Domain structure of the endothelial cell receptor thrombomodulin as deduced from modulation of its anticoagulant functions. Evidence for a glycosaminoglycan-dependent secondary binding site for thrombin

Abstract

Rabbit thrombomodulin (TM) influences blood coagulation by serving as a cofactor for thrombin-induced protein C activation (activity a), by directly affecting the procoagulant activity of thrombin (activity b) and by accelerating the inhibition of thrombin by antithrombin III (AT III) (activity c). Although high molecular weight cationic compounds, such as poly-L-lysine and the ionophore-releasate from human platelets, only partly affected activity a in a concentration-dependent manner, activities b and c, however, were almost totally inhibited by these cationic compounds. Likewise, a heparin- and dermatan sulfate-binding peptide which represents a portion of the glycosaminoglycan-binding domain of vitronectin (VN) selectively inhibited activities b and c, indicating the presence of clustered acidic domain(s) in TM responsible for these activities. While heparinase or heparitinase did not affect rabbit TM function at all, digestion of rabbit TM with chondroitin ABC-lyase abolished activities b and c, whereas activity a remained unaffected. Modification of rabbit TM with chondroitin ABC-lyase was associated with a decrease in molecular mass of the receptor by about 10 kDa and a 2- to 3-fold decrease in affinity to thrombin as deduced from direct binding studies. These results suggest that at least two acidic thrombin binding domains are present in rabbit TM, whereby a dermatan sulfate-like glycosaminoglycan moiety constitutes the secondary binding domain for thrombin, eliciting both the direct as well as the AT III-dependent anticoagulant function of rabbit TM (activities b and c) but not protein C activation (activity a). In contrast to rabbit TM, human TM isolated from placenta only showed weak activities b and c. These differences in reactivity of TM from different sources appeared to be due to the masking (or absence) of the proposed secondary thrombin binding site in human TM, since VN could be identified as a major contamination in the human TM preparation as revealed by enzyme-linked immunosorbent assay and Western blot analysis. In addition, the major part of human TM could be immunoprecipitated by monospecific antibodies to VN. These findings indicate a possible modulatory function for VN in the human thrombin-TM system.