A Simple Screening Assay for Certain Fibrinolysis Parameters (FIPA)

Thrombosis Research 97 (2000) 231–237

REGULAR ARTICLE

A Simple Screening Assay for Certain Fibrinolysis Parameters (FIPA) Thomas W. Stief1, Frauke Hinz1, Ju¨rgen Kurz1, Manfred O. Doss1 and Volker Kretschmer2 Institute of Clinical Chemistry, and 2Institute of Transfusion Medicine and Haemostaseology, University Hospital of Philipps University, D-35033 Marburg, Germany. 1

(Received 7 May 1999 by Editor J. Meier; revised/accepted 10 August 1999)

Abstract Hemostasis, the system of generation and degradation of thrombi, consists of coagulation and fibrinolysis. Whereas global assays to study coagulation have existed for many years, there has been no simple, rapid, and economic routine test for the plasmatic fibrinolysis parameters plasminogen activator inhibitor-1, ␣2-antiplasmin, plasminogen, and aprotinin. Here a fast functional global assay for these plasmatic fibrinolytic parameters is presented. However, the present assay is not sensitive to physiological concentrations of prourokinase or tissue-type plasminogen activator. The following assay conditions have been found to be optimal: 50 ␮L of citrated plasma is incubated with 50 ␮L of 10 IU urinary-type plasminogen activator (urokinase)/mL, 1.1 mmol/L tranexamic acid, 1% polygelin, 0.1% Triton X-100, phosphate-buffered saline, pH 7.4, for 20 min at 37⬚C (plasmin generation phase). Then 50 ␮L of 3 mmol/L HD-Nva-CHALys-pNA, 1.05 mol/L KCl is added, and ⌬A (405 nm)/10 min (37⬚C) is determined, by using a microAbbreviations: FIPA, fibrinolysis parameters assay; u-PA, urinary-type plasminogen activator (urokinase); t-PA, tissue-type plasminogen activator; MW, molecular weight; AP, ␣2-antiplasmin; a2M, ␣2-macroglobulin; PBS, phosphate-buffered saline; IU, international units; mA, milliabsorbance (A ⫻1000); PAI, plasminogen activator inhibitor; SCU, single chain urokinase; TAFI, thrombin activatable fibrinolysis inhibitor. Corresponding author: T. W. Stief, University Hospital of Philipps University, Institute of Clinical Chemistry, D35033 Marburg, Germany. Tel: ⫹49 (6421) 28 64471; Fax: ⫹49 (6421) 28 65594. E-mail:⬍[email protected]⬎.

titerplate reader (plasmin detection phase). The results are calibrated against pooled normal plasma (100% plasmatic fibrinolytic parameters activity). The intra- and interassay coefficients of variation have been found to be less than 5%. The detection limit (sensitivity) of the functional fibrinolysis assay is 5% of the normal plasmatic fibrinolysis parameters activity. The normal plasmatic fibrinolysis parameters activity is 100%, ␴⫽25%. The plasmatic fibrinolysis parameters activity correlates negatively (r⫽⫺0.684) with the plasminogen activator inhibitor-1 activity of patient samples. The plasmatic fibrinolysis parameters assay is a simple, rapid, and economic functional test for several clinical relevant fibrinolysis parameters.  2000 Elsevier Science Ltd. All rights reserved. Key Words: Fibrinolysis; Plasma; Assay; Urokinase

H

emostasis is a balanced system of generation and degradation of thrombi, that is, coagulation and fibrinolysis. Thromboplastin time and activated partial thromboplastin time are screening assays for coagulation. Euglobulin lysis time, fibrin plate, or microtiter plate methods or the determination of D-Dimer are screening assays for fibrinolytic activity [1–6]. However, these methods are rather complicated and not very practical for routine use, or at least not economical. Therefore, a rapid functional screening assay has been developed that detects abnormalities of certain plasmatic fibrinolysis parameters (FIPA).

0049-3848/00 $–see front matter  2000 Elsevier Science Ltd. All rights reserved. PII S0049-3848(99)00165-6

232

T.W. Stief et al./Thrombosis Research 97 (2000) 231–237

1. Material and Methods Standard human plasma, two-chain urokinase (urinary-type plasminogen activator, Diagnostic Products Corporation (DPC) [u-PA]; 80% of the high molecular weight form), Berichrom PAI and polygelin (Haemaccel) were from Behringwerke (Marburg, Germany). Active human a2M, bovine serum albumin, trans-4-(aminomethyl)-cyclohexane carboxylic acid (tranexamic acid), guanidine and Triton X-100 were from Sigma (Deisenhofen, Germany). The chromogenic plasmin substrate HD-NvaCHA-Lys-pNA was from Pentapharm (Basel, Switzerland). Polyclonal antibodies against human thrombin activatable fibrinolysis inhibitor (TAFI) were from Haemochrom (Essen, Germany). Plasminogen depleted plasma, ␣2-antiplasmin depleted plasma, active human ␣2-antiplasmin, and human single chain urokinase (SCU)–PA were from American Diagnostica, Greenwich, USA. Aprotinin was purchased from Bayer (Leverkusen, Germany), PAI-2 from Calbiochem (Frankfurt, Germany), and unfractionated heparin from Hoffmann-La Roche (Basel, Switzerland). Glu-plasminogen was from Chromogenix, (Mo¨lndal, Sweden), PAI control plasmas (1 u-PA inhibiting unit⫽6.3 t-PA-inhibiting units, considering a molecular weight [MW] of 54000 D for u-PA and one of 68000 D for t-PA) were from Biopool International (Umea, Sweden). The standard FIPA is performed as follows: 50 ␮L of plasma are incubated with 50 ␮L of 10 IU u-PA/mL, 1.1 mmol/L tranexamic acid, 1% polygelin, 0.1% Triton X-100, PBS, pH 7.4, for 20 minutes at 37⬚C. Then 50 ␮L of 3 mmol/L HD-NvaCHA-Lys-pNA, 1.05 mol/L KCl is added, and ⌬A (405 nm)/10 minutes (37⬚C) is determined by using a microtiter plate reader (DPC, Los Angeles, CA, USA). Addition of 0.45 mol/L guanidine to the substrate reagent (0.15 mol/L final test concentration) inhibits fibrinolysis activation, thus improving the kinetic-linearity. The result is calibrated against a pooled 100% normal plasma. This chromogenic substrate was chosen because in a final assay concentration of 300–400 mmol/L KCl and 100–200 mmol/L guanidine it is not cleaved by plasma kallikrein.

1.1. Optimization of Urokinase Activity Normal human plasma, containing 1.3 urokinase inhibiting units of PAI-1/mL (A; 100% control),

Table 1. Test scheme FIPA 50 ␮L plasma 50 ␮L 10 IU/mL u-PA, 1.1 mmol/L tranexamic acid, PBS, pH 7.4 Incubation of 20 min (37⬚C) 50 ␮L 3 mmol/L HD-Nva-CHA-Lys-pNA, 1.05 mol/L KCl Determination of ⌬ A (405 nm)/10 min

plasma containing 3.8 units of PAI-1/mL (B; pathologic PAI-1 plasma [7–9]), and plasma containing 6.3 units of PAI-1/mL (C; severely pathologic PAI-1 plasma) were analyzed in the FIPA. The urokinase activity was considered to be optimal, which results in (1) an induced plasmin activity for plasma A of greater than 25 milliabsorbance (mA)/ minute; (2) an induced plasmin activity for plasma B of 75% of plasma A and for plasma C of 50% of plasma A; and (3) a maximal (A-C)/A value.

1.2. Optimization of Tranexamic Acid Concentration Standard human plasma and plasma with 50% ␣2antiplasmin (AP) activity were incubated as indicated in Table 1, varying the tranexamic acid concentration. The concentration was considered to be optimal, which resulted in (1) a maximally inducible plasmin activity; and (2) a pathologically increased FIPA result in the plasma with 50% AP activity.

1.3. Test Validation The FIPA was tested for the influence of sample PAI-1, PAI-2, plasminogen, AP, aprotinin, SCUPA, or TAFI. Therefore, 50 ␮L of normal human plasma, containing different amounts of (1) PAI-1 (0–10 [u-PA inhibiting] U/mL); (2) PAI-2 (0–10 [u-PA inhibiting] U/mL); (3) plasminogen (0–2 plasma units [U]/mL ⫽0–200%); (4) AP (0–2 plasma units [U]/mL ⫽0–200%); (5) aprotinin (0– 500 kallikrein-inhibiting U/mL; in presence of unfractionated heparin (0–1 IU/mL); or (6) SCU-PA (5, 50, 500 ng/mL) was incubated as indicated in Table 1. To investigate the influence of TAFI on the FIPA, 50-␮L samples of 100% control plasma or samples of pooled serum were preincubated for 30 minutes at 37⬚C with 0, 100, or 200 ␮g polyclonal anti-TAFI antibodies as indicated in Table 1. To

T.W. Stief et al./Thrombosis Research 97 (2000) 231–237

233

investigate the stability of the individual fibrinolytic factors over the reaction time, we preincubated normal human plasma for 0, 20, 40, or 60 minutes at 37⬚C before the FIPA was performed. The effect of the inhibitors AP and a2M on the FIPA also was evaluated in a purified system: 50 ␮L of 0.1 g/L Glu-plasminogen in PBS, 1% bovine albumin, supplemented with 0–1 U/mL human AP or with 0–1 U/mL human ␣2-macrogobulin was incubated as described in Table 1.

1.4. Test Precision and Test Sensitivity Ten series of 10-fold determinations of the FIPA for 50%, 100%, and 150% plasmas were performed to determine intra- and interassay coefficients of variation (test precision). The sensitivity (detection limit) of the FIPA was defined as the threefold standard deviation of a 0% FIPA sample (plasminogen depleted plasma).

1.5. Normal Range Normal plasmas (n⫽126) from healthy blood donors (aged 18–65 years) were assayed in the FIPA. Venous blood was drawn without prolonged stasis into plastic tubes containing 1:10 106 mmol/L sodium citrate, followed by hard centrifugation (10 min at 2500⫻g) to prevent (PAI-1 releasing) platelet contamination. The plasma was immediately separated and analyzed. The FIPA results were tested for normal distribution and expressed in a histogram.

1.6. Correlation between Functional PAI-1 and FIPA Activity PAI-1 and FIPA activity was determined in 134 randomly collected patient samples submitted for laboratory analysis. The correlation factor (r) was calculated according to Spearman.

2. Results A simple, rapid, and economic FIPA has been developed. The assay consists of addition of twochain u-PA to plasma. After a plasmin-generating incubation, a chromogenic plasmin substrate is added, and the plasmin cleavage of this substrate

Fig. 1. Optimization of urokinase activity. Normal human plasma (100% control) containing 1.3 U/mL PAI-1 (diamonds), plasma containing 3.8 U/mL PAI-1 (triangles), and plasma containing 6.3 U/mL PAI-1 (squares) were analyzed in the FIPA.

is determined as an increase in absorbance at 405 nm by using a microtiterplate reader. The optimal u-PA activity added to 50 ␮L of sample is 0.5 IU (u-PA reagent consisting of 10 IU/mL u-PA; Figure 1). At this u-PA activity, the assay allows sensitive detection of relevant changes of PAI-1 activity. In addition, a valid result can be obtained within 30 minutes: FIPA activity of 100% control plasma ⫽ 270⫾5 mA/10 minutes. A modified FIPA version with an u-PA reagent containing only 5 IU/mL u-PA would be even more sensitive to PAI-1 increases; however, the incubation time has to be doubled, which gives rise to unspecific reactions (e.g., u-PA/ antithrombin [10,11]) and retards the hemostaseological diagnosis. The optimal test concentration of tranexamic acid at physiologic assay pH is 0.35–0.7 mmol/L (0.7–1.4 mmol/L tranexamic acid in the u-PA reagent; Figure 2). This tranexamic acid concentration results in a pathologically increased FIPA result (about 150%) in plasma with 50% of AP activity. Without tranexamic acid, hardly any plasmin activity could be detected, indicating a strong control of fibrinolysis activation by AP [4]. Tranexamic acid only partially suppresses the activity of AP. Thus, the FIPA depends on the main inhibitor of plasmin, since a change in the remaining AP activity is reflected by the FIPA (Figure 3). The FIPA is also sensitive to abnormal plasminogen activity: ⬎25% changes of the normal plasma concentration of AP or plasminogen are detected in the FIPA (Figure 3).

234

T.W. Stief et al./Thrombosis Research 97 (2000) 231–237

Fig. 2. Optimization of tranexamic acid concentration. Normal human plasma (diamonds) and plasma with 50% AP activity (squares) were assayed in the FIPA, varying the final assay concentration of tranexamic acid. One hundred percent control ⫽ FIPA with an u-PA reagent containing 1.1 mmol/L tranexamic acid.

The FIPA is also sensitive to PAI-1. Figure 4 demonstrates the assay dependence of PAI-1: pathologically increased PAI-1 activities of greater than 4 U/mL resulted in a greater than 25% decreased FIPA. Increased PAI-2 activities are only detected if PAI-2 is increased to greater than 10 U/mL. The activity of the drug aprotinin can be monitored by the present assay (Figure 5). FIPA in presence or absence of aprotinin is independent of therapeutic doses of plasmatic unfractionated heparin. In a purified system, 1 U/mL AP still resulted in a FIPA activity of 36% when compared with the 0 U/mL AP sample (zero control) (Figure 6).

Fig. 3. Test influence of plasmatic AP and plasminogen. Fifty microliters of normal human plasma containing 0–2 plasma U/mL plasminogen (diamonds) or 0–2 U/mL AP (squares) was analyzed in the FIPA.

Fig. 4. Test influence of plasmatic PAI. Fifty-microliter samples of human plasma containing 0–10 U/mL PAI-1 (diamonds) or 0–10 U/mL PAI-2 (squares) was analyzed in the FIPA.

The FIPA is only slightly sensitive to the broad spectrum inhibitor a2M: 1 U/mL of a2M decreased the FIPA activity in a purified system by 21%. A combination of 1 U/mL AP and 1 U/mL a2M in the purified system resulted in a FIPA activity of 44% when compared with the zero control (i.e., a2M protected 8% of the generated plasmin molecules from inhibition by AP). This limited effect of a2M on the plasmatic fibrinolysis system is consistent with the relatively small inhibitory activity of a2M in a previously described purified urokinase/plasminogen system [11]. The FIPA is not sensitive to TAFI [12]: addition of 100 ␮g or 200 ␮g of polyclonal anti-TAFI antibodies to 50 ␮L of control plasma decreased the

Fig. 5. Test influence of aprotinin (and heparin). Fiftymicroliter samples of normal human plasma, supplemented with 0–500 kallikrein-inhibiting U/mL aprotinin and 0 IU/ mL (diamonds), 0.25 IU/mL (filled squares), 0.5 IU/mL (triangles), or 1.0 IU/mL (open squares) unfractionated heparin was analyzed in the FIPA.

T.W. Stief et al./Thrombosis Research 97 (2000) 231–237

235

Fig. 6. Test influence of purified a2M (or AP). 50 ␮L samples of 0-1 U/mL human a2M (䊏), or AP (䉬) in PBS, 1% albumin, was analyzed in the FIPA.

Fig. 7. Normal range of FIPA activity. Normal plasmas (n⫽126) from healthy blood donors (aged 18–65 years) were assayed in the FIPA.

FIPA activity by 10.5% or 15.5%, respectively. Anti-TAFI antibodies given to pooled serum decreased the FIPA activity by 7.8% or 11.6%, respectively. Furthermore, the FIPA is not sensitive to slight changes (⫾50%) in plasmatic SCU-PA; only SCU-PA activities greater than 10-fold the normal plasma concentration are detectable by the FIPA: 100% plasma supplemented with 50 ng/mL SCU-PA resulted in a FIPA of 124%, a plasma supplemented with 500 ng/mL SCU-PA in 371% FIPA. The inactive SCU-PA seems to be converted to u-PA by plasmin being generated in the first FIPA incubation period. The concentration of active tissue-type plasminogen activator (t-PA) in plasma is very low (0.5 IU/mL) [9]; the FIPA uses the omega-amino acid tranexamic acid for partial neutralization of plasmatic AP, which is an inhibitor of t-PA. Therefore, FIPA—like other complex fibrinolysis assays [6]—is not sensitive to plasmatic t-PA. The fibrinolytic factors investigated in the FIPA are stable for 1 hour at 37⬚C. Thus, in the 20-minute reaction time of the plasmin-generating phase of the FIPA, the fibrinolysis factors that influence the FIPA do not change. The intra- and interassay coefficients of variation are less than 5%. The detection limit (sensitivity) of the FIPA is 5% of the activity of normal control plasma. The normal FIPA activity is 100%; ␴ ⫽ 25% (Figure 7). The clinical diagnosis for 19 of 134 patients (14.2%) with severely depressed FIPA activity was: sepsis (n⫽7), myocardial infarction (n⫽4), malignancy (n⫽4), liver cirrhosis (n⫽2), and pulmonary thromboembolism (n⫽2). The FIPA correlated negatively with PAI-1 activity in

the 134-patient samples (r⫽⫺0.684, p⬍0.05; Figure 8).

3. Discussion A rapid, simple, and economic functional screening assay for certain fibrinolysis parameters in plasma was developed. The principle of this new test is the activation of fibrinolysis by addition of a defined amount of u-PA. In a first step, u-PA interacts with the plasminogen activator inhibitors PAI-1 and PAI-2, mainly, however, with PAI-1, since PAI-2 is present almost only in pregnancy plasma. u-PA activates plasmatic plasminogen to plasmin. Decreased plasma concentrations of plasminogen or altered concentrations of histidine-rich glycoprotein, a possible inhibitor of the plasminogen activa-

Fig. 8. Correlation betweeen functional plasmatic PAI-1 and FIPA. In 134 randomly collected patient samples submitted for routine analysis, plasmatic PAI-1 activity and FIPA activity were determined. Both parameters correlated negatively with (r⫽⫺0.684, p⬍0.05).

236

T.W. Stief et al./Thrombosis Research 97 (2000) 231–237

tion reaction, could be reflected by the present test. Finally, plasmin is inactivated by AP and to a much lesser degree by a2M. Normally, this plasmin–AP reaction is so efficient and fast that hardly any detectable active plasmin remains. Therefore, plasmatic AP has to be partially neutralized, using the omega-amino acid tranexamic acid. Tranexamic acid (20 ␮mol/L) reduces the k1 app of the plasmin–AP reaction by 50% [4]. Remaining plasmin then is detected by means of cleavage of a chromogenic plasmin substrate, the only significant cost factor of the FIPA: a FIPA in duplicate determination costs less than one US dollar, which allows the usage of the FIPA in hemostaseological screening of patients. Pathological changes in plasmatic AP or plasminogen can be diagnosed by a significant change of the FIPA result. Consequently, in patients with bleeding disorders due to insufficient plasmatic AP activity (⬍50% of normal plasma) [13–18], a strongly increased FIPA activity might be detected. In contrast, pathological activities of plasmatic PAI-1 are detected only by the FIPA if PAI-1 is pathologically increased. A decreased PAI-1 activity [19–23] would not change the FIPA by more than 20%. However, given the hemostaseological importance of elevated PAI-1 activities and the relative lack of importance of decreased PAI-1 activities [24], the detection of elevated PAI-1 levels seems to be of greater clinical relevance than that of decreased PAI-1 activities. In conclusion, the rapid, simple, and economical FIPA detects abnormalities in the interaction of the clinically relevant plasmatic fibrinolysis factors PAI-1, AP, and plasminogen and can be used to monitor the efficacy of aprotinin. This work is part of the medical academic thesis of F. Hinz.

3.

4.

5.

6.

7.

8.

9.

10.

References 11. 1. von Kaulla E, von Kaulla KN. Detection of hypercoagulability states by a modified ethanol gelation test and its correlation with other coagulation procedures. Am J Clin Pathol 1974;61:810–22. 2. Jespersen J, Astrup T. A study of the fibrin plate assay for fibrinolytic agents. Optimal con-

12. 13.

ditions, reproducibility and precision. Haemostasis 1983;13:301–15. Fareed J, Hoppensteadt DA, Leya F, Iqbal O, Wolf H, Bick R. Useful laboratory tests for studying thrombogenesis in acute cardiac syndromes. Clin Chem 1998;44:1845–53. Bachmann F. The plasminogen-plasmin enzyme system. In: Colman RW, Hirsh J, Marder VJ, Salzman EW, editors. Hemostasis and thrombosis: Basic principles and clinical practice. 3rd ed. Philadelphia: Lippincott; 1994. p. 1592–622. van Giezen JJ, Chung A Hing JE, Vegter CB, Bouma NN, Jansen JW. Fibrinolytic activity in blood is distributed over a cellular and the plasma fraction which can be modulated separately. Thromb Haemost 1994;72:887–92. Amiral J, Malmejac A, Gin H, Pannell R, Vissac A-M, Seigneur M, Scarabin P-Y, Boisseau M, Guize L, Gurewich V. Evaluation of the fibrinolytic potential on plasma:physiological and pathological variations, and associations with cardio-vascular disease risk factors. Fibrinol Proteolysis 1999;13(Suppl 1):1–10. Stief TW, Lenz P, Becker U, Heimburger N. Determination of plasminogen activator (PAI) capacity of human plasma in presence of oxidants: A novel principle. Thromb Res 1988; 50:559–73. Chmielewska J, Ranby M, Wiman B. Evidence for a rapid inhibitor to tissue plasminogen activator. Thromb Res 1983;31:427–36. Rosen S, Wejkum L, Billing-Claeson S, Ghosh R, Grdic K, Chmielewska J, Meijer P, Kluft C, Tengborn L, Conkie J, Walker I. Evaluation of a bioimmuno assay for t-PA activity and its relation to PAI-1 activity and antigen levels. Fibrinol Proteolysis 1998;12:340–6. Clemmensen I. Inhibition of urokinase by complex formation with human antithrombin III in absence and presence of heparin. Thromb Haemostas 1978;39:616–23. Stief TW, Martin E, Villalobos E, Yoldi F, Rodriguez JM. Specificity of functional determination of urokinase in human plasma using selective oxidants. Fibrinolysis 1990;4(Suppl 2):148–9. Nesheim ME. TAFI. Fibrinol Proteolysis 1999; 13:72–7. Kettle P, Mayne E. A bleeding disorder due to

T.W. Stief et al./Thrombosis Research 97 (2000) 231–237

14.

15.

16.

17.

18.

19.

deficiency of alpha 2-antiplasmin. J Clin Pathol 1985;38:428–9. Kordich L, Feldman L, Porterie P, Lago O. Severe hemorrhagic tendency in heterozygous alpha 2-antiplasmin deficiency. Thromb Res 1985;40:645–51. Knot EA, ten Cate JW, Lamping RJ, Gie LK. Alpha 2-antiplasmin: Functional characterization and metabolism in a heterozygote deficient patient. Thromb Haemost 1986;55:375–8. Holmes WE, Lijnen HR, Nelles L, Kluft C, Nieuwenhuis HR, Rijken DC, Collen C. Alpha 2-antiplasmin Enschede: Alanine insertion and abolition of plasmin inhibitory activity. Science 1987;238:209–11. Kluft C, Nieuwenhuis HR, Rijken DC, Groeneveld E, Wijngaards G, van Berkel W, Dooijewaard G, Sixma JJ. AP Enschede: Dysfunctional alpha 2-antiplasmin molecule associated with an autosomal recessive hemorrhagic disorder. J Clin Invest 1987;80:1391–400. Williams EC. Plasma alpha 2-antiplasmin activity. Role in evaluation and management of fibrinolytic states and other bleeding disorders. Arch Intern Med 1989;149:1769–72. Schleef RR, Higgins DL, Pillemer E, Levitt LJ.

20.

21.

22.

23.

24.

237

Bleeding diathesis due to decreased functional activity of type 1 plasminogen activator inhibitor. J Clin Invest 1989;83:1747–52. Dieval J, Nguyen G, Gross S, Delobel J, Kruithof EK. A lifelong bleeding disorder associated with a deficiency of plasminogen activator inhibitor type 1. Blood 1991;77:528–32. Reilly TM, Mousa SA, Seetharam R, Racanelli AL. Recombinant plasminogen activator inhibitor type 1: A review of structural, functional, and biological aspects. Blood Coagul Fibrinol 1994;5:73–81. Lee MH, Vosburgh E, Anderson K, McDonagh J. Deficiency of plasma plasminogen activator inhibitor 1 results in hyperfibrinolytic bleeding. Blood 1993;81:2357–62. Fay WP, Parker AC, Condrey R, Shapiro D. Human plasminogen activator inhibitor-1 (PAI-1) deficiency: Characterization of a large kindred with a null mutation in the PAI-1 gene. Blood 1997;90:204–8. Carmeliet P, Stassen JM, Schoonjans L, Ream B, van den Oord JJ, De Mol M, Mulligan RC, Collen D. Plasminogen activator inhibitor-1 gene deficient mice. II. Effects on hemostasis, thrombosis, and thrombolysis. J Clin Invest 1993;92:2756–60.