- Top of page
- Materials and methods
Summary. In patients classified with type 1 and type 3 von Willebrand disease missense mutations resulting in the loss of cysteine residues in the D3-domain (multimerization area) and in the carboxy-terminus (dimerization area) of the von Willebrand factor (VWF) have been identified. We have investigated how these structural changes result in a quantitative VWF deficiency and how they interfere with the dimerization and multimerization processes. The effect of mutations in the multimerization area (C1130F, C1149R) and in the dimerization area (C2671Y, C2739Y, C2754W) of human recombinant VWF were investigated in transient transfection assays in 293T cells. All mutations resulted in reduced secretion of VWF in the medium and in intracellular retention. The amino-terminal mutants C1130F and C1149R showed impaired multimerization by lacking high molecular weight (HMW) multimers, in cotransfection experiments with wild-type (wt) VWF, the multimeric pattern was consistent with the pattern in the heterozygous type 1 patients. The carboxy-terminal mutants C2739Y and C2754W showed strongly reduced to nearly absent secretion of VWF, consistent with type 3 VWD. The multimeric pattern of C2739Y and C2754W is characterized by the absence of HMW multimers, an excess of monomers and intervening odd-numbered multimeric bands, indicating a dimerization defect. The carboxy-terminal mutant C2671Y is different, with mildly reduced secretion, intermediate intracellular retention and a normal multimerization pattern. We conclude that, in accordance with a phenotype of quantitative VWF deficiency, all cysteine mutants show impaired secretion, although the decrease of VWF in vitro appears lower than in the patients, suggesting additional, possibly heightened clearance, mechanisms in vivo.
The von Willebrand factor (VWF) is a high molecular weight multimeric glycoprotein (0.5–10 × 106 Da) with adhesive properties. VWF mediates adhesion of platelets to the vessel wall and platelet–platelet aggregation. VWF forms a non-covalent complex with factor VIII. The unprocessed preproprotein of 2813 amino acids (aa) is targeted to the endoplasmatic reticulum (ER) by the 22-aa long signal peptide. In the ER, the proVWF is glycosylated and the subunits are linked pairwise through covalent disulfide bonds at the carboxy-terminus. The proVWF dimers are further modified when passing through the ER and the Golgi apparatus. Multimers are formed from proVWF dimers by disulfide bonds at the amino-terminal end of the subunits. After formation of multimers, the propeptide is cleaved off, resulting in VWF multimers consisting of an even number of mature VWF subunits (2050 aa) [1,2]. The VWF protein contains a high number of cysteine residues (8.2%), all of which participate in forming intra- or interchain disulfide bonds .
von Willebrand disease (VWD) is the most common inherited bleeding disorder. It is caused by dysfunctional VWF or by a deficiency of VWF. VWD is divided into three groups : type 1 refers to partial VWF deficiency; type 3 is characterized by an almost complete VWF deficiency; and type 2 involves all functional defects of the VWF protein. The molecular basis of the disease has been elucidated for most type 2 VWD variants . The molecular basis of type 1 and type 3 VWD has been difficult to characterize, because mutations are not restricted to a specific region in the VWF gene.
Although type 1 and type 3 VWD are both characterized by a deficiency of VWF, the underlying genetic mechanisms appear to be different. Type 1 VWD has an autosomal dominant inheritance pattern; however, recessive inheritance has also been described . Inheritance of type 3 VWD is autosomal recessive. The heterozygous carriers of type 3 mutations (mainly null alleles) usually have only mild or no bleeding symptoms and roughly 50% reduction of VWF levels. Therefore, these carriers of type 3 VWD are different from most type 1 VWD patients who have VWF levels well below 50%. We previously hypothesized that type 1 VWD could possibly be caused by missense mutations, resulting in a dominant negative defect, i.e. mutant subunits interacting with normal subunits leading to a reduction of more than 50% of VWF levels. Following this hypothesis we previously identified two missense mutations, C1130F and C1149R. These mutations both occur in the D3-domain, which is involved in multimerization . C1149R has been shown to cause intracellular retention of VWF [7,8] (nomenclature of mutations according to reference ).
While mutated cysteines in the multimerization domain cause a dominant negative type 1 defect, we hypothesized that loss of cysteines in the dimerization domain (the 151 carboxy-terminal amino-acid residues ) may also be responsible for quantitative defects. If mutant proVWF subunits are unable to form dimers, it is possible that only wild-type homodimers are formed (in the heterozygous state) and, if only these normal homodimers were transported to the Golgi, a reduction of VWF by 50% would be seen. This would mimic the effect of a null allele. Using this hypothesis, we have previously investigated VWD type 3 patients and identified a C2671Y mutation at the carboxy-terminus . Other C-terminally located cysteine mutations have also been identified in type 3 VWD patients [12–15].
To further investigate the VWF defect due to a loss of cysteines in the multimerization and dimerization domains, we have expressed five different cysteine mutations; the amino-terminal C1130F and C1149R mutations, identified in type 1 VWD and the C-terminal C2671Y, C2739Y and C2754W mutations, identified in type 3 VWD.
- Top of page
- Materials and methods
We have investigated the effect of the loss of cysteine residues, involved in either dimerization or multimerization, on the quantity and quality of the VWF produced. The two mutations in the D3-domain, C1130F and C1149R, were identified in type 1 VWD patients. Their phenotypes are characterized by a dominant inheritance pattern with high penetrance and VWF antigen levels between 10 and 15 IU dL−1. We postulated that the strikingly low levels of VWF:Ag could be explained by the interaction of the mutant subunit with the normal subunit, resulting in retention of normal subunits in the ER and the Golgi, causing a dominant negative defect. We have previously reported that the mutation C1149R decreases the secretion of coexpressed normal VWF and causes intracellular retention and degradation of VWF [7,8]. In this study we performed expression studies of both C1149R and C1130F. Similar results were obtained for both mutations (Figs 1–3): impaired secretion (∼20% of wt) and intracellular retention (∼280% of wt) of VWF. In the 1 : 1 ratio (wt/mutant) cotransfection, mimicking the heterozygous type 1 patients, secretion of VWF was 69% and 63% for C1130F and C1149R, respectively, and both cotransfections demonstrated dose-dependent intracellular retention of VWF. We have demonstrated that the reduced secretion is caused by intracellular retention and not by reduced expression of the mutant construct. Both in the present and in the previous study of C1149R  we did not find the decrease of VWF that is observed in the plasma of heterozygous patients with C1130F or C1149R. The more pronounced decrease of plasma VWF in the patients could be explained by mechanisms not reflected by the 293T cell system, such as an increased clearance of mutant VWF by VWF cleaving protease and/or regulated secretion.
Both C1130F and C1149R VWF show abnormal multimeric patterns in cotransfections (Figs 2 and 3), illustrating a dominant negative effect at the qualitative level. However, the quantitative results support an additive rather than a dominant negative model as the observed amounts of VWF secreted or retained can be predicted by the summation of single construct transfections of wt or mutant constructs alone. Bodo et al.  showed proteosomal degradation of intracellularly retained C1149R in Fur4BHK cells with stable expression of C1149R. This degradation could explain the dominant negative effect of the mutation C1149R observed in patients. Our 293T cell system may not reflect this dominant negative aspect due to the transient nature of the transfection and the collection of medium at a time when production of VWF was still linear. In the linear phase, the contribution of intracellular degradation may be negligible.
The intracellular retention and impaired multimerization of both C1130F and C1149R are probably due to disturbed protein folding. The decrease of secreted VWF C1149R was previously shown not to be due to the effect of the resulting unpaired cysteine 1169, which normally pairs with C1149. It was shown that expression of the double mutant C1149R + C1169S did not restore the defect of C1149R alone . Another argument for a conformational change is the fact that cysteine 1130 and 1149 are likely to form intrachain bonds  and are not directly involved in the intersubunit bonds required for multimerization. Dong et al.  demonstrated that even mutating the three cysteines (C1222, C1225, C1227) involved in the formation of intersubunit disulfide bonds did not influence assembly or secretion of VWF, indicating that the conformation of VWF as determined by intrachain bonds is more important for normal multimerization than the individual interchain bonds. Finally, C1130F and C1149R involve mutations to bulky or charged amino acids, predicting major effects on the conformation.
We have also investigated mutated carboxy-terminal cysteines that may be involved in dimerization. Three mutated cysteine residues (C2671Y, C2739Y and C2754W), all described in type 3 VWD patients (one reported by us  and two by others [12,13]), were studied. Most mutations described in type 3 VWD are null alleles . We hypothesized that these mutated cysteines at the carboxy-terminus could mimic null alleles when the mutated protomer is not able to dimerize, when it is retained and degraded in the ER and when, subsequently, only homodimers from the normal allele are routed to the Golgi and secreted. This would result in a level of about 50% normal VWF (in a heterozygous carrier of such a mutation), which is the same as in carriers of a null allele. In recombinant expression a striking difference in VWF levels and multimer patterns was observed for C2671Y compared with the two other carboxy-terminal mutants. C2671Y showed completely normal dimerization and multimerization (Figs 1 and 4). The apparent lack of influence of C2671Y on dimerization may be explained by its location. C2671 lies within the 151 carboxy-terminal amino acids that have been demonstrated by Voorberg et al.  to be required, and sufficient, for dimerization. C2671 was therefore anticipated to be important for dimerization. However, only the carboxy-terminal 90 amino acids are homologous to the cysteine knot (CK) family of proteins . Therefore, it is possible that only amino acids 2724–2813, excluding C2671, are essential for dimerization. The VWF:Ag level in the patient carrying the C2671Y mutation in compound heterozygosity with a complete VWF gene deletion, corresponding to homozygosity for C2671Y, is only 2%, whereas the expression in medium of 293T cells is 40% of wt. This discrepancy can be explained by increased physiologic proteolysis of VWF C2671Y, that was demonstrated in this patient's plasma , but which is lacking in the medium of the in vitro expression system. Our results suggest that the mutation C2671Y interferes with intracellular routing, has little or no influence on dimerization and is possibly more sensitive to proteolytic degradation in plasma.
The missense mutations C2739Y and C2754W were shown to interfere with dimerization. Excess of mainly intracellular monomers and odd-numbered multimers was found for both mutations (Figs 1, 5 and 6). This indicates amino-terminal pairing of mutant monomers in the Golgi. The pattern with odd-numbered multimers in the heterozygous transfection of C2754W was previously described by Schneppenheim et al. and demonstrated in plasma of a heterozygous carrier . Odd-numbered multimers were not observed in medium of C2739Y cotransfections. The difference in secreted multimers of C2739Y and C2754W may lie in the survival of the mutant protomers in the cell. It is unlikely that a missense mutation affects translation; we therefore assume that the mutant monomer is misfolded, is retained in the cell and cleared at a high rate as has previously been described . The linear increase in the total amount VWF in C2754W cotransfections compared with the fixed low value in C2739Y cotransfections indicate that the C2739Y mutation causes a more profound conformational change than the C2754W. This would explain the different rate of clearance and the impossibility for C2739Y to pass the ER and participate in multimerization in the Golgi, whereas C2754W may pass the ER.
Both C2739Y VWF and C2754W VWF were retained in the cell and secreted at very low levels in medium, 1 and 8%, respectively (Fig. 1), which corresponds to a type 3 phenotype. Further, the cotransfection experiments of mutant and wt in a 1 : 1 ratio showed about 50% VWF secretion for C2739Y and C2754W, which agrees with the phenotype of a heterozygous carrier of type 3 VWD. However, heterozygosity for mutations of some cysteine residues in the CK domain results in type 2A (formerly subtype IID) VWD. Expression studies have shown a similar dimerization defect [13,25] as in the mutants in this study. This apparent phenotypic discrepancy may be explained by differences in the involvement of cysteines in either inter- or intrachain disulfide bonds. All mutations that have been described in association with a type 3 phenotype, C2739Y , C2754W , C2804Y  and C2806R  are involved in intrachain disulfide bonds , whereas the mutations associated with a 2A/IID phenotype involve interchain disulfide bonds: C2771Y , C2771S  and C2773R [13,26]. We hypothesize that the loss of a cysteine involved in an intrachain disulfide bond considerably disturbs the conformation of the VWF subunit, leading to hampered dimerization and rapid degradation in the ER. The amount of abnormal monomer contributing to the total amount of VWF would be small and would not cause a dominant negative effect. This would result in a quantitative defect with little influence on multimeric structure, i.e. a type 3 phenotype in homozygotes and only 50% reduction of VWF in heterozygotes, mimicking the effect of null alleles. On the contrary, the mutations interfering with interchain bonds do not disrupt the conformation of the monomer, but do reduce the covalent dimerization of the subunit . As the conformation of the monomer is not dramatically affected, it is possible that a higher fraction of monomers escape degradation in the ER and is routed to the Golgi. Since the proportion of mutant monomer to normal monomer is higher in this case, the multimerization will be significantly influenced by amino-terminal disulfide bonding of the excess mutant monomers resulting in a type 2A/IID VWD multimer pattern.
In conclusion, we have shown that the cysteine mutations investigated in this study cause quantitative VWF deficiency, although the effect is not as strong as that observed in patients. The discrepancy in VWF levels of C1130F, C1149R and C2671Y observed in vitro vs. in vivo might be explained by physiological factors such as heightened clearance. The loss of cysteines 2739 and 2754 corresponds with the type 3 VWD phenotype and the apparent lack of influence of the C2671Y mutation on dimerization is most likely due to its location just outside the conserved CK domain.