News and Publications

The Antithrombotic Protein C Pathway

J.H. Griffin, B.N. Bouma,* H. Deguchi, J.A. Fernández, A.J. Gale, M.J. Heeb, Y.M. Lee, L. Mosnier, N. Pecheniuk, X. Xu, S. Yegneswaran

* University Hospital Utrecht, Utrecht, the Netherlands

Coagulation pathways, fibrinolysis pathways, and anticoagulant mechanisms must act in concert to prevent bleeding and yet not produce harmful blood clots. The protein C pathway provides physiologic antithrombotic activity by inactivating coagulation factors Va and VIIIa. Our research concerns the blood coagulation pathways, the compensatory anticoagulant protein C pathway, and the molecular mechanisms or defects related to the development of thrombosis.

Advances in the past year include preparation of recombinant factor V mutants that shed new light on mechanisms of factor Va inactivation, discovery of a glucosylceramide deficiency associated with venous thrombosis, and evidence that certain neutral glycosphingolipids surprisingly enhance the anticoagulant activity of the protein C pathway.

MECHANISMS OF FACTOR Va INACTIVATION

Procoagulant factor Va is inactivated via limited proteolysis by activated protein C (APC) due to 3 cleavages in the A2 domain, at Arg306, Arg506, and Arg679. The mechanisms by which these cleavages cause loss of factor Va activity are unclear. Previous data indicated that cleavages at Arg506 and Arg306 occur near binding sites for factor Xa (downregulation of factor Xa-binding mechanism), whereas cleavage at Arg306 severs covalent linkage of the A2 domain to factor Va such that the A2 domain fragments are dissociated from the rest of the factor Va molecule (A2 domain dissociation mechanism). To clarify the roles of the A2 dissociation and downregulation of factor Xa-binding mechanisms, we engineered a disulfide bond into recombinant factor Va mutants between the A2 and A3 domains (designated A2-SS-A3).

The disulfide bond in A2-SS-A3 prevented dissociation of the C-terminal half of the A2 domain after cleavage at Arg306 by APC. Even after cleavage by APC at Arg306, the mutant Q506-A2-SS-A3-Va, which retained the A2 domain, was resistant to APC inactivation. On the basis of a variety of experiments, our data emphasize that APC cleavages intrinsically cause substantial loss of factor Va activity that can be overcome by high concentrations of factor Xa. Furthermore, dissociation of the A2 domain after proteolysis at Arg306 in factor Va by APC provides an irreversible mechanism for factor Va inactivation.

GLUCOSYLCERAMIDE DEFICIENCY

To assess the relationship between venous thrombosis and plasma glucosylceramide or phosphatidylethanolamine, plasma levels of glucosylceramide and phosphatidylethanolamine were determined for 70 patients with venous thrombosis and in 70 subjects without venous thrombosis (control subjects). Mean levels of glucosylceramide (P < .001) but not phosphatidylethanolamine (P = .48) were lower in patients than in control subjects. As a measure of relative risk, the odds ratio for deep vein thrombosis in patients with glucosylceramide levels below the 90th percentile of control subjects was 5.7 (95% CI = 2.3-14).

To assess the influence of glycolipids on anticoagulant response to APC/protein S in clotting assays, we depleted endogenous plasma levels of glucosylceramide by treatment with glucocerebrosidase or added exogenous purified glucosylceramide or other neutral glycolipids to plasma. Treatment with glucocerebrosidase reduced plasma sensitivity to APC/protein S in parallel with glucosylceramide reduction. Exogenously added glucosylceramide and the homologous glucosyl-containing globotriaosylceramide dose-dependently prolonged tissue factor-induced clotting of normal plasma in the presence but not the absence of APC/protein S, suggesting that glucosylceramide and globotriaosylceramide can enhance protein C pathway activities. Thus, deficiency of plasma glucosylceramide may be a risk factor for venous thrombosis.

ENHANCEMENT OF ANTICOAGULANT ACTIVITY BY NEUTRAL GLYCOSPHINGOLIPIDS

We also studied the effects of incorporation of neutral glycolipids into phospholipid vesicles on APC/protein S activity. Inactivation of purified factor Va by APC was dose-dependently enhanced by the presence of glucosylceramide in vesicles containing 70%-90% phosphatidylcholine (PC) and 10% phosphatidylserine (PS). Inactivation of factor Va was further enhanced by protein S in a dose-dependent fashion, and this protein S-dependent enhancement was greater with vesicles containing 80% PC, 10% PS, and 10% glucosylceramide than with vesicles containing 90% PC and 10% PS. Factor Va inactivation by APC/protein S was also dose-dependently enhanced by incorporation of 15%-20% trihexose-containing globotriaosylceramide into vesicles with 70%-90% PC and 10% PS. In contrast, incorporation of 1%-20% galactosylceramide into PC/PS vesicles resulted in minimal if any enhancement of factor Va inactivation. In factor Xa-1-stage clotting assays with normal plasma, incorporation of glucosylceramide or globotriaosylceramide, but not galactosylceramide, into PC/PS vesicles dose-dependently prolonged clotting times in the presence of APC, without affecting baseline clotting times.

Thus, incorporation of certain but not all neutral glycosphingolipids into PC/PS vesicles enhanced anticoagulant activity of APC/protein S in plasma and purified systems. This surprising finding that neutral glycolipids in PC/PS vesicles enhance the anticoagulant protein C pathway provides a mechanistic rationale for the discovery that deficiency of plasma glucosylceramide is associated with increased prevalence of venous thrombosis.

On the basis of our laboratory and clinical results, we hypothesize that certain neutral glycosphingolipids that can function as anticoagulant lipid cofactors are localized in microdomains (so-called lipid rafts) on the external leaflet of cell membranes or on lipoproteins that enhance anticoagulant reactions in vivo because they provide specific sites for the antithrombotic action of the protein C pathway. We further hypothesize that defects in these antithrombotic microdomains predispose to venous thrombosis, and possibly to arterial thrombosis.