FRETS-VWF73, a first fluorogenic substrate for ADAMTS13 assay


Koichi Kokame, National Cardiovascular Centre Research Institute, 5-7-1 Fujishirodai, Suita, Osaka 565-8565, Japan.


A plasma metalloprotease, ADAMTS13, cleaves von Willebrand factor (VWF) multimers and downregulates their activity in platelet aggregation. Functional ADAMTS13 deficiency leads to the accumulation of hyperactive large VWF multimers, inducing a life-threatening disease, thrombotic thrombocytopenic purpura (TTP). Although measuring ADAMTS13 activity is important in TTP diagnosis, existing methods require time and skill. Here, we report a fluorescence resonance energy transfer (FRET) assay for ADAMTS13 activity. We developed a synthetic 73-amino-acid peptide, FRETS-VWF73. Cleavage of this substrate between two modified residues relieves the fluorescence quenching in the intact peptide. Incubation of FRETS-VWF73 with normal human plasma quantitatively increased fluorescence over time, while ADAMTS13-deficient plasma had no effect. Quantitative analysis could be achieved within a 1-h period using a 96-well format in commercial plate readers with common filters. The FRETS-VWF73 assay will be useful for the characterization of thrombotic microangiopathies like TTP and may clarify the importance of ADAMTS13 activity as a predictive marker for various thrombotic diseases.

Thrombotic thrombocytopenic purpura (TTP), a syndrome characterized by thrombocytopenia and microangiopathic haemolytic anaemia, is often associated with neurological dysfunction, renal failure and fever (Moschcowitz, 1924; Moake et al, 1982). Although most patients with TTP experience these crises idiopathically throughout adulthood, some patients present with neonatal onset and frequent relapses, also called Upshaw–Schulman syndrome (USS). Recent genetic studies have revealed that the majority of USS patients are homozygous or compound heterozygous for a critical mutation of the ADAMTS13 gene (Levy et al, 2001; Kokame et al, 2002; Antoine et al, 2003; Assink et al, 2003; Savasan et al, 2003; Schneppenheim et al, 2003; Matsumoto et al, 2004; Pimanda et al, 2004). ADAMTS13 encodes a plasma metalloprotease of the ADAMTS family (Levy et al, 2001; Soejima et al, 2001; Zheng et al, 2001; Plaimauer et al, 2002; Banno et al, 2004). Many patients with acquired TTP possess inhibitory auto-antibodies against ADAMTS13 (Furlan et al, 1998; Tsai & Lian, 1998). As the clinical characteristics of TTP are similar to those of other microangiopathic haemolytic anaemias, such as haemolytic uremic syndrome (HUS) and disseminated intravascular coagulation (DIC), an assay measuring ADAMTS13 activity would be an useful tool for appropriate diagnosis and treatment of TTP.

ADAMTS13 cleaves the peptidyl bond between Y1605 and M1606 in the A2 domain of von Willebrand factor (VWF) (Dent et al, 1990; Tsai et al, 1994; Furlan et al, 1996; Tsai, 1996), which circulates in plasma as large multimeric forms, ranging in size from 500 to 20 000 kDa. Functional ADAMTS13 deficiency can lead to the accumulation of large, hyperactive VWF multimers. A method to measure VWF-cleavage activity of ADAMTS13 was originally developed by Furlan et al (1996) and Tsai (1996), in which purified human VWF multimers were incubated with plasma in the presence of either urea or guanidine. The reaction products were separated by sodium dodecyl sulphate (SDS)-agarose (Furlan et al, 1996) or SDS-polyacrylamide (Tsai, 1996) gel electrophoresis, followed by Western blotting analysis with anti-VWF antibodies. Although these methods have significantly increased our understanding of the role of ADAMTS13 in TTP pathogenesis, they are not widely used at the clinical level because of technical complications.

Several groups have attempted to develop more simple and rapid diagnostic procedures for clinical use, including a collagen-binding assay (Gerritsen et al, 1999), an immunoradiometric assay using two site-directed VWF antibodies (Obert et al, 1999) and a ristocetin-cofactor assay (Böhm et al, 2002). Multicentre comparison studies of these different assays showed varied performance but supported the usefulness of the ADAMTS13 assay for TTP diagnosis (Studt et al, 2003; Tripodi et al, 2004). These assays, however, still demand complicated procedures and highly specialized materials. Therefore, a more rapid, reliable and convenient method of measuring VWF activity is eagerly awaited.

As chromogenic substrate assays are used in the clinical measurement of protease activities, initial studies were sought to identify a short oligopeptide that can be specifically cleaved by ADAMTS13 (Furlan & Lämmle, 2002). As these attempts have systematically failed, the cleavage at Y1605–M1606 of VWF probably depends on both the specific residues in the vicinity of the scissile bond and more remote sequences. Recently, we have succeeded in creating a recombinant substrate encompassing the shortest region of VWF that serves as a specific substrate for ADAMTS13 (Kokame et al, 2004). The peptide substrate, designated VWF73, contains 73-amino-acid residues of VWF from D1596 to R1668. In this study, we have chemically modified VWF73 to facilitate the quantitative measurement of ADAMTS13 activity in a single-step procedure.

Materials and methods


The fluorogenic substrate, FRETS-VWF73, was chemically synthesized by Thermo Electron GmbH (Sedanstrasse, Ulm, Germany) and the Peptide Institute, Inc. (Osaka, Japan). It was dissolved in 25% dimethyl sulphoxide/water to prepare the 100-μmol/l stock solution. Human plasma was obtained by centrifugation from whole blood that was treated with a 1/10 volume of 3·8% sodium citrate as an anti-coagulant. A protease inhibitor cocktail (Sigma, St Louis, MO, USA) used in the cleavage experiments contained 1 mmol/l 4-(2-aminoethyl)benzenesulphonyl fluoride, 15 μmol/l pepstatin A, 14 μmol/l trans-epoxysuccinyl-L-leucylamido(4-guanidino)butane, 36 μmol/l bestatin, 21 μmol/l leupeptin and 0·8 μmol/l aprotinin at a final concentration.

Fluorescent assay to measure the ADAMTS13 activity

Pooled human plasma (a range of 0–8 μl as a standard), or 4 μl of each test plasma, was diluted in 100 μl of assay buffer (5 mmol/l Bis–Tris, 25 mmol/l CaCl2, 0·005% Tween-20, pH 6·0) in a 96-well white plate (Sumitomo Bakelite, Tokyo, Japan). Then, 100 μl of 4 μmol/l FRETS-VWF73 in the assay buffer was added to each well. Fluorescence was measured at 30°C in a Wallac 1420 ARVO multilabel counter (PerkinElmer Japan, Yokohama, Japan) equipped with a 340-nm excitation filter and a 450-nm emission filter. Fluorescence was measured every 5 min. The reaction rate was calculated by linear regression analysis of fluorescence over time from 0 to 60 min using the prism software (GraphPad Software, San Diego, CA, USA).

Preparation of recombinant ADAMTS13 (rADAMTS13)

HeLa cells were cultured in Dulbecco's minimal essential medium (Invitrogen, Carlsbad, NM, USA) supplemented with 10% fetal bovine serum in humidified air with 5% CO2 at 37°C. To produce rADAMTS13, the human ADAMTS13-expression plasmid was transfected into the subconfluent cells using FuGENE6 (Roche Diagnostics, Indianapolis, IN, USA), as described previously (Kokame et al, 2002; Matsumoto et al, 2004). Following a 4-h incubation, the culture medium was replaced with serum-free OPTI-MEM I medium (Invitrogen) and the culture was incubated for 44 h. The medium was collected and concentrated to one-eighth the original volume using Centricon YM-30 (Millipore, Billerica, MA, USA). As a negative control, a series of operations was performed in parallel as for the untransfected cells.

Subject population

The Suita study participants were arbitrarily selected from the municipality population registry of Suita city, stratified by gender and 10-year age groups. The basic sampling of the population started in 1989 with a cohort study base (Mannami et al, 1997). In the present study, 100 consecutive samples were selected from this population as a control group. This study was approved by the ethical committee on human research of the National Cardiovascular Centre. Written informed consent was obtained from all subjects prior to testing.


Design of the fluorogenic substrate for ADAMTS13

To utilize fluorescence resonance energy transfer (FRET) to measure ADAMTS13 activity, we chemically synthesized a fluorogenic peptide, FRETS-VWF73 (Fig. 1), containing the 73-amino acids from D1596 to R1668 of VWF. Within this peptide, the Q1599 residue at the P7 position was converted to a 2,3-diaminopropionic residue (A2pr) modified with a 2-(N-methylamino)benzoyl group (Nma). The N1610 residue of the P5′ position was converted to A2pr modified with a 2,4-dinitrophenyl group (Dnp). When the Nma group is excited at 340 nm, fluorescence resonance energy is transferred to the neighbouring quencher, Dnp. If the bond between Y1605 and M1606 is cleaved, the energy transfer quenching the fluorescence does not occur, allowing the emission of fluorescence at 440 nm from Nma.

Figure 1.

Structure of FRETS-VWF73. Within the 73-amino-acid peptide sequence, corresponding to the region from D1596 to R1668 of von Willebrand factor (VWF), Q1599 and N1610 were substituted with A2pr(Nma) and A2pr(Dnp) respectively. The arrowhead indicates the site cleaved by ADAMTS13.

Cleavage of FRETS-VWF73 by plasma ADAMTS13

To explore the cleavage activity present in plasma, FRETS-VWF73 was incubated with normal human plasma in a fluorescent plate reader. Emission at 450 nm increased with time, indicating that FRETS-VWF73 was cleaved between the two A2pr residues by a plasma component (Fig. 2A). The increase of fluorescence was not inhibited by the addition of a protease inhibitor cocktail (mixed inhibitors effective against a broad range of serine proteases, cysteine proteases, aminopeptidases and acid proteases), but was completely inhibited by a divalent cation chelator (EDTA), suggesting that cleavage was mediated by the plasma metalloprotease, ADAMTS13, with minimal contribution of other plasma proteases. In fact, neither thrombin nor plasmin (5 μg/ml each, Sigma) increased fluorescence of FRETS-VWF73 (data not shown). The incubation of FRETS-VWF73 with plasma from an ADAMTS13-deficient patient showed no increase of fluorescence (Fig. 2A). The addition of ADAMTS13-deficient plasma to the normal plasma did not interfere with the cleavage of FRETS-VWF73 by the normal plasma (data not shown).

Figure 2.

Cleavage of FRETS-VWF73 by ADAMTS13. (A) Fluorescence changes in FRETS-VWF73 during incubation with plasma. FRETS-VWF73 was incubated with normal plasma from a single donor in the absence (closed circles) or presence of either protease inhibitors (open circles) or EDTA (open triangles). The substrate was also incubated with ADAMTS13-deficient plasma from a congenital thrombotic thrombocytopenic purpura patient (closed triangles). Fluorescent emission at 450 nm was measured at the indicated times. A representative of three repetitive experiments is shown. (B) Fluorescence changes in FRETS-VWF73 during incubation with recombinant ADAMTS13 (rADAMTS13). FRETS-VWF73 was incubated with the conditioned medium of HeLa cells transfected with (closed circles) or without (open circles) ADAMTS13-expression plasmid DNA. Values shown are the mean and SD (n = 3).

To verify further the cleavage by ADAMTS13, the substrate was incubated with the conditioned medium of cultured HeLa cells (Fig. 2B). Incubation with the medium of ADAMTS13-transfected cells showed the time-dependent increase of fluorescence, whereas the incubation with the medium of untransfected cells did not. All these data supported the conclusion that ADAMTS13 specifically cleaved FRETS-VWF73.

Plasma-dose dependency

FRETS-VWF73 cleavage was quantitatively dependent on plasma dosage (Fig. 3). We monitored fluorescence increase in the presence of variable volumes of normal plasma to the reaction mixture. The fluorescence over time increased with increasing plasma in a dose-dependent manner (Fig. 3A). To compensate for any differences in background fluorescence derived from plasma itself and to calculate the initial reaction rate, we estimated the slopes of the fluorescence over time using time points 0 and 60 min from a linear regression. These slopes (reaction rates) were then plotted against the plasma dosage (Fig. 3B). The data points fitted to a non-linear regression, indicating that ADAMTS13 activity in sample plasma could be estimated from the fluorescence reaction rate.

Figure 3.

Plasma-dose dependency of FRETS-VWF73 cleavage. (A) Fluorescence was measured at 5-min intervals after the addition of FRETS-VWF73 to 0–6 μl normal plasma. (B) The reaction rates of time points 0 and 60 min were plotted against plasma dosage. Values were fit to a non-linear regression curve.

Optimization of the FRETS-VWF73 assay

We next optimized reaction conditions to increase both the sensitivity and rapidity of measurement (Fig. 4). As ADAMTS13 requires divalent metal ions for proteolytic activity, we monitored the cleavage of FRETS-VWF73 by plasma in the presence of various metal ions (Fig. 4A). Ca2+ and Ba2+ ions were the most favourable for the reaction, although Mg2+ and Zn2+ also enhanced ADAMTS13 activity. In contrast, Mn2+ and Ni2+ could not activate the reaction, consistent with previous reports (Furlan et al, 1996; Tsai, 1996). Testing of various Ca2+ ion concentrations revealed that a range of 10–50 mmol/l Ca2+ was optimal for the reaction (Fig. 4B). We also examined the effect of differing NaCl concentrations, determining that lower concentrations provided more rapid cleavage (Fig. 4C), as seen in previous reports (Furlan et al, 1996; Kokame et al, 2004). The pH optimum for the FRETS-VWF73 assay was approximately 6·0 (Fig. 4D), which differed from previous studies reporting an optimal pH of 8·0–10·0 for the cleavage reaction (Furlan et al, 1996). This inconsistency may be a result of different reaction conditions, such as the presence or absence of denaturants. In addition, substitution of Q1599 and N1610 to A2pr(Nma) and A2pr(Dnp), respectively, may affect the cleavage pH dependency. Alternatively, pH dependency of the assay might be affected not only by the cleavage efficiency, but also fluorescence emission, because most fluorescence reactions are highly pH dependent. Regardless, these data indicated that the FRETS-VWF73 assay was most efficient in the reaction buffer containing 5 mmol/l Bis–Tris, 25 mmol/l CaCl2 and 0·005% Tween 20 at pH 6·0.

Figure 4.

Optimization of the FRETS-VWF73 assay. (A) To determine metal ion dependency, FRETS-VWF73 was incubated with normal plasma from a single donor in the presence of the indicated divalent ions. (B) To measure Ca2+-concentration dependency, FRETS-VWF73 was incubated with plasma in the presence of 0–100 mmol/l CaCl2. (C) Ion-strength dependency was determined by incubating FRETS-VWF73 with plasma in the presence of 0–1000 mmol/l NaCl. (D) To measure pH dependency, FRETS-VWF73 was incubated with plasma in the indicated pH buffer. The reaction rates of time points 0 and 60 min are shown with the mean and SD (n = 3).


We examined inter-run reproducibility of the FRETS-VWF73 assay. Plasma-dose dependency in the optimized condition was observed independently seven times. Each regression curve corresponded well with the other curves, indicating that the assay was obviously reproducible (data not shown). The relative ADAMTS13 activities of three different plasma samples were also measured independently seven times, where the activity of pooled plasma was normalized as 100%. The mean ± standard deviation (SD) values of the three samples were 113·9 ± 2·4, 62·5 ± 2·1 and 22·3 ± 1·4% (n = 7), respectively, indicating that the inter-assay variation was significantly small. The coefficients of variation of the three samples were 2·1, 3·4 and 6·3% (n = 7) respectively.

Plasma ADAMTS13 activity of patients and healthy individuals

To evaluate the FRETS-VWF73 assay for potential clinical use, we measured the relative ADAMTS13 activity in 78 plasma samples from various patients and 100 healthy individuals (Fig. 5A). The relative activities were estimated from the activity of pooled plasma prepared from all the 100 healthy individuals (66·0 ± 11·7 years old). Plasma samples from congenital TTP patients, homozygotes or compound heterozygotes of critical ADAMTS13 mutations (Kokame et al, 2002; Matsumoto et al, 2004), all exhibited very low (<1%) or undetectable activities. The majority (33 samples) of plasma samples obtained from 41 patients with idiopathic TTP also showed low (<5%) or undetectable activities. The most possible explanation would be a deficiency of plasma ADAMTS13 level or generation of auto-antibodies against ADAMTS13, although there may be some other factor, such as auto-antibodies, that bind to the substrate and protect it from being cleaved. In contrast, plasma from parents or siblings of congenital TTP patients, heterozygotes of ADAMTS13 mutations, exhibited on average approximately half the activity (59·0 ± 14·4%) of healthy individuals, while the plasma of patients with HUS showed substantial activity (73·2 ± 32·3%). Thus, the FRETS-VWF73 assay can be used to measure ADAMTS13 activity for TTP diagnosis in clinical samples.

Figure 5.

ADAMTS13 activity in plasma from patients and healthy individuals. (A) Relative ADAMTS13 activity was measured using the FRETS-VWF73 assay. FRETS-VWF73 was incubated with 4 μl of plasma from seven congenital thrombotic thrombocytopenic purpura (TTP) patients, 41 idiopathic TTP patients, 12 haemolytic uremic syndrome patients and 18 heterozygotes of ADAMTS13 with critical mutations. Plasma samples from randomly selected 100 healthy individuals were also examined. The relative ADAMTS13 activities were estimated from the standard curve, which was drawn up on the basis of the reaction rates of the pooled plasma prepared from all the 100 healthy individuals. (B) Association of ADAMTS13 activity with gender. Plasma ADAMTS13 activity in the healthy individuals was plotted by gender (45 men and 55 women). Bars indicate the means. Unpaired t-test identified significant differences between men and women (P = 0·0016). (C) Association of ADAMTS13 activity with age. The plasma ADAMTS13 activity correlated significantly with age (P < 0·0001). A line of best fit is indicated.

Association of ADAMTS13 activity with gender and age

The measured ADAMTS13 activities of plasma samples from 100 healthy individuals (45 men aged 67·4 ± 11·5 years old and 55 women aged 64·9 ± 11·8 years old) were plotted according to gender (Fig. 5B). Comparison of the ADAMTS13 activities between men (97·9 ± 19·2%) and women (113·5 ± 27·1%) using the unpaired t-test demonstrated a significant difference between these groups (P = 0·0016), suggesting that the ADAMTS13 activity of women should be significantly higher than that of men. Examination of the effect of age on ADAMTS13 activities using Spearman's rank correlation revealed a significant correlation (r = −0·396, P < 0·0001) (Fig. 5C). The slopes of best fit in linear regression analysis were −0·894 ± 0·196, with R2 values of 0·175 (P < 0·0001), suggesting that plasma ADAMTS13 activity should decrease with advancing age, at least after the early 40s.


The Y1605–M1606 bond is inaccessible in native VWF and made sensitive to ADAMTS13 by denaturation and shear force. Structural modelling has suggested that the bond is buried in the core β-sheet of the VWF A2 domain (Jenkins et al, 1998; Sutherland et al, 2004). This partially explains the requirement for denaturants or shear force in the hydrolysis of the Y1605–M1606 bond by ADAMTS13. VWF73, corresponding to the C-terminal two-fifths of the A2 domain, can be efficiently cleaved by ADAMTS13 in the absence of denaturants and shear force (Kokame et al, 2004), suggesting that the N-terminal three-fifths of the A2 domain may prevent ADAMTS13 from accessing the cleavage site. A recent study indicated that the VWF A1 domain inhibits cleavage of the A2 domain by ADAMTS13; binding of platelet glycoprotein Ibα to the A1 domain appears to relieve the inhibition (Nishio et al, 2004). As VWF73 is a relatively small substrate, cleavage is less likely to be affected by other molecules. Therefore, VWF73 is an appropriate core for the convenient single-step fluorogenic assay for ADAMTS13 activity developed in this study.

Being a chemically modified version of VWF73 containing A2pr(Nma) and A2pr(Dnp), FRETS-VWF73 was a good substrate for ADAMTS13 cleavage, suggesting that Q1599 at the P7 position and N1610 at the P5′ position are not essential for the cleavage. We also examined the substitution of N1602 at the P4 position to A2pr(Nma). Although the peptide could be cleaved by plasma ADAMTS13, the efficiency was lower than that of the original FRETS-VWF73 (data not shown). The shorter distance of the modified residue from the cleavage site may interfere with efficient cleavage by ADAMTS13.

Enzymatic studies of ADAMTS13 will progress using FRETS-VWF73 as a model substrate in the future. The previously established substrate, purified plasma VWF, is comprised of non-uniform multimers with multiple cleavage sites. In contrast, FRETS-VWF73 is a monomeric molecule with a single cleavage site, facilitating the determination of cleavage kinetic parameters. No denaturants are required for the reaction, making this assay more closely reflect the physiological conditions. Although the optimal cleavage of FRETS-VWF73 still requires a hypotonic environment, isotonic solution gives approximately 80% of the activity observed in NaCl-free conditions (Fig. 3C) for kinetic analyses. VWF73, however, is not suitable for studying the functions of the other VWF domains, such as A1 and A3.

The greatest impact of the FRETS-VWF73 assay will be as a potential clinical diagnostic test. Unlike previous assays, the assay is a simple procedure, requiring no special reagents or equipment except a fluorescence spectrophotometer. These advantages may popularize ADAMTS13-activity measurement at the clinical level. The best possible application will be the appropriate diagnosis of TTP. The FRETS-VWF73 assay could be useful also for selecting curative plasma before administration to patients, as ADAMTS13 activity in the general population varies widely (Fig. 5). The selection of high-titre plasma may improve the responses of patients to plasma infusion or exchange treatment.

The relationship between ADAMTS13 deficiency and TTP is more complicated than originally thought (George & Vesely, 2004; Zheng et al, 2004); the problem may be because of symptomatical and pathological variety and diagnostic criteria of TTP. Not all patients with TTP present the classical five features of disease, thrombocytopenia, microangiopathic haemolytic anaemia, neurological dysfunction, renal failure and fever. Although severe ADAMTS13 deficiency is observed in most patients with idiopathic TTP without pre-existing medical conditions (Furlan et al, 1998; Tsai & Lian, 1998), the association between ADAMTS13 deficiency and TTP is unclear in less highly selected patient groups (Veyradier et al, 2001; Vesely et al, 2003). ADAMTS13 measurement cannot be used to predict exactly response to plasma exchange in patients that are clinically diagnosed with TTP (Vesely et al, 2003). An accurate ADAMTS13 assay may help to categorize TTP patients into subgroups and help establish objective diagnostic criteria.

What should be the cut-off value of ADAMTS13 activity for the diagnosis of TTP or ADAMTS13 deficiency? The present and previous studies (Mannucci et al, 2001; Veyradier et al, 2001; Böhm et al, 2002) showed a wide distribution of the ADAMTS13 activity in the healthy population. Further, we showed that the ADAMTS13 activity was associated with gender and age. As we used pooled plasma that was derived from relatively older individuals as a standard, the apparent ADAMTS13 activity of patient plasma may be over estimated in the present study. To determine the universally applicable cut-off value, the definition of standard plasma will be of primary importance. The availability of purified or recombinant ADAMTS13 may help the standardization of ADAMTS13 assay. The gender- and age-oriented distribution of ADAMTS13 activity will need to be determined in the general population. Although the FRETS-VWF73 assay detected significant activity in some idiopathic TTP patients, the value was evidently lower than the lowest activity of 100 healthy individuals (Fig. 5A). Therefore, the cut-off value (for instance, the mean −2 SD % of normal activity) may have to be determined considering gender and age. The FRETS-VWF73 assay, suitable for high-throughput measurement, would accelerate such a population study.


We thank Dr Masanori Matsumoto and Dr Yoshihiro Fujimura for patient plasma, Dr Kenji Soejima for ADAMTS13-expression plasmid, Dr Masahiko Tsunemi for critical discussion and the members of the Satsuki–Junyukai for attending the project. This work was supported in part by grants-in-aid from the Ministry of Health, Labour and Welfare of Japan; the Ministry of Education, Culture, Sports, Science and Technology of Japan; the Program for Promotion of Fundamental Studies in Health Science of the Pharmaceuticals and Medical Devices Agency (PMDA) of Japan; and Mitsubishi Pharma Research Foundation.