Incorporation of Vitronectin into Fibrin Clots EVIDENCE FOR A BINDING INTERACTION BETWEEN VITRONECTIN AND (cid:1) A/ (cid:1) (cid:1) FIBRINOGEN*

Vitronectin is an abundant plasma protein that regulates coagulation, fibrinolysis, complement activation, and cell adhesion. Recently, we demonstrated that plasma vitronectin inhibits fibrinolysis by mediating the interaction of type 1 plasminogen activator inhibitor with fibrin (Podor, T. J., Peterson, C. B., Lawrence, D. A., S., Shaughnessy, S. G., D. M., Butcher, and J. (2000) J. Biol. Chem. 275, 19788–19794). The current studies were undertaken to further examine the interactions between vitronectin and fibrin(ogen). Comparison of vitronectin levels in plasma with those in serum indicates that (cid:2) 20% of plasma vitronectin is incorporated into the clot. When the time course of biotinylated-vitronectin incorporation into clots formed from 125 I-fibrinogen is monitored, vitronectin

Vitronectin is a multifunctional plasma glycoprotein that participates in the regulation of coagulation, fibrinolysis, and the complement cascade (reviewed in Refs. 1 and 2). Vitronectin also regulates cell adhesion and pericellular proteolysis on surfaces of cells and extracellular matrices (1)(2)(3)(4)(5). Like fibrinogen, vitronectin is found in plasma at micromolar concentrations (6), and is stored in megakaryocyte and platelet ␣-gran-ules (7)(8)(9). In plasma, vitronectin circulates as a native, monomeric form that is a mixture of 72-kDa single-chain and two-chain disulfide-linked species (10 -12). Under normal conditions, less than 3% of plasma vitronectin is comprised of more reactive oligomeric forms that display enhanced affinity for heparin or heparin-like molecules, and for the conformationsensitive monoclonal antibody 8E6 (10 -12).
During acute phase response, plasma vitronectin levels increase (6), with a relative increase in the percentage of oligomeric vitronectin (13). Levels of the oligomeric forms of vitronectin in serum relative to plasma also increase; indicating the process of coagulation alters vitronectin structure and function (10 -12, 14). The altered, oligomeric form of vitronectin is generated, at least in part, by interactions with other plasma proteins such as thrombin-antithrombin complexes (11,12,14,15, and complement C5b-9 complexes (11,12,16).
Oligomeric vitronectin and vitronectin⅐PAI-1 complexes accumulate in atherosclerotic plaques and at sites of vascular injury, tissue damage, or inflammation (1, 2, 28 -32). Immunohistochemical studies have localized vitronectin along fibrin strands of thrombi formed in vivo (33), and in vitro (34,35), suggesting that vitronectin binds to fibrin. Recently, we demonstrated that fibrin-associated vitronectin mediates the cooperative multivalent binding of PAI-1 to fibrin (35). Moreover, using confocal immunofluorescence microscopy, we found PAI-1 co-localized with vitronectin on the fibrin fibrils of plasma clots. A unique organization of the fibrin-associated Vn-PAI-1 distribution revealed insights into the nature of the vitronectin-binding sites on fibrin. Thus, we observed the fibrin-associated vitronectin and PAI-1 to be distributed in periodic aggregates along fibrin fibrils and at sites of fibril branching. This conspicuous pattern of staining is suggestive of a vitronectin interaction with the fibrinogen ␥Ϫchains, and particularly the ␥A/␥Ј variant form of fibrinogen that represents about 5-10% of the total fibrinogen in plasma (36). The fibrinogen ␥A/␥Ј arises from alternative mRNA processing, and differs structurally in that the carboxyl-terminal sequences 408 -411 of the ␥A/␥Ј chain are replaced in the ␥A/␥ Ј variant by an anionic 20 residue sequence.
The purpose of this current study was to further characterize the interactions between vitronectin and fibrin by: (a) comparing total vitronectin antigen levels in plasma with those in serum, (b) quantifying the binding of vitronectin to purified fibrin, and (c) pursuing a morphological analysis of vitronectin in clots formed from plasma or purified fibrinogen. We present evidence that vitronectin associates with fibrin clots due to its preferentially binding to the carboxyl-terminal ␥Ј chain of the fibrin(ogen) ␥A/␥Ј chain variant.
Vitronectin Concentrations in Platelet-poor Plasma (PPP) and Serum-Blood was collected from 6 healthy volunteers into 0.1 volume of 0.13 M trisodium citrate. After centrifugation at 1800 ϫ g for 15 min at 4°C, PPP was harvested and stored in aliquots at Ϫ70°C until needed, whereas serum was prepared by adding 20 mM CaCl 2 (final concentration) and allowing the samples to clot for 2 h at 37°C. Prior to quantifying plasma and serum levels of vitronectin using an immunoassay for vitronectin (6), all samples and vitronectin standards were dialyzed against 6 M urea for 18 h at 23°C to denature vitronectin (37)(38)(39), and diluted with PBS containing 3% BSA, 0.1% Tween 80, 5 mM EDTA, 20 units/ml aprotinin, and 0.05% sodium azide (Dilution buffer). The standard curve is linear at vitronectin concentrations ranging from 2.0 to 125 ng/ml (r ϭ 0.982), and the inter-and intra-assay coefficients of variation are 13.5 and 8.4%, respectively. As controls, plasminogen, fibrinogen, and albumin concentrations in plasma and serum also were measured. Plasminogen and fibrinogen concentrations were quantified using the IL Test TM kit (Instrumentation Laboratory Co., Lexington, MA) and STA TM assay (Diagnostica Stago, Asnieres-Sur-Siene, France), respectively, on an Automated Coagulation Analyzer (model MDA-180, Organon Teknika Inc., Scarborough, ON). Albumin was quantified with the Spectrum TM kit using an EPX Analyzer (Abbott Laboratories Ltd., Mississauga, ON).
Immunoblotting for Vitronectin in Solubilized PPP Clots-PPP clots were extensively washed with TBS, and solubilized in 0.2 ml of 1 ϫ Laemmli sample buffer comprised of 50 mM Tris, pH 6.8, 2% SDS, 10% glycerol, and 0.001% bromphenol blue in the presence or absence of N-ethylmalamide and 5% 2-mercaptoethanol (40). After boiling for 1 h, the undissolved material was sedimented by centrifugation, and 0.1-ml aliquots of the supernatants were subjected to SDS-PAGE analysis on 7.5% slab gels. Separated proteins were transferred to nitrocellulose membranes, and after blocking with 1% casein, 0.05% Tween 20 in PBS (Blotting Buffer). The membranes were incubated with 5 g/ml of biotinylated SAHVn IgG, washed with Blotting Buffer, and developed with alkaline phosphatase-conjugated streptavidin, and then the substrate p-nitrophenyl phosphate.
Isolation of Vitronectin from Human Plasma-Native vitronectin was isolated from PPP by affinity chromotography using SAHVn IgG coupled to Affi-Gel affinity resin (Bio-Rad) (8,35). For some experi-ments, native vitronectin was converted to the oligomeric form by treatment with 6 M urea in phosphate-buffered saline (PBS), pH 7.4, at 37°C for 1 h, followed by dialysis against PBS. Vitronectin preparations were subjected to PAGE analysis in the absence or presence of SDS, and conformational changes were assessed by measuring their relative affinities for heparin-Sepharose and for the conformation-dependent monoclonal antibody 8E6. Fibrinogen was rendered vitronectin-free by immunoaffinity chromatography using immobilized SAHVn IgG and was characterized as described (35).
Preparation of Labeled Fibrinogen and Vitronectin-Fibrinogen trace-labeled with Na 125 I using Iodo-Beads (specific activity ϳ100 Ci/ mg) was Ͼ95% clottable (41). Native vitronectin was biotinylated (8), or labeled using 125 I-Bolton-Hunter reagent (ICN Biomedicals, Mississauga, ON) to a specific activity of ϳ200 Ci/mg (42). Radiolabeled native vitronectin was converted to the oligomeric form by treatment with 6 M urea.
Biotinylated Vitronectin Incorporation into Purified Fibrin Clots-Purified fibrin clots were formed at 37°C from a TBS buffer solution (0.2 ml) containing 3 M fibrinogen, 100,000 cpm of 125 I-fibrinogen, 150 nM biotinylated native vitronectin, and 2 units/ml thrombin. At intervals, D-Phe-Pro-Arg-chloromethyl ketone (PPACK) was added to 0.05 mM, and fibrin was pelleted by centrifugation at 16,000 ϫ g for 20 min. The radioactivity in the fibrin clot supernatant were measured in a ␥-counter, whereas the amount of biotinylated-vitronectin in the clot supernatant at each time point was quantified relative to the starting solution at time 0 by titrating the samples with Dilution buffer, followed by incubation on SAHVn IgG-coated microtiter wells for 1 h. After washing, bound biotinylated-vitronectin was detected using streptavidin-conjugated alkaline phosphatase, and the hydrolysis of the p-nitrophenyl phosphate chromogenic substrate was monitored at 405 nm on a Microtek plate reader.
Incorporation and Diffusion of 125 I-Vitronectin within Purified Fibrin Clots-Purified fibrinogen (3 M) was clotted with thrombin (2 units/ml) around inoculation loops (41) in the presence of increasing concentrations of 125 I-vitronectin in TBS containing 0.01% Tween 20. Clots were formed in the absence or presence of 10 mM Ca 2ϩ and Factor XIII (10 g/ml) to examine the effects of cross-linking on vitronectin incorporation. After incubation for 2 h at 37°C and washing, 125 I-vitronectin incorporation into the clots was determined by counting the radioactivity of the clots and the supernatant. Clots were then solubilized with 0.2 ml of Laemmli sample buffer in the absence or presence of 5% 2-mercaptoethanol. After boiling for 1 h, half of each sample was subjected to SDS-PAGE (7.5% polyacrylamide gel), and the dried gels exposed to autoradiography film.
The rate of 125 I-vitronectin diffusion out of the clots was measured by incubating clots formed in the absence of factor XIII in a 50-ml conical centrifuge tube containing 5 ml/clot of buffer consisting of either 20 mM Tris, 150 mM NaCl, 0.01% Tween 20, pH 7.4, or the same buffer containing 2 M NaCl. All conditions were performed in triplicate. At time 0, 0.5 ml of the bathing buffer was put into four test tubes and clots added to three of these. All samples were counted, and then all clots and bathing buffer returned to the 50-ml tubes. The clots were monitored in the same fashion for various times, and the radioactivity in the bathing buffer was subtracted from the average radioactivity of clots, and the clot radioactivity at each time point was then expressed as a percentage of their initial radioactivity.
Binding of 125 I-Vitronectin to Purified Fibrin-50-l aliquots of varying concentrations of a fibrinogen solution were added to wells of 96-well plates, and fibrin matrices were formed by clotting the fibrinogen with 1.0 units/ml thrombin and 10 mM CaCl 2 . After incubation for 3 h at room temperature, the plates were stored at 4°C until use. To block nonspecific binding, 100-l aliquots of PBS containing 3% BSA and 0.05% Tween 20 were added to wells for 2 h at 37°C, and after washing, increasing concentrations of 125 I-vitronectin in Dilution buffer were added. After incubating for 1 h at 37°C, the plates were washed, and the bound 125 I-vitronectin quantified. To examine the specificity of binding, experiments were repeated in the presence of a 10-fold molar excess of unlabeled vitronectin. Control experiments were done using BSA-coated microtiter wells, and the 125 I-vitronectin to BSA-coated wells was subtracted from that bound to fibrin to control for nonspecific binding.
Confocal Microscopic Image Analysis-For fluorescence confocal laser microscopic analysis of vitronectin distribution in plasma and purified fibrin clots, 150 l of PPP or purified fibrinogen (3 M) were placed on APTEX-coated coverslips and clotted with 1 unit/ml of thrombin and 10 mM CaCl 2 . To directly visualize the fibrin fibrils, a 1:20 molar ratio of FITC-conjugated fibrinogen (Molecular Bioprobes) per mole of native fibrinogen was added. Samples also were prepared with various concentrations of biotin-labeled vitronectin or BSA prior to clotting. After incubation at 37°C for 2 h, the biotinylated proteins were detected by incubation for 20 min with streptavidin-conjugated Texas Red rhodamine. After a brief rinsing, the clots were mounted using Permafluor mounting medium, and then visualized using a Zeiss LSM 510. Dual wavelength images were acquired using an argon ion laser (488 nm excitation), a helium/neon ion laser (543 nm excitation) and two matched long pass barrier filters for FITC (515-525 nm emission) and TxR (575-640 nm emission) images. Immunofluorescence detection of PAI-1 in clots formed from normal plasma was conducted as previously described (35).

Coagulation-induced Reduction in Plasma Vitronectin Levels Reflects Its Incorporation into the Plasma Clot Matrix-Initial
experiments were undertaken to determine whether clotting of plasma influences vitronectin antigen levels. Vitronectin, plasminogen, fibrinogen, and albumin antigens were quantified, and levels in plasma and serum from six healthy volunteers were compared (Fig. 1A). Affinity purified sheep anti-vitronectin IgG (SAHVn IgG) was used as a capture antibody in the vitronectin immunoassay. This antibody, like the conformationally sensitive mAb 8E6, preferentially binds to the oligomeric, heparin-binding form of vitronectin. Because samples can contain mixed vitronectin conforms, all the samples were denatured with urea to convert the vitronectin to its oligomeric form. The mean vitronectin concentration in serum was ϳ20% lower than that in plasma (paired t test; p Ͻ 0.05). Likewise, plasminogen and fibrinogen levels in serum are 40 and 98% lower in serum than in plasma. In contrast, the level of albumin in serum and plasma is similar.
To determine whether the lower level of vitronectin in serum relative to plasma reflects vitronectin incorporation into the clot, equal volumes of plasma, serum, or solubilized plasma clots were subjected to SDS-PAGE, followed by immunoblot analysis and scanning densitometry. Samples fractionated under nonreducing conditions confirmed the presence of clot-associated native vitronectin and vitronectin multimers with electrophoretic mobilities similar to those of vitronectin in plasma and serum (Fig. 1B, arrows a and b). Under reducing conditions, multimeric vitronectin dissociates into 72-and 62-kDa vitronectin subunits (Fig. 1B, arrows C). The presence of N-ethylmalamide had no effect on the mobility of vitronectin, indicating that the change in the apparent M r of vitronectin after clotting was not the result of disulfide-mediated binding interactions. Thus, these experiments suggest that vitronectin binds to the clot matrix.
Time tion of 125 I-fibrin(ogen) into the clots was mirrored by an increase in the level of biotinylated-vitronectin incorporation ( Fig. 2A). 125 I-Fibrinogen incorporation reaches a plateau by 30 min, with 50% of the maximum incorporation achieved at ϳ12 min. In contrast, vitronectin incorporation reaches a plateau within 7.5 min, with 50% of the maximum vitronectin incorporation occurring at Ͻ2.5 min. Thus, the rate of vitronectin incorporation into the clots appears to exceed the rate of fibrin incorporation in the initial 5-min period so that the vitronectin: fibrin ratio is initially over 1 ( Fig. 2A, left of dashed line).
The non-reduced SDS-PAGE gels in Fig. 2B suggest that, as early as 1 min after the addition of thrombin, vitronectin multimers and monomers are incorporated within the solubilized fibrin clots. A portion of the oligomeric vitronectin is too large to enter the separating gel (Fig. 2B, arrow a). After reduction, the majority of fibrin-associated vitronectin oligomers dissociate into native subunit forms, although there is evidence of residual vitronectin that migrates with the same apparent M r as vitronectin dimers (Fig. 2B, arrow b), as well as thrombincleaved forms of vitronectin (Fig. 2B, arrow c). The rate of 125 I-fibrinogen incorporation into clots was similar in the absence or presence of vitronectin, indicating that vitronectin does not influence fibrin(ogen) incorporation (not shown).
Vitronectin Incorporation into Fibrin Clots Is Not Factor XIII-dependent-Studies were next undertaken to determine whether factor XIII-mediated cross-linking influences this interaction because both fibrinogen and vitronectin contain potential factor XIII-mediated cross-linking sites (43,44). Clots were formed with 3 M fibrinogen in the presence of increasing concentrations of 125 I-labeled vitronectin, and in the absence or presence of factor XIII and calcium. Vitronectin incorporation into clots was similar in the absence or presence of activated factor XIII (Fig. 3). Moreover, the biphasic nature of the doseresponse curves is consistent with the presence of two binding interactions between vitronectin and fibrin.
Although the binding isotherms in Fig. 3 indicate that factor XIII cross-linking activity does not govern the quantity of vitronectin incorporated into fibrin clots, it is still possible that factor XIII could mediate vitronectin cross-linking to fibrin. To address this issue, we analyzed clots and clot supernatants by SDS-PAGE analysis followed by autoradiography (Fig. 3, inset). Native 125 I-labeled vitronectin consists of predominately monomeric vitronectin forms (Fig. 3, inset, lane 1, arrow c). After clotting in the presence or absence of factor XIII, the 125 I-labeled vitronectin in reduced samples of clot supernatants and clot extracts migrates predominately as native vitronectin (Fig. 3, inset, lanes 2-5, arrow c), although small amounts of oligomeric vitronectin are also detected (Fig. 3, inset, lanes 2-5,  arrow b). Minor amounts (Ͻ5%) of cross-linked vitronectin multimers that remain in the stacking gel are found in the extracts of clots formed in the presence of factor XIII (Fig. 3, inset, lane 5, arrow a), but not in its absence.
Vitronectin Incorporation into the Fibrin Clot Is Specific-To exclude the possibility that vitronectin is trapped in the fibrin clots, we compared the rates of 125 I-labeled vitronectin diffusion from clots with that of albumin and thrombin. Over 95% of the 125 I-labeled ovalbumin, a protein that does not bind to fibrin, diffuses out of the clot by 1 h (Fig. 4B). In contrast, at least 50% of the 125 I-labeled vitronectin remains clot-associated at 2 h even in the presence of high salt (Fig. 4A). Thrombin, a protein known to bind to fibrin, also remains clot-associated, but unlike vitronectin, thrombin diffusion is accelerated in the presence of 2 M NaCl (Fig. 4C).
Our studies thus far demonstrate that vitronectin binds to fibrin during the process of fibrinogen polymerization. To further characterize the vitronectin-fibrin(ogen) interaction, we quantified the binding of a fixed concentration of 125 I-labeled vitronectin (100 nM) to microtiter wells coated with increasing concentrations of fibrin (0.09 -6.0 M). Fig. 5A illustrates that both native (monomeric) and urea-treated (oligomeric) vitronectin conforms bind to fibrin in a dose-dependent, saturable fashion. Oligomeric vitronectin binds with an estimated B max of 1.1 nM, and a K d of 0.52 M, while the native vitronectin conform binds fibrin with an estimated B max of 0.9 nM, and a K d of ϳ0.61 M, indicating similar binding affinities for the two forms. The stoichiometry between vitronectin:fibrin is difficult to quantify in these experiments with pre-formed fibrin because the homogeneity, and exposure of binding sites in the microtiter plate is uncertain. To further examine the specificity of the vitronectin interactions with fibrin, we quantified the binding of increasing concentrations of oligomeric 125 I-labeled vitronectin (0.5-270 nM) to wells pre-coated with a fixed concentration of fibrin (375 nM) (Fig. 5B). A 10-fold molar excess of unlabeled vitronectin inhibited binding of 125 I-labeled vitronectin by 75%. These results are consistent with the presence of a limited number of saturable vitronectin-binding sites on fibrin. Moreover, the Scatchard plot of the specific binding curve yields an upward-convex curve that is consistent with a nonlinear, cooperative binding process (Fig. 5B, inset).
Confocal Microscopic Examination of Vitronectin in Fibrin Clots-We have previously reported that fibrin-bound vitronectin mediates the binding of PAI-1 to fibrin clots (35). Confocal immunofluorescence microscopic imaging of fibrin-associated PAI-1 in clots formed from normal plasma reveals an intense punctate staining for PAI-1 that is distributed with a notably periodicity along the length of the fibrin fibrils (Fig. 6A, arrows), and at sites of fibrin fibrils branching. To visualize vitronectin interactions with fibrin, and determine how it may regulate the periodic distribution of PAI-1 on fibrin, we used confocal microscopy to examine the structure of fibrin-associated vitronectin in unfixed, wet-mounted clots formed in the presence of FITC-conjugated fibrinogen and Texas Red rhodamine/biotin-labeled vitronectin. As controls, clots were formed in the presence of FITC-fibrinogen but without labeled vitronectin (Fig. 6, B and C), or in the presence of biotin-labeled vitronectin but without FITC-fibrinogen (Fig. 6, D and E). These images illustrate that there is no fluorescence emission crossover between the two fluorochrome capture channels, and underscore the morphological differences between the linear network of fibrin fibrils versus the globular vitronectin aggregates. Fig. 7A is a dual red/green overlay image of a FITC-fibrinogen-labeled clot formed in the presence of Texas Red rhodamine-labeled/biotin-BSA. The lack of any significant red fluorescence in these images confirms that that the BSA does not associate with fibrin. In contrast, purified fibrin clots formed in the presence of similar molar ratios of biotin-labeled vitronec- tin and fibrinogen reveal distinct foci of variable-sized vitronectin aggregates that cluster around fibrin fibrils, particularly at points of fibril branching and overlap (Fig. 7B, large arrows). Closer inspection of the vitronectin distribution within successive Z-plane optical sections reveals fibrin-bound, biotin-labeled vitronectin is distributed at regular intervals along the length of FITC-fibrin fibrils (Fig. 7B, small arrows). These periodic points of vitronectin on fibrin are seen to coalesce to form the branching, globular vitronectin clusters. To further investigate the phenomenon of periodicity, purified fibrin clots were formed with a lower molar concentration of biotin-labeled vitronectin to fibrinogen (1:30) so as to minimize formation of fluorescent aggregates. Under these conditions, more punctate forms of fibrin-bound vitronectin (red) are observed (Fig. 7C,  arrows). Likewise, punctate forms of vitronectin also are observed when plasma containing trace amounts of biotin-labeled vitronectin is clotted (Fig. 7D, arrows). Table I represents the results from morphometric measurements of the interval distances between fibrin-bound vitronectin foci in purified and plasma clots. The average distance between adjacent vitronectin foci on purified fibrin fibrils is 1.07 m (Ϯ0.25 m), and is not significantly different from that measured in plasma clots (1.13 Ϯ 0.19 m). Interestingly, these measures of periodicity coincide with those observed for fibrin-bound PAI-1 (Fig. 6A,  arrows).
Interaction of Vitronectin with ␥A/␥Ј Fibrinogen-Our find- ings that vitronectin associates with fibrinogen during coagulation, and that vitronectin is bound with a distinct periodicity along fibrin fibrils, suggest that vitronectin binds only to a subset of fibrin(ogen) molecules. The linear length of fibrin (ogen) is 45 nm (44,45). Our measured periodicities of fibrinassociated vitronectin ranges from 0.8 to 1.4 M. Taken together, these data suggest that vitronectin binds to approximately one in every 18 -31 fibrin(ogen) molecules that could be randomly incorporated along the length of fibrin protofibrils. One possible candidate for specific vitronectin binding is the highly conserved ␥A/␥Ј fibrinogen variant that represents 5-10% of the total circulating fibrinogen. The more anionic ␥A/␥Ј variant can be isolated from ␥A fibrinogen by ion exchange chromatography on DEAE-Sepharose (36). Western blot analysis of fractionated fibrinogen indicates that most of the plasma vitronectin which contaminates commercial fibrinogen preparations (ϳ0.1 g of vitronectin/mg of fibrinogen) co-purifies with the ␥A/␥Ј fibrinogen (Fig. 8A). A smaller proportion of vitronectin also is detectable in the peak I fraction, but is only visible when the gel is overloaded. To explore the possibility that ␥A/␥Ј fibrinogen preferentially binds to vitronectin, we quantified the incorporation of 40 nM 125 I-labeled vitronectin into clots formed in from increasing concentrations of either ␥A/␥A or ␥A/␥Ј fibrinogen. Vitronectin pref-

TABLE I Periodicity of fibrin-bound vitronectin foci in purified fibrin and platelet-poor plasma clots formed in the presence of biotin-labeled
vitronectin Clots were prepared from either purified fibrinogen (3 M) or PPP containing 0.15 M FITC-conjugated fibrinogen and 0.1 M biotin-labeled vitronectin. The biotin-labeled vitronectin was then detected after clotting using streptavidin-conjugated Texas Red rhodamine as described under "Experimental Procedures." Images were captured using a Zeiss LSM510 confocal microscope, and the interval distance between adjacent foci of labeled vitronectin on fibrin fibrils quantified from randomly selected fields of view, as illustrated in Fig. 7 erentially binds to clots formed from ␥A/␥Ј fibrinogen, and the sigmoidal dose-response curve is again consistent with a cooperative binding interaction (Fig. 8B). Moreover, the half-maximal vitronectin binding occurs with 0.6 M ␥A/␥Ј fibrinogen, a value similar to the K d measured for total fibrinogen (Fig. 5A).
Although the ␥A/␥A fibrinogen also binds vitronectin, it does not saturate under the conditions of these experiments.

DISCUSSION
This report is the first comprehensive examination of the direct binding interactions between vitronectin and fibrin(ogen). The notion that plasma vitronectin binds to fibrin clots is supported by our observations of lower levels of vitronectin in serum than plasma, of the presence of vitronectin in solubilized plasma clot extracts, of the localization of vitronectin on fibrin fibrils in plasma clots, as well as of the vitronectin-dependent binding of plasma PAI-1 to fibrin (35). To further characterize the nature of these binding interactions, we used direct binding measurements and morphological studies with purified vitronectin and fibrinogen in solution, and on microtiter plates.
Vitronectin Is Incorporated into Fibrin Clots-Vitronectin incorporation into clots, like that of thrombospondin (46), another adhesive glycoprotein, is non-saturable and factor XIIIindependent. Kinetic studies indicate that the incorporation of vitronectin into clots is not necessarily dependent on the pres-ence of pre-formed fibrin as the initial rate of precipitable vitronectin incorporation into clots exceeds the rate of fibrinogen incorporation during the early phases (Ͻ5 min) of coagulation. Also, additional vitronectin incorporation is not observed even after fibrinogen polymerization is complete. However, this does not exclude the possibility that vitronectin interacts with pre-formed fibrin because oligomeric vitronectin binds specifically to fibrin-coated microtiter wells. Additional binding of vitronectin to the fibrin matrix occurs via cooperative binding interactions between the fibrin-bound vitronectin and native vitronectin subunits. This type of positive cooperativity binding process may account for the proposed two binding site interactions, and is consistent with the previously described mechanism of the concentration-dependent, urea-induced formation of vitronectin polymers (47).
Vitronectin Associates with Fibrinogen during Coagulation-Several lines of evidence indicate that vitronectin co-polymerizes with fibrin. First, vitronectin incorporation into clots is non-saturable, and is directly related to the concentrations of vitronectin and fibrinogen. Second, disulfide-linked vitronectin multimers bind to purified fibrin clots in a cooperative manner, a characteristic consistent with that of an accreting polymer, like fibrinogen. Third, it is unlikely that the vitronectin is trapped in the clots because diffusion of clot-associated FIG. 8. Vitronectin association with the ␥A/␥ fibrinogen. Purified fibrinogen was fractionated by ion exchange chromatography using DEAE-cellulose, and resulted in two distinct fractions, known as peak I and peak II, which differ with respect to their ␥ chains. The peak I fibrinogen contains two ␥A chain dimers, whereas peak II fibrinogen is a heterodimer containing one ␥A chain, and one ␥ Ј chain. Panel A, fibrinogen samples were fractionated by SDS-PAGE (reduced), and the gels were either processed for Coomassie Blue staining for visualizing the fibrinogen chains, A␣, B␤, and ␥A or ␥Ј (total fibrinogen, 15 g/lane; peaks I and II fibrinogen, 9 g/lane), or the samples (3 g/lane) were transferred to nitrocellulose membranes, and processed for Western blot analysis using rabbit antisera directed against either human ␥Ј fibrinogen, or vitronectin. Panel B, purified fibrin clots were formed from a TBS buffer solution (0.2 ml) containing various concentrations of either peak I or peak II fibrinogen, and 40 nM cpm of 125 Ivitronectin. After 2 h at 37°C, the thrombin was neutralized with PPACK, the insoluble fibrin clots precipitated by centrifugation, and the quantities of radioactivity in the supernatant and precipitate were determined. Data represents the concentration of fibrin-bound vitronectin versus concentration of fibrinogen added from one of three representative experiments.
vitronectin is consistently lower than that of ovalbumin, and levels of binding of vitronectin remain stable. Albumin is not incorporated into clots formed from plasma or purified fibrinogen. Finally, direct morphological examination of clot-associated vitronectin structure reveals a branching network of globular polymeric aggregates.
A New Function of ␥A/␥Ј Fibrinogen in Hemostasis-Confocal microscopic studies reveal fibrin-bound vitronectin aggregates clustered at intervals along the length of fibrin fibrils, and extending laterally between adjacent fibrin fibrils, particularly at sites of fibril branching or overlap. Morphometric analysis reveals an average periodicity of fibrin-bound vitronectin of 1.1 Ϯ 0.3 m, suggesting that vitronectin binds to specific, repeating domains along the fibrin polymers. Our studies indicate that the vitronectin, which is a trace contaminant in commercial fibrinogen preparations (35), co-elutes with peak II fibrinogen, and clots formed from ␥A/␥Ј fibrinogen incorporate significantly more vitronectin than clots formed from peak I fibrinogen. These results strongly support the hypothesis that the anionic sequence within the carboxyl termini of the fibrinogen ␥Ј chain contains the major vitronectinbinding site for fibrin.
The fibrinogen ␥A/␥Ј variant is found in 5-10% of circulating fibrinogen levels in humans (48 -50), and has unique interactions with proteins that regulate fibrin formation. Fibrinogen ␥A/␥Ј accelerates thrombin-mediated factor XIII activation as a consequence of possessing binding sites for factor XIII (51) and thrombin (52). Moreover, the binding of vitronectin to fibrinogen ␥A/␥Ј may serve an anti-fibrinolytic function by localizing PAI-1 on fibrin fibrils. Furthermore, it remains speculative whether the presence of vitronectin in proximity of the thrombin-binding sites on the fibrinogen ␥A/␥Ј variant may also serve a regulatory role in the vitronectin-dependent interactions of thrombin with anti-thrombin (15) and PAI-1 (24).
Proposed Model for Coagulation-induced Vitronectin Association with Fibrinogen-Our findings suggest plasma vitronectin interacts with circulating ␥A/␥Ј fibrinogen prior to clotting, and additional vitronectin incorporation occurs post-clotting. Moreover, the binding of vitronectin to fibrin may not be limited to its pre-clotting interactions with ␥A/␥Ј fibrinogen as vitronectin also binds to clots formed from peak I fibrinogen, and to pre-formed fibrin surfaces.
Recent studies with vitronectin-deficient mice support the notion that plasma vitronectin has complex effects on thrombogenesis. Thus, investigators have identified a previously unexpected antithrombotic effect of vitronectin at sites of plateletrich thrombosis (53). These authors postulate that the effect is caused, at least in part, by vitronectin-mediated inhibition of thrombin-fibrinogen interactions, a phenomenon that may be related to the binding of vitronectin to fibrin. On the other hand, in another vascular injury model of occlusive thrombus formation, the absence of vitronectin inhibits reocclusion, and modulates endogenous fibrinolysis (54). This may be related to the recent findings that arterial thrombi in vitronectin-deficient are unstable and frequently embolize (55). Thus, incorporation of vitronectin into fibrin clots is likely to play a multifunctional role in regulating hemostasis, fibrinolysis, and cell adhesion/migration during thrombosis, angiogenesis, and wound healing.