Factor V Is a Substrate for the Transamidase Factor XIIIa*

Coagulation Factor V (Mr = 330,000), upon cleavage by thrombin, produces Factor Va, which is composed of two subunits with Mr values of 94,000 and 74,000, along with two activation fragments possessing no known function. Studies were undertaken to assess the ability of the transamidase Factor XIIIa to covalently incorporate the lysine analogs [3H]putrescine and dansylcadaverine into the thrombin-cleaved (activated) and unactivated forms of human and bovine Factor V. The incorporation of either probe into thrombin-activated Factor V proceeded at an initial rate approximately twice that for unactivated Factor V. The extent of the incorporation of [3H]putrescine or dansylcadaverine into activated or unactivated human Factor V was identical; 4 mol of either probe per mol of Factor V. In the case of bovine Factor V, however, while 4 mol of probe were bound per mol of the unactivated pro-cofactor, 5 mol of either lysine analog were covalently linked to 1 mol of thrombin-cleaved Factor V. Polyacrylamide gel fluorography, immunoaffinity chromatography, and immunoprecipitation identified the largest activation fragment of human Factor V (Mr = 150,000) and bovine Factor V (Mr = 120,000) to contain the sites of incorporation of the covalently bound probes. High molecular weight, apparently covalent polymers of Factor V were produced by the action of Factor XIIIa on activated and unactivated human or bovine Factor V. The absence of either probe in the reaction mixtures did not appear to allow an enhancement of protein polymerization.

= 150,000) and bovine Factor V (M, = 120,000) to contain the sites of incorporation of the covalently bound probes. High molecular weight, apparently covatent polymers of Factor V were produced by the action of Factor XIIIa on activated and unactivated human or bovine Factor V. The absence of either probe in the reaction mixtures did not appear to allow an enhancement of protein polymerization.
Transamidases (transglutaminases) are calcium-dependent enzymes that catafyze an acyl transfer reaction between the y-carboxamide group of a peptide-bound glutamine residue and certain primary amines such as putrescine, cadaverine, histamine, or lysine residues of peptides (1,Z). These enzymes have been reported to cross-link keratin in hair follicles and epidermal cells (3, 4). The calcium-induced aggregation and cross-linking of erythrocyte membrane proteins have also been attributed to a transglutaminase (5,6 ) .
Plasma Factor XIIIa is a transamidase responsible for cross-linking aggregated fibrin monomers (7,8), fibrin to fibronectin (91, fibrin to von Willebrand Factor,' fibronectin to collagen (lo), and az-plasmin inhibitor to fibrin (11). Platelet Factor XIIIa has been postulated to function in cross- linking membrane proteins to cytoskeletal proteins such as actin and myosin (12). Activation of plasma Factor XIII, a tetrameric protein (a2b2), proceeds with the initial thrombincatalyzed release of a 36-amino-acid activation peptide from the NW, terminus of each "a" chain (13). The calcium-dependent dissociation of the "b" chains from the dimer of modified chains follows, unmasking each active center thiol group (14,15). Platelet Factor XIII, existing only as an "a" chain dimer, is also activated following thrombin cleavage and a subsequent conformational change (15).
Bovine Factor V exists as a single-chain plasma and platelet protein with a relative molecular weight (lw,) of 330,000. Upon activation with thrombin, the pro-cofactor is proteolytically processed, producing fragments with M, values of 120,000, 94,000, 74,000, and 71,000 (16). The active form of Factor V, Factor Va, is composed of the noncovalently associated M, = 94,000 and 74,000 peptides which represent the amino and carboxyl termini (respectively) of the pro-cofactor molecule (26). Factor Va, along with Factor Xa, platelet membrane, and/or certain phospholipi~, forms the thrombin-generating prothrombinase complex (17). The remaining fragments of the thrombin-cleaved Factor V (app~ximately 50% of the pro-cofactor) are activation peptides of M , = 120,000 and 71,000 in bovine Factor V and 150,000 and 71,000 in human Factor V (18,19). Previous studies have shown that once Factor V is cleaved by thrombin, the M , = 94,000 and 74,000 chains of Va bind to lipid membranes, while the activation peptides fail to do so (20)(21)(22). It is also known that full prothrombinase activity can be produced in the absence of the activation peptides of Factor V (23). The substrates of Factor XIIIa all possess elongated conformations (24-28) and are either associated with platelet membrane or are secreted by stimulated platelets (27,(29)(30)(31). Since it has been known that Factor V has an elongated shape (32, 33) and that it has been found in platelets (34), it was decided to evaluate the ability of Factor XIIIa to use Factor V as a substrate.
We present evidence which indicates that both human and bovine Factors V possess a region that contains accessible glutamine residues for the transamidase incorporation of two lysine analogs by Factor XIIIa. The incorporation of either 13H]putrescine or dansylcadaverine identified only the large activation peptide to contain the glutamine residues accessible for transamidase cross-linking. Evidence is also presented to suggest the formation of high molecular weight stable crosslinked products of Factor V.
Bovine Factor V was isolated by the procedure of Nesheim et al. (35), human Factor V according to Katzmann et ai. (18), and CYthrombin was purified according to Lundblad et al. (36). Human Factor XI11 was isolated from plasma as previously described by Skrzynia et al. (37).

Methods
Activation of Factor XIII-Twenty five pl of a 200 pg/ml Factor XI11 solution were added to 100 pl of Trislsaline buffer (0.02 M Tris-HCl, 0.15 M NaCl, pH 7.4). Six pl of 1.0 M CaC1' were added, as well as 5 pl of 50 units/ml a-thrombin. The solution was heated at 37 "C for 20 min in order to activate Factor XIII. Ten pl of 1 mM PPACK, a thrombin inhibitor, were added to one-half of the incubation mixtures. After 5 min at room temperature, 15 pl of [3H]putrescine (15 Ci/mmof, 2.0 X 10" M) or 20 pl of 2 mM dansylcadaverine, 12 pl of 0.2 M dithiothreitol, and 125 pl of Factor V (0.6 mg/ml in Tris/saline) were added. Incubations were performed in a 37 "C water bath.
Incorporation of PHfPutrescine-An analytic procedure by Lorand et al. (1) was employed to assess the incorporation of ['HJputrescine into Factor V. At various times, 30 pl of each sample were placed on discs of Whatman No. 3MM filter paper and immersed in cold 10% trichloroacetic acid for 20 min. The discs were then transferred to 5% acid, soaked for 10 min, and washed three times each in ethano1:acetone (1:l) and 100% acetone. They were air-dried and counted in a Beckman scintillation counter.
Incorporat~n of Dansykadauerine-At the same times used for the [3H]putrescine incorporation, the samples were each transferred to a quartz cuvette and the fluorescence emission at 500 nm was measured using a Perkin-Elmer MPF-44B fluorescence spectrophotometer. Excitation was set at 360 nm. The incorporation of dansylcadaverine exhibits an overall fluorescence emission enhancement, as well as a blue shift of maximum emission (546 + 515 nm) (2). In order to quantitate the amount of dansylcadaverine incorporated at apparent saturation, the sample was eluted through a P6 (gel filtration) resin (Bio-Rad) to separate unbound probe from that bound to Factor V. The protein concentration was calculated using the extinction coefficient (E&%,,=) of 9.6 (35) and the dansylcadaverine concentration using of 4,300 M" (38). For the dansylcadaverine absorbance measurement, the samples were first made to 6 M guanidine HCl. The dansylcadaverine eluting in the void volume of P6 with Factor V was confirmed to be covalently bound, due to the fact that no fluorescence was observed in the dye front of sodium dodecyl sulfate gel analysis of the samples.
Polyacrylamide Gel Electrophoresis (PAGE)-Protein samples were analyzed by slab gel electrophoresis in the presence of sodium dodecyl sulfate (SDS) according to Laemmli (39). Electrophoresis was performed on 5-15% polyacrylamide gels. After incubation, each sample was dialyzed against 20 mM Tris-HC1, pH 7.4, and lyophilized. Dried samples were dissolved in 0.024 M Tris-HC1, pH 6.8, containing 1% SDS, 10% glycerol, and 0.002% bromphenol blue and heated at 95 "C for 5 min. Proteins were visualized in the gels by staining with Coomassie Blue R-250.
[3H]Putrescine-labeled protein was identified by fluorography after soaking the gel in an enhancing solution (DuPont), vacuum-drying, and placing it at -80 "C against Kodak XAR-1 film for 24-48 h.
Immunoaffinity Chromatography-A monoclonal antibody that binds the 74,000-dalton chain of human and bovine Factor Va (40) was immobilized onto Sepharose CL-4B and used to survey the distribution of [3H]putrescine incorporation into human and bovine Factor V. The samples were loaded onto the 1-X IO-cm column and the matrix was washed with Tris/saline buffer containing 5 mM CaC12. After the Am values of the fractions returned to baseline, Tris/saline, 5 mM CaClz with 1.5 M NaCl, was introduced. Fifty pl of each 1.0-ml fraction were placed into the scintillation counter. The matrix was regenerated by washing with starting buffer.
Immunoprecipitation-Immunoprecipitation studies, conducted in order to immunochemically identify any probe-incorporated peptides, were performed according to Foster et al. (41). After incubation, [$HI putrescine samples were diluted as necessary in radioimmunoassay buffer (0.075 M Tris-HC1, 0.075 M NaCl, 1% Triton X-100, pH 7.0, containing 10 mg/ml bovine serum albumin). Reaction mixtures were The abbreviations used are: PPACK, D-phenylalanyl-L-prolyl-Larginine chloromethyl ketone; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. prepared by incubating 0.2 ml of antigen solution with 0.2 ml of purified murine monoclonal immunoglobul~n (0.1 mg/ml in radioimmunoassay buffer) and incubated at room temperature for 30 min after which 0.05 ml of rabbit anti-mouse antiserum was added to each tube. After 30 min, the immunoprecipitates were collected by centrifugation, washed once with radioimmunoassay buffer, and dissolved in 0.2 ml of 0.2 N acetic acid. Thirty pl of each dissolved pellet were counted for radioactivity.

RESULTS
Lysine analogs such as [3H]putrescine and dansylcadaverine are useful when attempting to identify peptides containing glutamine residues that are accessible to a transamidase such as Factor XIIIa. The [3H]putrescine and dansylcadaverine incorporation time courses are shown in Fig. 1. The incorporation of either probe into the Factor V pro-cofactor proceeded at a rate approximately one-half that of thrombin-cleaved Factor V. Since thrombin is required for Factor XI11 activation (13) and Factor V is cleaved and activated by thrombin (16), the thrombin inhibitor PPACK was used to preserve the status of pro-cofactor (Factor V) in some of the samples.
At the end points of the incubation time courses, the amounts of [3H]putrescine radioactivity bound to protein on filter paper indicated that 4 mol of putrescine were incorporated into 1 mol of human pro-cofactor or thrombin-cleaved Factor V. While 4 mol of putrescine were bound per mol of bovine pro-cofactor V, the data indicated that 5 mol of probe were bound per mol of thrombin-cleaved bovine Factor V peptide mixture. It appears that the thrombin cleavage of bovine Factor V reveals 1 more glutamine residue accessible for trans~idation, while the activation of human Factor V produced no additional putrescine incorporation sites. In order to test the possibility that PPACK might somehow have caused the decreased rate of Factor XIIIa activity in the thrombin-inhibited samples (Fig. l), parallel incubations of Factor XIIIa with dimethylcasein (Sigma) were performed. The presence of PPACK did not retard the rate of dansylcadaverine incorporation into casein (data not shown), indicating PPACK to lack influence on Factor XIIIa function.
SDS gels were used to resolve the pro-cofactor and the thrombin activation products of Factor V. The gels revealed that the presence of PPACK had prevented significant degradation of the pro-cofactor throughout the 20-h incubations (Fig. 2). Fluorographs of the gels (Fig. 2) showed that with the exception of radiolabeled material retained at the top of the 3% stacker gel, the only [3H]putrescine-labeled material existed with M, = 120,000 in thrombin-cleaved bovine Factor V, 150,000 in cleaved human Factor V, and 330,000 in PPACK-treated human and bovine Factor V samples. The data show that in each situation, the rate of incorporation of probe into activated Factor V was approximately twice that of the pro-cofactor. In following dansylcadaverine incorporation, the relative fluorescence intensity was measured with an emission of 500 nm and an excitation of 360 nm. The designation Va on the right ordinate axis refers to thrombin-activated Factor V peptide mixtures. completely destained. These "ghosts" have been understood to be the carbohydrate-rich, non-Coomassie Blue-staining, large activation fragments.  In contrast to human Factor V, the antibody column binds only the two chains of bovine Va and not the bovine procofactor V molecule. Fig. 3B illustrates that appropriately all of the radiolabeled material in the bovine procofactor V samples failed to bind to the column, as did the labeled component of thrombin-cleaved bovine Factor V samples. The antibody matrix did however retain the M, = 94,000 and 74,000 chains of bovine Va in the thrombin-activated samples and the bound Factor Va was shown to lack significant radioactivity. Thus, the elution of [3H]putrescine-labeled bovine pro-cofactor and thrombin-activated Factor V demonstrated the absence of f3H]putrescine incorporated into the Factor Va (M, = 94,~0-74,000) complex (Fig. 3B).

Time, hours
Since the SDS-PAGE fluorescence, fluorographic, and immunoaffinity evidence indicate that the large activation fragment. of human and bovine Factor V to exclusively possess the incorporated probes, immunoprecipitation employing murine monoclonal antibodies to the activation peptide itself were included in this study. Table I  fied at least 90% of the [3H]putrescine-labeled protein in each activated Factor V solution to be the large activation peptide. These results indicated that the incorporation of putrescine did not significantly inhibit antibody binding to the respective epitopes and that the site of incorporation is the large, poorly staining midregion of the Factor V molecule. Fig. 4 illustrates the arrangement of the Factor V peptides produced by thrombin cleavage. The M, = 94,000 and 74,000 chains of Factor Va bind to phospholipid vesicles or platelets (20) and form a high affinity receptor for Factor Xa (17). The fate or function of the Factor V activation peptides has not been elucidated. Since only 51% of the Factor V molecule is employed in the expression of Factor Va in the prothrombinase complex, it would not seem unreasonable to consider there to be another role(s) for the relatively significant amount of protein represented by the activation peptides.

DISCUSSION
Factor V exists as an asymmetric molecule with a relatively large axial ratio ( s~~,~ = 9.2) (32) and is found in platelets, as well as plasma (34). Since most of the substrates for Factor XIIIa are also elongated in conformation (24-28) and exist in platelets (27,(29)(30)(31), Factor V was tested as a potential substrate for Factor XIIIa through the use of the lysine analogs [3H]putrescine and dansylcadaverine. Factor V was demonstrated to contain glutamine residues accessible to transamidation by Factor XIIIa. When probe-inco~orated, thrombin-activated human or bovine Factor V was analyzed by gel electrophoresis and fluorography, the incorporated material was observed to exist in primarily two forms. In human and bovine Factor V samples, all had probe-labeled, polymerized material that failed to enter the 3% stacker gel.
In the thrombin-inhibited Factor V samples, the majority of the labeled protein existed with an M , = 330,000. The thrombin-cleaved samples produced probe-incorporated material with M, values of 150,000 in human and 120,000 in bovine samples.
Since the denaturing polyacrylamide gels revealed a fraction of the radiolabeled material in each sample to be too large to enter the 3% stacker gel, it seems apparent that, as is the case with fibronectin (45) and von Willebrand Factor,'  ' nHFV, anti-human Factor V. and 74,000) that comprise Factor Va, and two activation peptides designated C1 and F. It is the large, heavily glycosylated C , peptide that apparently contains the only glutamine residues of Factor V that are accessible for transamidation.
the Factor XIIIa-induced covalent polymerization of Factor V (330,000) or its large activation peptide is not completely favored. In order to be a suitable substrate for Factor XIIIa, a protein must possess glutamine residues accessible for transamidation. In order for a protein to be cross-linked in homopolymers, it must also furnish lysine residues that align with the acyl-doning glutamine groups when the monomers are arranged in an extended configuration. In the presence of Factor XIIIa, the a and y chains of fibrin are linked differently. The a chains are cross-linked into large polymers, while the y chains orient themselves in an antiparallel manner and are cross-linked into dimers (46). This isologous arrangement of the y dimer supposedly involves all substrate lysine and glutamine residues; thus, no more are available for further cross-linking. If the latter behavior was the case for Factor V or the large activation peptides, the dimerized proteins would be visualized in the stacker gel (pro-cofactor molecule), or with an M, = 220,000 to 300,000 in the resolving gel (large activation peptides). Although dimerization was not apparent, probe-labeled material retained in the upper portion of the 3% stacker gel demonstrated the presence of high molecular weight covalent polymers. The existence of the polymeric Factor V material indicated that Factor XIIIa had cross-,inked pro-cofactor V and the large activation peptide into arge homopolymers.
The presence of the lysine analogs at the concentrations rsed in these studies did not prevent Factor V p o~y m e r i~t i o n rpon incubation with Factor XIIIa, as evidenced by Coomasie Blue R-250 stained patterns of SDS gels containing Factor 7 and thrombin-activated Factor V, with and without ['HI ,utrescine or dansylcadaverine. That Factor XIIIa can covamtly polymerize Factor V in the presence of the lysine nalogs suggests that there are lysine residues that can, to an xtent, compete with the probes for coupling to the accessible lutamine residues. Factor V does not appear to self-associate noncovaiently at plasma concentrations. Thus, the observation of Factor V cross-linked polymers raises the question of its crosslinking to other Factor XIIIa substrates. The maximum incorporation of either probe did not prevent the proteolytic processing of human or bovine Factor V by thrombin (although some large polymerized material remained after thrombin treatment). This suggests that the regions of Factor V that are involved in thrombin recognition and binding probably do not contain glutamine groups that were accessible for cross-linking to the probes.
The lack of incorporation of the probes into either chain of Factor Va is not an unusual situation. The substrate range for Factor XEIIa, as well as other transamidases, is relatively narrow (1, 2). With many proteins, incorporation of a lysine analog by a transamidase proceeds only after the substrate's native conformation has been disrupted by chemical modification such as succinylation (2).
Factor XIIIa has been shown to cross-link fibrin into homopolymers, fibrin to a,-plasmin inhibitor, and fibrin to fibronectin, as well as cross-link fibronectin to collagen (7-Il).' A recent report' demonstrated plasma von Willebrand Factor to be a substrate of Factor XIIIa and capable of being cross-linked to fibrin during gel formation. Another recent study reported that human Factor V and guinea pig megakaryocyte Factor V are cross-Iinked by Factor XIIIa (47). The von Willebrand Factor, as well as fibronectin, a*-plasmin inhibitor, and t.hrombospondin, have all been observed to cross-link to fibrin (9,11,46).' All of these blood coagulation proteins are believed to contribute to the stability of the plate~et-~brin-enodothelium association. Since all of these Factor XIIIa substrates contribute to the construction of blood clots, it is conceivabje that the Iarge activation peptide of Factor V, now demonstrated to be a substrate of Factor XIIIa, may also be covalently incorporated into the fibrin network. This conjecture is the subject of 8 continuing investigation in this laboratory.
A c~n o~~d g m~n~~-W e are grateful to Drs. Paula Tracy and Robert Viskup for their vatuable ~ntributions of human Factor V.