Dingeman C. Rijken, Department of Hematology, Erasmus University Medical Center, Room Ee1393, Dr. Molewaterplein 50, 3015 GE Rotterdam, The Netherlands. Tel.: +31 10 408 94 48; fax: +31 10 408 94 70; e-mail: email@example.com
Summary. New thrombin activatable fibrinolysis inhibitor (TAFI) assays are necessary for studying the role of this fibrinolysis inhibitor in cardiovascular disease. The identification of a functional single nucleotide polymorphism (SNP) (1040C/T) leading to a TAFI-variant with increased stability but lower antigen levels has made the determination of functional activity even more essential. Therefore, we developed a new assay for the functional activity of TAFI in citrated plasma samples. This assay is based on the retardation of plasma clot lysis by TAFIa. TAFI activation was induced simultaneously with fibrin formation and lysis was mediated by rt-PA. The variability of other plasma components was minimized by a 20-fold dilution of the samples in TAFI-depleted plasma. Lysis times (−/+ potato carboxypeptidase inhibitor) and the TAFI-related retardation of clot lysis, the functional parameter of the assay, were determined in a group of 92 healthy volunteers, as well as TAFI antigen levels (electroimmunoassay) and two TAFI SNPs (−438A/G and 1040C/T). TAFI-related retardation was 19.8 ± 5.6 min (mean ± SD) and was correlated with the antigen level. The specific antifibrinolytic activity of TAFI was associated with the −438A/G and 1040C/T genotypes. Individuals with the 325Ile-variant had on average a 34% higher TAFI-specific antifibrinolytic activity than individuals with the 325Thr-isoform. The TAFI-related retardation in the two groups of individuals did not differ, as a lower level compensated for the higher specific antifibrinolytic activity of the 325Ile-isoform. This assay provides valuable information about the performance of different TAFI isoforms and constitutes a new method for studying the role of TAFI in cardiovascular disease.
Thrombin activatable fibrinolysis inhibitor (TAFI) or plasma procarboxypeptidase B was discovered about a decade ago and constitutes a connection between the clotting and fibrinolytic pathways [1–3]. Proteolytic cleavage of TAFI after Arg-92 by thrombin, plasmin, trypsin or the thrombin–thrombomodulin (TM) complex forms the active enzyme (TAFIa) and the glycosylated activation peptide [4–6]. The antifibrinolytic effect of TAFIa in vitro is dependent on the concentration of TAFI [7,8] as well as on the type and concentration of the plasminogen activator used . TAFIa removes C-terminal arginine and lysine residues present on partially degraded fibrin [7,10–12]. These residues have a pivotal role in plasminogen activation and in the lysis of a plasma clot. Consequently, TAFIa inhibits clot lysis.
There is evidence that a single nucleotide polymorphism (SNP) in the TAFI gene, which results in the substitution of a Thr by an Ile at position 325 of the protein , is located in a region that is related to the thermal stability of TAFIa and therefore to its antifibrinolytic potential. Hence, an Ile at position 325 in recombinant TAFI doubles its stability (half-life at 37 °C increases from 8 to 15 min) and results in an increased antifibrinolytic activity . However, the same SNP (1040 C/T) was found to be associated with TAFI antigen levels in plasma with the T allele, coding for the 325Ile-isoform, being linked to lower antigen levels . This probably arises from the fact that this SNP is in strong linkage disequilibrium with several other SNPs situated in the promoter region of the TAFI gene [13,15–17]. Indeed, the increased stability and activity of this isoform and its decreased antigen level pose a problem for the interpretation of results of TAFI assays.
Different methods have been employed for the determination of TAFI activity in plasma and recently, an assay was devised for the determination of endogenous TAFIa activity in plasma . Some of the TAFI activity assays used clot lysis times (LTs) [8,19,20]. However, most TAFI activity assays are based on the activation of TAFI zymogen and subsequent detection of TAFIa via the enzymatic cleavage of a small synthetic substrate [3,21–24].
Our aim was to develop an assay for the measurement of TAFI that would include TAFI activation, the limited stability of TAFIa as well as the removal of C-terminal lysines and arginines from partially degraded fibrin by TAFIa. This functional assay is based on the ability of activated TAFI to delay the lysis of a plasma clot mediated by r-tPA. A 20-fold dilution of the sample in TAFI-depleted plasma reduced the influence of other individual plasma components on lysis, which previously posed a problem for the specificity of lysis-based TAFI assays . In addition, this predilution assured the measurement of TAFI activity in a concentration range, which produced a proportional increase in response. By determining TAFI antigen levels as well as genotyping for the functional TAFI SNP (1040C/T–Thr325Ile) and a SNP in the promoter region (−438A/G) we were able to show that the response of the TAFI functional assay is dependent not only on the TAFI level but also on the stability and activity of the particular TAFI isoform.
In our view this novel TAFI functional assay can be used for the determination of potential TAFI activity in a variety of individual plasma samples and may be used for the identification and characterization of TAFI molecules with altered activity as well as for testing new pharmaceutical approaches for the modulation of the antifibrinolytic activity of TAFIa.
Actilyse (r-tPA) was kindly supplied by Boehringer Ingelheim (Ingelheim, Germany). Human thrombin was purchased from Sigma (St Louis, MO, USA) and both potato carboxypeptidase inhibitor (PCI) and human recombinant plasmin activator inhibitor 1 (PAI-1) were obtained from Calbiochem (La Jolla, CA, USA). For stability reasons, dilutions of the PAI-1 stock were performed in 20 mm Hepes, 1 mm EDTA, pH 6.5. Rabbit lung TM, with a specific activity towards thrombin of 1.2 units μg−1, was acquired from American Diagnostica Inc (Greenwich, CT, USA). Plasmin inhibitor (previously α2-antiplasmin) and human fibrinogen (plasminogen, vWF and fibronectin depleted) were obtained from Biopool (Umeå, Sweden) and Enzyme Research Laboratories (South Bend, IN, USA), respectively. Plasminogen was purified from human plasma by lysine-Sepharose chromatography . All the other reagents were from Merck (Darmstadt, Germany).
Pooled normal plasma
Blood was collected by venipuncture from 70 healthy individuals (30 men, 40 women, mean age 38.7 years) in plastic tubes containing 0.1 vol of 0.106 m trisodium citrate. Blood was centrifuged at 2000 g for 20 min at 20 °C. Women on oral contraceptives were excluded. The platelet-poor plasmas were pooled and used as calibrator for the Electroimmunoassay [24,26]. The pooled normal plasma was considered to contain 1 U mL−1 of TAFI.
A group of 92 healthy individuals (46 males and 46 females, mean age of 46.5 years, range 21–75 years) was recruited among hospital personnel and their families. All subjects gave informed consent. Blood was collected into 0.1 vol of 0.106 m trisodium citrate and centrifuged at 2000 g for 30 min at 4 °C. The plasma was stored at −80 °C in aliquots of 0.4 mL.
The TAFI-depleted plasma was prepared using an anti-TAFI IgG sepharose column as previously described by Van Tilburg et al. . Briefly, rabbit polyclonal anti-TAFI antibodies were coupled to CNBr-activated Sepharose-4B (Amersham Pharmacia Biotech, Uppsala, Sweden) as described by the manufacturer. Pooled normal plasma prepared from blood collected on citrate-phosphate-dextrose was applied to the anti-TAFI IgG sepharose column (10 mg IgG mL−1 sepharose) and the breakthrough fractions were tested for TAFI with a TAFI ELISA  and with the TAFI functional assay (see under). TAFI-depleted fractions were pooled and dialyzed against 50 mm Hepes, 100 mm sodium chloride, 20 mm sodium citrate (pH 7.4), in order to remove phosphate ions. The plasma was then aliquoted and frozen at −20 °C. TAFI-depleted plasma was used to dilute samples both in the TAFI functional assay and in the Electroimmunoassay.
TAFI functional assay
Individual citrated plasma samples were diluted 20-fold in TAFI-depleted plasma (a single batch of TAFI-depleted plasma was used to determine all individual samples). The diluted plasma samples (100 μL) were added to the wells of a microtitre plate containing 25 μL of a reaction mix. The mix was composed of thrombin (3.3 NIH units mL−1), TM (0.6 units mL−1), r-tPA (0.10 μg mL−1), CaCl2 (20 mm) and PCI (30 μg mL−1 where stated) in 50 mm Hepes buffer, pH 7.4 containing 0.1% w/v BSA. The concentrations between brackets refer to the final concentrations in the clotted plasma, which have been optimized as described previously . The wells were immediately covered with paraffin oil (Merck – No.107162) and the microtitre plate was placed in the prewarmed (37 °C) incubation chamber of a TECAN Sunrise® Microplate-reader (Giessen, The Netherlands). The optical density at 405 nm was monitored every minute for 150 min. A control with PCI was used for each sample. LT was defined as the time point corresponding to a 50% decrease in optical density and a sigmoidal regression was used for its determination. The TAFI-related retardation was determined for each sample by subtracting the LT in the presence of PCI (LT+PCI) from the LT in the absence of PCI (LT-PCI). Using this assay the intra- and inter-assay variability of the TAFI-related retardation of pooled normal plasma (20-fold diluted) were 7% and 13%, respectively (n = 30; using several batches of TAFI-depleted plasma).
TAFI Functional assay: dilution profile
Serial dilutions of citrated plasma in TAFI-depleted plasma were prepared and the TAFI-related retardation of clot lysis was measured as outlined above. The dilution profile was inspected for pooled normal plasma and for citrated plasma from two individuals who were homozygous for the −438 A/G and 1040 C/T TAFI polymorphisms. One individual had the −438 GG, 1040 CC genotype and the other the −438 AA, 1040 TT genotype.
TAFI antigen levels (Electroimmunoassay) and genotyping
Rocket immunoelectrophoresis  was used to measure TAFI antigen levels . The electroimmunoassay and genotyping were performed as described previously . The electroimmunoassay had an intra- and inter-assay variation coefficient of 6% and a detection limit of around 0.016 U mL−1 . The distribution of the individuals in the genotype subgroups according to gender was normal.
Values are expressed as mean ± SD or mean ± SEM as indicated. The normality of the distribution of each functional assay parameter was tested in the entire group and in the genotype subgroups using both parametric and non-parametric tests. One-way anova with a post-test for linear trend was used to look at the relation between TAFI levels and the TAFI genotype. Pearson's correlation coefficient was calculated to study the associations between the TAFI functional assay parameters and the TAFI Electroimmunoassay. Linear regression analysis was performed to evaluate the between-assay relationships in the TAFI genotype subgroups. Multivariate regression analysis was performed to assess the effect of TAFI levels and TAFI genotypes considered together. P-values < 0.05 were considered statistically significant.
A typical example of the clot lysis profiles obtained is shown in Fig. 1. The LT of TAFI-depleted plasma in the absence and presence of PCI were essentially identical (30.2 ± 3.6 and 31.6 ± 4.0 min, respectively, mean ± SD, n = 25 with two batches of TAFI-depleted plasma). Normal plasma diluted 20-fold in TAFI-depleted plasma showed a 1.5-fold prolonged LT, which was abolished by the addition of PCI. Undiluted normal plasma showed a twofold prolonged LT. Figure 1 also demonstrated that the levels of other constituents of plasma essential for clot formation and lysis were not appreciably altered by the immunoadsorption procedure, as the LT+PCI of the plasma remained constant before and after TAFI depletion.
From serial dilutions of pooled normal plasma in TAFI-depleted plasma we obtained the dilution profile for the TAFI-related retardation, shown in Fig. 2. Increasing TAFI concentrations produced an increase in retardation up to about 0.20 U mL−1. At higher TAFI concentrations a plateau in the response was reached.
TAFI functional assay
On the basis of the dilution profile of pooled normal plasma (Fig. 2) we used a 20-fold dilution of plasma samples in TAFI-depleted plasma in order to determine TAFI-related retardation in a relatively sensitive range. LTs were derived for each sample in the absence and presence of PCI. Normal distributions were obtained for all assay parameters determined in plasma from 92 healthy individuals. Mean values for LT-PCI and LT+PCI, and for TAFI-related retardation are presented in Table 1. No associations were found between the LT-PCI or LT+PCI or the TAFI-related retardation and gender. There was also no relationship between TAFI-related retardation or the LT-PCI and age. The LT+PCI, on the contrary, progressively decreased with age (r = 0.311; P = 0.003). In the genotype subgroups, normal distributions were observed for the LT-PCI, LT+PCI (not shown) and the TAFI-related retardation (Fig. 3, for 1040 CT genotype). The TAFI functional assay parameters were not associated with TAFI genotypes (−438A/G and 1040C/T, tested by anova).
Table 1. Distribution of −438 A/G and 1040 C/T SNPs of the thrombin activatable fibrinolysis inhibitor (TAFI) gene in 92 healthy individuals and mean values of TAFI functional assay parameters, TAFI antigen levels and TAFI-specific antifibrinolytic activity
LT -PCI (SD)
LT +PCI (SD)
TAFI-related retardation (SD)
TAFI antigen (SD)*
TAFI-specific antifibrinolytic activity (SD)
TAFI functional assay parameters (LT -PCI, LT +PCI, and TAFI-related retardation) expressed in min. TAFI antigen levels were determined with the electroimmunoassay and expressed in U mL−1. The TAFI-specific antifibrinolytic activity was expressed in arbitrary units (AU).
*TAFI antigen was significantly associated with TAFI genotypes, as previously described in Ref. . No association was found between the TAFI functional assay parameters and TAFI genotypes.
Influence of other plasma components on the TAFI functional assay
We investigated the influence of a number of plasma components, namely fibrinogen, plasminogen, plasmin inhibitor and PAI-1, on the TAFI functional assay parameters. The assay was performed essentially as described under methods. Each of the purified components was added in increasing amounts to normal plasma which was then diluted 20-fold in TAFI-depleted plasma. As shown in Fig. 4, fibrinogen (Fig. 4A), plasminogen (Fig. 4B) or plasmin inhibitor (Fig. 4C) had no effect on the TAFI functional assay parameters (P > 0.05 for all, repeated measurements test) in a wide concentration range, up to a threefold increase of the component concentration in the normal plasma. An increase in PAI-1 concentration in normal plasma up to 2000 ng mL−1 (100-fold) had no effect on the TAFI functional assay parameters (Fig. 4D–P > 0.05 for all, repeated measurements test). Only at PAI-1 concentrations of 10 000 ng mL−1 (500-fold increase) the TAFI functional assay parameters became considerably prolonged (96.6 ± 6.8 min against 46.1 ± 2.4 min for LT-PCI, 54.7 ± 0.2 min against 30.3 ± 2.8 min for LT+PCI, and 41.9 ± 6.8 min against 15.8 ± 3.7 min for TAFI-related retardation).
Correlation of TAFI functional assay parameters with TAFI antigen
Both the LT-PCI and LT+PCI displayed a weak correlation with the TAFI antigen levels (Table 1) determined with the Electroimmunoassay (r = 0.205; P = 0.05 and r = −0.245; P = 0.02, respectively). After correction for age, the association between LT+PCI and the TAFI antigen levels became weaker. TAFI antigen levels were not associated with age or gender. The TAFI-related retardation correlated significantly with the TAFI antigen levels (r = 0.357; P = 0.0005).
When the different genotypes were taken into account the correlation of the TAFI-related retardation with the TAFI antigen increased in strength for the −438GG (r = 0.497, P = 0.0003), −438AG (r = 0.517, P = 0.0011) and the 1040CC (r = 0.430, P = 0.003), 1040 CT (r = 0.490, P = 0.0015) genotypes (Fig. 5A,B). For the −438AA and 1040TT genotype subgroups no correlation was found between the TAFI-related retardation and the TAFI antigen levels, which may be due to the small sample sizes. The regression lines (Fig. 5B) are consistent with the notion that individuals with the 1040T allele (TAFI-325Ile) have higher TAFI activity (TAFI-related retardation) per TAFI antigen concentration than individuals with the 1040C allele (TAFI-325Thr).
Multivariate regression analysis illustrated that the TAFI antigen on its own explained 13% of the variation in the TAFI-related retardation and that the TAFI antigen together with the −438AG and 1040CT genotypes clarified around 22% of the variation. Furthermore, the assay variability could explain roughly 50% of the variation in the TAFI-related retardation. This leaves nearly 30% of variability unexplained and points to additional differences between TAFI functional activity and TAFI antigen level.
TAFI-specific antifibrinolytic activity
The TAFI-specific antifibrinolytic activity was defined as the ratio between the TAFI-related retardation (min) and the TAFI antigen level determined with the Electroimmunoassay (U mL−1) and values are given in arbitrary units (AU) (Table 1). The box plots in Fig. 6 depict the association of the TAFI-specific antifibrinolytic activity with the −438 A/G (Fig. 6A) and 1040 C/T (Fig. 6B) TAFI genotypes. From these results we concluded that individuals who are heterozygous or homozygous for the −438A allele or for the 1040T allele have an increased TAFI-specific antifibrinolytic activity (by anova, post-test for linear trend: r = 0.291; P = 0.004 and r = 0.306; P = 0.003, respectively). In addition, from the mean TAFI-specific antifibrinolytic activities in the TAFI genotype subgroups we can estimate the increase in antifibrinolytic activity because of the 1040 C/T TAFI SNP. On average, the subgroup with 325Thr/Ile (1040CT) and 325Ile/Ile (1040TT) showed respectively 14% and 34% greater specific antifibrinolytic activity than the subgroup with 325Thr/Thr (1040CC).
Dilution profile of TAFI variants
Dilution profiles were constructed with plasma from individuals with the 1040CC genotype and 1040TT genotype (Fig. 7) in order to further examine the difference in TAFI-specific antifibrinolytic activity. Plasma of the CC individual displayed a response similar to that of the pooled normal plasma (Fig. 2), in agreement with the high frequency of the C-allele in a normal population (Table 1). Plasma of the 1040TT individual, on the contrary, exhibited a considerably higher response and plateau than that of the 1040CC individual, corroborating a higher specific antifibrinolytic activity of the 325Ile-isoform.
This distinct response means that pooled normal plasma cannot be used for the calibration of this TAFI functional assay.
Thrombin activatable fibrinolysis inhibitor is a carboxypeptidase B zymogen present in plasma, which has the capacity to suppress fibrinolysis. This is accomplished by removing newly plasmin-generated C-terminal lysine and arginine residues in fibrin that function as high affinity binding sites for plasminogen [28,29], thus leading to the downregulation of plasmin generation [11,12,30].
To incorporate this role of TAFI in fibrin clot lysis into functional TAFI measurements, a novel assay was developed for the determination of TAFI functional activity in plasma. In order to mimic physiological conditions as much as possible, a high plasma concentration was used in the assay (80% plasma, final concentration). Fibrin generation and simultaneous TAFI activation were achieved by the addition of thrombin, TM and calcium. The conditions for optimal TAFI activation, i.e. maximal retardation of clot lysis, have been described previously . The lysis of the plasma clot is mediated by r-tPA that is included in the reaction mix and not in the plasma sample in order to minimize its inhibition by plasma proteinase inhibitors (such as PAI-1) prior to the assay measurements. Subsequently, the lysis profile is obtained by monitoring the OD at 405 nm while incubating the microtitre plate at 37 °C. Finally, every sample is diluted 20-fold in TAFI-depleted plasma and determined in the absence and in the presence of PCI. The difference between clot LTs in the absence and in the presence of PCI is thus specific for the contribution of TAFI. In order to study the assay performance we executed the TAFI functional assay in samples from 92 healthy individuals.
As shown in Fig. 4, the 20-fold dilution in TAFI-depleted plasma ensures a minimal interference of other plasma components on the determination of TAFI functional activity in different individuals. In the particular case of PAI-1 (Fig. 4D), it would be expected that when a 1:1 molar ratio between PAI-1 added and r-tPA used in the assay is reached an effect would be observed in the LTs. To our surprise, at a PAI-1 concentration as high as 2000 ng mL−1 no effect was detected. The explanation possibly lies in the particular conditions of the TAFI functional assay. The presence of thrombin/TM in the assay will activate protein C, which will form a tight 1:1 complex with PAI-1. The presence of an excess of thrombin may also lead to inactivation of PAI-1 via the formation of a ternary complex with vitronectin. Moreover, the assay is performed close to physiological conditions (pH 7.4 and 37 °C), conditions under which PAI-1 is known to rapidly lose activity.
We also determined the TAFI antigen levels and genotyped the 92 healthy individuals for the −438A/G and 1040C/T TAFI SNPs. Recently, it was shown that some anti-TAFI IgG preparations have different affinities towards the two TAFI isoforms (Thr325Ile) of the 1040C/T SNP [24,31]. This leads to a concentration- and genotype-dependent artefact in the TAFI antigen measurement. However, we have previously shown that some assays, among which the TAFI electroimmunoassay used in the present study, are insensitive to this artefact. By the use of such assays it was confirmed that the TAFI concentration in plasma is associated with the TAFI genotype, although TAFI antigen levels in 1040CC carriers were only 20%–30% higher than in 1040 TT carriers .
A weak correlation was found between the TAFI-related retardation and TAFI antigen levels (r = 0.357). Taking into account the TAFI genotype improved the relationship between the TAFI-related retardation and TAFI antigen. The genotypes comprising the stable TAFI variant (TAFI-325Ile) displayed the highest response (upper line –Fig. 5B). The dispersion found here between assays can only partially be accounted for by assay variability or by known TAFI variability (SNPs). This points to additional differences between TAFI molecules and/or their activity as well as to the need for more detailed information concerning the role of TAFI in distinct assays.
We studied the response of pooled normal plasma (Fig. 2) as well as two genotype-specific individual plasmas (1040 CC and TT genotype, Fig. 7). The genotype-specific plasma samples confirm a distinct response of samples with known differences in stability and activity with the more stable 325Ile-variant presenting a higher response and plateau than the 325Thr-variant. On the contrary, these results also indicate that pooled normal plasma cannot be used as the calibrator, as a combination of TAFI forms will generate an intermediate curve and will lead to under- or over-estimation of the individual samples.
Schneider et al.  have shown that different isoforms of TAFI lead to different retardation profiles with distinct plateaux. Using purified recombinant variants they reported that the 325Ile-variant displayed a 30%–50% higher antifibrinolytic effect, compared with the 325Thr-variant. We also found an increased TAFI-specific antifibrinolytic activity for the more stable 325Ile-variant (Fig. 7) and accordingly, the specific TAFI antifibrinolytic activity was associated with the TAFI genotype (Fig. 6). On average, the subgroup with 325Thr/Ile (1040C/T) and 325Ile/Ile (1040T/T) showed respectively 14% and 34% higher TAFI specific antifibrinolytic effect than the subgroup with 325Thr/Thr (1040C/C). This is consistent with the results of Schneider et al. , especially when we take into account that here individual plasmas are being studied. The mechanism behind the distinct plateaux found by Schneider et al. with recombinant TAFI variants and in this study with native TAFI variants from plasma remains unclear. During the preparation of this manuscript Leurs et al.  and Walker and Bajzar  demonstrated that TAFIa influenced clot lysis through a threshold-dependent mechanism. They showed that according to this mechanism not only the TAFI concentration, but also the time course of TAFI activation, the stability of TAFIa and the tPA concentration are determinant factors for the clot dissolution. This might help to elucidate the distinct retardation profiles observed in the present work (Fig. 7 and in Ref. ) as the more stable variant will remain longer above the threshold.
It is interesting to note that the 325Ile-variant with the higher specific TAFI antifibrinolytic activity is present at a slightly lower concentration in plasma (Table 1). This is explained by the linkage between the functional TAFI SNP (1040C/T) and SNPs in the TAFI promoter region. The two effects compensate for each other and, as a result, the TAFI functional activity (TAFI-related retardation) did not differ in the genotype subgroups (Table 1). If TAFI functional activity is found to be associated with risk for some cardiovascular disease, it will be interesting to perform TAFI genotyping and TAFI antigen determination in order to investigate the contribution of these factors to the risk.
Thus, the TAFI functional parameter of this new assay (TAFI-related retardation) seems to be related to a particular combination of factors (e.g. concentration, stability, activity) for each individual. The understanding of these particular combinations and their relation to disease states will be necessary to further elucidate the involvement of TAFI in cardiovascular disease.
We wish to thank Nico van Tilburg, Leiden University Medical Center, Leiden, The Netherlands, for his useful technical assistance.