Characterization of the Association of Tissue Factor Pathway Inhibitor With Human Placenta
Objective— Tissue factor pathway inhibitor (TFPI) is an endothelial-associated inhibitor of blood coagulation. Because the mechanism for attachment of TFPI to endothelium is not clear, we investigated its association with human placenta.
Methods and Results— Western blots demonstrate that treatment with phosphatidylinositol-specific phospholipase C (PIPLC) removes more placental TFPI than either PBS or heparin, a finding confirmed by immunohistochemistry. The amounts of heparin-releasable and PIPLC-releasable TFPI activity on placental endothelium were measured in placentas from 5 individuals. PIPLC removes >10-fold more TFPI activity from the placental fragments than 10 U/mL heparin and >100-fold more than 1 U/mL heparin. Pretreatment of the placental fragments with PIPLC increases the amount of heparin-releasable TFPI by ≈3-fold. An antibody specific for the C-terminal region of TFPI recognizes PIPLC-releasable TFPI in Western blots.
Conclusions— GPI-anchored TFPI is the predominant form on placental endothelium. Heparin-releasable TFPI likely represents only a small portion of the total TFPI on endothelium that remains attached to cell-surface glycosaminoglycans after cleavage of the GPI anchor by endogenous enzymes. The predominance of GPI-anchored TFPI suggests that heparin infusion does not significantly redistribute TFPI within the vasculature. The intact C-terminus in GPI-anchored TFPI indicates it is not directly attached to a GPI anchor. Rather, it most likely associates with endothelium by binding to a GPI-anchored protein.
Tissue factor pathway inhibitor (TFPI) is a trivalent Kunitz-type serine proteinase inhibitor of tissue factor–initiated blood coagulation. Its inhibitory activity is mediated through the first and second Kunitz domains that are active site-directed inhibitors of factors VIIa and Xa, respectively.1 TFPI is present in 3 pools within the vasculature. Its concentration in circulating plasma is ≈2.5 nmol/L; however, much of this TFPI is C-terminally truncated and associated with lipoproteins. Therefore, the plasma form of TFPI has reduced anticoagulant activity and is not thought to be an important inhibitor of tissue factor–initiated blood coagulation.2 A small amount of TFPI is secreted from thrombin-stimulated platelets and may inhibit tissue factor at sites of vascular injury.3 The largest pool of TFPI is produced by the endothelial cells and remains associated with the vascular endothelium,4,5⇓ where it simultaneously inhibits factors VIIa and Xa immediately after the activation of factor X by the factor VIIa/tissue factor catalytic complex.6 The key role of TFPI as an endothelial-associated inhibitor of blood clotting is demonstrated in individuals with abetalipoproteinemia who possess greatly decreased circulating TFPI concentrations but normal amounts of heparin-releasable TFPI. Because these patients do not have increased risk for thrombosis, the endothelial-bound TFPI is probably the most important physiological reservoir of TFPI.7
Its location on the endothelial surface is somewhat unusual because, structurally, TFPI resembles a secreted protein normally found in plasma. In addition to the 3 Kunitz domains, TFPI has an acidic N-terminal region and a highly basic C-terminal region. The C-terminal region has high affinity for heparin8 but contains neither the membrane-spanning amino acids typical of integral membrane proteins nor the hydrophobic sequence typically found in proteins directly attached to a glycosyl phosphatidylinositol (GPI) anchor.9 Thus, the mechanism for the binding of TFPI to endothelium is not entirely defined.
In vivo studies indicate that the plasma concentration of TFPI increases 2- to 4-fold after the infusion of heparin.7,10,11⇓⇓ Because the increase is rapid (<10 minutes), it seems that heparin-releasable TFPI is bound to the endothelium through nonspecific interactions between the basic C-terminal region and cell-surface glycosaminoglycans. Consistent with these data, in vitro studies have demonstrated that the binding of Escherichia coli–produced recombinant TFPI to cultured endothelial cells is prevented by heparin12 and that C-terminally truncated forms of TFPI have very low affinity for the cell surface.13,14⇓ However, studies examining endogenous TFPI produced by cultured endothelial cells have demonstrated that it associates with the cell surface via a GPI anchor in a manner that is not altered by heparin,5,15,16⇓⇓ whereas recombinant TFPI expressed in mammalian cells does not bind to the cell surface.5,17⇓ Because of these different means for the attachment of TFPI to cell surfaces, fresh human placenta was used to characterize how TFPI associates with endothelial surfaces in vivo.
Porcine heparin was obtained from Elkins-Sinn Inc. Protamine sulfate was obtained from American Pharmaceutical Partners, Inc. The chromogenic substrate for factor Xa (methoxycarbonyl-d-cyclohexylglycylglycylarginine-p-nitroanilide acetate) was obtained from American Diagnostica Inc. Human placentas were obtained from the Regional Medical Center. Placentas were refrigerated after delivery and used within 12 hours. The protocol for obtaining placentas was approved by both the Memphis VA Medical Center and the University of Tennessee Institutional Review Boards. No personal information or identifiers were associated with the placentas obtained.
The anti-TFPI 2H8 monoclonal antibody against the first Kunitz domain, the anti-TFPI 2B12 monoclonal antibody against the second Kunitz domain, the rabbit polyclonal antibody against total TFPI, and the rabbit polyclonal antibody against the 12 C-terminal amino acids of TFPI were gifts of Dr George Broze, Jr (Washington University, St Louis, Mo). The secondary antibodies directed against either mouse or rabbit IgG conjugated to horseradish peroxidase were obtained from Sigma Chemical Co. Phosphatidylinositol specific phospholipase C (PIPLC) was obtained from Molecular Probes, Inc. Recombinant human tissue factor was obtained from Ortho Diagnostic Systems Inc. Recombinant human ala-TFPI was a gift from the Chiron (Emeryville, Calif) and Searle (Skokie, Ill) corporations. Human factors VIIa and X were from Hematologic Technologies.
Cellular Membrane Preparations From Placenta
Ten grams of finely minced placenta was homogenized on ice for 3 minutes using a VirTishear tissue homogenizer (Virtis Co) in a total volume of 20 mL PBS containing 10 mmol/L EDTA, 1 mmol/L phenylmethyl sulfonyl fluoride, 50 μmol/L E-64, 0.005 mg/mL soy bean trypsin inhibitor, 0.0175 mg/mL benzamidine, and 0.005 mg/mL leupeptin. After a low-speed centrifugation for 20 minutes to remove large debris, 20 mL of 5 mmol/L phosphate buffer pH 7.4 was added to the supernatant to hypotonically lyse any remaining intact cells. After a 20-minute incubation in the hypotonic buffer, the sample was homogenized on ice for an additional minute and placental membranes collected by ultracentrifugation for 1 hour at 30 000 rpm in a SW40Ti rotor. The membranes were washed once and resuspended in 600 μL PBS. The membrane suspension was separated into 150-μL aliquots and incubated with PBS only, 1 U/mL heparin in PBS, or 1 U/mL PIPLC in PBS for 1 hour at 23°C. The membranes were removed by centrifugation and the supernatants were subjected to SDS-PAGE and Western blot analysis for TFPI, as described below.
SDS Polyacrylamide Gel Electrophoresis
Proteins were analyzed using continuous 5% to 15% linear acrylamide gradient gels in an ammediol/glycine buffer system. Before electrophoresis, samples were boiled in sample buffer containing 1% SDS for 3 minutes under nonreducing conditions.
Western Blot Analysis
After separation by SDS-PAGE, proteins were transferred to nitrocellulose and immunostained for TFPI with the indicated primary antibodies and then with anti-rabbit IgG conjugated to horseradish peroxidase. Proteins were visualized using SuperSignal Western blotting reagents (Pierce Co).
After treatment with PBS, 1 U/mL heparin, or 1U/mL PIPLC for 2 hours at 23°C, minced placental samples were fixed in 10% formalin. Four micron sections of the formalin-fixed and paraffin-embedded tissue were mounted on coated slides, air dried, deparaffinized, and hydrated. The sections were immunostained with the 2B12 anti-TFPI monoclonal antibody. The antibody was incubated at a dilution of 1:400 for 30 minutes at room temperature. After washing, secondary biotinylated rabbit anti-mouse IgG antibody was applied, followed by streptavidin-peroxidase conjugate. 3,3′-diaminobenzidine substrate was used as a chromogen. The sections were then counterstained with hematoxylin, dehydrated, and placed under cover glass.
Preparation and Treatment of Minced Placental Samples
Samples from freshly obtained placenta were rinsed with PBS and finely minced with a razor blade. The minced placenta was separated into 0.5-g samples and washed 7 times through repeated vortexing and centrifugation in ice-cold PBS containing 10 mmol/L EDTA (PBS/EDTA) to remove residual blood, which contains TFPI. The samples were then treated with 0.75 mL of PBS/EDTA to control for any residual plasma TFPI associated with the placental fragments or PBS/EDTA containing 1 U/mL heparin, 10 U/mL heparin, or 1 U/mL PIPLC for 2 hours at 23°C. In some experiments, samples were sequentially treated with PIPLC and 10 U/mL heparin or vice versa for 1 hour each at 23°C. The samples were washed one time with PBS/EDTA to remove residual PIPLC or heparin before the second reagent was added. After incubation with the various reagents, the samples were centrifuged to remove the placental fragments and the supernatants were assayed for TFPI activity as described below.
TFPI Activity Assays
Functional TFPI activity in the placental washes was determined using recombinant tissue factor, prepared according to the manufacturer’s instructions, and diluted 1:200 to make a working stock; 0.2 nmol/L human factor VIIa and 20 nmol/L human factor X and a sample of the various placenta washes were mixed in the presence of 50 mmol/L HEPES, 100 mmol/L NaCl, 10 mmol/L CaCl2, and 0.1% BSA (pH 7.4) and allowed to generate factor Xa for 30 minutes at 23°C. The reaction was quenched by addition of EDTA to 20 mmol/L final concentration. Factor Xa generated was determined by monitoring the cleavage of 500 μmol/L factor Xa substrate. The amount of TFPI in the placental samples was determined by comparison to a standard curve generated using known amounts of TFPI. The effects of heparin on the assay were reversed by the addition of 5 μg/mL protamine sulfate to all reactions. To validate that factor Xa is generated linearly over the 30-minute incubation period, the assay was performed with TFPI at concentrations spanning the standard curve. Samples were tested for factor Xa activity after 10, 20, and 30 minutes of incubation, confirming that the assay is linear for this time period. To control for interassay variability, 2 PIPLC-treated placenta supernatants were aliquoted and stored at −20°C. With each assay, a fresh aliquot of each was thawed and tested. Assays were repeated if the values for these 2 control samples were not within 2 standard deviations of their respective mean values. In some assays, TFPI activity was inhibited by the addition of either the 2H8 monoclonal or a polyclonal anti-TFPI antibody. Statistical analysis to determine probability values was performed using a single tailed t test.
Western Blot Analysis of TFPI Released From Cellular Membranes Prepared From Placental Tissue
Western blot analysis of placental TFPI was used as an initial, qualitative technique to characterize the attachment of TFPI to placental endothelium. To obtain sufficient amounts of TFPI to visualize on Western blots, cellular membranes were prepared from placenta and treated with PBS/EDTA, 1 U/mL heparin, or 1 U/mL PIPLC, an enzyme that specifically removes GPI-anchored proteins from the cell surface, as described in Methods. In blots treated with a polyclonal anti-TFPI antibody, progressively darker bands of identical molecular weight (≈41 kDa) are observed in the PBS/EDTA-, heparin-, and PIPLC-treated samples (Figure 1), demonstrating that GPI-anchored TFPI is the predominant form present in placenta.
Immumohistochemistry of Placental Fragments
After treatment with PBS/EDTA, 1 U/mL heparin, or 1 U/mL PIPLC for 2 hours at 23°C, samples of minced placenta were placed in 10% formalin and submitted for immunohistochemical staining with the 2B12 anti-TFPI monoclonal antibody (Figure 2). In PBS/EDTA- and heparin-treated samples, intense stain is seen in the endothelial cells of fetal capillaries and small arteries and veins. In the PIPLC-treated samples, the staining is almost completely removed from the endothelial cells of small blood vessels and capillaries, again demonstrating that GPI-anchored TFPI is the predominant form present in placenta. Although examination of the fetal capillaries best demonstrates the effects of PIPLC on endothelial TFPI, PIPLC-releasable TFPI is also present on syncytiotrophoblasts, cytotrophoblasts, and extravillius trophoblasts as well as some of the stromal cells of the villi.18
Characterization of the Attachment of TFPI to Placenta Using TFPI Activity Assays
Because the Western blot and immunohistochemical assays are difficult to accurately quantify, TFPI activity released from placenta was measured using factor VIIa/tissue factor inhibition assays. Incubation of minced placenta in different solutions is a simple means to examine the association of TFPI with the placental surface that uses fewer reagents than perfusion techniques. Additionally, because only 0.5 g of placenta is needed for each experiment, multiple different experiments can be performed in triplicate to fully characterize the TFPI in each individual placenta. Preliminary activity assays indicated that PIPLC removes more TFPI from the minced placenta than either PBS or heparin, as expected based on the results from the Western blot and immunohistochemistry. To confirm that TFPI, and not another placental proteinase inhibitor, such as TFPI-2,19 is producing the inhibitory activity observed, assays were performed using the PIPLC supernatant in the presence of the 2H8 monoclonal anti-TFPI antibody or an anti-TFPI polyclonal antibody. Both antibodies completely reverse the inhibitory activity observed to the level of the PBS-treated sample, indicating that the assay is specifically measuring TFPI released from the placental surface (Figure 3).
To quantify the amounts of heparin-releasable and PIPLC-releasable TFPI on the placental surface, 0.5-g samples of minced placenta were washed with PBS/EDTA to remove residual blood and incubated for 2 hours at 23°C with PBS/EDTA alone or containing 1 U/mL heparin, 10 U/mL heparin, or 1 U/mL PIPLC. The experiments were performed in triplicate and averaged to determine a final value for the amount of TFPI activity released from the placental fragments during treatment with the different reagents. This process was repeated for placentas obtained from 5 individuals. The amount of TFPI released from the washed placental fragments by 1 U/mL heparin, 10 U/mL heparin, or 1 U/mL PIPLC, after subtraction of the background TFPI released by PBS/EDTA alone (to control for residual plasma TFPI not removed in the washing steps), is presented in Table 1 along with the average values and standard deviation for data obtained from all 5 placentas. The data demonstrate that there is >10-fold more GPI-anchored TFPI than that released by 10 U/mL heparin and >100-fold more than that released by 1 U/mL heparin. Comparison of the amounts of TFPI removed from the individual placentas indicates that the amount of heparin-releasable TFPI varies up to 8.1-fold, whereas the amount of TFPI attached through a GPI-anchor varies up to 1.6-fold.
Effects of Sequential Incubation with PIPLC and Heparin or Vice Versa on the Release of TFPI From the Placental Surface
Samples, in triplicate, from each of the 5 placentas were sequentially incubated for 1 hour at 23°C with 10 U/mL heparin and 1 U/mL PIPLC or 1 U/mL PIPLC and 10 U/mL heparin (Table 2). Pretreatment with heparin has no effect on the amount of TFPI released by subsequent incubation with PIPLC, but pretreatment with PIPLC significantly increases the amount of TFPI released by subsequent incubation with heparin (P<0.005). Transfer of the PIPLC-treated placental fragments to a new tube before treatment with heparin demonstrated that the heparin-releasable TFPI was bound to the placental fragments and not the microcentrifuge tube.
GPI-Anchored Placental TFPI Contains an Intact C-Terminus
GPI-anchored proteins have an N-terminal leader sequence that directs the polypeptide to the endoplasmic reticulum. Within the endoplasmic reticulum, attachment of the GPI anchor to the C-terminus of the protein occurs through a transamidation reaction that displaces the C-terminal 17 to 30 amino acids of the protein. A common feature among the C-terminal peptides cleaved from proteins during GPI anchor attachment is a region of 15 to 20 hydrophobic residues.9 In contrast to other GPI-anchored proteins, TFPI has a very basic C-terminal region with no long stretches of hydrophobic amino acids, suggesting that it is not directly attached to a GPI anchor, or if it is, attachment occurs through a novel mechanism. To additionally investigate how TFPI is bound to the placental surface via a GPI anchor, Western blots were performed on samples prepared by incubation of cellular membranes with PBS/EDTA, 1 U/mL heparin, or 1 U/mL PIPLC using an antibody specific for the C-terminus of TFPI. As with the polyclonal antibody, this antibody detects progressively darker bands of identical molecular weight in the PBS, heparin, and PIPLC samples (Figure 4). These data indicate that the C-terminus is intact in both the heparin-releasable and the GPI-anchored forms of TFPI and demonstrate that TFPI is not directly attached to a GPI anchor. Additional bands of heparin and PIPLC-releasable TFPI are detected by C-terminal antibody in Figure 4 that are not present in Figure 1. The C-terminal antibody has been previously characterized20 and is specific for the C-terminal region of TFPI. Therefore, the additional bands represent variably glycosylated forms of TFPI rather than C-terminally truncated TFPI. The variability in the ability of these antibodies to detect the less prevalent forms of TFPI is a function of the affinity of the respective antibodies for TFPI and the length of time the film is exposed to the blot before development.
Endothelial-associated TFPI functions within the vasculature primarily to abrogate factor VIIa/tissue factor activity transiently present on endothelial cells or monocytes that have been stimulated with inflammatory cytokines. This activity is most clearly demonstrated in mice in which deletion of the first Kunitz domain (TFPI-K1) leads to death in utero from a consumptive coagulopathy.21 Additionally, when the TFPI-K1 deletion is bred into Apo-E null mice, the TFPI heterozygous animals have an increased atherosclerotic burden and increased plaque tissue factor activity compared with mice with normal amounts of TFPI, suggesting that TFPI also is an important regulator of thrombosis in diseased vessels.22 Because cultured endothelial cells can rapidly change phenotype on growth in tissue culture,23 we used minced, human placenta as a source of fresh, noncultured endothelial cells to characterize how TFPI is attached to the endothelium in vivo. The use of minced placenta, as opposed to perfusion of either the entire placenta or the cord vessels alone, is technically simple and allows for numerous experiments to be performed with the same placenta.
Comparison of the amount of TFPI activity released from minced placenta demonstrates that there is 10- to 100-fold more GPI-anchored TFPI than heparin-releasable TFPI on PBS/EDTA-washed placental fragments. Consistent with the TFPI activity assays, immunohistochemical analyses of the placental fragments demonstrate that almost all of the TFPI is removed from the placental vasculature with PIPLC. Heparin has very little effect on the amount or location of TFPI antigen detected by immunohistochemistry. The interindividual variation of the TFPI associated with the placental fragments from the 5 individuals studied was much less for GPI-anchored TFPI (1.6-fold) than for the heparin-releasable TFPI (8.1-fold). Although additional experiments are necessary to characterize the association of TFPI with the endothelium of other vascular beds, these data using placenta suggest that a GPI anchor is the predominant mechanism for the association of TFPI with endothelium and that heparin infusion does not induce a significant redistribution of the total TFPI within the vasculature. This conclusion is consistent with the recent findings of Adams et al,24 who measured the circulating TFPI concentration in 96 patients receiving heparin for cardiopulmonary bypass. Two patients with low amounts of heparin-releasable TFPI were identified, but neither experienced an adverse clinical outcome.
Sequential treatment of minced placenta with PIPLC followed by heparin revealed a heparin-releasable pool of TFPI that is not present before treatment of the placental fragments with PIPLC. These data suggest that heparin-releasable TFPI represents a fraction of the GPI-anchored TFPI that has had its GPI anchor cleaved but remains nonspecifically associated with cell-surface glycosaminoglycans. Because heparin-releasable TFPI has not been detected on cultured endothelial cells5,25⇓ and was released in only very small amounts from the washed placental fragments, it is likely that the heparin-releasable pool of TFPI readily dissociates from the cell surface and is removed during preliminary washing steps in in vitro studies. Because of the presence of plasma TFPI in the placental blood, it is not possible to quantify the amount of TFPI removed from the placental fragments during washing steps in the experiments presented here. However, after PIPLC treatment of the placental fragments, only a small amount (<20%) of the TFPI seems to remain nonspecifically associated with GAGs, as determined by the TFPI activity assays of samples treated sequentially with PIPLC and heparin. Because it is unlikely that large amounts of TFPI are separated from the GPI anchor by endogenous enzymes, the percentage of the total endothelial-associated TFPI that is heparin-releasable in vivo is probably much less than that observed in these in vitro experiments, where nearly all of the GPI-anchored TFPI is cleaved after treatment with PIPLC.
The biological mechanism for the association of TFPI with a GPI anchor is somewhat unclear. Lupu et al16 detected incorporation of [3H] ethanolamine, a component of the GPI-anchor, into TFPI removed from the cell surface and concluded that TFPI is directly attached to a GPI anchor. However, the C-terminal region of TFPI is very basic and does not contain the hydrophobic sequence of amino acids that is removed from GPI-anchored proteins on anchor attachment. Furthermore, the C-terminal TFPI Western blot analysis of TFPI released by PIPLC from cellular membranes prepared from placenta demonstrates that the C-terminal region of GPI-anchored TFPI is intact. Thus, if TFPI is directly attached to a GPI anchor, its attachment occurs through a novel mechanism. We hypothesize that TFPI tightly associates with a GPI-anchored binding protein on the surface of endothelial cells. Candidate GPI-anchored TFPI binding proteins include glypican-3,26 which we have shown binds to TFPI, and glypican-1,27 which has been shown to bind to another Kunitz-type inhibitor, the amyloid precursor protein inhibitor. One consequence of the attachment of TFPI to endothelium through a GPI-anchored binding protein is its localization into caveolae/lipid raft microdomains on the cell surface.15,16⇓ However, the physiological advantages for localization of TFPI to these domains and the role of the GPI-anchored binding protein in TFPI anticoagulant activity remain to be determined.
This study was supported by grants from the Department of Veteran’s Affairs (to Dr Mast) and the American Heart Association (to Drs Hall and Dietzen).
Received September 12, 2002; revision accepted October 2, 2002.
- ↵Novotny WF, Girard TJ, Miletich JP, Broze GJ Jr. Platelets secrete a coagulation inhibitor functionally and antigenically similar to the lipoprotein associated coagulation inhibitor. Blood. 1988; 72: 2020–2025.
- ↵Ameri A, Kuppuswamy MN, Basu S, Bajaj SP. Expression of tissue factor pathway inhibitor by cultured endothelial cells in response to inflammatory mediators. Blood. 1992; 79: 3219–3226.
- ↵Ott I, Miyagi Y, Miyazaki K, Heeb MJ, Mueller BM, Rao LV, Ruf W. Reversible regulation of tissue factor-induced coagulation by glycosyl phosphatidylinositol-anchored tissue factor pathway inhibitor. Arterioscler Thromb Vasc Biol. 2000; 20: 874–882.
- ↵Baugh RJ, Broze GJ Jr, Krishnaswamy S. Regulation of extrinsic pathway factor Xa formation by tissue factor pathway inhibitor. J Biol Chem. 1998; 273: 4378–4386.
- ↵Novotny WF, Brown SG, Miletich JP, Rader DJ, Broze GJ Jr. Plasma antigen levels of the lipoprotein-associated coagulation inhibitor in patient samples. Blood. 1991; 78: 387–393.
- ↵Hansen JB, Olsen R, Webster P. Association of tissue factor pathway inhibitor with human umbilical vein endothelial cells. Blood. 1997; 90: 3568–3578.
- ↵Sevinsky JR, Rao LV, Ruf W. Ligand-induced protease receptor translocation into caveolae: a mechanism for regulating cell surface proteolysis of the tissue factor-dependent coagulation pathway. J Cell Biol. 1996; 133: 293–304.
- ↵Lupu C, Goodwin CA, Westmuckett AD, Emeis JJ, Scully MF, Kakkar VV, Lupu F. Tissue factor pathway inhibitor in endothelial cells colocalizes with glycolipid microdomains/caveolae: regulatory mechanism(s) of the anticoagulant properties of the endothelium. Arterioscler Thromb Vasc Biol. 1997; 17: 2964–2974.
- ↵Ho G, Narita M, Broze GJ Jr, Schwartz AL. Recombinant full-length tissue factor pathway inhibitor fails to bind to the cell surface: implications for catabolism in vitro and in vivo. Blood. 2000; 95: 1973–1978.
- ↵Sprecher CA, Kisiel W, Mathewes S, Foster DC. Molecular cloning, expression, and partial characterization of a second human tissue-factor-pathway inhibitor. Proc Natl Acad Sci U S A. 1994; 91: 3353–3357.
- ↵Mast AE, Stadanlick JE, Lockett JM, Dietzen DJ, Hasty KA, Hall CL. Tissue factor pathway inhibitor binds to platelet thrombospondin-1. J Biol Chem. 2000; 275: 31715–31721.
- ↵Huang ZF, Higuchi D, Lasky N, Broze GJ Jr. Tissue factor pathway inhibitor gene disruption produces intrauterine lethality in mice. Blood. 1997; 90: 944–951.
- ↵Westrick RJ, Bodary PF, Xu Z, Shen YC, Broze GJ, Eitzman DT. Deficiency of tissue factor pathway inhibitor promotes atherosclerosis and thrombosis in mice. Circulation. 2001; 103: 3044–3046.
- ↵Cines DB, Pollak ES, Buck CA, Loscalzo J, Zimmerman GA, McEver RP, Pober JS, Wick TM, Konkle BA, Schwartz BS, Barnathan ES, McCrae KR, Hug BA, Schmidt AM, Stern DM. Endothelial cells in physiology and in the pathophysiology of vascular disorders. Blood. 1998; 91: 3527–3561.
- ↵Williamson TG, Mok SS, Henry A, Cappai R, Lander AD, Nurcombe V, Beyreuther K, Masters CL, Small DH. Secreted glypican binds to the amyloid precursor protein of Alzheimer’s disease (APP) and inhibits APP-induced neurite outgrowth. J Biol Chem. 1996; 271: 31215–31221.