The hyperfibrinolytic state of mice with combined thrombin-activatable fibrinolysis inhibitor (TAFI) and plasminogen activator inhibitor-1 gene deficiency is critically dependent on TAFI deficiency

Authors


Ann Gils, Laboratory for Pharmaceutical Biology, Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, Campus Gasthuisberg, O&N2, PB 824, Herestraat 49, B-3000 Leuven, Belgium.
Tel.: +32 16 32 34 36; fax: +32 16 32 34 60.
E-mail: ann.gils@pharm.kuleuven.be

Abstract

Summary.  Background: Mice with single gene deficiency of thrombin-activatable fibrinolysis inhibitor (TAFI) or plasminogen activator inhibitor-1 (PAI-1) have an enhanced fibrinolytic capacity. Objectives: To unravel the function and relevance of both antifibrinolytic proteins through the generation and characterization of mice with combined TAFI and PAI-1 gene deficiency. Results: Mating of TAFI knockout (KO) mice with PAI-1 KO mice resulted in the production of TAFI/PAI-1 double-KO mice that were viable, were fertile, and developed normally. In a tail vein bleeding model, the bleeding time and hemoglobin content of the TAFI/PAI-1 double-KO mice did not deviate significantly from those of the single-KO mice or of the wild-type (WT) counterparts. Interestingly, in ex vivo rotational thromboelastometry measurements with whole blood samples, TAFI KO mice and TAFI/PAI-1 double-KO mice were more sensitive to fibrinolytic activation with tissue-type plasminogen activator than WT or PAI-1 KO mice. This enhanced fibrinolytic capacity was confirmed in vivo in a mouse thromboembolism model, as shown by decreased fibrin deposition in the lungs of TAFI KO mice and TAFI/PAI-1 double-KO mice as compared with WT or PAI-1 KO mice.Conclusions: TAFI gene inactivation predominantly contributes to the increased fibrinolytic capacity of TAFI and PAI-1 double-gene-deficient mice, as observed in some basic thrombosis models.

Introduction

The hemostatic process preserves the integrity of the vascular system by maintaining a delicate balance between coagulation and fibrinolysis. Plasmin is the key protease responsible for the degradation of fibrin into fibrin degradation products. It is generated from circulating plasminogen by tissue-type plasminogen activator (t-PA). Fibrin degradation products expose C-terminal lysines that bind to lysine-binding sites of plasminogen and t-PA, thereby potentiating plasmin production at the surface of the fibrin clot. Inhibitors such as thrombin-activatable fibrinolysis inhibitor (TAFI), plasminogen activator inhibitor-1 (PAI-1) and α2-antiplasmin regulate fibrinolysis, albeit through different mechanisms.

TAFI, which is activated by trypsin-like enzymes such as thrombin, the thrombin–thrombomodulin complex, and plasmin, removes C-terminal lysines from fibrin, leading to reduced plasmin generation and consequent prolongation of the clot lysis time [1,2]. Two different pools of TAFI are present in blood, i.e. the plasma pool and the platelet pool, the latter being only secreted upon platelet activation. Platelet-derived TAFI has been identified at a concentration that corresponds to only 0.1% of the total amount of TAFI found in plasma; however, the concentration of TAFI inside the platelet is of the same order of magnitude as that in plasma [3].

PAI-1 is an essential serine protease inhibitor (serpin) of the fibrinolytic system, and acts by inhibiting t-PA, leading to reduced plasmin generation [4]. In blood, two different pools of PAI-1 exist, i.e. plasma and platelets. The amount of PAI-1 in plasma is rather low, but PAI-1 occurs mainly in the active conformation. Although platelets store higher amounts of PAI-1, they do not seem to contribute to plasma PAI-1 under normal conditions, as only ∼ 10% of platelet PAI-1 is in the active form [5]. However, it has recently been reported that the majority of platelet PAI-1 is active and can contribute to clot stabilization [6].

During activation, platelets secrete not only PAI-1, but also another serpin, named protease nexin-1 (PN-1), which also inhibits t-PA. Recently, it has been established that, besides platelet PAI-1, platelet PN-1 also has relevant antifibrinolytic properties in both human and mice [7]. Other known plasminogen activator inhibitors are PAI-2 (placental PAI, which is only detectable during pregnancy), PAI-3 (a non-specific PAI that inhibits many other serine proteases involved in blood coagulation and fibrinolysis), and neuroserpin (involved in the regulation of neurogenesis, probably by blocking the deleterious effect of t-PA) [8]. Another important fibrinolysis inhibitor is α2-antiplasmin, a serpin responsible for the inactivation of free circulating plasmin [9].

Previously, different groups generated and/or characterized TAFI, PAI-1 and α2-antiplasmin single-gene-deficient mice. Despite certain conflicting results, the general conclusion tends towards improved fibrinolytic activity in these mice [10–14]. In the present study, we aimed to enhance stimulation of the plasminogen activation pathway by combining TAFI and PAI-1 deficiency in mice. The initial scope of our study was to determine whether an overt phenotype would be displayed in these double-deficient mice (in contrast to the single-deficient mice). As these double-knockout (KO) mice were viable, were fertile, and developed normally, we continued our study by subjecting these mice to a select panel of in vitro and in vivo thrombosis models, in order to determine whether, upon triggering, the double-KO mice would behave differently from the single-KO mice, and whether one gene deletion would predominate over the other.

Materials and methods

Animals

The TAFI and PAI-1 single-gene-deficient mouse strains (100% C57BL/6J and 75% : 25% C57BL/6J/129SvJ background, respectively) were generated as previously described [10,12]. Before the start of our breeding program, the PAI-1 KO mice were backcrossed into a pure C57BL/6J background for two generations (C57BL/6J mice purchased from Janvier, Le Genest St Isle, France). TAFI and PAI-1 KO mice were subsequently intercrossed (P−2), and heterozygous mating (P−1) was established to obtain wild-type (WT) mice (TAFI+/+: PAI-1+/+), single-gene-deficient mice (TAFI−/−:PAI-1+/+ and TAFI+/+:PAI-1−/−) and double-gene-deficient mice (TAFI−/−: PAI-1−/−) in the same genetic background (P, parental lines). Mice heterozygous for TAFI and homozygous deficient for PAI-1 (TAFI+/−:PAI-1−/−) or vice versa (TAFI−/−:PAI-1+/−) were mated to increase the generation of mice with double gene deficiency. The four different genotypes (P) were further intercrossed to raise sufficient mice for in vivo evaluation (Fx progeny, x = 1–3). The state of health of the Fx progeny was judged by inspecting viability, general appearance, fertility, and growth.

Mice were kept in microisolation cages on a 12-h day–night cycle with water and food ad libitum. Housing and procedures involving experimental animals were conducted in accordance with institutional guidelines and on a license (P133-2010) approved by the Ethical Committee of the KU Leuven, Belgium.

Genotypic analyses

Genomic DNA was isolated from mouse tail biopsies as previously described [15], and used as a template in PCR to verify the genotypes. Each sample was subjected to four separate PCR reactions with the following primer sets, to detect (i) the TAFI gene, i.e. forward primer 5′-GCTCTGGTTCTCTGGTTGGT-3′ and reverse primer 5′-CAGTCTTCTATGGTAACAGC-3′ (annealing in introns 1 and 2 [ENSMUSG00000021999], respectively), detecting a 446-bp amplification product in TAFI+/+ and TAFI+/− samples, or reverse primer 5′-ACAAGATGGATTGCACGCAGG-3′ (derived from the phosphoglycerine thymidine kinase–neomycin expression cassette, pKONEO), detecting a 1278-bp PCR product in TAFI−/− and TAFI+/− samples; and (ii) the PAI-1 gene, i.e. forward primer 5′-GACCTTGCCAAGGTGATGCTTGGCAAC-3′ and reverse primer 5′-GAGTGGCCTGCTAGGAAATTACATTC-3′ (both annealing in exon 9 [ENSMUSG00000037411]), detecting a 416-bp amplification product in PAI-1+/+ and PAI-1+/− samples, or forward primer 5′-AATGTGTCAGTTTCATAGCC-3′ (derived from the neomycin expression cassette), detecting a 190-bp PCR product in PAI-1−/− and PAI-1+/− samples. PCR products were electrophoresed on a 1.5% agarose gel, post-stained with 2 × GelRed (Biotium, Hayward, CA, USA) in 1 m NaCl, and visualized with a UV transilluminator and the UVP Biodoc-It Imaging System (UVP, Upland, CA, USA).

TAFI, PAI-1 and α2-antiplasmin plasma levels

Following anesthesia (isoflurane; 2.5%), blood was collected via the retro-orbital vein on sodium citrate 3.8% (1 : 10 v/v). Plasma was prepared by centrifugation of blood at 2000 × g for 20 min. Mouse plasma levels of TAFI and PAI-1 were determined with previously described ELISAs [16,17]. An ELISA specific for the detection of free α2-antiplasmin was developed with mAbs directed towards human α2-antiplasmin that cross-react with mouse α2-antiplasmin, i.e. MAP18C6 as primary antibody and MAP27B12-HRP as secondary antibody. Recombinant (TAFI and PAI-1) or purified (α2-antiplasmin) mouse proteins were used for calibration.

Tail bleeding time

Mouse tail vein bleeding times were monitored with a tail-clipping assay, as described previously [18]. Bleeding time was recorded until first arrest of bleeding (no rebleeding within 30 s). In addition, the accumulated bleeding time was calculated as the sum of all bleeding episodes within a 30-min period. If no cessation of bleeding occurred within 30 min, 1800 s was defined as the (accumulated) bleeding time. Additionally, hemoglobin concentration was determined over a period of 60 min after tail-clipping. Subsequent to centrifugation (10 min at 2000 × g), blood cells were resuspended, and the hemoglobin content was measured on a Cell-Dyn 3500R counter (Abbott, Diegem, Belgium). If necessary, mouse tails were cauterized at the end of the experiment.

Rotational thromboelastometry (ROTEM) analysis

Blood samples of the four genotypes were collected by retro-orbital vein puncture into sodium citrate (3.8%, 1 : 10 v/v), and analyzed on a ROTEM delta instrument (Tem International, Munich, Germany). The samples (300 μL) were incubated with CaCl2 (final concentration, 20 mm), thromboplastin (final concentration, 7.7 pm; Innovin; Siemens, Hamburg, Germany), and increasing doses of recombinant t-PA (final concentration, 2–20 nm; Actilyse; Boehringer Ingelheim, Brussels, Belgium). The clot formation and lysis reaction was monitored for 2 h. Maximum lysis (ML, as percentage reduction of clot firmness within 1 h) was selected as the most appropriate parameter with which to define the lytic activity of the samples.

FeCl3-induced mesenteric thrombosis model

A mesenteric thrombosis model was set up as previously described [19], with slight modifications. Briefly, 4-week-old female anesthetized (Nembutal; 60 mg kg−1, intraperitoneal) mice were injected intravenously with rhodamine 6G chloride. A filter paper (2 × 3 mm) soaked in 5% FeCl3 was applied topically for 3 min on a mesenteric artery and vein. Thrombus formation and blood flow were monitored with a Leica DM IL LED Fluo microscope (× 10/0.25 objective) connected to a DFC350FX camera (Leica Microsystems, Wetzlar, Germany). Time to occlusion was defined as the time needed to form an occlusive thrombus. If no occlusion occurred within 1 h, the experiment was ended, and time to occlusion was defined as 60 min.

Thromboembolism model

A mouse thromboembolism model was used as previously described [20], with some modifications. Briefly, after injection of thromboplastin (2.5 μg kg−1, intravenous) to induce thromboembolism, fibrin of the mouse lungs was extracted, and concentrations were determined with an in-house ELISA, as previously described [20].

Statistical analysis

Quantitative data are shown as means and standard deviations, unless otherwise specified. Statistical analysis and data plotting were performed with graphpad prism 5 (Graphpad Software, San Diego, CA, USA).

To compare the segregation of male and female progeny of the heterozygous couples with the expected Mendelian ratio, a Fisher exact test was used. A chi-square test within the different homozygous breeding pairs was used to compare: (i) the number of pups that died before weaning age; and (ii) the death of female breeding pair partners.

Every statistical comparison of one of the next parameters between the different genotypes was preceded by a D’Agostino and Pearson normality test to check whether the values followed a Gaussian distribution. A one-way anova with Bonferroni’s multiple comparison test was used for statistical comparison of: (i) litter sizes, time between litters, and body weights; (ii) ML in ROTEM; and (iii) percentage of fibrin in the lungs of thromboembolism-induced mice. A Kruskal–Wallis anova with Dunn’s multiple comparison test was used for statistical comparison of: (i) TAFI, PAI-1 and α2-antiplasmin levels; (ii) tail vein bleeding times and hemoglobin content; and (iii) occlusion times in the mesenteric thrombosis model.

A dose–response (stimulation) fitting was used to calculate the EC50 of ML at increasing t-PA concentration.

Results

Phenotypic characteristics of TAFI and PAI-1 double-gene-deficient mice

After the heterozygous mating (P−1), the offspring appeared healthy, at least until the age of 12 months. The segregation of male and female progeny (Table 1) was not significantly different from the expected Mendelian ratio for the four expected genotypes (WT, TAFI KO, PAI-1 KO, and TAFI/PAI-1 KO), except for the male double-KO mice (0.82% vs. Mendelian ratio of 6.25%; < 0.05).

Table 1.   Segregation of male and female progeny after heterozygous mating (TAFI+/−:PAI-1+/− × TAFI+/−:PAI-1+/−)
 MaleFemaleTotalExpected
n % n % n %%
  1. KO, knockout; PAI-1, plasminogen activator inhibitor-1; TAFI, thrombin-activatable fibrinolysis inhibitor; WT, wild type. *Male double-KO progeny do not show Mendelian inheritance (< 0.05). †Other genotypes are defined as mice heterozygous for either the TAFI or PAI-1 gene, or both the TAFI and PAI-1 genes.

WT86.5643.17124.846.25
TAFI KO97.3897.14187.266.25
PAI-1 KO129.84118.73239.276.25
TAFI/PAI-1 KO10.82*64.7672.826.25
Other genotypes†9275.409676.1918875.8175.00
Total122100126100248100100

Subsequent parental breeding of the four genotypes resulted in a dropout of 18% of double-KO mice before weaning age (24 of 132 pups), which is statistically different from the other genotypes (< 0.001), i.e. 9.9% (eight of 81 pups), 7.4% (24 of 323 pups) and 5.2% (10 of 191 pups) for, respectively, the WT, TAFI KO and PAI-1 KO colonies. No significant differences in litter size or time between litters were observed, i.e. 7.8 ± 2.6, 7.0 ± 2.7, 6.2 ± 2.8 and 5.7 ± 2.4 pups per litter and 47.6 ± 16.1, 33.3 ± 13.7, 37.4 ± 12.7 and 39.9 ± 19.4 days between litters for WT, TAFI KO, PAI-1 KO and double-KO mice, respectively. However, there was a higher incidence of death of female breeding pair partners after delivery in the double-KO colony (48%; 12 of 25 females, < 0.001) than in the WT (7.7%; one of 13 females), TAFI KO (0%; none of 46 females) and PAI-1 KO (6.3%; 2 of 32 females) colonies. Unfortunately, we cannot be sure whether these deaths were correlated with delivery and/or nursery. No significant difference in body weight between the four genotypes was observed, either in 3-week-old or in 15-week-old male and female mice (data not shown).

TAFI, PAI-1 and α2-antiplasmin plasma levels

The levels of the three main antifibrinolytic proteins (TAFI, PAI-1, and α2-antiplasmin) were evaluated in plasma of the different homozygous genotypes with ELISA (Table 2). As expected, neither TAFI nor PAI-1 was present in the TAFI KO and PAI-1 KO mice, nor in the double KO mice. Similar TAFI levels were observed in PAI-1 KO and WT mice, and similar PAI-1 levels were observed in TAFI KO and WT mice. This indicates that there is no upregulation of PAI-1 in TAFI KO mice and vice versa. α2-Antiplasmin levels were similar in TAFI KO and double-KO mice, but were significantly lower than those in WT mice. This might be explained by higher plasmin generation in the absence of TAFI and subsequent capture of this circulating plasmin by α2-antiplasmin.

Table 2.   Mouse thrombin-activatable fibrinolysis inhibitor (TAFI), plasminogen activator inhibitor-1 (PAI-1) and α2-antiplasmin plasma levels determined by ELISA
 WTTAFI KOPAI-1 KOTAFI/PAI-1 KO
  1. KO, knockout; WT, wild type. n = 5–6. *< 0.05 as compared with WT mice.

TAFI (μg mL−1)3.0 ± 0.5<0.03*2.8 ± 0.9<0.03*
PAI-1 (ng mL−1)0.8 ± 0.40.4 ± 0.2<0.3*<0.3*
α2-Antiplasmin (μg mL−1)50.8 ± 8.436.2 ± 7.2*45.7 ± 3.637.0 ± 5.2*

Tail bleeding

So far, the reported effects of single TAFI or PAI-1 gene deficiency on tail bleeding have been diverse and dependent on the particular experimental setting used [11,21,22]. In our study, the effects of single and combined TAFI and PAI-1 gene deficiency on hemostasis were (re)evaluated in a tail bleeding time experiment (bleeding times are represented as medians; n = 10–28; Fig. 1). TAFI and PAI-1 single-KO mice showed a normal bleeding time (170 and 168 s, > 0.05, respectively) as compared with WT mice (270 s). Similarly, bleeding time was not substantially prolonged in double-KO mice (263 s, > 0.05). Furthermore, no significant difference (> 0.05) was observed in accumulated bleeding times (330 s, 630 s and 515 s for TAFI KO, PAI-1 KO and double-KO mice, respectively, vs. 398 s for WT mice). Hemoglobin contents were also similar, i.e. 0.046, 0.050 and 0.045 g dL−1 for TAFI KO, PAI-1 KO and double-KO mice, respectively, vs. 0.079 g dL−1 for WT mice (represented as medians, n = 4–5).

Figure 1.

 Tail bleeding times and accumulated bleeding times. No significant differences in bleeding times were observed between the four genotypes; n = 13–28. KO, knockout; PAI-1, plasminogen activator inhibitor-1; TAFI, thrombin-activatable fibrinolysis inhibitor; WT, wild type.

ROTEM analysis

To explore the effects on fibrinolysis of TAFI and/or PAI-1 gene inactivation, whole blood samples of the mice were analyzed by ROTEM. As derived from the ML shown in Fig. 2, blood samples from WT mice and PAI-1 KO mice showed similar lysis with increasing t-PA concentration. In blood samples from TAFI KO and double-KO mice, significant lysis already occurred at lower t-PA concentrations as compared with WT and PAI-1 KO mice; EC50 values were 3.76 ± 0.17 nm and 3.68 ± 0.23 nm for TAFI KO and double-KO mice, respectively, as compared with 9.52 ± 2.48 nm for WT mice (< 0.001), whereas, for PAI-1 KO mice, no significant difference from WT mice was observed (8.85 ± 0.28 nm, > 0.05). Moreover, no additional lytic effect could be observed in blood of double-KO mice as compared with TAFI KO mice, suggesting that PAI-1 deficiency does not contribute to the enhanced fibrinolytic capacity of the double-KO mice.

Figure 2.

 Maximum lysis in rotational thromboelastometry with increasing tissue-type plasminogen activator (t-PA) concentration. Maximum lysis is expressed as percentage reduction of clot firmness within 1 h after starting lysis. In blood samples from thrombin-activatable fibrinolysis inhibitor knockout (KO) (inline image) and double-KO mice (inline image), significant lysis occurred at lower t-PA concentrations (EC50 values of 3.76 ± 0.17 nm and 3.68 ± 0.23 nm, respectively) than in wild-type mice (inline image; EC50 > 10 nm) and plasminogen activator inhibitor-1 KO mice (inline image; EC50 > 10 nm); n = 1–7.

Mesenteric thrombosis model

In order to further investigate the influence of single and double TAFI and PAI-1 deficiency on hemostatic function in vivo, a mesenteric thrombosis model was implemented in which time from injury to occlusive thrombus formation was measured. No significant increase in time to occlusion of the vein (31, 29.5 and 27.5 min for TAFI KO, PAI-1 KO and double-KO mice, respectively, vs. 28.5 min for WT mice) or the artery (26.8, 23.2 and 19.5 min for TAFI KO, PAI-1 KO and double-KO mice, respectively, vs. 17.5 min for WT mice) could be detected (times are represented as medians, n = 5–13; Fig. 3).

Figure 3.

 FeCl3-induced mesenteric thrombosis model. No significant increase in time to occlusion (time needed to form an occlusive thrombus) of the vein (A) and artery (B) could be detected; n = 5–13. KO, knockout; PAI-1, plasminogen activator inhibitor-1; TAFI, thrombin-activatable fibrinolysis inhibitor; WT, wild type.

Thromboembolism model

A mouse thromboembolism model was used to examine the relative roles of TAFI and PAI-1 in fibrinolysis. The percentages of fibrin in the lungs of TAFI KO mice (17% ± 15%, < 0.05) and double-KO mice (26% ± 20%, < 0.05) were significantly lower than in WT mice (100% ± 32%; Fig. 4). In contrast, no difference was observed for PAI-1 KO mice (97% ± 57%, > 0.05). As the fibrin concentration in the lungs of double-KO mice was similar to that in TAFI KO mice, the profibrinolytic effect in the double-KO mice is thus predominantly attributable to the gene deletion of TAFI and not to that of PAI-1.

Figure 4.

 Fibrin concentration in the mouse lungs upon thromboembolism induction. Fibrin concentrations are expressed as percentage relative to the percentage obtained in the wild-type (WT) mice. The percentage of fibrin in the lungs of thrombin-activatable fibrinolysis inhibitor (TAFI) knockout (KO) and double-KO mice was significantly lower than in WT mice, whereas no difference for plasminogen activator inhibitor-1 (PAI-1) KO mice was observed. *Significantly different from WT mice (< 0.005); n = 5.

Discussion

In this study, we generated and characterized mice with double deficiency of TAFI and PAI-1 to evaluate whether combined inactivation would result in an overt phenotype. These double-KO mice were viable, developed normally, and were fertile, which is consistent with data previously obtained for the single-KO mice [10,12,23,24]. However, it is surprising that a higher frequency of death was observed for the female breeding pair partners and their pups before reaching weaning age. These observations suggest that the hyperfibrinolytic state has a more pronounced impact in extreme conditions. One possible hypothesis for the high dropout of double-KO mice before weaning age is that these pups complete embryonic development but experience fatal postnatal bleeding events, owing to unstable fibrin formation caused by combined TAFI and PAI-1 gene deficiency. These pups either die within a few days after delivery or make adaptations in order to keep homeostasis at a survival level without showing physical abnormalities.

We continued this study by subjecting the different genotypes to a select panel of in vitro and in vivo thrombosis models. In this way, we determined whether, upon triggering, the double-KO mice behave differently from the single-KO mice, and whether one gene deletion predominates over the other. We therefore selected a few basic thrombosis models.

Double gene deficiency did not prolong bleeding times or affect blood loss in the tail bleeding model, which is in line with the observed absence of hemorrhage in the single-gene-deficient mice. Given the high variability of the data, however, the statistical test used is only sufficiently powered to detect major differences in bleeding times between the genotypes. Previous studies on bleeding times and blood loss reported conflicting data for the single-gene-deficient mice, which might be dependent on the particular bleeding model used, the different strain, sex or age of the mice, or a lack of power of the study [11,21,25,26]. However, even if the test were standardized, results would have to be interpreted with caution, because unstable clots in the presence of vascular spasm may be sufficient to temporarily protect against bleeding [22].

Interestingly, fibrinolysis was markedly enhanced in the double-KO mice, as observed both ex vivo in ROTEM and in vivo in the mouse thromboembolism model. However, the profibrinolytic effect was detected in all mice lacking TAFI, regardless of the presence or absence of PAI-1. Ex vivo, TAFI deficiency in mice has previously been studied in thromboelastograph analysis, which has revealed enhanced lysis of whole blood clots [26], in agreement with our observations. In vivo, lack of TAFI or PAI-1 has previously been assessed in various mouse thrombosis models, with different outcomes. In a photochemical injury model of arterial thrombosis, no effect of TAFI deficiency on the rate of acute thrombus formation was observed. Moreover, the rate of spontaneous reperfusion determined from the size of the residual thrombus after 24 h was not accelerated in TAFI KO mice [23,27]. A similar model of venous thrombosis also indicated a lack of impact of TAFI deficiency [23,27]. In other models, a variety of agents were injected intravenously to induce thrombus formation, i.e. thrombin, lipopolysaccharides, and factor X coagulant protein. No significant differences in survival rate or the extent of fibrin deposition in the lungs of TAFI-deficient or WT mice were detected [23,27]. On the other hand, thrombus weight was reduced in an FeCl3-induced thrombosis model in the venous circulation, but not in the arterial circulation [26]. Furthermore, fibrinolysis was enhanced in a batroxobin-induced pulmonary embolism model of TAFI KO mice, observed as a reduction in radioactivity accumulation in the lungs after [125I]fibrinogen administration [28]. Concerning the lack of PAI-1 in mice, inconsistent data were obtained in an FeCl3-induced carotid artery injury model. One research group reported that the residual thrombus area was diminished in PAI-1 KO mice as compared with WT mice, but the mean times to occlusion did not differ significantly [29,30], whereas another group confirmed the reduced thrombotic occlusion but found longer occlusion times, owing to the lack of PAI-1 [31]. In photochemically induced arterial and venous thrombosis models, deficiency of PAI-1 also led to a decrease in thrombus size and a prolongation of the time to occlusion [17,32,33]. Furthermore, venous thrombi developed less frequently in PAI-1 KO mice than in WT mice upon endotoxin injection in the footpad [21].

Despite the enhanced fibrinolysis in the mouse thromboembolism model, no significant difference in arterial or venous occlusion time between double-KO mice and WT mice was observed in our mesenteric thrombosis model. Moreover, it was previously been shown that thrombus weight was reduced in TAFI KO mice as compared with WT mice in an FeCl3-induced vena cava thrombosis model with 3.5% FeCl3, whereas only a slight difference was observed with 5% FeCl3 [24]. The absence of a profibrinolytic effect in our model might be explained by some limitations: (i) the use as a thrombotic agent of FeCl3, as this causes aggressive oxidative stress, which can result in denaturation or alteration of hemostatic proteins; (ii) the use of occlusion time as a parameter, as this may not detect subtle effects on thrombosis; (iii) the relatively short observation period (1 h) used to study the effects of antifibrinolytic proteins after initiation of the fairly aggressive thrombotic stimulus; and (iv) the high variability of the data, which implies that the statistical test used is only sufficiently powered to detect major differences in occlusion times between the genotypes.

In humans, TAFI deficiency has not been described so far. On the other hand, excessive fibrinolysis resulting from decreased PAI-1 levels has been reported in a few patients, and was associated with bleeding complications in response to trauma or surgery [34–38]. Hence, the physiologic consequences of loss of PAI-1 in humans appear to be more severe than in mice. The contrast between humans and mice may be explained by significant interspecies differences in the fibrinolytic sytem, e.g. the subtle structural differences between human and mouse PAI-1, the shorter half-life of mouse t-PA in plasma, and the resistance of mouse plasma clots to t-PA [39,40]. In addition, PAI-1 levels in mice are lower than those in humans, i.e. more than 10-fold lower in plasma and 500-fold lower in platelets [41]. However, upon injury in mice, PAI-1 can be released from the vessel wall and from neutrophil granulocytes into the circulation, thereby generating high concentrations of PAI-1 in the vicinity of the developing thrombus [17]. Consequently, it is generally accepted that mice can be used as tool with which to study the role of PAI-1 in thrombosis models, but conclusions drawn from mouse studies cannot unambiguously be translated into humans, and further studies on the role of human PAI-1 and/or TAFI need to be considered.

Overall, we focused on the physiologic relevance of TAFI and PAI-1 in mouse thrombosis models that are related to thrombotic disease states. In our in vivo mouse thrombosis models, coagulation was triggered by FeCl3 (mesenteric thrombosis model) or thromboplastin (thromboembolism model) to initiate the coagulation cascade. In these models, we did not add additional t-PA. This is an important difference from some previous studies in which a role of PAI-1 in resistance to thrombolysis was observed when a thrombolysis step was included, i.e. infusion of t-PA, human plasminogen, and heparin [42]. Although it might be that not all of our thrombosis models necessarily involve fibrinolysis, we could demonstrate a significant effect of TAFI gene deletion in mice, whereas no effect of PAI-1 gene deletion was observed.

Taken together, our data indicate that TAFI gene deletion predominantly contributes to the increased fibrinolytic capacity of TAFI/PAI-1 double-KO mice. A similar phenomenon was previously observed in mice with α2-antiplasmin gene inactivation, i.e. the occurrence of significantly faster lysis of a pulmonary plasma clot, irrespective of the presence or the absence of PAI-1 [43]. This indicates that, in contrast to TAFI and α2-antiplasmin, PAI-1 gene deficiency in mice does not seem to play a major role in the dissolution of a blood clot after challenge. However, more extensive in vivo studies, especially long-term thrombosis models with non-destructive stimuli, need to be performed to confirm this hypothesis.

With the analysis of combined TAFI and PAI-1 gene deficiency in mice, a further delineation of the function and importance of these proteins is provided. Moreover, this study highlights the major contribution of TAFI deficiency to the enhanced fibrinolytic capacity of the double-gene-deficient mice in some basic thrombosis models.

Acknowledgements

The authors wish to thank C. Vranckx and I. Scroyen for assistance with the PAI-1 and α2-antiplasmin ELISAs.

Disclosure of conflicts of interests

This study was supported by grant G.0833.09N of the Fund for Scientific Research-Flanders (FWO-Vlaanderen). E. Vercauteren is a PhD fellow of the Agency for the Promotion and Innovation through Science and Technology in Flanders (IWT-Vlaanderen).

Ancillary