The contribution of the sinusoidal endothelial cell receptors CLEC4M, stabilin-2, and SCARA5 to VWF–FVIII clearance in thrombosis and hemostasis

Abstract

Quantitative abnormalities in factor VIII (FVIII) and its binding partner, von Willebrand factor (VWF), are associated with an increased risk of bleeding or thrombosis, and pathways that regulate the clearance of VWF-FVIII can strongly influence their plasma levels. In 2010, the Cohorts for Heart and Aging Research in Genome Epidemiology (CHARGE) on genome-wide association study meta-analysis identified variants in the genes for the sinusoidal endothelial receptors C-type lectin domain family 4 member M (CLEC4M), stabilin-2, and scavenger receptor class A member 5 (SCARA5) as being associated with plasma levels of VWF and/or FVIII in normal individuals. The ability of these receptors to bind, internalize, and clear the VWF-FVIII complex from the circulation has now been reported in a series of studies using in vitro and in vivo models. The receptor stabilin-2 has also been shown to modulate the immune response to infused VWF-FVIII concentrates in a murine model. In addition, the influence of genetic variants in CLEC4M, STAB2, and SCARA5 on type 1 von Willebrand disease/low VWF phenotype, FVIII pharmacokinetics, and the risk of venous thromboembolism has been described in a number of patient-based studies. Understanding the role of these receptors in the regulation of VWF-FVIII clearance has led to significant insights into the genomic architecture that modulates plasma VWF and FVIII levels, improving the understanding of pathways that regulate VWF-FVIII clearance and the mechanistic basis of quantitative VWF-FVIII pathologies.

1 |. INTRODUCTION

Plasma levels of coagulation factors, including the multimeric glycoprotein von Willebrand factor (VWF) and its binding partner factor VIII (FVIII), are influenced by the rate at which they are synthesized and secreted into the plasma and the rate at which they are cleared from the circulation. While plasma levels of VWF and FVIII can vary widely in the normal population (50%–200%), it is well recognized that low levels of VWF (“low VWF” phenotype and type 1 von Willebrand disease [VWD]) and FVIII (hemophilia A, type 2N VWD) associate with an increased risk of bleeding. Moreover, elevated VWF antigen (VWF:Ag) or FVIII activity (FVIII:C) levels increase the risk of both venous and arterial thrombosis. As plasma levels of VWF:Ag and FVIII:C have a heritability of approximately 65% and 58%, respectively [1–5], it follows that genetic variants that influence the rate of VWF-FVIII clearance can therefore modify the risk of bleeding or thrombosis. VWF serves as a carrier for FVIII and protects FVIII from accelerated clearance and proteolysis [6]. Plasma VWF levels correlate with plasma FVIII levels, with every 1% change in VWF:Ag levels resulting in a 0.54% change in FVIII:C [7]. Variants that modify plasma levels of VWF therefore tend to also associate with plasma FVIII activity, albeit with a decreased magnitude of effect [8,9].

Both observational and hypothesis-driven studies have elucidated the complex series of pathways by which VWF-FVIII is cleared from the plasma. Initially, macrophage-expressed receptors were the major focus of study [10]. However, the publication of the Cohorts for Heart and Aging Research in Genome Epidemiology (CHARGE) genome-wide association study (GWAS) meta-analysis in 2010 identified variants in the genes of 3 putative VWF-FVIII clearance receptors, C-type lectin domain family 4 member M (CLEC4M [CLEC4M]), stabilin-2 (STAB2), and scavenger receptor class A member 5 (SCARA5 [SCARA5]), expressed by sinusoidal endothelial cells in the liver or spleen [9]. A subsequent series of studies focusing on in vitro solid phase binding, cell-based assays, and animal models have confirmed the ability of these receptors to regulate the clearance of VWF-FVIII. Moreover, human genetic studies have demonstrated the influence of common and rare variants in the CLEC4M, STAB2, and SCARA5 genes on type 1 VWD, FVIII pharmacogenomics, and risk of venous thromboembolism (VTE).

2 |. MACROPHAGE AND HEPATOCYTE CLEARANCE OF VWF-FVIII

The role of the liver and spleen in regulating the clearance of VWF and FVIII has been confirmed by several animal studies using infused radiolabelled ligands [10,11]. While the spleen has a higher endocytic capacity for VWF and FVIII than the liver by tissue weight, the relative size and proportionate blood flow suggest that the majority of VWF-FVIII is cleared by cells in the liver. Macrophages, the predominant tissue-based component of the mononuclear phagocyte system, were the first cell type investigated to contribute to physiological VWF-FVIII clearance. Macrophage-mediated clearance of VWF-FVIII has been demonstrated by both immunohistochemical analyses of VWF-FVIII infused VWF knockout (KO) mice as well as VWF-FVIII clearance studies in macrophage-depleted mice [10,12]. Additionally, macrophages cultured ex vivo have been shown to bind and internalize VWF, and this interaction is generally enhanced under the application of shear stress or in the presence of ristocetin [13,14].

The first identified macrophage clearance receptor for FVIII was LRP1 (low-density lipoprotein receptor-related protein-1) [15,16]. Clearance studies featuring preinfusion of RAP (receptor-associated protein), an LRP1 ligand antagonist, and conditional LRP1 KO mice have confirmed the function of LRP1 in regulating the clearance of the VWF-FVIII complex in vivo [15,17,18]. Additional members of the low-density lipoprotein recepetor (LDLR) superfamily of receptors, also predominantly expressed on macrophages, have been implicated in VWF-FVIII clearance, although the evidence for their involvement is incomplete [19]. Studies have also identified the scavenger receptor SR-A1, and the lectin receptors MGL (macrophage galactose-like lectin), and Siglec-5 as macrophage receptors that may contribute to the clearance of VWF-FVIII [20–22].

There is more limited evidence that hepatocytes play a role in the clearance of aged VWF-FVIII. While human plasma-derived VWF infused into VWF KO mice associates weakly with hepatocytes immediately post-infusion, endogenous VWF has been shown to localize in vivo with murine hepatocytes [23,24]. Reports have also suggested that recombinant FVIII can also be internalized by hepatocytes in the absence of VWF [25,26]. Hepatocyte clearance of VWF-FVIII is hypothesized to be regulated by the asialoglycoprotein receptor (ASGPR), which binds galactose residues on VWF-FVIII glycoproteins [23,27]. As the N- and O-linked glycans of VWF are heavily sialylated [28,29], this interaction may require glycosidase remodeling of the VWF N-linked glycan structures through exposure to bacteria-derived neuraminidase during septic infection or the aging process of circulating plasma proteins [23,30].

3 |. SINUSOIDAL ENDOTHELIAL CELL CLEARANCE OF VWF-FVIII

The hypothesis that macrophage-mediated clearance pathways represent the dominant mechanism by which VWF-FVIII is removed from the plasma was challenged in 2010 with the publication of a GWAS meta-analysis by the CHARGE consortium [9]. Here, approximately 2.6 million single nucleotide variants (SNVs) were correlated with plasma VWF antigen or FVIII activity levels in a population of >23 000 individuals of European ancestry and confirmed in a replication cohort of 7600 participants. Plasma VWF levels were significantly associated with 8 loci, including VWF, ABO, STX2, STXBP5, CLEC4M, STAB2, SCARA5, and TC2N, and accounted for 12.8% of the variability in plasma VWF levels. Plasma FVIII activity was associated with SNVs at 6 loci, including VWF, ABO, STXBP5, STAB2, and SCARA5. A subsequent multiethnic association study involving an additional 46,000 normal participants of European, African, East Asian, and Hispanic ancestry provided additional confirmation of these associations [8]. Aside from the large influence of ABO, the majority of these variants had a more modest effect on plasma VWF and FVIII levels (3%–6%).

The relationship between plasma VWF and FVIII levels and variants in the VWF gene and ABO blood group locus has been previously described and serves as an important confirmation of the study’s validity [31–33]. Syntaxin-2 (STX2) and syntaxin binding protein 5 (STXBP5) are SNAP receptor (SNARE) family members expressed by endothelial cells and platelets and most likely regulate VWF secretory pathways, while the function of TC2N has yet to be described [34,35]. Importantly, CLEC4M, stabilin-2, and SCARA5 are all cell surface receptors expressed by sinusoidal endothelial cells in the liver, spleen, and/or lymphatics, suggesting that endocytic endothelial cells contribute significantly to the clearance of VWF-FVIII.

Sinusoidal endothelial cells within the liver and spleen are highly specialized and phenotypically distinct from the vascular endothelium, playing an essential role in the removal of aged plasma proteins and necrotic or apoptotic cellular materials from the blood [36]. In the liver, blood enters from the hepatic artery and the portal vein and mixes in the lumen of low-pressure vascular channels termed sinusoids (Figure 1A). Liver sinusoids feature a discontinuous lining composed of sinusoidal endothelial cells (or liver sinusoidal endothelial cells [LSECs]) and liver-resident macrophages or Kupffer cells. LSECs filter the passage of solute from the blood in the lumen to the Space of Disse and adjacent hepatocytes. LSECs contain numerous fenestrae, transcellular channels 50 to 150 nm in diameter, organized into groups known as sieve plates [37]. Sinusoidal endothelial cells have been reported to express numerous endocytic receptors, including CLEC4M and stabilin-2, and are reported to have the highest endocytic capacity of any cell in the body [36,37]. In the human spleen, littoral cells (LCs) are a specialized form of splenic vascular endothelial cells that line the sinuses and comprise 30% of the red pulp [38]. Splenic LCs have strong endocytic/phagocytic activities and express numerous endocytic receptors, including SCARA5, stabilin-1, stabilin-2, and the mannose receptor [38]. Functionally, splenic LCs are primarily thought to recognize and filter foreign antigens and senescent erythrocytes.

FIGURE 1.

Clearance of von Willebrand factor (VWF)–FVIII by sinusoidal endothelial cells (A) Clearance of VWF–factor VIII by endocytic cells in the liver. Blood enters the liver from the hepatic artery and portal veins and mixes in the lumen of the liver sinusoids. Liver sinusoids feature a discontinuous lining of sinusoidal endothelial cells and Kupffer cells or liver-resident macrophages. Liver sinusoidal endothelial cells are highly fenestrated, and the transcellular pores allow for the passage of solute from the lumen into the Space of Disse where it may interact with hepatocytes. (B) Immunohistochemical analysis of infused VWF in the murine liver and spleen. Human plasma-derived VWF was administered to VWF knockout mice via tail vein injection. Animals were euthanized 30 minutes post-infusion and perfused with phosphate-buffered saline and then formalin. Organs were collected, formalin-fixed, and paraffin-embedded. 3,3′-Diaminobenzidine peroxidase immunohistochemistry of VWF was performed using a polyclonal anti-VWF antibody on a Ventana Discovery Immunostainer and imaged using an Aperio ScanScope SC slide scanner as described [42]. In the liver, VWF staining (brown) is strongest in the sinusoidal endothelial cells and Kupffer cells and weaker in hepatocytes. In the spleen, VWF staining (brown) is strongest in the sinusoidal endothelial cells and other cells within the marginal zone (MZ). For both images, the scale bar represents 100 μm. RP, red pulp; WP, white pulp.

The ability of liver and splenic sinusoidal endothelial cells to regulate the clearance of VWF-FVIII has been demonstrated using a series of in vivo half-life and imaging studies [24]. Pretreatment of mice with low-dose cyclophosphamide, which is cytotoxic to the liver endothelium but not to Kupffer cells [39], significantly increased the half-life of human plasma-derived VWF by 64% (p < .0001) [24]. Moreover, immunohistochemical analysis of livers from VWF KO mice infused with either human plasma-derived VWF or recombinant human FVIII (VWF-free) has demonstrated colocalization of infused material with LSEC markers (Figure 1B) [24]. Murine LSECs isolated and cultured ex vivo bind and internalize both human and murine VWF and FVIII in a VWF-dependent manner. Interestingly, although murine spleens are asinusoidal [40], human plasma-derived VWF (and presumably the VWF-FVIII complex) can be bound and internalized by endocytic endothelium in the murine spleen [24,41]. Immunohistochemical studies of human liver and splenic tissues have provided further confirmation regarding the colocalization of VWF with markers of both liver and splenic sinusoidal endothelium and the CHARGE-identified receptors CLEC4M, stabilin-2, and SCARA5 [24,42].

4 |. CLEC4M

4.1 |. Receptor structure and function

CLEC4M (L-SIGN, DC-SIGNR, CD299) is a type II integral membrane protein with homology to the dendritic cell scavenger receptor DC-SIGN (CD-209). CLEC4M is expressed predominantly on the sinusoidal endothelium of the liver as well as the lymphatic endothelium [43,44]. CLEC4M is a calcium-dependent C-type lectin-containing an extracellular carbohydrate recognition domain that mediates its ligand binding activity (Figure 2) [44]. The carbohydrate recognition domain is supported by a polymorphic neck region that promotes extracellular tetramerization of CLEC4M and includes a variable number of tandem repeats (VNTR) polymorphism, featuring between 3 and 9 tandem repeats of 23 amino acids in length [43–45]. The N-terminus of CLEC4M consists of a transmembrane domain and an intracellular region that contains a triacidic cluster and dileucine internalization motifs.

FIGURE 2.

Structure of sinusoidal endothelial receptors C-type lectin domain family 4 member M (CLEC4M), scavenger receptor class A member 5 (SCARA5), and stabilin-2. CLEC4M is composed of an extracellular carbohydrate recognition domain that mediates its ligand binding activity; the carbohydrate recognition domain (CRD) is supported by a variable number of tandem repeats (VNTR) polymorphic neck region that promotes extracellular tetramerization. SCARA5 forms a trimeric complex at the cell surface. Similar to other Class A scavenger receptors, its extracellular region consists of a scavenger receptor cysteine-rich (SRCR) domain, a collagenous domain, and a spacer region. The extracellular portion of Stabilin-2 contains 4 regions of epidermal growth factor (EGF)–like tandem repetitive elements, 7 fasciclin (FAS-1) domains, and an X-link (extracellular link) carbohydrate-binding region. TMD, transmembrane domain.

CLEC4M is capable of binding to both the VWF-FVIII complex through direct interactions with VWF as well as directly to VWF-free FVIII [26,46]. Previous studies have suggested that high-mannose-type glycans represent the primary ligand for CLEC4M [47]. While FVIII contains 2 well-recognized N-linked high mannose glycans, the mannose content of the N-linked glycome of VWF is composed predominantly of complex-type mannose-containing glycans with only trace evidence of high mannose glycans [28,48]. The interaction between CLEC4M and VWF-FVIII is calcium-dependent and can be attenuated by the mannose polymer mannan and by endoglycosidase-mediated removal of the N-linked glycans on both VWF and FVIII, suggesting that CLEC4M can bind to both high mannose as well as mannose-containing glycans of complex type [26,46].

Previously, CLEC4M was thought to function as an adhesive pattern recognition receptor of pathogens and was not recognized to have endocytic activity [43,45]. In a series of human embryonic kidney (HEK) 293 heterologous cell expression studies, the ability of CLEC4M to endocytose both VWF and FVIII was observed [26,46]. CLEC4M localizes within lipid rafts at the cell surface, potentially creating high-avidity binding sites for its glycoprotein ligands. Upon engagement with CLEC4M, VWF and FVIII are immediately internalized by CLEC4M through clathrin-coated pits and transported first to early and then late endosomes. In the HEK 293 heterologous system, catabolism of FVIII but not VWF by lysosomes could be demonstrated, which may be an artifact of the cell system.

4.2 |. Animal models

Current understanding of the in vivo ability of CLEC4M to regulate the clearance of VWF-FVIII is limited by the absence of an orthologous receptor in mice and, thus, an appropriate murine KO model. Although mice express a series of SIGN receptors homologous to human CLEC4M and DC-SIGN [49], SIGNR1, which is expressed by LSECs and splenic marginal zone macrophages and binds to both mannosylated glycoproteins, does not influence the clearance of VWF-FVIII in vivo [24,50,51]. Importantly, however, transgenic expression of CLEC4M by murine hepatocytes via hydrodynamic injection results in reduced plasma levels of VWF (~50%) and FVIII (~65%) compared with controls [26,46].

4.3 |. Genetic associations in normal individuals

In the CHARGE GWAS, the intronic CLEC4M variant rs868875 was associated with plasma VWF but not FVIII levels, while the non-synonymous variant rs2277998 (p.Asp291Asn) was significantly associated with VWF but not FVIII in the multiethnic follow-up investigation (Table) [8,9]. Both the rs868875 and rs2277998 variants are in linkage disequilibrium with the CLEC4M neck region VNTR variant [46,52], which is known to strongly influence the ligand binding capabilities of CLEC4M, potentially through modifying the spatial orientation of the carbohydrate recognition domain with respect to the plasma membrane [47,53]. The VNTR variation may also influence the ability of CLEC4M to self-assemble at the cell surface, thus modifying ligand binding activity [50]. It therefore follows that the VNTR variation is likely responsible for the association between the CLEC4M gene and plasma VWF levels. Concordantly, in vitro studies have demonstrated that CLEC4M 4 and 9 VNTR variants have reduced VWF-binding abilities (62% and 45%, respectively) compared with 6- and 7-VNTR variant proteins [46].

TABLE.

Association of variants in CLEC4M, STAB2, and SCARA5 with plasma levels of von Willebrand factor-factor VIII complex, thrombosis, and factor VIII pharmacokinetics.

Gene	Variant	Study population(s)	Key observations	Reference
CLEC4M	rs868875	>23 000 normal Europeans	4% decrease in VWF:Ag (p = 1.3 × 109)	[9]
		364 patients with Dutch type 1/low VWF	4.3% decrease in VWF:Ag, 5.7% decrease in VWF:Act	[72]
		43 Canadian pediatric subjects with severe HA	39.5 mL/h (p = .039) increase in FVIII clearance	[82]
		43 adolescent/adult Spanish subjects with moderate/severe HA	0.33 mL/h/kg (p = .001) increase in FVIII clearance, 1.1 hours (p = 2.90 × 10−5) decrease in half-life	[83]
		801 French Caucasian subjects with VTE	Decreased association with VTE; OR, 0.81 (p = .0025)	[90]
	VNTR	318 Canadian patients with type 1/low VWF	VNTR 6 allele associated with unaffected status (z score =−2.59; p = .0096) and elevated VWF:RCo (z score = −2.26; p = .029) by FBAT	[46]
		43 Canadian pediatric subjects with severe HA	Associated with clearance (p = .003), k (p = .0004), half-life (p = .016)	[82]
	rs2277998	>46 000 normal multiethnic subjects	2.2% decrease in VWF:Ag (p = 6.5E−16), 1% decrease in FVIII:C (p = 4.4E−04)	[8]
STAB2	rs141041254	~25 000 normal multiethnic individuals	26.8% increase in FVIII:C (p = 2.4 × 10−7) and 33.7% increase in VWF:Ag (p = 2.1 × 10−8)	[62]
	rs12229292	>23 000 normal European individuals	3.1% increase in FVIII:C (p = 7.2 × 10−9)	[9]
		187 Canadian and American patients with type 1/low VWF	3.8% increase in VWF:Ag (p = .014)	[24]
		43 Canadian pediatric subjects with severe HA	Heterozygous association with FVIII clearance (p = .016) and AUC (p = .049)	[82]
	rs4981022	>23 000 normal European individuals	3.6% decrease in VWF:Ag (7.3×10−10)	[9]
		187 Canadian and American patients with type 1/low VWF	2.8% decrease in VWF:Ag (p = .041)	[24]
		>46 000 normal multiethnic individuals	3.5% increase in VWF:Ag (p = 6.6 × 10−41), 2.5% increase in FVIII:C (p = 3.0 × 10−20)	[8]
		43 Canadian pediatric subjects with severe HA	1.941 h (p = .007) decrease in FVIII half-life	[82]
SCARA5	rs2726953	>23 000 normal Europeans	4.5% (p = 1.3 × 10−16) increase in VWF:Ag	[9]
		>30 000 multiethnic VTE cases	Increased association with VTE; OR, 1.05 (p = 2.6 × 10−5)	[92]
	rs9644133	>23 000 normal Europeans	4.1% (p = 4.4 × 10−15) decrease in FVIII:C	[9]
		2440 German females	9.0% (p = .0068) decrease in FVIII:C	[68]
	rs4276643	>46 000 normal multiethnic subjects	2.3% (p = 1.3 × 10−19) decrease FVIII:C, 2.9% (p = 8.8 × 10−28) decrease VWF:Ag	[8]

Act, activity; Ag, antigen; AUC, area under the curve; FBAT, family-based association testing; FVIII, factor VIII; FVIII:C, FVIII activity; HA, hemophilia A; OR, odds ratio; RCo, ristocetin cofactor; VNTR, variable number tandem repeats; VTE, venous thromboembolism; VWF, von Willebrand factor.

5 |. STABILIN-2

5.1 |. Receptor structure and function

Stabilin-2 (FEEL-2, HARE) is a large type I integral membrane protein and class H scavenger receptor that is expressed on the sinusoidal endothelium of the liver, spleen, lymph nodes, and bone marrow [54]. Stabilin-2 comprises 4 regions containing EGF-like tandem repetitive elements, 7 fasciclin domains, and an X-link hyaluronic acid binding domain (Figure 2) [55]. Stabilin-2 shares a 55% amino acid identity with its homolog stabilin-1, but despite structural similarities, the 2 molecules have different ligand binding profiles. In addition to hyaluronic acid, stabilin-2 has been previously reported to bind and internalize acetylated low-density lipoprotein, advanced glycation end products, phosphatidylserine-positive erythrocytes, and glycosaminoglycans, such as chondroitin sulfate, dextran sulfate, and heparan sulfate [56,57]. Stabilin-2 has also been shown to facilitate the adhesion of lymphocytes to LSECs through interactions with the αMβ2 integrin [58]. Upon binding, stabilin-2 mediates the internalization of its ligands through clathrin-coated endocytic vesicles [59]. Engagement of stabilin-2 by a subset of its ligands has been reported to initiate signaling via extracellular signal–regulated kinases 1 and 2 (ERK1/2), resulting in NF-κB activation [56].

In vitro and in vivo animal models have confirmed that stabilin-2 functions as an endocytic and clearance receptor for human VWF-FVIII [24]. Heterologous expression of human or murine stabilin-2 by HEK 293 cells has demonstrated that stabilin-2 is capable of binding and internalizing human VWF and the human VWF-FVIII complex. Interestingly, stabilin-2 does not significantly interact with human FVIII in the absence of VWF or with recombinant murine VWF, suggesting that stabilin-2 regulates the clearance of human VWF only. While the binding sites between stabilin-2 and VWF remain uncharacterized, the removal of N-linked glycans on VWF enhances the ability of stabilin-2 to endocytose VWF, suggesting that stabilin-2 interacts with an amino acid sequence within the VWF molecule rather than via a lectin-mediated interaction.

5.2 |. Animal models

Stabilin-2 deficient mice have been previously reported as having no overt phenotype, although they display elevated levels of circulating hyaluronic acid when compared with controls [60]. Interestingly, stabilin-2 KO mice do not have significantly different levels of VWF:Ag when compared with wild-type C57Bl/6 mice or littermate controls, and no difference in the half-life of murine plasma-derived or recombinant murine VWF is observed in VWF/STAB2 double knockout mice as compared with VWF KO controls, an observation that is consistent with the in vitro cell binding data [24,61]. However, the half-life of human recombinant (72%, p = .0085) and plasma-derived VWF (112%, p < .0001) is significantly prolonged in VWF/STAB2 double knockout mice as compared with VWF KO controls, confirming the role of stabilin-2 as an in vivo clearance receptor for human VWF.

5.3 |. Genetic associations in normal individuals

Human genetic studies have now demonstrated that both common intronic STAB2 variants and rare exonic variants are associated with plasma VWF and FVIII levels in normal individuals. In the CHARGE GWAS meta-analysis, the STAB2 intronic SNVs rs4981022 and rs12229292 were associated with plasma levels of VWF and FVIII, respectively, while the subsequent multiethnic analysis confirmed the association between rs4981022 and VWF-FVIII and identified additional independent signals between STAB2 and plasma levels of VWF or FVIII (Table) [8,9]. Although the mechanistic relationship between these intronic variants and plasma VWF or FVIII levels is not currently understood, they may be in linkage disequilibrium with exonic variants that modify protein function, alter messenger RNA (mRNA) splicing patterns, or modify levels of gene expression through translation efficiency.

An association study between low-frequency variants and plasma VWF or FVIII levels in normal individuals identified the STAB2 p.(Glu2377Lys) variant as being associated with increased levels of VWF (33.6%) and FVIII (26.8%) [62]. Heterologous cell studies of the stabilin-2 p.Glu2377Lys variant protein demonstrated reduced expression compared with wild-type, suggesting that this variant is associated with increased plasma VWF-FVIII by decreasing stabilin-2 expression [24]. An exome sequencing study conducted in normal individuals further confirmed that certain likely nonbenign variants in STAB2 are associated with increased plasma VWF:Ag [63].

6 |. SCARA5

6.1 |. Receptor structure and function

SCARA5 is a class A scavenger receptor and type II membrane protein. Its extracellular region consists of an N-terminal scavenger receptor cysteine-rich (SRCR) domain, a collagenous domain, and a spacer region, followed by a transmembrane domain and a C-terminal intracellular region (Figure 2) [64]. While several alternative splice variants of SCARA5 have been identified, the functional consequences of these variants are currently unknown. Similar to other class A scavenger receptors, SCARA5 exists as a homotrimer and can bind to and endocytose numerous pathogen-associated molecular patterns and damage-associated molecular patterns, including polyanions, lipopolysaccharide, and microbes, although it does not interact with low-density lipoproteins [64,65]. SCARA5 has also been shown to regulate ferritin trafficking and iron delivery [65].

In vitro and ex vivo studies have demonstrated that SCARA5 can bind and internalize VWF and FVIII in a VWF-dependent manner [42]. This interaction is calcium and magnesium-dependent and can be fully inhibited by antibodies to either the collagenous domain or the SRCR domain of SCARA5, or attenuated by polyanionic ligands. While the SCARA5 collagenous domain is thought to be the major ligand-binding domain of class A scavenger receptors, the calcium-dependent interaction between VWF and SCARA5 suggests the involvement of the SRCR domain in mediating this interaction [66]. The expression of SCARA5 in human tissues is not well characterized, although mRNA and immunohistochemical analysis have detected SCARA5 expression in the red pulp, white pulp, and splenic marginal zones [38,42]. Gene expression profiles of purified splenic endothelial cells have demonstrated SCARA5 mRNA expression by CD31⁺/CD34−CD8α⁺ littoral endothelial cells, and immunohistochemical analysis has confirmed colocalization of VWF and SCARA5 within this cell type.

6.2 |. Animal models

Expression of SCARA5 in murine tissues is limited to fibroblasts and epithelial cells of the spleen, testis, heart, and brain [67]. Plasma VWF levels are not significantly different between wild-type and SCARA5 KO mice, and infused human VWF shows only a modest increase in half-life in the absence of SCARA5 expression (27%; p = .026) compared with that in controls, indicating that differences in SCARA5 expression profile between species may limit the utility of studying the clearance of VWF by SCARA5 in vivo [42].

6.3 |. Genetic associations in normal individuals

In the 2010 CHARGE GWAS, SCARA5 intronic variants rs2726953 and rs9644133 were associated with a 4.5% increase in plasma VWF and a 4.1% decrease in FVIII, respectively (Table) [9]. The association between SCARA5 and plasma VWF and FVIII was confirmed in a multiethnic population, where the rs4276643 variant was associated with decreased VWF and FVIII [8]. An additional analysis of the Gutenberg Health Study identified an association between plasma FVIII:C and the homozygous inheritance of the SCARA5 rs9644133 variant in female but not in male study subjects [68]. To date, little is understood regarding the mechanistic basis by which SCARA5 intronic variants may alter the expression or activity of the receptor or if this effect is sex-specific.

7 |. VWD OR “LOW VWF” PHENOTYPE

The genotype-phenotype relationship of type 1 VWD and “low VWF” conditions involving partial quantitative VWF deficiencies is characterized by incomplete penetrance and variable expressivity [69]. Approximately 35% of patients with partial quantitative VWF deficiency do not have a pathogenic coding sequence VWF gene variant, with a greater proportion of these individuals being of the “low VWF” phenotype [70,71]. Together, this suggests that loci external to the VWF gene can contribute to quantitative VWF deficiency. The influence of the CHARGE receptor variants on partial quantitative deficiencies of VWF has been subsequently evaluated by several research groups.

In a study of 318 Canadian and American patients with type 1 VWD/“low VWF” and 173 unaffected family members, family-based association testing demonstrated that the CLEC4M VNTR 6 allele was associated with unaffected status and elevated VWF ristocetin cofactor [46]. In a study of 106 Swedish type 1 VWD/low VWF cases, the CLEC4M rs2277998 variant and the VNTR 5/7 and 6/7 genotypes had an increased but nonsignificant frequency compared with normal controls [52]. Concordantly, CLEC4M rs868875 was associated with both VWF:Ag and VWF activity in a cohort of patients with type 1 VWD/“low VWF” enrolled in the Willebrand in the Netherlands study by regression analysis [72].

While the Willebrand in the Netherlands study analysis did not detect an association between the STAB2 rs4981022 variant and VWF:Ag, a population of 165 Canadian and American patients with type 1 VWD/“Low VWF” showed an association between both the STAB2 rs4981022 and rs12229292 variants and VWF:Ag [24,72]. In contrast, an association study between rare variants in VWF-modifying genes, including CLEC4M, STAB2, and SCARA5, in 104 Swedish type 1 VWD/low VWF cases did not detect a significant accumulation of rare or low-frequency variants compared with control populations [73]. The variability of these results likely relates in part to the relatively small number of subjects and differences in ethnic representation.

To date, no association between quantitative VWF deficiencies and variants in the SCARA5 gene has been reported [42]. It is unknown at this time whether clearance receptor variants may also influence the penetrance of these conditions or overall bleeding phenotype and require larger studies for investigation. Moreover, while the macrophage-expressed receptors LDLR and SRA-1 have been shown to contribute to the enhanced clearance of VWF in type 1C and type 2B VWD [20,74], the interactions between CLEC4M, stabilin-2, and SCARA5 and VWF gene variants with accelerated clearance are yet to be characterized.

8 |. FVIII PHARMACOKINETICS

The pharmacokinetics (PK) of infused FVIII concentrates vary widely within the hemophilia population, although the individual PK profile of a patient can remain relatively stable throughout their lifetime [75–78]. Genetic loci known to modify FVIII PK include VWF, ABO, TC2N, ASGR2, and LDLR, although they do not completely account for the variability of FVIII PK [79–83].

Several studies have now reported an association between the CLEC4M rs868875 variant and FVIII PK [82–85]. The FVIII half-life is decreased by 1.1 hours and clearance is increased by 0.3 mL/h/kg with each additional copy of the CLEC4M rs868875 allele [83]. Pediatric hemophilia A patients with the slowest rate of FVIII clearance had an increased frequency of the rs868875 variant compared with patients who had the fastest rate of FVIII clearance [82]. Interestingly, although a mechanistic relationship between CLEC4M and ABO has not been described, type O patients with at least one copy of the rs868875 nonreference allele displayed faster FVIII clearance and reduced half-life compared with non-O patients or patients homozygous for the rs868875 reference allele [85]. As the CLEC4M rs868875 variant is in linkage disequilibrium with the VNTR variant, pediatric and adult studies have also shown an association between the VNTR genotype and FVIII clearance [82,84].

Variants in the STAB2 gene have also been associated with FVIII PK in pediatric subjects [82]. The rs4981022 variant was associated with FVIII increased k and decreased half-life and area under the curve, while the rs12229292 variant was associated with slower FVIII clearance. Patients with the longest FVIII half-life demonstrated an increased frequency of the rs12229292 variant compared with the patients with the shortest FVIII half-life. Studies of FVIII PK in adolescent and adult subjects have not detected significant associations between STAB2 variants and FVIII PK, likely owing to the influence of aging on the heritability of VWF:Ag [83,84]. Additionally, these small studies may not be adequately powered to capture the relatively small effect that these common variants have on FVIII PK in adults. To date, no association between the SCARA5 CHARGE SNVs and FVIII PK has been identified [82–84,86]. While common CLEC4M and STAB2 variants appear to be important modifiers of FVIII PK, the influence of rare receptor variants remains to be assessed. Together, these studies demonstrate that FVIII PK is regulated by genetic modifiers of VWF and FVIII clearance and suggest novel approaches for the optimization of individualized FVIII dosing regimens and strategies for the design of novel extended half-life FVIII products.

9 |. THROMBOSIS

Elevated VWF and FVIII plasma levels are associated with an increased risk of venous and arterial thrombosis, and SNVs within the VWF and F8 genes modify the thrombosis risk [86–91]. Thus, it follows that SNVs at external loci that modify VWF or FVIII plasma levels may also influence thromboembolic disease. Several studies have now assessed the influence of the CHARGE receptors in thrombosis populations. In a VTE case-control cohort including 656 nonfatal incident VTE cases and 710 controls who were female subjects (aged 40–89 years) of European ancestry, the candidate VWF clearance receptor variants, including SCARA5 (rs2726953), STAB2 (rs4981022), and CLEC4M (rs868875), did not show an association with VTE [89]. In a single case-control cohort (801 cases vs 1150 controls of French Caucasian ancestry), the CLEC4M rs868875 variant was protective for VTE with an odds ratio (OR) of 0.81 (p = .0025), although this finding was not replicated in an independent cohort [90].

A large GWAS meta-analysis of VTE using Mendelian randomization analyses that included 18 studies totaling 30 234 VTE cases and 172 122 multiancestry controls (European or African American) [92] identified that the SCARA5 SNV rs2726953 (OR = 1.05; p = 2.6 × 10⁻⁵) and an independent intronic signal in NUGGC (OR = 1.07; p = 2.4 × 10⁻⁹), 48 kb downstream of the SCARA5 variant, were associated with VTE risk [92]. A phenome-wide association study identified circulating SCARA5 as a protective mediator in stroke [93]. The study included blood biomarker levels from 5 GWAS (predominantly European ancestry) and found circulating SCARA5 levels associated with a decreased risk of cardioembolic stroke (OR = 0.67; p = .001) and subarachnoid hemorrhage (OR = 0.67; p = .001) [93]. It is unclear from this investigation whether SCARA5 scavenger function is causally involved in stroke protection or if it serves as a surrogate marker of prothrombotic activity.

To discern the influence of rare genomic variants on the risk of VTE, whole exome sequencing of 393 patients with unprovoked VTE and 6114 control patients of European ancestry was performed to identify genes with an excess frequency of damaging variants [63]. Here, variants in STAB2 likely to be loss-of-function pathogenic changes associated with risk of VTE (OR = 3.37; p = 2.70 × 10⁻⁷). Overall, 7.8% of VTE cases and 2.4% of controls had a variant predicted to be damaging to stabilin-2 production or function. These observations were replicated in a larger Swedish population of 3177 subjects and over 26 000 controls, albeit with a lower effect size (OR = 1.31; p = .01) [94]. In vitro analysis confirmed impaired intracellular trafficking and therefore, reduced cell surface expression of 7 missense STAB2 variants in a heterologous cell line [63]. These data suggest that impaired endothelial clearance of VWF (±FVIII) contributes to the risk of thrombosis.

In support of these observations, studies using the inferior vena cava stenosis model of deep vein thrombosis demonstrated that stabilin-2 KO mice developed thrombi approximately 60% larger than wild-type littermate controls [61]. Thrombi from stabilin-2 KO mice were morphologically distinct compared with those from controls, with significantly increased leukocyte and citrullinated histone H3 content. Importantly, as stabilin-2 KO mice do not display elevated plasma VWF levels, the prothrombotic phenotype may be driven by elevated circulating leukocytes and enhanced plasma procoagulant activity. Conversely, in ApoE-KO and Ldlr-KO mouse models of atherosclerosis, antibody inhibition of stabilin-1 or stabilin-2 reduced atherogenesis [95]. As stabilin-2 is a multiligand scavenger receptor, the identification of additional factors that contribute to these observations remains an important subject of future studies.

10 |. VWF-FVIII IMMUNOGENICITY

The development of an immune response to infused FVIII or VWF-FVIII concentrates is a complication of both hemophilia A (~30% of patients) and type 3 VWD (~5% of patients) [96,97]. In addition to regulating clearance, ligand-receptor interactions involving the VWF-FVIII complex influence the immune response to infused factor concentrates. Internalization of VWF-FVIII by endocytic receptors expressed on antigen-presenting cells can result in the proteolysis and presentation of peptides on major histocompatibility complex II and the development of either tolerance or immunity to the protein antigens [96]. However, little is currently known regarding which antigen-presenting cells and receptors are essential for mediating these interactions. Interestingly, evidence in a murine model suggests that stabilin-2 deficiency is associated with a decreased immunogenic response to the infused human VWF-FVIII complex [24]. While the mechanistic basis of this interaction is not currently understood, it is possible that stabilin-2 expressing endothelial cells in the spleen can influence the localization of VWF-FVIII antigen within the marginal zone and B-cell compartments. Additionally, elevated levels of plasma hyaluronic acid associated with stabilin-2 deficiency or downstream signaling via extracellular signal–regulated kinases 1 and 2 (ERK1/2) may influence proinflammatory gene expression and modulate the immune response to these antigens. To date, an association between stabilin-2 and the immune response to VWF-FVIII has not been described in humans.

11 |. CONCLUSION AND FUTURE DIRECTIONS

In the years following the publication of the original CHARGE GWAS, numerous studies have confirmed the association between variants in the CLEC4M, STAB2, and SCARA5 genes and plasma levels of VWF/FVIII, including in multiethnic populations, and described the mechanistic relationship supporting these associations. Although CLEC4M, stabilin-2, and SCARA5 have all been characterized as endocytic receptors for VWF-FVIII, there is much that is still unknown regarding the biological basis of these interactions, including large-scale identification of VWF-FVIII modifying quantitative trait loci within these genes, identification of the specific binding site(s) between the receptors and their VWF or FVIII ligands, and understanding the relative contribution of each receptor to the clearance of VWF-FVIII in human subjects.

The importance of the CHARGE study, however, extends significantly beyond the identification of VWF-FVIII clearance receptors, and generating novel insights into quantitative VWF-FVIII pathologies, including type 1 VWD/low VWF, FVIII PK in hemophilia A, and thrombosis. These observations may ultimately improve molecular diagnostic algorithms for patients with quantitative VWF pathologies or the identification of patients most at risk of developing thrombosis. Finally, these studies may contribute to the improvement of novel extended half-life FVIII replacement products with reduced immunogenic effects or personalized therapies tailored to a patient’s pharmacogenomic profile.