Force Dependent Structural Changes of the Von Willebrand Factor A2 Domain Are Required for VWF Cleavage by ADAMTS13

Background: An important regulatory mechanism of VWF functional activity is the proteolysis of the VWF high molecular weight multimers by ADAMTS13. The ADAMTS13 specific proteolytic site is located in the VWF A2 domain between Y1605 and M1606. Homology models by Sutherland et al. (1) predict the ADAMTS13 cleavage site of A2 to be covered by the helices of the domain. Thus an unfolding or opening of the domain is necessary in order to ensure cleavage by ADAMTS13. The main difference between the three neighboring VWF A domains is a disulfide bond connecting the respective N and C termini of the A1 and A3 domain, which is lacking in A2. An obvious assumption is that the disulfide bonds lock A1 and A3 in the folded state, while A2 can unfold in response to shear stress in blood vessels. To test this, we examined if the mutation N1493C in A2 enables the formation of a disulfide bond analogous to A1 and A3, thereby preventing unfolding of the A2 domain and stabilizing VWF against ADAMTS13.

Methods: By in vitro mutagenesis of full length VWF we exchanged N1493 at the N-terminal site to Cysteine and C1670, one of two neighboring cysteines at the C-terminal site of the A2 domain, to Serine to allow creation of a Cysteine bonded loop (177 aa) in the A2 domain. Mutant and wt VWF were then digested by recombinant ADAMTS13 under denaturing non-reducing and reducing conditions, respectively, and analyzed by electrophoresis. Based on multiple sequence and structural alignments we created a homology model of the VWF A2 domain (residues 1488 to 1676 of human VWF) and the mutant A2 domain (N1493C and C1670S) from a human VWF A1 X-ray structure (PDB: 1AUQ). Our models cover a longer sequence (residues 1488 to 1676 of human VWF) than the model proposed by Sutherland et al.(1) The models were validated by molecular dynamics (MD) simulations. Force-probe MD simulations were applied to wildtype (wt) VWF to investigate unfolding.

Results and Discussion: Successful introduction of a disulfide bond into A2 was validated by VWF multimer analysis. An additional loop due to a disulfide bond resulted in faster migration of VWF, suggesting a more compact structure. Persistence of the large VWF multimers and the lack of proteolytic products after treatment with recombinant ADAMTS13 indicated resistance of the A2 mutant to ADAMTS13. Electrophoresis of ADAMTS13 treated A2 domain mutant VWF under reducing conditions demonstrated a majority of undigested VWF monomers suggesting that proteolysis is not only prevented by creation of a disulfide bond spanning the proteolytic site but indeed by structural changes of the A2 domain, making the proteolytic site less accessible for ADAMTS13 in general. Physiological blood flow exerts forces onto VWF, stretching the multimeric protein to an elongated conformation. Force-induced unfolding of A2 is observed in force probe MD simulations by subjecting the termini of the A2 domain to a pulling force mimicking the tension in stretched VWF. The ADAMTS13 cleavage site is located at β-strand 4 (β4), which is after stepwise unfolding of α6, β6, α5, β5, and α4 exposed and therefore accessible for ADAMTS13. This explains the size regulation of VWF by ADAMTS13: larger multimers involve higher pulling forces and thereby a higher unfolding ratio at a given shear flow. As a result, larger VWF is cleaved more readily.

Conclusion: Our combined experimental and theoretical approach suggests that locking of the VWF A2 domain by an artificial disulfide bond prevents a shear response and emphasizes the requirement of a force sensitive A2 domain structure for ADAMTS13 proteolysis. Respective experiments under shear conditions should confirm our hypothesis.