Arterial Flow Conditions Downregulate Thrombomodulin on Saphenous Vein Endothelium
Background—The antithrombogenic properties of venous endothelium may be attenuated when vein is implanted in the arterial circulation. Such changes may facilitate thrombosis, which is the final common pathway for saphenous vein arterial bypass graft occlusion.
Methods and Results—Using human saphenous vein in a validated ex vivo flow circuit, we investigated (1) the possibility that arterial flow conditions (mean pressure, 100 mm Hg, 90 cpm, ≈200 mL/min) alter the concentration of proteins involved in regulating thrombosis at the vessel wall and (2) the influence of ion channel blockade on such effects. Concentrations of thrombomodulin and tissue factor were quantified by Western blotting (ratio of von Willebrand factor staining) and immunohistochemistry (as a percentage of CD31-staining area). Thrombomodulin concentrations after 90 minutes of venous and arterial flow conditions were quantified by immunostaining (68.9±4.8% and 41.0±3.0% CD31, respectively; P<0.01) and by Western blotting (1.35±0.20 and 0.15±0.03 ratio of von Willebrand factor, respectively; P<0.01). The ability of endothelial cells to generate activated protein C also decreased from 62±14 to 19±10 ng · min−1 · 1000 cells−1 (P=0.01). The significant reduction in thrombomodulin was attenuated if calcium was removed from the perfusate but not by external vein stenting. Inclusion in the vein perfusate of drugs that reduce calcium entry (including Gd3+, to block stretch-activated ion channels, and nifedipine) abolished the reduction in thrombomodulin concentration observed after arterial flow conditions. In freshly excised vein, negligible concentrations of tissue factor were detected on the endothelium and concentrations did not increase after 90 minutes of arterial flow conditions, although the inclusion of nifedipine caused the immunostaining to increase from 3.0±0.4% to 8.5±0.7% CD31 (P<0.02).
Conclusions—In saphenous vein endothelium exposed to arterial flow conditions, there is rapid downregulation of thrombomodulin, sufficient to limit protein C activation, by a calcium-dependent mechanism.
In both the aortocoronary and infrainguinal circulations, thrombosis is the final common pathway leading to saphenous vein bypass graft occlusion. Normally, the endothelium provides an antithrombotic surface, with the expression of thrombomodulin and tissue plasminogen activator (tPA) on the endothelium contributing to this property. The preparation of vein for grafting causes considerable endothelial cell injury.1 In addition, the new hemodynamic forces placed on the endothelium, in vein grafts, have complex effects on gene expression.2 These hemodynamic forces may alter the expression of thrombomodulin, tPA, and tissue factor.3 4 5 6 In cultured bovine aortic endothelial cells, thrombomodulin protein concentration decreased 3- to 5-fold after 36 hours of exposure to high shear stress, with mRNA levels decreasing as early as 2 to 4 hours after the application of shear stress.6 In cultured human umbilical endothelial cells, both fluid flow and shear stress have been reported to stimulate mRNA levels and protein secretion of tPA.4 5 Similarly, although there is scant evidence for the expression of tissue factor in normal endothelium, in cultured cells the application of shear stress causes a transient increase in tissue factor procoagulant activity.6 In saphenous vein grafted into the arterial circulation, such changes could disturb the antithrombotic properties of endothelium in a manner that may predispose to bypass occlusion.
To facilitate investigation of the early adaptive responses of saphenous vein endothelium to arterial hemodynamics, we developed and validated an ex vivo arterial bypass circuit.7 External stenting of vein in vitro permits discrimination of effects attributable to increased flow and shear stress from effects attributable to circumferential deformation.7 Using this ex vivo circuit, we have been able to show rapid changes in the expression of endothelial proteins (eg, intercellular adhesion molecule–1, vascular cell adhesion molecule–1, and nitric oxide synthase) that participate in leukocyte adhesion.7 Here, we report on the changes in endothelial thrombomodulin and tissue factor expression, extending these studies to indicate the importance of ion channels in mechanotransduction events, when saphenous vein endothelium is subject to arterial flow conditions.
Saphenous vein was harvested from patients undergoing aortocoronary or infrainguinal bypass, amputation, or high ligation of saphenous vein for correction of varicose veins, as approved by the local ethical committee. The vein segment was sected to provide 2 long lengths and a vein ring from the center. The vein ring was suspended in an organ chamber to confirm the presence of functional endothelium and smooth muscle, as described previously.8 Diseased vein that did not respond to phenylephrine (10 μmol/L) with a contraction of >2g and then relax in response to A23187 (1 μmol/L) was discarded. Samples from recent smokers and patients with diabetes were excluded.
Modified Krebs’ solution (118.4 mmol/L NaCl, 4.7 mmol/L KCl, 1.2 mmol/L KH2PO4, 1.2 mmol/L MgSO4 · 0.7H2O, 11.1 mmol/L glucose, 24.9 mmol/L NaHCO3, 2.5 mmol/L CaCl2) was made fresh each day. Antibodies for immunohistochemistry and Western blotting were obtained as follows: platelet endothelial cell adhesion molecule–1 (CD31, monoclonal; R&D Systems), P-selectin and tPA (monoclonals; Serotec), tissue factor (monoclonal was the kind gift from Dr John McVey, Clinical Sciences Center, Royal Postgraduate Medical School, London, UK W12; polyclonal was the kind gift from Prof Y. Nemerson, Mount Sinai School of Medicine, New York), thrombomodulin (monoclonals 24FM and 3EZ were the kind gifts from Dr J. Amiral, Serbio Research Laboratory, Gennevilliers, France), and polyclonal antibodies to von Willebrand factor (vWF; DAKO). Antibodies for the characterization of isolated endothelial cells included vWF and CD31 (as above), anti–smooth muscle cell actin, and CD45 (DAKO). Enhanced chemiluminescence (ECL) Western blotting reagents were obtained from Amersham UK. Unless specified, other reagents were obtained from Sigma.
Saphenous vein (3 to 4 cm) was mounted in a retaining jig, placed in an in vitro flow circuit, and perfused with oxygenated Krebs’ solution, and the internal diameter was monitored as described previously.9 Veins were exposed either to pulsatile flow (90 cpm) at a mean pressure of 100 mm Hg (arterial flow) for 45 to 90 minutes (flow rate, 200 to 225 mL/min; shear stress, 0.26±0.09 N/m2) or to nonpulsatile flow at 20 mm Hg (venous flow) for 90 minutes (flow rate, 10 to 20 mL/min; shear stress, 0.021±0.011 N/m2). Some veins were placed inside a tube (2 to 4 cm in length) of externally supported polytetrafluorethylene, which was nonrestrictive but sized to limit circumferential distention of the vein during arterial flow (shear stress, 0.40±0.13 N/m2). In some experiments, the Krebs’ solution perfusing the vein was supplemented with ion channel blockers. After 45 or 90 minutes, the vein was removed, the ends were discarded, and the remainder was divided for histology or immunohistochemistry or used for harvesting of endothelial cells. The Krebs’ solution perfusing the vein was sampled at 15-minute intervals and later concentrated 100-fold through an Amicon filter with a 3000-Da exclusion limit, and the concentrates were stored at –70°C.
Vein specimens for immunohistochemical analysis were fixed in Zamboni’s solution and prepared, and serial cryostat sections were stained through use of the ABC immunoperoxidase method7 10 with the use of monoclonal antibodies at the following dilutions: to endothelial nitric oxide synthase (1:10 000), to P-selectin (1:1000), to tPA (≤1:10), to thrombomodulin 24FM (1:5000), and to CD31 (1:1000): Polyclonal antibodies to tissue factor were used at 1:25 dilution. The area of immunostaining was computed from serial sections as described previously.7 To allow for the small endothelium loss noted from CD31 immunostaining and the dilation that occurs in response to arterial flow, the staining areas for P-selectin, thrombomodulin, and tissue factor were expressed as a percentage of CD31 staining.
Endothelial cells were harvested from freshly excised veins and veins exposed to the different flow conditions as described previously.7 The yield of endothelial cells, estimated with the use of a hemocytometer, ranged between 5000 and 10 000 cells per vein and contained >95% endothelial cells, <2% leukocytes, and <2% smooth muscle cells. Cells were collected, as a pellet, and dispersed in lysis buffer (10 mmol/L Tris, pH 8, 1 mmol/L EDTA, 2.5% SDS, 5% mercaptoethanol) before SDS-PAGE (Phast System 8% to 25% gradient acrylamide gel) as described previously.7 The primary antibodies used were thrombomodulin at 1:200 dilution, monoclonal to tissue factor at 1:500 dilution, tPA at ≤1:25 dilution, CD31 at 1:1000 dilution, and vWF at 1:1000 dilution. Proteins were visualized with ECL and quantified with densitometric scanning. If possible, changes in thrombomodulin and tissue factor staining were standardized with respect to vWF staining. To minimize inconsistencies, paired samples always were processed together, there was simultaneous development of thrombomodulin or tissue factor and vWF, and only band densities within the linear ECL signal range were assessed. For experiments using A23187 or veins perfused with calcium-free Krebs’ solution, the staining of CD31 and thrombomodulin or tissue factor was performed on separate gels because of molecular weight similarities.
Activation of Protein C to Assess Thrombomodulin Activity
Endothelial cells were harvested through the use of collagenase digestion,7 resuspended, and washed 3 times in 50 mmol/L Tris buffer, pH 8.0, containing 2 mmol/L CaCl2, 0.1 mol/L NaCl, and 0.1% BSA. The cell suspension was incubated with 1.0 μg protein C and 0.1 μg thrombin (both from Enzyme Research Laboratories), and the activation of protein C was measured as described previously.11 The assay was calibrated using activated protein C (Enzyme Research Laboratories). The endothelial cells used in these assays were quantified using monoclonal antibodies to CD31. Briefly, cells were incubated at 4°C with antibody CD31 (diluted 1:80 in PBS containing 5% fetal calf serum, 1 μg/mL pepstatin,1 μg/mL leupeptin, and 0.1 mmol/L phenylmethylsulfonyl fluoride). Cells were harvested by centrifugation, washed 3 times in the same PBS-based buffer as above, sequentially incubated, and washed with biotinylated second antibody and a streptavidin-peroxidase conjugate before the development of the peroxidase activity with a chromogenic substrate. Cultured human saphenous vein endothelial cells (1000 to 25 000) were used to calibrate the assay, and cultured human aortic smooth muscle cells (25 000) were used as a negative control. The results are given as generation of activated protein C (ng · min−1 · 1000 cells−1).
Immunostaining areas (mean±SEM) and staining of Western blots were compared using Student’s t test for paired comparisons.
Expression of Thrombomodulin on Saphenous Vein Endothelium: Influence of Pulsatile Arterial Flow Conditions on Unstented and Stented Vein
The abundant immunostaining for thrombomodulin on the endothelium of freshly excised veins was unchanged after exposure to venous flow circuits for 90 minutes (Table 1⇓). After exposure to arterial flow conditions for 45 or 90 minutes, there was decreased thrombomodulin staining (Figure 1⇓), with a significant 2-fold decrease in the ratio of thrombomodulin to CD31 staining area (Table 1⇓). These changes were accentuated in Western blotting experiments (Figure 2a⇓). The staining ratio of thrombomodulin to vWF was similar in lysates of cells from freshly excised vein (1.5±0.2) and cell lysates from vein subjected to venous flow conditions (1.35±0.2) but decreased markedly after the vein had been exposed to arterial flow conditions for 90 minutes (to 0.15±0.03, P<0.01, 6 paired samples, compared with venous flow conditions). A reduction in thrombomodulin/vWF staining ratio also was evident after 45 minutes of arterial flow conditions (staining ratio, 0.60±0.1, P=0.02, 4 paired samples, compared with freshly excised vein). The temporal changes in thrombomodulin are summarized in Figure 3⇓.
There was no evidence of thrombomodulin in the concentrated vein perfusate after 90 minutes of arterial flow conditions by either Western blotting or dot-blotting (the minimum concentration that could be detected was 2 pg or 100 pg/mL), mitigating against surface shedding of thrombomodulin. External stenting of vein with polytetrafluorethylene to limit circumferential and radial deformation did not alter the reduction in thrombomodulin concentration observed after exposure to arterial flow conditions for 90 minutes. The immunostaining area for thrombomodulin in stented vein exposed to arterial flow conditions was 41.0±3.0 compared with 68.9±4.8 for stented vein subjected to venous flow conditions (P<0.01, 5 paired samples).
Activation of Protein C by Endothelial Cells Harvested After Venous and Arterial Flow Conditions
After 90 minutes of arterial flow conditions, the ability of isolated endothelial cells to activate protein C had decreased 3-fold compared with cells isolated after 90 minutes of venous flow conditions (from 62±6 to 19±4 ng · min−1 · 1000 cells−1, respectively, 6 paired samples, P=0.01; Figure 3⇑).
Expression of tPA and Tissue Factor in Saphenous Vein Endothelium: Influence of Pulsatile Arterial Flow
In freshly excised vein, there was scant endothelial staining for either tissue factor (Figure 4a⇓) or tPA, even at primary antibody concentrations of ≤1:10. There was no evidence of altered endothelial staining after exposure to arterial flow conditions for 45 or 90 minutes or to venous flow conditions for 90 minutes. In veins in which smooth muscle cells were present in the intima, staining for tissue factor in the intimal muscle cells was observed after exposure of veins to arterial flow conditions for 90 minutes (Figure 4b⇓). Tissue factor was more readily detected in endothelial cell lysates. There was no change in the tissue factor/vWF ratio, as observed with Western blotting, after 90 minutes of arterial or venous flow conditions (6 paired samples): the results are shown in Table 2⇓. Western blotting failed to detect tPA in endothelial cell lysates. Because tPA is secreted rapidly after synthesis, the presence of tPA in the vein perfusate was investigated with both Western blotting and dot blotting, but no tPA was detected. As reported previously, arterial flow conditions for 90 minutes had no effect on the endothelial concentration of P-selectin7 : baseline results for immunostaining area are given in Table 1⇑.
Ion Channel Blockade: Limitation of Flow Responses
First, it was necessary to establish whether the expression of the selected endothelial proteins was altered by ion channel blockers alone. In the absence of flow, the incubation of vein for 90 minutes in oxygenated Krebs’ solution supplemented with 3 mmol/L tetraethylammonium chloride (TEA) (to block K+ channels), 30 μmol/L glibenclamide (to block KATP channels), 10 μmol/L Gd3+ (to block stretch-activated cation channels), or 20 μmol/L nifedipine (to block voltage-gated calcium channels) did not alter the immunostaining area of any of the selected endothelial proteins (data not shown). For veins perfused under venous flow conditions for 90 minutes with Krebs’ with or without TEA, there were no changes in the immunostaining area of thrombomodulin or tissue factor (Table 1⇑). For the other ion channel blockers, there usually was insufficient length of vein to conduct venous flow condition controls, and only arterial flow condition experiments are reported in Tables 1⇑ and 2⇑, using vein incubated in Krebs’ supplemented with the appropriate channel blocker as the control. Inclusion of ion channel blockers in the vein perfusate did not alter the immunostaining area of P-selectin after 90 minutes of arterial flow conditions. The reduction in thrombomodulin concentration, after arterial flow conditions, was abolished when either Gd3+ (10 μmol/L) or nifedipine (20 μmol/L) was included in the vein perfusate: the results for immunostaining are shown in Table 1⇑, and the results for Western blotting are shown in Table 2⇑. Inclusion of TEA in the vein perfusate did not abolish the arterial flow–induced decreases in immunostaining area or the thrombomodulin concentration observed on Western blotting (Tables 1⇑ and 2⇑). Inclusion of nifedipine in the vein perfusate resulted in a significant increase in the endothelial concentration of tissue factor, with clear evidence of endothelial staining (Figure 4c⇑ and Table 1⇑). These results were confirmed by Western blotting. The tissue factor/vWF ratio in endothelial cell lysates increased from 0.25±0.06 to 0.35±0.10 (5 paired samples, P=0.04) from control vein and vein exposed to arterial flow conditions for 90 minutes, respectively (Figure 2b⇑ and Table 2⇑). The inclusion of glibenclamide in vein perfusate did not abrogate the changes in immunostaining area, observed after 90 minutes of arterial flow conditions, for any of the endothelial proteins assessed (Table 1⇑). The marked endothelial cell loss (>50%) after perfusion of vein, under arterial flow conditions for 90 minutes, with calcium-free solutions precluded the assessment of immunostaining areas under these conditions. However, remaining cells could be isolated for Western blotting and functional assays. The Western blotting experiments showed no reduction in thrombomodulin (n=3). The ability of cells, harvested after arterial or venous perfusion in the absence of calcium, to generate activated protein C was 84±14 and 60±8 ng · min−1 · 1000 cells−1, respectively (n=3).
Augmentation of Endothelial Cell Calcium Concentrations With A23187
Perfusion under venous conditions or incubation of vein rings with 0.5 μmol/L A23187 for 90 minutes effected a 2-fold reduction in immunostaining area for thrombomodulin. The staining area for thrombomodulin decreased from 73.5±6.1 in the absence of A23187 to 44.3±5.3 in the presence of 0.5 μmol/L A23187 (P<0.01, 5 paired samples, venous flow conditions for 90 minutes). Because A23187 stimulates the secretion of vWF, albeit at concentrations >1 μmol/L,12 Western blotting for CD31 was performed. Venous perfusion conditions with A23187 decreased the staining ratio (thrombomodulin/CD31) from 1.32±0.3 to 0.33±0.08 (4 paired samples, P<0.03). There also was a 2-fold reduction in the thrombomodulin/vWF staining ratio. After experiments in the presence and absence of A23187, the ability of isolated endothelial cells to generate activated protein C was 15±4 and 49±9 ng · min−1 · 1000 cells−1, respectively (n=3).
The prevention of thrombosis in saphenous vein bypass grafts is an important therapeutic goal that has provided the impetus for randomized trials of both antiplatelet and anticoagulant drugs after bypass surgery.13 14 15 16 The disappointing results of some of these trials, with respect to graft patency, emphasize the gaps in our understanding of how the antithrombotic surface of saphenous vein endothelium is altered when vein is implanted in the arterial circulation. Here, we show that the concentration of thrombomodulin on saphenous vein endothelium is downregulated rapidly in response to arterial flow conditions, whereas the changes in tissue factor expression are either slower or much more subtle. Our results also indicate that these changes in thrombomodulin concentration could be mediated by changes of calcium flux into the endothelium.
The antithrombogenic properties of thrombomodulin include the sequestration of thrombin and the activation of protein C by the thrombomodulin-thrombin complex.17 Protein C deficiency, traditionally associated with venous thrombosis, has been reported in association with peripheral arterial thrombosis,18 19 and recently, variations in the thrombomodulin gene have been associated with coronary artery thrombosis.20 In cultured cells, the thrombomodulin-protein C pathway has been shown to regulate the thrombogenic properties of endothelium under shearing conditions simulating both the venous and arterial circulation.21 Thrombomodulin is only 1 component of the antithrombogenic system of vascular endothelium: Other components include heparan sulfates, plasminogen activator, tissue factor inhibitor, and the endothelium-derived vasodilators nitric oxide and prostacyclin, many of which are regulated by shear stress. In addition, preparation of the vein for grafting injures the endothelium, which may expose underlying tissue factor or promote tissue factor synthesis. Nevertheless, downregulation of endothelial thrombomodulin may be an important factor predisposing to early vein graft occlusion.
We have shown downregulation of thrombomodulin on saphenous vein endothelium within 45 to 90 minutes of exposure to arterial flow conditions. This downregulation of thrombomodulin appears to result from the increase in shear stress rather than circumferential deformation because external vein stenting did not alter the rapid reduction in thrombomodulin concentration. This rapid reduction in thrombomodulin concentration by modest shear stress (≈0.4 N/m2) contrasts with the findings in cultured bovine aortic endothelial cells, in which reductions in thrombomodulin mRNA concentrations were not observed at comparable levels of shear stress.3 Even at higher levels of shear stress (1.5 to 3.6 N/m2), a reduction in thrombomodulin mRNA was not observed until after 6 to 9 hours, and a reduction in protein concentration was not observed until after 36 hours.3
For most of this study, we used immunostaining as the primary technique to investigate thrombomodulin expression and confirmed the direction of all the important findings by Western blotting. Because we failed to detect any thrombomodulin in the vein perfusate, thrombomodulin may be recycled and degraded. This would be consistent with the findings in cultured endothelial cells, in which cytokines and phorbol esters stimulated internalization and degradation of thrombomodulin to reduce surface activity.22 Equally, it is possible that our techniques were not sufficiently sensitive to record circulating thrombomodulin. Unfortunately, we have not been able to explore whether the arterial flow–induced downregulation of thrombomodulin on saphenous vein endothelium diminishes the antithrombogenic properties of endothelium in situ. Although others have quantified thrombomodulin activity on saphenous vein using chromogenic assays,23 this method could not be used in our system with the problems of endothelial cell loss and increase in vein diameter in response to arterial flow conditions. However, we have shown that the ability of endothelial cells to activate protein C was markedly reduced after 90 minutes of arterial flow conditions.
The responses of endothelium to hemodynamic forces occur over a time frame ranging from minutes to days.2 In the intermediate-early period (0 to 90 minutes), we could find no evidence to support the altered expression of tissue factor, and tPA was not detected at all. These may be later changes, as they are in cultured endothelial cells exposed to shear stress.3 4 5 6 In only a single condition, nifedipine in the vein perfusate, was significant tissue factor staining observed on endothelium.
Potassium channels have been considered important in transducing signals from increased shear stress into changes in endothelial gene expression.24 25 Perfusion of vein with either the nonselective potassium channel blocker TEA or the KATP channel blocker glibenclamide did not alter the downregulation of thrombomodulin in response to arterial flow conditions (Table 1⇑). A role for calcium in transducing signals from increased shear stress also should be considered because the application of shear stress to cultured endothelial cells increases intracellular calcium concentrations.26 Low concentrations of the calcium ionophore A23187 caused a significant reduction in thrombomodulin concentration. In contrast, perfusion of vein with gadolinium or nifedipine prevented the thrombomodulin downregulation. Perfusion of vein with calcium-free Krebs’ solution also prevented the downregulation of thrombomodulin, with endothelial cells retaining their full thrombomodulin functional activity. The effect of nifedipine, which selectively blocks L-type voltage-gated calcium channels, was unexpected. Although these channels are present in vascular smooth muscle, there is limited evidence for the presence of these channels in endothelium.27 However, in arterioles, electrotonic coupling from smooth muscle to endothelium has been reported, with elevation of calcium concentrations in smooth muscle, caused by vasoconstrictors, leading to increased endothelial calcium concentrations.28 A similar phenomenon may account for the effects of nifedipine we observe.
In summary, we have shown that there is rapid downregulation of thrombomodulin concentration and functional activity in the endothelium of saphenous vein exposed to arterial flow conditions. These changes are a response to increased flow or shear stress rather than circumferential deformation. The relevance of these findings to vein bypass graft occlusion is a topic of current investigation.
We thank all the cardiac and vascular surgeons who have provided us with saphenous vein, David Lane for constructive criticism, and the British Heart Foundation for research support.
↵1 These authors contributed equally to this work.
- Received August 13, 1998.
- Revision received November 11, 1998.
- Accepted November 11, 1998.
- Copyright © 1999 by American Heart Association
Resnick N, Gimbrone MA Jr. Hemodynamic forces are complex regulators of endothelial gene expression. FASEB J.. 1995;9:874–882.
Malek AM, Jackman R, Rosenberg RD, Izumo S. Endothelial expression of thrombomodulin is reversibly regulated by fluid shear stress. Circ Res. 1994;74:852–860.
Diamond SL, Eskin SG, McIntire LV. Fluid flow stimulates tissue plasminogen activator secretion by cultured human endothelial cells. Science. 1989;243:1483–1485.
Higman DJ, Strachan AMJ, Buttery L, Hicks RCT, Springall DR, Greenhalgh RM, Powell JT. Smoking impairs the activity of endothelial nitric oxide synthase in saphenous vein. Arterioscler Thromb Vasc Biol. 1996;16:546–552.
Matlori R, Hamilton KK, Gugate RD, McEver RP, Sims PJ. Stimulated secretion of endothelial von Willebrand factor is accompanied by rapid redistribution to the cell surface of the intracellular granule membrane protein GMP140. J Biol Chem. 1989;264:7768–7771.
Kretschmer G, Herbst F, Prager M, Sautner T, Wenzl E, Berlakovich GA, Zekert F, Marosi L, Schemper M. A decade of oral anticoagulant treatment to maintain autologous vein grafts for femoropopliteal atherosclerosis. Arch Surg. 1992;127:112–115.
van der Meer J, Hillege HL, Kootstra GJ, Ascoop CA, Mulder BJ, Pfisterer M, van Gilst WH, Lie KI. Prevention of one-year vein-graft occlusion after aortocoronary-bypass surgery: a comparison of low-dose aspirin, low-dose aspirin plus dipyridamole, and oral anticoagulants: the CABADAS Research Group of the Interuniversity Cardiology Institute of The Netherlands. Lancet. 1993;342:257–264.
Esmon CT. Thrombomodulin as a model of molecular mechanisms that modulate protease specificity and function at the vessel surface. FASEB J. 1995;9:946–955.
Ireland H, Kunz G, Kyriakoulis K, Stubbs PJ, Lane DA. Thrombomodulin gene mutations associated with myocardial infarction. Circulation. 1997;96:15–18.
Cadroy Y, Diquelou A, Dupouy D, Bossavy JP, Sakariassen KS, Sie P, Boneu B. The thrombomodulin/protein C/protein S anticoagulant pathway modulates the thrombogenic properties of the normal resting and stimulated endothelium. Arterioscler Thromb Vasc Biol. 1997;17:520–527.
Ohno M, Cooke JP, Dzau VJ, Gibbons GH. Fluid shear stress induces endothelial transforming growth factor beta-1 transcription and production. J Clin Invest. 1995:1363–1369.
Tsao PS, Lewis NP, Alpert S, Cooke JP. Exposure to shear stress alters endothelial adhesiveness: role of nitric oxide. Circulation. 1995;92:3513–3519.
Shen J, Luscinskas S, Connolly A, Dewey CF, Gimbrone MA. Fluid shear stress modulates cystolic free calcium in vascular endothelial cells. Am J Physiol. 1992;262:C384–C390.
Murphy HS, Maroughi M, Till GO, Ward PA. Phorbol-stimulated influx of calcium in rat pulmonary artery endothelial cells. Am J Physiol. 1994;267:L145–L151.
Dora KA, Doyle MP, Duling BR. Elevation of intracellular calcium in smooth muscle causes endothelial generation of NO in arterioles. Proc Natl Acad Sci U S A. 1997;94:6529–6534.