In recent years we, and others, have demonstrated that calmodulin is associated with membrane-proximal, positively-charged, amphipathic sequences in GPIbβ, GPV and GPVI in resting human platelets (Figs 2 and 3) [56,64,65]. Activation of platelets through GPIb–IX–V by thrombin or by ristocetin/VWF leads to the rapid dissociation of calmodulin from GPIbβ and GPV . Similarly, activation of platelets with the GPVI-specific agonist, CRP, led to the rapid dissociation of calmodulin from GPVI . We therefore investigated whether calmodulin dissociation acted as a trigger for ectodomain shedding in these receptors . In this regard, it has long been recognized that an ectodomain fragment of GPIbα (termed glycocalicin) is continuously and constitutively shed from platelets, and that glycocalicin circulates at high concentrations in plasma (∼3 μm) [2,34,66]. GPV has also been demonstrated to be shed from activated platelets, by a metalloproteinase-dependent mechanism . Calmodulin inhibitors such as W7 induce shedding of GPIbα and GPV, as well as GPVI [10, 67; unpublished observations]. Analysis with a rabbit polyclonal antibody directed against the GPVI cytoplasmic tail indicated that loss of intact GPVI correlated with the formation of a membrane-bound GPVI stump of ∼10 kDa molecular weight. The loss of intact receptor was blocked by EDTA, and by the generic metalloproteinase inhibitor, GM6001 . Shedding was also induced by treating intact platelets with the thiol-modifying reagent, NEM, suggesting shedding was ADAM dependent . Indeed, in mouse platelets, Nieswandt, Wagner et al. [13–15] have demonstrated using mice where ADAM17 is expressed in an inactive form in the hematopoietic compartment (the ADAM17-knockout mouse is perinatal lethal), that shedding of GPIbα and GPV are mediated almost exclusively by ADAM17, although it is unclear whether this also applies to human platelets. Treating platelets with the mitochondrial-targeting reagent, CCCP, mimicking platelet aging also induces ADAM17-mediated GPIbα shedding, and there is decreased GPIbα expression on aged platelets . Aspirin also promoted ADAM17-mediated shedding of GPIbα and GPV from human or mouse platelets, with increased levels of the respective ectodomain fragments occurring in plasma . The mechanism for this metalloproteinase-mediated shedding may involve acylation of ADAM17 and/or substrate(s), rather than the classical antithrombotic target for aspirin, cyclooxygenase-1 (COX-1), as shedding was normal in COX-1-deficient mice .
Consistent with the GPVI agonist-dependent loss of GPVI-associated calmodulin , GPVI agonists such as collagen, CRP and CVX induce a rapid loss of GPVI from intact platelets, and the appearance of a ∼55-kDa soluble fragment in the supernatant, relative to intact GPVI in platelets (∼62 kDa) [9–11]. Shedding was blocked by treatment of platelets with EDTA, or with GM6001. In contrast, other membrane receptors, such as PECAM-1, were not shed from the platelet surface under the same conditions . Nor were GPIbα and GPV shed under these conditions, suggesting a shedding mechanism specific to GPVI. In contrast, GPIbα and GPV are shed in response to platelet activation by low-dose thrombin, whereas GPVI is shed to a lesser extent by this agonist (unpubl. obs.). This and other evidence suggests that GPVI is shed by a different ADAM than ADAM17, possibly ADAM10 . GPVI agonist-induced shedding was dependent at least on early GPVI dependent signaling and was blocked by inhibitors of Src family kinases (PP2), Syk (piceatannol) and PI 3-kinase (wortmannin) . Calmodulin dissociation from GPVI on CRP-dependent platelet activation was also blocked by PP2 . In contrast, W7- and NEM-induced shedding of GPVI is activation independent, as W7 causes calmodulin/receptor dissociation and NEM directly activates surface ADAM activity [22,23]. GPVI shedding can also be artificially induced in vivo using the antimouse GPVI monoclonal antibody, JAQ1, which selectively depletes GPVI expression on mouse platelets . Alternately, human platelets injected into a NOD/SCID mouse can be depleted of GPVI in an activation-independent manner by antihuman GPVI antibodies .