Introduction
Diabetes mellitus is a major risk factor for the development of cardiovascular disease(s), and the morbidity and mortality associated with diabetes are frequently related to micro- and macro-vascular complications, characterized by accelerated atherothrombosis [
2]. Several mechanisms contribute to such a diabetes-associated prothrombotic state, including endothelial dysfunction, hypercoagulability and platelet hyperactivation [
6,
45]. Although it is well accepted that platelets are involved in the regulation of vascular homeostasis, exactly how they contribute to changes in the vascular wall is not fully understood. One mechanism by which platelets affect the vascular wall is through the release of factors stored in platelet granules. For example, following their release from α-granules chemokines; such as CXCL4 and CCL5, are deposited on the endothelial cell surface to initiate monocyte recruitment and diapedesis [
22]. Dense granule contents also appear to play a critical role in thrombosis and vascular remodeling, as Hermansky–Pudlak syndrome 3-deficient mice; which demonstrate impaired platelet dense-granule secretion, are protected from thrombotic arterial occlusion and the development of neointimal hyperplasia [
21]. In addition to the release of soluble factors, platelets can also affect vascular homeostasis through the release of platelet-derived microparticles (PMPs) that contain proteins and microRNAs that can be transferred to the vasculature [
44].
One group of platelet proteins that have been linked with platelet hyperactivation in the context of diabetes are the Ca
2+-dependent cysteine proteases or calpains [
39]. The latter are involved in several steps of platelet activation and calpain activation affects integrin signaling, aggregation, spreading, and granule secretion [
24]. Interestingly, calpains can also be secreted by platelets and have been detected in PMPs [
12,
33,
37]. Indeed, calpain activity in plasma correlates well with PMP levels and is significantly higher in plasma from diabetic subjects than from healthy volunteers [
37]. The current study was designed to determine the relevance of platelet-derived calpain 1 (CAPN1) in the vascular complications associated with diabetes by identifying new calpain target proteins on the surface of endothelial cells. Use was made of CAPN1
−/− mice and mice lacking CAPN1 specifically in platelets (CAPN1
ΔPF4 mice) to assess the importance of extracellular CAPN1 on endothelial cell activation and vascular inflammation.
Discussion
The results of this study indicate that platelet-derived CAPN1, carried by PMPs, cleaves PAR-1 on endothelial cells to initiate a cascade of events leading to the activation of TACE and the shedding of EPCR and TNF-α. The latter increases the expression of adhesion molecules and promotes vascular inflammation. It was possible to demonstrate correlative changes in circulating EPCR levels in plasma from healthy and diabetic subjects as well as non-diabetic and diabetic mice. Also in mice, calpain inhibition and the platelet-specific deletion of CAPN1 both prevented the diabetes-induced increase in circulating EPCR as well as the associated vascular inflammation. These data highlight a novel mechanism by which activated platelets can directly affect the homeostasis of the vascular wall to initiate the vascular complications associated with diabetes and vascular disease (see Online Fig. 10).
Calpains are involved in a variety of Ca
2+-regulated cellular processes by inducing the partial proteolysis of a broad spectrum of substrates [
16]. Although these effects have been mainly attributed to the activation of intracellular calpains, the proteases are also found in the circulation [
14,
34], where they have been linked with angiogenesis and vascular repair, in part by cleaving fibronectin and amplifying the effects of vascular endothelial growth factor [
27]. In the present study, CAPN1 carried by PMPs was found to induce the shedding of EPCR from the vascular wall. EPCR is a type I transmembrane protein that is important for the generation of the potent anticoagulant and cytoprotective protein; activated protein C (APC) [
13,
43]. While APC inactivates factors Va and VIIIa to exert its anticoagulant effects, its cell-protective actions are attributed to the cleavage of PAR-1 [
41], an effect that requires its binding to EPCR [
29,
40,
41]. This explains why EPCR has been generally classified as cytoprotective and why EPCR shedding has been associated with vasculopathy [
26,
42]. Although the EPCR was picked up by the proteomic approach used to identify endothelial cell surface proteins targeted by CAPN1, the protease was unable to cleave the EPCR in cell lysates, indicating that it was an indirect target. Rather, fitting with the fact that the shedding of the EPCR is controlled by TACE [
18,
35], it was possible to prevent the CAPN1-induced decrease in EPCR using a TACE inhibitor. Ours is not the first report to link calpain with the regulation of TACE as the Ca
2+ ionophore-induced, TACE-dependent shedding of glycoprotein Ibα in platelets was previously attributed to calpain activation [
47]. However, the authors of the latter study did not address the mechanisms involved or the physiological consequences in any detail. Perhaps the best known target of TACE is TNF-α [
28], which is synthesized as a 26 kDa transmembrane pro-protein that is bound to the endothelial cell surface. TNF-α released as a 17 kDa peptide into the extracellular space only after TACE activation. We found that TNF-α was present on the endothelial cell surface and that extracellular CAPN1 effectively decreased the cell-bound form of the protein. The consequence of this process was endothelial cell activation and the expression of ICAM-1. Such an increase in the levels of adhesion molecules on the surface of endothelial cells is a prerequisite for the adhesion of circulating monocytes and represents an early step in the development of inflammatory responses.
The next step was to identify a link between extracellular CAPN1 and the activation of a signaling cascade that could affect TACE activation. We focused on PAR-1, as this receptor has been previously linked with TACE-dependent EPCR shedding [
17]. Moreover, the activation of PAR-1 requires the proteolytic cleavage of the extracellular N-terminal domain of the protein, thus generating an amino terminus that functions as a tethered ligand to initiate signaling [
1]. CAPN1 was found to cleave the PAR1 receptor and generate PAR-1-derived peptides similar to those generated by thrombin, which cleaves the N-terminal domain of the protein at Arg41 [
46]. That a protease other than thrombin is able to activate the PAR-1 receptor is not that unusual, as other proteases can cleave the receptor, albeit at distinct sites. For example, the APC/EPCR complex can cleave PAR-1 at Arg41 and Arg46 [
40], while matrix metalloprotease-1 cleaves PAR-1 to create a longer tethered ligand [
23]. The different proteases activate different signaling pathways, depending on the ligands released. While thrombin-induced PAR-1 activation leads to Gq-dependent signaling resulting in increased intracellular Ca
2+, Rho activation as well as the phosphorylation of protein kinase C, ERK1/2 and AKT [
32], APC-induced signaling involves Gi proteins, Rac and even transactivation of the sphingosine 1-phosphate receptor [
8,
9]. We found that the CAPN1-induced signaling via the PAR-1 receptor resulted in the activation of ERK1/2 and RhoA but not AKT. Like thrombin, CAPN1 also disrupted endothelial cell barrier function to increase permeability as well as TACE activity. Such a mechanism very probably contributes to the diabetes-associated vascular leakage that characterizes diabetic microangiopathy.
Where does extracellular CAPN1 come from? CAPN1 has been identified in the platelet secretome [
11], but most circulating CAPN1 in diabetic individuals is contained in PMPs. Such microparticles are an effective way to transport and transfer biological information as they can fuse with the membrane of target cells to deliver their contents at the cell surface or can be internalized through presentation of specific antigens [
10,
25]. Although recombinant CAPN1 was used in many of the experiments performed in the present study, the effects of CAPN1 could be reproduced by microparticles generated in vitro from washed platelets. Given that platelets contain CAPN1 and CAPN2 [
24], and that PMPs represent up to 80% of all circulating microparticles [
10,
19], it seems reasonable to assume that PMPs are the main source of circulating calpains. However, not all PMPs are able to cleave PAR-1, possibly because not all PMPs carry high amounts of activated calpains. Importantly, we could show that microparticles isolated from healthy individuals had minimal effects on EPCR shedding while microparticles from diabetic subjects (which contain more active CAPN1) elicited a pronounced effect. In the present study, ionomycin was used to generate murine PMPs in vitro, a stimulus known to activate both CAPN1 and CAPN2 [
15,
37]. However, it was possible to demonstrate that PMPs from CAPN1
−/− mice were less effective at decreasing PAR-1 levels and increasing endothelial cell ICAM-1 expression than PMPs from wild-type mice. To demonstrate the in vivo relevance of the pathway described, mice were made diabetic with STZ. While the induction of diabetes led to the loss of the N-terminal of the PAR-1 receptor on aortic endothelial cells and a concomitant induction of ICAM-1, the animals given the calpain inhibitor were protected. Some authors have expressed concern about the use of STZ to induce diabetes [
5], but we were able to confirm our findings in diabetic (type I) Ins2
Akita mice which carry a mutation in the Ins2 gene [
48,
49]. Moreover, we have previously reported that calpain activation in platelets was found in both type 1 and type 2 diabetes [
37].
The calpain inhibitor used was previously reported to prevent platelet activation in diabetic mice [
37], as well as to prevent the diabetes-induced generation of platelet microRNAs that could potentially affect endothelial cell protein expression [
7]. Therefore, diabetes-induced changes in PAR-1 and ICAM-1 expression were studied in animals specifically lacking CAPN1 in platelets. The importance of platelet-derived CAPN1 in the vascular complications of diabetes was demonstrated by the fact that diabetes induction in CAPN1
ΔPF4 mice failed to increase circulating levels of the EPCR or to alter the surface expression of PAR-1 or ICAM-1 on endothelial cells. Our findings have a clear pathophysiological relevance as they imply that increased circulating levels of calpain-enriched PMPs in diabetic patients are responsible for the activation of the endothelial cell PAR-1 receptor. This in turn can initiate a sterile vascular inflammation via the release of TNF-α, to promote EPCR shedding and thus attenuate cytoprotective EPCR-APC signaling. As CAPN1 is ubiquitously expressed, it is clear that the activation of the protease in endothelial cells can also contribute to vascular dysfunction and the activation of endothelial cell CAPN1 was reported to cleave prostaglandin synthase in small arteries to decrease prostacyclin formation [
38]. However, the finding that platelet-derived calpain carried by PMPs circulates has important implications for the development of vascular disease, and can contribute to the spread of endothelial cell activation and vascular inflammation.