Correction of Bleeding Symptoms in von Willebrand Factor–Deficient Mice by Liver-Expressed von Willebrand Factor Mutants
Objective— von Willebrand Factor (vWF) structure-function relationship has been studied only in vitro. To investigate the physiological importance of particular vWF domains, we have introduced mutations into murine vWF (mvWF) cDNA inhibiting vWF binding to glycoprotein (Gp) Ib, GpIIbIIIa, and to fibrillar collagen.
Methods and Results— We delivered wild-type (WT) or mutant mvWF cDNA into vWF-deficient (Vwf−/−) mice using hydrodynamic injection and assessed whether hemorrhagic symptoms could be corrected. Hydrodynamic gene transfer resulted in high expression of plasma mvWF 24 hours after injection (438±63% for 50 μg of cDNA). Factor VIII activity was normalized in Vwf−/− mice injected with mvWF cDNA and multimerization was achieved. Bleeding time was corrected after injection of WT mvWF cDNA in Vwf−/− mice whereas noninjected mice did not stop bleeding. Injection of the GpIIbIIIa and the collagen binding mutants in Vwf−/− mice also resulted in a correction of bleeding time whereas mice injected with the GpIb binding mutant were bleeding for as long they were observed, although blood loss was decreased compared with noninjected mice (61±21 μL versus 232±63 μL).
Conclusion— Our model allows the rapid in vivo evaluation of specific mutations on plasma vWF function.
von Willebrand Factor (vWF) is a multimeric plasma glycoprotein (Gp) playing an essential role in adhesion and aggregation of platelets at sites of vascular damage. At high shear stress, vWF acts as a molecular bridge between components of the subendothelium such as collagen and platelet receptors GpIb and GpIIbIIIa.1 Furthermore, vWF acts as a carrier protein for coagulation factor VIII (FVIII) and protects it from premature clearance. vWF is produced and stored in endothelial cells and megakaryocytes and is released into the plasma as a series of multimers with a molecular weight ranging from 500 to 20 000 kDa, the largest multimers being the most biologically active. The mature subunit of vWF contains 2050 amino acid (aa) residues organized in 5 types of repeated domains arranged in the order D′-D3-A1-A2-A3-D4-B1-B2-B3-C1-C2-CK.2 A number of functional domains have been identified on vWF sequence, mostly through in vitro approaches. The A3 domain is important for vWF binding to collagen3 whereas the A1 domain contains the binding site of vWF to platelet GpIb.4 The binding site of vWF for its other platelet receptor, GpIIbIIIa, has been located near the carboxyl-terminal end of the C1 domain encompassing the RGD sequence.5,6
Further insight in the structure-function relationship of vWF was gained from the analysis of patients suffering from von Willebrand disease (VWD), the heterogeneous bleeding disorder resulting from defects in vWF gene. VWD can be classified in quantitative deficiencies (types 1 and 3) or qualitative abnormalities (type 2).7,8 Point mutations represent the molecular basis underlying VWD type 2.9 These molecular variants have proven useful in establishing the link between a specific functional domain and its physiological relevance. However, this approach is limited by the identification of patients. So far, no patient has been identified with an abnormal binding of vWF to GpIIbIIIa and only one patient with defective collagen binding has been reported, with slightly prolonged bleeding time.10 The question then arises whether such mutations exist or whether they lead to bleeding symptoms.
To bypass this limitation, we have developed a model to test the physiological importance of specific vWF mutations in vivo. This model is derived from the vWF-deficient mice which represent a model of VWD type 3.11 Various point mutations were introduced into murine vWF (mvWF) cDNA and injected into Vwf−/− mice by hydrodynamic tail vein injection.12,13 Transient high expression of wild-type (WT) or mutated vWF protein in the plasma allowed us to assess the efficacy of these different variants to correct the hemorrhagic symptoms of Vwf−/− mice.
Please see http://atvb.ahajournals.org for details.
Five-to 8-week-old vWF-deficient (Vwf−/−)11 and WT mice on a C57BL/6 background, bred in the IN SERM u770 animal facility, were used throughout this study. Housing and experiments were done as recommended by French regulations and the experimental guidelines of the European Community.
Hydrodynamic Injection of Plasmid DNA
Different amounts of plasmid DNA diluted in a volume of saline (0.9% NaCl) equivalent to 10% of the bodyweight were injected into the mouse tail vein within 5 seconds.12–14 No liposomes were used to facilitate the transfer of the cDNA.
Blood Collection and Determination of mvWF Levels by ELISA
Mice were anesthetized by intraperitoneal injection of tribromoethanol (0.15 mL per 10 g of bodyweight) and blood was collected by eye bleed in trisodium citrate (9 volumes of blood to 1 volume of 0.138 mol/L trisodium citrate). To obtain platelet poor plasma, blood samples were centrifuged at 1000g for 20 minutes at 22°C.
Plasma vWF concentration was measured by ELISA15 using a polyclonal antibody anti-human vWF (Dako) and a horseradish peroxydase conjugated polyclonal antibody anti-human vWF (Dako). Normal pooled plasma from 15 C57Bl/6 WT mice was used as a reference and set at 100%. Results are expressed as percentage of normal mvWF level.
vWF Structure Analysis
The multimeric structure of vWF was analyzed by 0.1% sodium dodecyl sulfate (SDS) and 1% agarose (GE Healthcare) gel electrophoresis.16 Multimers were visualized using an alkaline phosphatase-conjugated anti-human vWF polyclonal antibody.
Mouse tail-bleeding time was performed as described.11 Briefly, nonanesthetized mice were placed in a restraining device and 3 mm of the distal tail were cut using a scalpel. The amputated tail was immersed immediately in physiological saline at 37°C and bleeding time was measured from the moment of transection until first arrest of bleeding. Observation was stopped at 600 seconds when bleeding did not cease.
Quantitative Measurement of Blood Loss
Blood loss after mouse tail transection was measured according to Turecek et al17 with slight modifications. Mice were anesthetized by intraperitoneal injection of tribromoethanol and the tail was cut with a scalpel, using a template resulting in sections of fixed diameter (1.5 mm). The blood escaping from the wound was absorbed on a filter paper every 30 seconds for 10 minutes while being careful not to dislodge the forming clot. Extraction and lysis of erythrocytes from the filter paper were performed with 20 mL of 0.04% NH4OH solution for 2 hours at 22°C and the amount of hemoglobin was obtained by reading the absorbance at 416 nm. The volume of blood in each sample was calculated from a standard curve, which was obtained by dropping defined volumes of WT mouse blood to a filter paper and extracting hemoglobin as described above.
Determination of FVIII, FIX and FX Activities
FVIII, FIX, and FX activities were determined by a clotting assay. Plasma, obtained from mvWF cDNA-injected Vwf−/− mice, was mixed at different dilutions with Owren buffer (Diagnostica Stago). Normal pooled mouse plasma was used as a standard and set at 100%. 50 μL of diluted mouse plasma were incubated with 50 μL of human FVIII, FIX, or FX-deficient plasma (Diagnostica Stago) and 50 μL of activated partial thromboplastin time (aPTT) reagent (Diagnostica Stago) at 37°C for 4 minutes. Coagulation was initiated by the addition of 50 μL of pre-warmed calcium chloride (25 mmol/L). Results are reported as percentage of normal murine FVIII, FIX, or FX activity.
Data are expressed as mean values±SEM. Statistical analyses were performed by the Student unpaired t test or by 1-way analysis of variance (ANOVA). A probability value below 0.05 was considered significant. For ANOVA, in case of P<0.05, pair-wise comparisons against the control group were made. Corrections for multiple comparisons were made according to Dunnet. Bleeding times that exceeded 600 seconds were handled as 600 seconds.
Gene Expression of mvWF and Normalization of FVIII Activity After Hydrodynamic Gene Transfer in Vwf−/− Mice
To evaluate the dose dependency of mvWF expression after hydrodynamic injection–based gene delivery, male and female Vwf−/− mice were injected with 1, 2, 5, 10, 50, or 100 μg of pcDNA6-mvWF and blood was collected 24 hours after injection. As shown in Figure 1A, hydrodynamic gene transfer resulted in high expression of plasma mvWF antigen and interestingly, gene expression of mvWF was twice as potent in males than in females. In both sexes, gene expression was dependent on the amount of plasmid DNA injected and reached a plateau at 50 μg of pcDNA6-mvWF with mvWF mean expressions of 438±63% in males and 199±30% in females. We used only male mice for all subsequent experiments. In Vwf−/− mice, FVIII level is strongly reduced as a result of the lack of protection provided by vWF. To demonstrate an eventual normalization of FVIII activity, Vwf−/− mice were injected with various concentrations of mvWF cDNA and plasma was collected 24 hours later. In noninjected Vwf−/− mice, FVIII activity was less than 15% of WT mice (Figure 1B). In contrast, Vwf−/− mice injected with doses of 2 μg or higher of mvWF cDNA presented a normalization of FVIII activity with an activity close to 100% (Figure 1B).
We next assessed the time course of mvWF expression and FVIII activity after injection of 50 μg of mvWF cDNA in Vwf−/− mice. The peak plasma vWF level was reached 24 hours after injection and then decreased gradually to nearly undetectable levels within 96 hours (supplemental Figure I, available online at http://atvb.ahajournals.org). At 96 hours, FVIII activity was still 5 times higher than in noninjected mice. We have also investigated whether activity levels of 2 other coagulation factors, factor IX and factor X, were affected by hydrodynamic injection. Mean activity levels of 90.5±6.0% for factor IX and 134.3±13.6% for factor X were measured in Vwf−/− mice after injection of mvWF cDNA, therefore resulting in minimal modification in the normal activity of these coagulation factors.
Altogether, these results confirm that hydrodynamic gene transfer is an efficient technique to express mvWF in mice.
Multimeric Analysis of Plasma mvWF
To analyze the multimeric pattern of mvWF after hydrodynamic gene delivery, we injected 50 μg of mvWF cDNA in Vwf−/− mice and plasma was collected 24 hours later. 1% SDS agarose gel electrophoresis revealed the presence of multimerized mvWF (Figure 2). The balance between high and low molecular weight multimers was slightly different compared with normal murine plasma. Injected Vwf−/− mice presented a moderate decrease in high molecular weight multimers.
Effect of mvWF on Bleeding Time in Vwf−/− Mice After Hydrodynamic Gene Transfer
To assess whether hydrodynamic gene transfer of mvWF cDNA could correct the bleeding time, we injected various amounts of mvWF cDNA in Vwf−/− mice. Bleeding time was determined by transection of the tail 24 or 48 hours after injection. As shown in Figure 3, WT mice stopped bleeding in less than 100 s whereas noninjected Vwf−/− were bleeding during the complete observation period (600 s). To check whether hydrodynamic injection per se could affect bleeding time, we injected 50 μg of pcDNA6 vector at high pressure in WT mice and measured bleeding time 24 hours after injection. Surprisingly, we observed a tendency toward an increase in bleeding time in injected mice compared with noninjected mice (131±44 s versus 75±10, P=0.057; Figure 3). Regarding Vwf−/− mice, administration of 2 μg or 10 μg of mvWF cDNA had no significant effect on bleeding time. In contrast, the 2 highest doses, 50 μg and 100 μg of mvWF cDNA per mouse, significantly corrected bleeding time 24 hours after injection (P<0.001) and resulted in mean bleeding times of 169±69 s and 203±42 s, respectively. At these 2 doses the difference between WT-injected mice and Vwf−/− injected mice was no longer significant (P=0.72 for 50 μg and P=0.26 for 100 μg). At 48 hours, results were more heterogeneous with some mice managing to control their bleeding whereas others did not.
Expression and Characterization of mvWF Mutants
Vwf−/− mice injected with 50 μg of mutated mvWF cDNA presented mvWF antigen expression levels similar to those observed after injection of WT mvWF cDNA (supplemental Table I). Furthermore, the multimeric pattern of mutated mvWF was comparable to the pattern seen in Vwf−/− mice injected with WT mvWF cDNA (Figure 2). To ascertain that the GpIb (K1362A) and collagen (D1742A, S1783A, H1786A) binding mutations introduced in the mvWF cDNA efficiently abolished the binding of mvWF to platelet GpIb and collagen, respectively, we evaluated their binding to these ligands in vitro. Both mutants showed a complete absence of binding to their respective ligands whereas WT mvWF presented a dose-dependent binding to human collagen and was able to induce murine platelet aggregation in the presence of ristocetin (supplemental Figure II).
Effect of Mutated mvWF on Bleeding Time in Vwf−/− Mice After Hydrodynamic Gene Transfer
To examine whether RGG mutant or GpIb and collagen binding mutants could correct the bleeding time in Vwf−/− mice, we injected 50 μg of mutated mvWF in Vwf−/− mice. Injection of collagen binding mutant and RGG mutant in Vwf−/− mice resulted in a correction of bleeding time, with values of 189±60 s and 139±56 s, respectively, similar to the correction obtained with injection of WT mvWF cDNA (Figure 4). In contrast, mice injected with the GpIb binding mutant were bleeding for as long as they were observed (>600 s) except one mouse, which stopped bleeding at 312 s after tail transection.
Although mice injected with the GpIb binding mutant kept on bleeding, they seemed to loose less blood than noninjected mice. We therefore measured the volume of blood lost over a period of 10 minutes. As shown in Figure 5, nontreated Vwf−/− mice lost on average 232±30 μL of blood and were bleeding for as long as they were observed whereas WT mice lost 13±3 μL of blood and were bleeding between 2 and 4.5 minutes in this setting. Vwf−/− mice injected with 50 μg of mvWF cDNA significantly reduced their blood loss to 19±5 μL and were no different from WT mice in terms of bleeding time. Interestingly, although Vwf−/− mice injected with mvWF GpIb mutant did not stop bleeding, they bled significantly less (60±11 μL) than noninjected Vwf−/− mice. Mice injected with the vector alone or with 2 μg of mvWF cDNA lost as much blood as noninjected mice.
vWF is a very large protein and a number of functional domains involved in the binding to different ligands have been localized on its sequence, such as binding sites for platelet GpIb and GpIIbIIIa, FVIII, and collagen.1,2 vWF binding to GpIb is of critical importance for vWF function as evidenced by the identification of numerous VWD patients carrying mutations in the GpIb binding domain of vWF.8 Similarly, VWD patients carrying mutations in the FVIII binding domain of vWF express a bleeding phenotype close to hemophilia A, with decreased FVIII levels.18 However, the clinical relevance, in terms of hemorrhagic symptoms, of most other vWF interactions with various ligands has not been verified because of the combined lack of VWD patients with corresponding mutations and of adequate animal models. In this report using the hydrodynamic injection technique, we describe a murine model allowing the transient expression of different mutant vWF in the plasma and their capacity to correct bleeding symptoms of vWF-deficient mice.
We first validated our technique with WT vWF. Because of the absence of binding of murine GpIb to human vWF, we have cloned mvWF cDNA and ligated it into pcDNA6-V5-His expression vector. On injection of this construct in vWF-deficient mice, we observed a rapid and strong expression of vWF antigen in plasma. The antigen peak was observed 24 hours after injection, followed by a rapid decline of expression, consistent with previous reports using plasmids with the CMV promoter.19 Injection of 50 μg of mvWF cDNA resulted in plasma vWF antigen levels as high as 200% to 440% depending on the sex of the mice. These levels are much higher than those reported in the study by Pergolizzi et al where the same dose of mvWF cDNA only led to vWF expression levels of 40%.20 This variation is likely attributable to the different cDNA used in both studies. These authors have used a cDNA leading to expression of a murine vWF with an Arg instead of a Cys in position 799, a protein associated with intracellular retention and impaired multimerization.21 The 2-fold higher expression in males compared with females is interesting. Using human factor IX as a reporter gene, Feng et al have also reported a similar difference between sexes.22 A potential hypothesis could relate to anatomic or liver damage differences between males and females.
Further analysis of the vWF-deficient mice injected with WT mvWF cDNA revealed a normalization of FVIII levels and the presence of fully multimerized vWF although there was a moderate decrease in the highest molecular weight forms. After hydrodynamic gene transfer, the gene of interest is being expressed by liver hepatocytes (Figure 6).12,13 Because vWF is normally produced by endothelial cells and megakaryocytes, the slight change in multimeric profile is likely attributable to this switch in expression site. A previous report stating that administration of vWF cDNA in vWF-deficient mice results in a multimer pattern similar to that seen in WT mice shows only the presence of 5 multimer bands in the treated mice (as opposed to 15 in the present study), making it difficult to evaluate the quality of multimerization.20
The functionality of our model was demonstrated by the correction of bleeding time observed in vWF-deficient mice 24 hours after injection of 50 μg of WT mvWF cDNA. Despite high vWF antigen levels, the bleeding time in these mice remained slightly prolonged compared with WT mice. Similarly, injection of 10 μg of mvWF cDNA did not allow consistent rescue of the bleeding phenotype despite leading to expression levels around 200%. Several reasons can account for these observations: (1) hydrodynamic injection itself leads to longer bleeding time even in WT mice; (2) the very high molecular weight multimers are slightly decreased; (3) vWF is present only in the plasma. The absence of vWF in platelets, endothelial cells, and subendothelium may prevent an optimal rescue of the bleeding phenotype. However, plasma vWF appears to be sufficient to promote hemostasis in a mouse model. Transfusion of vWF into a pig with VWD also results in the correction of bleeding time showing that efficient hemostasis can be achieved in the absence of endogenous endothelial vWF if plasma vWF is present.23
The exclusive plasma expression of vWF in our model mimics the clinical situation where VWD patients are treated with vWF concentrates that replenish only the plasma compartment. Our model thus allows the direct comparison of hemostatic capacities of WT versus mutant plasma vWF by testing the ability of various mutants to correct bleeding time in vWF-deficient mice.
For the collagen binding mutant, we mutated residues in the A3 domain identified as being critical for binding to collagen types I and III.3 These 3 residues are conserved in the murine vWF sequence. Despite the absence of binding to collagen in vitro, this mutant proved able to correct bleeding time in vWF-deficient mice as efficiently as WT-vWF. Although this result suggests that vWF binding to collagens types I and III is not critical for efficient hemostasis, it does not exclude collagens as being relevant physiological ligands for vWF in the subendothelium. Indeed, vWF also binds collagen type VI through its A1 domain24,25 and it has been recently demonstrated that vWF-A1 domain can substitute for the A3 domain in supporting binding to collagens type I and III.26 These A1-mediated interactions as well as potential vWF interactions with other subendothelial constituents appear sufficient to support the formation of the hemostatic plug.
Mutation of the vWF RGD sequence into RGG was previously shown to abolish vWF binding to platelet GpIIbIIIa.6 This sequence is also conserved in murine vWF. We report here that this interaction is not strictly required to maintain a normal bleeding time in mice. This could explain why VWD patients with mutations in the RGD sequence have never been identified. Binding of GpIIbIIIa to its other ligands such as fibrinogen and fibronectin can thus compensate for the lack of vWF-GpIIbIIIa binding in our model.
The third mutant we tested in this study is the GpIb binding mutant K1362A. As previously described4 the human mutant does not bind human GpIb and we also checked that the murine mutant does not bind mouse GpIb. When expressed in the plasma, this mutant did not correct bleeding time in vWF-deficient mice, confirming the critical importance of the vWF-GpIb interaction for efficient primary hemostasis in vivo. However, vWF-deficient mice expressing this mutant seemed to lose less blood than their noninjected counterparts. Quantification of blood loss during a 10-minute period confirmed this observation. This rather surprising result can be examined in light of recent findings showing that GpIb deficiency is more severe than vWF deficiency in a murine model of arterial thrombosis, leading to the conclusion that GpIb is probably able to bind to a ligand different than vWF.27
In summary, we have described a model allowing the examination of the in vivo function of plasma vWF mutants and derivatives. Provided that mouse-human differences are taken into account, this method allows a rapid evaluation of the causative effect of various mutations found in VWD patient while avoiding the lengthy and expensive process of producing and purifying the recombinant mutant protein. The transient expression of the vWF transgene has so far limited our phenotypic analysis to bleeding time measurements but hydrodynamic injection using expression vectors with a liver-specific promoter has been described as leading to sustained expression of the transgene.19 Using such vectors will allow more in depth analysis of various vWF mutants in vivo.
We thank Nathalie Ba for help with preparing the liver sections.
Sources of Funding
This study was supported by Institut National de la Santé et de la Recherche Médicale and by a PhD thesis grant from Institut Servier and from Association Nationale de la Recherche Technique (to I.M.).
Original received August 13, 2007; final version accepted December 20, 2007.
Romijn RA, Westein E, Bouma B, Schiphorst ME, Sixma JJ, Lenting PJ, Huizinga EG. Mapping the collagen-binding site in the von Willebrand factor-A3 domain. J Biol Chem. 2003; 278: 15035–15039.
Matsushita T, Meyer D, Sadler JE. Localization of von willebrand factor-binding sites for platelet glycoprotein Ib and botrocetin by charged-to-alanine scanning mutagenesis. J Biol Chem. 2000; 275: 11044–11049.
Beacham DA, Wise RJ, Turci SM, Handin RI. Selective inactivation of the Arg-Gly-Asp-Ser (RGDS) binding site in von Willebrand factor by site-directed mutagenesis. J Biol Chem. 1992; 267: 3409–3415.
Lankhof H, Wu YP, Vink T, Schiphorst ME, Zerwes HG, de Groot PG, Sixma JJ. Role of the glycoprotein Ib-binding A1 repeat and the RGD sequence in platelet adhesion to human recombinant von Willebrand factor. Blood. 1995; 86: 1035–1042.
Sadler JE, Budde U, Eikenboom JC, Favaloro EJ, Hill FG, Holmberg L, Ingerslev J, Lee CA, Lillicrap D, Mannucci PM, Mazurier C, Meyer D, Nichols WL, Nishino M, Peake IR, Rodeghiero F, Schneppenheim R, Ruggeri ZM, Srivastava A, Montgomery RR, Federici AB. Update on the pathophysiology and classification of von Willebrand disease: a report of the Subcommittee on von Willebrand Factor. J Thromb Haemost. 2006; 4: 2103–2114.
Denis C, Methia N, Frenette PS, Rayburn H, Ullman-Cullere M, Hynes RO, Wagner DD. A mouse model of severe von Willebrand disease: defects in hemostasis and thrombosis. Proc Natl Acad Sci U S A. 1998; 95: 9524–9529.
Lenting PJ, Westein E, Terraube V, Ribba AS, Huizinga EG, Meyer D, de Groot PG, Denis CV. An experimental model to study the in vivo survival of von Willebrand factor. Basic aspects and application to the R1205H mutation. J Biol Chem. 2004; 279: 12102–12109.
Pergolizzi RG, Jin G, Chan D, Pierre L, Bussel J, Ferris B, Leopold PL, Crystal RG. Correction of a murine model of von Willebrand disease by gene transfer. Blood. 2006; 108: 862–869.
Lenting PJ, de Groot PG, De Meyer SF, Vanhoorelbeke K, Pruss C, Lillicrap D, Marx I, Denis CV. Correction of the bleeding time in von Willebrand factor (vWF)-deficient mice using murine vWF. Blood. 2007; 109: 2267–2268.
Denis C, Baruch D, Kielty CM, Ajzenberg N, Christophe O, Meyer D. Localization of von Willebrand factor binding domains to endothelial extracellular matrix and to type VI collagen. Arterioscler Thromb. 1993; 13: 398–406.
Mazzucato M, Spessotto P, Masotti A, De Appollonia L, Cozzi MR, Yoshioka A, Perris R, Colombatti A, De Marco L. Identification of domains responsible for von Willebrand factor type VI collagen interaction mediating platelet adhesion under high flow. J Biol Chem. 1999; 274: 3033–3041.
Bergmeier W, Piffath CL, Goerge T, Cifuni SM, Ruggeri ZM, Ware J, Wagner DD. The role of platelet adhesion receptor GPIbalpha far exceeds that of its main ligand, von Willebrand factor, in arterial thrombosis. Proc Natl Acad Sci U S A. 2006; 103: 16900–16905.