Vincent Jenkins, PhD, National Centre for Hereditary Coagulation Disorders, St James Hospital, James Street, Dublin 8, Ireland. Tel.: +353 1 416 2141; fax: +353 1 410 3570; e-mail: email@example.com
Summary. The interaction of factor VIII (FVIII) with von Willebrand Factor (VWF) is of direct clinical significance in the diagnosis and treatment of patients with haemophilia A and von Willebrand disease (VWD). A normal haemostatic response to vascular injury requires both FVIII and VWF. It is well-established that in addition to its role in mediating platelet to platelet and platelet to matrix binding, VWF has a direct role in thrombin and fibrin generation by acting as a carrier molecule for the cofactor FVIII. Recent studies show that the interaction affects not only the biology of both FVIII and VWF, and the pathology of haemophilia and VWD, but also presents opportunities in the treatment of haemophilia. This review details the mechanisms and the molecular determinants of FVIII interaction with VWF, and the role of FVIII–VWF interaction in modulating FVIII interactions with other proteases, cell types and cellular receptors. The effect of defective interaction of FVIII with VWF as a result of mutations in either protein is discussed.
The introduction of cryoprecipitate as a useful treatment for patients with constitutional bleeding disorders led to the identification of a novel protein complex containing both anti-haemophilia A activity, and platelet-binding properties. These discrete functions were subsequently shown to be mediated by two proteins circulating together in a single complex in normal plasma – namely anti-haemophilia factor VIII (FVIII) and FVIII-related antigen (FVIII-RAG) . The FVIII component of this complex corrected bleeding in haemophilia A, whilst the FVIII-RAG (or von Willebrand factor; VWF) component could correct the bleeding phenotype in patients with von Willebrand’s disease (VWD). It is now well recognized that FVIII and VWF constitute independent gene products with distinct functions. Nevertheless, the haemostatic activities and life-cycles of these glycoproteins in the normal circulation remain inextricably linked. Consequently, understanding the biochemical basis underlying the interaction between FVIII and VWF in human plasma is of direct translational significance.
Vascular injury leads to generation of a platelet plug at the site of injury, which is subsequently stabilized through activation of the coagulation cascade and formation of a cross-linked fibrin network. Platelet adhesion, activation, and aggregation, together with concurrent thrombin generation, are central events in this response. The FVIII–VWF complex plays critical roles in regulating both platelet responses and the normal coagulation cascade. First, increased local shear stress at sites of vascular damage results in VWF adhesion to the sub-endothelial matrix. This bound VWF can then tether circulating platelets, to initiate formation of the platelet plug. Second, as VWF and FVIII circulate as a single complex in normal plasma, the ability of VWF to interact with exposed subendothelial tissues at sites of vascular injury, also serves to significantly increase the local concentration of FVIII . Tissue-factor-initiated thrombin generation causes FVIII activation and release from VWF, as a result of limited proteolytic FVIII cleavage. Activated FVIII (FVIIIa) is then free to function as an effective procoagulant cofactor of factor IXa (FIXa) in the conversion of zymogen factor X (FX) to the enzyme FXa, increasing the catalytic activity of FIXa by several orders of magnitude . Consequently, the FVIII–VWF complex serves critical roles in mediating primary haemostasis, and coagulation.
FVIII and VWF biosynthesis
The FVIII mRNA and protein have been identified in many human tissues including the spleen, lung and kidney; however the liver is likely to constitute the primary source of FVIII synthesis in vivo [4–9]. The cell type(s) within the liver principally responsible for the synthesis and secretion of FVIII have not been clearly delineated. Although hepatocytes have been shown to synthesize FVIII, increasing evidence suggests that hepatic sinusoidal endothelial cells (EC) may be of prime importance. FVIII mRNA and protein has been demonstrated in both human and murine liver sinusoidal EC [5–7]; transplantation of normal murine sinusoidal EC into a mouse model of haemophilia A has recently been shown to effectively restore normal haemostasis .
In vivo biosynthesis of VWF is restricted to EC and megakaryocytes [11,12], though quantitative expression of VWF varies significantly between different vascular beds. Histological studies of animal tissue have shown that VWF expression is significantly higher in venous as compared with arterial EC, and also that secretion is increased in larger vessels [7,13,14]. Highest VWF levels were reported in the lung and brain, with very low levels of expression in the liver . Despite the association of FVIII and VWF in the peripheral circulation, there is no direct evidence to suggest that VWF and FVIII are actually synthesized together in any particular cell type in vivo. Nevertheless expression studies have shown that FVIII and VWF can be co-synthesized, transported to storage granules and released by endothelial cell lineages and megakaryocytes [15–17]. Numerous indicators suggest that limited co-expression may exist in vivo, it is well recognized that the administration of vasopressin or its pharmaceutical analogue desmopressin (DDAVP) results in a transient increase in VWF and FVIII levels. This ability has led to the widespread use of DDAVP in the treatment of patients with VWD and mild haemophilia A patients. Interestingly, recent studies have demonstrated that whilst liver transplantation cures haemophilia A, subsequent infusion of DDAVP in these patients produced a transient increase in plasma VWF levels, but did not further increase plasma FVIII levels , whereas non-haemophilic liver transplant recipients demonstrate responses in both VWF and FVIII following DDAVP. Furthermore DDAVP administration does not significantly increase plasma FVIII levels in patients with type 3 VWD . Cumulatively, these data further support the hypothesis that a co-synthesized, releasable pool of FVIII–VWF may indeed exist in vivo .
Determinants of FVIII and VWF interaction
The FVIII binds to VWF with high affinity (Kd approximately 0.2–0.5nm) [20,21]. Although the interaction between FVIII heavy and light chain increases the affinity of FVIII and VWF interaction, the VWF-interactive region of FVIII is entirely located within the FVIII light chain [22–25]. The N-terminal acidic region of the light chain between residues 1672–1689 is critical in mediating the interaction, with sulphation of residue Tyr1680 particularly important for VWF-binding [26–28]. Interestingly, in the recently described crystal structures of B-domainless FVIII, the acidic region of the light chain could not be resolved and remained disordered, suggesting that the interaction with VWF may be important in stabilizing this region (Fig. 1b) [29,30].
Following activation by thrombin, FVIII is cleaved at multiple sites within the heavy chain and at the N-terminal end of the light chain to generate FVIIIa (see Fig. 1). Removal of the N-terminal acidic region of the FVIII light chain significantly reduces the affinity of FVIIIa for VWF (1400-fold), thereby releasing FVIIIa from the complex .
In addition to the high-affinity acidic-binding region, the C2 and C1 domains of the FVIII light chain have also been shown to be important in determining interaction with VWF. Contributory regions within the FVIII C2 domain have been identified following characterization of different recombinant FVIII C2 domain variants, including some haemophilia-A-associated mutations (Fig. 1b). In particular, the residues Met3199/Phe2200 and Leu2251/Leu2252 have been shown to directly interact with both VWF and phospholipid surfaces . Cluster mutation of all four residues reduced the binding to phosphatidylserine containing phospholipids by approximately 95%, and reduced VWF-binding 20-fold. A role for the C1 domain in mediating the interaction between FVIII and VWF has also been described, a naturally occurring inhibitory antibody directed towards an epitope within the C1 domain has been shown to inhibit VWF interaction and the mutation Ser2119Tyr resulted in an 80-fold loss of affinity for FVIII–VWF binding . Given the dramatic loss of affinity of the FVIII light chain for VWF following thrombin cleavage at Arg 1689 and loss of the N-terminal acidic sequence, the importance of the C-domain regions in mediating FVIII–VWF binding appears incongruous. Nevertheless, specific residues within the C1 and C2 domains may play a critical role in mediating the ability of FVIII to form an initial complex with VWF.
The VWF precursor, named pre-pro-VWF, is composed of 2813 amino acids, of which 22 correspond to the signal peptide, 741 to the propeptide, and 2050 to the mature subunit. The pro-VWF consists of four repeated domains (A–D) (Fig. 2) . Pro-VWF undergoes extensive post translational modification prior to secretion, which results in the formation of highly glycosylated, high molecular-weight multimers. A striking feature of VWF glycosylation is the presence of ABO(H) determinants on a proportion of the N-linked chains, in keeping with the ABO blood group of the individual . Circulatory VWF is almost entirely of EC origin, being constitutively secreted toward the extracellular matrix and the plasma. About 5% of total VWF is retained the storage granules of endothelial cells and platelets, respectively, and is secreted upon adequate stimuli.
The FVIII binds to VWF within the first 272 residues of the mature N-terminal region of the VWF polypeptide (D’ and D3 domains within the corresponding to residues 763–1035) [35–39]. Cleavage of the propeptide from the mature polypeptide is required for FVIII binding, however the prior involvement of the propeptide in mature VWF processing increases the subsequent affinity of VWF for FVIII by approximately 10-fold [40,41].
Mutations in a restricted area of the VWF gene have been associated with markedly VWF-binding to FVIII, resulting in the autosomal recessive subtype 2N VWD (Normandy variant) [42–45]. Typically, patients with 2N VWD have VWF levels within the normal range with only FVIII levels reduced to below normal, such that basic laboratory and clinical parameters appear similar to mild haemophilia A. Certain DDAVP studies have demonstrated that the half-life of FVIII in these patients is significantly reduced (approximately 2–3 h) . Mutations resulting in 2N VWD are listed in the VWF mutation database (http://www.ragtimedesign.com/vwf/mutation/). In general the mutations result in amino acid substitutions that do not generally alter multimer structure, but rather reduce or abolish the ability to bind FVIII only, by mechanisms which are not yet clearly defined [42,47]. Notable exceptions are mutations which prevent cleavage of the propeptide from mature VWF at Arg760, and hence prevent FVIII binding ; and mutations which by introducing or abolishing cysteine residues in the D’ or D3 regions alter multimer structure and decrease VWF-binding to FVIII [49–51].
Plasma FVIII and VWF
In clinical practice, the mean plasma concentrations of both FVIII and VWF in the normal population are defined as 1 IU mL−1. Consequently, the ratio of FVIII to VWF is 1. However the molar concentrations of the two molecules in plasma are very different. Although the typical plasma concentration of FVIII is 100–250ng mL−1 (approximately 1 nm), the plasma concentration of VWF is approximately 8 μg mL−1 (approximately 50 nm) . Thus there is a 30–50 m excess of VWF to FVIII in normal circulation, such that not all VWF multimers contain FVIII [20,21]. In vitro experiments have shown that VWF can bind FVIII at a 1:1 molar ratio, indicating that each monomer has the ability to bind FVIII, though this ability likely requires a change in conformation of VWF [21,24,53].
Plasma FVIII and VWF levels vary over a wide range even amongst normal individuals (approximately 0.5–2 IU mL−1), according to blood group, age, race, and gender. ABO blood group constitutes an important determinant of plasma FVIII and VWF levels . FVIII:Ag and VWF:Ag levels are both reduced by approximately 25% in blood group O as compared with non-O (A, B, or AB) individuals [55–60]. Multivariate analysis has shown that the effect of ABO blood group on plasma FVIII levels is primarily mediated through an effect of blood group on plasma VWF:Ag levels. In addition, several studies have demonstrated that ratio of FVIII to VWF does not vary across different ABO blood groups [59,61]. However, a small but significant VWF-independent effect of ABO blood group on plasma FVIII levels has also been reported in a recent study of healthy family populations . FVIII and VWF:Ag are significantly higher (approximately 20%) in African-Americans as compared with similar caucasian populations, although the effect of ABO blood group is maintained [63–66]. In addition, plasma FVIII and VWF levels rise with increasing age in adults [64,67,68].
Functional effects of FVIII–VWF interaction
The FVIII–VWF complex has a direct role in both primary haemostasis and coagulation by mediating platelet–platelet and platelet–matrix interaction and in local generation of a fibrin clot by increasing FVIII concentration at the site of injury. However, the functional effects of this interaction extend beyond events at site of injury. In particular, interaction with VWF is a critical factor in increasing the circulatory half-life of FVIII.
FVIII–VWF half-life and clearance
It is well-established that interaction with VWF significantly increases FVIII survival in normal plasma. Previous studies have shown the half-life of infused FVIII concentrate in patients with type 3 VWD is only 2.5 h, as compared with approximately 12 h in patients with haemophilia A . A critical role for VWF in regulating FVIII catabolism has also been confirmed in animal studies. For example, infusion of purified porcine VWF into type 3 VWD pigs was sufficient to restore FVIII levels from approximately 25% to normality. Moreover, the increase in FVIII levels was not attributable to increased FVIII synthesis, as liver FVIII mRNA levels were not affected . Similarly, use of a high purity VWF therapeutic concentrate (containing very low levels of FVIII) in patients with VWD demonstrated that FVIII levels increased from very low to haemostastic levels within 6 h following infusion, and that FVIII levels were sustained for upto 24 h . Cumulatively, these data confirm that binding of FVIII to VWF is critical for normal survival of FVIII in the circulation. The mechanisms for maintaining FVIII half-life include stabilization of FVIII structure, prevention of cleavage and removal by cellular interactions as outlined below.
Effects of VWF interaction on FVIII stability, activation and inactivation
von Willebrand factor interaction maintains the stability of the FVIII heterodimer as demonstrated by in vitro expression studies in which the presence of VWF increased the yield of FVIII by fivefold [40,72]. VWF interaction with the FVIII light chain serves to enhance the rate of association of the FVIII heavy and light chains [22,73].
Anti-haemophilia factor VIII (FVIII) is activated by limited proteolytic cleavage by thrombin, and FVIIIa inactivated by activated Protein C (APC) cleavage. However FVIII can be activated and/or inactivated by a number of coagulation-related serine proteases, including FXa, APC and FIXa. The physiological relevance of these reactions remains unclear, however FVIII-binding to VWF protects against cleavage by these proteases with the exception of thrombin [74–77]. This protection is mediated by two mechanisms First, VWF-bound FVIII is unable to bind to phospholipid or platelets [78,79], second, direct protease-binding sites within the FVIII light chain are hidden whilst FVIII is in complex with VWF [80,81]. This protection from proteolysis serves to increase FVIII circulatory life-span.
FVIII–VWF clearance and cellular interactions
The VWF-bound or -unbound state of FVIII modulates FVIII cellular interactions and removal from the circulation. Several cellular receptors implicated in FVIII clearance have been described and extensively reviewed elsewhere (see ). In particular, the role of the low-density lipoprotein receptor-related protein (LRP), a member of the LDLR family and its effects on FVIII clearance, have been studied in vitro and in vivo in murine model studies. LRP is a multifunctional scavenger receptor abundant in the liver that can bind to at least 30 ligands with high affinity . FVIII can bind to LRP via the A3 1811–1818 region within light chain, and 484–509 region of the A2 domain within the heavy chain [84,85]. The latter site is cryptic and exposed only on activation of FVIII, whereas the LRP-binding site within the FVIII light chain is only exposed when FVIII is not bound to VWF . VWF does not bind to LRP, and because of the higher affinity of FVIII for VWF, prevents binding of bound FVIII to the receptor, suggesting that LRP-mediated clearance is of minimal importance in the FVIII life-cycle. However an LRP-knockout mouse model has a twofold increase in FVIII levels as compared with control mice, and an increased FVIII half-life, suggesting a significant role for LRP-related clearance mechanisms of FVIII . A recent hypothesis to resolve this apparent contradiction has been suggested by Lenting et al. . Because of high affinity of both molecules and the molar excess of VWF as compared with FVIII, almost all circulating FVIII is bound in complex with VWF. However a small (approximately 2%), but significant proportion circulates unbound, and it is this pool of free FVIII that is cleared by LRP-mediated mechanisms. Moreover, clearance of the free FVIII results in a shift in the balance of bound and free FVIII, and a further release of FVIII from VWF .
The close association of FVIII and VWF levels and half-life suggests that the remaining FVIII is cleared as part of the VWF complex. Clearance of the VWF complex from the circulation remains an enigma, however very recent data has thrown some light on possible mechanisms. Studies of cell types within the liver and spleen demonstrate that isolated FVIII, VWF and FVIII–VWF complex can be endocytosed by macrophages within these organs . Furthermore elegant studies of gene knockout mice show that an asialoglycoprotein receptor, the Ashwell receptor, located on hepatocytes modulates VWF clearance . The Ashwell receptor is composed of two transmembrane glycoproteins: asialoglycoprotein receptor-1 (ASPGR-1) and asialoglycoprotein receptor-2 (ASPGR-2). By studying Asgr-1 or Asgr-2 gene knockout of the respective receptors, Grewal et al. demonstrated an increase in VWF half-life with a corresponding 1.5-fold increase in circulatory plasma VWF and FVIII levels in mice lacking ASPGR-1 as compared with those lacking only ASPGR-2 or wild-type mice. Further studies are required, as the role of the Ashwell receptor in human VWF clearance is yet to be demonstrated, however these studies show an attractive candidate mechanism for control and clearance of the FVIII–VWF complex, which is directly related to the glycosylation status of VWF.
FVIII–VWF and ADAMTS-13
There is no evidence to suggest that the presence of FVIII affects the survival of VWF or alters VWF function. VWF levels are not reduced in patients with haemophilia A. Studies using the vasopressin analogue DDAVP show that VWF half-life in patients with mild haemophilia A are similar to those described for normal individuals levels . However, it has been recently suggested that FVIII does play a role in control of VWF multimer distribution by acting as a cofactor for the VWF-cleaving protease ADAMTS-13. Deficiency of this protease is associated with thrombotic thrombocytopenic purpura (TTP) (reviewed in ). ADAMTS-13 cleaves specifically between Tyr 1605 and Met 1606 in the A2 domain of VWF, although the efficiency of the enzyme is greatly dependent on VWF conformation. In vitro experiments have demonstrated that the presence of FVIII increased the cleaving efficiency of ADAMTS-13 on shear-stressed VWF by upto 10-fold . The experiments were performed using concentrations of FVIII and ADAMTS-13 manyfold higher than physiological levels, however half-maximal co-factor effects were observed at a FVIII concentration (approximately 3 nm), which are suggested to be at or near levels saturating in vivo VWF . Binding of FVIII to VWF is important in modulating the co-factor effect as FVIII mutants lacking the a3 acidic region showed no effect in enhancing cleavage. The physiological relevance of this mechanism remains to be determined; however FVIII may have a subtle role in the modulation of VWF structure.
FVIII–VWF and immune response
The development of an immune response to FVIII therapy is currently the most significant complication of haemophilia A treatment. Approximately, 25% of the patients with severe haemophilia A develop inhibitors. There is considerable debate as to whether the presence of VWF in therapeutic concentrations may play a role in preventing and/or overcoming inhibitor development [94–99]. Although it is beyond the scope of this review to discuss the clinical and therapeutic aspects of haemophilia care, studies have demonstrated a definite role for binding of FVIII to VWF in regulating immune recognition. The most common epitopes of inhibitory antibodies to FVIII are located in the A2 domain of the heavy chain, and the C2 domain of the light chain. A common epitope in the latter has been identified within the region spanning residues 2248 through 2312 . Antibodies to this epitope have been shown to inhibit FVIII-binding to both VWF and phospholipids, but show less reactivity to FVIII in complex with VWF. Similar inhibitory effects have been described for anti-C1 human antibodies. An anti-wild type FVIII antibody developed in a patient with mutation Arg2150His that prevented FVIII binding to VWF. Of note, the antibody did not cross-react with the mutant ‘self’ FVIII suggesting that the epitope was at or near Arg2150 .
In addition to concealing certain epitopes on FVIII, VWF interaction also prevents FVIII-binding to antigen-presenting cells (APCs) such as macrophages and dendritic cells and thereby subsequent activation of CD4+ T-cells . The macrophage mannose receptor CD206 present on macrophages and dendritic cells has been shown to mediate FVIII endocytosis. Mannose-terminating glycans are found on Asn239 in the heavy chain, and Asn2188 in the light chain. VWF-binding to FVIII prevents binding of FVIII to CD206 , thereby downregulating the immune ability to recognize FVIII. Thus binding of FVIII to VWF by preventing, upstream from the activation of immune effectors, the entry of FVIII in APCs may reduce its immunogenicity .
It is clear therefore that whilst the direct clinical evidence may be inconclusive, in vitro experiments indicate that formation of the complex with VWF has a direct effect on the immunogenicity of FVIII.
FVIII, VWF and gene therapy
Gene therapy for haemophilia is extensively studied as treatment requires restoration of circulating factor, but not necessarily to normal levels. Expression of even very low levels of the deficient factor would ameliorate the worst symptoms of the disease. The limited clinical trials to date have not shown extensive success, however more recent large-animal studies have demonstrated efficacious long-term expression of factor (see  for review). Most studies have focussed on expression and the secretion into the circulation of either FVIII or FIX. One distinct approach under investigation for therapy of haemophilia A is the expression and storage of FVIII in naturally VWF-expressing cells and tissues.
In vitro studies have shown that FVIII expression vectors can be introduced into and FVIII expressed in endothelial cell lines and megakaryocytes, where expressed FVIII is also trafficked to storage organelles and stored with VWF. Transfection of primary human umbilical vein endothelial cells (HUVECs), endothelial cells from human lung microvasculature and pulmonary arteries with a retroviral B domainless FVIII expression construct resulted in significant expression of FVIII . In addition to being secreted, FVIII was trafficked and co-localized, with VWF in Weibel–Palade (W–P) bodies. Stimulation of the endothelial cells resulted in the secretion of fully functional FVIII together with VWF from the W–P bodies.
Similarly, ectopic FVIII production has been stimulated in megakaryocytes. Transfection of a FVIII expression vector with a platelet-specific promoter resulted in the production and storage of FVIII in α-granules co-localized with endogenous VWF [16,17]. The resulting platelets thus contain FVIII and VWF in the α-granules. Transgenic mouse models demonstrated that the FVIII was co-localized with VWF in the α-granules of platelets, and that the platelet derived FVIII was partly efficacious in restoring haemostasis in a haemophilia mouse model on platelet activation . Notably, platelet derived FVIII was shown to be effective in the treatment of murine haemophilia models even in the presence of inhibitory antibodies [108,109]. Interestingly, cross-breeding experiment of VWF null mice with mice expressing platelet FVIII showed that the presence of VWF was not a requirement for FVIII trafficking to the α-granules, as FVIII was detected in the absence of VWF to 75% of levels in VWF+/+ mice .
Thus the expression of FVIII within VWF-expressing cells, especially megakaryocytes, for the several reasons outlined, offers an attractive strategy for gene therapy, and possibly overcoming the limitations for treatment of patients with inhibitors.
The life-cycle and function of FVIII is inextricably linked with that of VWF, such that in normal circulation the complex of FVIII–VWF can be considered as a single entity. The function of FVIII is enhanced by its delivery to site of injury by VWF, the half-life of FVIII is dependent on its interaction with VWF, being virtually identical when in complex, and as described, VWF interaction with FVIII protects the latter from a variety of proteolytic degradation and removal from the circulation. Very recent in vitro data suggests a role for FVIII in control of VWF multimer size, though this needs to be further investigated. Nonetheless although the structure and function of both molecules have been extensively studied, some questions remain: how and where the FVIII–VWF complex is formed remains poorly defined, and despite the in vitro ability of VWF to bind FVIII at 1:1 molar ratio, the circulatory ratio of approximately 50:1 is relatively constant, with increased VWF associated with increased FVIII. What the determinants are of this ratio and interaction are yet to be clearly elucidated. Nonetheless, in view of the distinct pathologies and complications of haemophilia A and VWD it is useful to remember that FVIII and VWF whilst being independent gene products, circulate as a complex. Consideration of the FVIII–VWF complex as an ‘unit’ may indeed may offer a simpler strategy for future haemophilia therapy.
PVJ is supported by a Bayer Haemophilia Award. JSOD is supported by a Science Foundation Ireland President of Ireland Young Researcher Award (PIYRA 06/Y12/0925), and a Health Research Board Project Grant Award (HRB RP/2006/44).
The authors stated that they had no interests which might be perceived as posing a conflict or bias.