The Reprolysin Jararhagin, a Snake Venom Metalloproteinase, Functions as a Fibrillar Collagen Agonist Involved in Fibroblast Cell Adhesion and Signaling*

The integrins α2β1 and α1β1 have been shown to modulate cellular activities of fibroblasts on contact with fibrillar collagen. Previously it has been shown that collagen binding to α2β1 regulates matrix metalloproteinase MMP-1 and membrane-type MT1-MMP expression. Jararhagin is a snake venom metalloproteinase of the Reprolysin family of zinc metalloproteinases, containing a metalloproteinase domain followed by disintegrin-like and cysteine-rich domains. Jararhagin blocks type I collagen-induced platelet aggregation by binding to the α2β1integrin and inhibiting collagen-mediated intracellular signaling events. Here we present evidence that, in contrast to the observations in platelets, jararhagin binding to the integrin receptor α2β1 in fibroblasts produces collagen-like cell signaling events such as up-regulation of MMP-1 and MT1-MMP. Inactivation of the metalloproteinase domain had no effect on these properties of jararhagin. Thus, in fibroblasts the snake venom metalloproteinase jararhagin functions as a collagen-mimetic substrate that binds to and activates integrins. Given the homology between the metalloproteinase, disintegrin-like and cysteine-rich domains of jararhagin and those of the members of the ADAMs (a disintegrin-like andmetalloproteinase) family of proteins, this work demonstrates the potential of the disintegrin-like/cysteine-rich domains in the ADAMs as cellular signaling agents to elicit responses relevant to the biological function of these proteins.

Adhesion of fibroblasts to native type I collagen is mediated by ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrin receptors (1,2). Recently, Knight et al. (3) have shown that the sequence GFOGER (O, hydroxyproline) in triple-helical collagen type I and IV is recognized by both ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins. Several studies have localized the binding site for collagen within the I-domain of the ␣-chain integrin subunit. The I-domain is composed of about 200 amino acids and shares homology with the von Willebrand factor A domain (4 -6).
When human dermal fibroblasts are grown in contact with fibrillar collagen type I, a series of events are triggered. Fibroblasts acquire phenotypic tissue-like characteristics that are not observed in fibroblasts grown as monolayer cultures on plastic or on monomeric collagen type I (7,8). When seeded into these loose networks of collagen fibrils, fibroblasts down-regulate type I collagen expression (9), induce MMP-1 1 synthesis (10), and activate pro-MMP-2 (11). Furthermore, it has been shown in fibroblasts that collagen binding to the ␣ 2 ␤ 1 integrin contributes to the reorganization and contraction of the collagenous matrix (12,13) and is responsible for the induction of MMP-1 synthesis (14,15). The down-regulation of type I collagen synthesis in this system was due to collagen binding to the ␣ 1 ␤ 1 integrin (14). Recently, we observed that, in addition to MMP-1, MT1-MMP is also induced on both the mRNA and protein levels by the ligation of the ␣ 2 ␤ 1 integrin receptor with fibrillar collagen (16). The synthesis of the ␣ 2 ␤ 1 integrin was found to be up-regulated in collagen lattices, whereas the expression of other collagen integrin receptors, such as ␣ 1 ␤ 1 and ␣ 3 ␤ 1 , was not affected (13).
The snake venom metalloproteinases (SVMPs) are members of the Reprolysin family (M13) of metalloproteinases. The ADAMs (a disintegrin-like and metalloproteinase)/MDC (metalloproteinase, disintegrin, cysteine-rich) group of proteins are also members of the Reprolysin family (17). The PIII class of SVMPs and the ADAMs share homologous metalloproteinase, disintegrin, and cysteine-rich domains (18). Based on these similarities, the SVMPs have served as early models for ADAM function. The PIII SVMPs have been demonstrated to be capable of proteolytically degrading extracellular matrix and inhibiting platelet aggregation by blocking collagen binding to the ␣ 2 ␤ 1 integrin on platelets (17).
Jararhagin, a hemorrhagic metalloproteinase from Bothrops jararaca, is one of the main venom components responsible for the local and systemic hemorrhage observed in envenomed humans (19). Jararhagin has been shown to degrade components of the basement membrane of the microvasculature and some plasma proteins important for hemostasis (20,21). Furthermore, it synergizes hemorrhage by inhibiting collagenstimulated platelet aggregation (22).
In the venom, the metalloproteinase domain of jararhagin is often proteolytically processed generating jararhagin-C, a frag-* This work was supported by Deutsche Forschungsgemeinschaft Grants KR 558/10-1 and BMFT/IDZ 10 (01 GB 950/4), by W. Sander-Stiftung (99.093.1), and by the Köln Fortune Project (86/1999). 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.
** To whom correspondence should be addressed: Tel.: 49-221-478-5407; Fax: 49-221-478-5949; E-mail: Cornelia.Mauch@medizin. uni-koeln.de. ment representing the disintegrin and cysteine-rich domains of jararhagin. The ␣ 2 ␤ 1 integrin can also interact with jararhagin-C, but the interaction seems to be weaker than with jararhagin (23), suggesting that additional N-terminal structures might be involved in jararhagin-␣ 2 ␤ 1 binding. Interestingly, synthetic peptides based on a sequence in the metalloproteinase domain of jararhagin have been shown to bind to the I-domain of the recombinant ␣ 2 integrin chain thereby preventing the binding of the ␣ 2 -I domain to collagens I and IV, and to laminin-1 (24).
In this study, we have investigated the ability of jararhagin to mimic fibrillar collagen interaction with fibroblasts to modulate the expression of the integrin ␣ 2 ␤ 1 and the matrix metalloproteinases MMP-1 and MT1-MMP. In contrast to previous studies performed in platelets, jararhagin binding to fibroblasts led to cellular activities similar to those induced by fibrillar type I collagen binding via the ␣ 2 ␤ 1 integrin. These results suggest that other disintegrin-like/cysteine-rich domain-containing proteins, such as the ADAMs, may be capable of not only binding to integrins, as has been shown, but also signaling via integrins to alter cellular events such as gene and protein expression.

MATERIALS AND METHODS
Antibodies and Reagents-The following antibodies were used: function blocking mouse monoclonal antibodies directed against the ␤ 1 (4B4; Coulter Corporation), ␣ 2 (P1E6; BIOMOL), and ␣ 3 (BIOMOL) integrin chains; monospecific mouse antibodies to MT1-MMP were raised against a peptide corresponding to the residues 160 -173 of human MT1-MMP (114 -1F2; Fuji Chemicals). Rabbit polyclonal antibodies to human MMP-1 were kindly provided by Dr. P. Angel (Deutsches Krebsforschungszentrum, Heidelberg, Germany). Function-blocking mouse antibodies raised against the human ␣ 1 integrin chain, mouse antibodies against human HLA-ABC, and antibodies directed against the ␤ 1 integrin subunit antibodies used for immunoblotting were from Chemicon. The rabbit polyclonal antibodies directed against jararhagin were a kind gift from Dr. R. D. G. Theakston (Liverpool School of Tropical Medicine, Liverpool, UK). Jararhagin was purified from the venom of B. jararaca as previously described (19). Inactivation of proteolytic activity was performed by a 5-min treatment with 5 mM 1,10phenanthroline at 37°C (25).
Cell Culture Conditions and Immunostaining-Human dermal fibroblasts obtained by outgrowth from explants were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2 mM glutamine, and 100 units/ml each of penicillin and streptomycin. Fibroblasts were used at passages 1-9. Three-dimensional collagen gels were prepared as described previously (9). Briefly, type I collagen (3 mg/ml, Vitrogen) and 10ϫ DMEM were combined in a 10:1 ratio and neutralized by the addition of 0.1 M NaOH. Fibroblasts were seeded into collagen gels at a density of 1 ϫ 10 5 cells/ml and incubated at 37°C. Alternatively, suspensions of fibroblasts were preincubated with jararhagin, 100 nM, for 20 min at 37°C before seeding into collagen gels. Rates of gel contraction were monitored by determining the remaining surface area.
For integrin binding experiments, cells were detached from confluent monolayer cultures by trypsinization, collected by centrifugation, and resuspended in growth medium. Before seeding onto tissue culture plastic plates, the cells were preincubated in the presence of different concentrations of antibodies for 20 min at 37°C. Cell viability was determined by trypan blue exclusion.
To detect ␣ 2 integrin, fibroblasts were cultured on chamber slides. After 48 h, cells were washed twice with phosphate-buffered saline (PBS) and then fixed with cold acetone. Staining was performed with rabbit anti-␣ 2 antibodies against the cytoplasmic domain of the protein (Chemicon) overnight at 4°C in PBS containing 4% BSA, followed by incubation with goat anti-rabbit-Cy3 antibodies (1:400 in PBS/2% BSA, Dianova) for 1 h at room temperature in a humidified chamber. Omission of the first antibody was used as a negative control.
Adhesion Assay-Semi-confluent fibroblast monolayer cultures were trypsinized, and the cells were washed with PBS and resuspended in DMEM containing 0.5% BSA and insulin, transferrin, and sodium selenite at concentrations as recommended by the manufacturer (Sigma). For competition assays, antibodies were added to the cell suspension before plating. Adhesion assays were performed as previously described (26). Briefly, 96-well microtiter plates were coated with 100 l of active or 1,10-phenanthroline-inactivated jararhagin (4 g/ml), monomeric collagen type I (40 g/ml), or 50% FCS at 4°C overnight. BSA coating and blockage of nonspecific binding sites were performed by 1-h incubation with heat-denatured BSA (1% BSA in Ca 2ϩ /Mg 2ϩ -free PBS) at room temperature. After washing the wells twice, cells (2 ϫ 10 4 cells/well) were seeded and incubated for 2 h at 37°C. Non-adherent cells were removed by washing twice with PBS, and adherent cells were fixed with 3% formaldehyde in PBS, pH 7.6, and stained with 0.5% crystal violet in 20% (v/v) methanol. The dye was released from the cells by addition of 0.1 M sodium citrate in 50% (v/v) ethanol. The optical density of the released dye solution was determined at 595 nm. Values were calculated relative to the values obtained for the control assays (FCS or jararhagin pre-coated plates), which were arbitrarily set as 100%. Statistical analysis was performed with the ANOVA Dunnett multiple comparison test.
Binding of Soluble ␣ 2 ␤ 1 to Immobilized Jararhagin-Recombinant soluble human integrin ␣ 2 ␤ 1 ectodomain heterodimers were prepared in insect cells using an expression plasmid in which the cytoplasmic and transmembrane domains were replaced by Fos and Jun dimerization motifs as described previously (27).
Microtiter plates were coated with jararhagin and bovine type I collagen at concentrations of 4 and 40 g/ml in Tris-buffered saline (TBS, 50 mM Tris-HCl, pH 7.4, 150 mM NaCl) containing 3 mM MgCl 2 and 0.1 M acetic acid. After overnight incubation, the wells were blocked with heat-denatured BSA and incubated with 6 g/ml soluble ␣ 2 ␤ 1 integrin in the absence or presence of 10 mM EDTA for 2 h at room temperature. Then the wells were washed twice, and substrate-bound integrin was detected by enzyme-linked immunosorbent assay using a rabbit anti-human ␤ 1 integrin antiserum and alkaline phosphatasecoupled anti-rabbit IgG antibodies as primary and secondary antibodies, respectively. para-Nitrophenylphosphate was used as the enzymelinked immunosorbent assay substrate with the product measured at 405 nm. Each value was measured in duplicate, and standard deviations were calculated.
Preparation of Cell Membranes and Western Blot Analysis-For preparation of crude plasma membranes, cells were washed twice with PBS and scraped off the plates with PBS containing the protease inhibitors aprotinin (10 g/ml), Pefabloc (0.25 mg/ml), and leupeptin (1 g/ml). Cell suspensions were subjected to three cycles of freeze-thaw in a dry ice-ethanol/37°C bath, and cell lysis was confirmed microscopically. Lysates were separated from cell nuclei by centrifugation at 500 ϫ g. Then the supernatant was centrifuged at 7,000 ϫ g for 15 min at 4°C. The crude plasma membranes were washed once with PBS/ inhibitors and, after centrifugation, resuspended in PBS. For preparation of total lysates, cells were washed twice with PBS and lysed in PBS containing 0.5% Nonidet P-40. Lysates were centrifuged at 15,000 ϫ g for 20 min at 4°C. Protein concentration was determined using a commercial assay (Bio-Rad).
For Western blotting, equal amounts of protein from the membrane preparations, lysates, or conditioned media were separated on 10% SDS-polyacrylamide gels under reducing conditions and transferred onto Hybond-C Super™ (Amersham Biosciences). After blockage of nonspecific binding sites with 5% skimmed milk in PBS containing 0.5% Tween (v/v), the blots were incubated with the primary antibodies overnight at 4°C. Bound primary antibodies were detected using a horseradish peroxidase-conjugated secondary antibody (1:2000, Dako) and visualized with the ECL system (ECL™, Amersham Biosciences).
Analysis of Jararhagin Binding to fibroblasts-Purified jararhagin was biotin-conjugated with biotin-XX sulfosuccinimidyl ester. Labeling and purification of biotin-labeled jararhagin were performed using the FluoReporter Mini-Biotin-XX protein labeling kit (Molecular Probes, Leiden, The Netherlands). The ability of jararhagin to displace bound, biotinylated jararhagin from fibroblasts was assayed by incubating fibroblasts in the presence of biotinylated jararhagin for 24 h followed by incubation with varying concentrations of unlabeled jararhagin for an additional 24 h. Cell surface binding of total jararhagin (jararhagin and biotinylated jararhagin) and biotinylated jararhagin after the 48 h treatment was determined by Western blotting of cell lysates prepared by washing twice with cold PBS and direct lysis in reducing sample buffer. After blotting, membranes were incubated with anti-jararhagin antibodies (for visualization of total jararhagin bound) or with extravidin-peroxidase (for visualization of only biotinylated jararhagin) (1: 1000, Sigma) for 1 h and detected as above described.
Zymographic Analysis-Cells were cultured as monolayers with or without jararhagin stimulation. At different time points, media were collected and separated (20 l/lane) on 10% SDS-polyacrylamide gels containing 1 mg/ml bovine gelatin (Sigma). Then gels were washed in 2.5% Triton X-100 for 30 min followed by an overnight incubation in metalloproteinase substrate buffer (50 mM Tris-HCl, pH 8.0, 5 mM CaCl 2 ) (28). Gels were stained with Coomassie Blue R-250 and then destained in water.

Jararhagin Supports Adhesion of Fibroblasts and Inhibits Collagen Lattice Contraction
Fibroblasts showed similar adhesion levels to jararhaginand type I collagen-coated dishes (Fig. 1A). No significant difference was observed between fibroblast adhesion to jararhagin or 1,10-phenanthroline-inactivated jararhagin. As shown in Fig. 1B, cell adhesion to jararhagin was reduced by ϳ30% in the presence of blocking antibodies directed against the ␣ 2 or ␤ 1 integrin subunits; whereas no inhibition was noticed with other function-blocking antibodies directed against ␣ 1 or ␣ 3 integrins, which both can serve as collagen receptors (2). Using a combination of both the ␣ 2 and ␤ 1 antibodies, cell adhesion to jararhagin was inhibited by up to ϳ60% in a dose-dependent manner.
When fibroblasts were pre-treated with jararhagin followed by incubation of the cells within collagen lattices there was a notable delay of lattice contraction (Fig. 2). At 24 h untreated fibroblasts contracted the gels to ϳ30% of the initial surface area. However, in the presence of 100 nM jararhagin, similar levels of contraction were not observed until 48 h. Cell viability was comparable in treated and untreated fibroblast cultures. Because contraction of fibrillar collagen type I lattices has been shown to be mediated by the integrin ␣ 2 ␤ 1 (12)(13)(14), these results suggest that jararhagin delays the contraction by interfering with the ␣ 2 ␤ 1 -collagen interaction.

Jararhagin Interaction with the Integrin Receptor
Previous studies have shown that binding of the ␣ 2 ␤ 1 integrin on platelets by PIII snake venom metalloproteinases results in an inhibition of the signaling events normally induced in collagen-stimulated platelets coupled with a potent inhibition of platelet aggregation (22,23). To determine whether jararhagin binds to cell surface proteins, both supernatants and crude fibroblast membranes were analyzed by SDS-PAGE after 24 and 48 h of incubation with jararhagin (Fig. 3). Immu-FIG. 3. Membrane preparations from fibroblasts bind jararhagin. Fibroblasts were cultured in monolayers (m) for 24 and 48 h in the absence or in the presence of 100 nM jararhagin (ϩJ). After 24 h, media were replaced and fresh jararhagin was added. After the indicated time points media were collected and stored at Ϫ20°C until use. In parallel, the cells were used for membrane preparations as described under "Materials and Methods." 20 l of the conditioned media (A) and 40 g of the crude membrane fractions (B) was resolved on 10% SDS-polyacrylamide gels under reducing conditions. After blotting, specific protein bands were visualized by immunodetection using polyclonal antibodies raised against jararhagin (5 g/ml). The jararhagin-specific bands of 55 and 33 kDa, the latter one presumably representing a degradation product, are marked by arrows. The molecular mass standards are indicated (kDa).

FIG. 1. Jararhagin is a cell-adhesive substrate.
A, fibroblasts were seeded on microtiter plates (2 ϫ 10 4 cells/well) coated with either 4 g/ml jararhagin or 1,10-phenanthroline-inactivated jararhagin, 40 g/ml collagen type I, or 50% FCS as described under "Materials and Methods." BSA-coated wells were included as a negative control. After 2 h, non-adherent cells were removed and the adherent cells were stained with crystal violet after fixation. The bars represent the mean Ϯ S.E. of the optical densities determined after release of the dye from three independent experiments performed in duplicates. The mean obtained for FCS-coated dishes (OD 595 nm of 0.66 Ϯ 0.08) was arbitrarily set as 100%. B, fibroblasts were incubated with monoclonal antibodies raised against the ␣ 2 , ␣ 1 , ␣ 3 , or ␤ 1 integrin chains (2.5 g/ml) or with a combination of antibodies raised against the ␣ 2 and ␤ 1 integrin chains (2.5 or 5 g/ml each). Then the cells were seeded on plates precoated with jararhagin (4 g/ml) for 2 h. HLA antibodies (5 g/ml) were used as a control. The bars represent the mean Ϯ S.E. of the optical densities of the released dye from three independent experiments performed in duplicate. The mean of the optical densities obtained for adhesion to jararhagin-coated wells without treatment (OD 595 nm of 0.5 Ϯ 0.09) was set arbitrarily as 100%. (*, p Ͻ 0.05; **, p Ͻ 0.01.)

FIG. 2. Jararhagin interferes with collagen gel contraction.
Human fibroblasts were grown in collagen gels in the absence (white bars) or in the presence of 100 nM (black bars). At the indicated time points the gel surface area was measured. Contraction is indicated as percentage of the initial gel surface area, which was set arbitrarily as 100%. The results represent the mean Ϯ S.E. of two independent experiments performed in triplicates.
noblotting using a jararhagin-specific antibody detected a band of 55 kDa indicating the presence of jararhagin in the cell culture supernatants. An additional band of ϳ33 kDa likely represents the proteolytic degradation fragment of jararhagin comprising the disintegrin-like/cysteine-rich domains (jararhagin-C) (Fig. 3A). In membranes isolated at 24 h, a weakly stained 55-kDa band corresponding to intact jararhagin was detected. The intensity of this band was significantly increased at 48 h (Fig. 3B). The additional band of 52 kDa might represent unreduced jararhagin or a proteolytically processed form (13). Additionally, these data suggest that jararhagin binding to cell surface proteins on fibroblasts offers protection from proteolytic degradation at the site producing jararhagin-C.
An assay was performed to demonstrate the specificity of binding of jararhagin to the cell surface. In these experiments, unlabeled jararhagin was used to displace biotinylated jararhagin from the cell surface of fibroblasts. As shown in Fig. 4A, unlabeled jararhagin is able to displace the cell-surface-associated biotinylated jararhagin in a concentration-dependent manner.
In Fig. 4B, the soluble ectodomain of the integrin ␣ 2 ␤ 1 binds to immobilized jararhagin. However, in contrast to collagen type I, the binding of the soluble ␣ 2 ␤ 1 receptor to jararhagin did not seem to be dependent upon the presence of divalent cations. This suggests the possibility of one or more different binding sites for jararhagin than that for collagen on the receptor molecule.

Jararhagin Treatment of Fibroblasts Grown in Collagen Gel Cultures Did Not Affect Collagen-induced Synthesis of MMPs
After 24 and 48 h of growth in collagen lattices, total RNA was isolated from fibroblasts pre-treated with 100 nM jararha-gin, and the transcript levels for MMP-1 and MT1-MMP were assessed by Northern blot analysis (Fig. 5). Control fibroblasts grown in collagen gels showed increased transcript levels for MT1-MMP and MMP-1 at 24 h with a further increase at 48 h culture. At both time points, pre-treatment with jararhagin did not result in significant differences of these transcript levels from those observed in the untreated cells. In addition, there was a similar increase in integrin ␣ 2 mRNA level from both untreated and jararhagin-treated fibroblasts. Therefore, pretreatment of fibroblasts with jararhagin had no apparent effect on fibroblasts grown within collagen lattices.

Jararhagin Treatment of Fibroblast Monolayer Cultures Results in Similar Changes as Observed for Fibroblasts Grown in Collagen Lattices
Morphology-Analysis of fibroblast cell morphology following treatment with increasing concentrations of jararhagin showed a characteristic elongated shape with protrusions of cell extensions identical to that reported for fibroblasts grown in collagen gels (9). Untreated fibroblasts maintained their characteristic spindle-like morphology with a flattened cell shape (Fig. 6).
MMP mRNA Expression-In contrast to fibroblast growth in collagen lattices, in which no significant alterations could be detected, in monolayer cultures treatment with jararhagin produced significant differences as shown in Fig. 5. In monolayer cultures, only very low levels of MT1-MMP and MMP-1 transcripts were observed. However, fibroblasts pre-treated with jararhagin displayed a strong induction of MT1-MMP and MMP-1 mRNA expression together with increased ␣ 2 -integrin transcript levels. These increases were apparent at 24 h, and by 48 h the increases were comparable to those obtained with fibroblasts cultured within collagen lattices. In addition, the induction of MMP mRNA levels was found to be concentrationdependent, showing maximal stimulation when the cells were pre-treated with 200 nM jararhagin (Fig. 7).
To test whether the metalloproteinase activity of jararhagin is required for the induction of MMP-1 and MT1-MMP expression, monolayer cultures of fibroblasts were treated with active or 1,10-phenanthroline-inactivated jararhagin (22). As shown in Fig. 8, treatment of fibroblasts with proteolytically inactive jararhagin resulted in no significant differences in MMP-1 and MT1-MMP mRNA levels when compared with fibroblasts treated with active jararhagin. The apparent slight reduction of the mRNA levels observed after treatment with inactivated jararhagin was due to the presence of the general metalloproteinase inhibitor 1,10-phenanthroline as shown by the control fibroblast treatment with 1,10-phenanthroline.
MMP Protein Expression-Western blot analysis of MT1-MMP in crude membrane preparations from cells treated with jararhagin displayed increased levels of the active 60-kDa MT1-MMP protein form at 24 h as well as at 48 h as compared with untreated fibroblasts (Fig. 9A). A low level of an additional immunoreactive protein of 63-kDa band corresponding to the unprocessed zymogen was also detected (33,34). This increase in MT1-MMP production in treated monolayer culture was paralleled by enhanced pro-MMP-2 activation as indicated by the appearance of the 62/59-kDa forms correspondent to the active enzyme (Fig. 9B). Analysis of MMP-1 protein in supernatants of cells treated with jararhagin showed increased inactive MMP-1 forms, 52/57 kDa, as well as appearance of active MMP-1, 42/47 kDa, at both 24 and 48 h treatment (Fig. 9C).
We also assessed the possibility that the proteolytic activity of jararhagin might be directly involved in MT1-MMP and pro-MMP-2 activation. Treatment of cell membranes, prepared from an MT1-MMP-stable expressing cell line containing only the pro-form of MT1-MMP (34) with jararhagin, did not result in activation of pro-MT1-MMP (data not shown). In addition, treatment of media containing pro-MMP-2 with jararhagin also failed to show activation of the zymogen (data not shown). These observations suggest that neither the activation of pro-MT1-MMP nor the activation of pro-MMP-2 was a result of the proteolytic activity of jararhagin.
␣ 2 ␤ 1 Protein Expression-As shown above, jararhagin did not block the ␣ 2 ␤ 1 -induced MMP-1 and MT1-MMP up-regulation in fibroblasts grown in collagen lattices. Surprisingly, in monolayer cultures, the addition of jararhagin resulted in increased transcript levels for the ␣ 2 -integrin subunit and MMP-1 and MT1-MMP indicating an activation rather than inhibition of this integrin receptor. It has been demonstrated that fibroblasts grown in collagen lattices respond with an up-regulation the ␣ 2 ␤ 1 integrin expression, whereas there is no up-regulation when fibroblasts are grown as monolayers (16). Treatment of fibroblasts with jararhagin followed by growth as monolayers caused a significant increase of ␣ 2 ␤ 1 -integrin immunostaining, whereas no specific staining was detected in untreated monolayer cultures (Fig. 10A). Western blot analysis of the ␣ 2 integrin subunit in cells treated with jararhagin displayed slightly increased protein levels at 24 h followed by a significant increase at 48 h treatment (Fig. 10B). The observed increase in ␣ 2 integrin levels displayed on the cell surface may explain the increase in jararhagin binding shown in Fig. 3.
These data indicate that treatment of fibroblasts with jararhagin gives rise to changes in gene expression and cellular phenotype similar to those observed for fibroblasts grown in collagen lattices. Therefore, in this system jararhagin is capable of acting as a collagen lattice mimetic by binding to the ␣ 2 ␤ 1 integrin to activate signal transduction.

DISCUSSION
In vivo, skin fibroblasts are surrounded by extracellular matrix components, including fibrillar collagens. As an approach to the in vivo situation, an in vitro model was established several years ago, which is believed to resemble some aspects of the in vivo environment (9,35). In the in vitro model, fibroblasts are embedded in a three-dimensional matrix consisting mainly of fibrillar type I collagen. The collagenous environment causes ␣ 2 ␤ 1 -mediated intracellular signaling events that result in changes in the expression of a variety of proteins, including MMP-1, MT1-MMP, and ␣ 2 ␤ 1 integrin (10,16,13). Recently, it has been demonstrated that fibroblasts, when adhered within three-dimensional matrices compared with adherence as monolayers, have very different focal adhesion complexes, morphologies, and biological activities (36). These results underscore the concept that cells respond differently to different architectures of the extracellular matrix to which they are adhering.
The snake venom metalloproteinase jararhagin inhibits collagen-induced platelet aggregation by binding to ␣ 2 ␤ 1 integrin, thereby blocking collagen binding and its subsequent cell surface receptor-mediated signaling (18,19). Our interests were in examining whether the PIII SVMP, jararhagin, could interfere with the ␣ 2 ␤ 1 -mediated changes in fibroblasts that are observed upon contact with fibrillar collagen.
We demonstrated that jararhagin does bind to the fibroblast cell surface, and, interestingly, the level of binding increases over time. This can be explained by the increase in cell surface expression ␣ 2 ␤ 1 receptor observed following initial incubation of fibroblasts with jararhagin. We consider the interaction of jararhagin with fibroblasts to occur via one or more specific interactions of jararhagin with the cell surface, because bound labeled jararhagin could be specifically displaced by jararhagin (Fig. 4A).
Fibroblasts binding to immobilized jararhagin was not dependent on the proteolytic activity of the SVMP (Fig. 8). Preincubation of the fibroblasts with antibodies against the subunits of the collagen binding integrins (␣ 2 , ␣ 1 , ␣ 3 , and ␤ 1 ) or combinations of antibodies against the ␣ 2 and ␤ 1 integrin chains blocked binding to a level of ϳ40% of the control (Fig.  1B). One possible explanation for the incomplete inhibition could be that, in addition to ␣ 2 ␤ 1 integrin, other integrins or matrix binding receptors may be involved in fibroblast adhe-sion to jararhagin. This is not necessarily surprising, given the multiple domains of the protein (metalloproteinase, disintegrins-like, and cystine-rich), each of which could be involved in cell surface interactions (37). We demonstrated that direct binding of the ectodomain of recombinant ␣ 2 ␤ 1 to jararhagin occurred in a non-cation-dependent manner (Fig. 4B). Most matrix components, including collagen, bind to the ␣ 2 ␤ 1 integrin in a cation-dependent manner (38). However, recent reports described the presence of a non-cation-dependent site on the I-domain of the ␣ 2 subunit that supports binding to pro-MMP-1 (39). Therefore, jararhagin binds to fibroblasts via interaction with the ␣ 2 ␤ 1 integrin, albeit at a site different from the cation-dependent site, and hence the anti-␣ 2 antibodies we used may not function as effectively to block binding at the non-cation-dependent site on the I-domain. This offers a possible explanation for the incomplete blocking of jararhagin to soluble ␣ 2 ␤ 1 . Several reports have indicated ␣ 2 ␤ 1 integrin as binding partner for jararhagin (18,19,23), and our results corroborate this. However, we cannot exclude that other receptors may also be involved in the binding of jararhagin to the cell surface.
In contrast to fibroblasts grown as monolayers, distinctive changes in gene expression were observed for fibroblasts grown in collagen lattices comprised of fibrillar type I collagen (16). Therefore, we anticipated that treatment of fibroblasts with jararhagin prior to growth in collagen lattices would compete with the collagen fibrils for ␣ 2 ␤ 1 binding and block the typical ␣ 2 ␤ 1 -mediated signaling responses observed for fibroblasts grown in collagen lattices. However, treatment of fibroblasts grown in collagen lattices with jararhagin failed to show any inhibition of the collagen-induced up-regulation of MT1-MMP and MMP-1 transcript levels. Jararhagin treatment did significantly delay collagen lattice contraction, a phenomenon that has been shown to be mediated by the integrin ␣ 2 ␤ 1 (11,13,14). From our data the binding of jararhagin to the ␣ 2 ␤ 1 integrin receptor at the concentrations tested could only partially compete with the native substrate collagen. Other investigators have shown a concentration-dependent inhibition of platelet adhesion to monomeric collagen in the presence of increasing amounts of jararhagin or jararhagin-C (23).
Although collagen binding was partially competed by jararhagin treatment of the fibroblasts (as indicated by the delayed collagen gel contraction), the collagen-induced MT1-MMP and MMP-1 mRNA expression was unchanged. This suggests that in the jararhagin-bound ␣ 2 ␤ 1 integrin population a similar signaling phenomenon was occurring that recapitulates the binding of the integrin to fibrillar collagen. This was further substantiated by the results in monolayer cultures, whereby jararhagin induced responses resembling those observed with fibroblasts grown in collagen lattices (16). These included the induction of MMP-1 and MT1-MMP expression and pro-MMP-2 activation. The apparent increase in MT1-MMP is modest; however, it was sufficient to produce the functional stoichiometry between MT1-MMP, TIMP-2, and pro-MMP-2 such that there is an overall increase in pro-MMP-2 activation as observed by zymography.
Because the effects observed in jararhagin-treated monolayer cultures are similar to those occurring within collagen gels, which are mediated by the engagement of ␣ 2 ␤ 1 integrin, we infer that jararhagin essentially mimics the effects of the physiological ligand. This is corroborated by the finding that jararhagin binds to the soluble ectodomain of the ␣ 2 ␤ 1 integrin.
Although jararhagin could support the binding of the ␣ 2 ␤ 1 ectodomain, it was somewhat surprising that this was not dependent on divalent cations, because EDTA did not inhibit the process. Although most of the ligand-integrin interactions are dependent upon divalent ions (38), there are recent reports suggesting that collagen binding to the ␣ 2 I-domain can occur in the absence of metal ions (40). These authors suggest that collagen binding could occur to the "open," metal-dependent, and "closed," metal-independent, conformations of the I-domain. Additional studies are planned to better characterize the binding site for jararhagin on the ␣ 2 ␤ 1 integrin.
The question as which region or regions of jararhagin is/are responsible for the activities observed on treatment of fibroblasts is unclear. Different regions of PIII SVMPs have been shown to bind to the ␣ 2 ␤ 1 integrin. One of these regions is represented by the ECD motif located in the disintegrin-like domain (41). The second region, the RKKH motif of jararhagin, is located in the metalloproteinase domain (24). A third possible site is the cysteine-rich domain. The recombinant cysteinerich domain of atrolysin A, a PIII SVMP isolated from Crotalus atrox venom, has been shown to inhibit collagen-stimulated platelet aggregation thereby indicating an ability to bind platelet ␣ 2 ␤ 1 integrin (37). Preliminary studies in our laboratory (data not shown) using venom proteins containing only the disintegrin-like/cysteine-rich domains have indicated that these domains can induce similar activities as observed with jararhagin when used to treat fibroblast. Therefore, it is likely that the proteinase domain does not play a significant role in these activities.
Data from Kamiguti and colleagues (25) indicated that inhibition of collagen-induced platelet aggregation occurs by binding of jararhagin to the I-domain of the ␣ 2 integrin subunit on platelets. In contrast, when fibroblasts are treated with jararhagin, activation of the ␣ 2 ␤ 1 integrin receptor was observed as evidenced by the up-regulation of MMP synthesis. Kamiguti and colleagues (22) also showed that inhibition of collageninduced platelets aggregation results in a reduced collagenstimulated phosphorylation of the tyrosine kinase pp72 syk .
In platelets, activation and aggregation induced by collagen depends on the cooperative action of ␣ 2 ␤ 1 , glycoprotein VI, and ␣ IIb ␤ 3 (42), and therefore the binding of jararhagin to ␣ 2 ␤ 1 may not be sufficient for effective signal transduction. Furthermore, binding of jararhagin to platelet ␣ 2 ␤ 1 may preclude the successful binding of collagen to ␣ 2 ␤ 1 and/or glycoprotein VI given the close proximity of these two receptors. These events could lead to the overall effect of inhibiting platelet aggregation. However, the situation with fibroblasts is rather different. The binding of jararhagin to ␣ 2 ␤ 1 integrin is sufficient to induce cellular signaling events essentially identical to that observed for fibrillar collagen. Therefore, it seems that, although different cells may use identical integrins to transduce signals, the presence of other receptors or cell signal pathways give rise to different activities. As shown by our experiments, in platelets jararhagin can engage the ␣ 2 ␤ 1 receptor to block platelet aggregation, yet in fibroblasts jararhagin promotes ␣ 2 ␤ 1 receptormediated signaling to promote the expression of MMP-1, MT1-MMP, and the ␣ 2 ␤ 1 integrin. Therefore, not unexpectedly, when considering the activities resulting from the binding of ligands to signal-transducing receptors, care must be given to fully understand the relationship of the receptor to other cell surface receptors as well as the pathways available for transmitting the signals.
Members of the ADAMs family of Reprolysins have been demonstrated to bind to various integrins (43). For instance, ADAM 12 and ADAM 15 have been shown to bind to the ␣ 9 ␤ 1 integrin through their disintegrin-like/cysteine-rich domains mediating thereby cell-cell contacts (44). In another case, ADAM 23 was shown to bind to the ␣ v ␤ 3 integrin and mediate cell interactions in cells of neural origin (45). Although this event promotes sperm-egg fusion, very little data is available on signal transduction following the ADAM-integrin interaction. Based on the data presented here, we would suggest that future investigations examining the potential of cell signaling following integrin engagement by an ADAM may be a productive path to fully understand the biological activities of the ADAMs.
In summary, our studies have demonstrated that the snake venom metalloproteinase jararhagin is a useful tool for studying the molecular basis of collagen-induced cellular activities in human skin fibroblasts for comparison and contrast with other cell types.