Measurement of the Free Form of TFPI Antigen in Hyperlipidemia
Relationship Between Free and Endothelial Cell–Associated Forms of TFPI
Abstract Tissue factor pathway inhibitor (TFPI), a protease with three tandem Kunitz-type (K1, K2, and K3) domains, inhibits the initial reaction of the TF-mediated coagulation pathway. TFPI occurs in a free and a lipoprotein-associated form in plasma as well as an endothelial cell–associated form on vascular walls. In a previous study we had demonstrated that free-form TFPI activity was lower in hyperlipidemic patients. In the present study we established a new enzyme immunoassay method for measuring free-form TFPI antigen; this new method uses a monoclonal antibody that recognizes the K3 domain of free-form TFPI but not lipoprotein-associated TFPI. Free-form TFPI antigen was significantly lower in hyperlipidemic patients compared with those in normolipidemic individuals. We applied this new method to measure the amount of endothelial cell–associated TFPI, which can be released by heparin injection, as “free-form TFPI.” We found that free-form TFPI antigen in plasma was positively correlated with the endothelial cell–associated form. These results indicate that both of these forms of TFPI are in equilibrium in vivo and that our new method can be used for assessing changes in the levels of endothelial cell–associated TFPI antigen and, hence, for assessing thrombotic tendencies in various disease states.
- Received September 12, 1995.
- Accepted March 26, 1996.
TFPI, a protease inhibitor with K1, K2, and K3,1 inhibits the initial steps of the extrinsic blood coagulation cascade by forming complexes with FXa, FVIIa, and TF.2 3 Most TFPI is synthesized by vascular ECs4 5 and is associated with these cells.6 7 8 TFPI from plasma can be separated by gel filtration into free and lipoprotein-associated forms, ie, an LDL/VLDL-associated form and an HDL-associated form. Although it was once believed that most TFPI was associated with lipoproteins in plasma and that only ≈5% of TFPI in plasma was in the free form,7 9 10 11 we recently demonstrated that ≈45% of plasma TFPI is in the free form by measuring each form of TFPI activity by gel filtration.12 We also found that free-form TFPI activity was lower in hyperlipidemic patients and in those who were undergoing repeated plasmapheresis. To further examine TFPI levels in various diseases we developed an EIA method. Using a monoclonal antibody that recognized the K3 domain of TFPI, we were able to detect only the free but not the lipoprotein-associated forms of TFPI in plasma.13 Although TFPI antigen levels in various diseases have been reported,14 15 16 17 only total plasma TFPI antigen was measured in these studies. Efforts have also been made to elucidate the relation between EC-associated TFPI and plasma TFPI in various human diseases18 19 and in diet-induced hypercholesterolemia in monkeys20 ; however, no correlation between the two forms of TFPI has yet been established. On the contrary, Hansen et al21 recently reported that downregulation of LDL did not affect the levels of EC-associated TFPI in hyperlipidemia. In the present study we established an EIA method specific for the free form of TFPI in plasma and applied this method to measure the levels of EC-associated TFPI, which was released by intravenous heparin injection. We then analyzed TFPI antigen in hyperlipidemic patients by this new method and compared these findings with those obtained by an earlier method. We discuss the relationship between free-form TFPI antigen in plasma and EC-associated TFPI antigen.
BSA and bovine milk proteins (Block Ace) were obtained from Boehringer Mannheim and Dai-Nippon Seiyaku, respectively. Rabbit anti-human rTFPI polyclonal antibody and mouse anti-human rTFPI monoclonal antibody were prepared as described.13 Horseradish peroxidase–labeled goat anti-rabbit IgG was purchased from Bio-Rad Laboratories, and an ELISA kit for total plasma TFPI antigen produced by Kaketsuken was purchased from Sanko Pure Chemicals. Porcine mucosal sodium heparin was purchased from Novo-Nordisk A/S. Cholesterol concentration was measured with a commercially available kit, the Cholesterol E test from Wako Junyaku. Multiwell plates (type H) were obtained from Sumitomo Bakelite. As described in a previous article,12 TFPI activity was measured by the ability of TFPI complex to inhibit activation of FX by TF-FVIIa complex. Free-form TFPI activity was measured after gel filtration of plasma through a column of Superdex 200 HR (26/60 from Pharmacia LKB Biotechnology).
Blood Sampling and Plasma Preparation
Blood specimens were obtained from normolipidemic individuals (41 healthy volunteers aged 22 to 64 years) and 45 hyperlipidemic patients (aged 8 to 80 years) at the National Cardiovascular Center Hospital, Osaka, Japan (Table⇓). Of the 45 hyperlipidemic patients 34 had not undergone plasmapheresis; their ages ranged from 23 to 80 years. Plasma from normolipidemic individuals and hyperlipidemic patients was collected 10 minutes after injection of heparin (30 U/kg IV). Before heparin injection, oral informed consent to participate in a clinical examination for measurements of lipoprotein lipase and coagulation factors (including TFPI) was obtained from all study subjects. The other 11/45 patients (aged 8 to 64 years) with heterozygous familial hyperlipidemia received repeated, biweekly plasmapheresis treatment, during which an apheresis column (Evaflux 5A from Kurare or Liposorber LA-15 from Kanegafuchi Kagaku) was used. Blood from these 11 patients was collected before heparin was infused. The diagnosis of hyperlipidemia was established according to the criteria of the Japan Atherosclerosis Society (total serum cholesterol concentration >5.7 mmol/L; Table⇓). Blood was collected into evacuated tubes containing trisodium citrate (final concentration, 0.38% [wt/vol]) and plasma was separated by centrifugation. Plasma specimens were stored at −80°C until assay.
EIA of TFPI Antigen
TFPI antigen was measured by a sandwich EIA method with monoclonal and polyclonal antibodies against human rTFPI as described.13 Each microplate well was coated with 50 μL of monoclonal anti-human rTFPI antibody (10 μg/mL in TBS). After a 2-hour incubation at room temperature each microplate well was soaked with 200 μL TBS containing diluted (×1/4) Block Ace. The plates were left for 2 hours at room temperature and then washed with TBS–Tween 20 (20 mmol/L Tris HCl, pH 8.0, containing 0.5 mol/L NaCl and 0.05% Tween 20). The rTFPI standard was serially diluted with Tris HCl buffer (0.5% BSA and 0.6 mol/L NaCl) and plasma was diluted (1/10 or 1/100) with Tris HCl buffer containing 0.6 mol/L NaCl. After addition of 50 μL of sample and rTFPI standard to each well, the plate was incubated overnight at 4°C and antigen measured as described.13 The amounts of TFPI antigen were calculated from a standard curve obtained from the serially diluted rTFPI. The level of TFPI antigen in the control plasma in each microplate was defined as 100%. The concentration of standard rTFPI was determined by titration with human FXa, with the assumption that 1 mol TFPI (Mr, 42 kDa) interacts with 1 mol FXa. The concentration of human FXa was determined by titration with p-nitrophenyl p′-guanidinobenzoate.
Fig 1⇓ shows the standard curve for the TFPI antigen assay in which TFPI antigen was measured at a range of 0.5 to 5.0 ng/mL. The intra-assay coefficient of variation was estimated to be 9.08% from 8 duplicate plasma samples. The inter-assay coefficient of variation was estimated to be 6.30% from 8 determinations of 1 plasma specimen.
Values were expressed as mean±SD. Pearson’s correlation coefficient was calculated for the different variables, and a value of P<.05 was considered statistically significant. The nonparametric Wilcoxon test was employed for statistical comparisons, with P<.05 indicating statistical significance.
Measurement of Free-Form TFPI Antigen in Human Plasma
In a previous study13 we had shown that a monoclonal antibody against rTFPI recognized the K3 domain of TFPI and that this antibody reacted with free-form TFPI only and not with lipoprotein-associated forms of TFPI. In the present study we first attempted to establish the optimum conditions for measurement of free-form TFPI antigen in human plasma by using monoclonal and polyclonal antibodies against rTFPI. BSA concentrations >0.5% in the buffer that was used for dilution of rTFPI were found to be optimal. For dilution of plasma there was no difference between BSA-free buffer and buffer with 0.5% BSA. This result indicated that albumin or other proteins in human plasma prevented the loss of TFPI antigen during the assay. In accordance with these results we selected the 0.5% BSA buffer for dilution of rTFPI and BSA-free buffer for dilution of plasma. However, BSA-containing buffer can also be used for dilution of both rTFPI and plasma.
Next we examined the effect of heparin on TFPI antigen level. When plasma was diluted in buffer containing 0.15 mol/L NaCl, the absorbance decreased with increasing amounts of heparin (Fig 2⇓). In contrast, when plasma was diluted with buffer containing 0.6 mol/L NaCl, the effect of heparin was not significant at concentrations <1 U/mL heparin (Fig 2⇓). We also examined the effect of salt concentration in the buffers for dilution of rTFPI and plasma. The difference between the buffer with 0.15 mol/L NaCl and that with 0.6 mol/L NaCl was not significant. However, reaction rates in buffers with 0.8 mol/L or 1 mol/L NaCl were clearly lower than those in buffer with 0.15 mol/L NaCl. In accordance with these results we selected the buffer with 0.6 mol/L NaCl for dilution of the samples to abolish the effects of heparin.
In another previous study12 we demonstrated that free-form TFPI activity could be measured quantitatively after gel filtration of plasma in a buffer with 1 mol/L NaCl. We investigated whether the same result could be achieved in a buffer with 0.6 mol/L NaCl, and we also examined whether heparin-releasable TFPI could be measured in its free form under these conditions. The elution profile of TFPI activity after gel filtration of plasma in the buffer with 0.6 mol/L NaCl was identical to that for the buffer with 1 mol/L NaCl, as we had reported previously12 (Fig 3⇓). When postheparin plasma was collected from 5 normolipidemic individuals and 5 hyperlipidemic patients and then subjected to gel filtration in buffer with 0.6 mol/L NaCl, an increase in TFPI activity was found for free-form TFPI only (Fig 3⇓). These results are consistent with those reported for buffers with 1 mol/L NaCl.12 Because our EIA method for TFPI antigen detects the free form only, these results indicate that heparin-releasable TFPI, ie, EC-associated TFPI, can be measured as “free-form TFPI” in a buffer with 0.6 mol/L NaCl.
We compared the concentration of TFPI antigen in plasma (as determined by our new EIA method) with the values for total plasma TFPI activity or free-form TFPI activity (as measured by the earlier method) in plasma from normolipidemic individuals (n=13), hyperlipidemic patients (n=12), and hyperlipidemic patients who had received repeated plasmapheresis treatment (n=8). TFPI antigen was positively correlated with free-form TFPI activity (r=.819, P<.0001), but the correlation of TFPI antigen with total plasma TFPI activity was poor (r=.488, P<.005) (Fig 4A⇓ and 4B⇓). On the other hand, we found no correlation between free-form TFPI activity and total plasma TFPI antigen (Fig 4C⇓). The correlation coefficient for the total plasma TFPI activity–total plasma TFPI antigen relation was calculated to be .418 (P<.05; Fig 4D⇓, dotted line). When two markedly different values were excluded (circles), the correlation coefficient increased to .615 (P<.0005; Fig 4D⇓, solid line). These results indicate that the best correlation was observed between free-form TFPI activity and free-form TFPI antigen. The reason for the relatively poor correlation between total plasma TFPI activity and free-form TFPI activity or antigen and that between total plasma TFPI activity and total plasma TFPI antigen will be discussed in a subsequent section.
We also compared TFPI antigen concentration, as determined by our new EIA method, with TFPI antigen concentration as measured with an EIA kit (Kaketsuken) that detects both lipoprotein-associated forms and the free form of TFPI (Fig 5⇓). The mean values for free-form TFPI antigen and total plasma TFPI antigen were 19.2 and 39.4 ng/mL, respectively. The ratio of free-form TFPI antigen to total plasma TFPI antigen was approximately 0.5. This result is consistent with our previous finding that showed ≈45% of total plasma TFPI activity to be in the free form.12 All of these results indicate that although the new EIA method specifically detects free-form TFPI, it can also be used to measure the EC-associated TFPI that is released after heparin injection.
Relation Between Free-Form TFPI Antigen and EC-Associated TFPI Antigen in Normolipidemic and Hyperlipidemic Subjects
Fig 6⇓ shows free-form TFPI antigen in plasma from normolipidemic individuals (n=41), hyperlipidemic patients (n=34), and hyperlipidemic patients who were receiving repeated plasmapheresis treatment (n=11). The amount of free-form TFPI antigen in hyperlipidemic patients was significantly lower than that in normolipidemic individuals (P<.001), whereas these values in hyperlipidemic patients who were receiving repeated plasmapheresis treatment were even lower (P<.0001). These results are consistent with our previous findings on the changes in free-form TFPI activity in hyperlipidemic patients: in that study12 we speculated that the decrease in free-form TFPI activity reflected a decrease in EC-associated TFPI activity. To confirm this hypothesis, we performed an experiment in which normolipidemic individuals (n=28) and hyperlipidemic patients who were not receiving repeated plasmapheresis treatment (n=13) were given heparin (30 U/kg IV). Blood was drawn 10 minutes after heparin was administered, and the TFPI antigen that had been released from ECs was assayed by our new EIA method. Free-form TFPI antigen in plasma was positively correlated with TFPI antigen released from ECs (r=.581, P<.0001; Fig 7A⇓, solid line). Because the value for one sample deviated markedly from the others, we excluded this outlier and recalculated the correlation coefficient, which increased to .707 (dotted line). On the other hand, the correlation of EC-associated TFPI antigen with total plasma TFPI antigen was less (r=.472; Fig 7B⇓). Free-form and EC-associated TFPI levels in normolipidemic individuals were 119±26% and 870±213%, respectively, and the corresponding values in hyperlipidemic patients were 85±22% and 679±226%. The difference in free-form TFPI level between normolipidemic and hyperlipidemic individuals and the difference in EC-associated TFPI level between them were found to be significant (P<.05 and P<.01, respectively). These findings led us to conclude that the changes in free-form TFPI generally reflected the changes in EC-associated TFPI and that both forms of TFPI in hyperlipidemic patients were usually lower than those in normolipidemic individuals.
In a previous study12 we had demonstrated that free-form TFPI activity in plasma was significantly lower in hyperlipidemic compared with normolipidemic individuals. From this finding we speculated that the decrease in free-form TFPI activity in hyperlipidemic patients might reflect a decrease in EC-associated TFPI. To confirm this idea we developed a new EIA method for both free-form and EC-associated TFPI in plasma, which employed a monoclonal antibody that recognized only free-form but not lipoprotein-associated TFPI antigen in plasma. Lipoprotein-associated TFPI did not react with the monoclonal antibody because its epitope (the K3 domain) was “masked” in lipoprotein-associated TFPI.13 In the present study we established the optimum conditions for measuring both free-form TFPI in plasma and that released after heparin injection, ie, EC-associated TFPI. As demonstrated previously22 23 TFPI has two heparin-binding sites, one in the K3 domain and the other in the carboxy-terminal basic domain. Therefore, we presumed that heparin could interfere with the interaction between TFPI and the monoclonal antibody by binding to the K3 domain. In fact, levels of TFPI antigen were markedly decreased by adding heparin to a buffer containing 0.15 mol/L NaCl (Fig 2⇑). To overcome this problem we examined the effect of salt concentration in the reaction mixture on the TFPI level, and 0.6 mol/L NaCl was found to be the optimal concentration.
As demonstrated in the present study, free-form TFPI antigen in plasma was correlated with free-form TFPI activity (Fig 4⇑). This result indicates that TFPI antigen has TFPI activity, although a monoclonal antibody against the K3 domain of TFPI was employed in the EIA method. However, it is clear that the EIA method does not discriminate full-length TFPI from TFPI with a deleted carboxy-terminal basic domain. That is, the measured levels of TFPI antigen in the present study include TFPI with a deleted carboxy-terminal basic domain (if any) and full-length TFPI. Recently Broze et al24 suggested that TFPI with a deleted carboxy-terminal basic domain and even a deleted K3 domain was present in normal plasma. However, for the following reasons, our results unequivocally indicate that the majority of free-form TFPI in normolipidemic and hyperlipidemic plasma is K3 domain–containing TFPI: (1) the TFPI antigen level in the free-form TFPI fraction obtained by gel filtration of plasma was not significantly different from that measured with the ADI kit for total plasma TFPI antigen, in which antibodies against K1 and K2 domains were used (data not shown); and (2) the finding that free-form TFPI antigen was ≈50% of the total TFPI antigen is consistent with our previous measurement of TFPI activity.12 Despite these results we cannot exclude the possibility that truncated forms of plasma TFPI may be lipoprotein associated in patients with other diseases, such as disseminated intravascular coagulation.
In contrast to the strong correlation between free-form TFPI antigen and free-form TFPI activity, the correlation between free-form TFPI antigen and total plasma TFPI activity was relatively poor (r=.488), as shown in Fig 4B⇑. The correlation between free-form TFPI activity and total plasma TFPI activity was also poor (r=.548, data not shown). Furthermore, the correlation between total plasma TFPI antigen and total plasma TFPI activity was not as high as expected (r=.615), as shown in Fig 4D⇑. It has been suggested that the anticoagulant activity of lipoprotein-associated forms of TFPI is markedly lower than that of free-form TFPI.10 When we compared the TFPI activity of LDL- and HDL-associated forms of TFPI with that of free-form TFPI antigen (using the EIA kit for total plasma TFPI after gel filtration of control plasma), we found that free-form TFPI activity was 6.7 times higher than that of the LDL-associated form and 2.5 times higher than that of the HDL-associated form in terms of nanograms of TFPI antigen (data not shown). We suspect that the discrepancy between TFPI activity and TFPI antigen may have been caused by the different specific activities of these forms in plasma. In particular, these levels are different in each hyperlipidemic patient and are probably different in normolipidemic individuals as well. Therefore, it is not surprising that the correlation between free-form TFPI and total plasma TFPI, in terms of their activity and antigen levels, was relatively poor. On the basis of these speculations, we recommend that free-form TFPI, and preferably lipoprotein-associated TFPI, be measured rather than total plasma TFPI to reveal the significant role of TFPI in various diseases.
It is generally agreed that most TFPI is synthesized in ECs and binds to proteoglycans on the cell surface and that some portion of the EC-associated TFPI is continuously released into plasma as the free form. Some of the free-form TFPI released from ECs then binds to lipoprotein particles (thereby becoming lipoprotein-associated TFPI) by some as-yet-unknown mechanism. On the basis of this consideration, we speculate that the decrease in free-form TFPI in the plasma of hyperlipidemic patients may have been caused by an increase in LDL-associated TFPI and/or a decrease in EC-associated TFPI. The relatively high LDL cholesterol levels in the hyperlipidemic patients would accelerate the conversion of free-form TFPI to lipoprotein-associated TFPI, as has been demonstrated in diet-induced hypercholesterolemia in monkeys20 and in familial hyperlipidemia in humans.25 26 The decrease in EC-associated TFPI may be due to EC dysfunction and/or “consumption” of the EC-associated TFPI owing to the hypercoagulability in these patients. Although free-form TFPI can be reconverted to EC-associated TFPI, lipoprotein-associated TFPI cannot bind to ECs because the heparin-binding sites of TFPI have been masked, as evidenced by their inability to bind to a heparin-conjugated column.10 Hansen et al21 have recently suggested that EC-associated TFPI and LDL-associated TFPI behave independently in the plasma of hyperlipidemic patients who are receiving lipid-lowering drugs. However, in hyperlipidemia it is very difficult to detect a decrease in the levels of EC-associated TFPI because they are markedly higher than those of plasma TFPI and the decrease is restricted to localized areas in the vascular wall. In fact, we found no significant difference in EC-associated TFPI levels between normolipidemic and diet-induced hypercholesterolemic monkeys.20 We speculate that a combination of the changes in the equilibrium between free-form and lipoprotein-associated TFPI in plasma and EC-associated TFPI caused decreases in both free-form TFPI in plasma and EC-associated TFPI. As demonstrated in Fig 7A⇑ free-form TFPI antigen in plasma was positively correlated with EC-released TFPI antigen, and an even stronger correlation was calculated by excluding one outlier. These results indicate that the free-form TFPI level in plasma generally reflects the level of EC-associated TFPI, although some individual cases show a poor relation between free-form TFPI and EC-associated TFPI owing to unknown mechanisms. We believe that further experiments on these TFPI antigens and use of our new EIA method may clarify the relation between free-form TFPI levels and EC-associated TFPI levels in various disease states.
Selected Abbreviations and Acronyms
|K1, K2, K3||=||Kunitz-type (domains) 1, 2, 3|
|(TF)PI||=||(tissue factor) pathway inhibitor|
This study was supported in part by a grant from the Japan Cardiovascular Research Foundation, by a grant-in-aid for Research Promoting Comprehensive Longevity Science, and by a Research Grant for Cardiovascular Diseases (5A-2) from the Ministry of Health and Welfare of Japan.
Wun TC, Kretzmer KK, Girard TJ, Miletich JP, Broze GJ. Cloning and characterization of a cDNA coding for the lipoprotein-associated coagulation inhibitor shows that it consists of three tandem Kunitz-type inhibitory domains. J Biol Chem. 1988;263:6001-6004.
Broze GJ, Warren LA, Novotny WF, Higuchi DA, Girard JJ, Miletich JP. The lipoprotein-associated coagulation inhibitor that inhibits the FVIIa-TF complex, also inhibits factor Xa: insight into its possible mechanism of action. Blood. 1988;71:335-343.
Bajaj MS, Kuppuswamy MN, Saito H, Spitzer SD, Bajaj SP. Cultured normal human hepatocytes do not synthesize lipoprotein-associated coagulation inhibitor: evidence that endothelium is the principal site of its synthesis. Proc Natl Acad Sci U S A. 1990;87:8869-8873.
Enjyoji K, Emi M, Mukai T, Kato H. cDNA cloning and expression of rat tissue factor pathway inhibitor (TFPI). J Biochem. 1992;111:681-687.
Novotny WF, Palmier M, Wun TC, Broze GJ, Miletich JP. Purification and properties of heparin releasable lipoprotein associated inhibitor. Blood. 1991;78:394-400.
Novotny WF, Girard TJ, Miletich JP, Broze GJ. Purification and characterization of the lipoprotein-associated coagulation inhibitor from human plasma. J Biol Chem. 1989;31:18832-18837.
Kokawa T, Abumiya T, Kimura T, Harada MS, Koh H, Tsushima M, Yamamoto A, Kato H. Tissue factor pathway inhibitor activity in human plasma: measurement of lipoprotein-associated and free form in hyperlipidemia. Arterioscler Thromb Vasc Biol. 1995;15:504-510.
Abumiya T, Enjyoji K, Kokawa T, Kamikubo Y, Kato H. An anti tissue factor pathway inhibitor (TFPI) monoclonal antibody recognized the third Kunitz domain (K3) of free form of TFPI but not lipoprotein-associated forms in plasma. J Biochem. 1995;118:178-182.
Novotny WF, Brown SG, Miletich JP, Rader DJ, Broze GJ. Plasma antigen levels of the lipoprotein-associated coagulation inhibitor in patient samples. Blood. 1991;78:387-393.
Leurs PB, Oerle RV, Hamulyak K, Wolffenbuttel BHR. Tissue factor pathway inhibitor activity in patients with IDDM. Diabetes. 1995;44:80-84.
Lindahl AK, Abildgaard U, Stokke G. Release of extrinsic pathway inhibitor after heparin injection increased response in cancer patients. Thromb Res. 1990;59:651-656.
Abumiya T, Nakamura S, Takenaka A, Takenaka O, Yoshikuni Y, Miyamoto S, Kimura T, Enjyoji K, Kato H. Response of plasma tissue factor pathway inhibitor to diet-induced hypercholesterolemia in crab-eating monkeys. Arterioscler Thromb. 1994;14:483-488.
Hansen J-B, Huesby KR, Huesby N-E, Sandset PM, Hanssen T-A, Nordoy A. Effect of cholesterol lowering on intravascular potential in type II hyperlipoproteinemia. Arterioscler Thromb Vasc Biol. 1995;15:879-885.
Sandset PM, Lund H, Norseth J, Abildgaard U, Ose L. Treatment with HMG Co A reductase inhibitors in hypercholesterolemia induces changes in the components of the extrinsic coagulation system. Arterioscler Thromb. 1991;11:138-145.
Hansen J-B, Huesby N-E, Sandset PM, Svensson B, Lyngmo V, Nordoy A. Tissue-factor pathway inhibitor and lipoproteins: evidence for association with and regulation by LDL in human plasma. Arterioscler Thromb. 1994;14:223-229.