Mouse models of von Willebrand disease


Cécile V. Denis, INSERM U770, 80 rue du Général Leclerc, 94276 Le Kremlin-Bicêtre cedex, France.
Tel.: +33 1 49 59 56 05; fax: +33 146 729472.


Summary.  von Willebrand disease (VWD), caused by quantitative or qualitative abnormalities in von Willebrand factor (VWF) is considered the most common inherited bleeding disorder in humans. Mild and severe quantitative defects in VWF cause VWD types 1 and 3 respectively, whereas qualitative abnormalities induce VWD type 2. VWD has also been diagnosed in a number of animal species such as dogs, pigs, cats and horses, as a result of naturally occurring mutations. More recently, murine models have drawn a great deal of attention. Their small size along with their well-defined genetic background makes them ideal tools to study the in vivo function of VWF. The most commonly used model is the VWF-deficient mouse engineered through homologous recombination. However, models resulting from changes in modifier genes indirectly affecting VWF have also been described. These various models have proven very useful in elucidating some aspects of VWF biology not easily addressed through in vitro approaches.

von Willebrand disease (VWD) is characterized by mucocutaneus bleeds, prolonged bleeding time and an autosomal pattern of inheritance [1]. The protein involved in VWD is von Willebrand factor (VWF), a multimeric glycoprotein the size of which varies from 500 kDa for dimers to 20 000 kDa for larger multimers. VWF is synthesized in megakaryocytes and endothelial cells. Megakaryocyte-derived VWF is stored in platelets alpha-granules, whereas VWF in endothelial cells can be directly secreted into the circulation or stored in Weibel–Palade bodies (WPBs). In case of vascular injury, VWF plays a key role in mediating platelet adhesion to the exposed subendothelium. Additionally, VWF mediates platelet-platelet interactions and carries coagulation factor VIII (FVIII). There are two main categories of murine models for VWD based on their origin, either spontaneous or genetically-engineered (see Table).

Table 1.   Murine models of VWD
ModelDescriptionEffect on VWFVWD typeContribution to researchReferences
RIIIS/J strainMutation in the glycosyltransferase B4GALNT2Impaired glycosylation leading to accelerated clearance
Levels decreased to 30%–50%
1VWF clearance
Intimal hyperplasia
ST3Gal-IV−/− miceGenetically induced modelImpaired sialylation leading to accelerated clearance
Levels decreased to 50%
1VWF clearance[8]
Asgr-1−/− miceGenetically induced modelDecreased clearance
Levels up to 150%
VWF clearance[9]
CASA/RkJ strainIdentification of several gene modifiers Mvwf2, Mvwf3 and Mvwf4R2657Q polymorphism leading to increased biosynthesisVWF synthesis[4,5]
VWF+/− miceHeterozygous state of the genetically engineered modelPresence of only one allele1Treatment of VWD with IL-11[26]
VWF−/− miceGenetically engineered modelComplete absence of VWF3VWF clearance[25]
Thrombus formation[6,11]
Ischemic stroke[19]
Creation of new transient models[7]
VWFR1326HGenetically engineered with a mutation at position 1326Does not bind to mouse GPIb. Binds to human GPIbScreening for anti platelet therapy[28]

Murine models for VWD

Naturally occurring models

RIIIS/J strain.  Plasma VWF levels among inbred mouse strains are highly variable [2]. Observation of a prolonged bleeding time (>15 min) first drew attention to the RIIIS/J strain. Further analysis showed a normal multimeric distribution of VWF with an antigen level down to 33–55% of the level found in normal mouse plasma. Ristocetin cofactor and FVIII activities were proportionately decreased. Genetic linkage analysis localized the defect at a locus distinct from the murine Vwf gene, leading to the discovery of the first gene modifier of VWF in mice named Mvwf1 for Modifier of VWF 1 [2]. Mvwf1 was identified as B4galnt2 a gene coding for an N-acetylgalactosaminyltransferase. In the RIIIS/J strain, a mutation in B4galnt2 causes a switch in the expression of the gene product from intestinal epithelium to vascular endothelium. Vascular B4galnt2 expression leads to aberrant glycosylation of VWF and to its rapid clearance from plasma via the asialoglycoprotein receptor (ASGPR). Interestingly, Mvwf1 was also found in 13 inbred mice including wild-derived strains, suggesting that this low VWF phenotype may confer an advantage to mice [3]. Such a mechanism could also contribute to the high prevalence of VWD in humans.

CASA/RkJ strain.  The CASA/RkJ strain exhibits VWF levels 8-fold higher than the A/J strain. A significant proportion of this difference is associated with Mvwf2, a gene modifier linked to the Vwf locus and leading to increased VWF biosynthesis/secretion in the CASA/RkJ strain [4]. Mvwf2 corresponds to a single nucleotide polymorphism associated with an amino acid change (R2657Q) in the Vwf gene. Two other putative modifiers, also correlating with increased VWF levels were identified in the same strain: Mvwf3 on chromosome 4 and Mvwf4 on chromosome 13 [5].

Genetically engineered mice

VWF-deficient mice.  Available for the last 10 years, mice deficient in VWF (VWF−/−) represent a good model of human type 3 VWD [6]. Disruption of the gene by insertion of a neomycin resistance gene resulted in the absence of detectable VWF or VWF propeptide as well as an 80% reduction of FVIII levels compared to wild type mice. VWF−/− mice present a highly prolonged bleeding time. Heterozygous mice display antigen levels down to 50%, FVIII levels between 50 and 60% of wild-type mice and a normal bleeding time.

Recently, the VWF−/− mice provided a basis to create transient models expressing VWF variants using hydrodynamic injection. In this model, a rapid injection of the murine Vwf cDNA under high pressure in the tail vein induces the intake of the gene by the liver, and results in expression and release of functional VWF into the circulation [7].

Other gene targeted mice.  Phenotypic analysis of mouse models engineered for research purposes independent of VWF revealed some surprising results, uncovering new VWF gene modifiers. For example, deficiency in the ST3Gal-IV sialyltransferase (ST3Gal-IV) induces a dominant 50% reduction in VWF plasma levels accompanied by a similar decrease in FVIII and a prolonged tail bleeding time [8]. By masking galactose linkages on VWF, ST3Gal-IV normally prevents accelerated clearance of VWF by the ASGPR.

Conversely, gene targeting of the Ashwell receptor (Asgr-1) leads to a 1.5-fold increase in VWF and FVIII plasma levels as well as a reduction in tail bleeding time, suggesting that this hepatocyte ASGPR contributes to VWF homeostasis [9]. Combined deficiency of ST3Gal-IV and Asgr-1 resulted in normalization of VWF plasma levels and bleeding time in mice.

Contribution of VWD murine models for investigation of VWF functions

VWF in platelet adhesion and thrombus formation

In a ferric chloride-induced injury in mesenteric arterioles, VWF−/− mice exhibit delayed platelet adhesion and thrombus formation. Thrombi can still form, but the persistence of high shear channels makes these thrombi ineffective in promoting vessel occlusion [10]. Interestingly, VWF also appears crucial in thrombus formation in mesenteric venules [11]. In a different model using laser-induced injury, VWF absence has no significant effect on thrombus growth or on platelet activation [7,12]. Further insight into the molecular basis of VWF role in thrombus formation was gained from VWF−/− mice transiently expressing VWF mutants in plasma. Mice expressing VWF mutants unable to bind to fibrillar collagens or to platelet glycoprotein (GP) Ib and GP IIbIIIa were tested in the ferric chloride-induced thrombosis model. All three interactions appeared critical for optimal thrombus formation [7]. In contrast, in the tail clip bleeding time assay, only the inhibition of the VWF-GPIb interaction led to absence of correction.

A distinct role for VWF in platelet adhesion independently of vessel wall injury, was also uncovered using VWF-deficient mice. Release of VWF from WPBs following endothelial stimulation leads to immediate and transient platelet interaction with the endothelium, allowing recruitment of platelets to sites of inflammation [13].

Other functions of VWF

VWF is likely to play a role in diseases where platelet adhesion and thrombus formation are involved. For example, studies performed in VWF-deficient pigs had suggested that absence of VWF was protective against atherosclerosis, a result that was later contradicted [14]. To clarify this issue, further studies were performed in VWF−/− mice backcrossed to atherosclerosis-sensitive strains, low-density lipoprotein receptor-deficient mice or apolipoprotein E-deficient mice. In the absence of VWF, the formation of atherosclerotic lesions was significantly delayed and their distribution was modified [15]. In patients with type 3 VWD, ultrasonography studies also revealed the presence of atherosclerosis lesions but whether their formation was delayed, like in mice, is not known.

The crucial role of VWF in the pathogenesis of thrombotic thrombocytopenic purpura (TTP) was addressed using VWF−/− mice backcrossed on a TTP-sensitive background, i.e. mice deficient in VWF-cleaving protease, ADAMTS13. Absence of VWF led to a complete protection against TTP onset induced by shigatoxin [16].

VWF−/− mice were also used to evaluate the implication of VWF in metastasis dissemination based on the observations that several receptors for VWF such as GPIIbIIIa, GPIb or vβ3 have been detected on tumor cells. Surprisingly, intravenous injection of tumor cells led to an increased number of metastasis in VWF−/− mice compared to wild-type mice, suggesting a protective role of VWF in tumor cells dissemination [17].

Other functions for VWF revealed with the mouse models include a protective role in experimental allergic encephalomyelitis and blood–brain barrier permeability [18], an important role in modulation of infarct volume after ischemic stroke induced by middle cerebral artery occlusion [19] and a potential implication in smooth muscle cells proliferation and intimal hyperplasia [20]. In sepsis, results varied according to the experimental model used. VWF deficiency does not affect mortality in LPS-induced endotoxemia [16], while it significantly improves survival in a cecum ligature and puncture model [21].

VWF−/− mice also revealed the indirect role played by VWF in inflammation, through P-selectin. Indeed, the regulated expression of P-selectin is impaired in VWF−/−mice secondary to the absence of WPBs which cannot form when VWF is not present [22].

Contribution of murine models to the understanding of VWF biology

Formation of WPBs

Endothelial cells obtained from VWF−/− mice do not contain WPBs. Besides VWF, WPBs contain a number of other proteins such as P-selectin, CD63, interleukin-8, osteoprotegerin and angiopoietin-2 [23]. In the absence of WPBs, P-selectin is mistargeted and leukocyte recruitment is impaired during acute inflammation in VWF−/− mice [22]. Additional unexpected pathological consequences will probably arise, as a consequence of mistargeting of other WPBs proteins. An indirect role of VWF in angiogenesis or in bone homeostasis would therefore not be surprising.

VWF compartments

VWF is present in plasma, endothelial cells, subendothelium and platelets. The relative importance of these different compartments is still not clear. VWF−/− mice can help address this issue. The role of platelet VWF could indeed be tested using crossed bone marrow transplantation. As for the role of plasma VWF, hydrodynamic injection experiments showed that bleeding time and thrombus formation could be restored with plasma VWF only [7]. It should be noted, however, that very high levels of VWF are usually obtained with this technique, potentially explaining this result. Indeed in humans with type 3 VWD, infusion of VWF does not correct bleeding time despite improvement of the hemorrhagic status of the patients [24].


Biodistribution experiments performed by injecting VWF in VWF−/− mice revealed that most of the protein is targeted to liver macrophages [25]. In vivo depletion of these macrophages led to doubling of endogenous VWF levels in wild-type mice and to slower clearance of infused VWF in VWF−/− mice. Receptors mediating VWF endocytosis remain to be identified. VWF−/− mice were also used to test clearance of mutant VWF reproducing defects found in VWD patients. Accelerated clearance resulting from four different mutations (R1205H, C1130F, C1149R and C2671Y) was confirmed as the underlying mechanism responsible for the VWD phenotype [25]. Also, our knowledge about the importance of VWF glycosylations in clearance was mostly gained from studies performed with different mouse models: the RIIIS/J, ST3Gal-VI and Ashwell receptor-deficient mice.

Use of VWD murine models in testing new treatments

Murine models of VWD have not been extensively used to test efficacy of new treatments. The observation that human VWF does not interact with mouse GPIb has hampered many studies in this regard. Desmopressin administration commonly used in human type 1 VWD also does not work very well in the mouse, although its use has been reported in RIIIS/J mice [20]. The efficacy of interleukin-11 to raise VWF and FVIII levels was first tested in murine type 1 VWD (VWF+/− mice) [26]. These studies paved the way for the clinical trials currently underway in patients with mild VWD [27]. Finally, hydrodynamic injection experiments demonstrated the feasibility of a gene therapy approach for VWD despite the large size of its cDNA [7]. Although not really a VWD model per se, a transgenic mouse in which VWF has been humanized to interact with human platelets will be very useful to test new anti-platelet therapies in thrombosis [28].

Concluding remarks

VWD mouse models have not only greatly contributed to improve our knowledge of VWF but also showed some limitations such as the lack of type 2 VWD models. In addition, experiments targeting VWF as an antithrombotic strategy were not feasible due to the species barrier. In this respect, a model expressing human VWF modified to interact with murine platelets would be of great interest. Future research directions will probably rely on the development of new models expressing VWF variants through knock-in technologies, allowing a subtle approach of VWF structure–function relationships.

Disclosure of Conflict of Interest

The authors state that they have no conflict of interests.