Expression of Primary Polymerization Sites in the D Domain of Human Fibrinogen Depends on Intact Conformation*

Fragments Dl and DD, plasmic degradation products of human fibrinogen and cross-linked fibrin, respec- tively, originate from the COOH-terminal domain of the parent molecule. Since a specific binding site for fibrin resides in the COOH-terminal region of the y chain, the primary structure of the two fragments was compared and their affinity for fibrin monomer measured. Fragments Dl and DD contained the same segments of the three fibrinogen chains, corresponding to the sequences a105-206, Fragment DD had a double set of the same chain remnants. Fragments Dl and DD inhibited polymerization of fibrin monomer in a dose-dependent manner; 50% inhibition occurred at a molar ratio of fragment to monomer of 1:l and 0.5: 1, respectively. To prevent fibrin monomer polymerization and render it suitable for binding studies in the liquid phase, fibrinogen was decorated with Fab fragments The binding of native and denatured fragments Dl DD was studied under the same conditions. In some experiments, a plasmic digest of fibrinogen was used in addition to purified fragment Dl.

Fragments Dl and DD, plasmic degradation products of human fibrinogen and cross-linked fibrin, respectively, originate from the COOH-terminal domain of the parent molecule. Since a specific binding site for fibrin resides in the COOH-terminal region of the y chain, the primary structure of the two fragments was compared and their affinity for fibrin monomer measured. Fragments Dl and DD contained the same segments of the three fibrinogen chains, corresponding to the sequences a105-206, B134-461, and 763-411. Fragment DD had a double set of the same chain remnants. Fragments Dl and DD inhibited polymerization of fibrin monomer in a dose-dependent manner; 50% inhibition occurred at a molar ratio of fragment to monomer of 1:l and 0.5: 1, respectively. To prevent fibrin monomer polymerization and render it suitable for binding studies in the liquid phase, fibrinogen was decorated with Fab fragments isolated from rabbit antibodies to human fragment Dl. Fibrinogen molecules decorated with 6 molecules of this Fab fragment did not clot after incubation with thrombin, and the decorated fibrin monomer could be used to measure binding of fragments Dl and DD in a homogeneous liquid phase. The data analyzed according to the Scatchard equation and a double-reciprocal plot gave a dissociation constant of 12 nM for fragment Dl and 38 nM for fragment DD. There were two binding sites/ fibrin monomer molecule for each fragment. After denaturation in 5 M guanidine HCI, the inhibitory function on fibrin polymerization was irreversibly destroyed. Denatured fragments also lost binding affinity for immobilized fibrin monomer. The preservation of the native tertiary structure in both fragments was essential for the expression of polymerization sites in the structural D domain.
Two major structural and functional regions are distinguished in the fibrinogen molecule. The E domain resides in the center encompassing the NH2 termini of the Aa, BO, and y polypeptide chains (1,2). The D domain involves the COOH termini of the p and y chains (1); it is recovered in fragment Dl ( M , 103,000) cleaved by plasmin from either fibrinogen or non-cross-linked fibrin (3,4). Degradation of cross-linked fibrin with plasmin results in the formation of fragment DD (5)(6)(7)(8). Although its amino acid sequence has not been completely determined, it was inferred from the polypeptide chain composition analyzed by polyacrylamide gel electrophoresis that fragment DD contains two D moieties linked covalently by t(y-glutamy1)lysine isopeptide bonds between the COOHterminal regions of the two y chain remnants (5). Fragments Dl and DD inhibit fibrin monomer polymerization (9,10). This phenomenon results from the presence of a fibrin polymerization site "a" localized in the COOH terminus of the y chain (11)(12)(13). The conclusion was derived mainly from the interaction of fragments Dl and DB with the process of fibrin clot formation. The two fragments have very similar a and / 3 chain remnants, but their y chains differ, containing at the COOH termini Val-411 and Lys-302, respectively (14). Since fragment Ds neither binds to immobilized fibrin monomer (11,15) nor inhibits fibrin monomer polymerization (9) it is apparent that a 109-residue segment of the y chain COOH terminus has an important contribution in the expression of fibrin polymerization site "a." The complementary site "A" appears in the E domain after removal of fibrinopeptides A by thrombin or batroxobin (12,13). The binding of sites A with a seems to propagate the assembly of fibrin clots.
The formation of fibrin oligomers in aqueous solution at neutral pH (16,17) complicates direct binding studies with fibrin monomer because of the phase transition. In this work we addressed the question of the affinity of fragments Dl and DD for fibrin monomer. The fragments were obtained from human fibrinogen and cross-linked fibrin, respectively, their polypeptide chains isolated, and the NH2-and COOH-terminai amino acid sequences determined. The characterized fragments Dl and DD were then tested for interference with the polymerization process and for binding to either immobilized or soluble fibrin monomer. Fibrin monomer was prevented from polymerizing by decoration with Fab fragments isolated from antibodies against fibrinogen fragment Dl. In this system, polymerization sites A were available on the decorated fibrin monomer and the complementary sites a on fragments Dl and DD. To correlate the inhibitory and binding functions with protein conformation, the fragments were studied before and after denaturation with guanidine HCl. MATERIALS  presence of 25 mM calcium chloride for 4 h at 37 "C. The digest contained homogeneous fragment Dl (18) and fragment E, which after thrombin treatment had the same electrophoretic mobility as fragment Ea (2).
Fragment Dl-This was obtained from a plasmic digest of human fibrinogen (100 mg) by gel filtration on a column (2.5 X 130 cmj of Sephadex G-100 (Pharmacia) in 10% acetic acid. Elution rate was 15 ml/h, and 3-ml fractions were collected. The peak containing fragment Dl was pooled and freeze-dried. Alternatively, nondenatured fragment Dl was obtained from a plasmic digest of fibrinogen by preparative electrophoresis on a Pevikon block (19). Purified fragment Dl was homogeneous in SDS-PAGE' and contained remnants of a (M, 12 Cross-linked Fibrin-This was prepared from human fibrinogen enriched with factor XIII, clotted, and freeze-dried as described previously (20).
Digestion of Cross-linked Fibrin-Digestion with plasmin (2 CTA units/g of fibrin) was done in the presence of 5 mM calcium chloride at 37 "C for 24 h as described before (20).
Fragment DD-This was purified from cross-linked fibrin digest by isolation of the DD . E complex using gel filtration on Sepharose CL-GB (Pharmaciaj as described before (20). The DD.E complex, which contained predominantly fragment El, was dissociated in 3 M guanidine HC1 buffer containing 1.5 M NaC1, pH 5.3, at 37 "c for 30 min. Under these conditions fragment DD precipitated. The precipitate was recovered by centrifugation at 3000 X g for 10 min, dissolved in 0.2 M Tris-HC1 buffer containing 6 M guanidine HC1, pH 8.4, and dialyzed against 3 changes (each 500 volumes) of the same buffer. Alternatively, nondenatured fragment DD was isolated from an exhaustively cross-linked fibrin digest using gel filtration on Sepharose CL-GB as described before (20). Purified fragment DD was homogeneous in SDS-PAGE and con- Antiserum to Fragment Dl-This was prepared in rabbits by intracutaneous injection of 100 pg of the fragment on a biweekly schedule in incomplete Freund's adjuvant. Monospecific IgG to determinants of the D domain exposed on the intact fibrinogen molecule was prepared according to the method of Shainoff and Braun (21). This immunochemically purified anti-Dl IgG was used to prepare Fab fragments, The purified IgG was digested with papain (22) at an enzyme to substrate ratio of 1:IOO (w/w). Antibodies were dissolved in 0.05 M sodium acetate buffer containing 0.01 M cysteine and 1 mM EDTA, pH 5.5, and digested at 37 "C for 2 h. The reaction was stopped by the addition of 0.01 M iodoacetamide in 0.2 M Tris-HC1 buffer, pH 8.2. The Fc portion of the IgG was removed by precipitation with IgG-Sorb (The Enzyme Center, Boston) at room temperature for 30 min followed by centrifugation. The Fab fragments were then dialyzed against 0.1 M sodium phosphate buffer, pH 7.1, and their concentration determined spectrophotometrically using an extinction coefficient of 1.5 at 280 nm for a 1 mg/ml solution.

Analytical Procedures
Amino Acid Analysis-This was performed on a Beckman model 119 automatic analyzer. The chains were hydrolyzed under vacuum with a hydrochloric acid/propionic acid mixture (5050, v/v) (Pierce) at 110 'C for 18,30, and 48 h.
NHz-terminal Sequences of Each Chain-These were obtained using phenylisothiocyanate exactly as described by Weiner and colleagues (23). Identification of dansyl derivatives obtained in 3 cycles was done by chromatography in four solvent systems on 5 X 5-cm polyamide sheets as described by Hartley (24).
COOH-terminal Amino Acid Sequenciq-This was done by carboxypeptidase Y (Pierce) digestion (14,25). Cleavage of the PDI, PDD, yD1, and yyoD chains was performed in 0.1 M sodium acetate containing 5 M urea, pH 5.9. The digestion of and CXDD chain remnants was performed in the same buffer but containing only 3 M urea. In all cases, the molar ratio of enzyme to substrate was 1:lOO (w/w). Cleavage of 3 amino acids was sufficient to define the COOH-terminal sequence in the complete sequence of the polypeptide chains of human fibrinogen.
Polyacrylamide Gel Electrophoresis-This was done in 9% gels containing 0.1% SDS either under nonreducing conditions or after reduction of the samples with 2-mercaptoethanol (26).
Labeling with lz5I of Fragments Dl and DD-This was done by the iodine monochloride technique (27) to give a specific radioactivity of approximately 100,000 cpm/pg or approximately 100 pCi/mg of protein.

Polymerization and Binding Studies
Fibrinogen Decorated with Fab Fragments-Two ml of fibrinogen (0.5 mg/ml) were mixed with anti-D1 Fab at a molar ratio of 1:6 and incubated at 4 "C for 3 h. Then, the Fab-decorated fibrinogen was digested at 37 "C for 2 h with thrombin using 0.4 unit/mg of protein.
Thrombin was then inactivated with &isopropyl fluorophosphate added to a final concentration of 1 mM. In some experiments, the decorated fibrin monomers were purified by a column gel filtration on Ultrogel AcA 22. The decorated fibrin monomers were dialyzed against 0.1 M phosphate buffer, pH 7.3, prior to mixing with various amounts of radioiodinated fragments Dl or DD and preincubation at room temperature for 30 min. The concentration of the decorated fibrin monomer was 0.25 p~ and that of lZ5I-D1 or "'1-DD varied from 0.025 to 5 p~. In the binding studies, aliquots of decorated fibrin monomer were mixed with increasing amounts of 1251-labeled fragment Dl or DD and incubated with Ultrogel AcA 34 (LKB) at room temperature for 2 h. The concentration of the radioactive fragments outside the gel in the presence and absence of the decorated fibrin monomers was determined. The amounts of bound and free peptides were calculated according to the method of Hirose and Kano (28) and the data analyzed according to the Scatchard equation (29) and a double reciprocal plot (30).
Aggregation of fragments Dl and DD in solutions of the same concentration as those used in the binding studies was analyzed in parallel experiments. The fragments at concentrations of 0.025-5 p~ were incubated with 50 mM glutaraldehyde at room temperature for 10 min. Unreacted glutaraldehyde was quenched with 1 M glycine and the incubation mixtures separated in nonreduced SDS-PAGE. Coomassie Brilliant Blue-stained gels were scanned (Photovolt) and the amount of oligomeric fragments calculated after subtraction of stain bound to control gels containing fragments cross-linked at 0.1 mg/ml protein concentration.
Treatment of Fragments Dl and DD with 5 M Guanidine HCI-Fragment Dl, either purified or present in the stage 3 digest of fibrinogen, and fragment DD (2 mg/ml) were dialyzed against 5 M guanidine HCl at room temperature for 5 h. Then they were exhaustively dialyzed at 4 "C against 0.05 M Tris phosphate buffer containing 0.15 M NaCI, pH 7.8.
Fibrin Monomers-These were prepared according to method of Belitser and colleagues (31). Polymerization of fibrin monomers generated from fibrinogen by thrombin or reaggregation of purified fibrin monomers was monitored spectrophotometrically at 350 nm as described previously (IO).
Immobilized Fibrin Monomer-This was obtained as described by Heene and Matthias (32) by treating fibrinogen coupled to CNBractivated Sepharose CL-4B (Pharmacia) with thrombin. One milliliter of packed gel contained approximately 20 mg of bound fibrin monomer. Binding was studied in 0.5 M Tris phosphate buffer containing 0.14 M NaCl and 0.005 M EDTA, pH 7.8. In a typical experiment, 2 ml of fragment Dl or DD (2 mg/ml) was incubated with T h e purified chains were homogeneous (Fig. 1, lanes 2-4 and  6-8); the electrophoretic mobility of the cy and [j remnants in both fragments was the same; the mobility of fragment DD y chain remnant (Fig. 1, lanc. 7) was slow due to its dimeric structure. The amino acid composition of t.he purified polypeptide chain remnants was very similar for fragments Dl and DD when resuks for the lat,ter were expressed as residues/0.5 mol of the fragment. The amino acid composition of the chain remnants was also in good agreement with that calculated from the sequences of human fibrinogen chains (33-36). NH,-and COOH-terminal amino acid sequences were determined to establish the exact length of each chain remnant, wit 11 the results summarized in Table I. The primary structure of fragment DD was identical to that of fragment Dl except   13, and y chain that the former contained a double set of the same segments of the N , 8 , and y chains.

Interference of Fragments Dl and DD uith Fibrin Pol?,m-
Prization-Both fragments Dl and DD used in these studies had a strong inhibitory effect on the rate of clot formation.
T h e effect of these fragments was analyzed over a wide range of concentration (Fig. 2). Fragments D, and DD efficiently inhibited both thrombin-induced fibrinogen polymerization (Fig. 2 A ) and polymerization of fibrin nlonomers (Fig. 2 R ) . A 50% decrease of the maximal reaction rate was observed at a molar ratio of the fragment t,o fibrin mnnomer of 0.5:l and 1:l for D D a n d D l , respectively. The inhibition appeared to be critically dependent upon the native conformation of bot,h derivatives. Fragments Dl and DD which had been exposed t o 5 M guanidine HC1 prior to equilibration in the polymerization buffer lost their abi1it.y t,o inhibit both thrombininduced fibrinogen clotting (Fig.  2 A ) and fibrin mnnomer polymerization (Fig. 2R). Denatured fragments had the same primary structure as the native ones, and, as judged by SDS-PAGE of nonreduced samples, there was no degradation or aggregation of polypeptide chains in either fragment, associated with denat,uration.

Binding of Fragments I), and LID to Immobilized Fibrin
Monomer-To confirm the importance of the native conformation of the fragments in determining affinity for fibrin, the binding of native and denatured fragments Dl and DD to immobilized fibrin monomer was compared. [intreated fragment Dl bound to the immobilized fibrin monomer as evidenced by the retention of protein applied in buffer 1 and its elution with buffer 2 (Fig. 3: bottom). The peak contained fragment D, as shown by SDS-PAGE of nonreduced samples (inset to Fig. 3). T h e txeatment, of fragment Dl with 5 M guanidine HCl, followed by dialysis, resulted in a loss of affinity for fibrin. Only 9% of the denatured fragment bound to fibrin monomer as compared to the untreated counterpart. Denaturation of fragment DD under the same conditions resulted in a total loss of affinity for fibrin monomer (Fig. 3,  top).
Binding Affinity of Primary Polymerization Sites-Spontaneous polymerization of fibrin monomers (16,17) does not allow measurement of binding constants by equilibrium techniques. Therefore, we attempted to block selectively polymerization sites a which are on the D domains of the fibrinogen molecules. For this purpose, Fab fragments were obtained   3 units/ml, and the absorbance increase recorded in a spectrophotometer at 350 nm. When fibrinogen was preincubated at a molar ratio of 1:6 with Fab fragments isolated from anti-Dl antiserum and then treated with thrombin, clot formation did not occur.
from affinity-purified anti-Dl antibodies. These antibody fragments recognized only the epitopes located on the surface of the fibrinogen molecule. The Fab fragments were potent inhibitors of fibrin monomer polymerization as assessed by absorption measurements at 350 nm. In the presence of a 6fold molar excess of anti-Dl Fab fragments, fibrinogen became incoagulable by thrombin (Fig. 4).
The incoagulable derivative of fibrin monomer was purified by gel filtration on a column of Ultrogel AcA 22. The elution profile showed the presence of high M, species and only a small amount of contamination by noncomplexed proteins. The composition and apparent molecular weight of the decorated fibrin monomers was tested by SDS-PAGE of nonreduced samples (Fig. 5). The decorated fibrin monomers had polymerization site a of the D domain blocked with Fab fragments, but the functional polymerization site A in the E domain was apparently available. This property allowed us to measure the binding affinity of fragments Dl and DD in a and Fab were observed. The elution profile of fibrinogen alone (0) from the same column serves as a control. homogeneous liquid phase system. The binding studies were performed using the gel equilibrium technique (28). The Fabdecorated fibrin monomer was prepared immediately before the binding experiments.
A Scatchard plot of the binding data gave values of Kd for fragment Dl and DD of the same order of magnitude, that is 12 and 38 nM, respectively. The values were calculated from the solid lines in Fig. 6 derived from experiments done at low concentrations of fragments. The limiting value of y, corresponding to the molar ratio of bound fragment to fibrin monomer was found to be equal to 2, indicating that there 9120

Fibrinogen Polymerization Sites
were two binding sites for each fragment on the decorated fibrin monomers. The course of the experimental points in the Scatchard plot obtained from experiments done at a high concentration of fragments showed the characteristic nonlinearity predicted by the simulation studies of Calvert and coworkers (37). These points are not connected with a line, although the experimental conditions were the same, and calculation gave Kd values of 1.6 and 0.8 pM, respectively, for fragments Dl and DD.
As postulated before (37), deviations from linearity in a plot of the binding data represent either cooperativity between binding sites or the presence of heterogeneous interactions. The latter may result from nonspecific aggregation of fragments Dl or DD, particularly at high protein concentrations. In order to explore the origin of the interaction, glutaraldehyde cross-linking of the interacting proteins was performed. This technique has been found to be useful in the analysis of quarternary structure of protein molecules (38). For this purpose, increasing concentrations of fragments Dl or DD, corresponding to those used in binding studies, were treated with 50 mM glutaraldehyde at room temperature for 10 min. Aliquots preincubated with 1 M glycine were analyzed by SDS-PAGE for the presence of oligomeric forms. The stained gels were scanned and the percentage of cross-linked fragments calculated.
It was found (Fig. 6, broken lines) that there was a linear increase in the amount of aggregated species of fragments Dl and DD starting from a concentration of approximately 1.5 p~. Above this concentration, dimeric, trimeric, and tetrametric derivatives of fragments Dl and DD were observed in SDS-PAGE of nonreduced glutaraldehyde-cross-linked samples. This fact demonstrated that fragments Dl and DD, at concentrations higher than 1.5 PM, tend to form aggregates and that this reaction may account for the nonlinearity observed during binding studies.

DISCUSSION
Fragment Dl, a terminal product of plasmic digestion of fibrinogen in the presence of Ca2+, binds to immobilized fibrin monomer and inhibits fibrin monomer polymerization. It loses its anticlotting activity when further digested to fragment D2 after chelation of Ca2+ by EDTA or EGTA. The production of fragment Dz from Dl is associated with the release of a peptide yAla-357-Val-411 (39) that seems to be involved in fibrin polymerization site a which is complementary to binding site A exposed by thrombin in the E domain.
The present results demonstrate for the first time that expression of the polymerization site depends upon the integrity of the D domain conformation. Dramatic changes in the inhibitory activity of fragments Dl and DD were observed during polymerization and binding studies when fragments denatured with guanidine HC1 were used. As judged by gel analysis, treatment of the fragments with this denaturant did not result in cleavage in the COOH-terminal region of the y chain. Spectropolarimetric studies showed that fragment D lost 95% of a-helical structure without any fragmentation after incubation in 5 M guanidine HC1 (40). Apparently, fragment Dl after denaturation contains the entire y chain peptide segment corresponding to the sequence yAla-63-Val-411. Therefore, a loss of the inhibitory activity in denatured fragments Dl and DD must have been caused by conformational changes occurring in and around the polymerization site.
Two hypotheses can be proposed to explain the importance of the native D domain conformation in the expression of polymerization site a. First, the polymerization site may be formed simply by a linear sequence of amino acid residues in a segment ofthe y chain. This segment would occupy a highly accessible location on the surface of the fibrinogen molecule. Second, the polymerization site may have a complex structure, requiring for complete expression the presence of several not necessarily contiguous peptide segments in close proximity. Both the simple and complex models can explain the effect of 5 M guanidine HCl on the inhibitory activity of fragments Dl and DD. Unfolding of the fragments after denaturation may result in steric hindrance of binding at the polymerization site resulting from the altered arrangement of the immediate sequence around the site. On the other hand, dissociation of separate peptide segments forming the polymerization site may cause the loss of inhibitory activity.
Direct binding studies using Ultrogel AcA 34 to separate free from bound ligand showed that there were two binding sites for fragments Dl and DD on a fibrin monomer decorated in the D domain with Fab fragments. The finding of two binding sites for fragment DD molecules on the decorated fibrin monomer was unexpected. Since fragment DD contains two identical subunits, it could be expected to express double polymerization sites "a,a" and form an equimolar complex with the decorated fibrin monomer. Since the cross-linking reaction occurs after completion of fibrin polymerization, the formation of two e-(y-glutamy1)lysine bonds may distort the distance between two a sites on the y chains. Thus, there could be an imperfect fit with two A sites in the native E domain. In addition, fragment DD, when released from crosslinked fibrin by plasmin, might undergo a conformational change which caused either shielding of one polymerization site or its dislocation. In either case, binding of fragment DD to the decorated fibrin monomer could only be accomplished through one polymerization site a. Thus, fragment DD would become functionally monovalent.
Plasmic digestion of fragment Dl in the presence of EDTA results in its degradation to fragment DS (41) and several peptides. Using affinity chromatography on immobilized fibrin monomer, we isolated peptide 7374-411 that inhibited fibrin polymerization and bound to a thrombin-treated NHzterminal disulfide knot with a Kd of 1.4 p~ (42). Horwitz et al. (43) confirmed this observation and obtained with staphylococcal protease peptide 7374-396 that inhibited polymerization in stoichiometric proportions. The findings prompted conclusions about the localization of polymerization site a in that segment of the y chain. However, two recent publications are at variance. Southan and colleagues (44) did not find any inhibitory activity either in peptides y303-356, y357-373, $374-405, or in their mixtures, leaning to our concept of complex tertiary-structured polymerization sites (45). Varadi and Scheraga (39) purified by high pressure liquid chromatography noninhibitory peptide 7374-411 and proposed that the segment y356/357-411 is important for maintenance of a polymerization site in this region. Regardless of the identification of peptide segments participating in polymerization site a, there is little doubt about its localization in the y chain between Ala-357 and the COOH terminus. However, contributions of other parts of the y chain to the expression of binding affinity should be taken under serious consideration.
Abnormal fibrinogen Milano I with the amino acid replacement yAsp-330 + Val (46) and fibrinogen Haifa, yArg-275 9 His (47) are characterized by defective fibrin clot formation and provide supportive evidence for long-range cooperativity of charged residues in an extended polymerization site.