Molecular Biology of ADAMTS13 and Diagnostic Utility of ADAMTS13 Proteolytic Activity and Inhibitor Assays

Abstract

ADAMTS13, a reprolysin-like metalloprotease, limits platelet-rich thrombus formation in the small arteries by cleaving von Willebrand factor (vWF) at the Tyr1605-Met1606 peptide bond. Deficiency of plasma ADAMTS13 activity, due to either an inherited or an acquired etiology, may lead to a potentially lethal syndrome, thrombotic thrombocytopenic purpura (TTP). Molecular cloning and characterization of the ADAMTS13 gene have provided further insight into the structure-function relationships, biosynthesis, and regulation of the ADAMTS13 protease, in addition to understanding the pathogenesis of TTP and perhaps other thrombotic disorders. ADAMTS13 consists of a short propeptide, a typical reprolysin-like metalloprotease domain, followed by a disinte-grin-like domain, first thrombospondin type 1 (TSP1) repeat, Cys-rich domain, and spacer domain. The carboxyl terminus of ADAMTS13 has seven more TSP1 repeats and two CUB domains. ADAMTS13 is synthesized mainly in hepatic stellate cells, but also in vascular endothelial cells. Recognition and cleavage of vWF require the proximal carboxyl terminal domains, but not the middle and distal carboxyl terminal domains. Cleavage of vWF appears to be modulated by shear force, binding to platelet or platelet glycoprotein-1bα, heparin, inflammatory cytokine (interleukin-6), and chloride ion. At the site of thrombus formation, the ADAMTS13 may be inactivated by thrombin, plasmin, and factor Xa. Having a sensitive and specific assay for ADAMTS13 activity is not only critical to understand the basic biology of ADAMTS13 protease, but also to facilitate a more timely and accurate clinical diagnosis of TTP, and to initiate potentially life-saving plasma exchange therapy. Although many assays have been developed and tested for clinical applications, the fluorescent resonance energy transfer-vWF73 assay appears to be the simplest and most promising assay to date.

The 13th member of adisintegrin-like and metalloprotease with thrombospondin type 1 motifs (ADAMTS) family, ADAMTS13,^1–7 limits platelet aggregation and microthrombi in small arteries by cleaving vWF at the Tyr1605-Met1606 bond.^8,9 Deficiency of ADAMTS13 protease due to either an inherited⁵ or acquired etiology¹⁰ may result in formation of disseminated microvascular thromboses, exhibiting a potentially lethal syndrome, thrombotic thrombocytopenic purpura (TTP).^10–13

TTP had classically been diagnosed using a pentad of criteria: thrombocytopenia, microangiopathic hemolytic anemia (MAHA), fever, renal dysfunction, and neurological abnormalities.¹¹ However, the diagnosis has been made recently in patients with only thrombocytopenia and MAHA.^14–17 On a histological level, these symptoms are attributable to the presence of platelet- and von Willebrand factor (vWF) -rich micro-thrombi in the terminal arterioles and capillaries of distal organs, leading to organ failure, coma, and death.¹⁸ Until the late 1960s, TTP was virtually fatal, with a mortality rate of >90%,¹⁹ but with advances in plasma exchange, the survival rate from acute TTP approaches 80 to 90%.^20,21

Several reviews have summarized the roles of vWF and ADAMTS13 in TTP.^13,22–30 This article specifically focuses on certain aspects of ADAMTS13 biology: the molecular cloning, structure-function relationship, biosynthesis, regulation of proteolytic activity, ADAMTS13 mutations found in congenital TTP patients, and the diagnostic utility of assays for plasma ADAMTS13 proteolytic activity and inhibitors.

MOLECULAR CLONING OF ADAMTS13

In 1996, two groups attempted to isolate a protease from normal human plasma that cleaves vWF, although the attempts were not successful initially. However, much was learned about the biochemical properties of their partially purified products.^8,9 The protease displays a greater proteolytic activity when vWF is subject to high shear stress or denatured by urea or guanidine in the presence of low ionic strength. Divalent cations such as Ba²⁺, Zn²⁺, Ca²⁺ or Co²⁺, are necessary to activate the protease.^8,9 These early studies led to the development of several functional assays critical for determining the protease activity in human plasma or for tracking the activity in the fractions during subsequent chromatographic purification of the enzyme.^1,2

In 2001, the vWF-cleaving metalloprotease was finally isolated and its partial amino-terminal sequence was determined.^1,2 This led to the identification and classification of the vWF-cleaving metalloprotease as a new member of the ADAMTS family, designated ADAMTS13.^1–4 By a completely different approach, the identical ADAMTS13 gene was also cloned.⁵

The ADAMTS13 gene is located on chromosome 9q34, spanning 37 kb in length and containing 29 exons.^4,5 Several alternatively spliced variants were found during the cloning and sequencing.^3–5 Analysis of the translated product showed that ADAMTS13 belongs to the metzincin superfamily of zinc metalloproteases, which, thus far, contains 19 family members that share the common structure of a hydrophobic signal sequence, a propeptide, a metalloprotease domain, first thrombo-spondin-1 (TSP1) repeat, a disintegrin-like domain, a cysteine-rich domain, and a spacer domain.^3–5 The carboxyl terminus of ADAMTS13 contains seven more TSP1 repeats and two CUB domains, which are unique to ADAMTS13 and were previously identified in complement components c1r and c1s, urinary EGF, and bone morphogenetic protein.³¹

BIOSYNTHESIS OF ADAMTS13

Northern blotting detects a 4.7-kb mRNA in human liver,^3–5 3.5- to 4.7-kb mRNAs in livers of different mouse strains,^32,33 and a 2.4-kb mRNA in human placenta and skeletal muscle.⁴ In situ hybridization and immunohistochemical staining with a monoclonal antibody directed against the disintegrin domain of ADAMTS13 have localized the ADAMTS13 mRNA and protein, respectively, to hepatic stellate cells (the interstitial cells of liver parenchyma that are elongated and varied in size and shape).^34,35 ADAMTS13-expressing cells are also positive for α-smooth muscle actin, suggesting only activated hepatic stellate cells or myofibroblasts produce ADAMTS13.³⁴ Cell fractionation showed that full-length ADAMTS13 is synthesized and secreted from primary hepatic stellate cells from mouse³⁵ and rat (M. Niiya and X. Zheng, unpublished data, 2005). Other cell types such as hepatocytes, sinusoidal endothelial cells, or Kupffer cells do not express detectable amounts of ADAMTS13 intracellularly or in the culture medium.³⁵ Furthermore, cell lines derived from normal hepatocytes (THLE-3) or hepatocellular carcinoma (HuH-7) do not express ADAMTS13, but those derived from hepatic stellate cells of human (hTERT-HSC) and rat (CFSC-3H and -8B) do secrete active ADAMTS13 into the culture medium.³⁵ These data further support that activated hepatic stellate cells are the major ADAMTS13-producing cells in the liver. Activated hepatic stellate cells have been implicated in the development of liver fibrosis and cirrhosis^36–38 after various insults because of their ability to produce and to deposit extracellular matrix proteins such as fibril collagen^36–38 in the injured area. Thus, upon activation, the enhanced proteolytic activity of ADAMTS13 in the hepatic stellate cells may be associated with either formation or resolution of liver fibrosis. More studies are needed to delineate the relationship between ADAMTS13 protease and liver fibrosis.

In addition to hepatic stellate cells, platelets or other tissues may also produce and secrete ADAMTS13. Reverse transcriptase polymerase chain reaction (RT-PCR) using a single-strand cDNA can detect the ADAMTS13 mRNA transcripts in almost all major organ tissues, including human heart, lungs, pancreas, brain, kidneys, adrenal glands, placenta, uterus, ovaries, and prostate.^5–7 Such ubiquitous distribution of ADAMTS13 mRNA suggests that vascular endothelial cells may produce ADAMTS13. Our results (D. Shang, X.W. Zheng, and X. Zheng, unpublished data, 2005) showed that primary human umbilical cord vein endo-thelial cells, primary aortic endothelial cells, and microvascular endothelial cell lines (EVC304) all synthesize and secrete full-length ADAMTS13, that is, detected by RT-PCR, Western blotting with rabbit anti-ADAMTS13, and capable of proteolytic cleavage of vWF and vWF73 substrates. Considering the enormous surface area covered by endothelial cells, it is possible that the plasma pool of ADAMTS13 may be derived from the vascular endothelial cells, in addition to hepatic stellate cells.

In a pulse-chase experiment, ADAMTS13 is readily secreted and can be detected within 3 hours of chase in the conditioned medium of transiently trans-fected HeLa cells.³⁹ The secreted recombinant ADAMTS13 migrates at 190 kd on a sodium dodecyl sulfate (SDS) polyacrylamide gel,^39–41 similar to that of ADAMTS13 purified from plasma,^1–3 although the calculated molecular weight is only 150 kd.^3–5 The difference in molecular mass is attributed to glycosylation^4,39,42 and perhaps other posttranslational modifications.⁴ Recombinant ADAMTS13 secreted into the conditioned medium is resistant to endoH digestion,³⁹ but sensitive to N-glycanase F and sialidase,³⁹ suggesting the presence of a complex-type structure of N-linked and O-linked oligosaccharides. ADAMTS13 is localized to the endoplasmic reticulum and Golgi apparatus, consistent with the distribution of secretory proteins.⁴⁰ Despite structural homology to ADAMTS1, which interacts with the extracellular matrix via its spacer domain and thrombospondin type repeats,^43,44 ADAMTS13 expressed from transfected COS7 and RFL-6 cells does not bind detectably to either cell membrane or extracellular matrix.⁴⁰

The role of the propeptide in biosynthesis and secretion of ADAMTS13 has been investigated recently. The ADAMTS13 propeptide is unusually short (≈41 amino acid residues) compared with the typical ≈200 residues in other members of the ADAMTS or ADAM protease family.^45–47 This domain terminates in a typical propeptide cleavage sequence, RQRR⁷⁴. The furin consensus sequence is required for cleavage of ADAMTS13 propeptide,³⁹ since changing this sequence to KQDR results in secretion of pro-ADAMTS13 with intact propeptide attached.³⁹ For several related proteases, propeptide is required for folding and maintaining enzymatic latency during biosynthesis. However, deletion of the propeptide did not impair the secretion of active ADAMTS13 in transfected HeLa cells, suggesting that the ADAMTS13 propeptide is not required to maintain enzyme latency.³⁹

CLEAVAGE OF vWF BY ADAMTS13

The mature form of the ADAMTS13 protease starts with the metalloproteases domain that has three His residues (HEXXHXXGXXHD) that coordinate the essential Zn²⁺ or Ca²⁺ binding.⁴ The conserved Met249 that forms a Metturn and the residues forming a predicted Ca²⁺ coordination site (Glu83, Asp173, Cys281, Asp284) are also present in the metalloprotease domain.⁴ The binding sites for Ca²⁺ and Zn²⁺ are consistent with data showing that the enzyme is inhibited by their chelation.^40,41,48 At the concentration of 16 nM (three to four times higher than plasma concentration), metalloprotease domain alone does not cleave multimeric vWF or small peptidyl substrate GST-vWF73-H that contains 73 amino acid residues from the central A2 domain of vWF, flanked by a glutathione s-transferase (GST) at N-terminus and 6xHis epitope at C-terminus (Fig. 1).⁴⁹ However, with a prolonged incubation (16 to 24 hours), the metalloprotease domain does cleave GST-vWF73-H substrate at a site other than the Tyr-Met bond, generating a N-terminal fragment ≈3 kd smaller than expected,⁴⁹ suggesting the carboxyl-terminal domains are required for substrate recognition and specificity.

Figure 1.

Cleavage of von Willebrand factor (vWF) and vWF73 by ADAMTS13 and variants. (A) Schematic representation of the construct of full-length (FL) ADAMTS13 and carboxyl-terminal truncated ADAMTS13 variants. (B) Cleavage of FL vWF and GST-vWF73-H by ADAMTS13 and C-terminal truncated variants. delCUB, aal-1191; MDTCS, aal-685; MDTC, aal-554; MDT, aal-439; MD, aal-385; M, metalloprotease. (Adapted from Zheng X, Nishio K, Majerus EM, Sadler JE. Cleavage of von Willebrand factor requires the spacer domain of the metalloprotease ADAMTS13. J Biol Chem 2003;278:30136–30141 and Ai J, Smith P, Wang S, et al. The proximal carboxyl terminal domains of ADAMTS13 determine substrate specificity and are all required for cleavage of von Willebrand factor. J Biol Chem 2005;280:29428–29434.)

Remarkably, addition of one or several proximal carboxyl terminal domains (including the disintegrin domain, first TSP1 repeat, the Cys-rich domain, and spacer domain) to the metalloprotease domain of ADAMTS13 restores the specific cleavage of Tyr-Met bond of GST-vWF73-H substrate.⁴⁹ However, the proteolytic activity of the ADAMTS13 mutants toward both vWF and GST-vWF73-H (Fig. 1) or fluorescent resonance energy transfer (FRETS)-vWF73 is not fully restored until all the proximal carboxyl-terminal domains are added.⁴⁹ These data suggest that all of the proximal carboxyl-terminal domains participate in substrate recognition and are required for cleavage of vWF. Further addition of the TSP1 2 to 8 domains and two CUB domains does not increase proteolytic cleavage of vWF,^40,41,49 and GST-vWF73 or FRETS-vWF73.⁴⁹ Several strains of mice (BALB/c, C3H/He, C57BL/6, and DBA/2) lack the carboxyl-terminal domains from the seventh TSP1 repeat to the CUB domain.³³ These data support the notion that the middle and the distal carboxyl-terminal domains may be dispensable at least in vitro in humans and in vivo in mice.

ADAMTS13-vWF BINDING

Full-length ADAMTS13 can bind immobilized vWF with a stoichiometry of 1:2 (i.e., one ADAMTS13 molecule binds two vWF molecules), reaching an equilibrium in ≈2 hours, with a half-life of dissociation of ≈4 hours.^39,49 The estimated K_d for ADAMTS13 binding to vWF is ≈14 nM, comparable to K_m of 16 nM, determined independently.³⁹ ADAMTS13 truncated after the spacer domain binds vWF comparably to full-length ADAMTS13, with a K_d of ≈24 nM (Table 1). Removal of the spacer domain significantly impairs its binding to vWF,³⁹ suggesting that the spacer domain contributes substantially to ADAMTS13-vWF interaction. The metalloprotease domain at the highest concentration tested (≈220 nM) does not bind vWF detectably.³⁹

Table 1.

Binding of ADAMT13 and Various Mutants to vWF and GST-vWF73-H

	Kd (nm)
Constructs	vWF66	vWF7349
FL	14 ±1.2	4.6 ± 1.3
MDTCS	23 ±0.9	7 ± 1.9
DTCS	ND	13 ± 2.1
M	ND	ND
Dis	ND	489 ±42
TSP1	ND	136 ±4.0
CysR	ND	121 ±5.0
Spa	ND	108 ±14

Binding of FL ADAMTS13 and various ADAMTS13 mutants to the purified plasma vWF and GST-vWF73-H was determined by ELISA method and the values for K_d (s) were determined by fitting into a double-reciprocal version of the binding equation. MDTCS and DTCS indicate the ADAMTS13 mutant truncated after the spacer domain with and without the metalloprotease domain, respectively. M, D, TSP1, CysR, S were all tagged at the N-termini (except for M) by V5-His epitope derived from vector. M was tagged at the C-terminus by V5-His.

vWF, von Willebrand factor; FL, full-length; GST, glutathione s-transferase; MDTCS, XXX; ND, not determined; DTCS, XXX; M, metalloprotease; Dis, disintegrin domain; TSP1, thrombospondin type 1; CysR, Cys-rich domain; Spa, spacer domain; ELISA, enzyme-linked immunosorbent assay.

The binding interaction between ADAMTS13 and GST-vWF73-H has been determined recently. vWF73 (D1659-R1668) has been shown to be the minimal and essential peptide region of vWF being recognized and cleaved by ADAMTS13.⁵⁰ Full-length ADAMTS13 and ADAMTS13 truncated after the spacer domain bound GST-vWF73-H with K_d (s) of ≈4.6 and ≈7.0 nM, respectively.⁴⁹ The affinity of binding ADAMTS13 or the carboxyl-terminal truncated mutant to vWF73 is approximately three times higher than that to vWF,^39,49 suggesting that the domains surrounding the central A2 subunit of vWF negatively regulate the interaction of ADAMTS13 with vWF.

The relative contribution of each of the proximal carboxyl-terminal domains in substrate binding has also been determined using recombinant domains of ADAMTS13 expressed from Escherichia coli and re-folded in vitro. The binding of ADAMTS13 domains to GST-vWF73-H was performed on an enzyme-linked immunosorbent assay (ELISA) plate and in solution. The K_d values for the disintegrin domain (D), the first TSP1 repeat (T), the Cys-rich domain (C), and the spacer domain (S) or DTCS fragment to bind GST-vWF73-H are ≈489, ≈136, ≈121, and ≈108 or 13 nM, respectively (Table 1).⁴⁹ The metalloprotease domain does not bind detectably at the concentrations tested. These data suggest that each of the proximal carboxyl-terminal domains participates in substrate recognition and acts cooperatively to confer efficient binding, consistent with the proteolysis of vWF and vWF73 by various ADAMTS13 mutants described above.⁴⁹

REGULATION OF ADAMTS13 ACTIVITY

Cleavage of vWF by ADAMTS13 may be modulated by flow shear stress,^9,10,51 structural elements of vWF,⁴⁸ heparin sulfate, platelets or platelet glycoprotein-1bα (GP1bα),⁴⁸ sodium chloride,⁵² and inflammatory cytokines.⁵³ vWF is not cleaved under physiological conditions due to structural inaccessibility of the binding or cleavage site on the vWF molecule by ADAMTS13. Modeling of the vWF-A2 domain suggests that the Tyr1605-Met1606 bond lies buried within the core β sheet of the native structure.⁵⁴ Shear force and denaturing reagents, or binding of platelets that expose the ADAMTS13 binding and cleavage site on vWF, significantly enhance the cleavage of vWF by ADAMTS13.^8–10 However, even in the presence of denaturing reagents, the ADAMTS13 binding site appears not fully exposed. For example, the rate of cleavage of the A2A3 mutant is ≈10-fold higher than that of the A1A2A3 mutant,⁴⁸ suggesting that the A1 domain may block the access of ADAMTS13 to the cleavage site. This blockage can be removed by interaction with platelet GP1bα or unfractionated heparin.⁴⁸

Other factors in the milieu may regulate ADAMTS13-vWF interaction, such as chloride ions⁵² and hemoglobin.⁵⁵ The cleavage of vWF in the absence of NaCl at pH 8.0 occurs with apparent k_cat/k_m ≈3.4 ×10⁴ M ⁻¹s ⁻¹, but this value decreases ≈10-fold in the presence of 0.15 M NaCl.⁵² The inhibitory effect can be found with a relative potency of ClO^–₄ > Cl^– >F^–.⁵² Inhibition of ADAMTS13 activity has been observed by hemoglobin at concentrations exceeding 2 g/L, which occurred during severe hemolysis in a fatal case of congenital TTP.⁵⁵ Unlike many other members of ADAMTS family that form SDS-stable complexes with α₂-macroglobulin, a nonspecific protease inhibitor present in the serum of healthy individuals, ADAMTS13 does not bind α₂-macroglobulin detectably.⁴⁰ Inflammatory cytokines such interleukin-6, but not interleukin-8 and tumor necrosis factor-α, inhibit the cleavage of ultralarge vWF (ULvWF) under flow.⁵³ The findings suggest a potential linkage between inflammation and thrombosis that may be of therapeutic importance.

Thrombin, factor Xa (FXa) and plasmin may regulate ADAMTS13 activity at the site of thrombus formation. ADAMTS13 is proteolytically cleaved following activation of coagulation in human plasma.⁵⁶ Incubation of purified recombinant ADAMTS13 with various concentrations of human thrombin (0.9 to 90 nM) results in degradation of ADAMTS13 in a time- and concentration-dependent fashion.⁵⁶ Such cleavage can be abrogated when thrombin binds soluble thrombomodulin, but not heparin, suggesting the involvement of thrombin exosite I, but not exosite II, in ADAMTS13 recognition. Plasmin and FXa also cleave ADAMTS13 and give rise to a similar but not identical fragmentation pattern, resulting in loss of ADAMTS13 activity.⁵⁶ Plasmin cleaves ADAMTS13 faster than do thrombin and FXa,⁵⁶ suggesting a role of ADAMTS13 and vWF in tissue repair.

MUTATIONS OF ADAMTS13

To date, more than 70 mutations on ADAMTS13 gene have been described in patients with congenital or familial TTP (Table 2).^5,42,57–61 The majority of these mutations are missense mutations (≈50%), followed by splice site mutations, silent mutations, nonsense, and frameshifts.^5,13,42 The mutations are distributed throughout the ADAMTS13 gene with no apparent clustering or hot spots. Twelve missense mutations involving cysteine residues have been identified in patients with familial TTP (Table 2), suggesting the importance of proper formation of disulfide bridges in ADAMTS13 function.

Table 2.

Location and Functional Consequence of ADAMTS13 Gene Mutation

Exon/Intron	Nucleotide	Effect on Amino Acid	Domain	Activity in Human Plasma (unless “in vitro” is noted)	Secretion	Comment
1	19C >T5,57	R7W	SP			SNP57
2	130C >T57	Q44X	proP	<3%80
3	237C >G100	I79M	M	<5%
3	286C >G5	H96D	M	2–7%
3	304C >T5	R102C	M	2–7%
In 3	5′border In3 G >A63	Splice	M	<3%
4	354G >A5	Silent	M
In 4	414 + 1G >A68	Splice abolished	M	<3%
5	420T >C568	Silent	M			SNP5,42
	546C >T	Silent	M
6	577C >T68	R193W	M	<3%	325% of wildtype68
6	582C >T5	Silent	M
6	587C >T5	T196I	M	2–7%
6	607T >C100	S203P	M	<5%
In 6	686 + 1 >A68 or	Splice	M	<3%
	686 + 4T >G5
7	695T >A	L232Q	M	2–6%70
7	703G >C58	D235H	M
7		A250V63	M	<3%	Decreased63
7	718–724Δ58	Del G + C	M
7	788C >G70	S263C	M	<2–5%
7	803G >C42	R268P	M	<3%42; <5%100; in vitro: no activity42	No42,68
8	932G >A58	C311Y	Dis
In 8	987 + 11C >T5 or+69C >T5	Splice	Dis
9	1058C >T58	P353L	Dis
In 9	1092 +67G >A5	Splice	Dis
10	1169G >A70	W390X	Tsp1–1
10	1170G >C101	W390C	Tsp1–1	<3%
10	1193G >A5	R398H	Tsp1–1	2–7%
In 10	1244 +2T >G68	Splice abolished	Tsp1–1	<3%
In 10	1245–32C >G5	Splice	Tsp1–1
12	1342C >G5	Q448E	Cys	<3%5,69 ; in vitro, full activity with↑ lower molecular weight vWF42	yes42	SNP5,42
12	1345C >T42	Q449X	Cys	<3%69; in vitro: very low activity42	yes42
12	1370CT58	P457L	Cys
	1423C >T5,42	P475S	Cys	<3%5,42,69; in vitro: very low activity, with ↑ lower MW vWF subunits42	yes42	SNP5,42
13	1520G >A	R507Q	Cys	<5%100
	1523G >A	C508Y	Cys	<3%42; In vitro: no activity42	No42,68
13	1582A >G	R528G	Cys	2–7%5
In 13	1584 +5G >A5, or +106C >G5, or + 236T >C5	Splice	Cys	2–7%
15	1716G >A5	Silent	Cys			SNP5,42
15	1783–1784 TT del71	Frameshift	Sp	<0.1 U/mL
In 15	1787–26G >A5	Splice	Sp
16	1787C >T100	A596V	Sp	<5%
16	1852C >G5	P618A	Sp			SNP57
16	1874G >A5	R625H	Sp			SNP5
17	2017A >T	I673F	Sp	<3%	No68
17	2074C >T5	R692C	Tsp1–2	2–7%
18	2195C >T5	A732V	Tsp1–2			SNP5,57
19	2272T >C100	C758R	Tsp1–3	<5%
19	2279delG58	Frameshift	Tsp1–3
19	2280 C >T68	Silent	Tsp1–3			SNP5,57
19	2376del26 (2376–2401Δ)5	Frameshift	Tsp1–3	2–7%
20	2549–2550delAT70	Frameshift	Tsp1–4
21	2699C >T5	A900V	Tsp1–5			SNP5
21	2723G >A68	C908Y	Tsp1–5	<3%	No68
21	2728C >T70	R910X	Tsp1–5
22	2851T >G5	C951G	Tsp1–5	2–7%
In 22	2861 +55C >T5	Splice	Tsp1–6
23	2910C >T5	Silent	Tsp1–6
24	3070T >G5	C1024G	Tsp1–7	2–7%
24	3097G >A5	A1033T	Tsp1–7			SNP5
24	3100A >T70	R1034X	Tsp1–7	<2%
24	3108G >A5	Silent	Tsp1–7
25	Del 3252–3253CT	R1096X	Tsp1–8	<5%100
25	C3367T68	R1123C	Tsp1–8	<3%	No68
26	3638G >A5	C1213Y	CUB1	2–7%5
27	3735G >A101	W1245X	CUB1	<3%
27	3769–3770insT5	Frameshift	CUB1	2–7%
28	4006C >T57	R1336W	CUB2
In 28	4077 +32T >C5	Splice	CUB2
	4143–4144insA60	Frameshift	CUB2	<10%; in vitro activity equals wildtype60	14% of wildtype60
29	4221 C >A60	Silent	CUB2			SNP5,42

^*The missense (change in amino acid residue), nonsense (introduction of premature stop), frameshift/deletion, alternatively splicing and silent (no amino acid change) mutations are indicated in column 3.

SNP, single nucleotide polymorphism; SP, signal peptide; ProP, propeptide; M, metalloprotease, Dis, disintegrin; Tsp, thrombospondin; Cys, cysteine-rich domain; Sp, spacer. The superscript numbers in columns 2 through 7 are reference citations.

Half of the patients with familial TTP experience their acute episodes during childhood and half during adulthood.⁶² In the latter patients, these episodes may be accompanied by pregnancy, infection, or other conditions that may result in the production of inciting factors.⁶² Although most of these mutations have been identified in patients with familial TTP, it is unclear if they also contribute to acquired TTP, which represents the majority of TTP cases.

Several missense mutations have been characterized by site-directed mutagenesis of the ADAMTS13 gene, followed by transfection into different expression systems, and assaying for their biosynthesis, secretion, and activity in the conditioned medium. Several recombinant mutants including R193W,⁵⁹ A250V, R268P, and C508Y⁴² exhibit low or no secretion and little activity in the conditioned medium of transfected cells. The R193W⁵⁹ and the A250V⁶³ mutations exhibited decreased secretion as compared with the wild type ADAMTS13. The R193W⁵⁹ is close to the active site of the metalloprotease domain, whereas the A250V is near the Met²⁴⁹ necessary for Zn²⁺ binding, suggesting that the small amount of protease secreted may also have compromised proteolytic activity. Similarly, the R268P substitution impairs secretion of the mutant protein from cells^42,59 perhaps due to the disruption of the α-helix structure, which may account for its diminished activity in plasma. On the other hand, the recombinant P475S is secreted normally in transfected cells, but shows low proteolytic activity toward vWF.⁴² Interestingly, approximately 9.6% of Japanese may be heterozygous for this polymorphism, which may be responsible for reduced protease activity.⁴² However, despite of a high prevalence of P475S in the Japanese population, such a mutation with low proteolytic activity is rare in whites (0.5%)⁶⁴ and in Chinese (1.7%).⁶⁵ Similarly, the nonsense mutation Q449X is also secreted normally in transfected cells, but has very low activity,⁴² consistent with reports that the spacer domain is necessary for ADAMTS13 activity.^40,41,49,66

Another recombinant missense, Q448E, which is secreted normally, has been reported as a single nucleotide polymorphism (SNP),^5,42 although it retains full activity in transfected cells.⁴² Although numerous SNPs have been reported (Table 2), none have yet shown a clear relationship to TTP,²⁸ although it is possible they may be associated with other conditions, such as coronary artery disease, in which ADAMTS13 activity is decreased.⁶⁷

Several frameshift mutations have been identified, and these aberrantly carboxyl-terminal truncated mutants may demonstrate either impaired secretion and/or low proteolytic activity. The 4143 to 4144insA mutation that deletes the second CUB domain exhibits impaired secretion, but retains proteolytic activity if the same amount of protein is tested.⁶⁰ Several ADAMTS13 gene mutations have been identified at splice sites, which disrupt the conventional GT-AG consensus splice sequences, and prevent production of normally spliced transcripts.^5,68 Such alternatively spliced forms of ADAMTS13 are present in normal human plasma,² although the biological functions of these variants are not known.

The in vitro expression studies indicate that ADAMTS13 mutations exhibit a heterogeneous phenotype, depending on the effects on biosynthesis, secretion, and protease bioactivity. These reflect the clinical heterogeneity seen among hereditary TTP patients, most of whom harbor compound heterozygous mutations. The parents of hereditary TTP patients (asymptomatic carriers) have about a 50% level of ADAMTS13 activity, although some asymptomatic carriers exhibit very low levels of ADAMTS13 activity (less than 10% of normal) who are compound heterozygotes such as R268P/P475S.⁶⁹ Of note, most of the described homozygous mutations, such as R692C,⁵ Q449X,⁴² 414 + 1G >A splice,⁶⁸ L232Q,⁷⁰ and 1783delTT,⁷¹ have some evidence of a consanguineous pedigree, and there are very few null alleles, suggesting that a complete lack of ADAMTS13 protease may be lethal.^{5,13,28,42,59}

DIAGNOSTIC UTILITY OF ADAMTS13 ASSAYS

ADAMTS13 activity in normal plasma is reported to be 50 to 178%.¹⁴ Severe deficiency of plasma ADAMTS13 activity toward vWF has been considered to be the primary cause of congenital and acquired idiopathic TTP,^{5,10,11,72,73} although mildly to moderately reduced levels of ADAMTS13 have been observed in conditions such as inflammation, coronary artery disease,⁶⁷ pregnancy, liver disease,⁷⁴ disseminated intravascular coagulation,⁷⁵ sepsis, and heparin-induced thrombocytopenia.⁷⁶ The clinical utility of the ADAMTS13 proteolytic activity assays remains the subject of considerable debate in recent years,^77–79 partly due to the variability of each assay. The assays can be categorized into direct versus indirect and static versus flow based, depending on the reaction conditions and how the cleavage of the substrate is determined.

Direct Assays

Direct assays focus on detection of cleavage products of a given substrate, whether it is a macromolecular vWF, a vWF A2 domain, or the small peptide GST-vWF-73 or FRETS-73. Agarose or polyacrylamide gel electrophoresis (PAGE), Western blotting, and FRETS techniques have been used to detect the cleavage products of the substrate. The FRETS-vWF73 detects proteolytic cleavage of the specific Tyr-Met peptide bond by ADAMTS13 in real time. By far, this assay may be the most sensitive and promising of all.

SDS-AGAROSE GEL ELECTROPHORESIS AND WESTERN BLOTTING

This assay was initially developed by Furlan et al⁸ to measure ADAMTS13 activity in plasma. After purified vWF is incubated with citrated plasma samples for approximately 24 hours, agarose gel electrophoresis and Western blotting with peroxidase-conjugated anti-vWF antibody are used to determine the proteolytic degradation of purified vWF multimers. The relative amount of ADAMTS13 activity is calculated from the reference of serially diluted normal human plasma.

This assay laid a strong foundation for understanding the pathogenesis of congenital or familial TTP^72,80–82 and acquired TTP.⁸⁰ It has also provided a tool to monitor the proteolytic activity of each fraction during the subsequent successful purification of ADAMTS13 protease from normal human plasma.² However, this assay is very complex and requires technical expertise in handling agarose gel electrophoresis, transfer and Western blotting.

SDS-PAGE AND WESTERN BLOTTING

Tsai et al^9,10,83 initially developed this assay based on detection of dimeric vWF fragments of 176 and 140 kd on denatured, but unreduced SDS-PAGE and Western blotting. This assay differs from the SDS-agarose method developed above by Furlan et al^8,72,80 in these ways: (1) no barium chloride activation is required; (2) the incubation is relatively short (1 to 3 hours at 37°C); (3) it is easier to handle gels from SDS-PAGE than agarose-gel electrophoresis; and (4) there is direct visualization of the cleavage product, eliminating the possibility of nonspecific degradation of the multimers. This SDS-PAGE assay detected severe deficiency of ADAMTS13 activity (<5% of normal) in all 39 samples from 37 patients with idiopathic TTP.¹⁰ In addition, the inhibitory activity against ADAMTS13 protease was detected in 26 of 39 samples (67%).¹⁰ This assay is sensitive enough to differentiate patients with congenital TTP from asymptomatic carriers and healthy individuals, and allows for the precise mapping of the defective ADAMTS13 gene.⁵

Alternatively, the epitope-tagged vWF fragment (His-A2 domain or GST-vWF73) could be digested by plasma ADAMTS13 within 1 to 3 hours.^49,50,84 The cleavage of the Tyr-Met bond is determined by Western blotting with an anti-His⁸⁴ or anti-GST^49,50 antibody. The assay using the vWF A2 domain detected 13 of the 16 samples (81%) with an activity of <6%.⁸⁴ The GST-vWF73 substrate is not cleaved by plasma from patients with congenital or acquired TTP, but is cleaved by plasma from patients with hemolytic uremic syndrome, suggesting that GST-vWF73 is a specific substrate for ADAMTS13. However, truncated ADAMTS13 mutants after the Cys-rich domain may cleave GST-vWR73.⁴⁹ Therefore, normal proteolytic cleavage does not rule out the possibility of the functional abnormality of ADAMTS13 protein due to mutations or acquired autoantibodies that are directed against the middle and distal carboxyl terminus. Again, the cleavage of a small recombinant domain or a peptidyl substrate does not require denaturing reagents and is free of contaminating ADAMTS13 in the substrate.

FRETS ASSAY

FRETS-vWF73 is a chemically modified vWF73 derived from the central A2 domain of vWF (D1596-R1668) containing A2pr (Nma) and A2Pr (Dnp; Fig. 2). It is readily cleaved by normal human plasma, but not by ADAMTS13-deficient plasma.⁸⁵ It could differentiate congenital and acquired TTP from patients with hemolytic uremic syndrome (HUS) and heterozygous mutations of ADAMTS13 gene or healthy individuals.⁸⁵ The proteolytic cleavage of the FRETS-vWF73 is completely blocked by ethylenediaminetetraacetic acid.⁸⁵ Compared with the collagen-binding assay (described below) and the GST-vWF73 assay (described above), FRETS-vWF73 is superior. It is sensitive and specific, requiring only 2 to 5 μL of citrated plasma for each activity assay, and the cleavage of the substrate can be monitored in real time with a result in ≈1 hour.⁸⁵ The FRETS-vWF73 can identify more TTP patients with severe ADAMTS13 deficiency and positive inhibitor, which can be verified independently by Technozyme-ELISA (Baxter, Orth, Austria) and immunoprecipitation and Western blotting methods (S. Shelat, P. Smith, J. Ai, and X. Zheng, unpublished data, 2005).

Figure 2.

Schematic representation of fluorescent resonance energy transfer (FRETS)-von Willebrand factor (vWF) 73 substrate. The amino acid residues Q1599 and N1610 within the vWF peptide (D1595-R1668) are replaced by A2pr (Nma) and A2pr (Dnp), respectively. When the Nma group is excited at 340 nm, fluorescence resonance energy is transferred to the neighboring quencher, Dnp. If the Tyr-Met peptide bond (indicated by arrow) is cleaved by ADAMTS13, the energy transfer quenching the fluorescence does not occur, allowing the emission of fluorescence at 440 nm from Nma (Adapted from Kokame K, Nobe Y, Kokubo Y, et al. FRETS-VWF73, a first fluorogenic substrate for ADAMTS13 assay. Br J Haematol 2005;129:93–100.)

Indirect Assays

Indirect assays depend on measuring the residual substrate or disappearance of the substrates. This type of assay may be influenced by non-specific proteolysis of vWF by other enzymes in plasma, unless they are inactivated by incubation of plasma with Pefobloc^73,80,81,86 or other serine protease inhibitors.^17,87

COLLAGEN-BINDING ASSAY

Initially described by Gerritsen et al⁸⁸ and modified by many others,^17,74,87,89 this assay is based on preferential binding of large vWF multimers to type III human collagen. After incubation of purified vWF or ADAMTS13-depleted plasma with patient’s plasma or normal human plasma (as control) in the presence of 5 mM Tris-HCl (pH 8.0), 1.5 M urea, and 10 mM BaCl₂ for 16 to 24 hours, the amount of residual vWF is determined by binding to human collagen type III. Bound vWF is quantified by peroxidase-conjugated anti-vWF immunoglobulin G (IgG) on an ELISA plate, followed by a densitometry measurement.^17,88–90

This assay is relatively straightforward to perform and can accommodate many samples simultaneously. It is quite robust in the detection of severe deficiency of ADAMTS13 (16 of 20 patients) and high titer inhibitor (seven of 16 patients) in idiopathic TTP.¹⁷ It correlates with other indirect assays well.⁹¹ The presence of inhibitor detected by this assay correlates with more relapses,^17,87 whereas patients with severe ADAMTS13 deficiency but no detectable inhibitor have a better response and outcome,¹⁷ suggesting that the assay may be used for making a diagnosis, monitoring therapy, and predicting outcome.

RISTOCETIN-INDUCED AGGREGATION

After digestion of the purified vWF in the buffer (similar to that in collagen-binding assay), the residual vWF is determined by a ristocetin cofactor activity in a platelet aggregometer. The method is reproducible as shown by low intra- and interassay coefficients of variation (2.8% and 7.5% for normal samples, respectively, and 8.7% and 12.9% for abnormal samples, respectively).⁹² Furthermore, it detects severe ADAMTS13 deficiency in patients with acute, classic TTP. The majority of patients with low titer inhibitor respond to plasma exchange treatment with increases in ADAMTS13 activity, whereas ADAMTS13 deficiency persists in some patients with high titer inhibitor despite clinical remission.⁹² This assay has not been widely used because only limited samples can be analyzed at one time on the aggregometer.

FLOW-BASED ASSAYS

Two types of assays use flow conditions for measuring ADAMTS13 activity. First, Shenkman et al⁹³ used a cone and plate(let) analyzer to evaluate the ability of TTP plasma to increase deposition on a polystyrene surface from normal plasma under flow conditions. This assay is sensitive and specific for detection of ADAMTS13 deficiency and inhibitors.⁹³ Increased deposition of normal platelets was observed following the addition of plasma in all 15 acute TTP patients, but only in three of 14 patients in remission.⁹³ The IgG from plasma or serum of three of five acute TTP patients enhanced platelet deposition,⁹³ suggesting the possibility of detecting autoantibodies against ADAMTS13 by this method. Second, Dong et al⁹⁴ described that ULvWF multimers, upon release from cultured endothelial cells, form extremely long, platelet-decorated string-like structures on the surface of endothelial cells under fluid shear stress. When perfused with normal human plasma, the string-like structures disappear rapidly within a few minutes, indicating the cleavage of ULvWF by ADAMTS13. Under flow conditions, cleaving ULvWF occurs at 1000-fold faster kinetics than that observed under static conditions. However, plasma from patients with idiopathic TTP does not cleave these platelet-decorated strings.

ENZYME-LINKED IMMUNOSORBENT ASSAY

This assay uses epitope-tagged recombinant fragment (vWF A2 domain or vWF73) as a substrate.^84,95 The substrate is immobilized to a microtiter plate with an antibody to one tag. After incubation of the immobilized substrate with plasma samples, an antibody to another tag detects the residual substrate. The proteolytic activity of ADAMTS13 is inversely proportional to the residual substrate concentration. This type of assay is relatively simple and sensitive. For example, the ADAMTS13 activity in patients with TTP detected by vWF A2 domain-ELISA is 2.4 ± 0.7%, compared with 40 ± 4.2% in normal human plasma.⁸⁴ The vWF73-ELISA can differentiate TTP patients from carriers of ADAMTS13 gene mutation and normal healthy individuals.⁹⁵

The ADAMTS13 Inhibitor Assay

ADAMTS13 autoantibodies may be classified into two groups: inhibitory antibodies or noninhibitory antibodies, depending on whether the autoantibodies block proteolytic cleavage of vWF (or other peptidyl substrate) in an in vitro assay. Antibodies that react proximally to ADAMTS13 may block proteolytic activity in vitro, whereas antibodies that target to the middle and distal carboxyl terminus of ADAMTS13 may not, based on the structure-function analysis of ADAMTS13.^40,41 Thus far, all inhibitory antibodies react with the Cys-rich and spacer domains with some specificity toward the catalytic domain and other parts of the molecule,⁹⁶ supporting the critical function of the Cys-rich and spacer domains in substrate recognition.^40,41,49

All of the functional assays for ADAMTS13 activity are capable of detecting the inhibitory autoantibodies with a wide range of sensitivities. The FRETS-vWF73 appears, by far, the most sensitive test among three assays we evaluated (collagen-binding assay, GST-vWF73, and FRETS-vWF73) to identify the inhibitory autoantibodies (S. Shelat, P. Smith, J. Ai, and X. Zheng, unpublished observation, 2005). An ELISA assay along with Western blotting using the recombinant ADAMTS13 as an antigen has been developed recently, which can detect both inhibitory and noninhibitory autoantibodies.⁹⁷ IgG antibodies are found in 97% of untreated patients with acute acquired TTP whose plasma levels of ADAMTS13 are below 10%.⁹⁷ The corresponding prevalence of IgM antibodies is 11%.

The presence of ADAMTS13 autoantibodies are rather specific for making a diagnosis of acquired TTP.^10,78 In addition, a high titer of inhibitory autoantibodies correlates with more relapses. Adjunct immune therapies such as rituximab, an anti-CD20 chimeric monoclonal antibody, or cyclophosphamide, may be considered once inhibitory autoantibodies are identified in patients with acute TTP who do not adequately respond to plasma exchange or are chronically relapsed.^17,87,98,99 Therefore, a robust ADAMTS13 inhibitor assay is critical for understanding of the mechanism of TTP and for tailoring therapy.

In conclusion, molecular cloning and characterization of ADAMTS13 gene and protein structure have opened a new avenue for study of the biology and biochemistry of the ADAMTS13 protease. Development of a sensitive and specific assay for ADAMTS13 activity and inhibitor would not only help to understand the pathogenesis of TTP and other thrombotic disorders, but also to facilitate a more timely and accurate clinical diagnosis, which is crucial for initiating and tailoring therapy in patients with TTP.