The coding sequence for human vWF was identified in a cDNA library derived from endothelial mRNA by several independent groups in 1985.^3,4,5,6 This cDNA sequencing demonstrated an 8.7-kb mRNA with an open reading frame that encodes a 2,813-amino-acid protein, including a signal peptide of 22 amino acids, a large propolypeptide of 741 amino acids, and a mature vWF molecule containing 2,050 amino acids. FIGURE 13.2 demonstrates the relationship of this cDNA to the vWF protein. Using the current numbering system, the initiating methionine codon is defined at nucleotide number one, and this methionine is also identified as amino acid number one. Analysis of the protein structure suggested areas of internal homology that have been termed A, B, C, and D domains.^7,8 The A repeats have sequence similarity to complement Factor B, type VI collagen, chicken cartilage matrix protein, and the “I-domain”-containing integrin α-subunit (Mac-1, VLA2, LFA, p 150, 95).^9,10 Areas of the C repeat are similar to segments of procollagen and thrombospondin.¹¹ Several functions of vWF have been mapped to specific domains. The D′ and D3 domains are important in FVIII binding, and the D3 domain is also essential for multimerization. The A1 domain contains the site for glycoprotein GPIbα binding. vWF binding to collagen occurs through interaction with the A1 and A3 domains. The C1 domain contains an RGDS sequence that allows vWF to bind glycoprotein GPIIb/IIIa. The site for proteolysis by the vWF-cleaving protease, ADAMTS13, is located within the A2 domain.¹²

After the coding sequence for vWF was identified, the entire human vWF gene has been cloned and much of its structure defined. The gene is about 178 kb long and contains 51 introns.¹³ The localization and sequence of the intron-exon boundaries of these 51 introns have been determined. The gene for vWF has been localized to the short arm of chromosome 12, although a pseudogene, representing approximately the middle third of the vWF gene, has been identified and mapped to chromosome 22.^14,15 The presence of this pseudogene can cause problems in determining the genetic structure of some of the vWD variants, but various strategies have been developed to avoid this problem.¹⁶ This pseudogene has about 98% homology to its corresponding portion of the authentic vWF gene on chromosome 12.¹⁴

Various genetic alterations of the vWF gene have been reported and include deletions, insertions, point mutations, alternative splice junctions, and premature transcription termination sites. Many of the molecular mutations reported are included in exon 28.¹⁷ This exon encodes the region of vWF that interacts with platelet glycoprotein Ib (GPIb) and encodes the portion of the vWF protein containing the important N-terminal multimerization site(s)⁷ and an important site where a metalloprotease cleaves vWF between Y1605 and M1606.¹⁸ Most of the type 2A and 2B mutations have been localized within exon 28.¹⁷ Type 2N mutations are clustered within the D′ and D3 domains of vWF.^{19,20,21,22,23} Although a number of patients with type 1 vWD do not appear to have identifiable mutations in the vWF coding region, mutations that have been identified in type 1 vWD do not appear to cluster in any one domain, but are located throughout the vWF protein.^24,25,26

Within intron 40, there are repetitive elements of variable length that have been termed VNTR (variable number of tandem repeats) or STR (short tandem repeats).^27,28 These repetitive elements can be amplified to define the allelic inheritance for unusual genetic phenotypes. These STRs may also be helpful for intrauterine diagnosis using amniocytes or chorionic villi cells with subsequent gene amplification of segments of the vWF gene using the polymerase chain reaction. Caution in amplifying this region must be exercised because of instability of this locus.²⁹

Significant progress has been made in defining the upstream regulatory elements that control vWF expression. Several consensus sequences for cis-acting elements have been identified in the immediate upstream promoter region and first exon including
two GATA-binding consensus sequences.³⁰ The endothelial cell-specific expression appears to be regulated by a repressor-derepressor mechanism and includes both an NF1-binding site and an Oct-1-binding site.^{30,31,32,33,34} A more complex mode of transcriptional regulation through vascular bed-specific signaling pathways has been recently defined. A 2,200 base pair 5′ flanking sequence together with the first intron and exon direct expression to microvascular endothelial cells of skeletal muscle and heart.³⁵ Recently, cell type-specific regulation of vWF expression by the E4BP4 transcriptional repressor has been identified.³⁶ Within the heart, a cardiomyocyte-dependent signaling pathway through platelet-derived growth factor (PDGF) has been defined.³⁷ Several single-nucleotide polymorphisms at nucleotides -1793, -1234, -1185, and -1051 have been associated with plasma vWF:Ag levels.^38,39 Thus, vWF gene is regulated by both cell-specific elements as well as the environment in which these cells are growing.^31,40

Numerous laboratories have purified normal human vWF, and there is general agreement concerning its structure and function relationships.^{41,42,43,44,45,46,47} It is composed of a series of high molecular weight multimers that range between approximately 600,000 (vWF dimer) and 20 million Da. Using low-concentration agarose gels (0.65% to 1%), these multimers can be resolved, with the lowest band representing the dimer. Using high-concentration agarose gels (2% to 3%), each of these individual multimers can be resolved into several additional bands that are thought to result from subunits that have undergone partial proteolytic cleavage or contain variable amounts of modified carbohydrate.

When the disulfide bonds are reduced, a single predominant band is seen on polyacrylamide gel electrophoresis with an apparent molecular weight estimated to be approximately 220,000, plus some smaller bands caused by proteolytic degradation.⁴⁸ The complete protein sequence has been determined
and is identical to the protein sequence derived from the cDNA sequence with a molecular weight of 225,663 for just the amino acid sequence of the mature vWF.^3,4,5,49 The carbohydrate component is estimated to add approximately 10% to 15% to the molecular mass; thus, the true molecular weight of the monomer is approximately 255,000, and its migration in polyacrylamide gel underestimates its true molecular weight.

The vWF multimers are extraordinarily large; the biggest are considerably larger than some virus particles. Electron microscopy of the vWF molecule after rotary shadowing reveals a flexible filamentous structure with irregularly spaced nodules that have been reported from several groups.^50,51,52 These filaments are approximately 20 to 30 Å wide and as long as 11,500 Å, with an average length of 4,780 Å; for comparison, fibrinogen is 475 Å long. Under native conditions, these long filamentous strands are presumed to be wound together in a compact molecule. During activation, shear stress, or other in vivo modifications, portions of the vWF may be unwound, allowing exposure of multiple interactive sites that may participate in platelet adhesion. Decreasing the size of the multimers by partial reduction of disulfide bonds results in a dramatic decrease in the platelet-agglutinating activity.^50,53,54

This molecular weight-dependent heterogeneity of vWF is important when one compares the results of assays based on immunologic and platelet-agglutinating activities. Although immunoassays recognize large and small multimers, estimates of vWF activity using the ristocetin cofactor (vWF R:Co) assay predominantly measure the function of high molecular weight multimers.^54,55,56,57 In contrast, estimates of vWF activity using botrocetin-induced platelet binding correlate better with the immunoassays but may not represent functional activity in vivo.⁵⁸ Individuals with hereditary conditions such as type 2A
vWD, in which there is a reduction in multimeric size, often have more severe bleeding symptoms than individuals with a similar reduction in protein concentration but normal multimeric size (type 1 vWD); thus, in vivo, multimeric size may correlate with function.⁵⁵

Endothelial cells and platelets store vWF in secretory granules that are termed Weibel-Palade bodies (endothelial cells) and α-granules (megakaryocytes and platelets).^60,64,65 These secretory granules can be induced to release their contents by a variety of agonists such as calcium ionophore A23187, thrombin, adenosine diphosphate (ADP)/epinephrine (platelets), and histamine (endothelial cells).^66,67,68 In vivo, the vasopressin derivative desmopressin acetate (DDAVP) can induce the release of vWF into plasma.⁶⁹ The cell from which it is released is presumably the endothelial cell, as evidenced by a lack of reduction in platelet vWF after DDAVP therapy and a loss of Weibel-Palade body vWF on immunostained tissue sections obtained after DDAVP administration.⁷⁰ Much of our understanding of these endothelial cell events comes from expression studies in model cell lines and is summarized in the following paragraphs.^45,46,71,72

vWF is synthesized as a 2,813-amino acid pre-pro-vWF molecule and contains a 22-amino acid signal peptide, 741-amino acid propeptide, and 2,050-amino acid mature vWF protein (FIGURE 13.3). vWF is subjected to extensive intracellular modifications. In the endoplasmic reticulum (ER), the signal peptide is removed, disulfide bonds are formed, and protein folding occurs. vWF is a cysteine-rich protein, containing 234 cysteine residues. While early studies indicated these cysteines were all involved in disulfide bonds, more recent studies using more sensitive techniques suggest that there may be a number of reactive free sulfhydryls in plasma vWF.^73,74 Although a number of disulfide linkages have been mapped, the majority of disulfide mapping in vWF has not been accomplished.^75,76,77 The process of protein folding and disulfide bonding has to be remarkably complicated given the large number of cysteines in vWF. The pro-vWF forms C-terminal dimers in the ER. This results in a pro-vWF C-terminal-linked dimer, the protomeric species that later forms larger vWF multimers. This dimerization is dependent upon the last 151 amino acids of the mature vWF protein.^78,79 These C-terminal sequences may serve a role in retaining monomers in the ER until they are either dimerized or degraded. The N-terminal portion of vWF is less important for dimerization and the large vWF propeptide is not required for dimerization to occur.^80,81,82

Upon transport through the Golgi compartments, vWF undergoes carbohydrate modification. The mature vWF subunit is heavily glycosylated with 10 O-linked and 12 N-linked oligosaccharides.^49,83,84 This glycosylation accounts for approximately 18% to 19% of the total protein mass. Sulfation of the vWF molecule has also been demonstrated.⁸⁵ In the Golgi, the C-terminal-linked pro-vWF dimers form N-terminal-linked multimers that may exceed 20 million Da in size. The vWF propeptide (vWFpp), a protein initially described in plasma as VW AgII, is also proteolytically cleaved in the trans-Golgi. Processing of pro-vWF into vWFpp and the mature vWF is carried out by a paired amino acid-cleaving enzyme (PACE), or furin, that cleaves the propeptide from the mature vWF.^86,87 The site of propeptide cleavage is targeted by the sequence motif Arg-Xxx-Arg/Lys-Arg at the C-terminal end of the propeptide.⁸⁷ Because patients with hereditary persistence of pro-vWF have normal multimers, cleavage of the propeptide does not appear to be a prerequisite for multimerization or secretion.⁸⁸ The cleavage of propeptide appears to be essentially completed in the Golgi, prior to regulated storage.^80,89,90 Both portions of the molecule, the high molecular weight multimerized vWF and the cleaved propeptide, are stored in the Weibel-Palade body secretory granule.

The assembly of vWF into N-linked multimers is a complex process involving interchain disulfide bonds. The first interchain disulfide bonds are formed during dimerization and involve cysteine residues located in the C-terminus region. Multimerization of C-terminal dimers involves cysteine residues located in the N-terminus region (D′ and D3 domains).^45,91 C-terminal dimerization is independent of multimerization. Although both processes involve the formation of disulfide bonds, C-terminal dimerization of vWF is completed in the ER, while N-terminal multimerization is carried out in the Golgi. The two processes are compartmentalized implying different enzymes or mechanisms regulate these steps. In the ER, neutral pH and the presence of necessary oxidoreductase enzymes such as protein disulfide isomerase promote disulfide bonding.^92,93 The acidic pH and lack of chaperones in the Golgi provide a desolate environment for disulfide bonding. However, ER-to-Golgi transport vesicles are unlikely to be able to accommodate the very large vWF multimers; thus, the formation of the large multimers is delayed until transport to the Golgi is complete. Initial glycosylation of the vWF protein and a low pH in the trans-Golgi network have been shown to be important for normal multimerization of vWF.⁴⁶ The acidic pH in the trans-Golgi network appears to be of particular importance for multimer formation. The assembly of functional vWF multimers from purified vWF subunits can be triggered in vitro under low pH conditions, similar to the pH in the Golgi.⁷³ The large propeptide plays an essential role in the vWF multimerization process. Deletion of either of the D1 or D2 propeptide domains, or the entire propeptide, results in a complete loss of normal vWF multimerization.^82,94 When the propeptide and the “mature” vWF molecule (propeptide-deleted) were expressed as two separate gene constructs, normal multimerization was observed indicating the propeptide can function in trans to facilitate multimerization.⁸² The propeptide may facilitate vWF multimerization through areas of the propeptide that contain vicinal
cysteine motifs similar to disulfide isomerase.⁹⁵ The propeptide was shown to act as an oxidoreductase, thus promoting the multimerization of vWF in the Golgi apparatus.⁹⁶ Additional studies identified two amino acids in the D3 domain, Cys-1099 and Cys-1142, that participate in this process.⁷⁶

Both vWF and vWFpp are stored for regulated secretion in Weibel-Palade bodies (endothelial cells) and alpha-granules (megakaryocytes/platelets). The Weibel-Palade body was first described nearly 50 years ago by Weibel and Palade.⁹⁷ Weibel-Palade bodies are large cigar-shaped granules that are the hallmark of endothelial cells. This secretory organelle is used for the storage of vWF (amongst other proteins) in endothelial cells and can be very rapidly triggered to release its contents.^98,99 Weibel-Palade bodies store components that are important to hemostasis, inflammation, and angiogenesis. vWF is the main component protein within Weibel-Palade bodies and plays an important role in the biogenesis of this organelle as discussed below.

The large vWF propeptide also plays a critical role in the regulated storage of vWF. vWF expression studies utilizing the AtT-20 and HEK293 cell lines have examined the regulated storage of vWF.^{80,94,100,101,102} After the propeptide is furin cleaved in the Golgi, the propeptide and vWF remain noncovalently associated and are trafficked to regulated storage in a propeptide-dependent process. Expression studies have shown that deletion of either of the D domains abolishes the sorting of vWF to storage granules.⁹⁴ When the entire propeptide is deleted (Δpro), regulated storage of the “mature” vWF protein is also abolished. In contrast, the propeptide sorts to endogenous storage granules when it is expressed alone in AtT-20 cells. When the vWF propeptide and mature vWF (propeptide-deleted vWF or Δpro) are expressed together as two separate gene constructs, normal
granular storage of vWF is reestablished.^80,103 The propeptide contains the necessary sequence or conformation for sorting to storage granules and secondarily cotraffics mature vWF multimers through noncovalent association.⁸⁰ The nature of the noncovalent association between propeptide and mature vWF has been further defined. Two amino acids, arginine 416 within the propeptide and threonine 869 in the mature vWF protein are essential for noncovalent association and regulated storage of vWF.¹⁰⁴ The putative sorting signal for the propeptide may been located in the D2 domain. By using a series of propeptide truncation cDNA constructs (FIGURE 13.4), the sorting signal has be localized within amino acids 387 to 545 of the propeptide. The sorting signal within the vWF propeptide has been used to traffick an unrelated, nonsecretory granule protein (C3α) to the regulated storage pathway in AtT-20 cells and endothelial cells.⁹⁰

Although the propeptide plays a critical role in both the multimerization and storage of vWF, the two events are clearly independent. Expression of vWF with disrupted vicinal cysteine motifs demonstrated a lack of vWF multimerization but maintenance of vWF regulated storage.⁹⁵ This was further demonstrated by expression studies utilizing human and canine vWF cDNA constructs. The canine propeptide was able to facilitate multimerization of human propeptide-deleted vWF (Δpro) but could not traffic the human vWF to regulated storage.⁸⁰ In a study utilizing the cross-species storage difference, a large number of human/canine chimeric constructs were expressed and examined for multimerization and regulated storage. No correlation was found between vWF multimerization and granular trafficking.¹⁰⁴ In many cases, normally multimerized vWF was not stored in granules and conversely, several vWF species that did not form multimers were trafficked to storage granules. A number of vWF mutations have been identified in vWD patients that disrupt vWF multimerization but did not affect the regulated storage of vWF.^101,105 There have been few patient mutations reported that maintain normal vWF multimerization but eliminate regulated storage and secretion.^102,106,107

The regulated storage of vWF in endothelial cell Weibel-Palade bodies and platelet alpha-granules is important physiologically. Patients with type 1 vWD are often treated with the vasopressin analog DDAVP (1-desamino-8-D-arginine-vassopressin). Administration of DDAVP results in a rapid increase in plasma vWF levels resulting from release of vWF from Weibel-Palade bodies.^108,109 vWF plays an essential role in the formation of Weibel-Palade bodies in endothelial cells. In the vWF-deficient mouse (e.g., type 3 vWD), no Weibel-Palade bodies can be detected in endothelial cells and the membrane protein P-selectin is routed to lysosomes.¹¹⁰ Similarly, cultured endothelial cells harvested from type 3 vWD dogs demonstrated a lack of Weibel-Palade bodies.¹¹¹ Expression studies using the endothelium from the type 3 vWD dogs have demonstrated that expression of vWF protein is necessary for the formation of Weibel-Palade bodies that contain the membrane protein, P-selectin (FIGURE 13.5).¹¹¹ Both propeptide and mature vWF are necessary for formation of Weibel-Palade bodies: neither the propeptide nor mature vWF (Δpro) were sufficient for granule formation. Expression studies using C-terminal truncations of vWF showed that a vWF construct consisting of vWFpp through the vWF A1 domain stored vWF as tubules that correlated with the elongation of Weibel-Palade bodies.¹¹² More recently, vWF tubule assembly has been demonstrated in vitro using only purified vWFpp and D-D3 domains of vWF, narrowing the required domains for Weibel-Palade body formation to the vWF D1-D3 domains.¹¹³ These studies also indicated that vWF may serve an active role in the recruitment of the leukocyte receptor, P-selectin to the Weibel-Palade body membrane.¹¹⁴ While vWF is believed to be the most abundant protein in the Weibel-Palade body, additional components include P-selectin, the tetraspanin CD63, IL-8, osteoprotegerin, α1,3-fucosyltransferase VI, endothelin-1, eotaxin, and t-PA.^{115,116,117,118,119,120,121,122,123} Although the site of synthesis of FVIII is unclear, FVIII can be stored together in a Weibel-Palade body when expressed with vWF.¹²⁴ Thus, vWF-dependent Weibel-Palade body biogenesis has significant biologic implications.

Although there are many similarities between endothelial cell Weibel-Palade bodies and platelet alpha-granules, there also several differences. While the hallmark characteristic of the endothelial cell Weibel-Palade body is its elongated, cigar shape, alpha-granules are not elongated. While both granules contain vWF and P-selectin, the alpha-granule contains several proteins not found in Weibel-Palade bodies such as fibrinogen, coagulation factor V (FV), platelet factor 4, thrombospondin-1, and fibronectin.¹²⁵ Unlike endothelial cell Weibel-Palade bodies, platelets do not require the synthesis of vWF to form the alpha-granule. In vWF-deficient mice¹¹⁰ or dogs (Haberichter,
unpublished data), P-selectin-containing alpha-granules are still present in platelets and megakaryocytes. Therefore, vWF is required for the biogenesis of the Weibel-Palade body but not the alpha-granule.

On appropriate agonist stimulation, endothelial cells and platelets release both vWF and the propeptide from Weibel-Palade bodies and α-granules, respectively. The cleaved propeptide remains associated with vWF under acidic conditions similar to those observed in the late trans-Golgi network and Weibel-Palade bodies, but dissociates from vWF when exposed to the higher pH found in plasma. The propeptide self-associates as a noncovalent dimer in plasma.¹²⁶ vWF can be secreted both luminally into plasma and abluminally into the subendothelial cell matrix.⁷¹ Although the propeptide is also secreted, its extracellular function has not been delineated.