Conformational changes in fibrinogen elicited by its interaction with platelet membrane glycoprotein GPIIb-IIIa.

The binding of fibrinogen to membrane glycoprotein GPIIb-IIIa on activated platelets leads to platelet aggregation. This interaction results in conformational changes in fibrinogen as evidenced by the expression of receptor-induced binding sites, RIBS, epitopes which are expressed by the bound but not the free ligand. In the present study, two RIBS epitopes have been localized. One sequence resides at gamma 112-119 and is recognized by mAb 9F9; the second is the RGDF sequence at A alpha 95-98 and is recognized by mAb 155B16. These epitopes are also exposed by adsorption of fibrinogen onto a plastic surface and digestion of the molecule by plasmin. Proteolytic exposure of the epitopes coincides with cleavage of the carboxyl-terminal aspects of the A alpha-chains to form fragment X2. The inaccessibility of the RGDF sequence at A alpha 95-98 in fibrinogen suggests that this sequence does not participate in the initial binding of the molecule to GPIIb-IIIa. The location of these RIBS epitopes suggests a model in which binding of fibrinogen to its receptor alters the conformation of the carboxyl-terminal aspects of the A alpha-chains, exposing the sequences which reside in the coiled-coil connector segments between the D and E domains of the molecule. These sequences may then serve as epitopes and may mediate unique functions of the receptor-bound molecule.

** Present address: Center for Thrombosis and Vascular Biology, 44195.
Small peptides which correspond to three regions of primary structure within fibrinogen are recognized by . These recognition peptides are: the COOH terminus of the y-chain, as exemplified by 7400-411 (HHLGGAKQAGDV); the RGDF sequence at Aa 95-98; and the RGDS sequence at A a 572-575. These recognition peptides not only inhibit binding of the parent molecule to the receptor but also interact directly with GPIIb-IIIa. Nevertheless, the role of these sequences in the intact fibrinogen molecule in receptor recognition is circumstantial. In particular, the role of the RGDF sequence at Aa 95-98 remains uncertain.
Binding of fibrinogen to GPIIb-IIIa results in conformational changes in both the occupied receptor and bound ligand (reviewed in Ref. 5). The occurrence of such conformational alterations is evidenced by the selective reaction of certain monoclonal antibodies (mAb) with epitopes expressed preferentially by the occupied receptor (ligand-induced binding site) (6)(7)(8) or the bound ligand (RIBS). RIBS epitopes include those recognized by mAb 9F9 and anti-Fg-RIBS-I (9,10). These mAb interact preferentially with fibrinogen bound to the activated platelets as opposed to the molecule in solution. These and certain other mAb also preferentially recognize fibrinogen deposited onto other surfaces such as plastic microtiter plates (11,12). The epitope for anti-Fg-RIBS-l resides, at least in part, in the COOH-terminal aspects of the y-chain at residues 373-385 (10). The epitope for mAb 9F9 has not been localized. In preliminary studies, we demonstrated that mAb 9F9 and anti-Fg-RIBS-I did not compete with one another for binding to immobilized fibrinogen, suggesting that their epitopes are distinct.
To elucidate further the nature of conformational changes occurring in fibrinogen bound to GPIIb-IIIa, we have sought to localize the epitope for mAb 9F9 in fibrinogen. The location of this epitope has allowed us to predict that the RGDF sequence at A a 95-98 would not be readily available in soluble fibrinogen, and this prediction has been directly tested. Thus, our results have direct bearing on the mechanism of fibrinogen binding to GPIIb-IIIa. Finally, we have sought to determine whether the interaction of fibrinogen with other integrins leads to similar conformational changes in the receptor-bound ligand.

Purification of Fibrinogen and Generation of Proteolytic F r q -
ments-Fibrinogen was purified from fresh human blood by differ-2 1080 entia1 ethanol precipitation (13). The protein was stored at -70 "C, thawed at 37 "C before use, and labeled with lZ5I as previously described (14). Fragment Dl00 (Mr 100,000) was prepared by digestion of 15 mg/ml fibrinogen in 0.05 M Tris-HC1 buffer, pH 7.4, containing 0.1 M NaCl, 1 mM NaN,, 5 mM CaC12, 7 pM merthiolate with human plasmin (Sigma) at 13 units/g of fibrinogen at 37 "C. Digestion was terminated after 24 h by addition of Trasylol (FBA Pharmaceutical, New York, NY) at a ratio of 250 KIU/unit of plasmin. The digest was subjected to gel filtration on Sephacryl S-200 (Pharmacia LKB Biotechnology Inc.), and fractions containing fragment Dl00 were pooled. After this purification step, fragment D100, although not devoid of fragment E3, maintained its native conformation (15). To prepare fragment D80 (Mr 80,000), fibrinogen was digested in the presence of 5 mM EDTA at 37 "C for 21 h. Fragment D39 ( M , -39,000) was produced by digestion of fragment D80 with plasmin in the presence of 2 M urea (16).
Monoclonal Antibodies-Monoclonal antibodies 9F9 (9), 155B16 (17), 134B29 (17) and anti-Fg-RIBS-1 (10) have been previously described in separate publications. MAb 9F9 is the antibody with RIBS-like properties, exhibiting preferential reactivity with plateletbound but not free fibrinogen and also reacting selectively with fibrinogen adsorbed onto a solid surface. MAb 155B16 was raised against the peptide (KK)TTNIMEILRGDFSS, corresponding to the fibrinogen Aa 87-100 sequence. This antibody recognized immobilized fibrinogen in ELISA, and this interaction was inhibited by RGDF but not RGDS peptides. MAb 134B29 was raised against the peptide SSTSYNRGDSTFESK, corresponding to the fibrinogen Aa 566-580 sequence. MAb anti-Fg-RIBS-1 was produced against human fragment DD and recognizes the fibrinogen 7373-385 sequence. MAb 3Gl1, also raised against human fragment DD, interacts with soluble and immobilized fibrinogen. MAb 169.1 reacts with fibrinogen ychain and was kindly provided by Dr. Howard Soule of Corvas International, Inc. (La Jolla, CA). Anti-a& (Mac-1) mAb was OKM-1 (18). MAb LM 609 directed against a& was provided by Dr. Cheresh, The Scripps Research Institute. All mAb (except mAb 169.1 and LM 609) were purified by affinity chromatography on Protein A-Sepharose (Pharmacia LKB Biotechnology Inc.) and were homogeneous as judged by SDS-PAGE. MAb 9F9,155B16, and anti-F-RIBS-1 were conjugated with N-hydroxysuccinimide-biotin ester (Calbiochem) as described (19). F(ab'), fragments of mAb 155B16 were prepared as described (20).
Peptides-A series of overlapping peptides spanning fibrinogen ychain 89-211 were synthesized by solid-phase technology using an Applied Biosystems Model 430 Peptide Synthesizer. Stock peptide solutions were prepared in PBS at nominal concentrations of 2-3 were dissolved in dimethyl sulfoxide and then diluted 30-fold with mM. Some peptides (yl,y-2, y-4,y-9,y-13) were poorly soluble and PBS.
Immunochemical Analyses-For solid-phase immunoassays, polystyrene microtiter plates were coated for 16 h with fibrinogen (100 pl/well at 2 pg/ml) at 4 "C or with synthetic peptides at 100 p~ at 37 "C. The wells were postcoated with 1% BSA in PBS for 1 h at 22 "C. The plates were washed with PBS containing 0.05% Tween 20 and incubated with the mAb, in either unlabeled or a biotinylated form, for 2 h at 37 "C. After 4 washes, goat anti-mouse IgG or avidin, conjugated to alkaline phosphatase (Zymed Laboratories, South San Francisco, CA, and Calbiochem, respectively) was added and incubated for 1 h at 37 'C. After washing, mAb binding was detected by reaction with p-nitrophenyl phosphate, measuring the absorbance at 405 nm.
Competitive ELISA was performed as follows: the selected mAb was diluted so as to be the limiting component in the assay (typically 1 pg/ml, final concentration). Different amounts of the test competitor (fibrinogen or its proteolytic fragments) were mixed with the mAb and 100-pl aliquots were then added immediately to the wells of fibrinogen-coated microtiter plates. Development of the assay proceeded as described above.
TO assess the interaction of mAb 155B16 with fibrinogen in solution, experiments were performed as described (9). Unlabeled mAb 155B16 or 3Gll (0.2 phf) were incubated with 1z61-fibrinogen (0.1-0.3 p~) in PBS for 2 h at 22 "C. Then 100-pl aliquots of the incubation mixtures were added to the wells of microtiter plates coated with goat anti-mouse IgG, 10 pg/ml (Boehringer Mannheim), and postcoated with 1% BSA. After 2 h incubation at 22 "C, the wells were washed with PBS + 0.05% Tween 20 and counted for Iz5I in a y-counter. The amount of Iz5I-fibrinogen bound onto the wells in a complex with each mAb was calculated. Gel Electrophoresis and Western Blotting-SDS-PAGE was performed under nonreducing or reducing conditions in Laemmli buffer system (22) using 7,9, or 15% gels. Proteins in the gels were electrophoretically transferred onto Immobilon-P membranes (Millipore), and the membranes were incubated with the selected mAb (10 pg/ ml). Bound mAb was detected by reaction with a peroxidase-conjugated second antibody (Bio-Rad) followed by addition of the substrate 4-chloro-1-naphthol.
For dot-blot analyses, individual drops of peptide solutions (5 X 10 pl) at 2 mg/ml were applied to Immobilon AV Affinity (Millipore) or Immobilon-P membranes and allowed to dry. The strips were then treated according to manufacturer's protocol.
Platelet Aggregation-Platelets were collected from fresh aspirinfree human blood, anticoagulated with acid/citrate/dextrose in the presence of 2.8 p~ prostaglandin El, and isolated by differential centrifugation followed by gel-filtration on Sepharose 2B in divalent ion-free Tyrode buffer, pH 7.2, containing 0.1% BSA (Sigma). The reaction mixture (total volume 0.4 ml) consisted of 1 x 10' platelets/ ml, 0.3 p~ fibrinogen, 1.0 mM CaCl2, varying concentrations of F(ab'), of mAb 155B16 in Tyrode's buffer. Aggregation was initiated by addition of 10 p~ ADP and monitored at 37 "C as a decrease in light transmission through the platelet suspension using a dual sample aggregometer (Sienco, Morrison, CO).
Flow Cytornetry-FACS analyses were performed as described (9). To 5 pl of gel-filtered platelets (2.5 X 106/ml) in a buffer containing 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl,, 5.6 mM glucose, 1 mg/ml BSA, 3.3 mM NaH2P04, 20 mM Hepes, pH 7.4 (isotonic Hepes buffer), 100 pg/ml fibrinogen, selected concentrations of mAb 155B16 or anti-Fg-RIBS-1 in biotinylated form, ADP and epinephrine (10 p~) were added, and the mixture was incubated for 15 min at 22 "C without stirring. The final incubation volume was 65 p1. Then 4 p1 of 1:lO dilution of avidin-FITC conjugate (Pierce Chemical Co.) was added, and samples were incubated an additional 15 min at 22 "C in the dark. The reaction was stopped by adding 0.5 ml of isotonic Hepes buffer. In controls, 1 p~ PGI, was substituted for the stimuli. Samples were analyzed by flow cytometry using a Becton-Dickinson FACStar instrument. The population of platelets representing single particles was gated. Binding of the mAb was expressed as the mean fluorescence intensity of the particle in arbitrary fluorescence units.
To assess the binding of mAb to endothelial cells, second-passage cultured human umbilical vein cells were detached by a 5-min treatment with 0.02% EDTA. Cells were recovered by centrifugation, resuspended in serum-free RPMI 1640, and distributed at 7 X lo5/ well in microtiter plates (Costar, V-bottom). After incubation for 30 min at 22 "C with normal human serum (1:5), the cells were washed twice with RPMI 1640 and incubated with 10 p~ PPACK, 2.5 mM CaCl,, and 100 pg/ml fibrinogen for 30 min at 22 "C. After removal of excess unbound fibrinogen by centrifugation and washing of cells, varying concentrations (50-200 pg/ml) of the test mAb 9F9 or 155B16 were added. After 30 min at 22 "C, the cells were washed once in PBS and resuspended in 50 pl of goat anti-mouse IgG conjugated with FITC (1:25, Zymed Laboratories). After 20 min at 22 "C in the dark, the cells were recovered by centrifugation, resuspended in PBS, and analyzed by flow cytometry.
The incubation mixture for binding of the mAb to THP-1 monocytoid cells (1.2 X 106/well) was composed of 10 p~ PPACK, 2.5 mM CaCL, in RPMI 1640. Cells were stimulated with 10 p~ formylmethionyl-leucyl-phenylalanine (Sigma) immediately before the addition of 250 pg/ml fibrinogen. After incubation for 20 min at 22 "C, the cells were centrifuged, resuspended in RPMI 1640, and incubated with different concentrations of biotinylated or unlabeled mAb 155B16 or 9F9 for 20-30 min at 22 "C. Subsequent steps were performed as described above for endothelial cells. In parallel, the binding of IZ5I-fibrinogen to THP-1 and endothelial cells was assessed as described previously in detail (16).

The 9F9 Epitope is Exposed at Early Stages of Fibrinogen
Degradation-Initial immunoblotting analyses had demonstrated that the 9F9 epitope is present in the D fragment of fibrinogen (9). The results shown in Fig. 1 verify this localization and further indicate that proteolysis of fibrinogen by plasmin exposes this epitope. Utilizing fibrinogen immobilized onto microtiter plates as the target antigen, a state in which the 9F9 epitope is exposed, the capacity of intact fibrinogen and two D fragments, DlOO and D80, to inhibit antibody binding was evaluated. Soluble fibrinogen produced only modest inhibition of antibody binding to immobilized fibrinogen. At 200-fold molar excess of soluble fibrinogen to the mAb (770-fold excess of soluble to immobilized fibrinogen), soluble fibrinogen produced less than 40% inhibition. In contrast, fragments DlOO and D80 were considerably more effective in inhibiting binding of mAb 9F9 to surface-bound fibrinogen. These two D fragments produced 50% inhibition of binding at about 50-fold molar excess. Accordingly, the 9F9 epitope falls into the category of a cleavage-associated neoantigen, an epitope which becomes expressed during the proteolytic digestion of fibrinogen (23).
To define which step in fibrinogen degradation is responsible for exposure of the 9F9 epitope, the inhibitory effect of serial plasmin digests of fibrinogen in the solid-phase immunoassay was determined. Fibrinogen was digested with plasmin at 0.2 unit/ml of 8 mg/ml fibrinogen in the absence of calcium. At selected time points, digestion was terminated by addition of Trasylol, and aliquots of the initial fibrinogen and the digests were tested as inhibitors of mAb 9F9 binding to immobilized fibrinogen. At the chosen concentration (100 pg/ ml), nondigested fibrinogen inhibited the binding of mAb 9F9 by approximately 10% (Fig. 2 A ) , whereas the 1-and 2-min digests produced extensive (>84%) inhibition. The inhibitory potency of the digests remained maximal throughout the 10min digest and then gradually declined. Electrophoretic analysis of these digests is shown in Fig. 2B. The 1-min digest, which exhibited near-maximal expression of the 9F9 epitope, contained predominantly fragments XI and X,. Full expression of the epitope was retained with subsequent digestion resulting in the disappearance of fragment X2 and formation of fragment Y in the 5-and 10-min digests. The decrease of 9F9 expression appeared to be associated with the accumulation of fragment DlOO and E. Thus, the expression of the 9F9 epitope during proteolysis correlated with sequential cleavage of COOH-terminal aspects of the Aa-chains and formation of fragments X, and X,. epitope, fibrinogen was digested with plasmin in the absence of CaCl, to generate a variety of D fragments. These digests were subjected to SDS-PAGE, transferred onto Immobilon-P membranes, and tested for reactivity with mAb 9F9 by immunoblotting (Fig. 3, lane 1). All fragments of the D-family (D100, D92, D80, D62) reacted with the mAb. A derivative of fragment D of lower molecular weight was produced by digestion of fragment D80 with plasmin in the presence of 2 M urea. This fragment of apparent M, 39,000 (D39) also reacted with mAb 9F9 (Fig. 3, lanes 4 and 5 ) . When fibrinogen was reduced and its constituent Acu-, Bj3-, and y-chains electrophoretically separated, mAb 9F9 bound selectively to the y-and 7'-chains (Fig. 3, lanes 2 and  3). The position of these y-chain variants was confirmed with mAb 169.1 and anti-Fg-RIBS-I, the antibodies with known specificity for the y-chain of fibrinogen (not shown). In reduced D39, a band which reacted with mAb 169.1 also reacted with mAb 9F9 (Fig. 3, lanes 6-8). Collectively, these data support the conclusion that the 9F9 epitope resides in the NH2-terminal part of the y-chain of fragment D. The y-chain remnant of D39 has Met-89 as its NH2 terminus (16)  11,000-13,000). Therefore, the 9F9 epitope resides within To more precisely locate the 9F9 epitope, 13 overlapping synthetic peptides corresponding to 789-211 were tested for their ability bind mAb 9F9 in ELISA and dot-blot formats. Only two overlapping peptides 7-3, 103-119 HDSSI-RYLQEIYNSNNQ, andy-4,112-126 EIYNSNNQKIVNLKE bound the mAb 9F9 in both analyses (Fig. 4, A and B). Two peptides (7-8 and y 9 ) , spanning the region 160-190, showed minor and variable activity in the ELISA but were always negative in the dot-blot analyses. The limited solubility of 7-3 and 7-4 precluded their use as competitors in immunoassays. Inhibition was not observed on an occasion where weak reactivity with 7-8 was observed. On the basis of immunoreactivity of the two overlapping peptides, 7-3 and 7-4, with the mAb, the minimum 9F9 epitope resides, at least in part, within 7112-119 and has a sequence of EIYNSNNQ.
MAb 155B16 to the RGDF Sequence ut A a 95-98 Does Not Recognize Soluble Fibrinogen-In the generally accepted models of fibrinogen, the 9F9 epitope at 7112-119 resides in the coiled-coils and is in close spatial proximity to the RGDF sequence at Aa 95-98 (24). If the RGDF sequence is also hidden, its role in mediating fibrinogen binding to GPIIb-IIIa on platelets would be unlikely. MAb 155B16 was raised to a synthetic peptide corresponding to Aa 87-100 (17) and specifically recognized the RGDF sequence. This mAb reacted with fibrinogen deposited on the surface of polystyrene microtiter plates, and this interaction was not inhibited by soluble fibrinogen (Fig. 5). In comparison with mAb 9F9 which interacted weakly with soluble fibrinogen, mAb 155B16 binding to immobilized fibrinogen was unaltered by the soluble molecule. As a control, soluble fibrinogen did neutralize mAb 134B29, which recognizes the RGDS sequence at A a 572-576.
Two additional experiments confirmed that mAb 155B16 did not interact with fibrinogen in solution. 1) MAb 155B16 did not immunoprecipitate radiolabeled fibrinogen (Fig. 6A). In this experiment, 3Gl1, a control mAb to fibrinogen, immunoprecipitated 1.3 x 10' cpm of '251-fibrinogen, whereas mAb 155B16 precipitated 6.3 x lo3 cpm (in the absence of antibody, nonspecific precipitation of fibrinogen was 6.4 X lo3 cpm). 2) MAb 155B16 did not form a complex with 789-210. The y-chain stretch from 89 to 211 was synthesized as a series of overlapping peptides (denoted as peptides 7-1-7-13). A, immunoreactivity of mAb 9F9 with D39 and synthetic peptides in ELISA. Peptides were adsorbed onto Titertek microtiter plates at 100 p~ (100 pl) and binding of mAb 9F9 (1 pg/ml) was tested. B, dot-blot analysis of mAb 9F9 binding to synthetic peptides 7-1-7-13. Peptides were applied onto Immobilon AV affinity membrane at 2 mg/ml (50 pl), allowed to dry, blocked with 5% dry nonfat milk. After a 2-h incubation with mAb 9F9 (5 pg/ml), reactions were developed using peroxidase-conjugated second antibody.

tion. A , immunoprecipitation of '251-fibrinogen with mAb 155B16 (a)
and 3Gll (W). MAbs were incubated with lZ5I-fibrinogen overnight at 4 "C and immune complexes were precipitated with Protein A. After washing of the pellets, bound 1251-fibrinogen was counted. Control, lZ6I-fibrinogen incubated with buffer (0). B, binding of mAb-fibrinogen complexes to IgG-coated microtiter plates. Goat anti-mouse IgG was adsorbed onto the wells of microtiter wells, and aliquots of mixtures of mAb 155B16 (0) or 3Gll (0) with '"I-fibrinogen (preincubated for 2 h at 22 "C), were added. After 2 h at 22 "C, the wells were washed and counted for lZ5I in a y-counter. The amount of ' "1fibrinogen bound to each antibody was calculated from the counts/ min and specific activity of fibrinogen. anti-mouse IgG. As shown in Fig. 6B, the amount of labeled fibrinogen captured onto the wells from mixtures containing mAb 155B16 and fibrinogen was significantly less than that captured with 3Gll. In control experiments, no significant difference in binding of biotinylated mAb 155B16 and 3Gll to goat anti-mouse IgG was observed, and no binding of lZ5Ifibrinogen to the wells in the absence of antibodies was detected.
Exposure of the RGDF Epitope during Proteolysis of Fibrinogen by Plasmin-The effect of plasmin cleavage of fibrinogen on expression of the RGDF epitope recognized by mAb 155B16 was assessed. Proteolysis of fibrinogen resulted in exposure of the RGDF sequence similarly to the expression of 9F9 epitope at 112-119 (Fig. 2 A ) . Two differences in the expression of the RGDF and 7112-119 epitopes were noted. First, in the 2-min digest which fully expressed the 7112-119, the RGDF epitope was only partially exposed. Full exposure coincided with the formation of fragment X z . Second, prolonged digestion of fibrinogen did not impair the expression of RGDF epitope; the 1.5-h digest, containing fragments D100, E, and low molecular weight peptides had the same inhibitory activity as the 10-min digest containing predominantly fragment X2.
The experiments described above demonstrate that two different procedures, adsorption of fibrinogen on plastic surface and its limited digestion with plasmin, result in the exposure of the Aa 95-98 and 7112-119 sequences. We sought to determine whether adsorbed fibrinogen fully expressed these epitopes. Fragment X2 was used as a reference for full expression of both epitopes. Fibrinogen and fragment X z were adsorbed onto the wells of microtiter plates at the same concentration (2 pg/ml). A similar amount of each protein was adsorbed as determined with a mAb that reacted equally with the parent molecule and fragment X (not shown). The extent to which mAb 155B16 and 9F9 reacted with these target antigens then was assessed (Fig. 7). Fibrinogen expressed both epitopes to a lesser extent than fragment X z , but the differences were more pronounced for the RGDF epitope (4.96 f 0.78 for mAb 155B16 and 1.35 k 0.08 for mAb 9F9 when expressed as the ratio of binding to fragment X p versus fibrinogen). Thus, the binding of fibrinogen to the plastic surface has differential effects on exposure of these hidden epitopes.
Exposure of the RGDF Sequence in Platelet-bound Fibrinogen and Its Role in Platelet Aggregation-Experiments were performed to determine if the RGDF sequence behaved as a RIBS epitope; i.e. was exposed as a consequence of fibrinogen binding to activated platelets. The binding of mAb 155B16 to fibrinogen on the surface of stimulated platelets was detected by flow cytometry. The histograms shown in Fig. 8A provide direct evidence for binding of mAb 155B16 to platelets. The binding was dose-dependent and reached plateau at 50-60 kg of mAb/ml (Fig. 8B). Nonstimulated platelets did not bind mAb 155B16, even in the presence of fibrinogen (Fig. 8B). The extent of expression of the epitope in platelet-bound fibrinogen was evaluated by comparing the interaction of mAb 155B16 and anti-Fg-RIBS-1. The affinity of both mAb in ELISA was similar provided that the appropriate target antigen was used to coat the microtiter plates: fibrinogen in the case of mAb anti-Fg-RIBS-1 and fragment X p for mAb 155B16 (Fig. 9). However, the binding of the two mAb to fibrinogen associated with its platelet receptor differed by 3.88 f 0.54-fold (n = 5). Thus, the RGDF sequence becomes partially available to mAb upon fibrinogen binding to platelets.
We sought to determine whether the partial expression of the RGDF sequence upon fibrinogen binding to GPIIb-IIIa was correlated with a role of the sequence in platelet function. F(ab')z fragments of mAb 155B16 were tested for their effects on ADP-induced platelet aggregation. At the maximal concentration testable, 6.7 k~, the F(ab'Iz fragments produced 25% inhibition. This level of inhibition was not increased by 2.0 7

Mab 155816
Mab 9F9 FIG. 7. Differential expression of RGDF and 9F9 epitopes in fibrinogen and fragment Xa adsorbed onto a plastic surface. Fibrinogen (W) and fragment X p (H) were deposited onto the polysterene microtiter plates at the same concentration (2 pg/ml, 100 pl) and ELISA was performed using mAb 155B16 and 9F9. Five-min plasmic digest of fibrinogen was used as source of fragment X, (conditions of digestion and composition of digest are described in the legend to Fig. 2). preincubation of the F(ab')* fragments with platelets, ADP, and fibrinogen under nonstirring conditions prior to initiation of aggregation by stirring. Under similar conditions, mAb 9F9 had no effect on platelet aggregation.
Effect of Fibrinogen Binding to Other Cells on Exposure of Aa 95-98 and 7112-119"The effect of receptor-mediated fibrinogen binding to other cells on expression of the 9F9 and RGDF epitopes, i.e. whether the receptor occupancy induces similar conformational changes as when fibrinogen binds to platelets, was assessed. Fibrinogen binds to THP-1 cells via a M j 3 2 (25) and to endothelial cells in suspension via a& (26,27). This was confirmed by blocking the binding of fibrinogen to THP-1 cells by mAb OKM-1 (25) and to endothelial cells by mAb LM 609 (26). These mAb inhibited fibrinogen binding to the appropriate cell types by >60%. By FACS analyses, such as described above with platelets, no binding of mAb 155B16 or 9F9 was detected at mAb concentrations ranging from 50 to 200 pg/ml. Nevertheless, in parallel experiments, the binding of radiolabeled fibrinogen to both cell types was verified (Table I). Thus, the interaction of fibrinogen with GPIIb-IIIa on platelets is unique in inducing exposure of A a 95-98 and 7112-119.

DISCUSSION
In this study, we have verified that the interaction of fibrinogen with GPIIb-IIIa on the surface of activated plate- Microtiter plate wells were coated with 100 pl of 2 pg/ml fibrinogen (target antigen for mAb anti-Fg-RIBS-1) or with 100 p1 of 2 pg/ml fragment X? (target antigen for mAb 155B16), postcoated with 1% BSA, and immunoreactivity of both mAbs at 1 pg/ml was determined as absorbance at 405 nm. Right panel, binding of mAb 155B16 (m) and anti-Fg-RIBS-1 (H) to fibrinogen on the surface of stimulated platelets. Gel-filtered platelets were incubated with fibrinogen (100 pg/ml) and 20 pg/ml biotinylated mAb 155B16 or biotinylated anti-Fg-RIBS-1 in the presence of ADP and epinephrine, followed by addition of FITC-avidin, dilution of reaction mixture, and determination of mean fluorescence. 3.1 X 10' "THP-1 and endothelial cells were incubated in the presence of 250 or 100 pg/ml fibrinogen, respectively, under conditions described under "Materials and Methods." Binding of mAb 155B16 at a concentration of 60 pg/ml to activated platelets is expressed as mean fluorescence/cell. There was no binding of antibody at a concentration of 200 pg/ml to THP-1 and endothelial cells.
* Cells were incubated in the presence of 100 pg/ml of '"I-Fg under the same conditions as for FACS analyses, and binding of "'I-Fg to the cells was determined as described (16). Specific binding is defined as the component of total binding which was inhibited by excess nonlabeled fibrinogen. The specific binding was 74% of total binding for THP-1 cells, 91% for endothelial cells, and 93% for platelets. lets induces conformational changes in bound ligand and have demonstrated the extensive nature of these alterations. Two sequences which become exposed upon fibrinogen binding to its receptor have been localized 7112-119, which is recognized by mAb 9F9, and Aa 95-98, which is recognized by mAb 155B16. The epitopes within these two sequences exhibit generally similar properties. Namely, they are not expressed by fibrinogen in solution but become partially or fully available when fibrinogen: 1) is bound to its receptor on stimulated platelets; 2) is adsorbed onto a plastic surface; and 3) is cleaved by plasmin. With respect to plasmin cleavage, both sequences are exposed at early stages of digestion, and maximal expression coincides with the formation of the X fragments. The diversity of events exposing these distinct epitopes suggests that fibrinogen is organized in a structure which permits a global and readily inducible conformational transition.
We postulate that this proposed conformational transition involves changes in the relationship between the carboxyltermini of the Aa-chains and the core D and E domains of fibrinogen. The symmetrical fibrinogen molecule is composed of three pairs of constituent chains: the Aa-, and 7chains (24). The NHz-termini of all six chains assemble to form the central E domain while the COOH-terminal parts of BB, y, and the central part of the Aa-chain form the two distal D domains. Each of the two sets of 7112-119 and A a 95-98 sequences resides in the "connector" segments which link the central E domain with one D domain. Biophysical and biochemical analyses indicate that the COOH-terminal parts of Aa-chains (aC domains) are highly flexible. Immunoelectron microscopy has visualized the aC domains in very divergent organizations (28-32). At times, the aC domains may fold back toward the E domain in such a way that they interact with each other and the E domain to create a ''compact" organization. Data from scanning microcalorimetry support this model (33). In contrast, the aC domains also have been visualized as extending from the ends of the two D domains to create an "extended" organization (31). We propose that the aC domains shield the connector segments between the D and E domains, including 7112-119 and Aa 95-98, limiting access of antibodies and other macromolecules to this region. Binding of fibrinogen to the platelet surface and to certain other surfaces influences the conformation or the conformational transitions of the aC domains and allows access of macromolecules to the connector regions. Proteolysis of the aC domains also eliminates the shielding phenomenon. This hypothesis appears consistent with the behavior of the epitopes contained within 7112-119 and Aa 95-98 and also provides an explanation for the sequential nature of plasmin cleavage of fibrinogen. The aC domains shield the main plasmin cleavage sites between the D and E domains; and, only after removal of the aC domains, do these sites become susceptible to proteolysis. Consistent with this interpretation, the same pattern of degradation of fibrinogen, initial cleavage of the aC domains followed by cleavage within the connectors, is observed with the majority of proteases, including enzymes with distinct peptide bond preferences (34-42).
As proposed by Doolittle and colleagues (43), each connector segment of fibrinogen is composed of a similar number of amino acid residues from each constituent chain. Each chain is predicted to be in the form of an a-helix over a significant portion of this amino acid stretch and is probably intertwined with one another to create a coiled-coil. The coiled-coils are interrupted by several short nonhelical segments, including the protease-sensitive area which is ultimately cleaved to yield the D and E fragments. 7112-119 lies at one of these interruptions in the 7-chain helix. Amino acid sequences at sites of changes in secondary structure frequently serve as epitopes, and this may account for the recognition of the sequence by 9F9. Furthermore, hydrophobicity plots (44) of 789-211 indicate that 7115-120 is the most hydrophilic segment in this region, an additional basis for this segment becoming an epitope upon removal of the aC domain shields. While 7112-119 may be a particularly attractive target for eliciting a mAb, according to the shielding model, sequences throughout the connectors would exhibit the properties of being exposed upon fibrinogen binding to platelets, other surfaces, and being cleavage-associated neoantigens. Two recent observations are consistent with this hypothesis. First, a mAb raised against synthetic peptide corresponding to Aa 148-160 within the connector region did not recognize soluble fibrinogen but did react with fibrinogen after proteolytic or chemical degradation (45). This mAb also reacted with fibrin, raising the possibility that the fibrin transition is still yet another mechanism for inducing the conformational changes which expose the connector regions. Second, mAb DD-3B6, which has been utilized to detect fibrin degradation products in plasma, i.e. does not react with fibrinogen, was mapped recently to 782-178 (46).
The limited availability of Am 95-98 RGDF on the surface of intact fibrinogen and the alterations in its exposure by physiological events has significant implications with respect to the role of this sequence in platelet function. If the RGDF sequence in fibrinogen is inaccessible to a mAb, it will probably also be inaccessible to GPIIb-IIIa. Thus, it is unlikely that the RGDF sequence contributes significantly to the initial binding of fibrinogen to its platelet receptor. This conclusion is supported by recent data using anti-idiotypic antibodies against PAC1, a function-blocking mAb to . Certain of these antibodies did recognize RGDcontaining peptides but, as evaluated by electron microscopy, bound infrequently in the vicinity of the RGDF sequence. Furthermore, no association between GPIIb-IIIa and fibrinogen was observed at this position by direct electron microscopic examination of the complex (48). Nevertheless, binding of fibrinogen to platelets resulted in partial exposure of the RGDF sequence. The extent of exposure of the RGDF sequence was -25%, and platelet aggregation was inhibited to a similar extent by mAb 155B16 to this sequence. Cheresh et al. (17) noted that this same mAb also blocked platelet adhesion to fibrinogen by a similar percentage. It is interesting to speculate that RGDF-mediated interactions might contribute to the variability in platelet organization within a thrombus (49) and that the exposure of this sequence might create a site of interaction for RGD-recognizing integrins on other cell types, leading to their incorporation into the thrombus. Deposition of fibrinogen onto the surface of microtiter wells also resulted in partial exposure of the RGDF sequence. This observation, in concert with previous data indicating that the fibrinogen assumes different conformations on different surfaces (50), suggests that studies of the adhesion of integrinbearing cells to fibrinogen may be markedly influenced by the selected substratum and that the extent of exposure of the RGDF sequence in deposited fibrinogen could control the thrombogenicity of surfaces. While the above discussion emphasizes the diversity of molecular events which induce the appropriate conformational transitions of the aC domains to expose the connector segments, specific initiating events are required. Evidence for the selective induction of the appropriate conformational transitions is indicated by the failure of fibrinogen bound to THP-1 and endothelial cells to express either the 9F9 of 155B16 epitopes. The binding of fibrinogen to endothelial cells is mediated by avp3 (the vitronectin receptor) (26,27) and to THP-1 monocytic cells by LYM& (25). Even though both receptors are integrins and a& and GPIIb-IIIa even share the same p subunit, it is only upon binding of fibrinogen to GPIIb-IIIa that 7112-119 and A a 95-98 are exposed as RIBS epitopes. This difference is consistent with the reported differences in the recognition specificities of these three integrin receptors for fibrinogen (51). However, this interpretation is viewed as being tentative since the density of receptors and the time course of epitope exposure may influence detection of the RIBS epitopes on these other cell types.
In summary, the present study supports and extends the concept that the interaction of fibrinogen with its platelet receptor induces conformational changes in bound ligand. The consequence of these transitions is the expression of RIBS which can elicit and be detected by anti-RIBS antibodies. The present study suggests an entire region composed of segments of all three constituent chains of the fibrinogen molecule, the connector segment, which may be targets for anti-RIBS antibodies. The demonstration that the RGDF sequence behaves as a RIBS necessitate reconsideration of the structure and function of this sequence with respect to its role in platelet aggregation and other cellular interactions. The anti-RIBS characterized in this study are also antibodies to cleavage-associated neoantigens. The existence of antibodies to cleavage-associated neoantigens in other ligands (e.g. complement components) (52) suggests that these antibodies may also be anti-RIBS. Moreover, this parallel between cleavage-associated neoepitopes and RIBS epitopes suggests a strategy for eliciting anti-RIBS in other ligand-receptor systems in which proteolytic digests of the ligand may be used as immunogen. This approach would be extremely useful in circumstance when the identity of the receptor is unknown or its availability is limited.