von Willebrand factor: an emerging target in stroke therapy

Abstract

Thrombus formation is of paramount importance in the pathophysiology of acute ischemic stroke. Current antithrombotics used to treat or prevent cerebral ischemia are only moderately effective or bear an increased risk of severe bleeding. von Willebrand factor (VWF) has long been known to be a key player in thrombus formation at sites of vascular damage. While the association between VWF and coronary heart disease has been well studied, knowledge about the role of VWF in stroke is much more limited. However, in recent years, an increasing amount of clinical and preclinical evidence has revealed the critical involvement of VWF in stroke development. This review summarizes the latest insights into the pathophysiologic role of VWF-related processes in ischemic brain injury under experimental conditions and in humans. Potential clinical merits of novel inhibitors of VWF-mediated platelet adhesion and activation as powerful and safe tools to combat thromboembolic disorders including ischemic stroke are discussed.

Introduction

Ischemic stroke is a devastating disease that represents the primary reason for sustained disability and the second leading cause of death worldwide.¹ Eighty percent of strokes are caused by arterial occlusion of cerebral arteries whereas the remaining 20% are caused by intracerebral hemorrhages. Currently, the only established therapeutic option for acute stroke is rapid thrombolysis using the clot-breaking agent tissue plasminogen activator (t-PA) in order to achieve recanalization of occluded cerebral vessels. However, due to the increased risk of bleeding associated with late t-PA administration, intravenous t-PA is recommended only within the limited therapeutic time window of 4.5 hours post-stroke and thus is available to less than 10% of patients.² A recent trial to extend the therapeutic window up to 9 hours by use of recombinant desmoteplase, a novel plasminogen activator, failed³ as did a trial using the defibrinogenating agent ancrod.⁴

In terms of secondary stroke prevention the situation is very similar. Antiplatelet agents such as acetylsalicylic acid (ASA), dipyridamole, and the platelet P2Y12 receptor inhibitor clopidogrel show only limited efficacy and substantially increase the risk of fatal bleeding. This holds also true for anticoagulants, particularly warfarin, and even the introduction of novel substance classes such as direct thrombin inhibitors (e.g. dabigatran) or FXa blockers (e.g. rivaroxaban, apixaban) could not overcome the threat of hemorrhage. These limitations emphasize the need for a better understanding of the pathophysiologic mechanisms of thrombus formation in acute ischemic stroke⁵, in order to successfully improve treatment.

VWF: role in hemostasis, thrombosis and inflammation

In 1926, Finnish physician Erik von Willebrand reported a new type of inherited bleeding disorder that was distinct from hemophilia A.⁶ Thirty years later, the plasma protein that is central to the disease was identified and was named von Willebrand factor (VWF). Now, more than half a century later, much of the structure and function of VWF has been elucidated and its role in maintaining the delicate balance between bleeding and thrombosis has become an intriguing subject. VWF is a large, multimeric glycoprotein. Along with serving as a protective carrier molecule for clotting factor VIII, its main function is mediating initial platelet adhesion at sites of vascular injury. Indeed, whereas this is a prerequisite for normal hemostasis, adhesion of platelets is also the first step in thrombosis and an important mediator of inflammation.⁷

The critical role of VWF in normal hemostasis is exemplified by von Willebrand disease (VWD). VWD is the most common inherited bleeding disorder in humans, caused by quantitative or qualitative defects in VWF.⁸ Bleeding symptoms range from mild (type 1) to severe (type 3) and include mucosal hemorrhages, such as epistaxis, menorrhagia, and bleeding from the gums and gastrointestinal tract.

VWF is exclusively synthesized in endothelial cells and megakaryocytes and circulates as multimers of varying size (up to 20,000 kDa). The multi-domain structure of the monomeric VWF building blocks (Figure 1) is fundamental to the function of VWF (Figure 2). Rapid binding of VWF (A3 domain) to exposed fibrillar collagen type I and III immobilizes VWF at sites of vascular damage. This and/or high shear blood leads to a conformational changes, exposing the binding site for platelet glycoprotein (GP)Ibα in the VWF A1 domain. The reversible nature of the GPIbα-VWF A1 interactions allows deceleration and rolling of platelets, which, especially under high shear forces, is necessary for the definitive arrest. Firm adhesion of platelets at the site of vascular injury is further supported by engagement of the platelet collagen receptors (GPVI and integrin α₂β₁) and leads to platelet activation. Upon platelet activation, soluble platelet agonists such as ADP, ATP and thromboxane A2 are released and platelet integrins like GPIIb/IIIa shift to a high-affinity state. Subsequent platelet aggregation is promoted by binding of activated platelet GPIIb/IIIa to its primary ligand fibrinogen and to the Arg-Gly-Asp (RGD) sequence found in the C1 domain of VWF. Additional incoming platelets are recruited to the growing thrombus, primarily via engagement of their GPIbα receptors (Figure 2). Hence, the GPIb complex is essential for both initial platelet adhesion to sites of vascular injury and recruitment of new platelets to the growing thrombus.

Figure 1. Schematic representation of VWF.

VWF is a multimeric protein composed of dimeric building blocks. The VWF domains, the binding sites of major binding partners and the cleavage site for ADAMTS13 are indicated. Adapted from De Meyer et al.⁸

Figure 2. Schematic representation of VWF-mediated platelet adhesion and aggregation.

VWF that is immobilized on collagen with its A3 domain or released from endothelial Weibel-Palade (WP) bodies recruits platelets by binding with its A1 domain to platelet GPIbα. Together with fibrinogen (not shown), VWF further cross-links platelets via binding of its C1 domain to platelet GPIIb/IIIa.

Besides its role in thrombus formation, VWF has also been shown to support inflammatory processes. In vitro experiments demonstrated that VWF promotes leukocyte adhesion by acting as a ligand for the leukocyte receptors P-selectin glycoprotein ligand 1 and β2 integrin.⁹ VWF-bound platelets also were shown to support leukocyte tethering and rolling under high shear stress¹⁰. More recently, Petri et al. showed that VWF promotes the extravasation of leukocytes from blood vessels in a strictly platelet and GPIbα dependent way.¹¹

ADAMTS13: a biological regulator of VWF activity

VWF activity is correlated with multimer size, with ultra-large (UL) multimers spontaneously binding to platelets. UL-VWF is released from endothelial and platelet storage granules upon stimulation with various thrombogenic or inflammatory secretagogues, but also during hypoxia.^{12, 13} To prevent spontaneous thrombosis, ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type 1 repeats, 13) cleaves the Y1605-M1606 bond in the VWF A2 domain (Figure 1 and 2), thereby digesting UL-VWF into smaller, less reactive molecules. As seen in thrombotic thrombocytopenic purpura (TTP), ADAMTS13 deficiency is associated with thrombotic occlusion of microvessels from multiple organs, including the brain.^14–16 Experimental studies in mice have demonstrated the antithrombotic and anti-inflammatory properties of ADAMTS13.^{17, 18}

VWF and stroke risk

Because of the pivotal role of VWF in platelet adhesion and thrombus formation, it seems logical to anticipate a correlation between high plasma levels of VWF and development of cardiovascular disease. While this relationship has been well studied for coronary heart disease¹⁹, much less is known about the association between VWF levels and stroke. When comparing stroke patients with healthy controls, several studies found an association between high VWF levels and stroke,^20–28 and even different etiologic subtypes of ischemic stroke.²⁹ However, increased VWF levels are a well-known marker of endothelial activation and/or dysfunction and a conclusive interpretation of many of these case-control studies on the causative or consequential nature of high VWF levels in stroke patients has been obscured by the post-stroke time point of VWF measurement. By also measuring the VWF propeptide, a recent investigation of two independent case-control studies (COCOS and ATTAC) elegantly indicated that endothelial cell activation and subsequent Weibel-Palade body secretion is indeed an important mechanism underlying the association between total VWF levels and the occurrence of a first ischemic stroke.³⁰ Several prospective studies, such as the recent longitudinal analysis embedded in the Rotterdam study³¹, clearly identified high plasma levels of VWF as a strong predictor of stroke.^32–37 Importantly, correlation between VWF levels and stroke-related mortality persisted after adjustment for common risk factors such as age, stroke severity, and atrial fibrillation indicating that VWF is an independent predictor.³⁸ On a genetic level, VWF polymorphisms have been identified that significantly raise the risk of ischemic stroke.^39–41 Not all of these are associated with higher VWF levels, suggesting that other mechanisms, such as increased VWF activity, could contribute to the risk of stroke as well.

A major down-regulator of VWF activity is ADAMTS13. Accordingly, low levels of ADAMTS13 come along with an increased risk of cardiovascular disease, including ischemic stroke.^{25, 28} Analogous to VWF, single nucleotide polymorphisms in ADAMTS13 were found to be associated with the incidence of ischemic stroke in a Swedish population.⁴²

Together, these proof-of-principle studies established that plasma levels of VWF and/or ADAMTS13 are associated with the risk of stroke in the general population. However, determination of VWF and/or ADAMTS13 levels on a regular basis to judge the risk of thromboembolic disease in individual patients cannot be recommended until larger prospective trials and standardized test systems are available.

Experimental studies: role of VWF in acute stroke

The importance of VWF as a risk factor for stroke occurrence and mortality in humans recently also stimulated experimental studies in models of acute stroke. Using a mouse model of transient middle cerebral artery occlusion (tMCAO), we showed that mice that are deficient in VWF due to VWF gene abrogation (VWF^−/−)⁴³ are protected from brain ischemia/reperfusion injury.^44–46 Infarct sizes in VWF^−/− were ~60% of the infarct volumes in wild-type controls one day after tMCAO, which was accompanied by reduced fibrin accumulation in the infarcted brain hemisphere. Accordingly, neurological scores assessing motor function and coordination were significantly better in VWF^−/− mice compared to controls. Importantly, genetic disruption of VWF did not increase the risk of intracerebral bleeding in the context of ischemic stroke.⁴⁵ Reconstitution of plasma VWF by hydrodynamic gene transfer fully restored the susceptibility of VWF^−/− mice to cerebral ischemia underlining the causative role of VWF in this setting.^{44, 45} This is in line with the well-established antithrombotic effects of VWF deficiency in several experimental arterial and venous thrombosis models.^{43, 47–50} Further illustrating the critical role of VWF in ischemia/reperfusion injury are the findings that ADAMTS13^−/− mice are more susceptible to focal cerebral ischemia.^{46, 51} These mice developed significantly larger infarctions, with an increased accumulation of immune cells and thrombi in the ischemic brain tissue, resulting in more severe neurological deficits.⁵¹ On the other hand, intravenous administration of recombinant ADAMTS13 into wild-type mice immediately before reperfusion significantly reduced infarct volume.⁴⁶

By reconstituting VWF^−/− mice with different VWF mutants, we recently showed that binding of VWF to both collagen and GPIbα, but not to GPIIb/IIIa, are mandatory steps in stroke development.⁴⁴ The involvement of collagen and GPIbα-mediated platelet adhesion in stroke is corroborated by the findings that blocking platelet collagen receptor GPVI or GPIbα also confers a protective effect in the mouse tMCAO model.⁵² Blockade of GPIIb/IIIa did not affect stroke size and led to an increased incidence of intracerebral hemorrhage, whereas blocking of GPIbα or GPVI did not increase the frequency of intracerebral bleeding.⁵² Finally, mice in which downstream signaling of GPIb via phospholipase D1 is abrogated and mice in which the extracellular part of GPIb is replaced by human interleukin-4 receptor (GPIbα/IL4Rα)⁵³ are also protected against focal cerebral ischemia without causing excessive bleeding (⁵⁴ and SFDM and DDW, unpublished observations, 2010). These observations further underline that blockage of the GPIbα-VWF axis or collagen-platelet axis might be a safe approach in ischemic stroke.

Inhibitors of VWF: a promising class of antithrombotics on the brink of reaching the clinic

From the above, it is clear that pharmacological interference in VWF-mediated platelet adhesion and thrombus formation could have clinical benefit as a promising strategy in stroke treatment. Although no such VWF-blockers have yet achieved regulatory approval for marketing, there are promising preclinical and clinical studies that demonstrate the antithrombotic potential of agents that inhibit VWF function by blocking the VWF-collagen or VWF-GPIbα interaction (Figure 3). In this section, we will discuss candidate molecules that could prove useful in stroke therapy based on the encouraging results they have demonstrated in the inhibition of VWF-mediated thrombosis. These inhibitors include monoclonal antibodies against VWF (82D6A3, AJvW2 and its humanized form AJW200) or GPIbα (6B4 and its humanized form h6B4), the nanobody^® ALX-0081, the aptamer ARC1779, and the recombinant GPIbα fragment GPG-290 (Table 1).^55–57 A detailed overview of the key features of each of these inhibitors is given in Table S1 (please see http://stroke.ahajournals.org).

Figure 3. Schematic representation of mode-of-action of various VWF inhibitors.

VWF-mediated platelet adhesion can be blocked by inhibiting binding of VWF to either collagen or GPIbα, or by cleaving VWF by ADAMTS13.

Table 1. Inhibitors of VWF-mediated platelet adhesion.

A detailed description of each of these inhibitors is given in Table S1 (please see http://stroke.ahajournals.org).

Compound	Description
82D6A3	Monoclonal antibody against VWF A3 domain that inhibits binding of VWF to collagen
6B4 h6B4	Fab-fragment of a monoclonal antibody against platelet GPIbα that inhibits binding of VWF to GPIb
AJvW2 AJW200	Monoclonal antibody against VWF A1 domainα that inhibits binding of VWF to GPIbα
GPG-290	Chimeric recombinant protein containing gain-of-function GPIbα fragment that binds to A1 domain thereby inhibiting binding of VWF to GPIbα
ARC1779	Aptamer against VWF A1 domain that inhibits binding of VWF to GPIbα
ALX-0081 ALX-0681	Nanobody® against VWF A1 domain that inhibits binding of VWF to GPIbα
rADAMTS13	Recombinant protein that cleaves VWF multimers by cleaving the Y1605-M1606 bond in the VWF A2 domain

While most attention has been focused on the development of VWF-GPIbα inhibitors, the anti-VWF antibody 82D6A3 is different in that it inhibits the binding of VWF to collagen. Preclinical studies in baboons have demonstrated that 82D6A3 has a strong antithrombotic efficacy.⁵⁸ This observation indicates that, despite the existence of other binding partners for VWF in the extracellular matrix, VWF binding to fibrillar collagen has an important role in mediating thrombosis. A humanized version of 82D6A3 has been constructed for further preclinical and clinical testing.⁵⁹

AJvW2, AJW200, 6B4 and h6B4 are monoclonal antibodies designed to disrupt the VWF-GPIbα interaction by binding to the VWF A1 domain and GPIbα respectively. Both antibodies have been tested extensively in preclinical thrombosis models.^60–67 The clinical efficacy and tolerance of AJW200 has been demonstrated in human volunteers without bleeding complications.⁶⁸

GPG-290 is a homodimeric recombinant fragment of human GPIbα composed of the first 290 amino acids and conjugated to human IgG1Fc to form a homodimer. GPG-290 contains two gain-of-function mutations (G233V/M239V) resulting in its enhanced affinity for the VWF A1 domain (and possibly other GPIb ligands). Its antithrombotic activity was demonstrated in preclinical murine and canine models of arterial and venous thrombosis.^{48, 69, 70}

The aptamer ARC1779 represents an interesting new class of inhibitors. Aptamers are nucleic acid macromolecules that tightly bind to a specific molecular target.⁷¹ In solution, the chain of nucleotides forms intra-molecular interactions that fold the molecule into a complex three-dimensional shape that enhances affinity for its target molecule. A potential advantage of this class of molecules is that they can be inactivated by a complementary aptamer antidote if necessary. ARC1779 is a 40-nucleotide DNA/RNA aptamer conjugated to a 20 kDa PEG to enhance its pharmacokinetic properties. ARC1779 binds to the VWF A1 domain, and was reported to effectively inhibit thrombus formation in several preclinical settings.^{72, 73} The aptamer was well tolerated in healthy volunteers in a Phase I trial. Importantly, in a Phase II trial, ARC1779 effectively increased platelet counts in critically ill TTP patients by blocking spontaneous VWF-mediated platelet aggregation^74–76 and even prevented desmopressin-induced thrombocytopenia in patients with VWD type 2B.⁷⁷ High affinity aptamers to murine VWF have recently been developed, which will allow the investigation of anti-VWF aptamers in murine preclinical models of thrombosis, including stroke.⁷⁸ Interestingly, a recent trial showed that ARC1779 was able to reduce cerebral emboli signals in patients undergoing carotid endarterectomy.⁷⁹ Nanobodies are antibody-derived therapeutic proteins that contain the structural and functional properties of naturally occurring heavy-chain antibodies. The cameloid bivalent Nanobody^® ALX-0081 specifically targets exposed GPIbα binding sites in VWF. By blocking GPIbα binding to VWF, ALX-0081 showed strong antithrombotic potency in preclinical baboon studies and in blood obtained from patients undergoing percutaneous coronary intervention (PCI).^{80, 81} In a Phase I clinical study, this nanobody was found to be well tolerated and safe. ALX-0081 (and ALX-0681, a subcutaneous formulation of ALX-0081) entered a Phase II study in patients undergoing PCI and recently also a Phase II study in TTP patients.

Apart from blocking binding of VWF to either collagen or GPIbα, another way of reducing VWF activity is decreasing its size by ADAMTS13. Recombinant ADAMTS13 is being developed as a new therapeutic agent and, as mentioned earlier, was protective in a preclinical mouse model of ischemic stroke.⁴⁶ Recombinant human ADAMTS13 has received orphan designation for the treatment of TTP (EU/3/08/588).

Potential use of VWF antagonism in stroke therapy

Since VWF antagonists are starting to enter the first clinical trials, it is interesting to speculate on the potential role of these new drugs in stroke therapy. Inhibitors of the VWF-GPIbα or VWF-collagen interaction have no direct thrombolytic properties, so their use is unlikely to result in dissolution of an already existing thrombus in the acute setting of ischemic stroke. However, when used in combination with t-PA, VWF antagonists could prevent ongoing microvascular thrombus formation during the reperfusion phase after successful or spontaneous thrombolysis^{82, 83}, reducing the occurrence of re-thrombosis and/or secondary stroke progression. Whether dual therapy of VWF inhibitors and t-PA would allow the use of lower and thereby safer doses of t-PA remains to be established. The use of t-PA-dependent experimental thromboembolic stroke models will have to shed more light on the possible beneficial effects of combining thrombolytic therapy with VWF inhibition. Former attempts to prevent the apposition of further clots into an already existing thrombus during the early stage of ischemic stroke by applying heparins failed.⁸⁴ While heparins could counteract deterioration of stroke symptoms in some studies, the net clinical benefit was outweighed by an increased risked of severe hemorrhages. Consequently, full-intensity parenteral anticoagulation with heparins is no longer recommended in the acute phase of cerebral ischemia.⁸⁵ In view of the anticipated safer benefit-over-risk ratio in terms of bleeding, it will be interesting to see whether VWF antagonists are more successful than heparins during the acute phase of ischemic stroke. Nevertheless, close CT or MRI monitoring of potential hemorrhagic transformation should accompany potential future applications of VWF inhibitors in stroke patients. Interestingly, recombinant ADAMTS13 promoted thrombus dissolution in injured mouse mesenteric arterioles¹⁷, calling for further studies of the thrombolytic potential of ADAMTS13 in acute ischemic stroke.

With regard to stroke prevention, VWF inhibition could become useful for short-term therapy during procedures associated with increased risk of cerebral thromboemoblism, such as angiography, carotid endarterectomy, or heart surgery under conditions of extracorporeal circulation. Promising results were recently obtained with ARC1779, which significantly reduced cerebral embolization in patients undergoing carotid endarterectomy.⁷⁹ In the absence of oral formulations, the potential of VWF blockers to prevent stroke or systemic embolism in chronic conditions, e.g. in patients with atrial fibrillation, cannot yet be judged.

Conclusion

There is now compelling evidence on preclinical and clinical levels that interactions between the cerebral blood vessels and platelets employing VWF, GPIbα and collagen are instrumental in ischemic brain disease. Early clinical testing of lead compounds targeting VWF-mediated platelet adhesion in healthy individuals or selected disease states points towards a favorable bleeding risk profile and higher efficacy as compared to established antithrombotic drugs. Based on the encouraging results in rodent stroke models, larger translational studies in non-human primates and proof-of-principle clinical trials are now warranted to further judge the risk-to-benefit ratio of interfering with the early steps of platelet activation. If these trials are as promising as the current experimental data, an exciting decade exploring VWF-targeted stroke strategies lies ahead. Back in 1926, Dr. von Willebrand may not have fully grasped what great implications his seminal report⁶ on a bleeding disease would have on our understanding and treatment of thrombotic disorders. More than ninety years later, stroke may very well become one of them.