The analysis of FN synthesized and secreted by control and EDS fibroblasts cultured with and without 5 g/ml COLLV and COLLIII was performed as follows: control, EDSI and EDIV cells were washed with PBS twice and collected 24 h after seeding, sonicated for 30 s, and centrifuged at 12,000 rpm for 30 min. Cell extracts and complete medium were collected in the presence of protease inhibitors (25 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM aminobenzamidine); 20 g of total proteins were diluted in Laemmli's buffer and separated by electrophoresis in 8% SDS-PAGE under non-reducing conditions. After nitrocellulose membrane (Schleicher & Schuell) transfer, the membranes were blocked overnight at 37°C with 3% nonfat milk (w/v in 0.1 M Tris-HCl, pH 8.1), immunoreacted for 2 h at 37°C with anti-FN f33 (25) mAb (final concentration 1 g/ml in 0.3% BSA, 0.01% NaN 3 in PBS), washed 3 ϫ 10 min in 20 mM Tris-HCl, pH 7.4, 150 mM NaCl (TBS), 0.1% Tween 20 (v/v) (TBS-T) at room temperature, and reacted for 2 h at 37°C with horseradish peroxidase-conjugated anti-mouse IgG.
The analysis of integrins synthesized by control and EDS fibroblasts was performed on cell cultures grown for 48 h in complete medium. The cells were kept at 4°C overnight in 0.5% (v/v) Triton X-100, 20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM MgCl 2 , 10 g/ml leupeptin, 4 g/ml pepstatin, and 0.1 KIU/ml aprotinin. After centrifugation at 14 000 rpm, 10 min at 4°C, the supernatants were recovered, 3 mg of proteins were immunoprecipitated with anti-␣ 5 ␤ 1 and anti-␣ v ␤ 3 mAbs (10 g of Ab/1 mg of protein) at ϩ4°C for 3 h in the presence of protein G-Sepharose beads (Pierce) diluted 1:1 in TBS and the immunocomplexes were recovered by centrifugation. After washing, the bound proteins were recovered by boiling in 1% SDS. Equal amounts of extracts were separated by 7% SDS-PAGE under non-reducing conditions, transferred to nitrocellulose sheets and reacted for 2 h at room temperature with anti-␣5␤1 and anti-␣v␤3 mAbs (1 g/ml) diluted in TBS-0.1% Tween 20 (TBS-T). The detection was performed incubating 2 h at room temperature with horseradish peroxidase-conjugated anti-mouse secondary Ab diluted in TBS-T.
Detection of FN and integrin subunits was carried out using the enhanced chemiluminescence (ECL) method (Pierce). The bands obtained were evaluated as IOD by the Gel-Pro 3.1 Analyzer software (Media Cybernetics, Silver Spring, MD) or by Analytical Imaging Station (AIS) software (Imaging Research INC., St. Catherine, Ontario, Canada).

Mutations in COL5A1 and COL3A1 Genes Affect the Organization of COLLs into the ECM and Clustering of ␣ 2 ␤ 1 Integrin
Receptor in Cultured Fibroblasts-We analyzed the assembly of COLLV and COLLIII in the ECM in EDSI and EDSIV fibroblasts, which carry mutations in COL5A1 and COL3A1, respectively, as compared with control fibroblasts. IF analysis on control fibroblasts shows that these cells organize COLLV and COLLIII in large fibrils (Fig. 1A, a and d) distributed in the intercellular spaces. On the contrary, EDSI-COL5A1 mutated cells do not organize COLLV in the ECM and retain this protein in the cytoplasm (Fig. 1A, b). These cells also show a large amount of COLLIII in the cytoplasm (Fig. 1A, e). EDSIV-COL3A1-mutated cells lack a COLLIII-ECM (Fig. 1A, f) but organize COLLV fibrils, as well as control fibroblasts (Fig. 1A,  c). Fig. 1B shows the quantitative evaluation by image analysis of COLLV and COLLIII signals detected by IF in control and EDS fibroblasts and shows that reduced amounts of each COLL molecules are present in EDS mutants and are mainly localized inside the cells.
These results show that both mutations present in the EDS strains analyzed lead to the defective organization of the abnormal protein into the ECM of each cell type and to the intracellu- lar storage of the abnormal proteins. The IF analysis of COLLI, the main collagen synthesized by skin fibroblasts, shows that in control and EDS fibroblasts this protein is not organized in the ECM and is mainly detectable in the cytoplasm (Fig. 1A, g-i). In particular, in EDS cells the amount of intracellular COLLI is higher, compared with control fibroblasts. Furthermore, we analyzed by IF the distribution in EDS cells of the major fibroblast COLL receptor, the ␣ 2 ␤ 1 integrin. This receptor, which is widely distributed in the plasma membrane of control cells (Fig. 1A, l), is strongly reduced in both EDS cell strains (Fig. 1A, m and n). FACS analysis shows that 91% of control fibroblasts are ␣ 2 ␤ 1 integrin-positive and their fluorescence intensity (FL1), ranging between 10 1 and 10 3 , is higher than that measured in negative controls (minus primary antibody) (Ͻ10 1 ) (Fig. 2). Only 3.3% EDSI and 3.9% EDSIV cells expose in membrane the ␣ 2 ␤ 1 integrin receptor (FL1 Ͼ 10 1 ) (Fig. 2). An analysis of the distribution of ␣ 1 -containing integrins, by IF and by FACS analysis with an anti-␣ 1 mAb, disclosed comparable patterns in control and EDS cells (data not shown). Because the ␣ 1 subunit has been found only in combination with the ␤ 1 subunit (48), this indicates that the ␣ 1 ␤ 1 integrin, acting as a COLL receptor, is not affected in EDS cells.
Similar COLLs and ␣ 2 ␤ 1 integrin patterns were observed in both EDS strains following a 2-day treatment with ascorbic acid, indicating that the defects observed in these cells do not depend on this cofactor involved in COLLs maturation (Figs. 1A and B and 2, A and B).
EDS Fibroblasts Lack the FN-ECM and Organize ␣ V ␤ 3 Instead of ␣ 5 ␤ 1 Integrin-In the control and EDS strains reported here, FN-ECM organization and the distribution of FN receptors, the ␣ 5 ␤ 1 and ␣ v ␤ 3 integrins, were also analyzed, in comparison with control fibroblasts. IF analysis, performed with an anti-FN polyclonal Ab, shows that control fibroblasts assemble FN in a fibrillar network overlaying the cells (Fig. 3A, a), whereas EDSI (Fig. 3A, d) and EDSIV (Fig. 3A, g) cells lack a fibrillar FN-ECM, and organize only a few FN fibrils in the extracellular spaces. In both EDS cell types, FN is stored in the cytoplasm. Quantitative evaluation by image analysis of FN detected by IF shows that in both EDS cells a double amount of protein, compared with control fibroblasts, is present (not shown). However, in control cells FN is mainly assembled in the ECM, whereas in EDS cells FN is mainly stored in the cytoplasm. The distribution and the level of FN in control and EDS cells were also analyzed by Western blotting, performed under non-reducing conditions with the f33 anti-FN mAb, on complete media, deoxycholate-insoluble and deoxycholate-soluble fractions collected from 2 to 48 h after cell seeding. Fig. 3, B and C show different distribution and levels of FN in the extracellular and in the intracellular compartment of control and EDS fibroblasts. In particular, the level of FN in the culture medium of control fibroblasts is 5-14-fold higher than that evaluated in EDS cells at 48 h after seeding (Fig. 3B, a and  C, a). Increasing amounts of FN are detectable in the medium of control fibroblast from 2 to 48 h after seeding, whereas the amount of FN in EDS cells increases from 2 to 16 or 24 h (EDSIV and EDSI cells, respectively) and decreases at 24 and 48 h (Fig. 3B, a and C, a). The deoxycholate-insoluble fraction of control cells contains increasing amounts of FN from 2 to 16 h; after this time, the FN level slightly decreases (Fig. 3B, b  and C, b). The deoxycholate-insoluble fraction of both EDS cells contain low levels of FN, which are detectable only at 24 and 48 h after seeding (Fig. 3B, b and C, b). These data are in agreement with the IF reported in Fig. 3, A and B and are confirmed by the IF analysis of the FN present in the deoxycholate-insoluble fraction of the three cell strains (Fig. 3D). Finally, the levels of FN present in the deoxycholate-soluble fraction of control cells decreases as a function of the time after seeding, whereas it increases, from 2 to 48 h, in both EDS cells (Fig. 3B, c and C, c). At 48 h the amount of FN in EDS cells is about 3-and 14-fold higher than in control cells. In the EDS deoxycholate-soluble fractions degradation of FN to a different extent also occurs (Fig. 3B, c). Taken altogether, these data confirm the accumulation of FN detected in EDS cells by IF (Fig.  3A) and also show a reduction of FN in the extracellular compartment of these cells, either in the soluble form, present in the culture medium, or in the insoluble form organized in the ECM.
The distribution of the most abundant FNR expressed by cultured fibroblasts, the ␣ 5 ␤ 1 integrin, analyzed by IF with an anti-␣ 5 ␤ 1 FNR mAb, that recognizes both subunits, shows that in control fibroblasts (Fig. 3A, b), these integrin patches are distributed over the whole cell surface either in focal or in fibrillar adhesions (7). On the contrary, EDS cells (Fig. 3A, e and h) show rare ␣ 5 ␤ 1 integrin patches, preferentially distributed in focal adhesions and at the boundary between cells, where few FN fibrils are organized (Fig. 3A, d and g). IF analysis of ␣ v ␤ 3 integrin FNR distribution in control fibroblasts shows rare dot-like patches localized at focal adhesion sites (Fig. 3A, c) and large integrin patches either in the focal or in the fibrillar adhesions in EDSI and EDSIV cells (Fig. 3A, f and  i). The cytoplasmic/nuclear storage of ␣ v ␤ 3 integrin detected in control fibroblasts is reduced or absent in EDS cells. Fig. 3E shows that the average number of ␣ v ␤ 3 integrin patches organized per cell by control fibroblasts is 7 Ϯ 2. This number increases 31-47-fold in EDS cells. Western blotting analysis, performed with specific Abs against ␣ 5 ␤ 1 and ␣ v ␤ 3 integrins expressed by control and EDS fibroblasts, shows that while the ␣ 5 ␤ 1 integrin is 2.3-2.4-fold higher in control cells, compared with EDS fibroblasts, ␣ v ␤ 3 integrin is 2.8-fold higher in EDS cells, compared with control fibroblasts (Fig. 4). These data are in agreement with those obtained by IF analysis (Fig. 3A).
The evaluation of ␣ 5 ␤ 1 integrin by FACS analysis confirms its reduction in EDS cells; indeed, only 2% of EDSI and 4% of EDSIV cells show a FL1 ranging between 10 1 and 10 2 , while 98% of control fibroblasts are positive for this integrin (peak at 10 2 ) (Fig. 2, A and B). In parallel with the ␣ 5 ␤ 1 integrin low levels , EDS cells show high ␣ v ␤ 3 FNR levels, compared with control fibroblasts which are almost negative for this integrin (Fig. 2, A and B). The FACS analysis also shows that the reduction of ␣ 5 ␤ 1 integrin in EDS fibroblasts is paralleled by an increase of ␣ v ␤ 3 integrin patches, compared with control fibroblasts. The levels of these integrins detected by FACS and IF analyses are not identical, because of the different and specific Abs used in each assay for the same receptor.
A comparable disorganization of COLL-and FN-ECM, as well as the peculiar distribution of their integrin receptors reported above, is detectable in other EDSI and EDSIV strains thus suggesting that this is a common feature of EDS cells carrying mutations in COLLV and COLLIII genes (Figs. 3, A-E  and 4).
␣ v ␤ 3 Integrin Is an FN and a VN Receptor in EDS Fibroblasts-Both EDS cell strains, with reduced ␣ 5 ␤ 1 integrin levels, show adhesion, proliferation, and migration properties comparable with those exhibited by control fibroblasts (Fig. 5). These results suggest that the ␣ v ␤ 3 integrin receptor expressed by EDS could be involved in cell adhesion, growth, and migration.
To verify whether ␣ v ␤ 3 integrin, expressed by EDS-FN Ϫ cells, is an FNR or a VN receptor, EDSI (Fig. 6) and EDSIV (not shown) cells were seeded on uncoated or FN-or VN-coated coverslips and, after 60 min, immunoreacted with anti-␣ v ␤ 3 mAb. Polylysine coating completely inhibits the adhesion and spreading of both cell types (Fig. 6, d and h). Both control and EDS cells adhere and spread more efficiently in the presence of FN and VN than in their absence. Control fibroblasts organize ␣ v ␤ 3 integrin patches mainly in the focal adhesions when seeded either on FN or VN (Fig. 6, b and c), indicating that in these cells the ␣ v ␤ 3 integrin is involved in FN and VN adhesion. In EDS fibroblasts, ␣ v ␤ 3 integrin is abundantly organized by cells seeded on uncoated and FN-coated coverslips (Fig. 6, e and f). In particular, in the presence of FN, ␣ v ␤ 3 integrin is distributed on the whole lower cell surface (Fig. 6, f), while in the presence of VN it is localized only in focal adhesions (Fig. 6,  g), confirmed by confocal microscopy (not shown). Similar results were obtained for EDSIV fibroblasts. Therefore, ␣ v ␤ 3 integrin organized by EDS cells is either an FN or a VN receptor: VN binding involves patches localized in focal adhesions, while FN binding mainly involves integrin patches distributed on the whole lower cell surface.
In order to verify whether the ␣ v ␤ 3 integrin patches organized in the fibrillar adhesions of EDS cells (Fig. 3A, f and i) are FNRs, EDSI cells were treated for 48 h with increasing amounts of purified human pFN (from 1 to 10 g/ml). Starting from 2.5 g/ml, the treatment induced the formation of a fibrillar FN-ECM (not shown). Therefore, both EDS cell strains were treated with 5 g/ml pFN. Exogenous pFN induces in EDSI and EDSIV cells the organization of a fibrillar ECM network, comparable with that assembled by control fibroblasts (Figs. 7  and 3A, a). However, the treatment does not induce in EDS cells the ␣ 5 ␤ 1 and ␣ 2 ␤ 1 integrin clustering and the ␣ v ␤ 3 integrin patch disorganization observed in control fibroblasts (Figs. 3A and 1A), as also detected by FACS analysis. Treatment of EDSI and EDSIV cells with pFN also fails to restore the organization of COLLV and COLLIII in the ECM. In control fibroblasts, treatment with pFN induces the organization of a thicker FN-ECM, the enhancement of ␣ 5 ␤ 1 and ␣ 2 ␤ 1 integrins and the organization of COLLIII, but not of COLLV, in the ECM (Fig. 7). Taken altogether, these data indicate that in EDS cells the ␣ v ␤ 3 integrin forming the fibrillar adhesions is an FNR capable of organizing the exogenous FN. Moreover, these data show that the FN-␣ v ␤ 3 binding is not sufficient to trigger ␣ 2 ␤ 1 integrin patch organization (Figs. 5-7).
Purified COLL Treatment Restores a Control Phenotype in EDS Fibroblasts-Western blotting analysis with specific anti- FIG. 4. Western blotting analysis of ␣ 5 ␤ 1 and ␣ v ␤ 3 integrins immunoprecipitated from control, EDSI and ED-SIV fibroblasts. A, equal amounts of immunoprecipitated cell proteins for each sample were immunoblotted with specific anti-␣ 5 ␤ 1 and ␣ v ␤ 3 integrins mAbs. B, quantitative evaluation by image analysis of the bands detected in A. FD, fold decrease obtained by normalizing the IODs over the highest values measured in control or EDS cells.
COLLV and anti-COLLIII Ab showed a slight reduction in the conditioned media of EDSI and EDSIV of COLLV and COL-LIII, respectively, as compared with control fibroblasts (not shown). In order to verify whether the defective COLL assembly in EDS fibroblasts (Fig. 1A) could be restored by exogenous addition of the COLL molecules they lacked, EDS fibroblasts were grown in medium containing human purified COLLV or COLLIII. Immunoreaction with polyclonal anti-COLLV Ab and anti-␣ 2 ␤ 1 integrin mAb was performed on EDSI cells, treated with increasing amounts of purified COLLV. Fig. 8, A and B shows that, starting from 2.5 g/ml, COLLV induced in EDSIdefective cells a COLLV-ECM, which increased with COLLV concentration. The organization of the COLLV-ECM was paralleled by the enhancement of ␣ 2 ␤ 1 integrin receptor in the plasma membrane of the treated cells. In particular, 10 g/ml COLLV induced an 11-fold enhancement of the ␣ 2 ␤ 1 integrin level, as evaluated by image analysis of the IF signals. Treatment with COLLV also induced in EDSI cells a reduction of the cell surface to values comparable to those of control fibroblasts and a more elongated phenotype, compared with untreated cells (Fig. 8B). Therefore, EDSI-and EDSIV-defective cells were treated with 5 g/ml COLLV or COLLIII, respectively. As already shown in Fig. 8, EDSI cells, after COLLV treatment, were capable of organizing this molecule in the extracellular spaces in a network (Fig. 9A, b) similar to that observed in control cells grown without and with COLLV (Figs. 1A, a and  9A, a). Likewise, treatment of EDSIV cells with COLLIII leads to the organization of this protein into an ECM (Fig. 9A, d) comparable to that observed in control fibroblasts without and with COLLIII treatment (Figs. 1A, d and 9A, c). Control fibroblasts in the presence of COLLV or COLLIII showed a slight increase of COLLV-and COLLIII-ECM (Fig. 9A, a and c and  9B), if compared with basal conditions (Fig. 1A, a and d and  1B). Treatment of EDS cells with COLLV and COLLIII also induced the organization of ␣ 2 ␤ 1 integrin patches to levels comparable with those observed in control fibroblasts treated with the purified molecules (Fig. 9A, e-h). These data were confirmed by FACS analysis showing that 97% of EDSI and EDSIV cells treated with COLLV and COLLIII, respectively, organized levels of ␣ 2 ␤ 1 integrin (Fig. 10) comparable to those measured in untreated control fibroblasts (Fig. 2). Therefore, in EDS cells the level of membrane-bound ␣ 2 ␤ 1 integrin receptor is regulated in the same manner by both COLLV and COLLIII ligands: the receptor clustering is up-regulated in the presence of an organized COLL-ECM and down-regulated in its absence.
COLL treatment also induces in EDS cells the FN-ECM reorganization (Fig. 9A, j and l and 9B). Western blotting analysis with the f33 anti-FN mAb shows that this event is associated with the presence of enhanced FN levels in the culture medium of EDS fibroblasts, which are comparable to those measured in control cells (Fig. 11). In parallel with the increase of extracellular FN, both EDS cell strains after COLLs treatment showed a strong reduction of intracellular FN (Fig. 11).
COLLV or COLLIII treatment induces in EDS cells not only the organization of the FN-ECM but also an ␣ v ␤ 3 /␣ 5 ␤ 1 integrin switch. In particular, the ␣ v ␤ 3 patches in EDSI and EDSIV COLL-treated cells are disorganized (Fig. 9A, r and t) and their number per cell decreases from 225 and 334, in EDSI and EDSIV, respectively (Fig. 3E) to 10 in both EDS cells, after COLL treatment (not shown). These values are comparable to those measured in control fibroblasts (Fig. 3E). At the same time, COLL treatment induced in EDS cells the formation of numerous ␣ 5 ␤ 1 integrin patches distributed over the whole cell surface (Fig. 9A, n and p). FACS analysis of these FNRs in EDS-COLL-treated cells showed that they reach levels (Fig. 10) similar to those observed in control fibroblasts (Fig. 2). In conclusion, the addition in the culture medium of EDS cells of purified COLL induces an ECM and an integrin profile comparable to those of control fibroblasts, probably acting on the up-regulation of the ␣ 2 ␤ 1 integrin receptor organization (Figs.  8 and 9 (A and B), 10, and 11). FIG. 5. In vitro EDS cell adhesion, proliferation, and migration assays. A, control fibroblasts (C), EDSI, and EDSIV cells were plated on culture wells coated with FN and counted after 30, 60, and 120 min. B, the mitogenic activity was tested by immunofluorescence detection of BrdUrd incorporation in synchronized C, EDSI, and EDSIV cells. C, number of control, EDSI, and EDSIV cells migrating through an 8-m filter was evaluated 6 h after seeding. The data represent the mean Ϯ S.D. of three replicate wells.
FIG. 6. The ␣ v ␤ 3 integrin is an FN and a VN receptor in ␣ 5 ␤ 1deficient EDS cells. The ␣ v ␤ 3 integrin receptor has been analyzed in control (C) and EDSI cells adhered to uncoated (-) and FN-, VN-or polylysine-coated substrate for 60 min by IF with anti-␣ v ␤ 3 integrin mAb. Scale bar, 5 m.

Perturbation of COLLs and Functional Blocking of Their Receptor Induce an EDS Phenotype in Control Fibroblasts-To
address further the role of COLLV and COLLIII in the organization of FN and in the induction of the EDS phenotype, the effects of polyclonal Abs directed against COLLV and COLLIII in control fibroblasts, organizing either COLLs or FN in the ECM, were investigated. Confluent control cells were treated with antibodies against COLLV or COLLIII blocking the sites of polymerization of these proteins. Both treatments induced in control cells an EDS phenotype, i.e. strong reduction of COLLs in the ECM, disorganization of the FN network, reduction of ␣ 5 ␤ 1 integrin patches, organization of ␣ v ␤ 3 patches and decrease of ␣ 2 ␤ 1 COLL receptor (Fig. 12). In particular, anti-COLLV Ab inhibits the organization of COLLV and COLLIII into the ECM, whereas anti-COLLIII Ab inhibits COLLIII, but not COLLV, assembly; again indicating that COLLV is necessary in cultured skin fibroblasts for COLLIII deposition but not vice versa.
Similar results were obtained in adherent control fibroblasts after functional blocking with a mAb directed to the ␣ 2 ␤ 1 integrin COLL binding site (46) (Fig. 13). Indeed, the lack of functional ␣ 2 ␤ 1 integrin does not affect the adhesive properties of control fibroblasts and induces the ␣ 5 ␤ 1 /␣ v ␤ 3 integrin switch, which is not observed after treatment of the cells with a polyclonal anti-␣ 2 cytoplasmic tail integrin Ab (Fig. 13, ϩ anti-␣ 2 ). These results demonstrate that in EDSI and EDSIV cells the ECM defect depends on the lack of organized COLLs in the ECM and that COLL disorganization modulates the level of its receptor. This event, in turn, induces the ␣ 5 ␤ 1 /␣ v ␤ 3 switch (Figs. 12 and 13).

DISCUSSION
In this work we studied the effect of mutations in COL5A1 and COL3A1 genes, leading in vivo to classical (EDSI) and vascular (EDSIV) EDS, on cultured skin fibroblast phenotypes. In particular, in these cells we studied the organization of COLLs and FN into the ECM, the distribution of their integrin receptors, and the adhesive and growth features in comparison with skin fibroblasts derived from a healthy donor. Control fibroblasts organize an ECM composed mainly of FN, but also containing COLLIII and COLLV fibrils and rare COLLI fibrils. EDS cells, carrying mutations in COL5A1 and COL3A1 genes, do not organize the aberrant proteins into the ECM, although they synthesize these molecules that are stored as intracellular deposits and partly secreted in the culture medium. In the absence of organized extracellular protein, the cytoplasmic FIG. 7. Effect of purified FN treatment on the FN-and COLLs-ECM organization and on FNR and COLL receptor cell membrane distribution in control and EDS fibroblasts. Control (C), EDSI (COLLV Ϫ ), and EDSIV (COL-LIII Ϫ ) cells treated with purified pFN added to the culture medium for 48 h were immunoreacted with polyclonal anti-FN, anti-COLLV, and anti-COLLIII Abs and with mouse anti-␣ 5 ␤ 1 , anti-␣ v ␤ 3 and anti-␣ 2 ␤ 1 integrin mAbs. Both EDS fibroblast strains were induced by FN treatment to organize a fibrillar FN-ECM without modifying the COLL assembly and the FNR and COLL receptor organization. Scale bar, 5 m.

FIG. 8. Dose response effect of COLLV on COLLV assembly and on ␣ 2 ␤ 1 COLL receptor integrin clustering in EDSI-defective cells.
A, IF was performed after treatment with 1-10 g/ml COLLV, with anti-COLLV Ab, and with anti-␣ 2 ␤ 1 integrin mAb on EDSI cells. Scale bar, 5 m. B, quantitative evaluation by image analysis of ␣ 2 ␤ 1 integrin, expressed as IOD, and of the cell area, expressed in pixels ϫ 10 8 and ϫ 10 3 , respectively. Fifty cells were measured in each sample. storage of COLLIII molecules is a common feature of in vitro grown skin fibroblasts derived from EDSIV patients, who probably carry different mutations in the COL3A1 gene that influence protein export (49,50). The aberrant COLLIII and COLLV proteins secreted by EDSIV and EDSI cells are not capable of organizing into fibrils detectable by IF analysis. The COL5A1 mutation present in EDSI cells not only reduces COLLV organization into the ECM but also affects the assembly of COL-LIII, showing that a missense mutation in a COL5A1 gene prevents the assembly into the ECM of another COLL. On the contrary, COL3A1 mutation present in EDSIV cells does not interfere with COLLV secretion and its organization in the ECM. These results, in combination with the finding that functional blocking of COLLV with a specific mAb abolishes not only COLLV but also COLLIII assembly in control fibroblasts, while COLLIII blocking inhibits COLLIII but not COLLV organization, lead to the conclusion that the deposition of COLLV is needed for the organization of COLLIII into the ECM.
In both EDS cell strains analyzed here the synthesis of altered COLL molecules also influences the assembly into the ECM of FN, a non-collagenic ECM protein. Indeed, in EDSI and EDSIV cells, FN is not organized in the extracellular microenvironment, although it is stored in the cytoplasm. The lack of an FN-ECM, here described in EDSI and EDSIV cells, has been reported as a common feature of skin fibroblasts from different EDS types (types I to VII and non-classified EDS), irrespective of their molecular defects (41,42). In both EDS strains COLL alterations also affect the level of soluble FN present into the extracellular environment, thus explaining the absence of an organized FN-ECM. That retention of intracellular FN in EDS cells is associated with disorganization of the FN-ECM has already been reported (43). In particular, induction of FN secretion with dexamethasone is capable of restoring a control FN-ECM in EDS fibroblasts. In this work we observed that treatment of EDS cells with the COLL molecule they fail to secrete, as a consequence of specific mutations, not only restores the COLL-ECM, but also induces the enhancement of FN in the culture medium and its organization into the ECM. Therefore, the assembly into the ECM of skin fibroblasts of structural proteins such as COLLIII, COLLV, and FN is influenced by their levels in the culture medium.
ECM components are known to modulate integrin expression (51); in particular, ␣ v ␤ 3 integrin is up-regulated by FN and fibrin and down-regulated by COLL in endothelial cells (37), and ␣ 5 ␤ 1 and ␣ 2 ␤ 1 integrins are highly induced by plateletderived growth factor when dermal fibroblasts are in a COLL gel (52). Thus the integrin-ECM interaction is reciprocal: integrins mediate cell attachment to the matrix, and the matrix, at least to a degree, controls the expression and activity of integrins (30). Therefore, EDS fibroblasts showing alterations in the ECM organization are a cellular model for the study of COLL and FN integrin receptor modulation, consequent to COLL and FN variation in the extracellular environment. We demonstrate that defective COLL assembly in EDS cells, or inhibition of COLL deposition by antibody perturbation in control fibroblasts, induces modifications of the FN integrin receptor repertoire: the decrease of ␣ 2 ␤ 1 and ␣ 5 ␤ 1 integrin receptors and the organization of ␣ v ␤ 3 FNR in the cell surface. The primary event consequent to COLLV or COLLIII defect in the ECM of cultured fibroblasts, is the reduction of the COLL receptor, the ␣ 2 ␤ 1 integrin, in the plasma membrane, as demonstrated by IF and FACS analysis and by ␣ 2 ␤ 1 integrin functional blocking with a specific antibody competing with its ligand(s), which induces the ␣ 5 ␤ 1 /␣ v ␤ 3 integrin switch in control fibroblasts. On the other hand, treatment of EDS cells with purified COLLIII and COLLV molecules induces the organization of ␣ 2 ␤ 1 and ␣ 5 ␤ 1 integrins and the disorganization of ␣ v ␤ 3 integrin. These data indicate that the ligand (COLL) specific for ␣ 2 ␤ 1 integrin receptor is a signal transduced inside the cell that induces the organization of ␣ 5 ␤ 1 and the disorganization of FIG. 9. Effect of COLLV and COL-LIII treatment on the organization of COLLV, COLLIII, ␣ 2 ␤ 1 integrin, FN, ␣ 5 ␤ 1 , and ␣ v ␤ 3 integrin in control and EDS fibroblasts. A, control (C), EDSI (COLLV Ϫ ) cells, grown in serum-free medium supplemented with COLLV, were immunoreacted with polyclonal anti-COLLV Ab (a and b), with anti-␣ 2 ␤ 1 integrin mAb (e and f), with anti-FN polyclonal Ab (i and j), with anti-␣ 5 ␤ 1 integrin mAb (m and n) and with anti-␣ v ␤ 3 integrin mAb (q and r). EDSIV (COLLIII Ϫ ), grown in serum-free medium supplemented with COLLIII, were immunoreacted with polyclonal anti-COLLIII Ab (c and d), with anti-␣ 2 ␤ 1 integrin mAb (g and h), with anti-FN polyclonal Ab (k and l), with anti-␣ 5 ␤ 1 integrin mAb (o and p) and with anti-␣ v ␤ 3 integrin mAb (s and t). Scale bar, 8 m. B, quantitative evaluation, performed by image analysis, of COLLV, COLLIII, and FN detected by IF.
The IOD values, expressed in pixels ϫ 10 8 , are average numbers obtained by evaluating the fluorescent signals in five fields of 1.3 ϫ 10 6 pixels for each sample. The average number of cells counted in each field was 6 Ϯ 2 control fibroblasts and 3 Ϯ 1 EDSI and EDSIV cells.
␣ v ␤ 3 integrin receptors binding to another ECM molecule.
Addition of purified FN into the culture medium of EDS cells induces the formation of an FN-ECM but fails to restore ␣ 5 ␤ 1 integrin patches, suggesting that ␣ v ␤ 3 -FN complexes do not trigger ␣ 2 ␤ 1 integrin recruitment in the absence of a COLL molecule organized into the ECM. Taken altogether, these data reinforce the role of ␣ 2 ␤ 1 integrin in the control of FNR recruitment and show that FNR switch is regulated not by FN present in the ECM but rather by an ␣ 2 ␤ 1 -transduced signal. Therefore, the absence of an integrin ligand not only affects its receptor organization but also triggers a cascade of events involving modulation of receptors having different ligand specificities. Although the intracellular mechanism by which the modulation of FNRs by ␣ 2 ␤ 1 integrin occurs has not been yet investigated, the data reported here indicate the existence of functional cooperation between these integrins: ␣ 2 ␤ 1 integrin in the absence of its ligand exerts inhibition of ␣ 5 ␤ 1 and activation of ␣ v ␤ 3 integrin organization in skin fibroblasts, underlining the existence of cross-talk events between integrins already reported in different cell models (40,(53)(54)(55)(56).
With respect to the ␣ v ␤ 3 integrin organized by EDS cells, this receptor supports cell adhesion and in vitro growth in the absence of either ␣ 2 ␤ 1 or ␣ 5 ␤ 1 integrin. Whereas in vivo the ␣ v ␤ 3 integrin is typically expressed by fibroblasts migrating into the provisional matrix of FN and fibrinogen during wound healing (36,37), in in vitro cultured fibroblasts, this integrin is a minor FNR mainly organized in the focal adhesions (7). In EDS cells, ␣ v ␤ 3 patches are organized not only in focal but also in fibrillar adhesions (7,57) distributed over the whole cell surface. This is a peculiar phenotypic trait of all EDS fibroblast types and therefore, together with the lack of the FN-ECM, it represents a marker of skin fibroblasts derived from EDS patients. Adhesion assay of EDS fibroblasts to FN and VN has shown that these cells engage ␣ v ␤ 3 integrin in focal adhesions for binding to either FN or VN and that the largest amount of ␣ v ␤ 3 integrin on the lower cell surface binds to FN. When the cells are maintained in complete medium, these ␣ v ␤ 3 integrin patches are distributed in the fibrillar adhesions, which are localized on the whole cell surface, and are capable of supporting the assembly of exogenous FN. However, in the absence of FIG. 10. Effect of COLLV and COLLIII treatment on the cell surface distribution of ␣ 5 ␤ 1 , ␣ v ␤ 3 , and ␣ 2 ␤ 1 integrin receptor in EDS skin fibroblasts, as detected by FACS analysis. EDSI and EDSIV cells were collected and immunoreacted with saturating concentrations of anti-␣ 5 ␤ 1 , anti-␣ v ␤ 3 , and ␣ 2 ␤ 1 FACS grade mAbs, prior to fluorescence flow cytometry analysis. The fluorescence intensities of the different antibodies, displayed as histograms in A, were subtracted from that of the negative control (-I ary Ab). B, percentage of positive cells evaluated in each sample.
high levels of FN in the medium, these ␣ v ␤ 3 integrin patches could, in EDS FN-ECM-deficient cells, be involved in cellular processes other than cell adhesion and spreading. Indeed, it is known that both ␣ v ␤ 3 and ␣ 5 ␤ 1 integrins can act synergistically with the epidermal growth factor receptor and transduce survival and proliferation signals from the ECM inside the cell (54,58,59). In EDS cells defective for the ECM and deprived of its survival signals (5,60,61), the ␣ v ␤ 3 integrin clustered in fibrillar adhesions might play an important role in the control of cell growth and survival.
The fact that the ␣ v ␤ 3 alternative FNR does not trigger FN organization into the ECM in EDS cells is probably due to the reduced amount of FN present in the culture medium of these cells, as compared with control fibroblasts. This hypothesis is supported both by the biochemical analysis of extracellular FN levels and by the exogenous FN assembly capability exhibited by EDS-␣ v ␤ 3 ϩ cells. FN bound to ␣ v ␤ 3 integrin is organized into the ECM even in the absence of a COLL-ECM, supporting the notion that deposition of FN into the ECM is independent and precedes COLL assembly (3,27,62). The FN-ECM organized by ␣ v ␤ 3 integrin is not sufficient to revert the EDS fibroblasts phenotype since it does not induce the up-regulation of ␣ 2 ␤ 1 integrin and the consequent organization of COLL into the ECM. These data show the existence of a hierarchy of events regulating ␣ 2 ␤ 1 , ␣ 5 ␤ 1 , and ␣ v ␤ 3 integrins in a coordinated and directional manner: secreted COLL binds to ␣ 2 ␤ 1 integrin and up-regulates its organization in plasma membrane; COLL receptor recruits ␣ 5 ␤ 1 FNR with the resulting disorganization of ␣ v ␤ 3 integrin; FN is secreted; the FN-ECM is assembled; and the COLL-ECM is assembled.
Because ␣ v ␤ 3 and ␣ 5 ␤ 1 integrins on fibroblast plasma membrane are required in different and specific stages in wound healing (36,38), our data give new insights into the understanding of the defective wound healing often observed in EDS patients and suggest the use of purified COLLs for restoring a normal wound-healing process. FIG. 11. Western blotting analysis of FN in control, EDSI, and EDSIV cells, before and after treatment with COLLV or COLLIII. A, Western blotting analysis with the f33 anti-FN mAb, performed under non-reducing conditions, of FN released in the culture medium (CM) or present in the cell extract (CE) of control (C), EDSI, and EDSIV cells cultured for 48 h in the absence and in the presence of 5 g/ml human COLLV or COL-LIII. Equal amounts of CM and CE proteins from all fibroblasts cultures were analyzed. The quantitative evaluation by image analysis of the FN bands shown in A, is presented in B. FI, fold increase obtained by normalizing the IOD measured in EDS cells, treated with COLLV or COLLIII, on the value measured in control cells under the same experimental conditions.