An Experimental Model to Study the in Vivo Survival of von Willebrand Factor

To explore the molecular basis of von Willebrand factor (VWF) clearance, an experimental model employing VWF-deficient mice was developed. Biodistribution was examined by the injection of radiolabeled VWF, which was primarily directed to the liver with minor amounts in other organs. Disappearance of VWF from plasma was characterized by a rapid initial phase (t½α = 13 min) and a slow secondary phase (t½β = 3 h), with a mean residence time (MRT) of 2.8 h. A similar clearance was observed for VWF consisting of only high or low molecular weight multimers, indicating that, in our experimental model, clearance is independent of multimeric distribution. This allowed us to compare the survival of full-length VWF to truncated variants. Deletion of both the amino-terminal D′-D3 and carboxyl-terminal D4-CK domains resulted in a fragment with a similar clearance to wild-type VWF. Deletion of only the D′-D3 region was associated with an almost 2-fold lower recovery and increased clearance (MRT = 1.6 h), whereas deletion of only the D4-CK region resulted in a significantly reduced clearance (MRT = 4.5 h, p < 0.02). These results point to a role of the D′-D3 region in preventing clearance of VWF. Furthermore, replacement of D3 domain residue Arg-1205 by His resulted in a markedly increased clearance (MRT = 0.3 h; p = 0.004). Therefore, this mutation seems to abrogate the protective effect of the D′-D3 region. In vitro analysis of this mutant also revealed a 2-fold reduced affinity for VWF propeptide at low pH, showing that mutation of Arg-1205 results not only in an increased clearance rate but is also associated with an impaired pH-dependent interaction with VWF propeptide.

von Willebrand factor (VWF) 1 is a multimeric plasma protein that participates in the hemostatic process. The absence of functional VWF is associated with an abnormal bleeding tendency known as von Willebrand Disease (VWD) (1,2). VWF contributes to hemostasis in a dual manner. First, it promotes the adhesion of platelets at sites of vascular injury by acting as a molecular bridge between the sub-endothelial collagen matrix and the platelet-surface receptor complex glycoprotein (Gp) Ib␣/IX/V (3). In addition, VWF serves as a carrier protein for coagulation factor VIII (FVIII) in the circulation. This chaperone function results in stabilization of the FVIII heterodimeric structure (4) and protection of FVIII from premature clearance by the low density lipoprotein receptor-related protein (5,6).
VWF is produced and stored in endothelial cells and megakaryocytes. It is synthesized as pre-pro-VWF, a single chain polypeptide with the domain structure D1-D2-DЈ-D3-A1-A2-A3-D4-B1-B2-B3-C1-C2-CK (7). The pro-VWF molecule is subject to extensive posttranslational modifications, including carboxyl-terminal dimerization (8). Because of proteolytic processing, pro-VWF is further separated into two polypeptides, the VWF propeptide (also known as VWF AgII) containing the D1-D2 region, and mature VWF, which comprises the remaining domains. The main function of the propeptide is to facilitate amino-terminal multimerization of mature VWF within the post-Golgi compartments and its targeting to Weibel-Palade storage bodies in endothelial cells (9,10). Thus, the propeptide and mature VWF appear to interact within cellular organelles, where conditions are slightly acidic (pH Ͻ 6.5). The extent of multimerization may differ between VWF molecules and is defined by the number of dimeric subunits included (between 2 and Ͼ20). As a result, VWF circulates within plasma as a heterogeneously sized protein with its molecular mass ranging from 0.5 ϫ 10 6 to over 5 ϫ 10 6 Da.
The average plasma level of VWF is ϳ10 g/ml (35 nM based on monomer concentrations), although a broad range is observed between individuals (11). A major part of this variation is determined by ABO blood type; the average VWF levels in persons with blood group O are ϳ25% lower than those in non-O individuals (12). However, other genetic (13,14) and non-genetic factors, such as age and physical activity (15,16), may also contribute to this variation. These factors may influence VWF levels by modulating the balance between its biosynthesis and clearance. In the last two decades, biosynthesis of VWF has been well studied (for recent reviews see Refs. 17 and 18). In contrast, little is known about the mechanism by which VWF is cleared from the circulation. However, this issue is becoming increasingly important, as illustrated by several recent reports suggesting that modified clearance plays a role in the pathogenesis of VWD type 1 (19,20).
One of the very few parameters that has been reported to date to influence VWF clearance is the VWF glycosylation profile (12,(21)(22)(23)(24). For instance, by using an animal model it has been shown that, in the absence of the enzyme ST3Gal-IV, the half-life of endogenous VWF is reduced 2-fold (24). Moreover, in a patient group referred to the hospital for real or suspected bleeding disorder, reduced ST3Gal-IV-mediated sialylation was observed to be associated with reduced VWF plasma levels (24). These findings suggest that carbohydrate components play a role in the clearance of VWF.
In addition to the carbohydrate side chains, polypeptide regions are also likely to contribute to the interaction with clearance receptors. However, no such region has yet been identified within the VWF molecule. Thus, the molecular basis of VWF clearance is still an unexplored issue in view of the responsible receptors as well as in view of the regions within VWF that mediate binding to these receptors. To gain more insight into the molecular aspects of VWF clearance, an experimental model was developed employing mice genetically deficient for VWF. Using this model, the in vivo survival of truncated as well as mutated variants of VWF was compared with that of full-length VWF. These experiments indicate that the DЈ-D3 region of VWF and, in particular, the D3-domain residue Arg-1205 contribute to the clearance process of VWF.

EXPERIMENTAL PROCEDURES
Mice-The VWF-deficient mice (25) and wild-type mice used in this study were on a C57BL/6J background and were used between 8 and 12 weeks of age. Housing and experiments were done as recommended by French regulations and the experimental guidelines of the European Community.
Proteins-For biodistribution experiments, a highly purified, plasma-derived (pd) human VWF concentrate (a kind gift of Dr. C. Mazurier, LFB, Lille, France) was labeled with Na 125 I (Amersham Biosciences) using the IODO-GEN method (Pierce). Routinely, this method resulted in preparations containing 2-4 ϫ 10 5 cpm/g VWF. Free iodine in final preparations was Ͻ5% as determined by precipitation with 20% trichloroacetic acid. VWF preparations enriched in either high or low molecular weight multimers were obtained by fractionating an intermediate pure VWF concentrate (Hemate-P, Behring, Marburg, Germany) by gel filtration chromatography using Bio-Gel A-15m (Bio-Rad). The factor VIII light chain was kindly provided by Dr. K. Mertens (Sanquin Research, Amsterdam, The Netherlands). Recombinant GpIb␣ (residues 1-290) was expressed and purified as described (26). Purified recombinant VWF propeptide was a generous gift of Dr. H. C. Raaijmakers (Utrecht University, The Netherlands). Human placenta collagen type III (catalog number C-4407) and bovine albumin (fraction V) were purchased from Sigma. Polyclonal antibodies against VWF and horseradish peroxidase-conjugated polyclonal antibodies against VWF were obtained from Dako (Glostrup, Denmark).
Biodistribution of VWF-VWF deficient or wild-type mice were injected intravenously with radiolabeled pd-VWF (5 g/mouse, corresponding to 1-2 ϫ 10 6 cpm/mouse) diluted in phosphate-buffered saline containing 3% albumin. At different time points (3,15, and 30 min and 1, 2, and 4 h after injection) mice were anesthetized with tribromoethanol (0.15 ml per 10 g of body weight) and bled by retro-orbital venous plexus puncture. Subsequently, the mice were sacrificed, and various organs (heart, lungs, liver, spleen, stomach, kidneys and left upper leg) were collected, which were analyzed for the presence of radioactivity. Three mice were used per time point.
Clearance of VWF in Mice-VWF-deficient mice were injected intravenously with wt-VWF or variants thereof (5 g/mouse) diluted in phosphate-buffered saline containing 3% bovine albumin. At different time points (3 and 30 min and 1, 2, 4, 8, and 24 h after injection), mice were anesthetized with tribromoethanol (0.15 ml per 10 g of body weight), blood was collected by eye bleed, and plasma was prepared as described (33). Three to six mice were used for each time point, and each mouse was bled only once. VWF antigen levels were quantified in a previously described immunosorbent assay using polyclonal antibodies (32). Normal pooled plasma from 40 healthy donors was used as a reference, assuming that 1 ml of pooled plasma contains 10 g of VWF (35 nM, based on a monomeric molecular mass of 250 kDa). To quantify antigen values of VWF/DЈ-A3 and VWF/DЈ-D3, polyclonal rabbit antibodies directed against the DЈ-D3 region were used as capturing antibodies. In addition, a purified preparation of VWF/DЈ-A3 was used as a reference.
Surface Plasmon Resonance Analysis-Surface plasmon resonance (SPR) binding assays were performed employing a Biacore2000 system (Biacore AB, Uppsala, Sweden). Collagen and GpIb␣ binding experiments were performed as described previously (26,32). For FVIII light chain and VWF propeptide binding experiments, wt-VWF, VWF/ R1205H, and VWF/A1-CK were immobilized on a CM5 sensor chip using the amine-coupling kit as instructed by the supplier (Biacore AB, Uppsala, Sweden). VWF/A1-CK was used as a control, as it lacks the interactive sites for the FVIII light chain and propeptide. Binding to wt-VWF and VWF/R1205H-coated channels was corrected for binding to the VWF/A1-CK-coated control channel. Binding to this control channel was Ͻ5% of binding to both other channels. Binding of the FVIII light chain to immobilized VWF was performed in 125 mM NaCl, 25 mM Hepes (pH7.4) at 25°C with a flow rate of 10 l/min. Regeneration of the surface was performed by subsequent application of the same buffer containing 1 M CaCl 2 . The interaction between the propeptide and VWF was analyzed employing various buffers such as 125 mM NaCl, 25 mM Hepes (pH 7.4 or 6.5) or 125 mM NaCl, 25 mM (CH 3 ) 2 AsO 2 Na (pH 5.8 or 5.2) at 25°C with a flow rate of 10 l/min. Regeneration of the surface was performed by subsequent application of 10 mM taurodeoxycholate, 0.1 M Tris-HCl (pH 9). At least two distinct chips were used for each pH to assess binding. Before and after experiments at pH 6.5, 5.8, and 5.2, the binding of propeptide to immobilized VWF was also examined at pH 7.4 to establish the integrity of the VWF surface. The system was primed each time the flow buffer was changed.
Data Analysis and Statistics-Analysis of pharmacokinetic and SPR data was performed using the GraphPad Prism program (GraphPad Prism version 4.0 for Windows, GraphPad Software, San Diego, CA).
Data obtained from SPR analysis were used for the calculation of the affinity constants (K D ) as follows. Responses at equilibrium (R eq ) derived from sensorgrams were plotted against protein concentration. The resulting binding isotherms were subsequently fitted to the equation R eq ϭ R eq,max ϫ [protein]/(K D ϩ [protein]) to obtain K D . Clearance data were fitted to the biexponential equation C p ϭ Ae Ϫ␣t ϩ Be Ϫ␤t to obtain A, ␣, B, and ␤. C p refers to the amount of residual VWF antigen in plasma relative to the amount injected. These parameters were used to calculate mean residence time (MRT) employing the equation MRT ϭ (A/␣ 2 ϩ B/␤ 2 )/(A/␣ ϩ B/␤). Furthermore, ␣ and ␤ were used to calculate half-lives of the initial and terminal phase, respectively, employing t1 ⁄2 ␣ ϭ ln 2/␣ and t1 ⁄2 ␤ ϭ ln 2/␤. Statistical analyses were performed by the Student's unpaired t test using the GraphPad InStat program (GraphPad InStat version 3.0 for Windows, GraphPad Software, San Diego, CA).

VWF Is Predominantly
Targeted to the Liver-To gain more insight into the clearance process of VWF, we first investigated the biodistribution of this protein by injecting radiolabeled pd-VWF (ϳ2 ϫ 10 6 cpm/mouse) in VWF-deficient mice. Mice were sacrificed after intravenous tail injection (time points 3, 15, or 30 min and 1, 2, or 4 h). Subsequently, organs were collected and analyzed for the presence of radioactivity. At the initial time points, minor amounts of radioactivity (between 0.4 and 2% of the amount injected) were detected in heart, lungs, stomach, spleen, kidneys, and left upper leg (Fig. 1A). These amounts remained below 2.5% up until 4 h after injection, except that some radioactivity (up to 10%) started to accumulate in the stomach between 1 and 2 h after injection. In contrast, within 3 min after injection 9 Ϯ 0.9% of the amount injected appeared to be present in the liver. This amount increased to 17 Ϯ 2% at 15 min after injection, followed by a gradual decline to 8 Ϯ 0.4% between 0.5 and 4 h. A similar organ distribution was observed in wild-type mice (Fig. 1B).
These data indicate that intravenously administered VWF is primarily targeted to the liver.
Biphasic Clearance of Recombinant wt-VWF-To investigate the removal of VWF from plasma, VWF-deficient mice were injected intravenously with recombinant wt-VWF. At indicated time points after injection, mice were bled and plasma was analyzed for VWF antigen. The recovery appeared to be 79 Ϯ 14% in samples obtained 3 min after injection (Table I). Graphic representation of the antigen values revealed that the protein disappeared from plasma in a biphasic manner, characterized by a rapid initial phase and a slow secondary phase (Fig. 2). Pharmacokinetic analysis allowed the calculation of the apparent half-lives, which were 12.6 Ϯ 0.9 min and 3.0 Ϯ 0.9 h for the rapid and slow phase, respectively. Furthermore, MRT was calculated to be 2.8 Ϯ 0.7 h (Table I).
High and Low Molecular Weight Multimeric VWF Display Similar Survival in Mice-Preparations containing multimerized VWF are heterogeneous, as the number of included subunits may vary between molecules (from 2 to Ͼ20). To address whether the clearance of VWF depends on its extent of multimerization, we assessed the survival of pd-VWF fractions enriched in either high (predominantly 14-mers and higher) or low (predominantly dimers and tetramers) molecular weight multimers (see inset, Fig. 3) using VWF-deficient mice. These experiments revealed that both VWF preparations were cleared from plasma in a similar manner (Fig. 3). Indeed, similar pharmacokinetic parameters could be calculated from the experimental data (Table I). Moreover, the clearance of both high and low molecular weight multimers was similar to that of recombinant wt-VWF. The possibility was considered that the clearance of larger multimers was similar to that of smaller multimers, because these large multimers were converted into smaller forms before removal from circulation. However, analysis of the plasma samples demonstrated that the large multimeric forms were not converted into smaller fragments before disappearance from the plasma (data not shown). Thus, the rate of VWF clearance is independent of its multimeric size in our experimental model.
Opposite Effects of Amino-and Carboxyl-terminal Truncations on the Plasma Survival of VWF-In view of the complex protein structure of VWF, it was of further interest to study the contribution of the various regions of the VWF molecule to its clearance. To this end, a number of recombinant VWF variants with amino-and/or carboxyl-terminal truncations were constructed (Fig. 4A). Purified proteins were administered to VWF-deficient mice by tail vein injection. With regard to a fragment consisting of the A1-A2-A3 domains selectively (i.e. VWF/A1-A3), a clearance pattern similar to wt-VWF was observed (Fig. 4B). As summarized in Table I, similar pharmacokinetic parameters were obtained for this variant and wt-VWF. In contrast, a variant with an amino-terminal truncation (designated VWF/A1-CK) displayed an initial recovery that was significantly lower than that observed for wt-VWF (43 Ϯ 16% versus 79 Ϯ 14% (p ϭ 0.01) for VWF/A1-CK and wt-VWF, respectively). In addition, VWF/A1-CK was cleared from plasma in the initial phase more rapidly than wt-VWF (Fig.  4B). Indeed, both t1 ⁄2 ␣ and MRT were significantly reduced for the truncated protein compared with wt-VWF (Table I).
For the variants with a deletion of the carboxyl-terminal part of the protein, designated VWF/DЈ-A3 and VWF/DЈ-D3, the initial recovery was indistinguishable from that of wt-VWF. However, in the rapid phase of the clearance, both variants disappeared slower from plasma than wt-VWF (Fig. 4B). For instance, 13 Ϯ 1% residual wt-VWF remained in plasma 1 h after injection, whereas almost 2-fold more antigen could be detected for VWF/DЈ-A3 (21 Ϯ 1.6%; p ϭ 0.0018) and VWF/  (Table I). Taken together, these data suggest the following. (i) Multiple areas may be involved in the direct interaction with components that mediate the removal of VWF from the circulation; and (ii) particular regions within the VWF molecule may play a regulatory role in its survival. One example hereof might be the DЈ-D3 region, which seems to protect VWF from premature clearance.
Replacement of Arg-1205 by His Results in an Increased Clearance Rate-The observation that the DЈ-D3 domain appears to be involved in the clearance of VWF may be of relevance with regard to VWF mutants with amino acid replacements in this region as found in a number of patients suffering from type 1 VWD. For instance, mutation of the D3-domain residue Arg-1205 to His has been suggested to modulate the survival of VWF in patients (19). To study the direct contribution of Arg-1205 to the clearance of VWF, recombinant VWF/ R1205H was expressed in baby hamster kidney-furin cells. Analysis of the conditioned medium revealed a normal pattern of multimerization for this mutant (Fig. 5A). However, when the purified protein was injected intravenously in VWF-deficient mice, VWF/R1205H disappeared remarkably rapidly from the circulation (Fig. 5B). One hour after injection, 2 Ϯ 0.1% was detected in plasma, compared with 13 Ϯ 1% for wt-VWF (p Ͻ 0.0001). Data analysis showed that t1 ⁄2 ␣ and t1 ⁄2 ␤ as well as the MRT were significantly reduced compared with the values obtained for wt-VWF (Table I). In contrast, when we tested the in vivo survival of another VWF mutant, i.e. VWF/D2509G comprising a mutation in the RGD motif, it appeared that this mutant was cleared from plasma to the same extent as wt-VWF ( Fig. 5; Table I). These data are compatible with the view that replacement of Arg-1205 by His specifically results in an accelerated clearance of VWF.
Replacement of Arg-1205 by His Leaves Binding to Collagen and GpIb␣ Unaffected-The finding that replacement of Arg-1205 by His impairs the in vivo survival of VWF prompted us to study this mutant in more detail by testing a number of TABLE I Pharmacokinetic parameters describing the clearance of VWF and its variants from plasma VWF-deficient mice were injected intravenously with VWF (5 g/mouse). Mice were bled at various time points after injection (from 3 min to 24 h), and plasma was prepared and analyzed for the presence of VWF antigen (33). Each mouse was bled once, and 3-6 mice were used for each time point. Residual VWF concentrations were plotted against time, and data were fitted to a biexponential equation to obtain various pharmacokinetic parameters, including MRT (see "Experimental Procedures"). Recovery represents the percentage of VWF present in plasma 3 min after injection relative to the amount injected. Data are presented as mean Ϯ S.D. p values were calculated using Student's unpaired t-test and represent comparison to wt-VWF.  2. Biphasic disappearance of wt-VWF from plasma. VWFdeficient mice were injected intravenously with purified wt-VWF (5 g/mouse). At the indicated time points blood samples were taken. The amount of residual wt-VWF antigen in plasma was quantified in an immunosorbent assay (see "Experimental Procedures"). Plotted is the percentage of residual VWF antigen in plasma relative to the amount injected versus time after injection. Pharmacokinetic parameters derived from these data are summarized in Table I . At the indicated time points, blood samples were taken. The amount of residual wt-VWF in plasma was quantified in an immunosorbent assay (see "Experimental Procedures"). Plotted is the percentage of residual VWF antigen in plasma relative to the amount injected versus time after injection. Pharmacokinetic parameters derived from these data are summarized in Table I. Data represent mean values Ϯ S.D. of three mice for each time point. Inset, purified samples were analyzed for multimeric composition by agarose gel electrophoresis as described (32). Multimeric distribution of high and low molecular weight VWF preparations are shown in the left and right lanes, respectively. The number of subunits included in each band are indicated. functional parameters. First, the binding of purified wt-VWF and VWF/R1205H to human collagen type III and to a recombinant VWF-binding fragment of GpIb␣ (residues 1-290) in the presence of botrocetin was examined by SPR analysis. Binding isotherms using the response at equilibrium were used to calculate apparent affinity constants (K D ), which are summarized in Table II. The steady-state binding data revealed that wt-VWF and VWF/R1205H display similar affinity for GpIb␣ (K D ϭ 30 and 27 nM for wt-VWF and the mutant, respectively) and for collagen (K D ϭ 2.9 and 3 nM for wt-VWF and the mutant, respectively). Apparently, replacement of Arg-1205 by His leaves complex formation through its A1 (GpIb␣) and A3 (collagen) domains unaffected.
Suboptimal pH-dependent Association of Propeptide to VWF/R1205H-The DЈ-D3 region, which contains residue Arg-1205, comprises interactive sites for the VWF propeptide as well as the light chain of factor VIII. Therefore, binding of VWF/R1205H to both ligands was tested. Steady-state binding analysis showed that VWF/R1205H and wt-VWF are indistinguishable in terms of affinity for the FVIII light chain (K D ϭ 7.9  5. In vivo survival of mutated VWF variants. A, the multimer composition of wt-VWF and VWF/R1205H in conditioned medium was examined by agarose gel electrophoresis as described (31). R1205H, VWF/R1205H; NP, normal pooled plasma; WT, wt-VWF. B, plasma clearance of VWF/R1205H (open circles) and VWF/D2509G (closed squares) was determined as described for wt-VWF in the legend of Fig.  2. Plotted is the percentage of residual VWF antigen in plasma relative to the amount injected versus time after injection. For clarity, only data between 3 and 120 min are shown. Data obtained for wt-VWF are included for comparison (closed circles). Pharmacokinetic parameters derived from the complete data set (from 3 min to 24 h) are summarized in Table I. Data represent mean values Ϯ S.D. of 3-6 mice for each time point.

TABLE II Affinity constants for the interactions between VWF and its ligands
Complex formation between VWF and human collagen type III, recombinant GpIb␣ (residues 1-290), the FVIII light chain, or the VWF propeptide was examined by SPR steady-state analysis. Response at equilibrium was plotted against concentration, and nonlinear regression analysis was performed to obtain affinity constants (K D ) (see "Experimental Procedures"). Each analysis was based on the response of 5-7 different concentrations. Values represent calculated K D Ϯ S.D. and 8 nM for wt-VWF and VWF/R1205H, respectively; Table II). With regard to the VWF propeptide, an SPR-based assay was developed to study its interaction with VWF. VWF was immobilized (14 fmol/mm 2 ) and incubated with purified recombinant VWF propeptide (280 nM). The interaction was examined at various pH values (between pH 5.2 and 7.4), because the binding of the propeptide has been reported to be pH-dependent (8). Whereas binding at pH 7.4 was weak, binding of the propeptide to wt-VWF was markedly more prominent at lower pH values (Fig. 6A). Increased binding was mainly due to a dissociation rate constant that was reduced one order of magnitude at lower pH values. Analysis of steady state binding isotherms revealed that propeptide binds to wt-VWF with low affinity at pH 7.4 (Table II). Reduction of the pH increases the affinity by Ͼ20fold (K D ϭ 2.5 and 0.1 M at pH 7.4 and 5.2, respectively).
As for wt-VWF, a pH-dependent interaction with propeptide was also observed for VWF/R1205H (Fig. 6B). Moreover, similar association and dissociation curves were obtained for the mutant and wt-VWF at pH 7.4 and 6.5 (Fig. 6). As summarized in Table II, both proteins proved similar in propeptide binding in terms of affinity under these conditions (Table II). In contrast, a less pronounced increase in association of propeptide to VWF/R1205H was observed when pH was further decreased to pH 5.8 and 5.2 (Fig. 6B). Indeed, the calculated affinity constants were 2-fold higher for the mutant when compared with wt-VWF (pH 5.8, K D ϭ 0.2 versus 0.4 M; pH 5.2, K D ϭ 0.1 M versus 0.2 M, for wt-VWF and VWF/R1205H, respectively). Apparently, replacement of Arg-1205 by His is associated with suboptimal binding of propeptide at low pH. DISCUSSION Elevated plasma concentrations of hemostatic proteins are often associated with an increased thrombotic risk, whereas reduced levels may result in a bleeding tendency. This indicates that steady state plasma levels of these proteins need to be tightly regulated. Plasma levels represent a balance between biosynthetic and catabolic pathways. With regard to VWF, its biosynthetic pathway has been well studied in the last two decades (17,18), whereas relatively little is known concerning its clearance. In the present study we used VWFdeficient mice as a model to get more insight into the molecular basis of how VWF is cleared from the circulation.
Infusion studies employing radiolabeled VWF revealed that this protein was predominantly targeted to the liver, whereas little contribution of other organs (including kidneys) to the uptake of VWF was detected (Fig. 1). In time, accumulation of radioactivity was observed in the stomach. Similar delayed accumulation of radioactivity in the stomach has been described for other proteins as well, including annexin V and insulin growth factor-binding protein-3 (34,35). In case of the latter, this accumulated radioactivity represents small degraded peptides that are excreted from the liver and reabsorbed by the stomach. Considering the time course by which the radioactivity appears in liver and stomach, it seems likely that a similar process is present for VWF. Whether VWF is taken up in the liver in a receptor-mediated manner is unclear from the present study. However, in view of its relatively large size and rapid accumulation in the liver, it seems conceivable that VWF is removed from the circulation by an active, receptordependent process.
We considered the possibility that the clearance of VWF is partially dependent on the presence of VWF in the circulation or extravascular space. However, a similar biodistribution was observed in normal and VWF-deficient mice in terms of organ distribution and time course (Fig. 1). VWF is cleared from plasma in a biphasic manner characterized by a rapid initial phase and a slow terminal phase (Table I). The MRT was calculated to be 2.8 h. In this respect, human VWF is cleared from the murine circulation to the same extent as reported previously for human VWF in rats (22). The half-lives of VWF in these rodents are considerably shorter compared with the half-life of VWF in humans, which is ϳ14 h (36,37). Similar differences in the clearance rate between species have been reported for numerous other proteins, including coagulation factor IX. Whereas its half-life in humans is ϳ18 h (38), it is reduced 7-fold in mice (39). Probably, this reflects the basal metabolic rate, which is increased in mice.
One intriguing observation relates to the similar survival of high and low multimeric VWF preparations (Fig. 4), which suggests that the clearance of VWF is independent of its multimeric size in our experimental model. This observation is opposite to the findings that highly multimerized human VWF is cleared more rapidly than its low multimeric counterpart in rats as well as in dogs (22,40). It might be that high molecular weight multimers of human VWF are converted into smaller multimers in dogs and rats. Such conversion could not be observed in our experiments, nor could we detect proteolytic degradation of VWF in murine plasma (data not shown). This is in line with the notion that human VWF is not proteolyzed by the murine ADAMTS-13 protease. 2 Alternatively, high molecular weight human VWF may display enhanced binding to the platelet GpIb␣/IX/V receptor in rats and dogs but not in mice. Indeed, it has been reported that human VWF is not recognized by murine GpIb␣ (41). This multimerization-independent sur-2 C. V. Denis, unpublished observation. . Ligand solution was replaced with buffer to initiate dissociation. For both panels, the signal is indicated in resonance units (RU) and is corrected for aspecific binding to the VWF/ A1-CK control channel, which was Ͻ5% of the wt-VWF-and VWF/ R1205H-coated channels. vival of VWF in our experimental model offers the big advantage that we were able to directly compare the clearance of fully multimerized VWF to that of short recombinant variants (Fig. 4).
Monomeric and dimeric variants of VWF have been compared with wt-VWF to identify regions involved in the clearance of VWF. It appears that the clearance process is complex in that it involves different regions of the VWF molecule. Elements that can be gathered from our data are the following. (i) In view of the similar clearance rates for full-length wt-VWF and the monomeric VWF/A1-A3 fragment ( Fig. 4B; Table I), it seems likely that a receptor-recognition site is present in the A1-A3 region. (ii) With regard to the D4-CK region, we found that its absence is associated with a slower clearance (Fig. 4C), suggesting that deletion of these domains leads to the removal of a receptor-interactive site. However, when the D4-CK domains were linked to the A1-A3 region (i.e. deletion of the DЈ-D3 region), this protein was cleared more rapidly than the A1-A3 region itself ( Fig. 4B; Table I). This points to a regulatory role of the D4-CK region by enhancing A1-A3 domainmediated clearance. Our data do not allow us to distinguish between both possibilities, which are not mutually exclusive. (iii) Considering the clearance rate of the VWF/DЈ-D3 variant, it seems plausible that a receptor-recognition site is present in the DЈ-D3 region as well. Furthermore, the DЈ-A3 fragment (i.e. deletion of the D4-CK domains) is cleared less rapidly than the A1-A3 fragment, which could be compatible with the DЈ-D3 region serving a regulatory role by reducing the clearance rate.
The involvement of the DЈ-D3 region in VWF clearance is supported by a recent report by Casonato and co-workers (19). They observed that patients harboring an Arg-1205 to His mutation within the D3 domain display an increased disappearance of their endogenous VWF upon desmopressin treatment. Noteworthily, from that report it remained unclear whether the decreased survival is the result of the R1205H replacement, an additional VWF gene mutation (only part of the VWF coding sequence was analyzed), or even a gene defect in a receptor involved in the cellular uptake of VWF. Our studies using purified recombinant VWF/R1205H reveal clearly that in our experimental model this mutant is removed from the circulation more rapidly relative to wt-VWF (Fig. 5B), resulting in a 10-fold reduced MRT (Table I). This demonstrates that the R1205H mutation per se is associated with an increased clearance rate and explains why levels of VWF are reduced in patients harboring this mutation. Of interest is the recent description of a mutation in the D4 domain of VWF, which seems to have a similar effect on VWF clearance (42). It is to be expected that more VWF mutations causing abnormal clearance will be discovered. Such an effect has probably been underestimated to date because of the impossibility of testing this function in vitro. Only with the help of an animal model will it be possible to identify these new types of variants, which could be responsible for a more general mechanism of VWD.
The mechanism by which the Arg-1205 to His mutation affects the clearance of VWF remains to be determined. One explanation is that the mutation may alter an existing receptor-recognition site by increasing the affinity for that receptor and, subsequently, the clearance rate. It should be noted that our data do not distinguish between the mutation being part of a receptor-binding site or affecting a distantly located interactive site in an allosteric manner. In vitro analysis of binding to GpIb␣, collagen, and the FVIII light chain (mediated by the A1-, A3-, and DЈ-D3 domains, respectively) demonstrated that wt-VWF and the mutant VWF/R1205H are indistinguishable in complex formation with these ligands (Table II). These data show that the Arg-1205 to His mutation does not affect dra-matically the structure and function of VWF. However, with regard to propeptide binding, we found that the Arg-1205 mutation resulted in an impaired pH-dependent interaction ( Fig. 6 and Table II). As the Arg-1205 residue is not part of the recently described propeptide binding site (residues Thr-866 to Ile-907) (43), it is possible that the Arg-1205 to His mutation displays allosteric effects.
It is tempting to consider potential functional consequences of the reduced affinity for propeptide by VWF/R1205H in view of the physiological role of the VWF/propeptide interaction. VWF propeptide facilitates multimerization of VWF in the post-Golgi compartments and subsequently associates with highly multimeric VWF in storage organelles in endothelial cells (10). A reduced affinity for propeptide may reduce the stability of these storage organelles and result in a premature release of the highly multimerized VWF molecules into the circulation. This could contribute to the increased amount of ultralarge VWF multimers in plasma, as is found in some (but not all) patients with the R1205H mutation (44 -47).