Factor XIII Val34Leu polymorphism and γ-chain cross-linking at the site of microvascular injury in healthy and coumadin-treated subjects

Summary

Fibrin cross-linking by activated factor (F)XIII is essential for clot stability. In vitro, a common Leu34 polymorphism of the FXIII A-subunit increases the rate of thrombin-mediated FXIII activation, but not cross-linking activity upon complete FXIII activation. The effect of FXIII Val34Leu polymorphism on fibrin(ogen) cross-linking in vivo when vascular injury triggers the blood coagulation has not been studied yet. Using quantitative immunoblotting with antibodies raised against FXIII A-subunits, fibrinogen and γ-γ-dimers, the rates of FXIIIA cleavage and fibrin(ogen) cross-link formation in the fluid phase of 30-second blood samples collected at the site of microvascular injury were compared in the Leu34-positive and −negative healthy individuals and patients on long-term oral anticoagulation. In addition to accelerated FXIII activation, in healthy subjects the presence of FXIII Leu34 allele was associated with increased soluble γ-γ-dimer formation by 40% (1355±17 μg/L for Leu34 carriers vs 804.3±17 μg/L for Leu34 non-carriers; p=0.028) at the site of microvascular injury. This solution phase effect was abolished in coumadin-treated patients (369.4±75.9 μg/L for Leu34 carriers vs 290.5±35.9 μg/L for Leu34 non-carriers; p>0.05). The present study indicates that the Leu34 allele affects soluble γ-γ-dimer formation in untreated individuals, but not in those receiving acenocoumarol. Our data may help elucidate the impact of the FXIII Val34Leu polymorphism on fibrin crosslinking in vivo and its modulation by oral anticoagulants.

Introduction

The final steps in the blood coagulation process, fibrinogen (Fbg) conversion to fibrin (Fn) and Fn monomer cross-linking by activated factor (F) XIII involve a sequence of complex enzymatic reactions leading to the formation of a hemostatic plug, which is relatively resistant to mechanical and enzymatic degradation (see for review 1,2).

The Fbg molecule, comprised of three pairs of polypeptide chains, denoted Aα, Bβ, and γ, is a principal substrate for thrombin. Thrombin-mediated release of fibrinopeptide A (FPA) and –much slower - B (FPB) from the amino-termini of the Aα- and Bβ-chains of Fbg, respectively, results in the formation of Fn monomer with the structure (α, β, γ)₂ that can polymerize via noncovalent interactions between the D domains and the central E domains of Fn monomers, resulting in double-stranded fibrils that associate laterally forming thick fibers. The noncovalent Fn polymer is stabilized by FXIIIa. A protransglutaminase, FXIII occurs in plasma as a tetramer A2B2, while in platelets only the A-subunit homodimer is present. Active FXIIIa is formed by thrombin cleavage at Arg37 in the A-subunits with subsequent calcium-mediated dissociation of B-subunits, that serve as a carrier of the catalytic A-subunit and contributes to Fbg binding of FXIII (3,4). There is evidence that thrombin-catalyzed cleavage of one A-subunit is sufficient to induce full transglutaminase activity in the presence of Ca⁺⁺ (3). Factor XIII activation is enhanced (about 100-fold) by Fbg and Fn, largely due to acceleration of the proteolytic cleavage of the 4 kD activation peptide (3). FXIIIa first catalyzes the formation of ɛ-amino(γ–glutamyl)lysine isopeptide bonds at the D domains between associated fibrin units, rather than between the soluble Fbg molecules (2). Cross-links can be formed either between the intact Fbg molecules, giving rise to γ-γ–dimers or Fbg-Fn heteropolymers, soluble and polymerized Fn. The covalent cross-linking of γ–chain carboxyl-terminal fragments occurs between lysine residues at position 406 of one chain and glutamine residues at position 398/399 of another (3,4). Intermolecular cross-linking between α chains in their mid- and carboxyl-terminal portions results in a much slower oligomer and polymer formation than γ-dimerization. Other cross-links, γ–multimers are formed through interfibril band formation within hours to days, while minutes to hours are necessary in vitro to create intramolecular cross-links between the carboxyl-termini of α- and γ-chains, the fibrin α–γ–heterodimers (1,2). Fibrin α–polymers and cross-links between γ–dimers, leading to the formation of trimers and tetramers, results eventually in the mature fibril network of the fibrin clots (2). FXIIIa also mediates cross-linking of fibronectin and α₂-antiplasmin to the fibrin α–chain(5). Plasminogen bound to Fn is activated by plasminogen activators to form plasmin which cleaves Lys-Arg bonds in the Fn molecules (1). However, FXIIIa-mediated Fn binding of α₂-antiplasmin facilitates the inhibition of plasmin and increases fibrin clot resistance to lysis (2).

The proper balance between Fn formation, its cross-linking and degradation is necessary to protect the vascular system from excess blood loss and also from obstructed blood flow. It is widely believed that efficient hemostasis occurs only at vascular lesions where tissue factor (TF) is exposed, platelets rapidly aggregate, thrombin is formed, FXIII is activated, and fibrin is anchored to the vessel wall (1,2). From previous studies (6,7) we have shown that high molecular weight cross-linked products are in solution prior to visible clot formation. To the best of our knowledge, there are scant data on the actual pattern of FXIII activation and cross-linking of fibrin(ogen) in vivo.

Factors that can modulate fibrin(ogen) cross-linking can be genetic and environmental. The most extensively studied modulator has been a common G→T mutation in the codon 34 of the FXIII A-subunit gene that results in a valine to leucine replacement (8). The Leu34 allele, which is present in approximately 25 % of Caucasian populations (4), has been reported to protect against not only thromboembolic events, such as myocardial infarction (7), or stroke (9), but also venous thrombosis (10,11), although in some populations, a prevalence of the Leu34 allele was not lower among patients with arterial or venous thrombosis than in healthy controls (12). Furthermore, it has been suggested that the Leu34 allele represents a risk factor for intracerebral hemorrhages (13). Paradoxically, the Leu34 allele has been demonstrated to accelerate FXIII activation two- to three-fold in vitro compared with the Val34Val variant (4). Recently, we have shown that the Leu34 allele increased FXIII activation in vivo at the site of microvascular injury (14). In purified Fbg-FXIII-thrombin-calcium systems, the Leu34 allele has been associated with enhanced γ–dimerization, although quantitative analysis of this reaction has not been reported (15,16). Accelerated activation of the FXIII Leu34 variant is associated with inhibited lateral aggregation of fibrin fibers and results in the formation of clots with thinner fibers that are less porous and presumably less susceptible to lysis (4). It is unclear how the FXIII Val34Leu mutation affects γ-γ–dimer formation in the blood in contact to the surface of damaged vascular wall in vivo. Additionally, the effect of reduced thrombin generation, as a result of oral anticoagulation, on FXIII activation and γ-γ–dimer formation in relation to the FXIII Leu34 polymorphism has not yet been reported.

To address these issues, we adopted a model of blood coagulation at the site of microvascular injury that involves the contribution of blood cells, platelets, and more importantly, components of the damaged vessel wall, especially endothelial cells and the subendothelial region. This model offers unique opportunities for the qualitative and quantitative evaluation of blood coagulation triggered by exposure of TF following vascular injury (17,18) that mimicks the in vivo conditions. The data presented here indicate that the FXIIIVal34Leu polymorphism enhances soluble γ–chain cross-linking in vivo, while oral anticoagulation abolishes this effect. These findings extend our understanding of a role of the Leu34 polymorphism in clot formation both in healthy and coumadin-treated subjects.

Methods

Subjects

Blood samples were obtained from 8 apparently healthy individuals (2M, 6F), aged 31–46 (mean, 40) years, with no history of thrombotic or bleeding disorders. Subjects with overt atherosclerotic vascular disease were excluded. None of the individuals took any medication during a previous month prior to the study. Coagulation profiles were normal: Fbg, measured by nephelometry (Dade Behring, Marburg, Germany), 2.57±0.14 g/L (±SEM); prothrombin time (International Normalized Ratio [INR], 1.05±0.02), and activated partial thromboplastin time (aPTT), 33.5 ±0.86 s.

To determine the effect of oral anticoagulation on cross-linking, we analyzed samples obtained from 8 patients (7M, 1F), aged 29–67 (mean, 54) years, who were taking acenocumarol (average, 20.5 months) for prevention of recurrent venous thromboembolism (n=6) or atrial fibrillation (n=2). Patients with diabetes, renal insufficiency, cancer, history of stroke or myocardial infarction were excluded from the study. On the day of the study, the results of Fbg, INR, and aPTT were the following: 2.7±0.19 g/L; 2.35±0.16; 42.6±2.2 s, respectively. The study was approved by the University Ethical Committee and all subjects gave informed written consent.

Model of microvascular injury

Evaluations of coagulation products were performed in the blood samples collected at 30-second intervals from forearm skin incisions, made using a Simplate II device (Organon Teknika, Durham, NC) at the inflation of the sphygmomanometer cuff at 40 mm Hg, as previously described in detail (16,18). Bleeding time, expressed as a mean of two values from two incisions, was similar in healthy and coumadin-treated individuals (medians, 400 and 465 seconds, respectively). Blood was collected by means of heparinized tubes (Kabe Labortechnik, Numbrecht-Elsenroth, Germany) into Eppendorf tubes containing anticoagulants according to the procedure described previously (16,18). After an immediate centrifugation at 3000xg at 4 °C for 20 minutes, supernatants were aliquoted (one portion mixed with SDS) and frozen at −80 °C; the tests were performed on freshly thawed samples. We measured levels of thrombin-antithrombin complexes (TAT) in the blood sample collected at the last minute of bleeding, using a commercially available ELISA kit (Dade Behring, Marburg, Germany).

Immunoblotting

Following electrophoretic separation using 4–12% gradient gels and transfer to nitrocellulose membranes (Bio-Rad), quantitative Western blotting was performed, as described previously (16,18,19).

For immunoblotting, we used mouse monoclonal antibodies kindly provided by B. Kudryk from the New York Blood Center (NY): antibody 2G10-1 (immunogen –Fn γ-γ-dimer; final concentration 5 μg/ml in TBST) cross-reacting with γ15-35; antibody Fgn 3A and Fgn 2e (final concentration 2.5 μg/ml in TBST) both reacting with the Aα-chain and the α-chain of human Fbg as determined by Western analyses(5). A rabbit polyclonal antibody raised against human FXIIIA (D4679) was generously provided by Dr. Paul Bishop (ZymoGenetics, Seattle, WA) (5). Horseradish peroxidase-labeled secondary IgG antibodies were purchased from Southern Biotech (Birmingham, AL).

Concentrations of FXIIIA/Aα and γ-γ-dimers were estimated by comparisons with purified human FXIII (gift of Haematologic Technologies, Essex Junction, VT) and in-house prepared cross-linked fibrin standards, respectively, as described previously (18,19). Band intensities from Western blots were analyzed densitometrically by the National Institutes of Health Image Program.

Genotyping

The FXIII Val34Leu polymorphism was determined by the polymerase chain reaction followed by the restriction fragments length polymorphism analysis (20).

Statistical analysis

Data are presented as mean ± SEM or otherwise stated. Intergroup differences were analyzed using the Mann-Whitney U test. Spearman’s correlation rank test was used to evaluate relationships between variables. A p-value <0.05 was considered significant.

Results

FXIII activation

We studied four healthy Val34Leu heterozygotes and four healthy Val34Val homozygotes. The density of a band of a M_r= 83 kD, migrating identically with the purified FXIII A-subunit on immunoblots, decreased gradually in blood obtained at the site of microvascular injury and became undetectable somewhat faster in the Leu34 carriers as compared to the Val34Val individuals (112.5±18.9 vs 165±26.1 s; p=0.1). The initial concentration of FXIIIA was similar in both groups (93.8±0.51 and 92.9±0.42 nmol/L, respectively). The rate of FXIIIA removal from the fluid phase of blood samples obtained within the first 2 minutes was markedly higher in the Leu 34 carriers than in the Val34Val homozygotes (1.5±0.5 vs 0.27±0.1 nmol/L/s; p=0.03; Fig.1A), which corroborates our previous findings (14) and in vitro studies (15,16). The time of detection of FXIIIAa (M_r =79 kD) was significantly accelerated in the Leu34 carriers compared with individuals homozygous for the Val34 allele (97.5±18.9 vs 195±26 s; p=0.04). Moreover, the rate of increase in FXIIIAa concentrations was higher by 43% in the former than in the latter group (0.44±0.06 vs 0.25±0.04 nmol/L/s, respectively; p=0.02; Fig.1C).

Fig. 1.

Time courses of the removal of factor (F)XIIIA (closed symbols) and appearance of FXIIIAa (open symbols) measured in consecutive blood samples obtained every 30 seconds at the site of microvascular injury using quantitative immunoblotting (see Methods) in healthy individuals (panel A and C) and anticoagulated patients (panel B and D) both homozygous for the Val34 allele (circles) and those heterozygous for the Leu34 allele (squares). Data are expressed as medians. The rate of these processes was significantly higher in the Leu34-positive subjects than in the Leu34-negative individuals.

We studied five Val34Leu heterozygotes and three Val34Val homozygotes treated with acenocoumarol. The time of disappearance of FXIIIA from blood collected at the site of injury in the former and latter group (170±26.4 vs 144±22 s; p>0.05). This was also true for the time of detection of FXIIIAa in the same samples (108±27.8 vs 160±26.5 s; p>0.05). In anticoagulated individuals, FXIIIA levels within the first 30 seconds after vascular injury were similar to those calculated for healthy subjects and showed no difference related to the presence or absence of the Leu34 allele (92.7±0.52 vs 92.3±0.48 nmol/L, respectively). The maximum velocity of decrease in FXIIIA levels in the blood obtained from skin wounds was similar in the Leu34 carriers and Val34Val homozygotes on anticoagulation (2.35±0.6 vs 2.05±0.6 nmol/L/s, respectively; Fig. 1B). The time-course of FXIIIAa formation did not differ between the Leu34-positive and −negative patients receiving acenocoumarol (0.22±0.02 vs 0.24±0.02 nmol/L/s; p>0.05; Fig. 1D). However, the maximum rate of increase in FXIIIAa levels observed at the end of bleeding was significantly lower in coumadin-treated Leu34 carriers comparing with those from the healthy group (0.22±0.02 nmol/L/s vs 0.44±0.06 nmol/L/s; p=0.03), while in the subjects homozygous for the Val34 allele, the rates of this process at the site of injury were similar regardless of the anticoagulation status (p>0.05).

γ-chain cross-linking

Immunobloting with the anti-γ-γ-dimer antibody 2G10-1, using the fluid phase of blood samples collected at the site of hemostatic plug formation, showed intense bands of M_r ~93 kD with migration equivalent to the Fn standard, ie. a single intense band of M_r ~93 kD corresponding to γ-γ-dimers (Fig. 2). These bands were detected coincidentally (at 2.5 minutes after injury) with FXIIIAa, which catalyzes γ-γ-dimer formation, consistent with the identity of the 93 kD band. Monomeric γ-chains (M_r ~46 kD), but not the Fbg Aα- and β-chains, were also stained by this antibody (Fig. 2). A marked decrease in the densities of bands, corresponding to Fbg γ-chains, was correlated with increasing γ-γ-dimer levels over time (n=8; r=−0.43; p=0.01). We also observed bands of a M_r ~110 kD (lane 1 to 4; Fig. 2). In most subjects, the density of these bands decreased over time and became undetectable at approximately 191.3±17.9 seconds. Based on a calculated electrophoretic mobility, we have assumed that the bands (~110 kD), absent in the Fn standard, while probed with the anti-γ-γdimer antibody (Fig. 2), may correspond to Aα-γ-heterodimers, produced by the action of tissue transglutaminase (tTG) from the damaged vessels. A similar electrophoretic mobility has been reported for Aα-γ-heterodimers (Aα+γ=66+46 kD), formed from Fbg in the presence of human erythrocyte TG (22). However, immunostaining with the anti-Fbg-Aα-chain antibodies (3A and 2e) showed Fbg Aα-chains (M_r ~66 kD), but not the bands of a M_r ~110 kD (data not shown). Identification of these latter bands, however, was beyond scope of the present study.

Fig. 2.

Representative immunoblots of fibrin(ogen) crosslinking products in healthy individuals (subject with the Val34Val genotype in 1–9 lanes A and that with the Val34Leu genotype in 1–9 lanes B). Consecutive blood samples obtained every 30 seconds at the site of microvascular injury (to 270 seconds in lane 9) were separated on a 4 to 12% linear gradient SDS-PAGE gel under reducing conditions (2% β-mercaptoethanol) and probed with the anti-γ-γ-dimer antibody (see Methods). Molecular weight standard (Bio-Rad) was marked on the left. Fibrin standard (100 ng/lane) is seen in the lane marked with STD.

Assuming that alterations in FXIII activation related to the FXIII Val34Leu polymorphism affect γ-γ-dimer formation also at the site of vascular injury as they do in vitro (15,16), we focused on the time-course of γ-γ-dimer formation. Soluble γ-γ-dimers appeared almost at the same time in Val34Val homozygotes and the Val34Leu heterozygotes (lane 6 and 5, Fig. 2, respectively). Quantitative γ-dimerization began almost simultaneously in both subgroups (142.5±22.5 vs 157.5±14.4 s, respectively; p>0.05). The rate of γ-γ-dimer formation was somewhat higher in the Leu34 carriers than in the Val34Val homozygotes (30.6±7.5 vs 18.1±2.1 μg/L/s; p>0.05). The concentration of this cross-linking product in the last minute before cessation of bleeding was significantly elevated in the Leu34 carriers compared with the Val34Val homozygotes (1355.6±177.9 vs 804.3±50.3 μg/L; p=0.028; Fig. 3). At the end of bleeding in the Simplate model, TAT levels did not differ among the Leu34-positive and −negative subjects (28.2±2.8 vs 26.3±2.4 nM; p>0.05).

Fig. 3.

Time-courses of the soluble γ-γ-dimer formation in 30-second blood samples collected at the site of microvascular injury, calculated on the basis of quantitative analysis of the immunoblots stained with the anti-γ-γ-dimer antibody. Data are given as medians. γ-γ-dimers are formed at a lower rate in anticoagulated subjects than in normal individuals. Maximum levels of these cross-links were higher in the healthy subjects heterozygous for the Leu34 allele than in those homozygous for the Val34 allele, with no such effect when comparing with respective groups treated with acenocoumarol.

In patients on long-term oral anticoagulation, there was no significant difference in the time at which the band corresponding to γ-γ-dimers became detectable on immunoblots between the non-carriers and carriers of the Leu34 allele (192±15.3 vs 180±34.6 s, respectively; p>0.05; data not shown). The maximum rate of increase in γ-γ-dimer formation was similar in the coumadin-treated Val34Val homozygotes and Val34Leu heterozygotes (7±7.5 vs 8.6±2.0 μg/L/s, respectively; p>0.05; Fig. 3). Levels of γ-γ-dimers were 290.5±35.9 μg/L for the Val34Val homozygotes and 369.4±75.9 μg/L for the Val34Leu heterozygotes (p>0.05) within the last minute of blood collection from skin incisions. Levels of TAT in shed blood, measured just before cessation of bleeding, were almost identical in the coumadin-treated Leu34 carriers and Val34 homozygotes (12.2±1.8 vs 13.1±2.2 nM, p>0.05). Only in the presence of the Leu34 allele, the differences in maximum γ-γ-dimer levels (p=0.01) and in the rate of its increase in blood samples obtained using the Simplate model (p=0.03) between healthy and anticoagulated subjects were significant (Fig. 3).

Discussion

The present study is the first to evaluate the extent and time-course of Fbg/Fn cross-linking in vivo when blood coagulation is triggered by TF exposure following vascular injury and a large amount of Fn is rapidly formed. We demonstrated that apart from a visible clot formation at margins of dissected vessels, blood flowing in proximity of a growing hemostatic plug contains measurable amounts of soluble FXIIIA and γ-γ-dimers that can be detected using Western blotting.

Our data indicate that the FXIII Leu34Val polymorphism exerts significant effects not only on thrombin-mediated FXIIIA cleavage, but also soluble γ-γ-dimer formation in the microvasculature. This effect has been demonstrated in a model in which all components of the vascular and subvascular tissues, along with blood hemostasis, are involved. Under these conditions, FXIIIA removal from the blood (involving binding to Fn, platelets and vessel wall components, along with thrombin-mediated FXIII activation [3]), generation of its soluble activated form and γ-γ-dimer formation, measured in the fluid phase of blood sampled from standardized incisions, proceed rapidly and almost simultaneously. Soluble FXIIIAa formation proceeds within 2 minutes with subsequent soluble γ-chain dimers formation as early as at 2-3 minutes after vascular injury while γ-chain cross-linking occurs in vitro within 5 to 10 minutes (16). The rapid γ-chain dimer formation at the site of hemostatic plug formation may result at least in part from platelet FXIII release and its activation in loco; a rapid platelet activation has been convincingly demonstrated in the Simplate model (18).

The cross-linked γ-γ-dimers are most likely not degradation products, but a soluble dyad of plasma protein generated by the action of TG activity on the surface of the growing hemostatic plug and may represent an indirect measure of FXIIIAa activity (23). The electrophoretic mobility of γ-γ-dimers detected in the microvasculature was identical to that of γ-γ-dimers in purified systems (15,24) and the Fn standard. Moreover, in the blood obtained from skin wounds, levels of a major fibrin degradation product, D-dimer, have been found to be below the detection limits, 0.1 μg/mL (A. Undas, unpublished data). We showed that final concentrations of soluble γ-γ-dimers at the sites of microvascular injury ranged between 1–2 mg/L. Elevated plasma Fbg γ-γ-dimer levels have been reported in patients with atherosclerosis and thromboembolism, with hardly detectable amounts of γ-γ-dimers in healthy subjects (25). Gerner et al. (26) observed high concentrations of these cross-links in the plasma of healthy controls, estimated at 5.5±3 mg/L by using integration of two-dimensional spot intensities. These discrepancies in results of the quantitative analysis could be attributed to different methodological approaches.

In the current study, we provide evidence for increased γ-γ-dimer formation in vivo in the presence of the FXIII Leu34 allele. In contrast to previous in vitro studies (15,16), the onsets of quantitative γ-γ-dimer formation at the site of injury were similar regardless of whether Leu or Val was at position 34. However, the rate of γ-γ-dimerization was significantly faster in individuals heterozygous for the Leu 34 allele than in those homozygous for the Val34 allele. Moreover, at the end of bleeding, γ-γ-dimer levels were higher by 40% in the former than in the latter group, which may result in an earlier stabilization of hemostatic plugs and more effective protection from fibrinolysis in vivo in the Leu34 carriers. Until now, in vitro experiments analyzed homozygous forms of the FXIII Val34Leu variants and their influence on Fn cross-linking (15,16). We report now that the Val34Leu heterozygous variant can significantly affect γ-γ-dimer formation in vivo like that homozygous for the Leu34 allele in vitro (16). Our findings supports the concept of a marked influence of the FXIII Val34Leu polymorphism on a biological activity of FXIII on γ-dimerization and probably the architecture of the Fn formed.

A novel observation is the abolition of the Leu34-related acceleration in FXIIIA cleavage and γ-dimerization during long-term treatment with acenocoumarol, which suggests that low thrombin levels alter the impact of the FXIII Val34Leu polymorphism on Fn formation in vivo. In the current study, coumadin-treated subjects (mean INR 2.35) had maximum TAT levels in shed blood by about 50% lower as compared to the values for the healthy individuals studied. This reduction in thrombin generation in the microvasculature during oral anticoagulation is almost identical to that reported in healthy subjects receiving acenocoumarol in a study by Kyrle et al. (27) who used a similar model of blood coagulation. In purified systems, with lower thrombin concentrations, the accumulation of γ-γ-dimers proceeded at a slower rate, but the FXIII Leu34 allele was still associated with a higher rate of this reaction (15). Our data might suggest that oral anticoagulation has the potential to abolish the Leu34Val polymorphism associated differences in the rate of soluble FXIIIA generation and γ-chain cross-linking at the site of vascular injury, by slowing both reactions in the Leu34 carriers. From the clinical point of view, it would be of interest to investigate the rates of recurrent thrombotic events among patients positive and negative for the Leu34 allele during oral anticoagulant therapy.

A limitation of the current study is a small number of individuals studied. Larger populations, including subjects homozygous for the Leu34 allele should be evaluated to validate our observations. Other methodological issues that should be taken into account in our blood clotting model include a role of Fn polymerization in FXIII activation and γ-γ-dimer formation. Although Fn polymerization determines a cross-linking process, at the site of vascular injury Fbg and prothrombin, as a zymogen for thrombin, are provided by the circulating blood in proximity of a clot constantly growing until bleeding stops. Thus in the Simplate model, Fn is most likely generated in increasing amounts throughout the bleeding period. Likewise, there is evidence that FXIII activation can be significantly reduced while the rate of Fn formation decreases at lowering thrombin generation (3), thrombin is increasingly formed at the injured microvasculature until the cessation of bleeding, as we have shown in previous reports (17,19). A dynamic model of blood coagulation applied in the present study, though it evalautes only soluble proteins or peptides, has, therefore, important advantages over static systems using selected purified proteins because in vivo, until blood flows, there are sufficient amounts of all components necessary for the efficient hemostasis at the site of injury. In contrast, we acknowledge the fact that the structure of the clot attached to the wound margins, including the extent of cross-linking, cannot be analyzed by means of the Simplate model.

We observed unexpected dense bands of a M_r ~110 kD, containing γ-chain, immediately after disruption of the blood vessel wall. It might be speculated that these bands represent products of tissue transglutaminase, present in skin blood vessels and released upon vascular injury (28). Tissue TG produces preferentially the Fbg Aα- and Aα-γ-chain complexes (29). However, these γ-chain containing cross-links have not been convincingly identified by us and further study is necessary to confirm this hypothesis.

In summary, our results indicate that the common FXIII Leu34 polymorphism is associated with a 40% increase in γ-γ-dimer formation detectable in the blood collected at the site of microvascular injury in healthy individuals. Importantly, long-term oral anticoagulation abolishes the impact of the Leu34 allele both on the cleavage of FXIIIA and soluble γ-γ-dimer formation in vivo. Our data from the model of microvascular injury point to the importance of all components of the hemostatic system in a localized process such as Fn formation and its stabilization. We believe that this study may help clarify the actual role of the FXIII Val34Leu polymorphism in atherosclerosis or thrombosis and highlights the significance of oral anticogulation in thrombotic patients. To what extent the Leu34 allele may modulate effects of this therapy in various clinical settings warrants further investigation.