Characterisation of W1745C and S1783A, Two Novel Collagen Binding Defects in the A3 Domain of Von Willebrand Factor

Investigation of three families with von Willebrand disease showed that haemorrhagic symptoms were associated with disproportionately reduced collagen binding activity whilst Ristocetin co-factor activity was commensurate with antigen and multimeric analysis was normal. Genetic analysis revealed heterozygosity for two novel mutations in two of the families: W1745C in exon 30 and S1783A in exon 31. In the third family the affected individuals were heterozygous for a previously-described mutation: S1731T in exon 30 but two unaffected individuals also carried this mutation. All three mutations lie in the A3 domain containing the main collagen binding site in VWF. In patients’ samples VWF:CB activity was measured using human type I and type III collagen. Patients heterozygous for W1745C and S1731T showed a reduction in binding to both collagens but more marked reduction in binding to type III collagen. Heterozygosity for S1731T resulted in mild impairment of type I collagen binding but normal binding to type III collagen. Site-directed mutagenesis was used to generate vectors containing the three mutations (S1731T, W1745C and S1783A) and also one containing a W1745A mutation. Mutated VWF was expressed in HEK293T cells both singly and in co-transfection with a wild-type VWF (wtVWF) vector. All VWF mutants were expressed at a similar rate to wtVWF. Multimeric analysis demonstrated that all the mutants had a similar multimeric structure compared to recombinant wtVWF. However recombinant-wtVWF (wtVWF) had a lower collagen binding to VWF antigen ratio (CB:Ag) compared to plasma VWF (0.39 type I collagen and 0.45 type III collagen vs >0.7 for plasma VWF). This is most likely due to the slight shift towards lower molecule weight multimers seen with recombinant VWF. CB:Ag ratios for the recombinant VWF showed the same pattern of binding to collagen type I and III as the clinical samples. The W1745A mutant demonstrated a similar CB:Ag ratio to W1745C. Kinetic analysis of binding to type I collagen demonstrated that W1745C, W1745A and S1783A did not bind and that S1731T bound with significantly less affinity compared to wtVWF (K_D,app 27.1 ± 0.5nM and 7.3 ± 0.8nM respectively). Analysis of binding to type III collagen demonstrated that W1745C and W1745A both bound with ~ 8-fold reduced affinity (K_D,app 16 ± 2.6nM and 21.3 ± 6.3nM) but wtVWF and S1731T bound with similar affinity, (K_D,app 2.0 ± 0.1nM and 3.7 ± 0.85nM respectively). Analysis of the crystal structure of the VWF A3 domain showed that W1745 may interact with Y1780 and we noted the mutation Y1780A has also been shown to significantly reduce collagen binding. Measurement of free thiols present in VWF demonstrated that the new cysteine residue in W1745C is not involved in disulphide bond formation. These results indicate that it is the loss of W1745 rather than the creation of a new cysteine residue that is responsible for the loss of collagen binding activity. We therefore hypothesised that W1745 and Y1780 participate in an internal aromatic interaction that helps to maintain the structural configuration of A3. We sought confirmation by expressing another mutant; W1745F, replacing the tryptophan with another aromatic amino acid. As predicted this did not significantly affect collagen binding. In conclusion, our findings demonstrate that type 2 VWD may be arise from mutations in A3 causing abnormal collagen binding without other functional defects or abnormalities in multimer formation. This type of VWD may be under-recognised unless laboratories measure binding to both types I and III collagen. Mutations in A3 yield insights into the structural requirements for collagen binding may have differential effects on binding to collagen types I and III and can result in variable clinical phenotypes. Some mutations may not be consistently associated with bleeding symptoms.