Cell Adhesion to Fibrillin-1 Molecules and Microfibrils Is Mediated by α5β1 and αvβ3 Integrins*

Fibrillins are the major glycoprotein components of microfibrils that form a template for tropoelastin during elastic fibrillogenesis. We have examined cell adhesion to assembled purified microfibrils, and its molecular basis. Human dermal fibroblasts exhibited Arg-Gly-Asp and cation-dependent adhesion to microfibrils and recombinant fibrillin-1 protein fragments. Strong integrin α5β1 interactions with fibrillin ligands were identified, but integrin αvβ3 also contributed to cell adhesion. Fluorescence-activated cell sorting analysis confirmed the presence of abundant α5β1 and some αvβ3 receptors on these cells. Adhesion to microfibrils and to Arg-Gly-Aspcontaining fibrillin-1 protein fragments induced signaling events that led to cell spreading, altered cytoskeletal organization, and enhanced extracellular fibrillin-1 deposition. Differences in cell shape when plated on fibrillin or fibronectin implied substrate-specific α5β1-mediated cellular responses. An Arg-Gly-Asp-independent cell adhesion sequence was also identified within fibrillin-1. Adhesion and spreading of smooth muscle cells on fibrillin ligands was enhanced by antibody-induced β1 integrin activation. A375-SM melanoma cells bound Arg-Gly-Asp-containing fibrillin-1 protein fragments mainly through αvβ3, whereas HT1080 cells used mainly α5β1. This study has shown that fibrillin microfibrils mediate cell adhesion, that α5β1 and αvβ3 are both important but cell-specific fibrillin-1 receptors, and that cellular interactions with fibrillin-1 influence cell behavior.

Fibrillins are large cysteine-rich glycoproteins and the major constituents of fibrillin-rich microfibrils of the extracellular matrix (ECM) 1 (1,2). They are multidomain molecules, containing 47 epidermal growth factor (EGF)-like domains and 8-cysteine (TB) modules. Fibrillin-1 and fibrillin-2 are both expressed during fetal development, but fibrillin-1 isoform is by far the most abundant isoform in adult tissues (3,4). The possibility that fibrillins interact with cell receptors was suggested by electron microscopy of vascular and other tissues showing juxtaposition of extracellular microfibrils and cellular dense plaques (5) and by the discovery that fibrillins contain the Arg-Gly-Asp (RGD) putative cell attachment motif (6). Integrins are ␣␤ heterodimeric transmembrane receptors that mediate cell adhesion to the ECM, usually through the characteristic RGD motif, and have widespread essential functions in development, tissue organization, and the immune system (7). Integrin-mediated cell interactions regulate cell adhesion and migration and initiate signaling pathways that lead to reorganization of actin cytoskeleton and cellular proliferative and secretory responses.
Fibrillin-1 contains one RGD sequence in the fourth TB module (6). Fibrillin-2 contains two RGD sequences, one in the same position as the fibrillin-1 RGD sequence in the fourth TB module and the other in the third TB module where it is surrounded by hydrophobic amino acids (8). Fibrillin-3 also has two RGD sequences, one in the 19th cbEGF repeat and a second in the fourth TB module (9). The RGD motifs located in the fourth TB modules are surrounded by polar and charged amino acid residues, suggesting that the motif is solvent-exposed. This RGD tripeptide is located in the middle of a 13-to 20-amino acid sequence that is flanked on both ends by cysteine residues. By analogy with the homologous sixth TB module of fibrillin-1 (10), disulfide bonding of these cysteines in the fourth module would produce a finger-like loop structure with the RGD near the end of the loop and available for cellular interactions. The RGD in the third TB motif in fibrillin-2 does not mediate cell adhesion, probably due to its inaccessibility (11).
A cell adhesion role for the RGD motif in the fourth TB module of fibrillin-1 and fibrillin-2 has been described using recombinant fibrillin-1 peptides (12,13), or synthetic fibrillin-1 RGD peptides and fibrillin molecules purified from tissues using a reductive denaturing protocol (11). These studies identified the integrin receptor ␣ v ␤ 3 as the major receptor mediating adhesion to these molecular fibrillin-1 ligands. One group found that purified ␣ 5 ␤ 1 integrins, when immobilized to wells, did not bind recombinant fibrillin-1 peptides (12). The ␤ 1 integrin subunit bound a recombinant fibrillin-1 TB4 module expressed in bacteria (13). Fetal bovine chondrocytes were stimulated to bind fibrillin-1 using ␣ 5 ␤ 1 after antibody activation (11).
In early studies, we showed that isolated fibrillin-rich microfibrils support attachment of vascular smooth muscle cells (14).
Here we have investigated the molecular basis of integrin-* This work was funded by the Medical Research Council, British Heart Foundation, Biotechnology and Biological Sciences Research Council, and Engineering and and Physical Sciences Research Council. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Both authors contributed equally to this work. ‡ ‡ To whom correspondence should be addressed. mediated adhesion to fibrillin microfibrils and molecules using human dermal fibroblasts, smooth muscle cells, HT1080 fibrosarcoma cells, and A375-SM melanoma cells. We have demonstrated that ␣ 5 ␤ 1 and ␣ v ␤ 3 are both important receptors for fibrillin ligands, and that RGD-dependent cell adhesion to fibrillin-1 influences cell shape and migration, focal complex formation, signaling, and ECM deposition. We have also identified a cell adhesion site within fibrillin-1 exons 41-52 that mediates cell adhesion in a non-RGD, non-cation, and nonheparan sulfate-dependent manner. These studies have provided new insights into the molecular basis of fibrillin interactions with cells and suggest a key role for microfibrils in regulating cell behavior in connective tissues.
Recombinant Fibrillin-1 Protein Fragments-Recombinant human fibrillin-1 protein fragments PF9 (amino acids 1528 -1807), PF10 (amino acids 1689 -2126), and PF11 (amino acids 1528 -2126) (6) were expressed and purified using the mammalian episomal expression system pCEP-pu/AC7 (obtained from Dr. R. Timpl, Munich, Germany) and 293-EBNA cells (Fig. 1A). The pCEP-pu/AC7 vector had been modified by incorporation of N-terminal His 6 following the signal peptide, which facilitated rapid peptide purification by nickel chromatography. Sitedirected mutagenesis was also used to mutate the Arg-Gly-Asp (RGD) sequence in PF9 to RDG. Following sequencing, this mutant (designated RDG PF9) was expressed and purified. The protein fragments were all N-glycosylated, had the correct molecular mass as judged by SDS-PAGE in the presence or absence of 10 mM dithiothreitol, size fractionation on Superdex 200 HR 10/30 columns, and laser light scattering, and they bound calcium, because electrophoretic shifts were apparent following EDTA treatment, as previously shown (19). Fibrillin-1 protein fragments expressed in a similar system are correctly folded (20 -22).
Microfibril Purification-Native microfibrils to be used as ligands in cell adhesion assays were isolated and purified from aortae of second trimester fetal calves obtained from a local abattoir, using a modification of a well defined methodology (Fig. 1B) (23,24). Briefly, microfibrils were isolated in the void volume of a Sepharose CL-2B column or a Sephadex G-25 column and then purified using a CsCl density gradient protocol. Fractions were extensively dialyzed into buffer (0.15 M NaCl, 0.05 M Tris-HCl, pH 7.4, 2 mM Ca 2ϩ ) at 4°C for 18 h. Western slot blots of CsCl fractions, prepared as previously described (24), showed that fibrillin-1 was present in the CsCl density gradient fractions 9 and 10 ( Fig. 1C). Microfibril purity of these fractions was, in all preparations, confirmed by electron microscopy (Fig. 1D) or by tapping mode atomic force microscopy in air using a Multimode AFM with a Nanoscope IIIa controller at a resonant frequency of 250 -300 kHz.
Cell Attachment and Spreading Assays-Standard cell attachment assays and cell attachment inhibition assays were performed using microtiter plates (Costar Corp.) according to well defined methodologies (25). Purified microfibrils, recombinant fibrillin-1 protein fragments, and fibronectin adhesion ligands were diluted in column buffer (0.4 M NaCl, 0.01 M Tris-HCl, pH 7.4, 2 mM Ca 2ϩ ), adsorbed to wells of 96 well microtiter plates, then incubated at room temperature for 1 h. The buffer and unbound ligand were aspirated from wells, and nonspecific binding was blocked by the addition of 200 l/well of 10 mg/ml heatdenatured BSA diluted in 130 mM NaCl, 3 mM Tris-HCl pH 7.4, at room temperature for 1 h. Cell suspensions were prepared by washing the cell layers with trypsin-EDTA and incubating with 1 ml of trypsin-EDTA/ 10-cm 2 flask area at 37°C for no more than 4 min. Resulting cell suspensions were neutralized with 5 volumes of DMEM containing 10% fetal calf serum, 2 mM glutamine, then centrifuged at 800 ϫ g for 5 min. Cell pellets were resuspended in 5 ml of warm DMEM/HEPES gassed with 5-10% (v/v) CO 2 , and cell numbers were calculated using a hemocytometer. Cell suspensions were centrifuged at 800 ϫ g for 3 min, and the cell pellets were then resuspended in an appropriate volume of warm DMEM/HEPES (contains Mg 2ϩ , Ca 2ϩ , and Mn 2ϩ ) gassed with 5-10% (v/v) CO 2 to give a final cell density of 5 ϫ 10 5 cells/ml. The cell suspensions were then incubated for 20 min at 37°C with the lid off in a 5% CO 2 incubator. The BSA-blocking solution was aspirated from the ligand-coated wells, which were then washed with 200 l of column buffer. To determine cell attachment to adhesion ligands, 100 l of cells (mixed prior to use by gentle pipetting) was added to the appropriate wells. Alternatively, to examine the effects of an exogenous agent on attachment, 50 l of 2ϫ exogenous agent (antibody, peptide, or heparan sulfate) followed by 50 l of 2ϫ cells were added to the wells. As controls when examining the effect(s) of exogenous agents, 50 l of column buffer followed by 50 l of 2ϫ cells were added to control wells.
To estimate 100% attachment, cells were initially diluted to 5 ϫ 10 5 cells/ml then further diluted to 20, 50, and 100% of the working cell suspension using warm DMEM/HEPES gassed with 5-10% (v/v) CO 2 , and 100 l of cells was added to uncoated, unblocked wells. The microtiter plates were incubated for 20 min at 37°C in a 5% CO 2 incubator with the lid removed. Cells in the wells to be used for determining 100, 50, and 20% cell attachment were fixed by adding 10 l of 50% (w/v) glutaraldehyde. Non-adherent and loosely attached cells were removed from the other wells by gentle tapping of the plate and aspirating the solution. Attached cells were then fixed by adding 100 l of 5% (w/v) glutaraldehyde per well. The microtiter plates were incubated at room temperature for 30 min then washed three times with 200 l of dH 2 O. Cells were stained by adding 100 l of 0.1% (w/v) crystal violet in 0.2 M MES, pH 5.0 to each well and incubating the microtiter plates at room temperature for 1 h, then wells were washed once with 200 l of dH 2 O, followed by three times with 400 l of dH 2 O. The dye was solubilized in 100 l of 10% (v/v) acetic acid, and the absorbance at 570 nm measured on a Dionex MRX II microtiter plate reader. Background crystal violet staining readings were subtracted from all experimental and 100% attachment results. Data from 20, 50, and 100% cell number standards were plotted (A 570 versus cell density) using Microsoft Excel. The slope of the graph was determined from a linear regression, which was then used to express data as percent attachment. In all experiments, triplicate or quadruplicate wells were prepared. To examine the effects of cations on attachment, 50 l 2x EDTA or PBS containing cations followed by 50 l 2 x cells resuspended in PBS minus cations (mixed prior to use by gentle pipetting) were added to ligand-coated and control wells after BSA-blocking.
To determine cell spreading, the wells of microtiter plates were ligand-coated and BSA-blocked as for cell attachment assays. The cells were trypsinized, quenched, and counted as before, then adjusted to 2 ϫ 10 5 cells/ml with warm DMEM/HEPES gassed with 5-10% (v/v) CO 2 . 100-l aliquots of cells were added to the appropriate wells. As negative controls cells were added to uncoated, BSA-blocked wells. The plate was incubated for 40 min at 37°C in a 5% CO 2 incubator with the lid removed. The cells were immediately fixed with the addition of 10 l of 37% formaldehyde directly to the well for 20 min. The formaldehyde was aspirated, and the wells were filled with PBS before layering a glass coverslip onto the plate. The level of cell spreading was determined by phase contrast microscopy.
FACS Analysis-Cell suspensions were prepared by washing confluent cultured cell layers with PBS and incubating with 1 ml of 5 mM EDTA in Hanks' buffered saline solution/10-cm 2 flask area at 37°C for no more than 30 min. The resulting cell suspensions were neutralized with 5 volumes of DMEM-5 (Dulbecco's minimal essential medium, including 25 mM Hepes, 500 mg/liter glucose, 4 mg/liter pyridoxine) supplemented with 10% (v/v) fetal calf serum and 2 mM L-glutamine and centrifuged at 800 ϫ g for 5 min. The cell pellets were resuspended in FIG. 1. A, domain organization of the recombinant fibrillin-1 protein fragments compared with full-length fibrillin-1. SDS-PAGE analysis of nickel affinity-purified recombinant fibrillin-1 protein fragments under non-reducing (i) or reducing (ii) conditions or after reduction and pretreatment with peptide:N-glycosidase F (iii) (peptide:N-glycosidase F is indicated with an asterisk). B, protein concentration profile of CsCl density-purified microfibrils. (The arrow indicates fractions containing microfibrils.) Diamonds, CsCl concentration; squares, protein concentration. C, dot blot detection of fibrillin-1 from CsCl density-purified protein fractions using the anti-fibrilllin-1 antibody 11C1. D, electron microscopy of CsCl-purified microfibrils taken from fraction 10.
1 ml of supplemented DMEM/HEPES, and the cell density was calculated using a Neubauer hemocytometer. The cell suspension was centrifuged at 800 ϫ g for 3 min, and the cell pellets were resuspended in an appropriate volume of supplemented DMEM/HEPES to give a final cell density of 1 ϫ 10 7 cells/ml. Cells (50 l) were added to FACS tubes followed by 50 l of primary antibody (K20, JBS5, JA218, 12G10, LM609, or mouse IgG control) diluted to 10 g/ml in PBS-2 containing 0.02% (w/v) sodium azide and incubated at 4°C for 1 h. The cellantibody mixtures were centrifuged at 800 ϫ g for 4 min, and the cell pellets were washed three times in PBS-2 (Dulbecco's phosphate-buffered saline (without Ca 2ϩ and Mg 2ϩ )) containing 1% (v/v) fetal calf serum. The resulting cell pellets were resuspended in 50 l of PBS-2 containing FITC-conjugated anti-mouse secondary antibody (1:200 dilution) and 10% (v/v) human serum and incubated at 4°C for 45 min. The antibody-cell mixtures were centrifuged at 800 ϫ g for 4 min, and the cell pellets were washed twice in PBS with cations containing 1% (v/v) fetal calf serum and once with PBS with cations. Cells were fixed by the addition of 100 l of 2% (v/v) formaldehyde followed by 400 l of PBS with cations and incubated at room temperature for 20 min.  anti-mouse IgG-conjugated rhodamine (1:100 dilution) was used. Cell cultures were then visualized using a Leica DM RXA microscope fitted with a UV lamp and appropriate filters. Basic image acquisition and analysis were performed using IP Lab version 3.2. Advanced image analysis was performed using Adobe Photoshop version 6.0.

HDF Attachment and Spreading on Microfibrils-
To determine whether assembled native fibrillin-rich microfibrils support HDF adhesion and spreading, microfibrils were immobilized onto 96-well microtiter plates and used as ligands in cell attachment and spreading assays (Fig. 2). Linear coating efficiencies within the concentration range 0 -20 g/ml were confirmed by enzyme-linked immunosorbent assay, and microscopy analysis confirmed that microfibrils readily adhered to the wells. Highest percentage of cell attachment was obtained using a ligand coating concentration of 2 g/ml, which was then used in all subsequent assays. Fibronectin was used as a positive control and BSA as a negative control. HDF attached strongly to microfibrils ( Fig. 2A) and spread well on microfibrils and on fibronectin (Fig. 2, B and C).
HDF Attachment and Spreading on Recombinant Fibrillin-1 Protein Fragments-To compare HDF attachment to microfibrils and to molecular fibrillin-1, cell adhesion to three overlapping fibrillin-1 recombinant protein fragments (Fig. 1A), two of which contained the RGD sequence, was examined (Fig.  2). The RGD-containing fibrillin-1 recombinant protein fragments PF9 and PF11 supported strong cell attachment, although slightly lower than microfibrils ( Fig. 2A). The non-RGD-containing fragment PF10 also supported some cell adhesion, but this was much lower than PF9 and PF11. The RGD-containing protein fragment PF11 had the highest cell adhesion activity, followed by the RGD-containing protein fragment PF9 and then by peptide PF10. HDF spread rapidly on fibrillin-1 fragments PF9 and PF11 but very little on fragment PF10 (Fig. 2, B and C).
HDF Attachment to Fibrillin-1 Ligands Is Divalent Cationand RGD-dependent-Divalent cations are known to regulate integrin-ligand interactions (25). To establish whether integrin receptors mediate cell adhesion to fibrillin-1 molecules and microfibrils, divalent cation dependence was assessed. For HDF attachment to microfibrils and RGD-containing protein fragments PF9 and PF11, calcium was unable to support attachment, whereas magnesium and especially manganese strongly enhanced binding (Fig. 3A). Addition of cations had no effect on HDF attachment to PF10 (data not shown).
RGD dependence of cell attachment was then examined directly using synthetic peptides to the fibrillin-1 and fibrillin-2 RGD sequences, and antibodies raised to these peptides. Immunoblotting demonstrated high immunoreactivity of each fibrillin isoform antibody for its own peptide but not for the other fibrillin peptide. HDF attachment to microfibrils was inhibited in a dose-dependent manner by both fibrillin-1 and fibrillin-2 RGD antibodies (Fig. 3B) and by synthetic peptides (not shown). HDF attachment to the RGD-containing recombinant fibrillin-1 protein fragments PF9 and PF11 was strongly inhibited by the fibrillin-1 RGD antibody and synthetic peptide but not the corresponding fibrillin-2 RGD reagents (Fig. 3B). The low level of HDF attachment to fibrillin-1 protein fragment PF10 was unaffected by addition of either peptides or antibodies. These studies confirmed that HDF attachment to fibrillin-1 fragments PF9 and PF11 is RGD-dependent and showed that ϳ45% of attachment to microfibrils is RGD-dependent with contributions from both fibrillin-1 and fibrillin-2.
Integrin Receptors Mediating HDF Attachment to Fibrillin-1 Ligands-TheintegrinreceptorsresponsibleformediatingRGDdependent cell attachment to microfibrils were identified in cell attachment-inhibition studies (Fig. 4A). Microfibrils immobilized onto 96-well microtiter plates were used as ligands in inhibition assays of HDF cell attachment, with antibodies to integrin receptors at final concentrations of 20 g/ml (␣ v ␤ 3 (LM609), ␣ 5 (JBS5 and mAb16), ␣ 2 (JA218), and ␤ 1 (mAb13 and K20)). Antibodies JBS5 and mAb 16 (block ␣ 5 integrin subunit) both strongly inhibited cell attachment, suggesting that ␣ 5 ␤ 1 integrin mediates HDF attachment to microfibrils. Addition of JBS5 resulted in 84% inhibition of attachment and of mAb16 led to 97% inhibition. mAb 13, which inhibits ␤ 1 integrins, inhibited cell attachment by 78%. No inhibition was observed using the K20 anti-␤1 antibody, as expected, because this antibody does not block integrin function. A contribution to cell attachment was also shown for ␣ v ␤ 3 , because antibody LM609 inhibited cell attachment by 41%. Integrin ␣ 2 ␤ 1 was not involved, because antibody JA218 did not inhibit cell attachment. Antibody dose-response curves confirmed the specific antibody effects and that ␣ 5 ␤ 1 and ␣ v ␤ 3 integrin receptors are responsible for the RGD-mediated HDF attachment to microfibrils (Fig. 4B).
FACS analysis confirmed the presence of ␣ 5 ␤ 1 and ␣ v ␤ 3 receptors on HDF cells (Fig. 4C). Abundant ␤ 1 integrin was detected as well as ␣ 2 and ␣ 5 subunits and low levels of ␣ v ␤ 3 integrin. The possibility that cell adhesion to ␣ 5 ␤ 1 might be due to endogenous production of fibronectin within the 20-min cell adhesion assay was excluded using several approaches. Sitedirected mutagenesis of the RGD motif in protein fragment PF9 to RDG, which is ineffective in binding integrins, showed that the RGD-dependent cell adhesion activities recorded for these protein fragments are due specifically to the fibrillin-1 RGD motif (Fig. 5, A and B). Moreover, mAb 333, which blocks fibronectin attachment to ␣ 5 ␤ 1 , had no effect on HDF attachment to the fibrillin-1 PF9 and PF11 fragments (Fig. 5C). Immunohistochemistry failed to detect any fibronectin by 20 min, with pericellular fibronectin only detected by 60 min.
Effects of Heparan Sulfate on Cell Adhesion to Microfibrils-Heparan sulfate and heparin have previously been shown to mediate binding to three fibrillin-1 regions (amino acids 40 -450, 1528 -2731, and 1028 -1486), the first two of which are calcium-independent interactions (27). The potential for heparan sulfate to inhibit HDF adhesion to microfibrils was therefore examined. Addition of heparan sulfate resulted in 18% inhibition of cell binding to microfibrils.

Cellular Consequences of Attachment to Fibrillin Ligands-
When the morphology of HDF plated on microfibrils for up to 6 h was compared with that of HDF on fibronectin, pronounced shape differences were observed (Fig. 6). The cells on microfibrils were less extensively spread and had very prominent actin filaments and some ruffled edges. The cells on fibronectin were highly spread with well organized cytoskeletal structure. Immunocytochemical analysis revealed increased fibrillin-1 immunoreactivity when plated on microfibrils relative to cells on fibronectin (Fig. 6). On binding microfibrils, focal complexes containing FAK and paxillin were apparent as previously shown (11), and focal adhesion molecules paxillin and FAK were phosphorylated (not shown). These experiments confirmed that adhesion to microfibrils results in cell signaling, cytoskeletal organizational changes, and extracellular matrix expression.
Adhesion of Smooth Muscle Cells, HT1080 Fibrosarcoma Cells, and A375-SM Melanoma Cells to Microfibrils and Fibrillin-1 Protein Fragments-Adhesion of primary human coronary artery smooth muscle cells, HT1080 fibrosarcoma cells, and A375-SM melanoma cells to microfibrils and fibrillin-1 protein fragments was also examined (Fig. 7). HT1080 cells exhibited 20% cell attachment to PF11 (Fig 7A). This adhesion was strongly inhibited by anti-integrin ␣ 5 antibody mAb16 and anti-integrin ␤ 1 antibody mAb13, but anti-integrin ␣ V␤3 antibody LM609 had little effect on cell adhesion. A375-SM melanoma cells exhibited 65% attachment to PF11 (Fig. 7B). This adhesion was strongly inhibited by anti-integrin ␣ v ␤ 3 antibody LM609, but anti-integrin ␣ 5 antibodies had little effect on cell adhesion. For smooth muscle cells, the percent cell attachment to fibrillin-1 protein fragments and microfibrils was lower than for HDF, but after pre-treatment with the ␤ 1 integrin activating antibody, TS2/16, activated SMC ␤ 1 integrins bound and spread on microfibrils and on fragments PF9 and PF11 (Fig. 7,  C and D). DISCUSSION Fibrillins serve an essential role linking extracellular fibrillin-rich microfibril bundles and elastic fibers with cells of connective tissues such as skin and blood vessels. Loss of cellmicrofibril interactions in Marfan syndrome, a severe heritable disease caused by mutations in fibrillin-1 (28), and in aging following microfibril degeneration may lead to altered cellmatrix interactions, which, in turn, could modify cell behavior. Studies have shown that recombinant fibrillin-1 fragments (12,13) and fibrillin molecules purified using a denaturing protocol (11) can interact with cells, and ␣ v ␤ 3 integrin was identified as the major receptor. An aim of this study was to determine whether connective tissue cells also bind assembled microfibrils and the molecular basis of these interactions. We have shown that cells strongly adhere and spread on microfibrils that, in addition to ␣ v ␤ 3 , ␣ 5 ␤ 1 is an important receptor for microfibrils as well as for recombinant fibrillin-1 molecules, and that RGD-dependent adhesion to fibrillin ligands influences cell shape and ECM deposition.
Although cell adhesion studies generally utilize molecules or peptides as ligands, this study has examined cell adhesion to native assembled fibrillin-rich microfibrils which are very large polymers with a complex bead ultrastructure (1,20,26,29). Using these physiological ligands, we have shown strong cation-and RGD-dependent adhesion of HDF to microfibrils, which is comparable to their adhesion to a 50-kDa fibronectin fragment that contains the central cell binding domain. Because interactions between cells and microfibrils are inhibited by fibrillin-1 and fibrillin-2 RGD peptides, the RGD sequence within the fourth fibrillin TB module must be accessible within assembled microfibrils. Such a location is consistent with pro- posed models of fibrillin alignment in microfibrils, which predict that the RGD motif is within the interbead region (1,20,26,30,31). Strong inhibition of microfibril adhesion was seen with both fibrillin-1 and fibrillin-2 RGD peptide antibodies, which is consistent with fetal aortic microfibrils containing both molecules (3,4). Thus for HDF, RGD-dependent adhesion to microfibrils, as well as to recombinant fibrillin-1 RGD protein fragments, is mediated mainly by 7. A, anti-integrin antibody inhibition of HT1080 cell attachment onto 20 g/ml PF10, PF11, or 50K fibronectin (type III repeats 6 -10). Unshaded bars, no antibody; light gray, mAb16 the inhibitory anti-␣ 5 antibody; mid gray, LM609 the inhibitory anti ␣ V ␤ 3 antibody; dark gray, LM609 and mAb16 combined. Error bars indicate Ϯ S.D. B, anti-integrin antibody inhibition A375-SM melanoma cell attachment onto BSA, PF11, or 50K fibronectin (type III repeats 6 -10). mAb16 is an inhibitory anti ␣ 5 antibody, and LM609 is an inhibitory anti-␣ V ␤ 3 antibody. Unshaded bars, no antibody; light gray, mAb16 the inhibitory anti-␣ 5 antibody; mid gray, LM609 the inhibitory anti-␣ V ␤ 3 antibody; dark gray, LM609 and mAb16 combined. Error bars indicate Ϯ S.D. C, human coronary artery smooth muscle cell adhesion to PF9, PF10, and PF11 in the presence (dark gray bars) or absence (unshaded bars) of the ␤ 1 activating antibody TS2/16. The non-functional ␣ 2 antibody 10A4 (mid gray bars) was included as a control. receptors also bind these ligands. Adhesion to integrin ␣ 5 ␤ 1 is well known to be highly conformation-dependent (32). Smooth muscle cells adhered strongly to ␤ 1 integrins only after antibody activation, indicating cultured cell-specific differences in ␣ 5 ␤ 1 -activated status. A similar effect was noted for bovine auricular chondrocytes (11). Integrin-dependent HDF and smooth muscle cells spreading on fibrillin ligands showed that cell signaling events caused cytoskeletal changes. Prominent differences in cell shape when plated on microfibrils or on fibronectin indicated subtle differences in cellular responses to adhesion to these two RGD ligands even though adhesion was, in both cases, mediated largely by ␣ 5 ␤ 1 integrin. On microfibrils, HDF were partially migratory with some ruffled edges, whereas on fibronectin the cells were well spread. These differences have physiological implications for cell morphology and signaling when adherent to fibronectin or microfibrils in vivo.
Interactions between fibrillin ligands and ␣ 5 ␤ 1 integrin were confirmed using a fibrillin-1 RDG mutant which, as expected, showed ablated cell adhesion. These data allowed us to exclude other ECM molecules such as fibronectin as a potential ␣ 5 ␤ 1 ligand that might have been secreted during the 20-min cell adhesion assays. Moreover, mAb 333, which blocks fibronectin binding to integrins (33), resulted in no reduction of cell adhesion to fibrillin-1, although in fibronectin control experiments adhesion was ablated. Immunofluorescence studies also failed to detect fibronectin expression within the timeframe of the assays. RGD-dependent integrin ␣ 5 ␤ 1 binding to fibronectin is known to involve recognition of a PHSRN synergy site in the ninth fibronectin type III repeat (34). No similar synergy sequence is present within the flanking cbEGF domains in fibrillin-1, so any synergy site in this molecule must be a different sequence or a conformation-dependent epitope. A recent study of ␣ 5 ␤ 1 -mediated binding to fibronectin has highlighted that the RGD domain serves to activate and align the ␣ 5 ␤ 1 -fibronectin interaction, whereas the synergy site provides the mechanical strength to the interaction (32). Because fibrillin microfibrils serve an important mechanical role in connective tissues (35), it will be of interest to determine the molecular basis of their mechanical coupling to cells.
Because RGD-mediated cell adhesion accounted for ϳ50% of cell adhesion to assembled microfibrils, we examined whether microfibrils also interact with heparan sulfate, which, in turn, would suggest interactions with cell surface heparan sulfate proteoglycan receptors. Three heparan sulfate binding sequences have been localized to large regions of fibrillin-1 using recombinant fibrillin-1 peptides (27), but it is not clear whether these sequences are exposed in microfibrils or available for interactions with cell surface receptors. The N-and C-terminal heparan sulfate binding regions, in particular, may be located in the microfibril beads and/or involved in fibrillin-1 assembly (1,20,26,36). Our data show that ϳone-fifth of HDF attachment to microfibrils was inhibited by heparan sulfate. The RGD-independent adhesion to fibrillin-1 exons 41-52 that we have identified was not inhibited by heparan sulfate, so the molecular basis of this adhesion is unknown but could possibly be mediated by an integrin receptor that does not recognize the RGD motif. We have shown that cell adhesion to fetal aortic microfibrils involves interactions with fibrillin-1 and fibrillin-2, but interactions with other potential microfibril-associated molecules shown to interact with cells in vitro may also, if present, contribute. These include tropoelastin, which binds cells via the elastin-binding protein (37), and MAGP-2, which binds cells via ␣ v ␤ 3 (38).
HDF attachment to microfibrils and to fibrillin-1 protein fragments, in addition to promoting cell spreading and migra-tory morphology, also enhanced the extracellular deposition of fibrillin-1 relative to deposition when attached to fibronectin. This effect suggests a positive feedback mechanism for fibrillin-1 expression. The integrin ␣ 5 ␤ 1 plays a central role in pericellular fibronectin assembly, mediating conformational changes that allow linear accretion (39). It will be of interest to determine whether ␣ 5 ␤ 1 -mediated adhesion to newly synthesized molecules also influences fibrillin microfibril assembly.
The physiological importance of fibrillin interactions with cells is clearly highlighted in blood vessel walls. Medial elastic fiber laminae have an outer mantle of microfibrils that is juxtaposed to smooth muscle cells at dense plaques. In a mouse model of Marfan syndrome, microfibril defects resulted in an unusually smooth surface of elastic laminae, loss of cell attachments normally mediated by fibrillin-1, and alterations in smooth muscle cell morphology and expression profiles (40). Subendothelial microfibrils deposited during aortic development interact with endothelial cells at dense plaques (5) and are critical in anchoring endothelial cells to the vessel wall. Juxtaposition of microfibrils and fibroblasts in skin and ligament further confirm the physiological importance of microfibril-cell interactions.
In summary, we have shown that assembled fibrillin-rich microfibrils are important adhesion ligands for connective tissue cells, ␣ 5 ␤ 1 is a major receptor for fibrillin-1 molecules and microfibrils, and RGD-dependent adhesion to fibrillin ligands influences cell phenotype. An implication of this study is that reduced or defective microfibrils in Marfan syndrome may directly alter cellular phenotype.