Cysteine-mutations in von Willebrand factor associated with increased clearance

Authors


Peter J. Lenting, Department of Haematology, Laboratory for Thrombosis and Haemostasis, G.03.647, University Medical Center Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands.
Tel.: +31 30 250 7610; fax: +31 30 251 1893; e-mail: p.j.lenting@lab.azu.nl

Abstract

Summary. Background: von Willebrand disease (VWD) is a bleeding disorder caused by the decrease of functional von Willebrand factor (VWF). Low levels of VWF can result from decreased synthesis, impaired secretion, increased clearance or combinations thereof. Several mutations lead to impaired synthesis or secretion of VWF, however, little is known about the survival of VWF in the circulation. Objectives: To evaluate the effect of several VWF mutations on VWF clearance. Patients/methods: The effect of three cysteine-mutations (C1130F, C1149R or C2671Y) on the in vivo survival of VWF was studied in patients carrying these mutations and in a VWF-deficient mice model. Results: In patients carrying these mutations, we observed increased propeptide/mature VWF ratios and rapid disappearance of VWF from the circulation after desmopressin treatment. Detailed analysis of in vivo clearance of recombinant VWF in a VWF-deficient mice model revealed a fourfold increased clearance rate of the mutants. The mutations C1130F, C1149R and C2671Y are each associated with reduced survival of VWF in the circulation. Detailed analysis of the recombinant mutant VWF demonstrated that increased clearance was not due to increased proteolysis by ADAMTS-13. We did not identify functional or structural characteristics that the mutant proteins have in common and could be associated with the phenomenon of increased clearance. Conclusions: Cysteine-mutations in VWF may result in reduced in vivo survival. The observation that various mutations are associated with increased in vivo clearance may have major implications for the therapeutic strategies that rely on the rise of endogenous VWF after desmopressin administration.

Introduction

von Willebrand factor (VWF) is a plasma glycoprotein that plays a dual role in hemostasis. Firstly, VWF forms a complex with coagulation factor VIII (FVIII), which is required for FVIII survival in vivo. Secondly, VWF contributes to platelet adhesion and aggregation by acting as a molecular bridge between subendothelial collagen and platelets [1].

von Willebrand factor is produced in megakaryocytes and endothelial cells, where it is subjected to extensive post-translational modifications, including proteolytic separation of the precursor molecule in a propeptide and mature VWF. VWF maturation involves C-terminal dimerization and propeptide-dependent N-terminal multimerization via the formation of covalent intermolecular cystine-bonds [1]. The ability of VWF to support platelet adhesion and aggregation increases with multimer size [1]. In the circulation, multimer size is controlled by several mechanisms, including proteolytic cleavage by ADAMTS-13 [2,3]. VWF is also subjected to N- and O-linked glycosylation and is one of the rare plasma proteins that carry the blood group antigens A, B and H [4]. The VWF glycosylation profile is one of the determinants of VWF survival in the circulation [5,6]. Besides altered glycosylation, amino acid substitutions in VWF may also affect clearance. Recently, we have used a mouse-model to demonstrate that the mutation R1205H leads to a strongly reduced survival of VWF in the circulation [7].

Defects in the VWF gene result in a bleeding disorder (von Willebrand disease, VWD) with variable penetrance. VWD can be categorized in qualitative VWF defects (type 2) and quantitative VWF deficiency (type 1 and type 3; partial and virtually complete deficiency, respectively) [1]. Most mutations in the human mutation database that are associated with the quantitative deficiencies originate from null alleles (gene deletions, stop codons, frame shifts, and splice mutations), but missense mutations have also been reported. Some of these mutations lead to replacement of cysteine-residues, like C1130F, C1149R and C2671Y [8,9]. Mutations C1130F and C1149R are both located in the VWF D3-domain and are recognized as dominant-negative mutations [9,10]. Patients heterozygous for C1130F and C1149R display a pronounced quantitative VWF deficiency, a prolonged bleeding time and a history of moderate bleeding [9,11,12]. Mutation C2671Y is located toward the carboxyterminus of the molecule in the connective region between the C2- and CK-domains. The mutation has been reported for a single patient who is compound heterozygous for this mutation and a gene deletion. The patient displays a type 3 phenotype with VWF antigen (VWF:Ag) levels of 0.02–0.04 U mL−1, and part of the residual circulating VWF protein consists of proteolyzed products [8,13].

Analysis of recombinant variants of VWF/C1130F and VWF/C1149R revealed that these mutants lacked high molecular weight multimers, which defect could be corrected upon co-expression with recombinant wild-type VWF (wt-VWF). In contrast, VWF/C2671Y showed a normal multimerization pattern. All three mutations resulted in reduced secretion of VWF due to intracellular retention, and, at least for VWF/C1149R, proteosomal degradation [9,10,12]. However, the plasma VWF antigen (VWF:Ag) levels found in the patients were much lower than the VWF levels observed in the in vitro expression experiments [12]. This indicates that the intracellular retention and degradation of the mutants only partly explain the low VWF levels in the patients. In the present study, we explored the possibility that modified clearance also contributes to the reduced VWF levels in these patients. By investigating the behavior of endogenous VWF upon desmopressin treatment of patients as well as the survival of recombinant VWF mutants in an experimental model employing VWF-deficient mice, we have indeed obtained evidence that increased clearance contributes to the reduced VWF levels in vivo.

Materials and methods

Patients, mutations and phenotypic tests

The C1130F and C1149R mutations were originally identified in patients classified as type 1 VWD, characterized by very low VWF:Ag levels (0.10–0.15 IU mL−1) [9]. The C2671Y mutation was found in a type 3 VWD patient, compound heterozygous for this mutation and a deletion of the other allele [8]. Plasma was collected on several occasions from seven patients and two unaffected family members (Table 1). An infusion with 1-deamino-8-d-arginine vasopressin (DDAVP; 0.3 μg DDAVP kg−1 bw) was performed after informed consent. Blood samples were analyzed for VWF:Ag by an immunosorbent assay, VWF ristocetin cofactor activity by aggregometry using fixed human platelets and FVIII activity using a one-stage clotting assay. The concentrations of VWF-propeptide antigen were analyzed in an immunosorbent assay as described [14,15]. A plasma pool containing 6.3 nm of propeptide and 50 nm of VWF (monomer-concentration) was used as a standard.

Table 1.  Patient characteristics
VWFFamily and patients*FVIII:C (IU mL−1)VWF:Ag (IU mL−1)VWF:RCo (IU mL−1)Blood group
  1. *Patients and family members are indicated according to the original articles [8,9,11].

  2. Unaffected family members.

  3. FVIII:C, FVIII activity; VWF:Ag, VWF antigen; VWF:RCo, VWF ristocetin cofactor activity.

C1130F/wtC Proband0.190.14<0.20O
C sister0.2–0.250.13–0.16<0.20O
B proband0.450.13–0.21<0.20A
C1149R/wtIII-40.18–0.210.10–0.21<0.20A
IV-10.22–0.290.11–0.22<0.20A
IV-20.35–0.740.15–0.35<0.20O
III-51.170.83–0.910.93O
IV-31.090.86–1.330.91A
C2671Y/delA III-20.170.10<0.20O
Normal range 0.5–20.4–20.5–2 

Mice

The VWF-deficient [16] and wild-type mice were on a C57BL/6J background and were between 8 and 12 weeks old. Housing and experiments were done as recommended by French regulations and the experimental guidelines of the European Community.

Proteins

Recombinant Glycoprotein-Ibα (GpIbα, residues 1–290) was prepared as described [17]. The GpIbα antibody (2D4) was a gift of H. Deckmyn (Kortrijk, Belgium). Botrocetin was from Kordia (Leiden, The Netherlands). Purified recombinant VWF-propeptide was prepared as described [15]. Recombinant B-domain deleted FVIII (Refacto) was from Wyeth. Human collagen type III (catalogue-number C-4407) and bovine albumin (fraction V) were from Sigma. Human albumin (fraction V) was from MP Biochemicals (Irvine, CA, USA). Polyclonal antibodies (unlabeled and peroxidase-conjugated) against VWF were from Dako (Glostrup, Denmark).

Recombinant VWF

All plasmids were constructed using conventional techniques and all constructs were sequenced before transfection. Plasmids pSVH-VWF encoding wt-VWF, rVWF/C1130F, rVWF/C1149R and rVWF/C2671Y were transiently expressed in 293T human kidney cells [12]. These recombinant variants were used to analyze ADAMTS-13-mediated proteolysis. Plasmids pNUT-VWF encoding wt-VWF, rVWF/C1130F, rVWF/C1149R and rVWF/C2671Y were stably expressed in baby hamster kidney-cells overexpressing furin for proper removal of the propeptide [7,18,19]. cDNA encoding wt-VWF was subcloned into pcDNA6 for co-transfection to rVWF/C1149R-expression cells, allowing selection using blasticidin to establish stable cell-lines expressing heterozygous wt-VWF-rVWF/C1149R. All recombinant variants were purified from conditioned serum-free medium, and used in the various functional assays and to determine VWF clearance in mice.

ADAMTS-13-mediated proteolysis

The pcDNA3.1-ADAMTS-13 expression vector (provided by J. Voorberg, CLB research at Sanquin, Amsterdam, The Netherlands) was used for transient expression of recombinant ADAMTS-13. Conditioned medium was collected and Pefabloc SC (Roche Diagnostics, Mannheim, Germany) was added to a final concentration of 1 mm. Medium was concentrated ∼10 times by centrifugation in 5 mm of Tris (pH 8.0) using Macrosep concentrators (cut-off value 100 kDa; Pall Gellman Laboratory, Ann Harbor, MI, USA). ADAMTS-13 protease activity was determined using previously described methods and conditions [20,21]. The sensitivity of a fixed concentration of mutant, cotransfected or wt-VWF was assessed with a 1:20 dilution of ADAMTS-13 in 5 mm of Tris (pH 8.0) with 12.5 mm of BaCl2 to initiate the reaction. The final concentration of urea in the reaction was 0, 0.1, 0.5, 1.0 or 1.5 m. Aliquots were taken after 0, 8 and 24 h incubation at 37 °C and the reaction was stopped with EDTA. Subsequently, the VWF multimeric structure was analyzed on denaturing, non-reducing agarose gels [1.5% (w/v)] [12]. In all tests, the same batch of concentrated ADAMTS-13 was used.

Clearance of purified recombinant VWF in mice

Clearance of recombinant wt-VWF and mutant VWF has been analyzed as described [7,16]. Three to six mice were used for each time-point, and each mouse was bled only once. Human VWF:Ag levels were quantified as described [14].

GpIbα binding

Monoclonal anti-GpIbα antibody 2D4 was immobilized in microtiter-wells (Costar, Cambridge, MA, USA) in 50 mm of NaHCO3 (1.0 μg mL−1, overnight at 4 °C), which were then blocked for 1 h at 37 °C with PBS/3% (w/v) bovine albumin/0.1% (v/v) Tween-20. Recombinant GpIbα was added (0.1 μg mL−1 for 2 h at 37 °C) and subsequently increasing concentrations of VWF (0–4 nm) were applied in the presence of botrocetin (2 μg mL−1) and incubated for 2 h at 37 °C. After washing, wells were incubated with peroxidase-labeled polyclonal anti-VWF antibodies (1.3 μg mL−1 for 1 h at 37 °C), and bound VWF was detected by measuring peroxidase activity using o-phenylenediamine as a substrate.

αIIBβ3-dependent platelet adhesion to immobilized VWF

Perfusion with platelets (shear-rate 1600 s−1 ) over VWF-coated cover slips was performed as described elsewhere [22]. The amount of platelet-adhesion was evaluated using computer-assisted analysis with optimas-6.0 software (Dutch-Vision-Systems, Breda, The Netherlands), and was expressed as the percentage of surface-coverage. Perfusions were performed 6–9 times for all variants.

Surface Plasmon Resonance analysis

Several binding assays were performed employing a Biacore2000 system (Biacore AB, Uppsala, Sweden). Binding of propeptide and FVIII to immobilized VWF or its mutants was investigated as described [7], except that recombinant B-domain deleted FVIII (Refacto) was used instead of FVIII light chain. Binding of VWF or its derivatives to immobilized collagen type III was performed as described [14].

Data analysis and statistics

Analysis of protein-interaction assays and clearance data obtained from mouse-experiments was performed using the GraphPad Prism program (GraphPad Prism version 4.0 for Windows, GraphPad Software, San Diego, CA, USA) as described [7]. Clearance data obtained from the patients, after subtraction of the basal VWF and propeptide level from the post-DDAVP values, were fitted to the monoexponential equation Ct = Aeαt to obtain α. Ct refers to the plasma concentration of VWF or propeptide at time-point t after DDAVP infusion, and the apparent half-life was calculated from the equation t1/2 = ln 2/α. Statistical analyses were performed by the Student's unpaired t-test using the GraphPad InStat program (GraphPad InStat version 3.00 for Windows).

Results

Increased VWF-propeptide/mature VWF ratio

von Willebrand factor is secreted simultaneously with its propeptide at equimolar concentrations and they circulate at a distinct ratio [15,23]. The patients in this study display VWF:Ag levels that are well below the normal range (Table 1). In contrast, the VWF-propeptide levels were less reduced, resulting in an increased ratio of propeptide over VWF:Ag (Fig. 1). These ratios were increased threefold for patients with the mutations C1130F and C1149R compared with the normal population. The ratio was also increased in the compound heterozygous patient (C2671Y/deletion) (Fig. 1). These data suggest a reduced half-life of VWF, or, alternatively, a prolonged half-life of propeptide.

Figure 1.

Propeptide/VWF ratio in VWD patients. The average ratio of propeptide over mature VWF for normal controls (N1), unaffected family members from the family with the C1149R mutation (N2), and patients harboring the C1130F, C1149R or C2671Y/del mutation at basal level are depicted. The number of measurements (n) is indicated, with the number of patients in brackets. Differences between groups were tested with the unpaired two-tailed t-test. Data represent mean ± S.D., and data of N1 were obtained in a previous study [23]. NS, not significant (P > 0.05).

Increased clearance of VWF upon DDAVP treatment

To test whether the modified ratios result from a change in half-life of propeptide or mature VWF, plasma levels of VWF and propeptide were monitored at different time-points after DDAVP infusion. Healthy subjects and patients showed a rapid increase of the propeptide level, which was followed by a monophasic decay (Fig. 2A). The half-life of propeptide in patients with C1130F or C1149R was similar to healthy subjects, whereas the half-life of propeptide in the patient with C2671Y was somewhat longer (Fig. 2A, Table 2). VWF showed a similar rapid increase followed by a monophasic decay (Fig. 2B). In the patient group, VWF was cleared much quicker than in the group of healthy subjects. Indeed, the estimated half-life of VWF in the patients with mutations C1130F, C1149R and C2671Y was reduced four- to fivefold (Fig. 2B, Table 2). It should be noted that the estimated half-lives were based on a limited number of time points, which prevent accurate calculation of the half-lives. Nevertheless, these data strongly indicate that the mutations C1130F, C1149R and C2671Y are associated with an increased clearance of mature VWF.

Figure 2.

Survival of endogenous propeptide and VWF after DDAVP administration. Blood was collected before and after DDAVP infusion (0.3 μg kg−1 bw). Samples were analyzed for propeptide (Panel A) and VWF (Panel B) levels. Shown are the average ± S.D. for controls (n = 9) and data from single experiments in patients. Data of the normal controls were obtained in a previous study [23]. The apparent half-lives are listed in Table 2.

Table 2.  Apparent half-lives of propeptide and VWF after DDAVP treatment
VWFt1/2 propeptide (h)t1/2 VWF (h)Blood group
Normal2.37.0 
C1130F3.21.6O
C1149R2.61.5A
C2671Y6.01.3O

Mutations at C1130, C1149 and C2671 leave ADAMTS-13-dependent proteolysis unaffected

Rapid clearance could be due to increased susceptibility of VWF mutants to degradation by ADAMTS-13. The susceptibility of the recombinant mutants for ADAMTS-13 was therefore assessed in degradation assays. Representative results from the protease assay of rVWF/C1130F and rVWF/C2671Y are shown in Fig. 3. No or little degradation was detected in the absence of urea. However, the presence of 1.0 m urea resulted in a gradual disappearance of the higher multimers (Fig. 3). Due to impaired multimerization of rVWF/C1130F and rVWF/C1149R, the sensitivity for ADAMTS-13 could only be interpreted for the respective cotransfections with wt-VWF, which are also more representative for the situation in the heterozygous patients (Fig. 3). At all urea concentrations and at all incubation times tested, the VWF mutants were indistinguishable from wt-VWF and no evidence was found that the accelerated disappearance from the circulation is caused by an elevated susceptibility toward ADAMTS-13-mediated cleavage.

Figure 3.

ADAMTS-13-mediated proteolysis of recombinant wild-type and mutant VWF. rVWF/C1130F (panel A), rVWF/C2671Y (panel B), wt-VWF (WT) alone or VWF obtained from co-transfections (Co-tf) were incubated with recombinant ADAMTS-13 for 0, 8 or 24 h (t) in the absence or presence of 0.1, 0.5, 1.0, 1.5 m of urea. The results from 0 and 1.0 m of urea are depicted. The degree of proteolysis was assayed on SDS-agarose gel electrophoresis under non-reducing conditions. The results for rVWF/C1130F are representative for rVWF/C1149R (data not shown).

Increased clearance of VWF mutants in VWF-deficient mice

The clearance of mutant VWF was studied in a murine model in which VWF clearance is independent of ADAMTS-13-mediated proteolysis and independent of the extent of VWF multimerization [7]. VWF mutants produced by stable cell-lines were purified from conditioned medium. Multimer analysis revealed a normal multimeric pattern for rVWF/C2671Y. rVWF/C1130F and rVWF/C1149R consisted mainly of low molecular weight multimers, whereas co-expression of rVWF/C1149R with wt-VWF resulted in the reappearance of higher multimers. These patterns are similar to that of the mutants produced by transiently transfected cells (Fig. 3; [12]). Purified proteins were injected intravenously in VWF-deficient mice and residual plasma-levels were measured. The mutants and wt-VWF displayed a similar recovery after injection (Table 3). In contrast, all four mutants were cleared from the circulation more rapidly than wt-VWF (Table 3, Fig. 4). In the murine system, wt-VWF disappears from plasma in a biphasic manner, characterized by a rapid initial phase and a slow secondary phase [7]. As for wt-VWF, the mutants disappeared in a biphasic manner, but data analysis revealed that the initial rapid phase (t1/2α), the slow secondary phase (t1/2β) and mean residence time were significantly reduced compared with the values obtained for wt-VWF (Table 3). Thus, the mutations C1130F, C1149R and C2671Y per se are associated with accelerated clearance of VWF.

Table 3.  Pharmacokinetic parameters of the clearance of recombinant VWF in mice
 Recovery (% of injected)Mean residence time (h)t1/2α (min)t1/2β (h)
  1. *Data were obtained during a previous study [7]. Wt-VWF is reference for comparisons. rVWF/D2509G and rVWF/R1205H are shown to illustrate the normal and decreased half-life of other recombinant mutants.

wt-VWF*79 ± 142.8 ± 0.712.6 ± 0.93.0 ± 0.9
rVWF/D2509G*82 ± 5 (P > 0.05)2.3 ± 0.5 (P > 0.05)12.0 ± 2.2 (P > 0.05)2.2 ± 0.2 (P > 0.05)
rVWF/C1130F72 ± 6 (P > 0.05)0.7 ± 0.2 (P = 0.007)6.0 ± 1.4 (P = 0.002)1.1 ± 0.5 (P = 0.032)
rVWF/C1149R67 ± 10 (P > 0.05)0.8 ± 0.4 (P = 0.011)7.5 ± 2.5 (P = 0.028)1.1 ± 0.5 (P = 0.032)
wt-VWF-rVWF/C1149R75 ± 9 (P > 0.05)0.4 ± 0.1 (P = 0.0042)5.5 ± 0.4 (P = 0.0002)0.7 ± 0.1 (P = 0.012)
rVWF/R1205H*97 ± 8 (P > 0.05)0.3 ± 0.1 (P = 0.004)7.6 ± 0.2 (P = 0.0007)0.3 ± 0.03 (P = 0.007)
rVWF/C2671Y63 ± 8 (P > 0.05)0.7 ± 0.1 (P = 0.006)8.9 ± 1.5 (P = 0.022)0.7 ± 0.2 (P = 0.012)
Figure 4.

In vivo survival of recombinant VWF mutants. VWF-deficient mice were injected intravenously with purified wt-VWF, rVWF/C1130F (Panel A), rVWF/C1149R (Panel B), wt-VWF-rVWF/C1149R (Panel B) or rVWF/C2671Y (Panel C) at a concentration of 5 μg/mouse. Plotted is the residual VWF:Ag in plasma relative to the amount injected vs. time after injection. For clarity, only data between 3 min and 4 h are shown. Data for wt-VWF were obtained in a previous study [7], and are shown for comparison. Pharmacokinetic parameters derived from the complete data set (from 3 min to 24 h) are summarized in Table 3. Data represent the mean ± SD of 3–6 mice for each time-point.

Functional characterization of recombinant mutants rVWF/C1130F, rVWF/C1149R and rVWF/C2671Y

Because the accelerated clearance of the mutant VWF is due to the mutations per se, we examined whether the mutants have functional or structural characteristics in common that point to a region in VWF responsible for the increased clearance. The results of a number of functional parameters (i.e. FVIII-, GpIbα- and collagen-binding, platelet-adhesion under flow-conditions) are summarized in Table 4. Briefly, mutant rVWF/C2671Y was similar to wt-VWF in all assays tested. Mutant rVWF/C1130F displayed normal GpIbα-binding and platelet-adhesion capacity. Furthermore, FVIII-binding was reduced fivefold for this mutant, while collagen-binding was reduced to a level similar to a dimeric control protein. Mutant rVWF/C1149R was most severely affected; apart from collagen-binding that was reduced to the level of the dimeric control, all other functions were considerably reduced compared with the control proteins.

Table 4.  Functional analysis of recombinant VWF mutants
 GpIbα binding Half-maximal binding (nm)Collagen binding KD,app (nm)FVIII binding KD,app (nm)Platelet adhesion Surface coverage (%)
  1. *Dimeric control was: rVWF/D′-A3 for collagen binding, rVWF/A1-CK for platelet-adhesion and rVWF/D′-D3 for FVIII binding. All dimeric controls have been described previously [7]. n.d., not determined.

wt-VWF1.2 ± 0.29.4 ± 0.60.51 ± 0.0477.3 ± 6.3
rVWF/C1130F1.0 ± 0.147.4 ± 1.02.3 ± 0.180.9 ± 8.8
rVWF/C1149R3.9 ± 1.153.4 ± 1.410.6 ± 0.449.4 ± 16.1
rVWF/C2671Y1.0 ± 0.28.3 ± 0.60.33 ± 0.0279.2 ± 9.8
Dimeric control*n.d.67.5 ± 1.20.38 ± 0.0670.5 ± 6.4

In view of the multimerization-defects of rVWF/C1130F and rVWF/C1149R, we also tested the interaction with the propeptide. This interaction mediates intracellular multimerization and targeting of mature VWF to the storage-organelles and occurs under slightly acidic conditions. The interaction between VWF and propeptide was investigated at various pHs (Fig. 5). For wt-VWF, a pH-dependent increase in affinity was observed with lower affinity at higher pH (pH 7.4, KD,app = 1.0 ± 0.1 μm and pH 5.2, KD,app = 0.08 ± 0.02 μm). A similar pattern was found for rVWF/C2671Y. However, binding of propeptide was markedly reduced for mutants rVWF/C1130F and rVWF/C1149R (KD,app > 2 μm at all pHs tested). Again, rVWF/C1149R was less efficient compared with rVWF/C1130F. These data indicate that mutations within the D′-D3 domain are not only associated with increased clearance, but also may lead to suboptimal interactions with ligands that bind to this particular region of the VWF molecule.

Figure 5.

Binding of propeptide to immobilized VWF. wt-VWF, rVWF/C1130F, rVWF/C1149R, and rVWF/C2671Y immobilized onto a CM5 sensor-chip (10–12 fmol mm−2 ) were incubated with various concentrations of purified propeptide (0–1.0 μm). Incubations were performed in 125 mm of NaCl, 25 mm of HEPES [pH 7.4 (Panel A) or pH 6.8 (Panel B)], 125 mm of NaCl, 25 mm (CH3)2AsO2Na [pH 5.8 (Panel C) or pH 5.2 (Panel D)] at a flow rate of 10 μL min−1 for 4 min at 25 °C. Response at equilibrium (indicated as mol propeptide/mol VWF) is plotted against concentration of propeptide. Please note the different scales of the Y-axes. Data represent the mean of duplicate experiments (range <10%.).

Discussion

Circulating levels of VWF need strict regulation, as levels that are too low are associated with an increased bleeding tendency [1], whereas levels that are too high predispose to an increased risk of cardiovascular mortality [24]. Low levels of VWF can result from decreased synthesis, impaired secretion, increased clearance or combinations thereof. Several mutations lead to impaired synthesis or secretion of VWF, however, little is known about the relationship between amino acid variations in VWF and its survival in the circulation. In the present study, we evaluated the effect of three VWD-associated mutations on the survival of VWF. These mutations were selected on basis of discrepancies between VWF levels measured in in vitro expression experiments and the relatively lower actual antigen values in the patients’ plasmas [9,10,12]. Such discrepancies may indicate the presence of mechanisms other than the synthesis and secretion that contribute to the low VWF levels.

We have obtained several lines of evidence of increased clearance of mutants VWF/C1130F, VWF/C1149R and VWF/C2671Y. Firstly, we analyzed in the patients the propeptide/VWF:Ag ratio, which is the resultant of a dissimilar survival of these proteins in the circulation [15,23]. The VWF:Ag levels were much more reduced than propeptide levels, resulting in ratios that were increased up to threefold compared with normal individuals or unaffected family members (Fig. 1). Secondly, VWF disappeared from the circulation four- to fivefold more rapidly in the patients upon DDAVP-treatment (Fig. 2B). Thirdly, we performed detailed analysis of in vivo clearance of recombinant VWF in a model employing VWF-deficient mice [7]. Both the initial and secondary phase of the clearance of rVWF/C1130F, rVWF/C1149R and rVWF/C2671Y were accelerated (Fig. 4), which was reflected by a mean residence time that was reduced fourfold (Table 3). Thus, the mutations C1130F, C1149R and C2671Y are each associated with a reduced survival of VWF in the circulation. Because also the heterozygous wt-VWF-rVWF/C1149R mimicked the increased clearance of the protein in the patients, it seems that at least some of these mutations may have a predominant effect on the clearance of the mutated protein. Furthermore, we have recently shown that also mutation R1205H results in an increased clearance of VWF [7]. A preliminary report has recently described a mutation S2179F in the VWF D4-domain, which seems to have a similar effect on VWF clearance [25]. This indicates that several mutations in VWF predispose to increased clearance. Indeed, a decreased survival of VWF upon DDAVP treatment in a cohort of VWD-type 1 patients has recently been reported [26].

This conclusion is of importance with respect to treatment of VWD patients. The goal of treatment is to correct the deficiency of VWF and FVIII, either by transfusion with plasma-derived FVIII/VWF concentrates or DDAVP administration. The latter leads to the release of endogenous VWF from storage organelles, which are associated with a three- to fivefold increase of VWF levels. However, if due to a mutation the survival of endogenous VWF is reduced, the initial response to DDAVP may be normal, but the effect is only short-lasting because the endogenous VWF is cleared rapidly. In some of these patients, it may be more appropriate to treat with FVIII/VWF concentrates. As also put forward by Brown et al. [26], the clinical practice to judge the effectiveness of DDAVP treatment by the initial rise of VWF may be insufficient for several patients as it does not take clearance into account. Therefore we suggest that, after a DDAVP test-infusion, VWF:Ag is monitored over a sufficient length of time and we propose the implementation of the propeptide/VWF:Ag ratio in the diagnosis of VWD.

The question remains why the various mutations are associated with increased clearance of the VWF protein. It has recently been shown that the presence of a Tyr to Cys polymorphism at position 1584 results in an increased susceptibility to the VWF-cleaving protease ADAMTS-13 [27]. As the patient harboring the C2671Y mutation displays relatively high plasma levels of VWF degradation products [13], we tested whether increased clearance could result from an increased susceptibility to the ADAMTS-13-protease. However, we did not find evidence of increased proteolysis (Fig. 3). Moreover, normal proteolysis was also reported for VWF/R1205H [28], another mutant that displays increased clearance [7]. Final evidence that the increased clearance of the various mutants is independent of ADAMTS-13 was provided by the rapid clearance in the mouse model, as murine ADAMTS-13 does not recognize human VWF. It should be noted that our study has not examined the potential role of thrombospondin-1, a plasma depolymerase that is able to modulate VWF multimer size [29].

Increased clearance may also be due to a disturbed structural integrity of the mutant proteins, rendering them susceptible to quality control mechanisms that remove anomalous proteins from the circulation. We have tried to identify whether the mutant proteins have functional or structural characteristics in common that may be associated with the phenomenon of increased clearance. Several parameters were examined, but none of these functional tests pointed to a particular region within the VWF molecule that seems to be consistently associated with abnormal clearance. Mutant rVWF/C2671Y was normal for all functions, indicating that this mutation results in local changes within the molecule only. This could suggest that this part of VWF contributes to interactions with clearance receptors. All other mutations that we tested so far, i.e. rVWF/C1130F, rVWF/C1149R and rVWF/R1205H, are located within a relatively short stretch in the D3 domain. However, the impact of these mutations on VWF function differs considerably. Whereas mutant rVWF/R1205H has normal interaction with FVIII [7], both rVWF/C1130F and rVWF/C1149R display impaired FVIII binding (Table 4). Mutations C1130F and C1149R further resulted in a lack of proper multimerization (Fig. 3; [12]), which was not seen with rVWF/R1205H [7]. A combined effect on FVIII binding and multimerization has been described for several other mutations, and seems to be related to a local distortion of the secondary structure [30]. Incomplete multimerization is a consequence of a suboptimal interaction with the propeptide, which is involved in intracellular multimerization of VWF [31,32]. Indeed, mutants rVWF/C1130F and rVWF/C1149R displayed defective binding of propeptide (Fig. 5). Thus, the lack of multimerization observed for these mutants can readily be explained by an almost complete absence of propeptide binding.

In conclusion, our study shows that certain amino acids changes within the VWF molecule are associated with increased clearance. This may have major implications for the therapeutic strategies that rely on the temporary rise of endogenous VWF after DDAVP administration. New therapeutic approaches based on the inhibition of VWF clearance could be developed as an adjuvant to optimize treatment of these patients. However, more insights into the molecular mechanisms mediating VWF clearance are needed in this regard.

Addendum

P.J. Lenting, and C.V. Denis: conception and design of the study, performing experiments, analysis of results, drafting and final version of the manuscript.

J.C.J. Eikenboom: conception and design of the study, analysis of results, drafting and final version of the manuscript.

C.J. van Schooten, P. Tjernberg, E. Westein, V. Terraube and M.J. Hollestelle: design of the study, performing experiments, analysis of results, drafting and final version of the manuscript.

G. Castaman, and J.A. van Mourik: data collection of patients and normal controls respectively, responsible for DDAVP infusion experiments, revision of the draft and approval of final manuscript.

H.L. Vos, R.M. Bertina, and H.M. van den Berg: design of the study, revision of the draft and approval of final manuscript.

Acknowledgements

This study was supported by grants from the von Creveld Foundation (to PJL and MHvdB), GEHT-ISTH (to VT), INSERM-NWO/ZonMW (#910–48–603 to CVD and PJL), INSERM (Avenir-research grant to CVD), NWO/ZonMW (#906–26–209 to JCJE) and the van den Tol Foundation (to JCJE). We thank J. Voorberg and B. Luken for providing the pcDNA3.1 ADAMTS-13 construct, H. Deckmyn for providing antibody 2D4 and T. Lisman for providing purified recombinant propeptide.

Ancillary