Defects in vein valve PROX1/FOXC2 antithrombotic pathway in endothelial cells drive the hypercoagulable state induced by trauma and critical illness.

Objectives: Deep venous thrombosis (DVT) causes significant morbidity and mortality after trauma. Recently, we have shown that blood flow patterns at vein valves induce oscillatory stress genes, which maintain an anticoagulant endothelial phenotype that inhibits spontaneous clotting at vein valves and sinuses, is lost in the presence of DVT in human pathological samples, and is dependent on expression of the transcription factor FOXC2. We describe an assay, modifying our mouse multiple injury system, which shows evidence of clinically relevant microthrombosis and hypercoagulability applicable to the study of spontaneous DVT in trauma without requiring direct vascular injury or ligation. Finally, we investigated whether these model findings are relevant to a human model of critical illness by examining gene expression changes by quantitative polymerase chain reaction and immunofluorescence in veins collected from critically ill.

Methods: C57/Bl6 mice were subjected to a modified mouse multiple injury model with liver crush injury, crush and pseudofracture of a single lower extremity, and 15% total blood volume hemorrhage. Serum was assayed for d-dimer at 2, 6, 24, and 48 hours after injury by enzyme-linked immunosorbent assay. For the thrombin clotting assay, veins of the leg were exposed, 100 μL of 1 mM rhodamine (6 g) was injected retro-orbitally, and 450 μg/mL thrombin was then applied to the surface of the vein with examination of real-time clot formation via in vivo immunofluorescence microscopy. Images were then examined for percentage area of clot coverage of visible mouse saphenous and common femoral vein. Vein valve specific knockout of FOXC2 was induced with tamoxifen treatment in PROX1 Ert2Cre FOXC2 fl/fl mice as previously described. Animals were then subjected to a modified mouse multiple injury model with liver crush injury, crush and pseudofracture of a single lower extremity, and 15% total blood volume hemorrhage. Twenty-four hours after injury, we examined the valve phenotype in naive versus multiple injury animals, with and without loss of the FOXC2 gene from the vein valve (FOXC2 del ) via the thrombin assay. Images were then examined for proximity of clot formation to the valve present at the junction of the mouse saphenous, tibial, and superficial femoral vein and presence of spontaneous microthrombi present in the veins before exposure to thrombin. Human vein samples were obtained from excess tissue preserved after harvest for elective cardiac surgery and from organ donors after organ procurement. Sections were submitted for paraffin embedding and then assayed by immunofluorescence for PROX1, FOXC2, thrombomodulin, endothelial protein C receptor, and von Willebrand's factor. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee, and all human studies reviewed and approved by the institutional review board.

Results: After mouse multiple injuries, enzyme-linked immunosorbent assay for d-dimer showed evidence of products of fibrin breakdown consistent with formation of clot related to injury, fibrinolysis, and/or microthrombosis. The thrombin clotting assay demonstrated higher percentage area of vein covered with clot when exposed to thrombin in the multiple injury animals compared with uninjured (45% vs. 27% p = 0.0002) consistent with a phenotype of hypercoagulable state after trauma in our model system. Unmanipulated FoxC2 knockout mice manifest increased clotting at the vein valve as compared with unmanipulated wild type animals. After multiple injuries, wild type mice manifest increase clotting at the vein after thrombin exposure ( p = 0.0033), and equivalent to that of valvular knockout of FoxC2 (FoxC2del), recapitulating the phenotype seen in FoxC2 knockout animals. The combination of multiple injuries and FoxC2 knockout resulted in spontaneous microthrombi in 50% of the animals, a phenotype not observed with either multiple injuries or FoxC2 deficiency alone (χ 2 , p = 0.017). Finally, human vein samples demonstrated the protective vein valve phenotype of increased FOXC2 and PROX1 and showed decreased expression in the critically ill organ donor population by immunofluorescence imaging in organ donor samples.

Conclusion: We have established a novel model of posttrauma hypercoagulation that does not require direct restriction of venous flow or direct injury to the vessel endothelium to assay for hypercoagulability and can generate spontaneous microthrombosis when combined with valve-specific FOXC2 knockout. We find that multiple injuries induce a procoagulant phenotype that recapitulates the valvular hypercoagulability seen in FOXC2 knockout and, in critically ill human specimens, find evidence for loss of oscillatory shear stress-induced gene expression of FOXC2 and PROX1 in the valvular endothelium consistent with potential loss of DVT-protective valvular phenotype.