TO THE EDITOR:
Understanding the differences between physiological hemostasis and pathological thrombosis at molecular and cellular levels is a long-standing goal of platelet and coagulation biology. Such a distinction will pave the way for designing strategies to pharmacologically inhibit thrombosis without an increased concomitant tendency to bleed. Given that bleeding is the most common adverse effect associated with antithrombotics, with some cases being life-threatening, it is essential to distinguish between hemostasis and thrombosis.
A principal difference is that hemostasis and thrombosis do not occur in the same local environment; while hemostasis occurs within tissue in the extravascular space, thrombosis is essentially an intravascular process (Figure 1). Additionally, although hemostasis and thrombosis share common pathways at a molecular level, increasing evidence has demonstrated that they are mechanistically distinct. For example, animal model gene knockout (KO) or inhibition of coagulation factor XI (FXI),1,2 FXII,3,4 glycoprotein VI (GPVI),5-8 water channel aquaporin-1 (AQP1),9 or protein kinase C alpha10 show no significant bleeding defects but markedly affect arterial thrombosis. Additionally, pharmacologic inhibition of FXI11,12 and FXII13 in preclinical trials provides evidence that hemostasis and thrombosis can be uncoupled. In this commentary, we propose the hypothesis that platelet procoagulant membrane ballooning and microvesiculation, collectively termed procoagulant membrane dynamics (PMD), which amplify the procoagulant response in platelets, could provide a mechanistic distinction between hemostasis and thrombosis, playing a critical role in thrombosis but not hemostasis (Figure 1). Importantly, although classical platelet hemostatic plug formation (principally aggregation and secretion) and procoagulation play important roles in both hemostasis and thrombosis, we propose that the principal procoagulant source differs for each event. As subsequently explained, platelet PMD may be the predominant contributor to procoagulation in thrombosis, whereas other cells may substantially contribute to the procoagulant response in hemostasis (Figure 1).
Platelet activation by GPVI leads to classical signaling switching on secretion and aggregation responses to generate a platelet hemostatic plug, which plays important roles in both hemostasis and thrombosis. However, GPVI signaling, particularly when combined with protease-activated receptor activation, can also result in a sustained increase in cytosolic calcium required for an additional response: procoagulant platelet formation.28-34 Procoagulant platelets are typified by an unusual and dramatic change in shape, membrane ballooning, and subsequent microvesiculation (PMD), paralleled by cyclophilin D-induced MPTP formation and pore opening leading to necrosis.31,35,36 These procoagulant ballooned platelets are evident in both arterial thrombi9,37 and in tissues in hemostasis38,39 (Figure 1). Previous studies have demonstrated that αII2bβ3 activation and platelet aggregation are uncoupled from platelet PMD,40-42 providing evidence of the ability of platelets to assume functionally segregated phenotypes after stimulation; however, this platelet response is likely equally evident during thrombosis and hemostasis. However, we propose that, unlike hemostasis, the procoagulant process underlying thrombosis, particularly arterial thrombosis, is reliant on platelet PMD, whereas a significant procoagulant contribution comes from cells other than platelets in hemostasis (Figure 1). The additional sources of negatively charged phospholipids in hemostasis include red blood cells,16-19 neutrophils,22 macrophages,20,21 monocytes,21,23 and endothelium.24-27 These are all important for tenase and prothrombinase complexes and for de-encryption of tissue factor43 and activation of associated FVIIa (Figure 1).
Therefore, targeting platelet genes and processes that regulate platelet membrane ballooning or microvesiculation may have preferential effects on thrombosis while sparing hemostasis. This is in contrast to drugs that inhibit pathways leading to platelet aggregation or secretion, which affects both thrombosis and hemostasis. So, for example, aspirin and P2Y12 blockers such as clopidogrel, ticlopidine, ticagrelor, prasugrel, and cangrelor are effective in dampening the aggregation response, have bleeding side effects associated with their antithrombotic activity, and humans lacking the P2Y12 receptor have a mild bleeding defect.44
Several procoagulation factors interplay during hemostasis and thrombosis, and both events require cell membrane driven-coagulation (Figure 1). However, coagulation mechanisms in either setting may be different. Key differentiators between hemostasis and thrombosis, which can be clinically monitored and targeted, are platelet PMD, the principal part of which is membrane ballooning or microvesiculation. The process is driven by the platelet GPVI receptor/collagen combination, usually in concert with other agonists, especially thrombin.28,37,45 We define platelet PMD as the composite morphological changes that platelet undergo to support thrombin generation. These include the formation of functionally segregated platelet phenotypes, membrane PS exposure, membrane ballooning, and the breakdown of the membrane into procoagulant microvesicles (Figure 1). Mechanistically, platelet membrane ballooning is mediated by salt (Na+ and Cl−) and water entry into the platelet, facilitated by the water channel AQP1.9 Water entry into platelets leads to the stretching of the platelet plasma membrane, the opening of mechanosensitive cation (TRPC6) channels,9 anadditional influx of extracellular Ca2+, and a sustained rise in cytosolic calcium. Concurrently, the combination of internal hydrostatic pressure and external osmotic pressure leads to irreversible inflation of the platelet membrane into balloons.37,46 Because of the Ca2+-dependent scramblase activity, which is likely to be transmembrane protein 16F (TMEM16F) principally, the result is PS exposure on the platelet membrane and the distinct phenotypic procoagulant transformation of activated platelet into ballooned-nonspread or balloon-and-procoagulant-spread platelets, and PS-exposing microvesicles.37,45 Platelet membrane ballooning (and likely microvesiculation) is enabled by microtubule disruption and effectively increases the catalytic surface area for the localized assembly of tenase and prothrombinase complex (the serine protease, factor Xa, and the protein cofactor, factor Va)46-48 and promotes thrombin generation on the platelet surface and the amplification of the clotting process47 (Figure 1).
We expect GPVI blockade to be potently antithrombotic, as clearly established in animal models.49-55 However, blockade of GPVI in humans is not as antithrombotic as predicted. For example, antithrombotic clinical studies of GPVI blocking agents, such as Revacept, have shown beneficial effects in patients with symptomatic internal carotid artery stenosis56 but not in those undergoing elective percutaneous coronary intervention for stable ischemic heart disease.57 However, further research is needed to explain this dichotomy. Furthermore, it is plausible that blocking GPVI may also hamper hemostasis and precipitate some bleeding.57-59 This is likely to be because platelet GPVI activation leads to platelet plug formation (mediated by aggregation and secretion) and significantly drives PS exposure and other features of platelet PMD,28,31,35,37,40,45,46 including membrane ballooning and microvesiculation,28,37 occurring in both hemostasis38,39 and thrombosis28,37 (Figure 1). Our view here is that antithrombotic approaches suppressing platelet PMD may spare hemostasis better because platelet PMD is more predominant as a procoagulant signal in thrombosis than in hemostasis (Figure 1). For example, data from AQP1 KO mice showed that partial blockade of platelet PMD (AQP1 deletion suppresses platelet microvesiculation but not ballooning) inhibits thrombosis in vivo while sparing hemostasis.9,60 In this model, membrane ballooning was likely rescued by compensatory mechanisms involving the isoform AQP7 expressed in platelets9; making AQP1 a subtle yet effective antithrombotic target compared to GPVI ablation.
Additionally, TMEM16F regulates Ca2+-dependent phospholipid scrambling for PS exposure and show its highest expression in platelets.61 Accordingly, the loss of the TMEM16F62 as in the human Scott syndrome63 ablates platelet PS exposure,64 microvesiculation64 and thrombosis,64 and precipitates a mild bleeding disorder.62 This is likely due to the fact that absence of TMEM16F has a marked effect on procoagulation in general, whether in platelet PMD,62,64,65 or other cells, and therefore effectively partially disables the procoagulant response in both hemostasis and thrombosis equally. However, hemostasis is only mildly affected, and, therefore, leads to a mild bleeding phenotype. Interestingly, although platelet-specific TMEM16F KO (Ano6−/−) mice lack platelet PMD and are protected from thrombosis, evidence is less clear about hemostasis, with 1 study showing no increased bleeding time64 whereas another study showed an increase.66 Methodological differences from this limited number of published studies67-69 may explain this, although it is important to note that global TMEM16F deletion was reported as both protecting from thrombosis and increasing bleeding risk.62,70
Recently, we reported that carbonic anhydrase (CA) inhibitors target procoagulant platelet formation and provide a new way to control platelet driven thrombosis without blocking essential platelet granular release.71 This approach opens new avenues for the development of novel antithrombotic agents that act distinctly from antithrombotics currently used in clinics, such as P2Y12 blockers (clopidogrel, prasugrel, ticagrelor, or cangrelor). However, CA inhibitors such as acetazolamide and methazolamide are relatively nonselective, and are known to block the activity of nearly all 13 isoforms of the CA enzyme ubiquitously expressed in various tissues of the human body.72 Of these isoforms, CA1, CA2, and C13 are expressed in platelets.71 Platelet pretreatment with acetazolamide suppresses total PMD, (PS exposure, microvesiculation, and ballooned platelet formation) and significantly attenuates in vivo thrombosis.71 Based on our current model, as explained above, we expect acetazolamide or methazolamide to precipitate mild bleeding along with its potent antithrombotic effects in vivo, due to a wide-ranging suppression of nonplatelet PS sources. Indeed, a published report indicates acetazolamide use in human may cause some increased bleeding under normobaric conditions.73 A carbonic-anhydrase-based antithrombotic approach, likely to spare hemostasis in addition to platelet secretion, may be one targeted to isoform CA13 preferentially expressed in platelets.61
Ultimately, there is a significant overlap between hemostasis and thrombosis in terms of cellular and molecular mechanisms as well as some degree of difference. Hemostasis and thrombosis require 2 principal activities: (1) platelet plug formation, mediated by “classical” platelet activation, aggregation, and secretion, and (2) procoagulation. Our proposal here is that although these have equivalent roles in both hemostasis and thrombosis, procoagulation is mediated by some distinct processes, in that it is principally provided by platelets in thrombosis, whereas in hemostasis, it is also provided by other cells (tissue and blood cells) that can expose PS. Platelets are unusual when undergoing marked membrane ballooning and microvesiculation driven by GPVI and mediated by salt and water entry. Thus, antithrombotic agents that target mediators of platelet ballooning or microvesiculation are likely to spare hemostasis. Therefore, a novel drug class, namely, antiprocoagulant antithrombotics, exemplified by water channel inhibitors, may be important for development.
Acknowledgments: A.W.P. and I.H. are funded by the British Heart Foundation (PG/17/62/33190, SP/F/21/150023, and PG/21/10760) and Wellcome Trust funding (#: 219472/Z/19/Z), United Kingdom. E.O.A. is supported by the Cumming School of Medicine, McCaig Institute for Bone and Joint Health, and Libin Cardiovascular Institute, University of Calgary, Calgary, AB, Canada.
Contribution: E.O.A., I.H., and A.W.P. wrote the article.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Correspondence: Ejaife O. Agbani, Department of Physiology and Pharmacology, Cumming School of Medicine, University of Calgary, 3330 Hospital Drive NW, Calgary, AB T2N 1N4, Canada; e-mail: email@example.com; and Alastair W. Poole, School of Physiology, Pharmacology and Neuroscience, University of Bristol, Bristol, United Kingdom; e-mail: firstname.lastname@example.org.
Data are available on request from the corresponding author, Ejaife Agbani (email@example.com).