The distal carboxyterminal domains of murine ADAMTS13 influence proteolysis of platelet-decorated VWF strings in vivo


Karen Vanhoorelbeke, Laboratory for Thrombosis Research, IRC, K.U. Leuven Campus Kortrijk, E. Sabbelaan 53, 8500 Kortrijk, Belgium.
Tel.: +32 56 246019; fax: +32 56 246997.


Summary. Background: The multidomain metalloprotease ADAMTS13 regulates the size of von Willebrand factor (VWF) multimers upon their release from endothelial cells. How the different domains in ADAMTS13 control VWF proteolysis in vivo remains largely unidentified. Methods: Seven C-terminally truncated murine ADAMTS13 (mADAMTS13) mutants were constructed and characterized in vitro. Their ability to cleave VWF strings in vivo was studied in the ADAMTS13−/− mouse. Results: Murine MDTCS (devoid of T2-8 and CUB domains) retained full enzyme activity in vitro towards FRETS-VWF73 and the C-terminal T6-8 (del(T6-CUB)) and CUB domains (delCUB) are dispensable under these assay conditions. In addition, mADAMTS13 fragments without the spacer domain (MDT and M) had reduced catalytic efficiencies. Our results hence indicate that similar domains in murine and human ADAMTS13 are required for activity in vitro, supporting the use of mouse models to study ADAMTS13 function in vivo. Interestingly, using intravital microscopy we show that removal of the CUB domains abolishes proteolysis of platelet-decorated VWF strings in vivo. In addition, whereas MDTCS is fully active in vivo, partial (del(T6-CUB)) or complete (delCUB) addition of the T2-8 domains gradually attenuates its activity. Conclusions: Our data demonstrate that the ADAMTS13 CUB and T2-8 domains influence proteolysis of platelet-decorated VWF strings in vivo.


ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type 1 repeats, 13) is a plasma metalloprotease consisting of multiple domains, with at its N-terminus the metalloprotease domain (M), followed by the disintegrin domain (D), thrombospondin type 1 repeat (T), Cys rich domain (C), spacer domain (S), seven additional T and 2 CUB domains (Fig. 1A) [1]. mRNA transcripts of ADAMTS13 have been found in liver (hepatic stellate cells) [2], endothelial cells [3], platelets [4] and glomerular podocytes [5]. The only known substrate of ADAMTS13 is the multimeric plasma protein von Willebrand factor (VWF) [1].

Figure 1.

 mADAMTS13 and truncated mutants: schematic representation and Western blot analysis. (A) Schematic representation of WT mADAMTS13 and the seven C-terminally truncated mutants; the amino acid truncation site is indicated on the left. (B) Western blot analysis of recombinant WT mADAMTS13 and mutants in conditioned medium from HEK 293T cells. Culture media were analyzed by SDS-PAGE under reducing conditions, followed by Western blot using an anti-V5 HRP labeled antibody. The media of MDTC, MDT, MD and M were concentrated 20–40 times prior to analysis.

VWF is a glycoprotein [6,7] that is exclusively synthesized in endothelial cells and megakaryocytes. Endothelial VWF is either constitutively secreted or stored in Weibel-Palade bodies. These granules contain ultra-large VWF (UL-VWF) multimers with a molecular weight that can exceed 20 000 kDa. The adhesive activity of VWF multimers depends on molecular size, the largest multimers being the most active ones. To prevent spontaneous interaction with circulating platelets, UL-VWF multimers are, after release, rapidly processed by ADAMTS13 into smaller, less reactive multimers [8,9]. Deficiency of ADAMTS13 leads to accumulation of UL-VWF multimers in circulation, causing thrombotic thrombocytopenic purpura (TTP) [10].

The molecular mechanism of VWF proteolysis by ADAMTS13 is unique. Because the cleavage site Tyr1605-Met1606 in the VWF A2 domain is buried in the globular VWF molecule [11], the constitutively active ADAMTS13 in plasma can digest only the unfolded VWF protein. VWF unfolding can occur under high shear conditions [12], upon secretion from endothelial cells [13] or at sites of hemostatic plug formation [14]. Interestingly, unfolding of VWF does not only expose the scissile bond but also some of the exosites in VWF that interact with ADAMTS13. Extensive exosite interactions might guarantee substrate specificity and allow ADAMTS13 to anchor to different VWF conformations induced by fluid shear stress [15]. Indeed, high affinity interactions between the ADAMTS13 spacer domain and the C-terminal part of the unfolded VWF A2 domain contribute to substrate specificity [15,16]. In addition, low affinity interactions between the ADAMTS13 M, D and C domains and VWF amino acids close to the scissile bond guarantee specific hydrolysis of the Tyr1605-Met1606 bond [17,18]. Recently, a novel binding site in the C-terminal region of VWF (D4CK) interacting with distal domains in ADAMTS13 (T5-CUB) was identified. This interaction also occurs with globular, folded VWF and might serve as an initial docking site for ADAMTS13 to the C-terminal region of VWF [19,20]. The current knowledge of ADAMTS13’s mode-of-action is based on data in vitro. Data in vivo are largely lacking although they are important for the basic understanding of the physiological mechanism of ADAMTS13-mediated VWF proteolysis. In the present study, we used seven C-terminally truncated mutants of murine ADAMTS13 (mADAMTS13) and studied their ability to cleave VWF both in vitro and in vivo. We found that analogous domains in murine and human ADAMTS13 determine VWF proteolysis in vitro, confirming the value of using the mouse model to get insight into the function of both murine and human ADAMTS13 in vivo. Furthermore, intravital microscopy experiments, analyzing proteolysis of platelet-decorated VWF strings by the ADAMTS13 truncated mutants, revealed that the distal carboxyterminal domains of ADAMTS13 influence proteolysis of VWF in vivo.

Materials and methods


ADAMTS13-deficient (ADAMTS13−/−) and WT mice were bred from ADAMTS13B/CN2−/+ animals (gift from D. Ginsburg). All animal experiments were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of K.U. Leuven (Belgium).

Expression of WT mADAMTS13 and truncated mutants

Seven truncated mADAMTS13 mutants, each lacking a different number of C-terminal domains (Fig. 1A), were constructed and expressed in a stable HEK293T cell line (for detailed protocols see Supplementary Data S1). Conditioned media of mutants MDTC, MDT, MD and M were concentrated 20–40 times using a Vivaspin MWCO 10 kDa (Sartorius, Goettingen, Germany).

Analysis of expression of mADAMTS13 mutants in Western blot

WT mADAMTS13 and all mutants were analyzed by SDS-PAGE (10%) under reducing conditions and transferred to a Hybond C Extra membrane (GE Healthcare, Waukeska, WI, USA). After blocking with 5% skimmed milk (Nestlé, Vervey, Switzerland) in PBS, membranes were incubated for 1 h with anti-V5-HRP (1/3000 in PBS, 0.3% skimmed milk) (AbD Serotec, Kidlinton, UK). Blots were developed with the chemiluminescent ECL Plus detection system (GE Healthcare) and signals were detected using a luminescent image analyzer (LAS-4000 Mini, Fujifilm, Tokyo, Japan).

Concentration determination of WT mADAMTS13 and truncated mutants

The concentration of WT mADAMTS13 and truncated mutants was determined by Western blotting using known concentrations of recombinant human ADAMTS13, containing a V5-His-tag (huADAMTS13-V5-His), as a reference. Each sample was loaded on the gel in triplicate, together with serial dilutions of huADAMTS13-V5-His (10–40 nm). Concentrations of huADAMTS13 were determined in ELISA as previously described [21,22]). Proteins were detected with anti-V5-HRP and chemiluminescence as described above and quantified with the calibration curve of huADAMTS13-V5-His, using a luminescent image analyzer (LAS-4000 Mini) and software (Multi Gauge Ver3.x, Fujifilm) [22]. Concentrations of mADAMTS13 variants ranged from 1 to 30 nm.

To know how the concentrations of WT mADAMTS13 and its truncated mutants related to the concentrations of mADAMTS13 present in plasma, the concentration of WT mADAMTS13 was compared with the amount of mADAMTS13 present in mouse plasma using an immunoassay. The expression levels of WT mADAMTS13 relative to the amounts of ADAMTS13 present in plasma of WT mice was determined as follows. A 96-well microtiter plate was coated with the polyclonal rabbit anti-mADAMTS13L (gift from K. Soejima) at 5 μg mL−1 in PBS overnight at 4 °C. After blocking for 2 h with 3% skimmed milk (Nestlé), a dilution series of pooled mouse WT plasma (in PBS, 0.3% skimmed milk), used as a standard, and a dilution series of WT mADAMTS13 expression medium (in PBS, 0.3% skimmed milk) was added and incubated for 1 h at 37 °C. Bound ADAMTS13 was detected by biotinylated polyclonal anti-mADAMTS13S antibodies (10 μg mL−1 in PBS 0.3% milk) (gift from K. Soejima) and HRP-labeled streptavidin (1/15 000 dilution in PBS, 0.3% milk). Colorimetric development was with o-phenylenediamine dihydrochloride (Sigma-Aldrich, St Louis, MO, USA) in 50 mm phosphate-citrate buffer containing H2O2, pH 5.0. Reactions were stopped with 0.95 m sulphuric acid, and the absorbance at 490 nm was determined. The ADAMTS13 level in pooled mouse plasma was set at 100%.

Digestion of murine GST-VWF73

Purified GST-mVWF73 (3 μg) was incubated with conditioned medium containing WT mADAMTS13 or its truncated mutants and the cleavage products (28 kDa) were detected after Western blotting using an anti GST-HRP labeled antibody (for detailed protocols see Supplementary Data S1).

Digestion of VWF multimers

Purified partially denatured human VWF was incubated with conditioned medium containing WT mADAMTS13 or its mutants. The (176)2 kDa VWF dimer fragment was detected with anti-VWF HRP labeled antibodies (for detailed protocols see Supplementary Data S1).

Kinetics of FRETS-VWF73 cleavage

Conditioned expression medium of WT mADAMTS13 or its mutants was changed to reaction buffer (50 mm Hepes, 150 mm NaCl, 5 mm CaCl2, 1 μm ZnCl2, pH 7.4) using Zeba Spin desalting columns (Thermo Scientific, Waltham, MA, USA) and the concentration was determined by Western blotting, as described. Known concentrations of WT mADAMTS13 or its mutants were diluted in 100 μl 2× reaction buffer containing 1 mg mL−1 BSA in a preblocked white microtiter plate (Greiner Bio-One, Frickenhausen, Germany). Then, 100 μL of various FRETS-VWF73 [23] (Peptides International, Louisville, KY, USA) concentrations (between 0.2 and 1 μm in MQ-water) were added. Fluorescence intensities were measured with a FLUOstar OPTIMA (BMG Labtech GmbH, Offenburg, Germany) every minute for 1 h using excitation at 355 nm and emission at 460 nm at 37 °C. Complete digestion of the substrate was used to relate changes in fluorescence to product formation. The observed velocities as a function of substrate concentration were fit to the linear part of the Michaelis-Menten equation obtained at low substrate concentrations: vobs = kcat/KM*[E]*[S] by linear regression [15,24] using Microsoft Excel software, to calculate the specificity constant kcat/KM.

Analysis of variance was assessed by anova and a post-test using Tukey’s algorithm, defining P < 0.05 as significant.

Digestion of endothelial cell anchored VWF multimers by ADAMTS13 and its mutants in vivo

ADAMTS13−/− and WT mice, weighing 12–15 g, were anesthetized by intraperitoneal injection of 60 μg g−1 sodium pentobarbital (Nembutal®, Ceva Sante Animale, Libourne, France). Concentrated conditioned medium containing 2.5 times the normal plasma levels of WT mADAMTS13 or its mutants and rhodamine 6G (1 μg g−1 body weight) (Invitrogen, Carlsbad, CA, USA), to fluorescently label the platelets, were simultaneously injected via the retro-orbital sinus. An incision was made through the abdominal wall to expose the mesenteric blood vessels and a filter paper saturated with FeCl3 (10%) was applied topically for 2 min on a mesenteric microvessel, covering both vein (diameter of 180 ± 6 μm, n = 19) and artery (diameter of 109 ± 5 μm, n = 15). After removal of the filter paper, venules were scanned downstream of the application site to identify a region where platelet-decorated VWF strings appeared but thrombus formation was minimal. During the following 5 min, the lifetime of minimally 15 platelet-decorated VWF strings was determined per mouse, which we designated ‘string survival time’ (SST). For each experiment, the median was calculated and the mean ± SEM of these medians is reported. From start (retro-orbital injection) to finish (end of the SST determination) the experiment takes, on average, 16 min. Analysis of variance was assessed by anova and a post-test using Dunnett’s algorithm, defining P < 0.05 as significant.

VWF strings decorated with fluorescently labeled platelets were visualized using a Nikon eclipse TE200 inverted microscope (Nikon Instruments, Melville, NY, USA) (objective 20 ×) connected to a Hamamatsu CCD camera (ORCA-R2, Hamamatsu Photonics, Hamamatsu City, Japan). Videos were recorded using HCImage software version 2.0 (Hamamatsu Photonics).


Expression of wild-type mADAMTS13 and truncated mutants

Wild-type mADAMTS13 and seven C-terminally truncated mutants were constructed and cloned into the pcDNA6.1 expression vector (Fig. 1A). Stable HEK 293T cell lines were generated for all seven constructs following selection with Blasticidin. Secretion of WT mADAMTS13 and each mutant into the culture medium was assessed by Western blot analysis using anti-V5 HRP labeled antibodies (Fig. 1B). WT mADAMTS13 and all mutants except MDTC were secreted in the expression medium, albeit with different efficiencies. The immunoreactive bands of the truncated mutants ranged from 30 to 200 kDa, corresponding to their expected molecular weights (Fig. 1B). The secretion of the MDTC mutant in the expression medium was minimal, although an immunoreactive band for the MDTC mutant in the cell lysate was detected (data not shown). Hence, MDTC was not further used.

Proteolysis of VWF73 and VWF multimers by mADAMTS13 mutants under static conditions

Proteolytic activities of the different mADAMTS13 constructs were measured by the digestion (2 h) of the recombinant substrate, GST-mVWF73 [25,26]. Cleaving activity was detected for WT mADAMTS13 and mutants delCUB, del(T6-CUB), MDTCS and MDT but not for MD and M (Fig. 2A). Only after prolonged incubation (24 h), was significant proteolytic activity of MD and M constructs observed (Fig. 2B).

Figure 2.

 Qualitative analysis of proteolytic activity of recombinant mADAMTS13 variants, in vitro. (A and B) Purified GST-mVWF73 was incubated with conditioned expression medium of either WT mADAMTS13 or mutants for 2 h (A) or 24 h (B) in the presence (+) or absence (−) of EDTA. Samples were analyzed by SDS-PAGE under reducing conditions and detected by immunoblotting with anti-GST HRP labeled antibodies. Digestion of GST-mVWF73 results in the appearance and detection of the N-terminal GST-mVWF73 fragment of 28 kDa. (C,D) Purified human VWF was incubated for 1 h (C) or 24 h (D) in the presence of guanidine-HCl with conditioned media of either WT mADAMTS13 or truncated mutants. As a control, purified human VWF digestion was blocked by the addition of EDTA (+). Samples were analyzed by SDS-PAGE under non-reducing conditions followed by Western blotting and detected with anti-VWF HRP labeled antibodies. Digestion of VWF results in the appearance of a (176)2 kDa C-terminal VWF fragment.

The activity of WT mADAMTS13 and its truncated mutants was further investigated using human VWF multimers (1 h digestion) [27]. In accordance with the GST-mVWF73 data, WT mADAMTS13 and mutants delCUB, del(T6-CUB), MDTCS and MDT all showed proteolytic activity towards full-length VWF, while no activity could be detected for MD and M (Fig. 2C). Prolonged incubation (24 h) resulted in detectable activity for MD but not for M (Fig. 2D).

Because different concentrations of enzyme were present in the conditioned media used in the above assays, cleavage rates between the different mADAMTS13 mutants could only be compared qualitatively. We therefore determined the catalytic efficiency (kcat/KM) of WT mADAMTS13 and the different mutants (except for MD) using FRETS-VWF73 and standardized concentrations of all enzymes (Table 1). For WT mADAMTS13 the kcat/KM was found to be 3.4 ± 0.33 × 105 m−1 s−1. The kcat/KM values for delCUB and del(T6-CUB) were in the same range, delCUB was moderately decreased (1.5 ± 0.065 × 105 m−1 s−1) and del(T6-CUB) moderately increased (6.1 ± 0.20 × 105 m−1 s−1), while MDTCS had increased 4-fold compared with WT mADAMTS13. Although MDT cleaved GST-VWF73, it was 44 times less efficient in processing FRETS-VWF73 compared with WT mADAMTS13. Also the ADAMTS13 M mutant was significantly less efficient, in accordance with the data on cleavage of GST-VWF73 and multimeric VWF.

Table 1.   Specificity constants for WT mADAMTS13 and truncated mutants
Constructkcat/KM (105 m−1 s−1) (n = 3)
  1. Known concentrations of the ADAMTS13 constructs were incubated with various concentrations of FRETS-VWF73 at 37 °C. The observed rates of change in fluorescence intensity were analyzed as function of FRETS-VWF73 concentration. Data are mean ± SEM.

WT3.4 ± 0.33
delCUB1.5 ± 0.065
del(T6-CUB)6.1 ± 0.20
MDTCS15 ± 0.53
MDT0.078 ± 0.0014
M0.087 ± 0.0015

In conclusion, these data demonstrate that efficient proteolysis of VWF is conserved for the MDTCS fragment, demonstrating that the T2-8 and CUB domains are not needed for effective substrate turnover under these assay conditions. In addition, MDTCS is more active than WT mADAMTS13. However, removal of the spacer domain, as in MDT, (MD) and M, drastically reduces enzyme efficiency.

Interestingly, this characterization of mADAMTS13 variants in vitro for the first time reveals that the contribution of the different domains of mADAMTS13 to its activity is analogous to what has been observed for hADAMTS13 in vitro [15,28–30]. Hence, the mouse model is a valuable tool to study the physiological working mechanism of murine and human ADAMTS13.

Proteolysis of platelet-decorated VWF strings by mADAMTS13 variants in vivo

To unravel the role of the different ADAMTS13 domains in VWF processing in vivo, we studied the capability of mADAMTS13 variants to cleave VWF strings on the surface of endothelial cells in the mesenteric veins of ADAMTS13−/− mice. We used topical application of FeCl3 to damage/activate the mesenteric endothelium. Chauhan et al. [14] observed increased platelet binding to this activated subendothelium in ADAMTS13−/− mice. In addition, we observed that also in this model platelet-decorated VWF strings could be monitored on the activated endothelium of the venules, downstream of the FeCl3 application site. These platelet-decorated VWF strings had a length of 35–148 μm, appeared to be attached to one end and waved in the direction of the blood flow, as is observed after stimulation of the mesenteric venules with calcium ionophore (Fig. 3A,B and Supplementary Video S1). In addition, platelet-decorated VWF strings occasionally occurred in the arterioles running parallel to the venules after topical application of FeCl3 and could not be observed in stimulated venules of VWF−/− mice (Fig. 3C).

Figure 3.

 Platelet-decorated VWF strings on the surface of mesenteric endothelium, in vivo. Endogenous platelets and leukocytes were labeled by injection of rhodamine 6G. Mice were anesthetized and mesenteric blood vessels were exposed. A filter paper saturated with FeCl3 (10%) solution (A,C) or calcium ionophore (A23187) (10 μL of a 100 μm solution) (B) was topically applied and appearance of VWF strings (arrows) on the surface of endothelial cells in mesenteric veins was visualized by intravital fluorescence microscopy in ADAMTS13−/− mice (A,B) and VWF−/− mice (C).

As expected, platelet-decorated VWF strings survived longer in ADAMTS13−/− mice than in WT animals, having a string survival time (SST) of 14 ± 4.0 s (mean ± SEM) (n = 7) vs. 4.7 ± 1.1 s (n = 7), respectively (< 0.05, Fig. 4A and Supplementary Videos S1 and S2). Injection of mock expression medium in ADAMTS13−/− mice did not alter the SST (15 ± 1.8 s, Fig. 4B), whereas injection of WT mADAMTS13 significantly decreased the SST to 3.2 ± 0.63 s (n = 10), similar to that observed in WT mice (Fig. 4A). Taken together, these data demonstrate that SST is a reliable measure for ADAMTS13 function in vivo.

Figure 4.

 String survival time of platelet-decorated VWF strings in ADAMTS13+/+, ADAMTS13−/− mice or ADAMTS13−/− mice injected with recombinant mADAMTS13 variants. Mesenteric veins were exposed and endothelial cells were damaged/activated by topical application of FeCl3. Platelets were visualized by retro-orbital injection of rhodamine 6G delivered through the retro-orbital plexus and the lifetime of minimal 15 platelet-decorated VWF strings was monitored (string survival time, SST) (median SST was taken for each mouse). (A) SST in ADAMTS13+/+ (n = 10) and ADAMTS13−/− mice. (n = 10) (B) Equimolar concentrations (2.5 times normal plasma level) of WT mADAMTS13 (n = 12), delCUB (n = 6), del(T6-CUB)) (n = 7), MDTCS (n = 6) and MDT (n = 10) or mock (n = 32) were injected via the retro-orbital plexus in mADAMTS13−/− mice before exposure of the mesenteric veins. Enzyme concentrations were determined by Western blotting and immunodetection with anti-V5 HRP labeled antibodies using densitometry. Data are mean ± SEM. *P < 0.05.

Interestingly, upon injection of mADAMTS13 lacking the 2 CUB domains (delCUB), the SST (14 ± 3.8 s, n = 7)) did not change, indicating that the two CUB domains are necessary for ADAMTS13-mediated cleavage of platelet-decorated VWF strings in vivo. Further removal of the T6-8 domains (del(T6-CUB)) slightly recovered enzyme activity because a small decrease in SST was observed (9.7 ± 2.4 s, n = 14), Fig. 4B). Interestingly, however, additional deletion of the T2-5 domains (MDTCS) resulted in complete restoration of ADAMTS13 activity with a median SST of 3.0 ± 0.52 s (n = 6). ADAMTS13 activity, however, was undetectable when also the Cys-rich and spacer domain were absent (MDT) (SST of 16 ± 4.1 s, n = 10). In conclusion, these data demonstrate a regulatory effect of the C-terminal CUB domains on the digestion of the platelet-decorated VWF strings in vivo, an effect not seen in the digestion of VWF73 and VWF multimers in vitro (Table 1). Furthermore, MDTCS retains full enzyme activity in vivo while gradual addition of the T2-5 and T6-8 domains decreases the activity.


ADAMTS13 plays a critical role in the prevention of circulating UL-VWF multimers. The contribution of the different ADAMTS13 domains to the proteolysis of VWF in vivo remains largely unknown. Yet, unravelling the mode-of-action of ADAMTS13 in vivo is of crucial importance to our understanding of the physiological role of ADAMTS13 in hemostasis. These data in vivo can challenge, confirm or extend our current models based on data in vitro, under non-physiological conditions. Indeed, conditions of ADAMTS13 assays in vitro are very distinct from the physiological environment in which VWF multimers are naturally cleaved.

The goal of this study was to investigate the role of the different domains in ADAMTS13 in vivo using seven C-terminally truncated murine ADAMTS13 mutants (Fig. 1). However, no detailed analysis of mADAMTS13 variants is yet available, making a thorough study in vitro warranted prior to studies in vivo.

Proteolysis of VWF73 or VWF multimers by ADAMTS13 variants in vitro under static conditions, demonstrated that in mADAMTS13, as in hADAMTS13, the distal portion of mADAMTS13 is dispensable for digestion of VWF73, because removal of the CUB domains or T6-8 and CUB domains did not negatively influence the proteolysis of VWF73 or VWF multimers. Accordingly, also the naturally occurring truncated variant of mADAMTS13 found in several strains of mice, devoid of the two C-terminal T domains and CUB domains but containing at its C-terminus 16 amino acids from the IAP-retrotransposon, has the same activity as full-length mADAMTS13 in cleaving VWF73 [25,31]. In contrast, the spacer domain is essential for enzyme activity of both murine and human ADAMTS13. Indeed, mADAMTS13 variants devoid of the spacer domain (MDT, MD and M) have severely reduced catalytic efficiencies (Fig. 2, Table 1). It is well known that truncation of hADAMTS13, N-terminal of the spacer domain, severely affects enzyme efficiency in vitro under static conditions [15,28–30]. Indeed, the spacer domain facilitates VWF proteolysis because it harbors a major exosite, which interacts with an exosite at the C-terminus of the VWF A2 domain [15,16,30]. We furthermore report that the murine MDTCS mutant displayed a 4-fold increase in activity towards FRETS-VWF73 compared with WT mADAMTS13 (Table 1). Interestingly, this observation was also reported for human MDTCS under both static and flow conditions and already suggested a regulatory role for the C-terminal region in ADAMTS13 activity [31,32]. Only the isolated M domain of murine and human ADAMTS13 seemed to react differently. Prolonged incubation of GST-VWF73 with murine M did result in specific proteolysis of VWF while prolonged incubation with human M resulted in a cleavage of GST-VWF73 at site different from the predicted Tyr1605-Met1606 [29]. Altogether, our data demonstrate that murine and human ADAMTS13 are comparable, indicating that mice are a valuable model to unravel the mode-of-action of ADAMTS13 in vivo.

Hence we studied VWF processing by ADAMTS13 in stimulated mesenteric veins of ADAMTS13−/− mice using intravital fluorescence microscopy. Long-lived platelet-decorated VWF strings appear on the surface of the stimulated endothelial cells in ADAMTS13−/− mice while these strings were rapidly removed by ADAMTS13 in WT mice [14,33]. By measuring the lifetime of these platelet-decorated VWF strings, we were able to compare the activity of the different truncated C-terminal mADAMTS13 mutants in vivo. As expected, injection of WT mADAMTS13 into ADAMTS13−/− mice efficiently restored VWF string processing.

However, a first important observation was that removal of the CUB domains (delCUB) completely abolished VWF string processing. This strongly suggests that the CUB domains in mADAMTS13 positively regulate VWF string proteolysis, maybe by serving as a docking site for VWF multimers. Interestingly, the naturally truncated variant of mADAMTS13 was also shown to be less active than WT ADAMTS13 in thrombus size regulation in vivo [34]. In addition, Tao et al. [35] demonstrated that recombinant CUB1 and CUB1 peptides partially block hADAMTS13-mediated cleavage of platelet-VWF strings in vitro under flow. In addition, we and others recently demonstrated that the C-terminal hADAMTS13 domains (T5-CUB) [19,20] interact with the C-terminal region of soluble VWF (D4CK) [20] with moderate affinity and hence can serve as an initial docking site for VWF.

Analogous to our in vitro results, MDTCS retains full activity in vivo but removal of the Cys- and spacer domain (MDT) severely reduces the enzyme’s activity in vivo to undetectable levels. Interestingly, however, adding the T2-5 domains and the T6-8 domains gradually attenuated the activity of MDTCS. This indicates that the T2-8 domains might negatively regulate enzyme activity. A hypothesis is that the catalytic site and/or the spacer domain in ADAMTS13 are shielded by the T2-8 tail, an interaction that can be disrupted after initial docking by the CUB domains, allowing interaction of the catalytic site and spacer domain with the VWF A2 domain. Whether there is an interaction between the N- and C-terminal ADAMTS13 domains is currently under investigation and is not included in this study. In addition, we cannot exclude conformational differences between MDTCS, del(T6-CUB) and delCUB induced by the removal of several domains or the presence of cofactors in plasma interacting differently with the various mutants.

In conclusion, we performed a detailed analysis of the role of the different domains of mADAMTS13 in VWF processing. Interestingly, we found that analogous domains in murine and human ADAMTS13 modulate the processing of VWF73 and VWF multimers in vitro, indicating that mice provide a valuable model to study ADAMTS13’s function in vivo. We furthermore showed in a physiological activity assay in vivo that the CUB domains are needed to regulate ADAMTS13-mediated proteolysis of VWF strings in vivo, probably by serving as a docking site to VWF.


B. De Maeyer is supported by a PhD grant (IWT 61132) from the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen). H.B. Feys and S.F. De Meyer are fellows of the Research Foundation Flanders [Fonds voor Wetenschappelijk Onderzoek Vlaanderen (FWO)]. This work was supported by grants from the FWO (G.0299.06) and the KU Leuven (GOA/2004/09). We thank D. Ginsburg for providing the ADAMTS13B/CN2−/+ mice and mADAMTS13-pcDNA3.1 plasmid. We thank K. Soejima (The Chemo-Sero-Therapeutic Research Institute, Kumamoto, Japan) for providing the polyclonal anti-mADAMTS13S and anti-mADAMTS13L antibodies. We thank H. Pottel (Laboratory for Biophysics, KU Leuven, Campus Kortrijk) for help with statistical analysis.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.