P.D187H Mutation Impairs ADAMTS13 Activity and Secretion and Contributes to Thrombotic Thrombocytopenic Purpura in Mice

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Lirias Journal of Thrombosis and Haemostasis, 13: 283–292 DOI: 10.1111/jth.12804

ORIGINAL ARTICLE

The novel ADAMTS13-p.D187H mutation impairs ADAMTS13 activity and secretion and contributes to thrombotic thrombocytopenic purpura in mice

E. DE COCK,* C. HERMANS,† J. DE RAEYMAECKER,‡ K. DE CEUNYNCK,* B. DE MAEYER,* N. VANDEPUTTE,* A. VANDENBULCKE,* H. DECKMYN,* H. ROTTENSTEINER,§ M. DE MAEYER,‡ S. F. DE MEYER* andK. VANHOORELBEKE* *Laboratory for Thrombosis Research, KU Leuven Kulak, Kortrijk; †Division of Haematology, Haemostasis and Thrombosis Unit, Saint-Luc University Hospital, Brussels; ‡Biochemistry, Molecular and Structural Biology Section, Department of Chemistry, Laboratory of Biomolecular Modeling, KU Leuven, Leuven, Belgium; and §Baxter Innovations GmbH, Vienna, Austria

To cite this article: De Cock E, Hermans C, De Raeymaecker J, De Ceunynck K, De Maeyer B, Vandeputte N, Vandenbulcke A, Deckmyn H, Rottensteiner H, De Maeyer M, De Meyer SF, Vanhoorelbeke K. The novel ADAMTS13-p.D187H mutation impairs ADAMTS13 activity and secretion and contributes to thrombotic thrombocytopenic purpura in mice. J Thromb Haemost 2015; 13: 283–92. of ADAMTS13. The homozygous p.D187H mutation Summary. Background: Congenital thrombotic thrombo- down-regulated ADAMTS13 activity in vitro. Impaired cytopenic purpura (TTP) is characterized by mutations in proteolytic activity was linked to unstable Ca2+ binding the ADAMTS13 gene, which either impair protein secre- as visualized using a molecular dynamics simulation. In tion or influence ADAMTS13 (A Disintegrin-like And addition, the p.D187H mutation affects protein secretion Metalloprotease domain with ThromboSpondin type-1 in vitro.InAdamts13–/– mice, the homozygous p.D187H motif, member 13) activity. Phenotypic consequences of mutation reduced ADAMTS13 secretion and activity and these mutations have not yet been evaluated in animal contributed to TTP when these mice were triggered with models for TTP. Objectives: To identify the in vitro effect recombinant human von Willebrand factor. Conclusions: of a novel ADAMTS13 mutation and to investigate Our data indicate that the p.D187H mutation impairs whether this mutation induces TTP in vivo. Methods: All ADAMTS13 activity and secretion and is responsible for 29 ADAMTS13 exons with exon–intron boundaries of a TTP onset in mice. patient with pregnancy-onset TTP were sequenced. Wild- type and mutant ADAMTS13 proteins were both tran- Keywords: ADAMTS13 protein, human; ADAMTS13 siently and stably expressed in human embryonic kidney protein, mouse; animal model; mutation; thrombotic cells, and their activity was evaluated in vitro using fluor- thrombocytopenic purpura. escence resonance energy transfer and flow assays. Molec- ular dynamics simulations were performed to study Ca2+ stability. Adamts13–/– mice were hydrodynamically injected with wild-type and mutant expression plasmids Introduction and triggered with recombinant human von Willebrand When patients are deficient in the metalloprotease factor. Results: We identified a novel heterozygous ADAMTS13 (A Disintegrin-like And Metalloprotease c.559G>C mutation in exon 6 of the proposita’s ADAM- domain with ThromboSpondin type-1 motif, member 13), TS13 gene. This mutation resulted in a p.Asp187His sub- spontaneous microthrombi are formed leading to the rare stitution (p.D187H), which was located in the high but severe disorder thrombotic thrombocytopenic purpura affinity Ca2+-binding site in the metalloprotease domain (TTP) [1]. ADAMTS13 deficiency prevents trimming of ultra-large (UL) von Willebrand factor (VWF) multimers. Correspondence: Karen Vanhoorelbeke, Laboratory for Thrombosis These UL multimers spontaneously interact with platelets Research, IRF Life Sciences, KU Leuven Kulak, Etienne Sabbelaan and form aggregates that obstruct blood vessels in multi- 53, B-8500 Kortrijk, Belgium. ple organs, resulting in microvascular thrombosis, severe Tel.: +32 56 24 60 61; fax: +32 56 24 69 97. thrombocytopenia, and hemolytic anemia [1,2]. In the E-mail: [email protected] majority of TTP patients, ADAMTS13 deficiency is Received 9 May 2014 acquired due to anti-ADAMTS13 antibodies either neu- Manuscript handled by: C. Gachet tralizing the enzyme’s activity or accelerating its clearance Final decision: P. H. Reitsma, 16 November 2014 [3,4]. Fewer patients have congenital TTP, or Upshaw–

Schulman syndrome, defined by genetic mutations and 28 with a platelet count of 48 000 lL1. Pathological polymorphisms within the ADAMTS13 gene [5,6]. AD- examination of the placenta revealed signs of chronic AMTS13 deficiency alone might not provoke TTP, as anoxia. Postpartum, the platelet count decreased to some patients remain asymptomatic until early adulthood. 14 000 lL1 with clear signs of hemolysis and microangi- In these patients, a second hit or trigger such as acute opathy (schistocytes 4%, elevated lactate dehydrogenase, infections or pregnancy causes the onset of TTP [7]. low haptoglobin). Platelets then normalized within ADAMTS13 is located on chromosome 9q34 and con- 10 days with steroids alone. sists of 29 exons [8,9]. More than 150 different mutations When she was tested in 2006, her platelet count was (missense, nonsense, and splice site mutations; small dele- normal. Complete thrombophilia workup was performed tions and insertions) randomly distributed throughout the but did not reveal any acquired or congenital abnormality ADAMTS13 gene have been reported so far [10]. To date, with the exception of a slight increase in factor VIII approximately two-thirds of the congenital TTP patients (FVIII; 166%) and VWF (160% for the antigen and are compound heterozygous, while the remaining patients 168% for the activity measured by collagen binding bear homozygous ADAMTS13 mutations. In vitro char- assay) levels. Because of her history of microangiopathy, acterization of many of these ADAMTS13 mutations ADAMTS13 activity was measured (she was under remis- revealed that the majority impaired protein secretion, sion at that time) and found to be < 8% (TECHNO- although mutations that also influence ADAMTS13 activ- ZYMEâ ADAMTS13-activity ELISA; Technoclone, ity (e.g., p.Ser119Phe [11], p.Arg193Trp [12], p.Pro457Leu Vienna, Austria), and no anti-ADAMTS13 antibodies [13], and p.Arg1060Trp [14]) have been reported [5]. Phe- were detected in her plasma (mixing studies with normal notypic consequences of these mutations have never been plasma using the TECHNOZYMEâ ADAMTS13-activity tested in animal models. Such studies might nevertheless ELISA). There were no plasma samples available from provide valuable information on the contribution of the period of her pregnancy to test for ADAMTS13 anti- specific ADAMTS13 mutations to the pathophysiology gen and activity levels. From the family history it became of TTP. clear that her mother had three pregnancies, two of which In this study, the genetic defect in a pregnancy-onset were complicated by preeclampsia. There was, however, TTP patient was identified to correctly diagnose her with no history suspicion of microangiopathy in any other hereditary TTP. In addition, the expression and activity member of her family. Family screening for ADAMTS13 of the recombinant mutant protein were evaluated in deficiency has not yet been feasible. Because of the clini- vitro. Finally, we used a TTP mouse model in which cal expression and ADAMTS13 phenotype, the patient Adamts13–/– mice were triggered with recombinant was diagnosed with TTP. Mutation analysis of ADAM- human VWF (rVWF) [15] [16] to study the effect of the TS13 was performed in this study to confirm the diagno- ADAMTS13 mutant on the pathophysiology of TTP. sis of hereditary TTP. Informed consent was obtained in accordance with the Declaration of Helsinki. Patient, materials, and methods Identification of the mutation Case history Genomic DNA was extracted from the peripheral leuko- A now 41-year-old woman was initially referred to the cytes from the patient using the salting out procedure as hospital in 2006 for hematological assessment following described [17]. All 29 ADAMTS13 exons with exon– two complicated pregnancies. Her first pregnancy at the intron boundaries were amplified by polymerase chain age of 31 was complicated by mild thrombocytopenia reaction with primers as described elsewhere [18]. Prod- detected at week 12 with severe intrauterine growth retar- ucts were sequenced in both directions, and sequencing dation. HELPP (hemolysis, elevated liver enzymes, and a reactions were outsourced (GATC Biotech AG, Kon- low platelet count) syndrome was suspected, and medical stanz, Germany). pregnancy termination was performed at week 17. During her second pregnancy at the age of 34, abnormal uterine Expression of recombinant wild-type and p.D187H artery Doppler waveforms were found at week 26. She ADAMTS13 was admitted at week 28 for fetal monitoring. Fetal growth stagnation and mild thrombocytopenia (platelet The identified missense mutation p.D187H was intro- count of 108 000 lL1) with biological signs of hemolysis duced in the pcDNA6.1 ADAMTS13 expression vector and microangiopathy (elevated lactate dehydrogenase lev- using the QuickChangeTM XL Site Directed Mutagenesis els, reduced haptoglobin levels, but absence of schisto- Kit (Stratagene, La Jolla, CA, USA). Constructs were cytes) were confirmed. There was no disturbance of liver sequenced (GATC Biotech AG), and sufficient amounts enzymes, and proteinuria was absent. Treatment with of both ADAMTS13 wild-type (WT) and D187H DNA steroids was empirically initiated (methylprednisolone plasmids were obtained using the Qiagen Plasmid Mega 1mgkg1). Cesarean section was performed at week Kit (Qiagen, Venlo, The Netherlands). Human embryonic

© 2014 International Society on Thrombosis and Haemostasis ADAMTS13-p.D187H contributes to TTP in mice 285 kidney (HEK) 293T cells were stably transfected with relative to normal human pooled plasma (NHP) or nor- plasmid DNA encoding either WT or mutant D187H mal murine pooled plasma (NMP) (100%) [11,21]. ADAMTS13 using FuGENEâ HD transfection reagent To determine kinetic parameters of recombinant WT (Roche Applied Science, Penzberg, Germany), and cells and p.D187H ADAMTS13, varying concentrations of were selected on Blasticidin (Invitrogen, Carlsbad, CA, ADAMTS13 (ranging from 0.1 to 10 nmol L1) were USA). Recombinant proteins were produced and partially incubated with 2 lmol L1 FRETS-VWF73. The increase purified, and concentrations were determined in the AD- in fluorescence in function of time equals the rate of the AMTS13:Ag ELISA [19]. reaction. Concentrations for WT and p.D187H ADAM- Transient (co)transfections were performed to study TS13 resulting in measurable enzyme activities were used protein expression under heterozygous WT/D187H, to determine the catalytic efficiency (kcat/KM). Therefore, homozygous WT, or homozygous D187H conditions. a constant concentration of ADAMTS13 (1 nmol L1 Transfection efficiency was controlled by the addition of WT and 10 nmol L1 p.D187H) was incubated with 10 pmaxGFPâ (Lonza, San Diego, CA, USA) [20]. Briefly, different concentrations of FRETS-VWF73 (ranging from HEK293T cells were grown in Dulbecco’s modified Eagle 0.01 lmol L1 to 10 lmol L1). The rate of these 10 dif- medium (DMEM; ATCC, Manassas, VA, USA) with ferent reactions (dF/dt) was plotted in function of the 10% FCS and 1% P/S/F (Invitrogen) until 60–80% con- substrate concentration. dF/dt was converted to d[P]/dt fluency. Cells were transfected using jetPRIMETM trans- by determining the change in fluorescence which corre- fection reagent and a total of 4 lg plasmid DNA in sponds to complete turnover of the substrate. The curve TM jetPRIME reaction buffer (VWR International, Rad- was fitted to the Michaelis–Menten equation and Kcat/Km nor, PA, USA). Plasmid DNA was a mixture of WT/ was calculated using PRISM Version 5 software (GraphPad D187H/pmaxGFPâ DNA (1.8/1.8/0.4 lg), WT/ Software, San Diego, CA, USA) [11,21]. pmaxGFPâ DNA (3.6/0.4 lg), or D187H/pmaxGFPâ Cleavage of platelet-decorated VWF strings in flow was DNA (3.6/0.4 lg). Expression was done in DMEM con- evaluated as previously described [22]. Briefly, blood out- taining 1% P/S/F. Forty-eight hours after transfection, growth endothelial cells (BOECs) were stimulated with green fluorescent protein (GFP)-expressing cells were 25 lmol L1 histamine (Sigma) for 10 min and perfused visualized using a Nikon eclipse TE200 inverted fluores- with fluorescently labeled (100 nmol L1 3,30-dihexyloxa- cence microscope (Nikon Instruments, Melville, NY, carbocyanine iodide [Dioc6]; Invitrogen) washed platelets USA) to determine transfection efficiency. One hundred at a shear rate of 250 s1 for 60 s, followed by perfusion cells were randomly selected, and the number of fluores- with HBS with or without ADAMTS13 WT or p.D187H cent cells among these hundred cells was determined. (25 nmol L1) for an additional 360 s. Perfusion with Expression medium was harvested, and expression levels HBS alone and ADAMTS13 preincubated for 10 min of WT and p.D187H ADAMTS13 were determined using with the inhibitory anti-ADAMTS13 antibody 3H9 the ADAMTS13:Ag ELISA [19]. (67 nmol L1) served as negative controls. BOEC- anchored platelet-decorated VWF strings were visualized using a Nikon eclipse TE200 inverted fluorescence micro- ADAMTS13 activity tests scope (Nikon instruments), equipped with a 920 objective Activities of ADAMTS13 WT and p.D187H were evalu- and connected to a Hamamatsu CCD camera (ORCAâ- ated using the fluorescent resonance energy transfer R2; Hamamatsu Photonics, Hamamatsu city, Japan). Vid- (FRET) assay with the fluorescent FRETS-VWF73 sub- eos were recorded using HCIMAGE software (Hamamatsu strate (Peptides International; Louisville, KY, USA) as Photonics). Removal of all VWF strings in one view field described. Briefly, recombinant ADAMTS13 (0.5 nmol was followed in function of time and cleavage events, L1) or transgene mADAMTS13 present in plasma which represent ADAMTS13 activity, were counted. (30 lL) were incubated with 2 lmol L1 FRETS-VWF73 Experimental values were compared with the average in a HEPES-buffered saline solution (50 mmol L1 HE- activity level obtained, using WT ADAMTS13, which 1 1 PES, 5 mmol L CaCl2,1lmol L ZnCl2, 150 mmol was set at 100% (n = 5). L1 NaCl, pH 7.4 [HBS]), containing 1 mg mL1 bovine serum albumin (Sigma, St Louis, MO, USA). Digestion Modeling of the mutated ADAMTS13 Ca2+-binding site of FRETS-VWF73 by recombinant WT or p.D187H AD- AMTS13 or transgene plasma mADAMTS13 generated a An in-depth structural analysis of the ADAMTS13 fluorescent signal that was measured using the FLUOstar p.D187H mutation was performed starting with the AD- OPTIMA reader (BMG Labtech GmbH, Offenburg, Ger- AMTS13 homology model we published previously [11]. many). One hundred cycles of 150 s or 40 cycles of 300 s A high-affinity Ca2+-binding site involving D182 or were measured using excitation at 355 nm and emission E184, D187, and E212 was identified by Gardner et al. at 460 nm. Fluorescent intensities were depicted in func- [23]. Based on these Ca2+-coordinating amino acids, a tion of time. The slope of the resulting curve was used as homologous X-ray crystal structure was found in AD- a measure for enzyme activity. Activity was represented AMTS5 (PDB: 3HY7), containing a Ca2+ ion bound in a

© 2014 International Society on Thrombosis and Haemostasis 286 E. De Cock et al cavity similar to the predicted site in ADAMTS13 [24]. rate 24 hours after rVWF injection and analyzed for We optimized the three-dimensional structure of the platelet counts using a Hemavet hematology system ADAMTS13 binding site (residues I178–V195 and T211– (Drew Scientific Inc., Dallas, TX, USA). Plasma was L218) using the conformational search module in the taken to determine mADAMTS13 antigen and activity Molecular Operating Environment (MOE) package ver- levels by an in-house developed mADAMTS13:Ag sand- sion 2012.10 (Chemical Computing Group Inc., Montreal, wich ELISA and FRETS-VWF73 assay, respectively. All Quebec, Canada). This generated 1600 different confor- animal experiments were performed in accordance with mations, grouped in 57 clusters based on root-mean- protocols approved by the Institutional Animal Care and square deviation after fitting and visually inspected to Use Committee of the KU Leuven (Belgium). select a putative ADAMTS13 Ca2+-binding model. The ADAMTS13 p.D187H mutation was introduced with the mADAMTS13:Ag ELISA mutagenesis module of the PYMOL modeling software package version 1.5.0.4 (PyMOL Molecular Graphics Sys- A 96-well microtiter plate was coated overnight with the tem, Schrodinger,€ LLC). A 2-ns molecular dynamics in-house developed murine anti-mADAMTS13 monoclo- (MD) simulation was performed with the GROMACS nal antibody 20A10 at 5 lgmL1 in PBS. After blocking, 4.6.5 package to test the stability of the homology models plasma was diluted in PBS, 0.3% (m/v) skimmed milk, in both ADAMTS13 WT and the p.D187H mutant [25]. and the plate was incubated at 37 °C for 1 h. An in- house developed polyclonal rabbit anti-mADAMTS13 antibody was diluted in PBS, 0.3% (m/v) skimmed milk Construction of a murine p.D187H ADAMTS13 expression to a final concentration of 5 lgmL 1 and incubated for vector 1 hour. Bound antibody was detected with horseradish cDNA encoding WT mADAMTS13 was amplified from peroxidase–labeled goat anti-rabbit antibody (Jackson pcDNA6.1-mADAMTS13-WT [26] using Phusionâ High Immunoresearch Laboratories Inc., West Grove, PA, Fidelity DNA polymerase (New England BioLabs, Ips- USA) (1:25 000 in dilution buffer). Antibody binding was wich, MA, USA) and primers flanking the desired region. detected as described here earlier. NMP was used as a ref- The mADAMTS13 polymerase chain reaction product erence. was ligated into the pBS-II-SK-HCRHPi expression vector containing a liver specific promoter (HCRHPi: human Statistical analysis a1-anti–trypsin promoter, and a truncated 1.4-kb human FIX intron A [27]) using T4 DNA ligase (Roche Applied Data are given as mean SD. All data were analyzed Science) resulting in the pBS-II-SK-HCRHPi-mADA- using PRISM Version 5 software (GraphPad Software). MTS13 WT construct. The p.D187H missense mutation Mean values were compared using the one-way or was introduced into the pBS-II-SK-HCRHPi mADA- two-way ANOVA followed by the Bonferroni multiple com- MTS13 WT expression vector using the QuickchangeTM II parison post-test. A probability of < 0.05 was taken as XL Site Directed Mutagenesis Kit (Stratagene). Both con- significant for rejection of null hypothesis. structs were sequenced (GATC Biotech AG), and sufficient amounts of both mADAMTS13 WT and D187H Results DNA plasmids were obtained using the Qiagen Plasmid Mega Kit (Qiagen). Heterozygous mutation p.D187H identified in the ADAMTS13 metalloprotease domain Murine TTP model In 2014, when the patient was in remission, her ADAM- Adamts13–/– mice [28] (4–8 weeks, weighing 15–20 g) were TS13 antigen and activity were assayed again with our in- hydrodynamically injected with a total of 50 lg expres- house developed ADAMTS13 antigen assay [19] and the sion plasmids [27], either pBS-II-SK-HCRHPi mADA- FRETS-VWF73 assay, respectively. Her ADAMTS13 MTS13 WT, pBS-II-SK-HCRHPi mADAMTS13 D187H, antigen levels were severely reduced (17 1%, n = 3, or an equal mixture (i.e. 2 9 25 lg) of both, to obtain compared with NHP), and her ADAMTS13 activity was homozygous or heterozygous in vivo expression of WT below the detection limit of the assay (< 0.5%, n = 3 and p.D187H mADAMTS13. Three days later, TTP was compared with NHP). We next studied the presence of a induced in these mice by injecting them with a single possible genetic defect in ADAMTS13. Sequence analysis bolus injection of 2000 rVWF ristocetin cofactor units of all 29 exons of her ADAMTS13 gene revealed the pres- (rVWF:RCoU) per kg body weight of rVWF essentially ence of a heterozygous c.559G>C mutation in exon 6. as described [15] but with rVWF injected retro-orbitally. This missense mutation changes D 187 to H (p.D187H) As controls, Adamts13–/– mice were hydrodynamically in the metalloprotease domain of ADAMTS13. Interest- injected with 0.9% NaCl followed by rVWF or PBS injec- ingly, D187 is part of the high affinity Ca2+-binding site, tion. Blood samples were collected on 3.8% trisodium cit- which is situated close to the active site cleft [23].

Ca2+, thereby confirming previously published data on The p.D187H mutation attenuates ADAMTS13 activity in the importance of D187 in Ca2+ binding. vitro and induces unstable Ca2+ binding The p.D187H mutation reduces the catalytic efficiency of The p.D187H mutation impairs secretion in vitro ADAMTS13 To determine the catalytic efficiency (kcat/ Km) of p.D187H and compare it with that of WT AD- The effect of the p.D187H mutation on protein secretion AMTS13, a FRETS-VWF73 kinetic assay was performed. was tested in vitro by performing transient transfections Analysis of product turnover in function of substrate con- of heterologous HEK293T cells with both WT and centration revealed comparable Km values for p.D187H p.D187H ADAMTS13 expression plasmids. A GFP- and WT ADAMTS13 (0.93 0.76 and 0.92 0.77 lmol expressing plasmid was added in all experiments to deter- L 1, n = 4 respectively), indicating that binding affinity mine transfection efficiency, which was similar for both was similar. However, the kcat for p.D187H was severely the homozygous WT (32.6 5.9%, n = 3) and p.D187H reduced compared with WT ADAMTS13 (0.03 (26.1 1.4%, n = 3, P > 0.05) conditions. Interestingly, 0.03 and 0.30 0.11 s 1, n = 4 respectively). Hence, the relative ADAMTS13 antigen levels were significantly catalytic efficiency (kcat/KM) of p.D187H was 11-fold reduced in the homozygous p.D187H (40.4 9.3%, lower (11.0 3.6, n = 4) than that of WT ADAMTS13, n = 3) compared with the WT condition (100 14.9%, demonstrating that the mutation negatively affects n = 3, P < 0.01) (Fig. 2A), implying that p.D187H nega- substrate turnover and/or product release. tively influences protein secretion. In line with the data obtained with FRETS and flow assays together with the The p.D187H mutation impairs proteolysis of VWF strings reduced expression levels of p.D187H, the activity of the under flow conditions The activity of p.D187H ADAM- transiently expressed homozygous p.D187H enzyme TS13 was also studied in a more physiological assay in (20.2 0.7%, n = 3) was significantly reduced compared which the disappearance of platelet-decorated VWF with that of WT (100 4.6%, n = 3, P < 0.001, Fig. 2B). strings anchored to stimulated endothelial cells under flow Because the patient is heterozygous for the mutation, we conditions was followed in function of time. A decreased also investigated expression levels of WT/p.D187H pro- activity of p.D187H ADAMTS13 (54.4 9.1%, n = 5) teins. Transfection efficiency of heterozygous WT/ compared with WT ADAMTS13 (100.0 7.1%, n = 5, p.D187H ADAMTS13 plasmids (24.6 2.5%, n = 3) with the mean value of WT ADAMTS13 activity set as was also similar (P > 0.05) to those of homozygous WT 100%) was observed (Fig. 1A). Background activity, eval- or p.D187H plasmids. Interestingly, heterozygous WT/ uated by adding the inhibitory anti-ADAMTS13 antibody p.D187H expression levels (53.8 15.5%, n = 3) were 3H9 to WT ADAMTS13 (23.1 5.7%, n = 5) and using significantly lower than those of the WT condition HBS alone (17.8 7.5%, n = 5) was similar as described (P < 0.01) but comparable with those of p.D187H previously [22]. (P > 0.05, Fig. 2A). As expected, the activity of heterozygous WT/p.D187H ADAMTS13 (45.8 2.7%, n = 3) The p.D187H mutation induces unstable binding of was also significantly lower than that of WT (P < 0.001, Ca2+ Because it was previously described that D187 Fig. 2B). Hence, transient transfection experiments show plays an important role in Ca2+ binding [23], a homology that ADAMTS13 expression (53.8 15.5%) and activity model of ADAMTS13 was made to study the influence of (45.8 2.7%) levels of the heterozygous WT/p.D187H the p.D187H mutation on Ca2+ binding. Our constructed condition did not entirely mimic the patients’ situation ADAMTS13 Ca2+-binding model indicated that apart where only 8% ADAMS13 activity was measured. These from residues D182, E184, H187, and E212 [23], a water data suggest that another unidentified mutation might be molecule might be essential for proper Ca2+ binding as present in, for example, the promoter region of the AD- well (Fig. 1B). MD simulations (2 ns) for WT ADAM- AMTS13 gene of our proposita. TS13 suggested that the predicted Ca2+-binding site was stable as the position of the Ca2+ ion coordinated by the WT but not p.D187H ADAMTS13 rescues the TTP negatively charged amino acids and the water molecule – – phenotype of Adamts13 / mice remained unchanged during the entire simulation (Fig. 1B and Video S1). Using p.D187H ADAMTS13, however, We finally investigated whether the homozygous p.D187H MD simulations showed that the Ca2+-binding site mutation could induce TTP in vivo by making Adamts13–/ deformed due to extensive movement of the loop contain- – mice transgenic for the p.D187H mutant using the tech- ing the p.D187H mutation, ultimately leading to a dis- nique of hydrodynamic injection [27] and triggering the placement of the bound water molecule and considerable mice with rVWF [15]. The heterozygous condition was not fluctuations in the Ca2+ ion position compared with the investigated in vivo as our transient transfection experi- WT ADAMTS13 Ca2+-binding model (Fig. 1B and ments showed that this situation did not entirely mimic Video S1). Hence, these data suggest that the p.D187H the patients’ condition. First, plasmids were generated mutation gives rise to a decreased binding affinity for expressing WT and p.D187H mADAMTS13 under

*** A

100

***

50 % ADAMTS13 activity

0 WT D187H HBS WT+3H9

B WT D187H

E184 E184

E212 E212 H187 D187 Start D182 D182

E184 E184

E212 H187 Middle E212 D187 D182 D182

E184

E212 D182 H187 E212 End D187 D182

Fig. 1. The p.D187H mutation partially inhibits ADAMTS13 activity in vitro. (A) Stimulated endothelial cells were perfused with 3,30-dihexy- loxacarbocyanine iodide (DIOC)-labeled washed platelets at a shear rate of 250 s1 for 60 s. Resulting platelet-decorated von Willebrand factor strings were perfused with HEPES buffer (HBS) without (control) or with wild-type (WT) or p.D187H ADAMTS13 (25 nmol L–1) for an additional 360 s. Control reactions were performed with WT ADAMTS13 in the presence of the inhibitory anti-ADAMTS13 monoclonal antibody 3H9 (67 nmol L–1) or in the absence of ADAMTS13 (HBS buffer). Mean of WT data was set as 100%. Data are represented as mean SD 2+ (n = 5, ***P < 0.001, one-way ANOVA). (B) Molecular modeling of the ADAMTS13 metalloprotease domain shows unstable Ca binding in the p.D187H mutant. Representation of the ADAMTS13 metalloprotease domain homology model and, more specifically, the proposed high- affinity Ca2+-binding site. The Ca2+ (blue sphere)-binding site is formed by residues D182, E184, D187, and E212 (green sticks) and a water molecule (red/white sphere). Molecular dynamic simulation reveals a local unfolding due to electrostatic repulsion. Images at the start, middle, and end of the simulation are shown. Mutation of residue p.D187 to histidine leads to a deformed structure at the Ca2+-binding site (see middle and end images, and Video S1) and hence negatively influences the stability of Ca2+ binding.

AB** *** 150 150 ** ***

100 100

***

50 50 ADAMTS13 activity (%) ADAMTS13 antigen (%)

0 0 WT D187H WT + D187H WT D187H WT + D187H

Fig. 2. p.D187H influences protein secretion and activity in vitro. (A) Conditioned medium of transiently transfected HEK293T cells was analyzed with use of an ELISA. Antigen levels of cells expressing homozygous WT, homozygous p.D187H, or heterozygous p.D187H/WT AD- AMTS13 were measured. Mean of wild-type (WT) data was set as 100%. (B) Activity of the recombinant ADAMTS13 proteins (0.5 nmol L–1) was tested in a FRETS-VWF73 assay. Initial reaction rates were calculated and compared with NHP as a reference. Mean of WT data was set as 100%. Data represent mean SD (n ≥ 3). Significance levels are calculated using one-way ANOVA and represented as **P < 0.01 and ***P < 0.001. control of a liver-specific promoter (HCRHPi). Next, induce a TTP phenotype in vivo. In conclusion, the D187H and WT mADAMTS13 expression plasmids were p.D187H mutation contributes to TTP in mice under hydrodynamically injected in Adamts13–/– mice. In vivo homozygous conditions. expression of homozygous p.D187H mAdamts13 was significantly reduced (143 78%, n = 11, P < 0.001) com- Discussion pared with homozygous WT mAdamts13 (985 452%, n = 7) (Fig. 3A), confirming that the p.D187H mutation In this study, we investigated whether the ADAMTS13 indeed affects protein secretion. As expected, the p.D187H p.D187H mutation identified in a patient with pregnancy- activity (6.3 4.5%, n = 11 was significantly reduced onset TTP affects protein secretion and activity in vitro compared with WT (76.7 10.6%, n = 7, P < 0.001) and whether this mutation could account for a TTP phe- (Fig. 3B). notype in vivo. To the best of our knowledge, the Mice were subsequently triggered with 2000 U kg1 p.D187H mutation has never been associated with TTP. rVWF, and the degree of thrombocytopenia was mea- Recently, however, p.D187H was found as one of the rare sured, which was used as a parameter for the TTP pheno- coding SNPs in patients with deep vein thrombosis [29]. type [15]. Adamts13–/– mice expressing WT We show here that the p.D187H mutation reduced the mADAMTS13 and triggered with rVWF did not develop catalytic efficiency of ADAMTS13 by 11-fold. Our data TTP as their platelet counts (541 55 9 103 lL1, strongly corroborate with the finding that D187 was part n = 7, Fig. 3C, WT) did not differ from those of control of the high-affinity Ca2+ binding site of ADAMTS13 and Adamts13–/– mice that were not expressing ADAMTS13 crucial for enzyme activity as a D187A mutant had a 10- and were not triggered with rVWF (487 51 9 fold reduced kcat/KM compared with WT ADAMTS13 103 lL1, n = 8, Fig. 3C, C2). In contrast, Adamts13–/– [23,30]. The p.D187H mutant enzyme was, moreover, less hydrodynamically injected with 0.9% NaCl and triggered effective in cleaving full-length VWF under flow condi- with rVWF did develop TTP (189 80 9 103 lL1, tions. Further, molecular simulations demonstrated that n = 9, Fig. 3C, C1; P < 0.001 vs. WT and C2). Interest- the p.D187H mutation induces unstable Ca2+ binding, ingly however, mADAMTS13 p.D187H could not prevent which is in line with the reduced activity of the mutant the TTP phenotype as hydrodynamic injection of the cor- enzyme. The p.D187H mutation not only affected responding expression plasmid resulted in a significant ADAMTS13 activity but also negatively influenced pro- drop in platelet counts (309 79 9 103 lL1, n = 11, tein secretion as demonstrated by a severe reduction Fig. 3C, p.D187H) compared with control mice not trig- (60%) in p.D187H protein levels in the expression med- gered with rVWF (487 51 9 103 lL1, n = 8, ium of transiently transfected HEK293T cells compared P < 0.001, Fig. 3C, C2) and mice injected with WT with WT. Interestingly, however, transfection of the mADAMTS13 expression plasmid and triggered with p.D187H mutant under heterozygous conditions resulted rVWF (541 55 9 103 lL1, n = 7, Fig. 3C, WT). in ADAMTS13 antigen and activity levels of 53.8% and These data indicate that both the reduced expression of 45.8%, respectively, which was not in agreement with p.D187H (Fig. 3A) and its reduced activity (Fig. 3B) can the heterozygous p.D187H condition observed in our

AB 100 1500

1000 60

40 500

20 mADAMTS13 activity (%) ADAMTS13 antigen (%)

0 0 WT D187H WT D187H

*** C 800 *** *** ***

) 600 –1 L µ

** 3 400

Platelets (10 200

0 WT C1 D187H C2

DNA + – + –

0.9 % NaCI – + – +

rVWF + + + –

Fig. 3. p.D187H causes thrombotic thrombocytopenic purpura symptoms in vivo. Adamts13–/– mice were hydrodynamically injected with mADAMTS13 WT (WT, n = 7) or D187H (D187H, n = 11) expression plasmids or 0.9% NaCl (C1, n = 9 and C2, n = 8). Three days later, mice received either 2000 U kg1 rVWF or PBS via retro-orbital injection. One day later (or at day 4 after hydrodynamic injection) blood was taken retro-orbitally on EDTA for blood cell count and on sodium citrate for plasma preparation. (A) mADAMTS13:Ag levels were determined in plasma samples using an in-house developed ELISA. NMP was set as 100%. Data are represented as mean SEM. (B) Activity of mADAMTS13 in plasma samples was determined using FRETS-VWF73 as a substrate. The initial reaction rates were calculated and compared with NMP as a reference, which was set as 100% activity. Data represent mean SD. (C) Platelet counts were determined 24 h after challenge with rVWF. Data represent mean SD (n = 7–11). Signiﬁcance levels are calculated using one-way ANOVA and represented as **P < 0.01 and ***P < 0.001. rVWF, recombinant human von Willebrand factor. proposita where ADAMTS13 antigen and activity levels recombinant ADAMTS13 (expressed in the HEK293T were 17 1% and < 0.5%, respectively. These data sug- cells) and plasma ADAMTS13 (mainly produced by stel- gest that the patient might be compound heterozygous late liver cells) might explain in part the observed differ- for another, yet unidentiﬁed mutation probably in the ences in ADAMTS13 antigen levels of the p.D187H ADAMTS13 promoter region or introns, as they were not mutant in culture medium and plasma (53.8% and 17%, sequenced in this study. On the other hand, although the respectively). It was recently reported that not all N142 D187H mutation did not alter a glycosylation site in the and N146 residues were glycosylated in plasma VWF, mutant protein, differences in glycosylation between while these sites always contained N-linked carbohydrates

© 2014 International Society on Thrombosis and Haemostasis ADAMTS13-p.D187H contributes to TTP in mice 291 in recombinant ADAMTS13 expressed in HEK293 cells Disclosure of Conflict of Interests [31,32]. Because N-linked glycans provide stability to a protein, plasma ADAMTS13 might be less stable than H. Rottensteiner is an employee of Baxter Innovations recombinant ADAMTS13. Therefore, a combination of GmbH, Vienna, Austria. All other authors state that they the absence of carbohydrates on N142/N146 in plasma have no conflicts of interests. ADAMST13 with the presence of the D187H mutation in the Ca2+-binding site, which might also reduce protein Supporting Information stability, might lead to a more profound reduction in secretion of plasma p.D187H than of recombinant Additional Supporting Information may be found in the p.D187H. Nevertheless, by studying this novel p.D187H online version of this article: ADAMTS13 mutation in vivo in a mouse model of TTP Video S1. A molecular dynamics simulation of the [15], we could clearly demonstrate that the homozygous ADAMTS13 metalloprotease domain shows unstable p.D187H mutation severely impairs ADAMTS13 secre- Ca2+ binding in the D187H mutant. tion and activity, which were associated with TTP in these animals. In conclusion, we showed that the p.D187H mutation modulates ADAMTS13 expression and reduces ADAM- References TS13 activity by affecting the Ca2+-binding site. Using 1 Tsai H-M. Pathophysiology of thrombotic thrombocytopenic the in vivo mouse model, we could demonstrate that the purpura. Int J Hematol 2010; 91:1–19. homozygous ADAMTS13 p.D187H can indeed induce 2 Moake JL, Rudy CK, Troll JH, Weinstein MJ, Colannino NM, TTP when mice are triggered with rVWF. Whether our Azocar J, Seder RH, Hong SL, Deykin D. Unusually large proposita is compound heterozygous for another yet plasma factor VIII:von Willebrand factor multimers in chronic unidentified ADAMTS13 mutation remains to be deter- relapsing thrombotic thrombocytopenic purpura. N Engl J Med 1982; 307: 1432–5. mined. 3 Pos W, Luken BM, Sorvillo N, Kremer Hovinga JA, Voorberg J. Humoral immune response to ADAMTS13 in acquired throm- Addendum botic thrombocytopenic purpura. J Thromb Haemost 2011; 9: 1285–91. E. De Cock designed the research, performed experi- 4 Furlan M, Robles R, Solenthaler M, Lammle€ B. Acquired defi- ments, analyzed and interpreted data, and wrote the man- ciency of von Willebrand factor-cleaving protease in a patient with thrombotic thrombocytopenic purpura. Blood 1998; 91: uscript; C. Hermans provided and analyzed patient 2839–46. samples, wrote the case history section, and critically read 5 Lotta LA, Garagiola I, Palla R, Cairo A, Peyvandi F. ADAM- the manuscript; J. De Raeymaecker and M. De Maeyer TS13 mutations and polymorphisms in congenital thrombotic performed molecular modeling studies, wrote the corre- thrombocytopenic purpura. Hum Mutat 2010; 31:11–9. sponding sections, and critically read the manuscript; K. 6 Furlan M, Robles R, Solenthaler M, Wassmer M, Sandoz P, Lammle€ B. Deficient activity of von Willebrand factor-cleaving De Ceunynck designed the flow assay, performed experi- protease in chronic relapsing thrombotic thrombocytopenic pur- ments, and critically read the manuscript; B. De Maeyer pura. Blood 1997; 89: 3097–103. designed the mADAMTS13:Ag ELISA and critically read 7 Desch KC, Motto DG. Thrombotic thrombocytopenic purpura the manuscript; N. Vandeputte and A. Vandenbulcke in humans and mice. Arterioscler Thromb Vasc Biol 2007; 27: – provided excellent technical assistance; H. Deckmyn pro- 1901 8. 8 Levy GG, Nichols WC, Lian EC, Foroud T, McClintick JN, vided helpful discussions and critically read the manu- McGee BM, Yang AY, Siemieniak DR, Stark KR, Gruppo R, script; H. Rottensteiner provided rVWF and critically Sarode R, Shurin SB, Chandrasekaran V, Stabler SP, Sabio H, read the manuscript; S. F. De Meyer performed experi- Bouhassira EE, Upshaw JD, Ginsburg D, Tsai HM. Mutations ments, provided helpful discussions, and critically read in a member of the ADAMTS gene family cause thrombotic – the manuscript; K. Vanhoorelbeke designed the research, thrombocytopenic purpura. Nature 2001; 413: 488 94. 9 Zheng X, Chung D, Takayama TK, Majerus EM, Sadler JE, Fu- interpreted data, wrote the manuscript, and provided jikawa K. Structure of von Willebrand factor-cleaving protease funding. (ADAMTS13), a metalloprotease involved in thrombotic thrombocytopenic purpura. J Biol Chem 2001; 276: 41059–63. 10 Mansouri Taleghani M, von Krogh A-S, Fujimura Y, George Acknowledgements JN, HrachovinovaI,Kn obl€ PN, Quist-Paulsen P, Schneppen- heim R, Lammle€ B, Kremer Hovinga JA. Hereditary thrombotic The authors wish to thank H. Feys for the helpful discus- thrombocytopenic purpura and the hereditary TTP registry. sion on the kinetic analysis and F. Scheiflinger (Baxter Hamostaseologie 2013; 33: 138–43. Innovations) for material support. This research received 11 Feys HB, Pareyn I, Vancraenenbroeck R, De Maeyer M, Deck- funding from the Flemish government in the form of the myn H, van Geet C, Vanhoorelbeke K. Mutation of the H-bond acceptor S119 in the ADAMTS13 metalloprotease domain ‘Fonds voor Wetenschappelijk Onderzoek Vlaanderen’ reduces secretion and substrate turnover in a patient with con- (FWO grants G.0607.09 and G.0584.11) and from the genital thrombotic thrombocytopenic purpura. Blood 2009; 114: KU Leuven grant GOA/09/013. 4749–52.

12 Matsumoto M, Kokame K, Soejima K, Miura M, Hayashi S, ADAMTS13 protein-mediated proteolysis. J Biol Chem 2011; Fujii Y, Iwai A, Ito E, Tsuji Y, Takeda-Shitaka M, Iwadate M, 286: 36361–7. Umeyama H, Yagi H, Ishizashi H, Banno F, Nakagaki T, Miy- 23 Gardner MD, Chion CKNK, de Groot R, Shah A, Crawley ata T, Fujimura Y. Molecular characterization of ADAMTS13 JTB, Lane DA. A functional calcium-binding site in the metallo- gene mutations in Japanese patients with Upshaw-Schulman syn- protease domain of ADAMTS13. Blood 2009; 113: 1149–57. drome. Blood 2004; 103: 1305–10. 24 Tortorella MD, Tomasselli AG, Mathis KJ, Schnute ME, 13 Manea M, Kristoffersson A, Schneppenheim R, Saleem MA, Woodard SS, Munie G, Williams JM, Caspers N, Wittwer AJ, Mathieson PW, Morgelin€ M, Bjork€ P, Holmberg L, Karpman Malfait A-M, Shieh H-S. Structural and inhibition analysis D. Podocytes express ADAMTS13 in normal renal cortex and in reveals the mechanism of selectivity of a series of aggrecanase patients with thrombotic thrombocytopenic purpura. Br J Hae- inhibitors. J Biol Chem 2009; 284: 24185–91. matol 2007; 138: 651–62. 25 van der Spoel D, Lindahl E, Hess B, Groenhof G, Mark AE, 14 Tao Z, Anthony K, Peng Y, Choi H, Nolasco L, Rice L, Moake Berendsen HJC. GROMACS: fast, flexible, and free. J Comput JL, Dong JF. Novel ADAMTS-13 mutations in an adult with Chem 2005; 26: 1701–18. delayed onset thrombotic thrombocytopenic purpura. J Thromb 26 De Maeyer B, De Meyer SF, Feys HB, Pareyn I, Vandeputte N, Haemost 2006; 4: 1931–5. Deckmyn H, Vanhoorelbeke K. The distal carboxyterminal domains 15 Schiviz A, Wuersch K, Piskernik C, Dietrich B, Hoellriegl W, of murine ADAMTS13 influence proteolysis of platelet-decorated Rottensteiner H, Scheiflinger F, Schwarz HP, Muchitsch E-M. A VWF strings in vivo. JThrombHaemost2010; 8:2305–12. new mouse model mimicking thrombotic thrombocytopenic pur- 27 De Meyer SF, Vandeputte N, Pareyn I, Petrus I, Lenting PJ, pura: correction of symptoms by recombinant human ADAM- Chuah MKL, VandenDriessche T, Deckmyn H, Vanhoorelbeke TS13. Blood 2012; 119: 6128–35. K. Restoration of plasma von Willebrand factor deficiency is suf- 16 Vanhoorelbeke K, De Meyer SF. Animal models for thrombotic ficient to correct thrombus formation after gene therapy for thrombocytopenic purpura. J Thromb Haemost 2013; 11 (Suppl. severe von Willebrand disease. Arterioscler Thromb Vasc Biol 1): 2–10. 2008; 28: 1621–6. 17 Miller SA, Dykes DD, Polesky HF. A simple salting out proce- 28 Motto DG, Chauhan AK, Zhu G, Homeister J, Lamb CB, dure for extracting DNA from human nucleated cells. Nucleic Desch KC, Zhang W, Tsai H-M, Wagner DD, Ginsburg D. Shi- Acids Res 1988; 16: 1215. gatoxin triggers thrombotic thrombocytopenic purpura in geneti- 18 Kokame K, Miyata T. Genetic defects leading to hereditary cally susceptible ADAMTS13-deficient mice. J Clin Invest 2005; thrombotic thrombocytopenic purpura. Semin Hematol 2004; 41: 115: 2752–61. 34–40. 29 Lotta LA, Tuana G, Yu J, Martinelli I, Wang M, Yu F, Passa- 19 Feys HB, Liu F, Dong N, Pareyn I, Vauterin S, Vandeputte N, monti SM, Pappalardo E, Valsecchi C, Scherer SE, Hale W, Noppe W, Ruan C, Deckmyn H, Vanhoorelbeke K. ADAMTS- Muzny DM, Randi G, Rosendaal FR, Gibbs RA, Peyvandi F. 13 plasma level determination uncovers antigen absence in Next-generation sequencing study finds an excess of rare, coding acquired thrombotic thrombocytopenic purpura and ethnic dif- single-nucleotide variants of ADAMTS13 in patients with deep ferences. J Thromb Haemost 2006; 4: 955–62. vein thrombosis. J Thromb Haemost 2013; 11: 1228–39. 20 Plaimauer B, Fuhrmann J, Mohr G, Wernhart W, Bruno K, 30 de Groot R, Lane DA, Crawley JTB. The ADAMTS13 metallo- Ferrari S, Konetschny C, Antoine G, Rieger M, Scheiflinger F. protease domain: roles of subsites in enzyme activity and Modulation of ADAMTS13 secretion and specific activity by a specificity. Blood 2010; 116: 3064–72. combination of common amino acid polymorphisms and a mis- 31 Sorvillo N, Kaijen PH, Matsumoto M, Fujimura Y, van der sense mutation. Blood 2006; 107: 118–25. Zwaan C, Verbij FC, Pos W, Fijnheer R, Voorberg J, Meijer 21 Anderson PJ, Kokame K, Sadler JE. Zinc and calcium ions AB. Identification of N-linked glycosylation and putative cooperatively modulate ADAMTS13 activity. J Biol Chem 2006; O-fucosylation, C-mannosylation sites in plasma derived 281: 850–7. ADAMTS13. J Thromb Haemost 2014; 12: 670–9. 22 De Ceunynck K, Rocha S, Feys HB, De Meyer SF, Uji-i H, 32 Zhou W, Tsai H-M. N-Glycans of ADAMTS13 modulate its Deckmyn H, Hofkens J, Vanhoorelbeke K. Local elongation of secretion and von Willebrand factor cleaving activity. Blood endothelial cell-anchored von Willebrand factor strings precedes 2009; 113: 929–35.