ADAMTS13 and Its Variants Do Not Cleave Fibrinogen Directly

Dear Editor

von Willebrand factor (VWF) is the only substrate identified for ADAMTS13 to date [1]. How ADAMTS13 achieves such remarkable substrate specificity remains a mystery. Previous studies demonstrated that carboxyl-terminal domains including 2–8^th thrombospondin type 1 repeat (TSP1) and CUB domains are dispensable for cleavage of VWF substrate under various conditions [2, 3]. However, these carboxyl-terminal domains were later found to exhibit a disulfide bond reducing activity [4] and to interact with globular VWF via its D4 domain [5]. More recently, the TSP1 2–8 and CUB domains (T2C) were shown to have an inhibitory function through their interactions with the amino terminal domains, presumably the spacer domain [6]. This C-terminus-mediated inhibition of ADAMTS13 activity was alleviated by exposing the full-length protease to an acidic environment, by binding of a monoclonal antibody against the C-terminal domains, by removing the C-terminal T2C domains [6, 7], as well as by alteration of surface exposed residues in the spacer domain, which presumably interact with the CUB domains [7, 8].

An animal study demonstrates that ADAMTS13 may have thrombolytic activity by dissolving a preformed clot induced by FeCl₃ injury in venules [9]. Such thrombolytic activity cannot be adequately explained by the cleavage of VWF alone. South et al reported in the J. Thromb. Haemost. [2016;14(10): 2011–22] that an ADATMS13 variant lacking the C-terminal TSP1 2–8 repeats and CUB domains (S) or a gain-of-function ADAMTS13 variant (R568K/F592Y/R660K/Y661F/Y665F in the spacer domain) (GoF), first discovered in our laboratory [8], was able to cleave fibrinogen directly [10]. Such an activity was also observed with full-length ADAMTS13 (FL) after being bound by the VWF-D4 domain [10]. Authors further demonstrated that the GoF variant could impair fibrin formation in plasma, alter clot structure, increase clot permeability, and increase susceptibility of fibrin to tissue plasminogen activator (t-PA), thereby enhancing thrombolysis under flow [10]. Authors, therefore, concluded that the carboxyl-terminal truncated ADAMTS13 variant or the gain-of-function ADAMTS13 mutant and full-length ADAMTS13 bound by VWF-D4 are in their “open” conformations, capable of cleaving fibrinogen and fibrin [10].

To follow up this exciting, but somewhat intriguing report, we performed a series of experiments with two different preparations of human fibrinogen: a commercial fibrinogen (fgn-1) (Sigma, St. Louis, MO) used by South et al in the article [10] and a highly purified fibrinogen (fgn-2) (a kind gift from Dr. Joe Bennett from the University of Pennsylvania). Consistent with the results reported [10], an incubation of the commercial fgn-1 (1 mg/ml) with the ADAMTS13 variant truncated after spacer domain (S) or after the 8^th TSP1 repeat (T8), as shown in Fig. 1A, at the final concentration of 200 nM in 20 mM Hepes, pH 7.4 containing 150 mM NaCl, and 5 mM CaCl₂ at 37 °C for 16 hours resulted in proteolytic cleavage of fgn-1. The cleavage products fragment X (~240 kDa), fragment Y (~160-kDa), fragment D (~90 kDa), and fragment E (~50 kDa) were detected in the SDS-polyacrylamide gel electrophoresis with Coomassie blue staining. The digestion pattern was similar to that by plasmin (data not shown). The increase in cleavage products coincided with the loss of the intact fibrinogen (~340-kDa) (Fig. 1B & 1C & D, lanes 3–4). No proteolytic cleavage product was observed when fgn-1 was incubated with buffer alone (Fig. 1C& D, lane 1), FL (200 nM) (Fig. 1C & D, lane 2), and T8C (200 nM) (Fig. 1C & D, lane 5). To our surprise, the proteolytic cleavage of a highly purified fibrinogen fgn-2 by S and T8 was not observed under the same conditions (Fig. 1C & D, lanes 8–9), which was not different from that by a buffer alone (Fig. 1C & D, Lane 6), FL (Fig. 1C & D, lane 7), and T8C (Fig. 1C & D, lane 10). As a positive control, both preparations of fibrinogen showed high sensitivity to proteolysis by plasmin (data not shown), suggesting no intrinsic difference in these fibrinogen preparations except for their purity. Furthermore, the proteolytic cleavage of commercial fgn-1 by T8 was abolished by addition of a serine protease inhibitor, including D-phenylalanyl-L-prolyl-L-arginine chloromethyl ketone (PPACK) (100 μM) (Fig. 1E, lane 4), Pefablock (2 mM) (Fig. 1E, lane 5), and α₂-plasmin inhibitor (α₂-PI) (Fig. 1E, lane 6). These results suggest that the carboxyl-terminal truncated ADAMTS13 variant (S or T8) does not cleave fibrinogen directly; it does so through an activation of a serine protease.

Fig. 1. ADAMTS13 and its variants cleave fibrinogen via activation of plasminogen.

A. Schematic representation of a full-length ADAMTS13 and its variants (top) and SDS-polyacrylamide gel electrophoresis and Coomassie blue staining for the purified recombinant human ADAMTS13 and variants (bottom); B. Schematic representation of fibrinogen and its potential degradation products (Adopted from Walker et al, JBC, 1999;274:5201-12); C and D. Proteolytic cleavage of a commercial fibrinogen (fgn-1) and a highly purified fibrinogen (fgn-2) by ADAMTS13 and its variants after 16 hours of incubation, visualized by SDS-PAGE with Coomassie blue staining under non-reducing and reducing conditions, respectively; E. Inhibition of T8-mediated proteolytic cleavage of fgn-1 by serine protease inhibitors including PPACK (PP), pefabloc (PB), and α₂-plasmin inhibitor (PI); F. Western blotting with anti-plasminogen/plasmin IgG detected plasminogen in fgn-1, but not in fgn-2. Purified glu-plasminogen (plg) (0–200 ng/lane) was used for calibration. G. Time-dependent proteolysis of fgn-2 in the presence of 1 mg/ml of glu-plg by S (0–96 hour) and by FL (96 hours); H. Time-dependent cleavage of fgn-2 in the presence of glu-plg by T8 (0–24 hours) or T8C or FL (24 hours). The symbol asterisk indicates the presence of α₂-PI.

Plasmin was the primary suspect of such a serine protease that cleaves fibrinogen and fibrin. Western blotting analysis demonstrated the presence of plasminogen (plg) in fgn-1 (Fig. 1F, lane 6), but not in fgn-2 (Fig. 1F, lane 5). Indeed, an addition of glu-plasminogen (1 mg/ml) to fgn-2 dramatically accelerated the proteolytic cleavage of fgn-2 by constructs S and T8 in a dose- (not shown) and time-dependent manner (Fig. 1G & H). Addition of α₂-PI completely inhibited the proteolytic cleavage of fgn-1 and fgn-2 in the presence of glu-plasminogen. Interestingly, after 96 hours of incubation, FL was able to cleave fgn-2 in the presence of glu-plasminogen (Fig. 1G, lane 8), but not in the absence of glu-plasminogen (not shown). No autodegradation of fgn-2 or activation of glu-plasminogen was observed under the same conditions without ADAMTS13 or variants (not shown).

These results demonstrate that FL or its variants have no direct proteolytic activity towards fibrinogen or fibrin. The observed proteolysis of fgn-1 and fgn-2 in the presence of glu-plasminogen by ADATMS13 variants (S and T8) or FL after a prolonged incubation, which is sensitive to inhibition by a serine protease inhibitor, suggests that the potential activation of plasminogen (contaminated in fgn-1 or added to fgn-2) to plasmin may be responsible for the cleavage of fibrinogen or fibrin. The activation of plasminogen may be mediated by a contaminated t-PA or t-PA/plasmin-like protease in recombinant ADAMTS13 or variants and/or by recombinant ADAMTS13 or variants themselves. Further investigation of the interactions between ADAMTS13 (or variants) and various components in the fibrinolytic system including t-PA, plasminogen, and fibrinogen may shed new light on the functions of ADAMTS13. These findings will help understand the pathogenesis of thrombotic thrombocytopenic purpura and other arterial thrombotic disorders and in the design of a novel strategy for anti-thrombotic therapy.