Two novel mutations, Q1053H and C1060R, located in the D3 domain of von Willebrand factor, are responsible for decreased FVIII-binding capacity

Authors


Dr L. Hilbert, PhD, Unité de Développement Préclinique, Laboratoire Français du Fractionnement et des Biotechnologies, 59 rue de Trévise, BP 2006, 59011 Lille cedex, France. E-mail: hilbert@lfb.fr

Abstract

Summary. In type 2N von Willebrand disease (VWD), von Willebrand factor (VWF) is characterized by a markedly decreased affinity for Factor VIII (FVIII), and the mutations responsible are essentially located in the D′ domain of VWF. We report the identification, in seven unrelated French families, of two novel type 2N VWD mutations, Q1053H and C1060R (Gln290His and Cys297Arg in mature VWF sequence), in exon 24 of the VWF gene. These missense mutations have been identified in the heterozygous, homozygous or hemizygous states. Using site-directed mutagenesis and transient expression in COS-7 cells, we showed that both mutations, although located in the D3 domain of VWF, outside the tryptic fragment containing the FVIII domain, dramatically decrease the binding of VWF to FVIII. In contrast, the R924Q substitution, which was identified in a patient who was heterozygous for C1060R, was shown to be a polymorphism.

Von Willebrand factor (VWF) is a multimeric plasma glycoprotein (GP) that is required for normal haemostasis. VWF is synthesized by endothelial cells and megakaryocytes and originates from its precursor prepro-VWF, which is processed in VWF propeptide [amino acid residues (aa) 22–763] and mature VWF (aa 764–2813) (Bonthron et al, 1986). Mature VWF mediates the adhesion of platelets at sites of vascular injury and stabilizes factor VIII (FVIII).

Type 2N von Willebrand disease (VWD) refers to all qualitative variants with a markedly decreased affinity of VWF for FVIII (Sadler, 1994). This recessively inherited phenotype may be misclassified as mild haemophilia A because type 2N VWD patients are also characterized by reduced FVIII levels, normal VWF levels and normal VWF multimeric profiles (Mazurier, 1992).

Most type 2N VWD patients have been found to harbour missense mutations in the VWF D′ domain, encoded by exons 18–20 (Mazurier et al, 2001). These mutations induce substitutions of several specific aa in the N-terminal region of the mature VWF, which contains the FVIII-binding domain present on the tryptic fragment of VWF (aa 764–1035 of prepro-VWF; Foster et al, 1987). Two point mutations modifying aa 879 and 1225 in the VWF D3 domain, mostly encoded by exons 20–27, have been also identified (Jorieux et al, 1998; Allen et al, 2000).

Expression studies reproducing nucleotide (nt) changes in VWF cDNA have been essential to confirm the relationship between the identified molecular abnormality and the patient's phenotype. Such studies have shown that some aa changes, particularly the R854Q mutation, dramatically diminish the capacity of VWF to bind FVIII (FVIII-binding capacity, VWF:FVIIIB) (Cacheris et al, 1991; Kroner et al, 1991), whereas other aa substitutions, such as those affecting T791 or W816, completely abolish this binding (Tuley et al, 1991; Jorieux et al, 1992). In addition, these studies have shown that the T789A (Kroner et al, 1996) and R852Q (Kroner et al, 1991) substitutions do not affect VWF function and are, in fact, polymorphisms, whereas the R782W and H817Q substitutions, which have been characterized on the same allele in a type 2N patient, only slightly decreased the VWF:FVIIIB (Kroner et al, 1996).

We report the identification of two novel mutations in exon 24, Q1053H and C1060R (Gln290His and Cys297Arg in mature VWF sequence), in seven unrelated French families. One patient displayed the Q1053H mutation in the heterozygous state. Another patient exhibited the C1060R mutation in the homozygous state, whereas the other five were compound heterozygotes for the C1060R mutation and either the frequent R854Q mutation (two patients), the as yet unreported R924Q (Arg161Gln in mature VWF) substitution (one patient) or a quantitative defect (two patients).

Using site-directed mutagenesis and transient expression in COS-7 cells, we showed that the R924Q substitution is a polymorphism, whereas the Q1053H and C1060R mutations dramatically decreased VWF:FVIIIB even though they are located in the D3 domain of VWF, outside the tryptic fragment containing the FVIII domain. Therefore, our results confirm the importance not only of the D′ domain but also of some aa of the D3 domain in the VWF interaction with FVIII.

Materials and methods

Laboratory methods.  Blood samples were collected into 13 mmol/l sodium citrate, and tests were performed on citrated platelet-poor plasma samples stored frozen at −20°C. Bleeding times were measured by the Ivy method. FVIII coagulant activity (FVIII:C), VWF antigen (VWF:Ag) level and ristocetin cofactor activity (VWF:RCo) were assayed as described previously (Mazurier et al, 1990). Measurement of VWF capacity to bind FVIII was performed in a solid phase system (Jorieux et al, 2000). The multimeric composition of VWF was analysed after 0·1% sodium dodecyl sulphate (SDS)−1·5% agarose gel electrophoresis, visualization with alkaline phosphatase-conjugated immunopurified anti-VWF polyclonal antibodies (Hilbert et al, 1994) and subsequent scanning of the gels with a densitometer.

Molecular genetic studies.  After informed consent was obtained from the patients, genomic DNA was extracted from peripheral blood leucocytes according to standard procedures. Exons 18–24 of the VWF gene were amplified by polymerase chain reaction (PCR) and sequenced with a Sequenase 2·0 kit using the respective PCR limiting strand or antistrand primers as described previously (Jorieux et al, 2000). Analyses of polymorphism segregation [RsaI in exon 18, MspI in intron 19, HphI and BstEII in exon 28 and variable nuclear tandem repeat (VNTR) in intron 40] were performed in patient D. Patient E was studied by multiplex single-stranded conformation polymorphism (MSSCP) followed by further sequencing of the exons showing an abnormal migration in MSSCP gels.

Plasmid construction.  Construction of the vector containing the R854Q mutation has been reported previously (Jorieux et al, 2000). The same procedure with the Transformer site-directed mutagenesis kit (Clontech) was used to introduce the R924Q, Q1053H and C1060R substitutions in VWF cDNA. The resulting mutated vectors were digested with the two unique restriction enzymes HindIII–NotI (for R924Q) or NotI–AflII (for Q1053H and C1060R). The fragments containing the mutated aa were subcloned into plasmid pSVvWFA (Hilbert et al, 1995) that had been digested previously with the same two enzymes.

Transient expression of the mutated recombinant VWF (rVWF).  COS-7 cells (106 cells in 80-cm2 flasks) were transfected with 25 µg of normal or mutated vector by the diethyl aminoethyl (DEAE) dextran method as described previously (Hilbert et al, 1994). After 72 h expression, conditioned media were collected in the presence of phenylmethylsulphonyl fluoride (PMSF; 1 mmol/l) and benzamidine (10 mmol/l). The concentration of VWF:Ag in conditioned media was measured by enzyme-linked immunosorbent assay (ELISA) using rabbit anti-VWF polyclonal antibodies (Mazurier et al, 1980).

Recognition of rVWF by various monoclonal antibodies (mAbs).  mAbs 9311A2, 32B12 and 14A12 were generated by the Centre de Transfusion Sanguine, Lille, while mAb 418 was generated by the Institut National de la Santé et de la Recherche Médicale (INSERM) U143. mAb 418 reacts with the unreduced but not reduced SPIII-T4 fragment, inhibits VWF:FVIIIB, and its epitope has been mapped to aa 765–816 of prepro-VWF (Piétu et al, 1994). mAb 32B12 is also a potent inhibitor of VWF:FVIIIB, and its epitope has been localized on the aa sequence 814–823 of prepro-VWF, whereas mAb 14A12, the epitope of which has been localized between aa 904 and 983 of prepro-VWF, has no significant effect on VWF:FVIIIB (Jorieux et al, 1994). mAb 9311A2 recognizes the SpII fragment of VWF (aa 2128–2813 of prepro-VWF), reacts with all VWF multimeric forms and does not inhibit any VWF function. The ability of these mAbs to recognize wild-type (WT) and mutated rVWFs was measured by ELISA using a mAb as a capture antibody and peroxidase-conjugated rabbit anti-VWF polyclonal antibodies as detecting antibodies (Gu et al, 1997). Absorbance at 492 nm was converted to mU/ml VWF using the standard curve generated for each captured mAb using dilutions of a pool of normal plasmas.

Results

Patients

The phenotypic data for the propositus of each family are summarized in Table I. Bleeding symptoms were generally mild or moderate, but some patients required replacement therapy. Bleeding time was normal except for patients D and E. The propositi displayed a FVIII deficiency, but showed normal or subnormal levels of VWF:Ag and VWF:RCo (Table I). VWF:FVIIIB was dramatically decreased in patients B–F (Table I). In patients A and G, VWF:FVIIIB was moderately decreased and similar to that of a patient who was heterozygous for a single type 2N mutation (Table I). In patient A, VWF displayed a normal multimeric pattern with all the multimeric forms and even some suprahigh-molecular-weight (HMW) multimers, whereas in the other patients, VWF displayed a multimeric pattern with a more or less pronounced decrease in the highest molecular weight multimers, particularly those above 10-mer (Fig 1).

Table I.  Biological data and DNA analyses of patients.
PropositusSexBlood
group
Ivy's
bleeding
time (min)
Platelets
(× 109/l)
FVIII:Ag
(IU/dl)
FVIII:C
(one stage)
(IU/dl)
FVIII:C
(chromogenic)
(IU/dl)
VWF:Ag
(IU/dl)
VWF:RCo
(IU/dl)
VWF:FVIIIBDNA
abnormalities
  1. The number of determinations is indicated in parentheses.

  2. F and M indicate a female or male patient respectively.

AFO+5·25 ± 1 (2)185 ± 50 (2)25·5 ± 7 (2)28 ± 2·8 (2)72 ± 1·4 (2)69·5 ± 13·4 (2)Moderately decreasedQ1053H
BMO+6·5 (1)323 (1)16·3 ± 6·8 (3)12·5 ± 5·2 (4)9·5 ± 3·8 (4)65·8 ± 15·8 (6)51·5 ± 14·9 (6)Dramatically decreasedC1060R, R854Q
CFA+7 ± 0·5 (2)275 ± 35 (3)12 (1)10 (1)13·1 ± 7·9 (3)81·6 ± 3·8 (3)83 ± 13·8 (3)Dramatically decreasedC1060R, R854Q
DMB+10 (1)
> 15 (3)
153 ± 25 (2)13·5 (1)8·6 ± 1·8 (4)9·3 ± 5·3 (6)129·8 ± 32·6 (7)104·4 ± 10·1 (5)Dramatically decreasedC1060R, C1060R
EFA–> 20 (1)7·25 ± 1 (2)4·5 ± 0·7 (2)< 2 (1)
10 (3)
43·3 ± 9 (3)49 ± 5·2 (3)Dramatically decreasedC1060R, Y357X
FMO+7 ± 1 (2)246 ± 20 (2)5·5 (1)3 (1)7 ± 4·3 (6)38 ± 2·3 (5)24·7 ± 6·6 (4)Dramatically decreasedC1060R, deletion
GMA+9·5 (1)36·5 ± 5·6 (3)40 ± 10 (3)38 ± 13·8 (3)59·3 ± 3 (3)51 ± 1·4 (2)Moderately decreasedC1060R, R924Q
Normal individuals  < 10150–40050–15050–15050–15050–15050–150Normal 
Figure 1.

Multimeric pattern of normal and patient's plasma VWF. Plasma samples were electrophoresed in SDS−1·5% agarose gel, and VWF was visualized with alkaline phosphatase-conjugated anti-VWF polyclonal antibodies. The two outside lanes contain the pool of normal plasmas (NP); lanes A to G contain the plasma of families A to G propositi.

DNA sequencing

To identify the VWF gene abnormality responsible for the VWF:FVIIIB defect characterized in these patients, PCR-amplified exons 18–24, coding for the N-terminal region of mature VWF up to aa 311 (i.e. aa 761–1074 of prepro-VWF), were sequenced. The molecular abnormalities identified are reported in Table I. A 3159G → T transversion in exon 24, predicting the Q1053H substitution (Gln290His in mature VWF), was found in patient A. Sequence analysis of exon 20 in patients B and C revealed the 2561G → A transition on one allele, which induces the already identified and frequent R854Q mutation (Arg91Gln in mature VWF). Moreover, a 3178T → C transition in exon 24, predicting the C1060R substitution (Cys297Arg in mature VWF), was identified, also in the heterozygous state, in these two patients. The same nt substitution inducing C1060R abnormality was found in the homozygous or hemizygous state in patients D, E and F, whereas it was heterozygous in patient G. Patient G was also heterozygous for a 2771G → A transition in exon 21, changing the arginine 924 into glutamine (Arg161Gln in mature VWF). In patient E, MSSCP analyses and further sequencing enabled the identification of a 1071C → A transversion in exon 9, inducing a stop codon at position 357 of prepro-VWF. Analysis of polymorphism segregation indicated that patient D was homozygous for the C1060R substitution, whereas patient F displayed a large deletion of the VWF gene of at least exons 18–40.

Characteristics of mutated rVWF

To determine whether the identified nt changes, predicting the R924Q, Q1053H and C1060R candidate mutations, alter the interaction of VWF with FVIII, they were introduced into full-length cDNA, and the corresponding rVWFs, expressed by transient transfection into COS-7 cells, were analysed. The characteristics of these rVWFs were compared with those of WT rVWF (WTrVWF) and mutated Q854rVWF harbouring the most frequent type 2N VWD mutation. Normal and mutated vectors directed the expression of equivalent levels of rVWF (≈ 10 IU/dl/72 h) regardless of the nt change. Like Q854rVWF, H1053rVWF displayed markedly decreased (≈ 10%) VWF:FVIIIB compared with that of WTrVWF (Fig 2A). Whereas Q924rVWF displayed normal VWF:FVIIIB, R1060rVWF was completely unable to bind to rFVIII (Fig 2A). The multimeric structure of secreted Q854, Q924 and H1053rVWF was similar to that of WTrVWF (Fig 2B). In contrast, R1060rVWF displayed an abnormal multimeric pattern characterized by a decrease in multimers ≥ 10-mer (6·1%vs 39·4% for WTrVWF) and a parallel increase in protomer level (11·1%vs 5·8% for WTrVWF) (Fig 2B).

Figure 2.

Characteristics of rVWFs. COS-7 cells were transfected with the appropriate plasmid DNA, then placed in medium for 24 h after transfection. Conditioned media were harvested after an additional 72 h. (A) Binding of recombinant FVIII to wild-type and mutated rVWFs. Serial dilutions of plasma or conditioned medium samples were incubated into microtitre plates coated with anti-VWF polyclonal antibodies. A constant amount of rFVIII (0·1 U/ml) was then added, and the activity of FVIII bound to immobilized VWF was determined using a chromogenic assay. The amounts of VWF captured by immobilized antibodies were measured by ELISA using a peroxidase-conjugated mAb. Pool of normal plasmas (filled circles); mix (v/v) of normal plasma and plasma from a type 2N VWD patient with the R816W mutation (open circles); WTrVWF (grey circles); Q854rVWF (filled triangles; Q952rVWF (open triangles); H1053rVWF (filled squares); R1060rVWF (open squares). (B) Multimeric pattern of wild-type and mutated rVWFs. Lane 1, WTrVWF; lane 2, Q854rVWF; lane 3, Q952rVWF; lane 4, H1053rVWF; lane 5, R1060rVWF.

The reactivity of four anti-VWF mAbs was tested towards the various rVWFs. The results, presented in Fig 3, show that each of the mutants was captured equally well by all four mAbs except R1060rVWF, which was bound three times more by mAb 14A12.

Figure 3.

Recognition of wild-type and mutated rVWFs by mAbs. Conditioned media containing 10 mU/ml VWF:Ag were incubated for 1 h at 37°C into microtitre plates coated with different mAbs: 9311A2; 418; 32B12; 14A12. After washing, the bound rVWF was detected with peroxidase-conjugated anti-VWF polyclonal antibodies. Similar results were observed when two independent transfection media were tested in two separate experiments. The previously identified epitope of each mAb is given in parentheses.

Discussion

We report here the identification and expression of two novel mutations (Q1053H and C1060R) in the D3 domain of VWF inducing a decrease in VWF capacity to bind to FVIII.

In the first patient studied (patient A), with moderately decreased VWF:FVIIIB and only one Q1053H substitution in the heterozygous state, the decreased FVIII level and bleeding tendency could not be explained by the VWF gene defect. Further molecular work enabled the identification of an, as yet, unreported C/T transition in the FVIII gene promoter (at position −217 ahead of exon 1), in the heterozygous state, in this patient and her father (J.-M. Lavergne, INSERM U143, personal communication). Thus, patient A is likely to be a haemophilia A carrier who is also heterozygous for a type 2N VWD mutation.

On the other hand, the C1060R substitution was identified in either the homozygous or the heterozygous state in five unrelated type 2N VWD patients exhibiting dramatically decreased VWF:FVIIIB. Among the four patients heterozygous for C1060R, two (patients B and C) harboured the frequent R854Q type 2N VWD mutation, whereas the other two (patients E and F) displayed a type 3 VWD mutation (stop codon and deletion respectively) on their second allele. Both patients E and F displayed relatively low VWF levels, as generally observed in compound heterozygote type 2N/type 3 VWD patients, but only patient E had a prolonged bleeding time (recently confirmed by a prolonged closure time measured with a platelet function analyser). Prolonged bleeding time was also observed in patient D who was homozygous for the C1060R mutation and had normal VWF levels. This patient's defect may be attributed to his moderate thrombopenia (110–152 × 109/l platelets) associated with chronic hepatitis C and haemochromatosis. The C1060R abnormality was identified in addition to an, as yet, unreported R924Q substitution in patient G who displayed moderately decreased VWF:FVIIIB, VWF and FVIII levels.

We have transiently expressed the R924Q, Q1053H and C1060R candidate mutations in COS-7 cells and analysed the characteristics of the corresponding recombinant proteins to evaluate the relationship between the aa change and the defective VWF:FVIIIB. We showed that the R924Q substitution is a polymorphism because VWF:FVIIIB of the corresponding mutated rVWF is similar to that of WTrVWF. These results are in agreement with the phenotype and genotype reported in patient G: the C1060R mutation in the heterozygous state is responsible for moderately decreased VWF:FVIIIB of patient plasma VWF and does not induce a significant decrease in the FVIIII:C/VWF:Ag ratio. In contrast, both Q1053H and C1060R substitutions dramatically reduced VWF:FVIIIB, but not the rate of synthesis of mutated rVWFs. Previous expression of different mutated rVWF harbouring a type 2N VWD mutation has revealed that T791M (Cacheris et al, 1991) and R816W (Tuley et al, 1991) completely destroyed the FVIII-binding capacity of the corresponding rVWF; R854Q only markedly decreased this function (Kroner et al, 1991). VWF:FVIIIB was absent for R1060rVWF and markedly decreased (similar to that of Q854rVWF) for H1053rVWF. Thus, the C1060R mutation has a more deleterious effect on VWF:FVIIIB than the Q1053H mutation. Furthermore, in contrast to H1053rVWF, R1060rVWF showed an abnormal multimeric pattern characterized by a decrease in the highest molecular weight multimers, in agreement with that observed in the plasma VWF of patients harbouring this mutation. As cysteine residue 1060 is paired to C1084 in normal VWF (Marti et al, 1987), the substitution from Cys-1060 by an arginine residue may induce mispairing of the cysteine residues localized in the D3 domain and therefore affect the multimerization process.

To investigate the hypothesis of a conformation change, an antigen-capture experiment with four mAbs that recognize epitopes in the D′ (aa 769–865) and D3 (aa 866–1241) domains, was performed. There was no difference in the ability of these mAbs to bind WT and H1053rVWF. In contrast, R1060rVWF, which was equally well recognized by mAbs 9311A2, 418 and 32B12, displayed a dramatically increased affinity for mAb 14A12, although cysteine 1060 is located outside its epitope (aa 904–983). Interestingly, mAb 14A12 still recognized reduced rVWF and other type 2N rVWFs with impaired multimerization resulting from cysteine substitutions (unpublished observations). Thus, the enhanced affinity of R1060rVWF for mAb 14A12 is not caused by the impaired multimerization. Unfortunately, to date, there is no exact information available on the tertiary structure of the FVIII-binding domain of VWF, and only speculation can be made on the structural impact of the Q1053H and C1060R mutations.

In conclusion, the identification of the Q1053H and C1060R type 2N VWD mutations confirmed that the FVIII-binding capacity of VWF depends not only on the 272-aa N-terminal region of mature VWF, which contains the FVIII binding site. Some aa changes outside the 272-aa tryptic fragment may induce a conformational change in this part of the molecule that could be responsible for the loss of function of the FVIII-binding domain. This also indicates that the D3 domain is a hinge region, playing a major role in both VWF:FVIIIB and multimerization. Indeed, the C1060R mutation is the third mutation in the D3 domain, after D879N (Jorieux et al, 1998) and C1225G (Allen et al, 2000), that we have found to affect both VWF:FVIIIB and multimerization of VWF.

Acknowledgments

We wish to thank J.-M. Lavergne (INSERM U143) for sequencing the FVIII gene in patient A, V. Barylo, S. Belmont, F. Degallaix and D. Hoguet for their excellent technical assistance, and all the members of the French INSERM Network on Molecular Abnormalities in von Willebrand Disease.

Ancillary