Fibrinogen Is a Ligand for Integrin α5β1 on Endothelial Cells

Previous studies have shown that fibrinogen can associate with endothelial cells via an Arg-Gly-Asp (RGD) recognition specificity. In the present study, we have characterized the specificity of fibrinogen binding to endothelial cells under different cation conditions. Fibrinogen binding to suspended endothelial cells was selectively supported by Mn2+ and was suppressed by Ca2+. The Mn2+-supported interaction was completely inhibited by RGD peptides but not by αvβ3 blocking monoclonal antibodies. In contrast, the interaction was completely blocked by two α5β1 monoclonal antibodies. This interaction was not mediated by fibronectin bound to the integrin; could be demonstrated with purified α5β1; and also was observed with a second α5β1-bearing cell type, platelets. The binding of fibrinogen to α5β1 on endothelial cells in the presence of Mn2+ was time-dependent, specific, saturable, and of high affinity (Kd = 65 nM). By employing anti-peptide monoclonal antibodies, the carboxyl-terminal RGD sequence at Aα 572-574 was implicated in fibrinogen recognition by α5β1. Two circumstances were identified in which α5β1 interacted with fibrinogen in the presence of Ca2+: when the receptor was activated with monoclonal antibody (8A2) or when the fibrinogen was presented as an immobilized substratum. These results identify fibrinogen as a ligand for α5β1 on endothelial and other cells, an interaction which may have broad biological implications.

The luminal surface of endothelial cells (EC) 1 is continuously exposed to a high concentration of plasma fibrinogen (Fg). Disruption of the endothelium results in local thrombus formation, and Fg/fibrin accumulates at such sites of vascular injury. Based upon these proximal relationships, the molecular mechanisms and functional consequences of the interaction of Fg with EC have been topics of considerable interest and investigation (1)(2)(3)(4)(5)(6). Indeed, it has been shown that Fg can induce EC attachment, spreading, and migration (1,7) and can support an angiogenic response (8). Several distinct receptors have been implicated in mediating Fg binding to EC. These include ␣ v ␤ 3 (9, 10), a member of the integrin family of cell adhesion receptors, as well as several non-integrin binding sites (6,11). Transglutaminase-mediated Fg cross-linking to EC also has been demonstrated (12).
␣ v ␤ 3 interacts with Fg via an Arg-Gly-Asp (RGD) recognition specificity; i.e. RGD-containing peptides block Fg binding to this receptor (13). This tripeptide sequence is recognized not only by ␣ v ␤ 3 but also by several other integrins (14 -16), including ␣ 5 ␤ 1 , which serves as a fibronectin (Fn) receptor on EC and many other cell types (17)(18)(19)(20). Fg contains two RGD sequences within its A␣-chain: RGDF at A␣ 95-98 and RGDS at A␣ 572-575. Previous studies have shown that EC adhesion to immobilized Fg is blocked by a monoclonal antibody (mAb) to the carboxyl-terminal peptide containing RGD sequence at A␣ 572-574 (3). Moreover, this adhesion was blocked by mAbs against ␣ v ␤ 3 (9,10). Soluble Fg also binds directly to EC in monolayers or in suspension in a specific, saturable (1) and RGD inhibitable interaction (2).
Recently, we (21) and others (22) have shown that the capacity of purified ␣ v ␤ 3 to bind Fg is regulated by divalent cations: Mn 2ϩ supports Fg binding to the receptor, whereas Ca 2ϩ does not; and, in fact, Ca 2ϩ inhibits Fg binding to ␣ v ␤ 3 in the presence of Mn 2ϩ . This pattern of cation regulation is not unique to ␣ v ␤ 3 ; Mn 2ϩ enhances and Ca 2ϩ suppresses ligand binding to several integrins. For example, recognition of Fn (23,24) and anti-␤ 1 mAb 9EG7 (25) by ␣ 5 ␤ 1 is enhanced and/or induced by Mn 2ϩ and is inhibited by Ca 2ϩ . In view of the pivotal role of cations in regulating integrin specificity, we have re-examined the binding of Fg to EC, anticipating an interaction that would be favored by Mn 2ϩ and would be mediated by ␣ v ␤ 3 . Surprisingly, while Mn 2ϩ enhanced binding, the receptor mediating this interaction was not ␣ v ␤ 3 but was ␣ 5 ␤ 1 . Thus, a novel ligand has been identified for this receptor which may have broad biological implications.

MATERIALS AND METHODS
Purification and Radioiodination of Proteins-Fg was purified from human fresh-frozen plasma by differential ethanol precipitation and ammonium sulfate fractionation (26). Human Fn was isolated (27) and provided by Dr. Tatiana Ugarova (Cleveland Clinic Foundation, Cleveland, OH). The 120-kDa chymotryptic Fn fragment containing the central cell-binding domain was purchased from Life Technologies Inc. Fg and 120-kDa chymotryptic Fn fragment were labeled with Na 125 I (Amersham Life Science Inc.) using a modified chloramine-T method (28). The specific activity of 125 I-Fg ranged from 0.8 to 1.2 ϫ 10 Ϫ6 cpm/ molecule. ␣ 5 ␤ 1 was purified from human placenta as described (23). Briefly, placental tissue extract was applied to an affinity column of the 120-kDa chymotryptic Fn fragment, coupled to Sepharose, and the bound receptor was eluted with GRGDSP.
Fg Binding to Cells in Suspension-For binding of Fg to HUVEC in suspension, the adherent cells were washed twice with 10 ml of 10 mM HEPES, 150 mM NaCl, pH 7.5 (Buffer A), and detached by brief exposure (less than 1 min) to a 0.25 mg/ml trypsin, 0.01% EDTA solution (Clonetics Corp.). After neutralizing the trypsin, the cells were immediately centrifuged at 500 ϫ g for 15 min. Cell pellets were suspended, washed twice with Buffer A, and resuspended in 1 ml of Buffer A containing 1% BSA. HUVEC (1 ϫ 10 6 /ml) were preincubated with or without mAbs or peptides for 30 min at 22°C, and then 125 I-Fg was added and incubated at 37°C in the presence of a selected divalent cation in a final volume of 200 l by continuously rotating the tubes end over end at approximately 6 rpm. All buffers were pretreated with Chelex 100 (Bio-Rad) to remove undesired cations. At selected times, 50 l of the cell suspension were layered on 300 l of 20% sucrose in Buffer A containing 1% BSA in conical polyethylene microcentrifuge tubes (38,39). Bound ligand was separated from free by centrifugation for 2 min in a Beckman Microfuge 12. The tips of the tubes were cut off with a razor blade, and the radioactivity associated with the cell pellet was counted in a ␥-counter. Data were determined with triplicate measurements of each experimental point. The standard deviation of the triplicate measurements was less than 10% throughout the course of these studies. The binding of 125 I-Fg to washed platelets was performed as described previously (38), using prostaglandin E 1 (1 g/ml) and theophylline (1 mM) to maintain the cells in a resting state during isolation and ligand binding assays.
SDS-Polyacrylamide Gel Electrophoresis and Autoradiography-Radioactivity associated with cells was extracted into buffer containing 10 mM Tris-HCl, 2 mM EDTA, 1.25% SDS, 1.5 mM phenylmethylsulfonyl fluoride, 5 mM o-phenanthroline, 0.1% NaN 3 , 1 M leupeptin, 1 M pepstatin, and 100 KIU/ml aprotinin, pH 7.4. Samples were boiled for 2 min and electrophoresed under reducing conditions using the buffer system of Weber and Osborn (40) with a 5% acrylamide gel. Gels were stained with Coomassie Brilliant Blue or dried and subjected to autoradiography.
Solid-phase Ligand Binding Assay-The binding of Fg or the 120-kDa chymotryptic Fn fragment to purified and immobilized ␣ 5 ␤ 1 was performed as described (21). ␣ 5 ␤ 1 (70 g/ml) was diluted 1:35 with Buffer A, immobilized in 96-well microtiter plates at 200 ng/well, and incubated overnight at 4°C. After the plates were blocked with 20 mg/ml BSA in Buffer A, 125 I-labeled ligands were added in Buffer A containing the selected divalent ions at 1 mM concentrations and incubated for 180 min at 37°C. Nonspecific binding was measured by determining the ligand binding to BSA-coated wells at each cation condition, and these values were subtracted from the corresponding values for receptor-coated wells.
HUVEC Adhesion Assay-HUVEC were harvested as described and resuspended in 1 ml of Buffer A. Radiolabeling of HUVEC was performed by incubating with 0.5 mCi of Na 2 51 CrO 4 (DuPont NEN) for 30 min at 22°C. Fg (10 g/ml in Buffer A) was immobilized in 96-well Immulon-3 microtiter plates (Dynatech Laboratories Inc., Chantilly, VA) at 1 g/well and incubated overnight at 4°C. The plates were blocked with 1% BSA (heat-inactivated for 1 h at 56°C) in Buffer A for 90 min at 22°C. 51 Cr-labeled HUVEC cell suspension (7.5 ϫ 10 5 /ml in Buffer A containing 0.1% BSA), incubated with or without inhibitors in the presence of appropriate cations for 30 min at 22°C, was applied to Fg-coated wells and incubated for 60 min at 37°C. After washing four times, bound cells were lysed by adding 2% SDS, and radioactivity was measured in ␤-counter.

Manganese Supports Fibrinogen
Binding to HUVEC-To assess how divalent cations might affect the interaction of Fg with EC, 125 I-Fg was incubated with HUVEC in suspension with 1 mM concentrations of different cations. Binding was measured after 45 min at 37°C. As shown in Fig. 1A, Mn 2ϩ supported Fg binding, whereas Ca 2ϩ or Mg 2ϩ failed to enhance binding above the level observed in EDTA. Since Ca 2ϩ was unable to support binding, we considered whether it might interfere with Fg binding supported by 1 mM Mn 2ϩ as reported for several integrin-ligand interactions (22,24,41). Indeed, Ca 2ϩ did suppress Mn 2ϩ -supported Fg binding; 2 mM Ca 2ϩ inhibited 125 I-Fg binding by 64%. The effects of various Mn 2ϩ concentrations on Fg binding is shown in Fig. 1B. No interaction was observed at Mn 2ϩ concentrations below 20 M, but higher concentrations stimulated binding. Ca 2ϩ did not support binding in the 1 M to 1 mM range. In subsequent experiments, 1 mM Mn 2ϩ was chosen as a concentration which supported extensive Fg binding.
We sought to verify that the HUVEC-bound radioactivity was Fg. The radioactive ligand was bound to the cells for 10 or 45 min, and then the HUVEC lysates were subjected to SDSpolyacrylamide gel electrophoresis under reducing conditions followed by autoradiography (Fig. 2). The radiolabeled Fg gave three distinct bands, at the same positions as the A␣-, B␤-, and ␥-chains stained with Coomassie, and exhibited no detectable bands of higher or lower mobility. After a 10-min incubation, the 125 I-Fg in the HUVEC lysate produced bands at these same positions although the intensity of A␣-chain was reduced relative to the B␤and ␥-chains. This phenomenon also was reported by Dejana et al. (1). We found that the band corresponding to the A␣-chain became still less intense with the longer incubation time, whereas the B␤-chain remained unchanged. Coinciding with the loss of A␣-chain was the appearance of higher molecular weight bands. These results indicate that radioactivity which interacted with HUVEC in the presence of Mn 2ϩ was Fg, although the ligand did undergo time-dependent modification.
Fibrinogen Binds to ␣ 5 ␤ 1 on HUVEC-To establish the specificity of Fg binding to HUVEC in the presence of Mn 2ϩ , several competitors were tested for their ability to inhibit the interaction (Fig. 3A). A 100-fold excess of nonlabeled Fg inhibited the binding of the radiolabeled ligand by more than 80%. Furthermore, a RGD ligand peptide, GRGDSP, inhibited binding by 90%, while the control peptide, GRGESP, had no effect. H12, the dodecapeptide from the carboxyl terminus of the Fg ␥-chain (residues 400 -411), which effectively inhibited Fg binding to ␣ IIb ␤ 3 , had no effect on HUVEC binding of the ligand.
The stimulatory effect of Mn 2ϩ and the inhibition by the RGD peptide is consistent with a role of ␣ v ␤ 3 in Fg-HUVEC interaction. However, the binding was not inhibited by two ␣ v ␤ 3 blocking mAbs (Fig. 3B). MAb 7E3 and mAb LM609, which block Fg binding to ␣ v ␤ 3 (10,30), had no effect on 125 I-Fg binding to HUVEC. The functionality of these mAbs was confirmed by inhibition in HUVEC adhesion to Fg (see below). In considering other candidate Fg-binding sites, we examined the role of ␣ 5 ␤ 1 as this integrin also exhibits RGD recognition specificity and its function is stimulated by Mn 2ϩ (23,24). Indeed, as shown in Fig. 3B, inhibition by anti-␤ 1 mAb JB1a was significant (58%). Moreover, anti-␣ 5 mAb P1D6 and anti-␣ 5 ␤ 1 mAb JBS5 inhibited this interaction by 88 and 91%, respectively. Thus, ␣ 5 ␤ 1 is centrally involved in the Mn 2ϩ supported Fg binding to suspended HUVEC.
We sought to determine if ␣ 5 ␤ 1 could be implicated in Fg binding to another cell type. Platelets were used as a model.
Little 125 I-Fg bound to non-stimulated platelets in the presence of Ca 2ϩ (Fig. 4). This limited binding was inhibited by anti-␤ 3 mAb 7E3, and anti-␣ 5 ␤ 1 mAb JBS5 was without effect. Fg binding in the presence of Mn 2ϩ was three times higher than in the presence of Ca 2ϩ , and now anti-␣ 5 ␤ 1 mAb inhibited this interaction by 47%. The anti-␤ 3 mAb blocked binding by 65% and, in combination with anti-␣ 5 ␤ 1 mAbs, almost totally inhibited the interaction (92%). The number of Fg molecules bound per platelet in the presence of Mn 2ϩ is consistent with the reported number of ␣ 5 ␤ 1 receptors (42).
Fibrinogen Binds to Purified ␣ 5 ␤ 1 -Consistent with a role for ␣ 5 ␤ 1 in the interaction of Fg with HUVEC, binding of Fg to the purified receptor also was demonstrable. Purified ␣ 5 ␤ 1 was immobilized onto the wells of microtiter plates. Using the 120-kDa chymotryptic Fn fragment as a radiolabeled ligand, substantial binding was observed in the presence of Mn 2ϩ but not Ca 2ϩ , entirely consistent with the data of Mould et al. (24), and this interaction was more than 97% inhibited by GRGDSP and an anti-␣ 5 ␤ 1 mAb. Next, 125 I-Fg binding to ␣ 5 ␤ 1 was performed in the same conditions. As shown in Fig. 5, 4.9-fold higher binding was obtained in the presence of Mn 2ϩ than in the presence of Ca 2ϩ . This interaction in the presence of Mn 2ϩ was also completely blocked by GRGDSP (94%) and anti-␣ 5 ␤ 1 mAb (95%) (Fig. 5).
Effect of Fibronectin on Fibrinogen Binding to HUVEC-Fg binds to Fn (43,44), and secreted Fn may enhance HUVEC adhesion and spreading on Fg (45). Thus, Fn bound to ␣ 5 ␤ 1 might mediate the observed Fg binding to HUVEC. In this case, increased occupancy of ␣ 5 ␤ 1 by Fn should enhance Fg-HUVEC interaction. On the other hand, if Fg bound directly to the receptor, then occupancy by Fn should inhibit binding since both ligands bind via a RGD recognition specificity. To distinguish these possibilities, two sets of experiments were performed. First, the effect of Fn on 125 I-Fg binding to HUVEC was assessed. When both were added simultaneously to the HU-VEC, Fn produced marked (Ͼ85%) inhibition of Fg binding (Fig. 6). Second, HUVEC were preincubated with Fn, GRGDSP peptide, or both and then washed extensively before incubation with 125 I-Fg. Preincubation with Fn inhibited but did not enhance 125 I-Fg binding (Fig. 6). When the cells were preincu- bated with GRGDSP or Fn plus GRGDSP and then washed, only minor inhibition of Fg binding was observed (Fig. 6), suggesting that the GRGDSP and unbound Fn were effectively removed by the washing. Since Fn, either free or bound to the cells, did not enhance and did, in fact, inhibit 125 I-Fg binding, Fg appears to interact directly with ␣ 5 ␤ 1 on HUVEC.
Characterization of Fibrinogen Binding to ␣ 5 ␤ 1 on HUVEC-The time course of the binding of Fg to HUVEC via ␣ 5 ␤ 1 in the presence of Mn 2ϩ was examined over a 2-h time course with 125 I-Fg at 40 nM (not shown). Specific binding (inhibitable by 1 mM GRDGSDP) of 125 I-Fg to HUVEC reached a plateau level in 45-60 min. In subsequent experiments, 45 min was selected as the incubation time. The specific 125 I-Fg binding to ␣ 5 ␤ 1 on HUVEC was saturable (Fig. 7A). A Scatchard plot (Fig. 7B) of these data suggested that the interaction could be described by a single class of high affinity binding sites with an apparent dissociation constant (K d ) of 65.4 Ϯ 6.1 nM and with a maximum of 3.15 Ϯ 0.18 ϫ 10 5 binding sites per cell (n ϭ 3). This number of Fg-binding sites is similar to the previously reported number of Fn-binding sites per HUVEC (7.5 ϫ 10 5 ) (20).
Localization of the ␣ 5 ␤ 1 Recognition Sequence in Fibrinogen-The RGD sensitivity of Fg binding to ␣ 5 ␤ 1 suggests that one of the two RGD sequences in the A␣-chain or the carboxyl terminus of the ␥-chain is involved in receptor recognition. MAbs which recognize these sequences were employed to distinguish these possibilities. As shown in Fig. 8, anti-C significantly inhibited (73%) Fg binding to HUVEC, whereas anti-N produced a minimal effect (13%). MAb 4A5 also had no inhibitory effect on this interaction (Fig. 8) although, at the same concentration, this mAb inhibited 125 I-Fg binding to ␣ IIb ␤ 3 by 92% (data not shown). These results suggest that ␣ 5 ␤ 1 on HUVEC interacts with the carboxyl-terminal RGD sequence in its recognition of Fg.
Manganese-independent Recognition of Fibrinogen by ␣ 5 ␤ 1 on HUVEC-Two circumstances were identified in which Fg was recognized by ␣ 5 ␤ 1 in the absence of Mn 2ϩ . First, when Fg binding to HUVEC was tested in the presence of the stimulatory anti-␤ 1 mAb 8A2, an interaction was observed in the presence of Ca 2ϩ . 8A2, at a concentration (5 g/ml) reported to enhance Fn binding to ␣ 5 ␤ 1 (36,46), increased specific 125 I-Fg binding to HUVEC in the presence of 1 mM Ca 2ϩ by approximately 2-fold (not shown). Second, a role of ␣ 5 ␤ 1 in HUVEC adhesion to Fg was demonstrable in the presence of Ca 2ϩ . Adhesion of 51 Cr-labeled HUVEC to immobilized Fg was com- pared in the presence of Ca 2ϩ or Mn 2ϩ (Fig. 9). The number of cells adherent in the presence of Mn 2ϩ was 1.4 Ϯ 0.2-fold (n ϭ 5) higher than in the presence of Ca 2ϩ . Morphological differences also were noted under the two cation conditions: at 60 min, spreading of HUVEC was more prominent in the presence of Mn 2ϩ , whereas most of HUVEC were attached but not spread in the presence of Ca 2ϩ . The effects of inhibitors were tested under each cation condition. In the presence of Ca 2ϩ , GRGDSP peptide inhibited the adhesion of HUVEC to immobilized Fg by 96% (Fig. 9). Anti-␤ 3 mAb 7E3 inhibited this interaction by 75%, consistent with a reported role of ␣ v ␤ 3 in HUVEC adhesion to Fg (9,47). The ␣ v ␤ 3 specific mAb LM609 also inhibited adhesion to a similar extent (not shown). However, anti-␣ 5 ␤ 1 mAb inhibited the interaction by 27% and the combination of anti-␤ 3 and anti-␣ 5 ␤ 1 mAb completely blocked adhesion (Fig. 9). These results suggest that ␣ 5 ␤ 1 contributes to HUVEC adhesion to immobilized Fg in the presence of Ca 2ϩ . In the presence of Mn 2ϩ (Fig. 9), anti-␤ 3 mAb, anti-␣ 5 ␤ 1 mAb, or their combination resulted in minor inhibition although the interaction remained RGD inhibitable. DISCUSSION In the present study, we have characterized the binding of Fg to HUVEC and have examined how divalent cations regulate this interaction. The following conclusions are drawn from these analyses. First, the binding of Fg to HUVEC in suspension is selectively supported by Mn 2ϩ and is suppressed by Ca 2ϩ . Second, the receptor on HUVEC which mediates this interaction is ␣ 5 ␤ 1 , and the capacity of this integrin to recognize Fg is not restricted to HUVEC. Third, the binding of Fg to purified ␣ 5 ␤ 1 also was demonstrable. Fourth, the functional region in Fg to interact with ␣ 5 ␤ 1 on HUVEC is the carboxylterminal RGD sequence of the A␣-chain. Fifth, ␣ 5 ␤ 1 can recognize Fg in the presence of Ca 2ϩ when activated with mAb 8A2 or when Fg is presented as an immobilized substrate.
Integrin ␣ 5 ␤ 1 is a major receptor for Fn (17,48). Mould et al. (24) found that Mn 2ϩ supported Fn binding, that Ca 2ϩ did not, and that Ca 2ϩ strongly inhibited Mn 2ϩ -supported ligand binding. These investigators suggested that Ca 2ϩ and Mn 2ϩ recog-nize different cation-binding sites. This concept is endorsed in a study of osteopontin binding to ␣ v ␤ 3 by Hu et al. (41), who showed that Ca 2ϩ decreased the association rate of Mn 2ϩsupported ligand binding but had no effect on the dissociation rate, again suggesting that Mn 2ϩ and Ca 2ϩ bind to different sites in this integrin. Moreover, recognition of Fg by ␣ v ␤ 3 exhibits a similar pattern of cation-controlled recognition; i.e. Mn 2ϩ supports binding and Ca 2ϩ suppresses the interaction (21,22). Thus, the pattern of cation regulation of the Fg-␣ 5 ␤ 1 interaction is consistent with the binding of this ligand to other integrins (e.g. Fg-␣ v ␤ 3 ) and of other ligands to this integrin (e.g. Fn-␣ 5 ␤ 1 ). Although this interpretation is supported by our data and the literature, a differential effect of Mn 2ϩ and Ca 2ϩ on the conformation of Fg cannot be entirely excluded as Fg contains three calcium-binding sites (49,50).
Several Fg receptors on HUVEC have been identified, but our data appear to be the first to implicate ␣ 5 ␤ 1 . Transglutaminase activity has been implicated in Fg binding to HUVEC (12), and we did find evidence of A␣-chain cross-linking in the bound ligand. However, Ca 2ϩ suppressed Fg binding but optimally supports transglutaminase activity (12), suggesting that transglutaminase activity is not sufficient for Fg binding. Based upon our gel analyses, it is reasonable to propose that Fg binds to the cells and then its A␣-chains become cross-linked to one another or to other EC-associated proteins such as Fn. HUVEC adhesion to immobilized Fg involves ␣ v ␤ 3 (9, 10); our data support these previous studies but we also observed a contribution by ␣ 5 ␤ 1 in this adhesion. It is unclear why ␣ v ␤ 3 did not contribute to soluble Fg binding in the present study. Ligand binding to ␣ v ␤ 3 is controlled by the activation state of the receptor (51, 52) which may vary among HUVEC preparations. With respect to other Fg receptors on HUVEC, the expression of intercellular adhesion molecule 1, also shown to be a Fg receptor (6), is also dependent upon activation of EC and is expressed at very low levels on our unstimulated HUVEC preparations. Moreover, the nature of the ligand also may be influential. ␣ v ␤ 3 on HUVEC will only recognize soluble vitronectin when the ligand is presented in multivalent form (53). Fg also may exist in multiple forms which may influence its recognition by ␣ 5 ␤ 1 and ␣ v ␤ 3 . It will be important to consider whether ␣ 5 ␤ 1 can distinguish soluble Fg, immobilized Fg, or fibrin and plasmic degradation products of Fg/fibrin. Precedence exists for differential recognition of various forms of Fg and other ligands by integrins (54). Thus, the availability of specific receptors on the luminal surface, the activation states  9. Adhesion of HUVEC to immobilized Fg. Radiolabeled HUVEC were allowed to adhere to Fg-coated microtiter wells (1 g per well) for 60 min at 37°C in the presence of 1 mM Ca 2ϩ or 1 mM Mn 2ϩ . Prior to adhesion, cells were incubated with GRGDSP peptide (1 mM), anti-␤ 3 mAb 7E3 (20 g/ml), anti-␣ 5 ␤ 1 JBS5 (1/100 dilution), or control mouse ascites (Ctrl) (1/40 dilution) for 30 min at 22°C. After washing, bound cells were recovered by adding 2% SDS, and radioactivity was measured in ␤-counter. Adhesion was expressed as percent of the control without inhibitor. The data shown are means and S.D. from three to five experiments. of the EC and these receptors, the occupancy of these receptors by competing ligands, and the nature of the ligands may determine which receptors will mediate Fg recognition by HUVEC.
Dejana et al. (45) have shown that Fn secreted from HUVEC enhances HUVEC adhesion and spreading on Fg. This finding may suggest that the interaction of Fg with HUVEC is indirect and Fn mediates Fg-HUVEC interaction as a bridging molecule. We concluded that Fg binding to ␣ 5 ␤ 1 on HUVEC in suspension was not mediated by Fn because Fg-HUVEC interaction was not enhanced but was suppressed by Fn. We also found that mAb directed to the carboxyl-terminal RGD sequence of the A␣-chain of Fg significantly inhibited Fg binding to HUVEC. This mAb is specific for Fg and does not cross-react with other adhesive proteins, including Fn (3). Moreover, we showed direct binding of Fg to purified ␣ 5 ␤ 1 using a solid-phase ligand binding assay. These results indicate that Fg can bind directly to ␣ 5 ␤ 1 on HUVEC.
Two circumstances were identified in which Fg interacted with ␣ 5 ␤ 1 in the absence of Mn 2ϩ : when the receptor was activated by 8A2 or when Fg was immobilized. MAb 8A2 is directed to the ␤ 1 subunit (55) and stimulates the binding of multiple ligands to multiple ␤ 1 integrins (36,46). The adhesion of HUVEC to immobilized Fg was partially blocked by anti-␣ 5 ␤ 1 and completely inhibited by the combination of anti-␣ 5 ␤ 1 and anti-␣ v ␤ 3 mAb in the presence of Ca 2ϩ , implicating both receptors in HUVEC adhesion to Fg in the presence of Ca 2ϩ . This interaction was only partially inhibited by these mAbs in the presence of Mn 2ϩ . Since this interaction remained fully inhibited by RGD, one interpretation of this observation is that integrin(s) other than ␣ 5 ␤ 1 and ␣ v ␤ 3 may become involved in Fg recognition. ␣ v ␤ 1 and ␣ v ␤ 5 are potential candidates as both are known to be expressed on HUVEC (56). An alternative explanation is that Mn 2ϩ may increase the affinity of ␣ 5 ␤ 1 and ␣ v ␤ 3 for Fg to an extent that the mAbs cannot effectively compete for the receptors. The greater spreading of HUVEC on Fg in the presence of Mn 2ϩ is consistent with either explanation. Thus, in addition to physiologic and pathophysiologic conditions that elevate Mn 2ϩ concentrations (57), Fg may interact with the receptor when ␣ 5 ␤ 1 is present in an appropriate activation state and/or when Fg is presented in an appropriate conformation.
There are a wide variety of physiologic and pathophysiologic circumstances in which Fg and ␣ 5 ␤ 1 bearing cells come into close contact. An illustrative example is wound healing. In vascular injury, ␣ 5 ␤ 1 -bearing EC come in direct contact with Fg/fibrin, and ␣ 5 ␤ 1 -Fg interactions may facilitate re-establishment of a nonthrombogenic EC monolayer on the luminal wall of the blood vessel. In a healing cutaneous wound, granulation tissue gradually accumulates as fibroblasts migrate into the fibrin clot that initially fills the wound. Since the clot matrix also contains Fn (58), ␣ 5 ␤ 1 may facilitate migration into the wound through interactions with two ligands: Fn, perhaps the principal ligand, and Fg/fibrin, a new potential ligand identified here. Furthermore, the ␣ 5 ␤ 1 -Fg interaction, in addition to providing another method of attachment to Fg, may trigger unique intracellular signaling pathways. Thus, the recognition of Fg by ␣ 5 ␤ 1 demonstrated in this study may have broad biological implications and may be subject to complex control mechanisms.