In this study, it was clearly shown that, in order to determine the influence of VWF in the first step of haemostasis, solely the quantification of the platelet adhesion – using in vitro standard procedures – is not sufficient. Only the comprehension of the biophysical component of the fluid dynamics on the one hand, the integration of both collective and single molecule phenomena on the other hand, and finally the distinction between immobilised and soluble VWF opens up the possibility for a targeted investigation of the mechanistic background.
Platelet adhesion on immobilised VWF is a time dependent process. To study the initial step of platelet adhesion due to interactions of the VWF A1-domain and the platelet glycoprotein (GP) Ib, platelet activation was inhibited as previously described (26). Thus, platelet derived VWF was excluded as a further source of the mobile VWF fraction. As expected, biofunctionalisation with the deletion mutation del-A1 succeeded in coating but completely failed to bind platelets. Regarding the platelet adhesion capability on immobilised VWF, the deletion mutations del-A2 and del-A3 did not show any significant variation compared to wt VWF. As bleeding events are known to cause pathological high-shear conditions in the area directly affected by the damaged vessel (27), we next focused on short-termed VWF-platelet interactions upon different shear flow regimens. Under low- or intermediate-shear conditions, the impact of the mobile VWF fraction on platelet adhesion to VWF biofunctionalised surfaces was negligible. Nevertheless, in the entire absence of soluble VWF, platelet binding to the surface was significantly diminished upon high shear application (see Fig. 3). In contrast, in the presence of soluble VWF rising shear led to an enhanced platelet adhesion. In accordance with former publications (28, 29), platelet decorated strings appear within seconds (see Electronic Supplementary Material 1), most likely due to a recruitment of stretched mobile VWF to the platelet surface.
Although single platelet adhesion to the channel footprint and consecutive recruitment of soluble VWF are the likely prerequisites for aggregate formation, adhesion characteristics have no predictive power for the formation of rolling VWF-platelet aggregates. As shown for the selected deletion variants of the A-subdomains of VWF, a deletion of the A1-domain indeed led to a significant decrease in adhesion and a complete loss of aggregate formation. Nevertheless, while the impact of the A3-domain on adhesion and aggregate formation seemed to be marginal, a deletion of the A2-domain – although leading to a negligible change in platelet adhesion – induced a significant reduction in ɣcrit, i.e. an increase in the formation of VWF-platelet aggregates. Note that due to the absence of divalent cations during the whole course of microfluidic experiments, the activity of the VWF degradation enzyme ADAMTS-13, known to specifically interact with the VWF A2-domain, is inhibited to facilitate the concentration on the VWF interdomain affection. Although one could raise the hypothesis that adhesion and aggregate formation are regulated by specific binding sites on different parts along the molecule (e.g. the collagen binding site of the A3-domain) (30, 31), we here prefer to look at this molecule from a different more physical angle:
Our results strongly support a hypothetical scenario suggested earlier by Ruggeri et al. (6): Single platelets tethered to immobilised VWF function as nucleation centres. Soluble VWF binds to the immobilised platelet, in particular under high shear flow, thereby representing a nucleation centre itself supporting the growth of the aggregate. The aggregate formation can be explained by a three-step process and the “matching” of two timescales. First, the elongational contribution of the shear field supports the elongated state of VWF (Fig. 6a). The shear field also introduces the first timescale
(gamma representing the shear rate), the period of one rotation of a platelet exposed to the field. The second timescale τbind arises from the binding kinetics and corresponds to the time the elongated VWF typically resides on the platelet. Of course, in reality, all these quantities are thermodynamic quantities and will vary with temperature, pH, ion concentrations etc. in other words they are represented by their appropriate diagrams of state (32). Nevertheless, if τrot ˂ τbind the VWF will reside long enough on the platelet to wrap around it (Fig. 6b). This will effectively coat the platelets due to the enlarged contact area. Via VWF-VWF interactions the platelets can now begin to form aggregates, given a high enough concentration of platelets (Fig. 6c).
We must note that a recent work explaining how VWF adheres to surfaces at high shear rates can also be used to explain why aggregate formation occurs only in the presence of soluble VWF (33). In particular, at shear rates above 1,000 s− 1, platelets are exposed to shear forces in the range above of 10 pN. This force is around the rupturing force of the GPIb-A1 bond (34). Thus, in excess of this force platelets need a cooperative mechanism to bind. By forming aggregates, it is possible to overcome this limit since the binding to the substrate leads to a higher valence arising from the other platelets on the string. As long as the lifetime of the bonds is long enough, such aggregates will be able to bind and immobilise on the substrate.
The gain of function observed during aggregate formation for the del-A2 mutations reveals that the simple picture of altered VWF-platelet association (Fig. 2) leads to a false prediction of the aggregation behaviour. As shown in Fig. 5, the threshold shear rate at which aggregates form for wt VWF is of the order of 4,000 s− 1; however, in the case of the deletion of the A2-domain it is 2,500 s− 1. The ~ 40% change in this threshold shear rate can be accounted by an increase in the binding strength between the A1-domain and the platelet GPIb receptor. As has been previously speculated, we can confirm that the GPIb-A1 interaction is regulated by the presence of the A2-domain (21, 35). It is believed that the interaction is masked by the A2-domain yielding an effective lifetime of the GPIb-A1 bond that displays a “catch-bond” behaviour (36–38). In the absence of the A2-domain, the lifetime at lower shear forces of the GPIb-A1 bond was radically increased. By how much? Based on our previous work on aggregate formation (20), it is possible to measure such differences. In particular, in order to reduce the critical shear rate by x, the lifetime of the GPIb-A1 has to increase by the same factor x. By assuming that all prefactors remain the same, we find that the effective interaction strength between the GPIb and the A1-domain has to increase by ~ 1 kT, where k is the Boltzmann constant and T the temperature.
Such variation is small, yet has a dramatic impact on the clotting process. Thus, we predict that the A2-domain effectively “masks” the potential between the GPIb and the A1-domain by about 1 kT. While it would be hard to resolve, this change should not carry over into the high shear regimen as the shear forces will already be enough to pull apart the domains and reduce the masking ability of the A2-domain.