Genetic regulation of plasma von Willebrand factor levels in health and disease

Abstract

Plasma levels of the multimeric glycoprotein von Willebrand factor (VWF) are a complex quantitative trait with a continuous distribution and wide range in the normal population (50-200%). Quantitative deficiencies of VWF (<50%) are associated with an increased risk for bleeding while high plasma levels of VWF (>150%) influence the risk for arterial and venous thromboembolism. Although environmental factors can strongly influence plasma VWF levels, it is estimated that approximately 65% of this variability is heritable. Interestingly, while variability at the VWF gene can account for ~5% of the genetic influence on plasma VWF levels, other genetic loci also strongly modify plasma VWF levels. The identification of the additional sources of VWF heritability has been the focus of recent observational trait mapping studies including genome-wide association studies (GWAS) or linkage analyses, as well as hypothesis-driven research studies. Quantitative trait loci (QTL) influencing VWF glycosylation, secretion, and clearance have been associated with plasma VWF:Ag levels in normal individuals and may contribute to quantitative VWF abnormalities in patients with a thrombotic tendency or type 1 von Willebrand disease (VWD). Identification of genetic modifiers of plasma VWF levels may allow for better molecular diagnosis of type 1 VWD and identify individuals at increased risk for thrombosis. Validation of trait-mapping studies using in vitro and in vivo methodologies has led to novel insights into the life cycle of VWF and the pathogenesis of quantitative VWF abnormalities.

Introduction

Plasma levels of the multimeric glycoprotein coagulation factor von Willebrand factor (VWF) are a continuously distributed complex quantitative trait, ranging between 50 and 200% in the normal population (Figure 1A). Low VWF or von Willebrand disease (VWD), characterized by either qualitative or quantitative VWF deficiency (VWF:Ag < 50%), is the most prevalent hereditary bleeding disorder affecting between 0.1 – 1% of individuals [1-3]. In contrast, epidemiological studies have shown elevated VWF:Ag is an independent risk factor for venous thrombosis, stroke, and ischemic heart disease [4-6]. Despite evidence that VWF plasma levels are highly heritable, variants in the VWF gene account for only a small proportion of this effect. It is increasingly recognized that the phenotypic variability in VWF plasma levels is influenced by several quantitative trait loci (QTL). Large scale studies in recent years have begun to characterize the genetic architecture that regulates VWF plasma levels. The QTL that influence VWF plasma levels in normal and pathological states and the biological mechanism(s) by which they exert their effects will be described in this review.

Figure 1. VWF:Ag is a complex quantitative trait.

(A) Distribution of VWF:Ag levels in Canadian type 1 VWD patients and their unaffected (normal) family members [40,111]. (B) Heritability estimates of VWF:Ag [19-23]. (C) 65% of plasma VWF levels are heritable with variability at the VWF gene, ABO blood group locus, and others contributing to this phenotype [20,70]. (D) The VWF gene is comprised of 52 exons encoding a ~370 kDA mature protein that is comprised of multi-functional domains.

Genetic versus environmental influences on VWF plasma levels

Plasma VWF levels are influenced by genetic, pathological, hormonal, and environmental interactions. VWF is an acute phase protein and numerous inflammatory conditions are associated with elevated plasma levels of VWF [7-9], as pro-inflammatory mediators can stimulate VWF release from endothelial and/or platelet stores [10-12]. VWF levels may also be influenced by physiological events including the menstrual cycle, pregnancy, exercise, aging, and circadian rhythm fluctuation, or environmental exposures including cigarette smoke and air pollution [13-18].

Despite the influence of environmental factors on VWF levels, studies have demonstrated that VWF levels are also highly heritable. While a pedigree analyses has suggested that the genetic influence on VWF levels is approximately 30%, twin and sibling studies find this value is closer to 65% [19-23] (Figure 1B). The composition of the study population can influence this estimate with larger heritability found in populations with diverse genetic background and fewer environmental influences. It is estimated that variability at the VWF locus accounts for ~5% of the heritability of VWF, while the ABO blood group locus contributes approximately 25% [24]. Characterization of the sites that contribute to the unidentified heritability of VWF is thus a focus of ongoing study (Figure 1C).

Influence of variability at the VWF locus on plasma levels of VWF

The VWF gene was first cloned in the mid-1980s [25-28]; it spans 178 kb of genomic sequence and is located on the short arm of chromosome 12. The VWF coding region is comprised of 52 exons that range in size from 40 bp (exon 50) to 1.3 kb (exon 28) (Figure 1D) [29]. Analysis of the VWF gene sequence is complicated by an unprocessed pseudogene (exons 23-34) located on chromosome 22 [30]. Mechanistically, variants in the VWF gene can modify VWF plasma levels by influencing the synthesis, storage and secretion of VWF from endothelial cells or megakaryocytes/platelets, or by regulating the clearance of VWF from plasma.

Normal population:

Both the intronic regions and coding sequence of the VWF gene are known to be highly polymorphic. An analysis of the 1000 Genomes database identified a rate of ~2.5 VWF gene single nucleotide variants (SNVs) per individual, with 8.5% of individuals possessing a non-synonymous coding region variant [31]. The majority of variants (>75%) were rare or previously unreported with unknown functional consequences, although SNVs in the coding and promoter regions in normal individuals have been shown to modify VWF antigen and activity levels [32,33]. Recently, megakaryocyte VWF gene expression has been shown to be regulated by a common SNV located in a super enhancer region located ~55 kb upstream of the VWF gene [34].

Population based studies have shown that VWF levels are influenced by ethnicity, as African Americans have ~15% higher VWF levels than Caucasian populations [35]. Recent studies have highlighted the high incidence of ethnic-specific VWF sequence variations, with Africans having greater diversity in the VWF gene than non-Africans [31,36,37]. The highest level of diversity was found in the D′ and D2 domains [31], which may, in part, contribute to elevated VWF levels.

“Low VWF” or VWD populations (<0.5 IU/mL):

Type 3 VWD (VWF:Ag <0.03 U/mL), results from the inheritance of two mutant VWF alleles that cause severe deficiency of plasma VWF. Type 3 VWD can have either a recessive or co-dominant pattern of inheritance with obligate carriers displaying a type 1 VWD phenotype in 25-50% of type 3 families [38]. Type 3 VWD shows significant allelic heterogeneity, and large deletions, frameshift, missense, splice site, and nonsense mutations have been described throughout the gene. Approximately 20% of mutations are missense variants that presumably result in the expression of VWF protein whose biosynthesis and secretion is significantly disrupted. Thus, the mechanistic basis of type 3 VWD is complex, and may involve impaired synthesis, Weibel Palade body (WPB) formation, or secretion [39].

The clinical definition of partial quantitative VWF deficiency, historically termed type 1 VWD, has been variable, with a VWF:Ag cut-off level ranging between 0.03 and 0.3 – 0.5 U/mL used for studies investigating the molecular pathogenesis of this disorder. For these patients, pathogenic SNVs result in missense substitutions that occur throughout the coding region of the VWF gene, although some SNVs occur in the promoter, or at intron/exon boundaries (Figure 1D) [40-44]. While rare recessive cases of type 1 VWD have been reported, the disorder is predominantly inherited in an autosomal dominant manner [45]. Approximately 85% of type 1 VWD cases are associated with defective VWF synthesis, storage and/or secretion from the endothelium. Type 1 secretion variants tend to localize in the VWF D1, D2, and D3 domains [46], and can result in intracellular retention in the endoplasmic reticulum, irregular WPB formation or reduced secretion [47-49]. Rare cases of decreased VWF synthesis related to variants in the VWF promoter have also been described [50].

The nomenclature, type 1C VWD, while not a currently recognized ISTH classification, is often used to describe accelerated clearance variants which make up ~15% of all type 1 cases [51,52]. Studies using the VWF propeptide to antigen ratio (VWFpp/VWF:Ag) as a surrogate measure of VWF clearance have indicated that accelerated clearance occurs more frequently with more severe VWF deficient states [46]. Type 1C pathogenic variants are most frequently observed in the D3, A1, and D4 domains of VWF [46]. While little is currently known regarding the mechanistic basis by which these variants regulate VWF half-life, they may alter either the glycosylation or conformation of the VWF molecule, and likely enhance the affinity of VWF for one or more of its clearance receptors [53]. Type 2 VWD, which is predominantly characterized by platelet, collagen or FVIII-binding qualitative defects may also be complicated by impaired VWF secretion or accelerated clearance of the VWF-platelet complex [54,55]

The diagnosis of type 1 VWD may be complicated by variability in phenotypic penetrance and expressivity. For example, several studies have now documented that pathogenic variants identified in European type 1 VWD subjects can have minor allele frequencies of 10-20% in normal asymptomatic African American individuals [31,36,37]. While this phenomenon may be partially related to false attribution of variant pathogenicity, in vitro characterization demonstrates that some of these variants display minor quantitative defects. Thus, it seems likely that some mild type 1 variants co-segregate with additional ethnic-specific variants that modify their pathogenicity, or that the bleeding phenotype associated with these variants may be influenced by external environmental interactions. For example, gene-environment interactions have been shown to modify the type 1 VWD phenotype where age-related changes can normalize VWF:Ag levels in older individuals [56,57].

Recent recommendations from the UK Haemophilia Centre Doctors Organization and the US National Heart, Lung, and Blood Institute (NHLBI) have further sub-classified partial VWF quantitative deficiency into type 1 VWD (0.03 – 0.3 U/mL) and a “Low VWF” phenotype (0.3 – 0.5 U/mL) [58,59]. Overall, approximately 35% patients with partial quantitative VWF deficiency do not have an identified pathogenic variant in their VWF coding region or consensus splice sites with a lower proportion of coding region variants identified in “Low VWF” individuals [40-44,60,61]. While some the pathogenic variants have been hypothesized to be found in distal regulatory regions or deep intronic sequences of the VWF gene, linkage analysis performed on two separate, cohorts demonstrated that the proportion of families that show linkage to the VWF locus is ~0.44 [62,63]; thus, non-VWF gene QTL may contribute to the “Low VWF” phenotype in the absence of a pathogenic VWF gene variant.

Thrombosis population:

Epidemiological studies have demonstrated that elevated VWF levels are a risk factor for venous and arterial thrombosis, and that patients diagnosed with type 1 VWD have a decreased incidence of thrombosis [64]. SNVs found throughout the VWF gene, including the promoter, coding regions, and introns, have been associated with either elevated VWF:Ag levels, or risk for venous thrombosis or coronary heart disease [20,65-68] (reviewed in detail by Van Schie et al.[69]). For example, the common VWF non-synonymous variant c.2365A>G (p.Thr789Ala), which is in strong linkage disequilibrium (r=99%) with the synonymous variant c.2385T>C (p.Tyr795=), associates with both increased VWF levels and risk for VTE (OR=1.2) [67,70]. These variants influence VWF synthesis/secretion in a heterologous expression system and also demonstrate evidence of increased VWF half-life [71,72].

Identification of non-VWF Loci that influence VWF levels

Non-VWF gene QTL contribute to the majority of the heritability of VWF plasma levels in normal individuals. They may also modify the quantitative VWF abnormalities that contribute to either the type 1 VWD phenotype, or to a hypercoagulable state. Identification of these additional loci has been the focus of both candidate gene-driven studies where knowledge of VWF (patho)biology is used to identify genes that modify VWF levels, and observational trait mapping studies (such as genome-wide association studies (GWAS) or linkage analyses) where variants that influence VWF levels are assessed in an unbiased manner. Trait-based association studies have thus far provided an initial insight into the architecture of the genetic regulation of VWF plasma levels, and have documented a number of QTL, estimated their effect size, and investigated their potential to mediate epistatic and pleiotropic effects on VWF levels (Table 1). They may also support or bring into question previous hypothesis driven studies.

Table 1. Non-VWF and ABO-variants associate with plasma levels of VWF in normal individuals.

MAF = minor allele frequency. N/A=not available. For some genes the most significant SNVs only were chosen.

Gene	rfSNP ID	HGVS nomenclature	Position	MAF	Putative Mechanism	Study type	Refs
ACE	rs 1799752	c.644-119_644-118insG	Intron	2.59x10−5	Secretion?	Gene candidate	87
AVPR2	rs2071126	c.35G>A p.Gly12Glu	Exon 2	0.0455	Secretion	Gene candidate	85
BAI3	rs9363864	g.68182664G>A	3’ UTR		Unknown	GWAS	68
CLEC4M	rs868875	c.631+73A>G	Intron 4	0.262	Clearance	GWAS	63
FUT1	rs104894686	c.948C>G p.Tyr316Ter	Exon 5	8.24x10−6	Glycosylation	Gene candidate	77
FUT2	rs601338	c.461G>A p.Trp154Ter	Exon 2	0.39	Glycosylation	Gene candidate	77
LRP1	rs34577247	c.6238G>A p.Asp2080Asn	Exon 39	0.0149	Glycosylation	Gene candidate	101, 102
LRP1	rs1800127	c.650C>T p.Ala217Val	Exon 6	0.017	Clearance	Gene candidate	101, 102
SCARA5	rs2726953	c.242-21543C>T	Intron 3	0.309	Clearance	GWAS	63
ST3GAL4	rs2186717	c.−60-14159G>A	Intron 1	0.488	Glycosylation	Gene candidate and GWAS	83
STAB2	rs4981022	c.6987+378G>A	Intron 63	0.315	Clearance	GWAS	63
STAB2	rs141041254	c.7129G>A p.Glu2377Lys	Exon 65	8.5x10−4	Clearance	GWAS	107
STX2	rs7978987	c.786+1576C>T	Intron 9	0.346	Secretion?	GWAS	63
STXBP5	rs9390459	c.2445A>G p.Leu815=	Exon 23	0.442	Secretion	GWAS	63
TC2N	rs 10133762	c.−57+9946A>C	Intron 1	0.444	Unknown	GWAS	63
UFM1	rs17057285	g.38737821 A>C	200 kb upstream	0.036	Unknown	GWAS	67

To date, VWF modifying QTL have been identified in two genome-wide linkage analyses. The Genetic Analysis of Idiopathic Thrombophilia (GAIT) study [23], performed in 21 Spanish families (342 individuals) identified six QTL with LOD scores >1 that associate with plasma VWF levels, including the ABO locus at 9q34, as well as regions on chromosomes 1p36.13, 2q23.2, 5q31.1, 6p22.3, and 22q11.1. Importantly, the 2q12 locus was replicated in a separate genome wide linkage analysis of young siblings [20], and three of these regions show synteny with QTL linked to VWF levels in mice [73].

Linkage analysis data has now been complemented with a number of GWAS studies, the largest of which is the CHARGE (Cohorts for Heart and Aging Research in Genome Epidemiology) consortium GWAS meta-analysis [70]. Plasma levels of VWF and/or FVIII were correlated with ~2.6 million alleles in an initial discovery population of over 23,000 normal subjects of European ancestry collected from five studies, and replication analysis was performed in 7,600 additional subjects. VWF levels associated with over 400 SNVs at eight loci with a genome-wide significance threshold of 5.0x10⁻⁸: VWF, ABO, STAB2, SCARA5, TC2N, STXBP5, STX2, and CLEC4M. Together, these loci explained 12.8% of the variation in plasma VWF levels. The majority of these variants were located within introns and may mediate their influence through a number of potential mechanisms: altering mRNA splicing patterns or efficiency, modifying levels of gene expression, or at least as likely, may be in linkage disequilibrium with coding region variants that modify protein function. Non-synonymous variants may also influence protein translation through splicing effects, codon use bias or by altering mRNA stability. To date, the association between VWF and the novel loci identified in the CHARGE study has been replicated in several additional GWAS analyses, with a number of new candidate loci identified [74-76].

An updated GWAS meta-analysis from the CHARGE consortium, recently described in abstract form, included results from a discovery cohort of >46,000 normal individuals of European, African, East Asian, and Hispanic descent [77]. Eleven additional QTLs were found to associate with plasma VWF levels: ARSA, C2CD4B, DAB2IP, FCHO2, GIMAP7, HLA-DGA1, OR13C5, PDHB, RAB5C, ST3GAL4, TAB1/SYNGR1. The putative function of the majority of these novel loci appears to involve the regulation of VWF secretion from endothelial cells and/or megakaryocytes, however these associations await further confirmation and validation.

Both observational and hypothesis-driven genetic studies have the capability of identifying common variants that have both small and large influences on VWF plasma levels, and rare variants that impart a large influence (Figure 2). Variants external to the VWF gene may modify the severity or penetrance of type 1 VWD, or conversely, independently associate with elevated VWF levels or an increased incidence of thrombosis (Table 2). Complementary experimental studies are required to identify the mechanistic basis by which the GWAS-identified variants convey their effects, to understand the biological pathway that results in altered VWF plasma levels, and to rule out the possibility of false positive results. The majority of VWF QTL fall into three broad categories: genes that modify VWF glycosylation, secretion from endothelial cells or platelets, and clearance from the plasma.

Figure 2. Rare and common genetic variants influence VWF levels.

VWF plasma levels are influenced by both rare and common variants in the VWF gene, ABO blood group locus, and additional loci. LOF = loss of function.

Table 2. Influence of non-VWF and ABO-variants on quantitative VWF pathologies.

DVT = deep vein thrombosis. CHD = coronary heart disease.

Study design	Population	Gene(s) investigated	Major Finding(s)	Ref
Association of CHARGE SNVs with VTE	656 women with DVT, 710 controls	CHARGE SNVs	-SNVs in STXBP5 and VWF associate with DVT	60
SNVs associated with VTE, plasma VWF levels	1166 VTE patients, 1408 healthy subjects, 5 extended families	Linkage analysis, GWAS	-ABO locus, STAB-2, BAI3, 2p12-13 regions associate with risk for VTE -BAI3 SNV variants associate with VWF levels	68
Association between CHARGE SNVs and VWD (type 1& 2)	364 type 1 VWD, 240 type 2 Netherland VWD patients	CHARGE SNVs	-SNVs in STXBP5 and CLEC4M associate with low VWF levels in type 1 VWD	90
Association of CLEC4M variants with type 1 VWD	318 type 1 VWD patients and 173 unaffected family members	CLEC4M VNTR polymorphism and CHARGE SNV	-association of CLEC4M VNTR with unaffected individuals, VWF:RCo	103
Association of STXBP2 and STX2 variants with arterial thrombosis	463 arterial thrombosis patients, 406 controls	STXBP5 and STX3 CHARGE SNVs	-SNVs in STXBP5 and STX2 associate with VWF:Ag in patients with arterial thrombosis -SNVs in STX2 associate with risk for arterial thrombosis	92
Association of STXBP2 and STX2 variants with type 1 VWD	158 type 1 VWD patients	STXBP5 and STX2 CHARGE SNVs	-SNVs in STX2 associate with VWF:Ag in type 1 VWD -SNVs in STXBP5 associate bleeding score in females with type 1 VWD	91
Association between variants in VWF gene and CHD in younger individuals	421 young CHD patients and 409 healthy controls	VWF gene	-SNVs in VWF are associated with elevated VWF levels and risk for cardiovascular disease	59
Association between VWF promoter SNV and CHD	352 subjects with CHD, 736 controls	VWF promoter	-a SNVs in the VWF promoter is associated an increased risk of CHD in subjects with advanced atherosclerosis.	58
Association between CHARGE SNVs and VT	1744 VT patients and 1389 healthy controls	CHARGE SNVs	-TC2N varient associates with increased risk for VTE	117

Genes influencing VWF glycosylation

VWF is produced in endothelial cells and megakaryocytes in a complex biosynthetic process that involves C-terminal dimerization, and cleavage of an N-terminal pro-peptide, followed by N-terminal multimerization, with subsequent N- and O-linked glycosylation. VWF possesses 13 potential N-linked and 10 O-linked glycan sites that constitute approximately 20% of the molecular mass of the mature protein (Figure 3A). Glycomic analysis has revealed the majority of VWF N-linked glycans to be of complex type, with ABO(H) expressed as terminal sugars [78,79]. Mutagenesis studies have demonstrated that both N- and O-linked glycans on the VWF molecule are involved in regulating its secretion in a heterologous cell system and its clearance from the plasma [80,81].

Figure 3. VWF:Ag levels are influenced by VWF glycosylation.

(A) Post-translational modification of VWF includes the addition of 16 putative N-linked and 10 putative O-linked glycans. (B) ABO(H) blood group antigens are added to the VWF N- and O-linked glycans. (C) Dysfunction of FUT1/2 and ABO blood group status can influence VWF plasma levels. Bombay = FUT1 and FUT1/FUT2 null, Se=FUT2, se=FUT2 null.

FUT1/2:

FUT1/2 are fucosyltransferases that produce the H antigen by transferring a fucose to the glycan core structure (Figure 3B). FUT1 encodes the H enzyme that regulates H antigen formation on VWF, while FUT2 encodes the Se enzyme, and regulates H antigen formation on red blood cells, and secretions in the mucosa and gastrointestinal tract respectively. The core H antigen structure is then either modified by additional glycosyltransferases to create complex N-linked glycans. Individuals with both Bombay phenotype (FUT1/2 deficient), and para-Bombay phenotype (FUT1 deficient) have no VWF H antigen, and VWF plasma levels of ~70% (Figure 3C), suggesting that FUT1 fucosyltransferase activity is a significant regulator of plasma VWF levels [82]. While conflicting reports exist, FUT2 variants have been shown to also associate with VWF plasma levels [19,83,84]; the mechanism by which this effect is exerted is unknown, but may involve FUT2 modification of pathways that regulate VWF secretion or clearance.

ABO:

The ABO blood group locus is located on chromosome 9 and is highly polymorphic. The locus encodes either A or B glycosyltransferase alleles, which modify the core H antigen by adding N-acetylgalactosamine or galactose respectively to the core galactose (Figure 3B). Type O individuals do not encode a functional AB glycosyltransferase and thus express only H antigen. ABO was first identified as a modifier of VWF levels in the 1980s [24], and has been repeatedly identified as the locus with the strongest influence on plasma VWF levels in trait-mapping studies [20,70,74]. Approximately 25% of the variability in plasma VWF levels can be accounted for by the ABO system, with blood type O (H-antigen) associating with ~25% lower VWF:Ag levels than non-O individuals [24]. Concordantly, individuals heterozygous for O and either A or B glycosyltransferase alleles (AO or BO) have intermediate levels of VWF, while AB and AA individuals have high levels of VWF:Ag (Figure 3C).

The ABO blood group genotype can modify partial quantitative VWF deficiency, with type O individuals more likely to be diagnosed with either type 1 VWD or “low VWF”, and more prone to severe bleeding diatheses than non-O type 1 VWD patients [24], thus contributing to the variable expressivity of this condition. Conversely, the elevated levels of VWF:Ag observed in non-O individuals in part explains the increased risk for thrombosis associated with these blood types. While type O normal individuals and patients with the type 1C VWD “Vicenza” variant have a decreased VWF half-life in response to DDAVP treatment [85,86] the clearance pathway(s) that underlies this effect is currently unknown. Recent VWF half-life studies performed in type 3 patients suggest that non-VWF ABO(H) glycans may also contribute mechanistically to the VWF clearance phenotype [87].

ST3GAL4:

ST3GAL4 encodes ST3 beta-galactoside alpha-2,3-sialyltransferase 4, which facilitates the transfer of terminal sialic acid sugars to the N- and O-linked glycans of VWF. Desialylation of VWF has been associated with increased VWF clearance by the asialoglycoprotein receptor (ASGPR) (Figure 4B) and results in the decreased VWF half-life in humans post DDAVP administration [88,89]. In mice, enzymatically desialylated VWF also has a significantly decreased half-life [89], and ST3GAL4-deficient mice have low VWF levels and a bleeding phenotype [90]. Sequencing of the ST3GAL4 gene in normal individuals has identified six clustered SNVs in the first intron that associate with plasma levels of VWF [91] suggesting that polymorphisms in the ST3GAL4 gene can regulate the degree of VWF sialylation.

Figure 4. VWF:Ag levels are influenced by processes that regulate VWF synthesis/secretion and mechanisms that regulate VWF clearance.

(A) VWF synthesis/secretion can be influenced by variants in or adjacent to the VWF gene that regulate VWF transcriptional activity, mRNA stability, codon use, protein folding, and Weibel Palade Body packaging and secretion. Variants at additional loci including SNARE proteins can influence the secretion of VWF from endothelial WPB or platelet α-granules. (B) VWF clearance can be influenced by variants in the VWF amino acid sequence and/or glycome that modify the affinity of VWF for one or more clearance receptors. Additionally, variants in the clearance receptors for VWF that alter ligand binding or expression can modify VWF plasma levels. HL = hepatic lectin, CBD = carbohydrate binding domain, EGF = endothelial growth factor, VNTR = variable number of tandem repeats, SRCR = scavenger receptor cysteine-rich.

Genes influencing VWF secretion:

Upon synthesis, mature VWF is packaged into intracellular storage vesicles termed Weibel-Palade bodies (WPB) in endothelial cells or α-granules in platelets and released constitutively or on-demand upon hemostatic challenge (Figure 4A). Platelet activation and WPB secretion can be stimulated by a variety of pro-inflammatory secretagogues including thrombin and desmopressin that signal through endothelial cell surface receptors including PAR-1 and vasopressin receptors. Secretion of VWF from platelets and endothelial cells is regulated by SNARE (soluble NSF attachment protein receptor) complexes that direct the fusion of secretory vesicles with the cell plasma membrane.

AVPR2:

Stimulation of the arginine vasopressin 2 receptor (AV2R) with arginine vasopressin induces the exocytosis of WPB, releasing VWF into the circulation in an on-demand fashion [92]. The AVPR2 ligand desmopressin (or DDAVP) is used in clinical practice to transiently increase circulating VWF in type 1 VWD. The gain-of-function p.Gly12Glu variant in AVPR2, that enhances binding of arginine vasopressin, is associated with elevated plasma VWF levels in normal individuals [93]. Thus, patients with inherited nephrogenic diabetes insipidus associated with pathogenic variants in the AVPR2 gene are unresponsive to VWF release upon DDAVP administration [94].

ACE:

Angiotensin-converting enzyme (ACE) regulates blood pressure by converting angiotensin I to angiotensin II, a potent vasoconstrictor. An insertion/deletion variant in ACE has been described to increase the risk of cardiovascular disease, potentially through endothelial dysfunction and modification of plasma levels of coagulation factors, including VWF, in elderly patients with hypertension [95]. As high blood pressure is associated with increased VWF plasma levels, it is likely that dysfunction of ACE promotes the release of VWF from the damaged endothelium, although this mechanism has yet to be confirmed.

SNARE proteins:

STX2 (syntaxin 2), STXBP5 (syntaxin binding protein 5), and STXBP1 are members of the SNARE family expressed in endothelial cells and/or platelets. These proteins have all been implicated in regulating plasma VWF levels in both GWAS (STX2 and STXBP5) and hypothesis-driven studies (STXBP1) (Figure 4A). In endothelial cells, STXBP5 functions as a negative regulator of VWF exocytosis [96,97], while in platelets the same protein promotes the release of VWF from α-granules. Interestingly, STXBP5-deficient mice have elevated levels of VWF, but counterintuitively display prolonged bleeding times and impaired arterial thrombosis, that may be related to abnormal platelet granule formation and secretion [96]. Variants in STXBP5 have been associated with VWF levels and bleeding severity in type 1 VWD [98,99], as well as elevated VWF levels in patients with arterial thrombosis [100]. STXBP5 variants can also modify the risk for venous thrombosis [67], although this association may involve both regulation of VWF levels as well as additional procoagulant factors associated with WPB or α-granule secretion.

Similar to STXBP5, variants in STX2 have been shown to influence VWF levels in normal individuals [67] and arterial thrombosis patients, and independently associates with an increased risk for arterial thrombosis [70,100]. However, conflicting reports exist on the association between variants in STX2 and plasma VWF levels in type 1 VWD [98,99]. In endothelial cells, syntaxin-2 forms a complex at the cell membrane with STXBP1 and Slp4-a, a Rab27a effector that localizes to WPBs and mediates their exocytosis [101]. STX2 is also involved in vesicle exocytosis in other cell types. However, STX2-defiency does not influence VWF release from platelet α-granules [102] and the direct influence of syntaxin-2 on VWF secretion from endothelial cells has yet to be described.

While common variants in STXBP1 have not been identified in GWAS of VWF levels, rare variants in this gene may contribute to a “low VWF” phenotype. STXBP1 loss-of-function is associated with the rare, severe epileptic disorder, EIEE4 (early infantile epileptic encephalopathy-4), related to impaired neurotransmitter release [101]. Studies of endothelial progenitor cells derived from an EIEE4 patient demonstrated decreased VWF secretion (~50% of wild-type), and low VWF plasma levels, but no reported association with a bleeding phenotype.

Genes influencing VWF clearance:

Clearance of VWF and FVIII proceeds through carbohydrate, amino acid and protein conformation-recognition based pathways, making this a semi-selective process that features numerous receptor-ligand interactions (Figure 4B). A growing number of VWF clearance receptors have been identified through a range of genetic, molecular and cellular approaches including LRP1, the macrophage galactose type lectin, Siglec-5, galactin-3 and −5, CLEC4M, stabilin-2, and the asialoglycoprotein receptor [89,103-108]. In this review, only clearance receptors with genetic variation that modify VWF plasma levels will be described.

LRP1:

The macrophage-expressed LRP1 (low-density lipoprotein receptor related protein) is a member of the LDLR superfamily of receptors, and is comprised of repetitive clustered repeats (I–IV) containing ligand binding and EGF homology region sequences that bind VWF under conditions of shear [104]. In mice, macrophage LRP-1 deficiency increases plasma VWF and FVIII levels, while both the LRP-1 ligand RAP and macrophage LRP-1 deficiency increase VWF half-life [104]. Although conflicting data exists, the rare, non-synonymous SNVs in the LRP-1 gene, p.Asp2080Asn and p.Ala217Val, have been shown to be associated with plasma levels of VWF and FVIII [109,110]. Additional receptors in the LDLR superfamily may also regulate VWF clearance, however, to date, genetic variants at these loci that modify this phenotype have not been described.

CLEC4M:

The CHARGE GWAS identified two receptors, CLEC4M and stabilin-2, expressed on the sinusoidal endothelium of the liver that associate with VWF plasma levels [70]. CLEC4M (C-type lectin domain family 4 member M) was originally characterized as an adhesive receptor involved in regulating pathogen infection. CLEC4M is comprised of a mannose-binding lectin domain, and a polymorphic neck region comprised of a variable number of tandem repeats (VNTRs) that mediates homotetramerization of the molecule and is in strong linkage-disequilibrium with the CHARGE SNV. We have characterized CLEC4M as a lectin receptor that mediates the endocytosis of VWF and FVIII via interactions with N-linked glycans [111,112]. Moreover, variants in CLEC4M including the VNTR polymorphism and the CHARGE SNV associate with plasma VWF levels in type 1 VWD populations [98,111].

STAB2:

Stabilin-2 is a scavenger receptor comprised of a repeating series of EGF-like, and fasciclin-1 (FAS-1) domains with an X-link domain located proximal to its transmembrane region that is C-type lectin-like. Interestingly, while human and murine stabilin-2 have retained an approximately 80% amino acid identity, stabilin-2 deficient mice exhibit normal VWF levels and normal half-life of murine VWF [113]. However, the half-life of recombinant or plasma-derived human VWF is increased by approximately 2-fold in stabilin-2 deficient mice, suggesting that stabilin-2 regulates the clearance of human but not murine VWF.

The CHARGE STAB2 SNVs associate with VWF:Ag levels in the Canadian type 1 VWD population, and with low plasma VWF levels in a separate GWAS [74,113]. In contrast, neither STAB2 SNV associated with type 1 VWD in a Dutch cohort [98]; this observation may be linked with differences in the composition of the study populations. Pathogenic variants in the STAB2 gene have been found to associate with large elevations in plasma VWF:Ag levels (>33%) in both association and exome sequencing studies [114,115], and modify the ability of stabilin-2 expressing cells to bind and internalize VWF in vitro [113]. Pathogenic STAB2 variants have also been shown to associate with the incidence of VTE [117], however given the scavenger function of stabilin-2, altered clearance of non-VWF plasma coagulation factors or inflammatory regulators may also contribute to the pathogenic basis of this observation.

SCARA5:

SCARA5 (Scavenger Receptor Class A Member 5) is a pattern recognition receptor expressed on murine epithelial cells and interstitial fibroblasts in the spleen [118]. SCARA5 assembles at the cell surface as a homo-trimer and contains an extracellular collagenous and cysteine rich scavenger domain. SCARA5 can bind a range of ligands including ferritin, polyanions, and microbes [119,120]. Solid phase binding assays have confirmed that VWF binds to SCARA5, and HEK 293 cells transfected with the SCARA5 cDNA are capable of binding and endocytosing both VWF and FVIII in a VWF-dependent manner [121]. However, to date, there is little available knowledge regarding SCARA5 expression in humans, and the in vivo location of VWF-SCARA5 interaction is unknown.

Genes influencing VWF multimerization:

VWF circulates in the plasma as a series of multimers ranging in size between 0.5–20 mDa with larger multimers possessing increased hemostatic activity. Modification of VWF multimer size by ADAMTS13 and thrombospondin-1 has been proposed to regulate VWF clearance, but animal models have failed to demonstrate a conclusive influence of proteolysis on VWF clearance in vivo [122]. In human studies, SNVs in the ADAMTS13 gene have been associated with plasma VWF levels [20], however this signal has been attributed to the ABO locus, as both genes are located within 136 kb on chromosome 9. Interestingly, variants in both ADAMTS13 and TSP-1 genes have been associated with an increased risk for cardiovascular disease, presumably through the regulation of VWF multimeric size and hemostatic activity [123,124].

Unknown Mechanisms

For several of the novel genetic loci that are associated with plasma levels of VWF, the functional relationship between VWF and the gene product is currently unknown, highlighting the challenge of mapping intergenic variants to the correct locus and identifying of causal variants associated with a genetic signal. Associations with the chromosome region 2q12–2p13, and the TC2N gene have been replicated by independent studies, increasing the probability that these two loci represent true genotype-phenotype associations and are not artifacts. Other associations, including BAI3 and UFM1 have not been replicated to date and additional work is required to test their influence on VWF [74,75].

Chromosome 2q12–2p13:

Two linkage analysis studies have identified signals at chromosome 2q12–2p13 that associate with VWF levels and contributed to 19.2% of its variability [20,75]. This locus has not been detected in GWAS, which may be related to the presence of rare pathogenic variants in this region that strongly regulate VWF levels. The identified linkage interval on chromosome 2q12-2p13 contains over 100 genes in a region ~34 Mb in size, and VWF modifying candidates in this region include SNARE proteins and glycosyltransferases. A subsequent GWAS and linkage analysis of the VWF propeptide in the same study population revealed no association in the 2q12-2p13 region [72], suggesting that variants in this genomic region modify VWF clearance rather than synthesis/secretion.

TC2N:

The association between VWF/FVIII levels and variants in TC2N (tandem C2 domains, nuclear), were first identified in the CHARGE study and subsequently replicated in additional cohorts [70,75]. The C2 domains of TC2N feature a basic cluster that imparts nuclear localization capabilities [125], however, to date, the mechanistic basis by which TC2N modulates VWF plasma levels is unknown. Importantly, TC2N variants may associate with an increased risk for venous thrombosis [126].

Variants influencing FVIII plasma levels

VWF circulates in the plasma in a non-covalent complex with the coagulation cofactor FVIII. Binding of VWF to FVIII protects FVIII from accelerated proteolysis and clearance, and both the plasma levels of VWF, and the binding constant between VWF and FVIII strongly influence plasma FVIII levels. Therefore, QTL that modify plasma levels of VWF tend to concomitantly modify FVIII levels, and the pathophysiological influence of these QTL on bleeding or thrombotic tendencies may therefore also involve altered plasma FVIII levels. Qualitative VWF abnormalities that involve pathogenic variants in the VWF D’D3 region that impair FVIII binding result in type 2N VWD involving isolated FVIII deficiency.

Interestingly, infrequent F8 gene variants have been associated with plasma FVIII:C in normal individuals [70,115], while variability in the VWF gene and ABO blood group locus account for approximately 50% of the variability in plasma FVIII:C [109]. Recent studies have suggested that for every 1% change in plasma VWF levels, FVIII activity (FVIII:C) levels will change ~0.54% [127]. Therefore, it follows that the majority of VWF QTL also associate with FVIII activity, although the magnitude of influence may be lower [70,76,128]. Importantly, both ABO blood group status and other variants that modify VWF clearance influence FVIII replacement pharmacokinetics in hemophilia A patients [129-131]

Future directions

Trait mapping studies have thus far provided an unprecedented understanding of the genetic architecture that regulates plasma VWF levels, while cell and animal-model based confirmatory studies have provided novel insights into the life cycle of VWF as well as the pathogenetic basis of both VWD and thrombosis. To date, these data indicate that with the exception of the ABO locus, common variants associated with a number of VWF-modifying genes exert a relatively small effect on plasma VWF:Ag levels, while a smaller number of rare variants can have a larger influence on plasma VWF:Ag. However, a substantial proportion of the heritability of VWF levels currently remains uncharacterized. Future studies that will assess the influence of rare variants on plasma VWF levels through whole genome/exome analysis will increase our understanding of the genetic basis of pathological quantitative abnormalities of VWF plasma levels, including expressivity, penetrance, epistasis, and gene-environment influences. This may allow for the development of algorithms that improve the molecular diagnosis of type 1 VWD, provide further insights into the pathogenic basis of the “low VWF” phenotype, or improve the identification of individuals at-risk for thrombosis. Moreover, the development and improved accessibility of bleeding assessment tools (BATs) will increase our understanding of how these variants modify the risk for bleeding in individuals with inherited coagulopathies. These studies may also facilitate the development of personalized therapies with improved coagulation factor pharmacokinetic profiles for inherited or acquired bleeding disorders. Ultimately, technical advances in high throughout sequencing and increased availability of genomic information will improve the molecular diagnostic evaluation and clinical care of patients with quantitative VWF abnormalities.