An Integrated Study of Fibrinogen during Blood Coagulation*

The rate of conversion of fibrinogen (Fg) to the insoluble product fibrin (Fn) is a key factor in hemostasis. We have developed methods to quantitate fibrinopeptides (FPs) and soluble and insoluble Fg/Fn products during the tissue factor induced clotting of whole blood. Significant FPA generation (>50%) occurs prior to visible clotting (4 ± 0.2 min) coincident with factor XIII activation. At this time Fg is mostly in solution along with high molecular weight cross-linked products. Cross-linking of γ-chains is virtually complete (5 min) prior to the release of FPB, a process that does not occur until after clot formation. FPB is detected still attached to the β-chain throughout the time course demonstrating release of only low levels of FPB from the clot. After release of FPB a carboxypeptidase-B-like enzyme removes the carboxyl-terminal arginine resulting exclusively in des-Arg FPB by the 20-min time point. This process is inhibited by ε-aminocaproic acid. These results demonstrate that transglutaminase and carboxypeptidase enzymes are activated simultaneously with Fn formation. The initial clot is a composite of Fn I and Fg already displaying γ-γ cross-linking prior to the formation of Fn II with Bβ-chain remaining mostly intact followed by the selective degradation of FPB to des-Arg FPB.

Blood coagulation proceeds through a cascade of protein activation's that ultimately lead to the catalytic cleavage of fibrinogen (Fg) 1 by thrombin to the product fibrin (Fn). Fn is generated from plasma Fg (M r 340,000) which is found in blood plasma at ϳ3 mg/ml and exists as a symmetrical dimer consisting of A␣, B␤, and ␥ polypeptide chains linked by noncovalent and disulfide bonds (1)(2)(3)(4). The two carboxyl-terminal domains of the B␤ and ␥-chains of Fg are designated "D" while the central domain which contains the amino termini of all the chains is designated "E." Clot formation which has been extensively studied in anticoagulated plasma and purified Fg occurs in a series of steps (4) initiated by thrombin cleavage of the A␣ and B␤-chains of Fg. Cleavage at A␣-16 releases fibrinopeptide (FP) A to form Fn I. The release of two FPA peptides exposes a site in the E domain that aligns with a complementary site in the D domain to form overlapping fibrils (5). This is followed by cleavage at B␤-14 releasing the two FPB peptides to form Fn II. FPB release appears to allow for lateral aggregation of the protofibrils (6,7). The degree of lateral strand association contributes to the tensile strength of the clot, but its resistance to plasmin degradation is influenced mainly by covalent crosslinking. Cross-links are formed by the action of factor XIIIa (fXIIIa), a transglutaminase enzyme whose formation from zymogen fXIII (plasma concentration 90 nmol/liter) is also catalyzed by thrombin (8,9). FXIII consists of an A 2 B 2 tetramer where the A subunit is acted upon by thrombin releasing an NH 2 -terminal activation peptide. Covalent isopeptide crosslinks are formed between certain adjacent ␥-carboxamido and ⑀-amino groups of glutamyl and lysyl residues within the extreme carboxyl-terminal ␥-chains rapidly forming ␥-␥ dimers (10,11). The carboxyl-terminal ␣-chains are also cross-linked but these isopeptide bonds form more slowly (12,13). However, a recent study conducted in human plasma observed that ␣-polymers and A␣-polymers are already present at the point of gelation (14). The overall stability of the clot appears to be dependent upon the formation and orientation of the Fn monomers (15).
In vitro, Fn formation in native whole blood proceeds principally by the activation of thrombin by the intrinsic (contact) pathway of coagulation (16). However, the pathway relevant to physiological hemostasis is the extrinsic (tissue factor, TF) pathway (17) that proceeds through assembly of three membrane surface and vitamin K-dependent enzyme-cofactor complexes (18,19). The initiating complex, is formed when circulating blood containing factor VIIa comes in contact with TF membrane (20). The resulting complex activates zymogens factor X and factor IX (21)(22)(23) to the proteases which with their cofactors factor VIIIa and factor Va form the complexes which propagate the formation of thrombin. Studies of the relevant TF pathway in blood in vitro are only possible by suppression of the contact pathway. We have developed techniques by which selective inhibition of factor XIIa is accomplished by corn trypsin inhibitor (CTI) (24) permitting TF pathway analysis (25).
Fg activation products have been studied extensively (2,15,26) utilizing a variety of methods including high performance liquid chromatography (HPLC) (27)(28)(29)(30). Both in purified systems (31,32) and plasma that was treated with thrombin (33) FPs isolated by HPLC methods showed Fg acted upon by thrombin releases first FPA followed by a delayed release of FPB. Previous studies undertaken to quantitate FPs generated in whole blood (25, 34 -37) have utilized immunoassay techniques for FPA determination. Fg, FPA, thrombin levels, and platelet activation have been studied in whole blood in normal (25) and hemophiliac patients' (34). Studies on normal donors showed that significant amounts of Fg (80%) and FPA (45%) were incorporated into the initial clot at low levels of thrombin (ϳ15 nmol/liter) (25). Comparative studies on a factor VIIIdeficient patient (hemophilia A) showed increased clot time with decreased thrombin levels (1.9 nmol/liter/min versus normal 55 nmol/liter/min) characterized by a slower rate (ϳ30% of the normal) of Fn formation (34). Normal levels of FPA were reached by 20 min even with decreased thrombin generation. Therefore, the level of FPA generation in hemophiliacs does not explain why clots are more friable in these patients. Hence, we wanted to develop a system that can be used to observe other Fg degradation products (i.e. FPB) involved in clot formation and possibly explain why clots are unstable in hemophiliacs. Characterizing other FP products (i.e. FPB) have been attempted (38) but little information is available regarding FPB detection in whole blood. The results obtained present a somewhat different sense of observations from those observed with purified Fg and anticoagulated plasma.

Methods
Normal donors (age range [21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] with no history of blood disorders, regular aspirin, or drug use were recruited and advised according to a protocol approved by the University of Vermont Human Studies Committee. All individuals exhibited normal ranges for plasma Fg (147-340 mg/dl). Blood was collected by venipuncture and rapidly distributed to a series of test and control tubes containing 12.5 pmol/liter of TF relipidated in 25 nmol/liter of phosphatidylcholine/phosphatidylserine (75% phosphatidylcholine/25% phosphatidylserine as described previously (42)) and CTI (100 g/ml) (25). When used, EACA (50 mmol/liter), was added to the tubes prior to the addition of blood. Tubes were quenched over time with a mixture of coagulation inhibitors, 50 mmol/ liter EDTA with 20 mmol/liter benzamidine-HCl in HBS, pH 7.4, and 50 mol/liter FPRck Ϯ 50 mmol/liter EACA. Aliquots were collected either every minute from 0 to 10 or 0 to 2 then every 20 s from 2 to 6 min, followed by 12-, 14-, 16-, and 20-min time points. The zero time point contained the quench solution prior to addition of blood. A tube containing CTI and no TF was added as a control to determine the quality of the phlebotomy and the extent of contact pathway inhibition. Clot time was determined visually. Solid material was removed by centrifugation (15 min at 2000 rpm) and stored at Ϫ80°C. The fluid material was analyzed directly or aliquoted and stored at Ϫ80°C for further analysis.
Analysis of Whole Blood-SDS-PAGE (4 -12%) was performed according to the modified (25) Laemmli (43) procedure. High molecular weight standard mixtures  were loaded along with Fg standards (300 ng/lane), fXIII (50 ng/lane), or activated fXIII (fXIIIa) to allow for comparison on the immunoblots. The gels were transferred to nitrocellulose membranes (Bio-Rad) and subjected to semi-dry transfer for 3 h at 250 mAmp as described by Towbin et al. (44). The primary antibody was either ␣-Fbgn2E or polyclonal ␣-fXIII used at 5 g/ml, the secondary antibody (goat ␣-mouse IgG horseradish peroxidase or goat ␣-rabbit (HϩL) horseradish peroxidase) at 1:5000 dilutions, and the substrate for emitting light was Luminol (NEN Life Science Products Inc.). The blots were developed as described previously (25). Comparisons of fXIII conversion to fXIIIa, and Fg levels in solution were analyzed on the immunoblots. A relative fraction was calculated from Fg or fXIIIa present at each time point divided by total Fg or total fXIII A-chain (fXIIIA ϩ fXIIIAa) detectable.
FPs were isolated using a Waters Model 484 Controller and Model 510 Solvent Pumps monitored using a Model 481 detector at 214 nm. The column used for all analyses was a wide pore octadecyl (C 18 ) Bakerbond (4.6 ϫ 250 mm) (VWR). Peptide samples were eluted by using linear gradients of Buffer A: H 2 O, 0.05% trifluoroacetic acid and Buffer B: CH 3 CN, 0.05% trifluoroacetic acid. The gradient elution was as follows: 0 -5 min 100% Buffer A, 25 min 80% Buffer A, 50 min 60% Buffer A, 55 min 100% Buffer B. The data was analyzed using Labview (Version 4, National Instruments, Austin, TX).
FP standards were prepared from Fg (2.5 mg/ml) in HBS treated with thrombin (5 nmol/l), 37°C, 1 h. The reaction was stopped by the addition of HClO 4 (0.2 mol/l final) and the precipitated proteins removed by centrifugation (10 min at 14,000 rpm). The supernatant was subsequently treated with equimolar KOH and allowed to sit on ice for 30 min, removing the salt precipitate by centrifugation. The soluble material from the whole blood quenched sample time points were also treated with HClO 4 (0.2 mol/l final) and neutralized with equimolar KOH. The HPLC analyses of the FPs were conducted without further separation. The identities of the resolved fractions were confirmed by mass spectrometry, amino acid composition, and amino acid sequence analyses. Fractions corresponding to phosphorylated FPA (P-FPA), FPA, des-Ala FPA, FPB, and des-Arg FPB were identified by matrixassisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (linear model; PE Applied Biosystems, Foster City, CA). The matrix used was ␣-cyano-4-hydroxycinnamic acid, 70/30, H 2 O, CH 3 CN, 0.1% trifluoroacetic acid at a 1:1 ratio with the sample. Amino acid sequencing and composition were performed on the samples (Dr. Alex Kurosky University of Texas, Medical Branch, Galveston, TX). The concentrations of stock FPs were assessed by amino acid composition.
FPA and FPB standard calibration curves of micrograms of injected versus area were developed. Correlation coefficient values were calculated to be as follows: FPA r 2 ϭ 0.966, FPB r 2 ϭ 0.940. Whole blood samples were spiked at the zero time point with a known amount of FPs and percent recovery determined. Recovery of spiked whole blood samples was calculated to be 57%. Therefore, all samples were corrected for by this factor along with corrections for whole blood sample dilution from 1.0 ml of added quench reagents as described previously (34).
Commercial FPA and thrombin-AT-III (TAT) ELISA's were conducted according to the manufacture's protocols with corrections that have been previously described (34). Results were obtained and analyzed using a V max microtiter plate reader (Molecular Devices, Menlo Park, CA) equipped with SOFTMax version 2.0 software and an IBM Personal System 2 Model 30/286 PC. Samples and standards (minimum of 5 standard concentrations) were run in duplicate or triplicate. The concentration was determined by log-logit fit of concentration of stand-ard versus optical density as described by the manufacturer (Molecular Devices).
The clots were analyzed according to previously published procedures (45) as modified by our laboratory. The insoluble clotted samples were washed 2-3 times with 1 ml of 0.15 mol/liter NaCl and then allowed to sit in the salt solution (1 ml) for 12-15 h so that additional soluble material within the clots could diffuse into solution. The clots were rinsed with H 2 O to remove salt, lyophilized, and weighed. The dry clots were solubilized in 4 mol/liter urea, 10% SDS, 10% ␤-mercaptoethanol and analyzed by 5-15% SDS-PAGE. Products were visualized using a Coomassie Blue source or an ␣-FPB monoclonal antibody. Densitometry was performed on a Hewlett-Packard Scanjet 4C/T. Cross-linking results were expressed as a relative fraction of the dimer at each time point over the total (monomer ϩ dimer).

Isolation, Characterization, and Quantitation of FPs-Puri-
fied Fg treated with thrombin releases three major forms of FPA: an unmodified form ADSGEGDFLAEGGGVR (FPA) which constitutes ϳ70%, an NH 2 -terminal truncated form (des-Ala FPA) ϳ10%, and a product in which Ser-3 is phosphorylated (P-FPA) ϳ20%. This is followed by release of FPB (ZGVNDNEEGFFSAR) which should exist in this unmodified form ϳ95%. Fig. 1A displays an HPLC chromatogram depicting the forms of P-FPA, FPA, and FPB. The three forms of FPA while not completely resolved are all identified by MALDI-TOF mass spectrometry ( Fig. 2A). A whole blood assay performed as described ("Experimental Procedures") was analyzed for FPs ( Fig. 1B) following the procedure used for purified Fg. In contrast to the purified system an additional peak, subsequently identified as des-Arg FPB was present. Approximately 22 studies on blood from multiple individuals confirmed these results although quantitative variability was observed from individual to individual. Des-Arg FPB was formed at different rates depending upon the individual blood being tested.
The identities of all Fg products was confirmed by MALDI-TOF mass spectrometry (Fig. 2) and amino acid analyses. Des-Ala FPA ( Fig. 2A, a) shows a mass m/z ϭ 1471.75, FPA (b) m/z ϭ 1540.56, and P-FPA (c) m/z ϭ 1621.68. Des-Arg FPB is seen in panel B (d) with a mass m/z ϭ 1422.37 and its double ion at m/z ϭ 2867.85 (d*). FPB in seen in panel C (e) as a molecular ion of 1556.48. The differences in the values compared with the true molecular ion mass are due to the addition of Na ϩ present. Since MALDI-TOF is not a quantitative tool the relative areas of the peaks cannot be used for quantitation.
Whole Blood Analysis of FPs-Coagulation of blood from normal volunteers after initiation with 12.5 pmol/liter TF was quenched at time points from 0 to 20 min. In Fig. 3, for a typical individual the time points of overlaid HPLC chromatograms are illustrated as 0 (I), 5 (II), 10 (III), and 20 (IV) min. Clotting was visualized at 4 Ϯ 0.2 min while the CTI control (no TF) clotted at 13.25 min. FPA release was complete by 5 min, FPB was observed at lesser amounts, with des-Arg FPB already observed by 5 min. Over time FPB levels decreased and des-Arg FPB levels increased, resulting only in des-Arg FPB by 20 min. Similar results were obtained with des-Arg FPB levels in 15 other whole blood analyses on normal donors initiated with 12.5 pmol/liter TF, where the CTI controls ranged from 12 to 24 min and clot times were between 3 and 4.2 min.
The FPs were quantitated based upon calculated areas and standard curves were developed for FPA and FPB (Fig. 4). Clot time (CT) is represented by an arrow shown at 4 Ϯ 0.2 min. The maximum levels (15.8 mol/liter) (-⅐-⅐-) of FPs which could be expected were calculated from the patients Fg levels (268 mg/ dl) at the time of the blood draw. From this figure it is easy to see that (ϳ50%, 7 mol/liter) FPA (OE) was present prior to clot time. This result is comparable to what has previously been seen in whole blood assays using commercial ELISA kits (25,34). FPB (ࡗ) just starts to appear at clot time (Ͻ1 mol/liter) and is quickly cleaved to des-Arg FPB (f). Maximum FPB and des-Arg FPB levels are equivalent (ϳ5 mol/liter). Both the final levels of FPA (ϳ10 mol/liter) and FPB (ϳ5 mol/liter) at 20 min are below the expected FP levels (ϳ16 mol/liter), possibly an indication of other degradation processes not yet identified. FPA levels appear to reach a maximum at clot time (ϳ14 mol/liter) and begin to decrease with time to a final level of ϳ10 mol/liter. Overall, combined levels of released FPB are not equivalent to FPA levels. The delayed release of FPB after clot formation along with decreased overall levels is suggestive that not all FPB is being cleaved from the B␤-chain.
FPB Degradation to Des-Arg FPB-Previous studies have observed the formation of des-Arg FPB (31,32) and noted that the addition of EACA to the Fg solution prior to thrombin addition prevents the formation of des-Arg FPB. To test whether the enzymatic degradation of FPB occurs during the process of blood coagulation or after sample quenching, the following experiments were performed. 1) EACA (50 mmol/ liter) was added to the collecting tubes prior to the addition of blood; 2) EACA (50 mmol/liter) was added to the mixture of coagulation inhibitors used to quench the reaction. 3) An experiment was performed where no EACA was added. The overlaid HPLC chromatograms illustrated in Fig. 5 represent the 0-(Ia/b), 5-(IIa/b), 10-(IIIa/b), and 20-min (IVa/b) time points plotted as absorbance (214 nm) versus time (seconds). Panel A shows the experiment with EACA (50 mmol/liter) added prior to the distribution of blood. Panel B shows the experiment with no addition of EACA. FPA is seen first over FPB at 5 min in both panels A and B. FPA generation was unchanged throughout the course of these experiments. Therefore, EACA had no effect on FPA separation or stability. FPB remains uncleaved in panel A where EACA was added prior to the addition of blood. Without EACA, des-Arg FPB was formed with time (B, IIIb and IIIc). When EACA was added to the quench buffer, The soluble material was subsequently treated with HClO 4 (0.2 mol/liter), centrifuged to remove any precipitate, then neutralized with equimolar KOH. A 10-min time point that was isolated by the same methods as described in A is illustrated in B. Combined FPA is seen along with FPB and a second peak that was identified as des-Arg FPB.
des-Arg FPB was also observed (data not shown). Thus an enzyme is acting upon FPB during the course of blood coagulation and des-Arg FPB formation was occurring after the samples had been quenched. Whatever carboxypeptidase is cleaving the carboxyl-terminal arginine from FPB is selective. EACA is only inhibiting this carboxypeptidase from attacking FPB since FPA remained unchanged with/without EACA addition within the time frame of this experiment. In panel B (without EACA addition) there also appears to be several unresolved peaks present to the left of des-Arg FPB, that are not present in panel A. These could be other degradation products that are not yet identified, but are also inhibited by EACA.
Several carboxypeptidases can be responsible for causing this type of cleavage of FPB. These carboxypeptidases include: carboxypeptidase-B (46), carboxypeptidase-N (47), and carboxypeptidase-U (48). All are basic carboxypeptidases capable of cleaving carboxyl-teminal lysines or arginines. Previous studies showed that a potent carboxypeptidase-B inhibitor was unable to inhibit cleavage of FPB, unlike EACA (32). Carboxypeptidase-U or TAFI has previously been suggested to play a role in the premature lysis of clots from hemophilic plasma (49). In order to determine if TAFI was capable of cleaving FPB to des-Arg FPB we incubated TAFI (70 nmol/liter) with Fg (2.5 mg/ml), thrombin (10 nmol/liter), and thrombomodulin (4 nmol/ liter) for 35 min at 25°C. Results showed the same pattern of FP generation that was seen in whole blood (Figs. 3-5) (data not shown). The three forms of FPA were seen as well as FPB and des-Arg FPB. No apparent degradation of FPA was detectable during the allotted time interval of the experiment. Next we wanted to test activated TAFI with the substrates FPA and FPB directly to determine if TAFI was capable of cleaving the carboxyl-terminal arginine of FPA. Therefore, we incubated the stock FPs (10 mol/liter final) with activated TAFI (70 nmol/ liter final) 1 h at 22°C. From these experiments we were able to identify FPA cleavage products as well as FPB cleavage products from HPLC analyses (data not shown), indicating that FPA is also a substrate candidate for TAFI, but does not appear during the time course observed for our reactions. In contrast, FPB cleavage to des-Arg FPB occurs almost to completion during this time period. Fig. 6. The quenched time points from 1 to 20 min are illustrated above the Western blot. Clotting occurred at 4 Ϯ 0.2 min (arrow, CT). Fg/Fn is seen in solution (Fg) up to 4 min. At clot time Fg/Fn is almost totally (Ͼ95%) out of solution. These results are similar to what was seen previously in our laboratory on whole blood analysis (24). At 4 min, just prior to clot time there is also evidence of a high molecular weight crosslinked product of Fg/Fn (X-Link). Prior to clot time, thrombin is generated at low (Ն10 nmol/liter) levels. These levels of thrombin are enough to activate fXIII allowing transglutaminase activity to be present by 4 min. Since FPA release is already detected at 3 min, Fn is capable of forming complementary overlapping protofibrils at this time. Another possibility for the soluble cross-linked product is that Fg monomers are cross-linked by fXIIIa. Although from previous studies (50), it is known that Fg undergoes cross-linking more slowly than Fn, the preferred substrate for fXIIIA.

Analyses of Fg/Fn and fXIII Activation in Whole Blood-Analyses of Fg depletion from solution is seen in the Western blots in
Analyses of FPA versus fXIII activation in quenched time points from a whole blood series is depicted in Fig. 7. The Western blot developed using polyclonal ␣-fXIII is illustrated in the inset with time points from 0 to 5.7 min labeled above. Time points extending to 20 min were conducted but by 5.7 min conversion to the activated form of fXIII had reached a maximum. The top band seen in the immunoblot is the A-chain of fXIII (fXIIIA) and the bottom band is seen as the activated form of fXIII (fXIIIAa). A relative fraction of activated fXIII is calculated from this immunoblot. Clot time occurred at 3 Ϯ 0.04 min. The FPA data determined by HPLC methods and plotted as FPA (M) is presented up to 5.7 min, which is also at a maximum at this time. This graph shows a direct correlation (r 2 ϭ 0.9795) between fXIII activation and FPA release. The burst between 4 and 11 M FPA illustrates that fXIII is activated rapidly and coincidentally with FPA cleavage.
Analysis of the Clot for Cross-linking and FPB-The insoluble material (clots) from the whole blood were analyzed for cross-linking on reduced SDS-PAGE gels by direct staining methods ("Experimental Procedures") and Western blotting with monoclonal ␣-FPB. The washed clots were solubilized at approximately 2.5 mg/ml. The results are seen in Fig. 8  Previous studies have inferred that ␥-␥ dimer formation occurs only after Fn II formation, following the release of FPB (51). However, FPB release is just starting to occur at the time that ␥-␥ dimer formation is complete in this study. Therefore, it appears that the formation of cross-links is contemporaneous with clotting and requires only FPA release and initial protofibril overlapping. FPB release does not appear to be a prerequisite for ␥-␥ dimer formation.
The ␤-chain FPB content was also evaluated in these clots using an ␣-FPB monoclonal antibody (supplied by B. Kudryk) (Fig. 8B). The immunoblot is lined up directly below the SDS-PAGE of the cross-linking. Fg and Fn standard are seen to the left and right of the immunoblot, respectively followed by 4 -20min time points. FPB is seen present on the ␤-chain (labeled FPB, B␤-chain). This antibody (␣-FPB) will recognize FPB, des-Arg FPB, and any Fg fragment containing such. These lanes were loaded at ϳ50 ng/lane based upon a calculation for 2 molecules of FPB per Fg chain. The Fg standard shows the most FPB present, since none have been cleaved at this point. The purified Fn standard which should have FPB removed still shows a small percentage of FPB present on the ␤-chain. The samples obtained at and following clot time show that FPB is detected still attached to the ␤-chain. There appears to be little difference between the time points, suggestive that little further release occurs with time. These results are of particular significance since it has been assumed that FPB was removed from the clot in order to allow for lateral aggregation and ␥-␥ dimer formation to occur (51). These data are consistent with the results in Fig. 4 where total combined levels of FPB (des-Arg FPB ϩ FPB, 5 mol/liter) does not equal FPA levels (10 mol/liter) as well as the expected maximum levels of 16 mol/liter.
Summary of Fg Cleavage, FP Formation, and Cross-linking-The results from a whole blood experiment of Fg depletion, FP formation, and ␥-␥ cross-linking (Figs. 4, 6, and 8) are correlated in Fig. 9. Clot time (arrow, CT) is seen at 4 Ϯ 0.2 min. FPA formation is represented as FnI (OE), FPB formation (combined FPB and des-Arg FPB) as Fn II (f), and des-Arg FPB (Ⅺ) are plotted as fibrinopeptides (M) versus time (min). Fg (q) levels in the soluble phase and ␥-␥ dimer (ࡗ) formation in the insoluble phase are plotted as a relative fraction versus time (min) on the secondary y axis. It is clear to see that at clot time, Fn I has already formed (ϳ14 mol/liter) and is crosslinked (ϳ90%). This correlates with the Fg levels that have been depleted from solution (ϳ60%) and is present in the clot. At clot time, Fn II formation occurs after Fg is (Ͼ95%) in the clot and to a lesser extent then Fn I. This is followed by a second process that degrades FPB to des-Arg FPB. DISCUSSION This study represents an account of Fg/Fn processing during the biologically relevant TF induced clotting of non-anticoagulated, warm, whole blood. Previous studies which have been conducted to study Fg/Fn processing have either utilized purified Fg with/without cells (i.e. human umbilical vein endothelial cells or platelets) (2,15,26,52,53) or plasma or blood with chelators (i.e. EDTA, sodium citrate) present (30,33,35,37,38). Studies of blood processing in this manner provides useful insights but leaves open the question of what is actually occurring in native blood. Chelators influence cellular metabolism and numerous plasma protein functions ranging from vitamin K-dependent zymogen and fXIII activation to the cross-linking of Fg/Fn. Additional processes may also come into play with native blood that may not be observed in purified systems. Until recently, studying the biochemistry of the TF induced coagulation of non-anticoagulated whole blood had not been A relative fraction of fXIIIA/fXIIIAa was calculated by densitometry of a Western blot (inset) using a polyclonal antibody to fXIII. This antibody was specific for the unactivated (A) and activated (Aa) ␣-chain. A linear fit was performed on the data, r 2 ϭ 0.9795. The inset shows time points from 1 to 5.7 min, the A-chain of fXIII and its activated form are seen here. The remaining time points were on a second blot (not shown), at which time fXIII is mostly in the active form. feasible. The CTI inhibited whole blood model (34) allows study of the biologically relevant TF pathway under conditions that presumably closely approximate the clotting of native blood. Thus the inhibition of factor XIIa contact activation permits investigation of Fg/Fn reactions under near physiological conditions.
Robust quantitative methods have been developed to study Fg and reaction products formed during the TF-induced clotting of whole blood. Analyses of the cleavage products of Fg and the formation of Fn lead to the following findings. 1) The first cleavage detected, as anticipated, is at A␣-16 releasing FPA molecules with no apparent selectivity for the different forms (P-FPA, FPA, and des-Ala FPA). Quantitative cleavage occurred within 1 min of blood clotting. Initially detectable FPA release occurred when TAT levels were Ͻ2% of maximum (10 nmol/liter). FPA levels reached a maximum of 14 mol/liter (ϳ100%) then decreased with time to a final level of 10 mol/ liter by 20 min. 2) FXIII activation associated with a primary cleavage in the A-subunit at Arg-37 releasing an NH 2 -terminal activation peptide (54), is coincident with FPA release. FPA was first detectable (ϳ9%) when fXIII was already ϳ17% in its active transglutaminase form. 3) ␥-␥ Cross-links are formed prior to and coincident with clotting. Significant FPB removal occurs subsequent to FPA release, but after, clotting and ␥-␥ cross-linking were virtually complete. Lower levels (5 mol/ liter, ϳ33%) of FPB are detected, compared with the expected level. The cross-linked Fn clot contains significant amounts of intact B␤-chain. FPB release is thus not a prerequisite for ␥-␥ cross-linking. 4) Once FPB was released from Fg, a carboxypeptidase-B-like enzyme cleaves the carboxyl-terminal arginine yielding quantitative levels of des-Arg FPB. FPA also contains a carboxyl-terminal arginine but this carboxypeptidase-B-like enzyme appears to be selective for FPB in whole blood. The formation of des-Arg FPB was EACA sensitive.
Our results are summarized in a schematic representation (Fig. 10) of the Fg reaction during the clotting of whole blood. Fn I formation in whole blood begins with the generation of low levels (ϳ10 nmol/liter) of thrombin that simultaneously act upon Fg and fXIII. Thrombin cleaves FPA and exposes sites in the E domain of Fg allowing for complementary overlap of the D domain to form protofibrils. Cross-links are formed either between intact Fg molecules forming dimers or Fg-Fn heteropolymers or soluble cross-linked Fn. Thrombin continues to remove FPAs and coincidentally activates fXIII yielding an initial clot that is a composite of Fn I and Fg with ␥-␥ crosslinks. At this point significant FPB has not been released from the formed clot. Thrombin continues to remove the remaining FPAs and some of the FPBs to produce the final clot, which is composed of Fn I, Fn II with quantitative cross-linked ␥-chains. A carboxypeptidase-B-like enzyme specifically degrades FPB to des-Arg FPB.
Some of the results seen in this study of whole blood Fn formation are predicted from what has previously been reported in less complicated systems. FPA release occurs first followed by release of FPB (Figs. 4 and 5B), as has been shown to occur in numerous studies (1)(2)(3)(4)55). Mechanisms proposed to explain the more rapid cleavage at A␣-16 include kinetic models in which k A␣-16 Ͼ Ͼ k B␤-14 (27,56). In this theory, thrombin is equivalently accessible to both A␣-and B␤-chains and FPA is released faster because the thrombin cleavage rate constant is greater. A contrasting sequential model hypothesizes that FPA release must occur first in order to expose the B␤ peptide site to thrombin (57). Regardless of the model used, released FPA precedes FPB. In purified Fg and thrombin reaction systems, quantitative release of FPA and FPB occur. However, in the whole blood system most B␤ remains intact. Only about ϳ30% of FPB is released over a 20-min time interval. The unexpected early termination of FPB release is interesting since FPB release is terminated when thrombin levels are high (ϳ3 mol/liter) (25). Thrombin, which is also being consumed by AT-III, is still being produced in massive quantities. Therefore, it is difficult to explain why the release of FPB suddenly stops. The termination of B␤-chain cleavage and resulting non-quantitative cleavage is also curious since it has been thought that this cleavage is necessary for normal protofibril and fiber assembly. With FPB release preferentially affecting lateral aggregation (58).
In whole blood fXIII activation is detected prior to clot formation and coincident with the interval when the lowest levels of thrombin are detected. FXIII activation correlates with FPA release (Fig. 7) making transglutaminase activity available to cross-link the overlapped fibrils as soon as they are formed. The cross-linking of Fn by fXIIIa has been thought to be an important step which occurs subsequent to FPB and lateral aggregation, in reinforcing the structure of a thrombus (26). In contrast ␥-␥ cross-linking is nearly complete before FPB is released from the clot (Fig. 9) and is not a subsequent step in forming the thrombus. The presence of clot cross-links prior to the release of FPB suggests that lateral aggregation upon release of FPB is not required for transglutaminase activity to begin. The cross-linking of the clots prior to the release of FPB might conceivably suggest the creation of an environment which traps FPB in the formed Fn II polymers. This possibility is ruled out by immunoblotting which detects most FPB still attached as the B␤-chain in the clots even at the 20-min time point (Fig. 8B). Therefore, the exposed binding site of FPB (G-H-R) suggested to allow for lateral association (59) may occur only to a small degree. Clot organization in terms of FPA (Fn I) and FPB (Fn II) have been studied previously (15). Fibrin deposition has been directly correlated to FPA not FPB release at the onset of gelation (57). These results are comparable to the whole blood studies seen here, where Fn I formation (FPA release) appears to be crucial to Fn deposition versus Fn II (FPB release).
FPB cleaved from Fn I in the clot is cleaved by a carboxypeptidase-B-like enzyme that produces des-Arg FPB (Figs. 3 and  4). Des-Arg FPB has been previously observed but was thought to arise from the action of an irrelevant carboxypeptidase contaminant of assay samples (31,32). Des-Arg FPB has also been identified in plasma (60). One study analyzed conversion of FPB to des-Arg FPB in anticoagulated blood by adding carboxypeptidase-B and measuring the levels produced (38). The natural enzyme that causes this cleavage in our study has not been identified, apart from its inhibition by EACA (Fig. 5). Several carboxypeptidases with similar cleavage specificity to carboxypeptidase-B are capable of removing the carboxyl-terminal arginine. Carboxypeptidase-N (47) has been proposed to serve as a regulator for several blood peptides including peptides (FDP-6A and -6D) released from Fg/Fn in the initial stage of plasmin degradation (61) and EACA has been shown to be an inhibitor of carboxypeptidase-N (62). The TAFI or procarboxypeptidase-U is activated by thrombin-thrombomodulin and is thought to contribute to clot stability by interference in the plasmin dissolution of a clot by removing terminal lysines required for plasmin binding. It has also been identified as being cross-linked to fibrin during the latter part of the coagulation cascade (63). Previously published results show that EACA is a competitive inhibitor of TAFI (64) and an inhibitor of fibrinolysis (65). In our studies with purified Fg, thrombin, thrombomodulin, TAFI provides an activity with the appropriate specifications. From our whole blood studies with EACA addition (Fig. 5) and our in vitro studies with TAFI acting on Fg, and purified FPA and FPB, it is tempting to suggest that the agent responsible for the formation of des-Arg FPB is TAFI. The pattern of selective cleavage of FPB was observed in Fg experiments, although TAFI was also capable of cleaving isolated FPA. However, preliminary experiments using monoclonal ␣-TAFI (at 1.5 and 3 mol/liter final), which inhibits the formation of activated TAFI (41), was unable to prevent the cleavage of FPB. Potato carboxypeptidase inhibitor (1 mol/ liter) which has been shown to inhibit TAFI (64) was also unable to inhibit the cleavage of FPB. Overall, the significance of the cleavage specificity toward FPB and the identification of the carboxypeptidase-B-like enzyme can be a link between the balance of forming a thrombus and fibrinolysis and remains to be identified. Levels of FPA in whole blood/serum are decreased over time. The final levels of FPA detected at 20 min are ϳ30% less then those seen at 4.5 min. This observation was also seen in a whole blood study (25) in which FPA was measured by immunoassay. Together these observations suggest that there are also FPA degradation processes occurring. FPA also contains a carboxylterminal arginine that could be susceptible to cleavage by a carboxypeptidase-B-like enzyme. The main difference between FPA and FPB is that the majority of FPA is removed from the clot, followed by a slow degradation process. Whereas, FPB is released in small amounts and is degraded immediately.
This study leads to an insight into what is seen in whole blood when initiated with a small amount of TF. The techniques developed here are sufficiently robust for quantitation of FPs and soluble and insoluble Fn products in media as complicated as blood. From these results and using these types of analyses we are able to compare results that are seen in the synthetic/reconstituted models developed in our laboratory (66,67). These studies are conducted in the hopes to eventually explaining blood coagulation using non-invasive techniques in normal and diseased states. FIG. 10. Summary of whole blood clotting. At the onset of clot formation, thrombin simultaneously acts upon Fg (D-E-D) and fXIII. FPA (ϳ30 -40%) is released from the molecules allowing for complementary overlap of the exposed sites in the E domain with adjacent D domains from another Fg molecule. FXIII is being activated (fXIIIa) at approximately the same rate. The subsequent formation of the initial soluble Fn is seen to be cross-linked (DϭD). Thrombin continues to activate fXIII and cleave the remaining FPA molecules, yielding an initial clot that is a composite of Fg, Fn, and ␥-␥ cross-links, with FPB still attached to the B␤-chain. The initial clot is continuously acted upon by thrombin to release the remaining FPAs and some of the FPBs to yield a final clot. The released FPB is selectively acted upon by a carboxypeptidase-B like enzyme cleaving the carboxyl-terminal arginine to produce des-Arg FPB. Together these results suggest that Fn I, Fn II, and cross-linking are not seen as separate processes but appear to occur simultaneously.