Authors: Eleni Maneta, Evmorfia Aivalioti, Simon Tual-Chalot, Besa Emini Veseli, Aikaterini Gatsiou, Kimon Stamatelopoulos, Konstantinos Stellos
Categories: Immunology, endothelial dysfunction, immunothrombosis, neutrophil extracellular traps, sepsis, tissue factor
Source: Frontiers in Immunology
Sepsis is a life-threatening clinical syndrome characterized by multiorgan dysfunction caused by a dysregulated or over-reactive host response to infection. During sepsis, the coagulation cascade is triggered by activated cells of the innate immune system, such as neutrophils and monocytes, resulting in clot formation mainly in the microcirculation, a process known as immunothrombosis. Although this process aims to protect the host through inhibition of the pathogen’s dissemination and survival, endothelial dysfunction and microthrombotic complications can rapidly lead to multiple organ dysfunction. The development of treatments targeting endothelial innate immune responses and immunothrombosis could be of great significance for reducing morbidity and mortality in patients with sepsis. Medications modifying cell-specific immune responses or inhibiting platelet–endothelial interaction or platelet activation have been proposed. Herein, we discuss the underlying mechanisms of organ-specific endothelial dysfunction and immunothrombosis in sepsis and its complications, while highlighting the recent advances in the development of new therapeutic approaches aiming at improving the short- or long-term prognosis in sepsis.
Keywords: immunothrombosis, endothelial dysfunction, neutrophil extracellular traps, tissue factor, sepsis
The latest definition by the Third International Consensus Definition Task Force defines sepsis as a life-threatening organ dysfunction resulting from a dysregulated host response to infection (1). Despite a better understanding of the underlying pathophysiological mechanisms and improvements to existing therapeutic algorithms, morbidity and mortality in patients with sepsis still remain high. It is estimated that 2.8 million deaths per year are attributed to sepsis, with one in three hospital deaths related to it (2, 3). Sepsis can originate from any infecting organism, but the most frequently identified ones are Staphylococcus aureus (S. aureus), Pseudomonas species, and Escherichia coli (E. coli). Among different sites susceptible to infection, the main ones leading to sepsis are the lungs (64%), the abdomen (20%), the bloodstream (15%), and the renal and genitourinary tract (14%) (2).
Activated components of innate immunity act as the first line of host immune defense against invading pathogens (4). Apart from their direct role in clearing the microorganism, neutrophils, monocytes, and the complement can interact with platelets, and the coagulation cascade leading to the local formation of thrombi in microvessels. This process, known as “immunothrombosis,” contributes to the recognition and restriction of a pathogen’s spread, tissue invasion, and survival, thus consisting an essential component of intravascular immunity (5). Although this distinct innate immune mechanism seems initially beneficial for the host, uncontrolled activation of this process in sepsis can trigger aberrant thrombosis in the microcirculation. During immunothrombosis, immune cells interact also with endothelial cells, which acquire a proadhesive and prothrombotic phenotype (5). Indeed, coagulation abnormalities are common in patients with sepsis, with severity ranging from subclinical coagulation activation to devastating thrombotic and hemorrhagic complications (6). In sepsis, both the activation of the coagulation cascade and the dysregulation of the endogenous anticoagulant and fibrinolytic mechanisms lead to microvascular thrombosis and tissue ischemia, which in turn contribute to tissue injury and multiorgan failure (7). Specifically, high thrombin and d-dimer plasma levels, both markers of coagulation activation, are considered indicative of severe sepsis (8, 9). Accordingly, decreased plasma levels of natural anticoagulants, such as protein C, protein S, and antithrombin (AT), have been found in sepsis and are associated with worse disease prognosis (9–11). Disseminated intravascular coagulation (DIC) is estimated to occur in up to 30%–50% of patients with severe sepsis, and it is associated with a twofold higher mortality risk (12). DIC is characterized by endothelial dysfunction and dysregulation of procoagulant, anticoagulant, and fibrinolytic factors (8). In sepsis-associated DIC, immunothrombosis is the underlying pathophysiological mechanism that triggers the initiation of DIC (5). Most of our knowledge regarding immunothrombosis suggests that it is a process mainly taking place in the microvasculature, affecting both arterioles and venules.
Despite the high prevalence of coagulation abnormalities and their potential deleterious effect on the outcome of sepsis patients, the current treatment approach is generally focused on antibiotics, fluid resuscitation, vasopressors, as well as oxygen and ventilatory support (13). Apart from heparin’s use for venous thromboembolism (VTE) prophylaxis in patients with sepsis, effective treatments specifically targeting the process of immunothrombosis, aiming to abrogate the devastating complications while maintaining its beneficial effects for the host, do not yet exist (13). A better understanding of the underlying mechanisms and cellular interactions in sepsis-related immunothrombosis is crucial to identifying new therapeutic targets that would open the field for the development of new treatments in sepsis. Our study summarizes the current literature regarding the role of endothelial cells and innate immune cells in immunothrombosis in sepsis and discusses clinical and preclinical evidence on drugs targeting immunothrombosis in sepsis.
The process of thrombus formation inside a blood vessel, which can lead to partial or total vessel occlusion, requires platelets’ accumulation and activation as well as activation of the coagulation cascade (14). The binding of platelet glycoprotein Ib (GP Ib) receptor to von Willebrand factor (vWF) and the interaction of platelet glycoprotein IIb/IIIa (GP IIb/IIIa) receptor with fibrinogen are essential steps for platelet adhesion and aggregation, respectively (14). The coagulation cascade involves the activation of a series of clotting factors and is classically divided into three the extrinsic, the intrinsic, and the common pathway. The extrinsic pathway is initiated when tissue factor (TF) from the damaged vessel wall interacts with activated clotting factor VII (aFVII), while the intrinsic pathway begins when factor XII (FXII) interacts with the exposed collagen and becomes activated (aFXII). Both pathways end up in factor X (FX) activation, which in turn, along with activated factor V (aFV), cleaves prothrombin into thrombin. Afterward, thrombin activates fibrinogen into fibrin and also cleaves factor XIII (FXIII) into activated factor XIII (aFXIII), resulting in clot stabilization (common pathway) (15, 16). In normal hemostasis, along with the coagulation cascade activation, endogenous anticoagulant mechanisms are being recruited in order to prevent uncontrolled clot expansion and thrombosis. Tissue factor pathway inhibitor (TFPI), AT, activated protein C (APC), and protein S, as well as thrombomodulin (TM), which along with endothelial protein C receptor (EPCR) catalyzes the thrombin-mediated protein C activation, are some of the main physiologic anticoagulant mechanisms (17–20). Finally, under normal conditions, plasminogen is activated by tissue plasminogen activator (tPA) into plasmin, which catalyzes clot dissolution and fibrinolysis, while plasminogen activator inhibitor (PAI) serves as a tPA inhibitor (21, 22).
Many of the clinical manifestations related to sepsis can be attributed to endothelial dysfunction which constitutes one of the most prominent underlying pathophysiological mechanisms of sepsis and sepsis-related complications. Although a variety of inflammatory diseases are characterized by endothelial dysfunction, sepsis is linked to a distinctive procoagulant, proadhesive and apoptotic profile of endothelial cells (23). Normal endothelium contributes to vascular health through the expression of antithrombotic and anti-inflammatory molecules (23, 24). In sepsis, the endothelium is activated directly by pathogen-associated molecular patterns (PAMPs) of bacteria, viruses, and fungi or indirectly via neutrophil extracellular traps (NETs) and proinflammatory cytokines such as tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin-1 (IL-1) ( Figure 1A ) (24–27). NETs are extracellular structures that are composed of cell-free DNA (cfDNA), histones, and antimicrobial proteins and are primarily released through a cell death process called NETosis or alternatively, through a process that does not require cell death, named non-lytic NETosis (28). PAMPs activate endothelial cells via pattern recognition receptors (PRRs), among which toll-like receptors (TLRs) exert the most profound inflammatory response in sepsis (29). Interestingly, endothelial activation during sepsis, as indicated by high circulating levels of intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1), E-selectin, P-selectin, and vWF, is correlated with disease severity and mortality (30–32).
Figure 1 Endothelial dysfunction in sepsis. (A) PAMPs, TNF-α, IL-1, IL-6, Ang2, and NETs activate endothelial cells during sepsis. This results in intracellular signaling activation, leading to transcription factors activation, and mainly NF-κB. (B) Endothelial glycocalyx damage during sepsis is mediated through heparanase and ADAM15, which cleave heparan sulfate and CD44, respectively. Glycocalyx deconstruction results in increased inflammation, thrombosis, and hyperpermeability in sepsis. Furthermore, sepsis leads to endothelial cell junctions’ disruption, which subsequently leads to increased endothelial permeability. (C) Sepsis is characterized by dysregulated nitric oxide synthase activation, leading to the production of reactive nitrogen species. Moreover, infection triggers endothelial cells to produce ROS, which in turn destroy the endothelium. (D) Endothelium activation during sepsis results in the expression of E-selectin, P-selectin, VCAM-1, and ICAM-1 on the endothelial cell surface and subsequent leukocyte rolling, adhesion, and transmigration through endothelial cells. Leukocytes express specific ligands for binding with endothelial receptors such as PSGL-1 and ESL-1 for leukocyte rolling and αLβ2 and α4β1 integrins for leucocyte adhesion. (E) Activated endothelial cells during sepsis express molecules such as vWF and P-selectin on their surface, which mediate platelet aggregation. This leads to the formation of platelet-neutrophil aggregates and subsequent neutrophil activation. Weibel–Palade bodies contain procoagulant factors such as VIIIF, vWF, and P-selectin that can undergo exocytosis under sepsis conditions. The procoagulant phenotype of the endothelium is further characterized by increased production of TF and subsequent extrinsic coagulation pathway activation, decreased production of anticoagulant proteins (TFPI, EPCR, TM, PC), and increased production of PAI-1, which inhibits fibrinolysis. Ang2, angiopoietin 2; ADAM-15; disintegrin and metalloproteinase 15; AJ, adherens junctions; AP-1, activation of activator protein-1; eNOS, endothelial nitric oxide synthase; ESL-1, E-selectin ligand; GJ, gap junctions; gp130, glycoprotein 130; H
2O2, hydrogen peroxide; HO·, hydroxyl radical; ICAM-1, intercellular adhesion molecule 1; IL, interleukin; IL-1R, interleukin-1 receptor; iNOS, inducible nitric oxide synthase; IRF3, interferon regulatory transcription factor-3; NETs, neutrophil extracellular traps; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; ·NO, nitric oxide; ·NO2, nitrogen dioxide; N2O3, dinitrogen trioxide; O2·–, superoxide anion; ONOO−, peroxynitrite; PAI-1, plasminogen activator inhibitor-1; PAMPs, pathogen associated molecular patterns; PC, protein C; PRR, pattern recognition receptor; PSGL-1, P-selectin glycoprotein ligand-1; RNS, reactive nitrogen species; ROS, reactive oxygen species; sIL-6R: soluble interleukin-6 receptor; STAT3, signal transducer and activator of transcription 3; TF, tissue factor; TFPI, tissue factor pathway inhibitor; TJ, tight junctions; TNF-α, tumor necrosis factor alpha; TNF-α-R, TNF-α receptor; VIIIF, VIII factor; VCAM-1, vascular cell adhesion molecule 1; vWF, von Willebrand factor; WPBs, Weibel–Palade bodies.
One of the first signs of endothelial dysfunction in sepsis is endothelial glycocalyx (eGC) deconstruction (33) ( Figure 1B ). The endothelial glycocalyx is a carbohydrate-rich layer that covers the luminal surface of endothelial cells and serves numerous protective roles for the vascular system (34, 35). Specifically, eGC regulates vascular permeability and endothelial response to shear stress and inhibits leucocyte and platelet adhesion as well as coagulation cascade activation and hence microthrombus formation (36). During sepsis, distinctive mechanisms have been identified that promote eGC damage, which subsequently enhances endothelial dysfunction and sepsis complications. In vivo sepsis models suggest that sepsis-induced reduction in eGC thickness leads to increased albumin extravasation and circulating glycocalyx products such as hyaluronic acid and CD44 (37). The underlying mechanism can be explained by a cascade of reactions where disintegrin and metalloproteinase 15 (ADAM15) cleave CD44 of eGC, leading to an increase in plasma levels of soluble CD44 and hyaluronic acid. This is followed by a decrease in endothelial barrier integrity and a disruption of vascular endothelial (VE)-cadherin and β-catenin, which are essential components of adherens cell junctions (37). Similarly, endothelial tight and gap junctions were found to be disrupted in sepsis (38). A variety of eGC products are emerging as biomarkers in sepsis as they add important prognostic value. For example, increased syndecan-1 (SYN-1) plasma levels during sepsis are correlated with the requirement for renal replacement therapy, respiratory failure, multiple organ dysfunction syndrome (MODS), and predict coagulation failure and mortality (39–41). Sphingosine-1-phosphate (S1P) is another eGC component that predicts sepsis coagulation failure and mortality (40), while increased plasma endocan levels, another marker of eGC destruction, are also associated with the need for mechanical ventilation and with higher mortality in sepsis patients (42, 43). Angiopoietin-2 (Ang-2) is a growth factor produced by the endothelium under inflammatory conditions. It has been linked to increased vascular permeability in sepsis through enhancing endothelial heparanase secretion, which in turn cleaves heparan sulfate, the main component of eGC (34). Increased heparanase production by Ang-2 was shown to be mediated by the inhibition of endothelium-stabilizing receptor Tie2 signaling (34). Additionally, Tie2 tyrosine kinase inhibition by Ang2 in sepsis resulted in a procoagulant phenotype of endothelial cells (44). Interestingly, increased Ang-2 has been associated with higher mortality, respiratory, liver, and kidney failure, and predicted DIC severity in patients with sepsis (44–47).
Nitric oxide (NO) is a soluble gas produced by endothelial cells and studied in depth for its protective role on the vascular endothelium. NO promotes vasodilation and is capable of inhibiting platelet activation and adherence to the endothelium (24). NO has another crucial role in impeding leucocyte adhesion on the vascular wall (24). This is possibly mediated through the downregulation of P-selectin, chemokine, E-selectin, ICAM-1, and VCAM-1 expression in endothelial cells (24). In preclinical sepsis models, an imbalance between endothelial nitric oxide synthase (eNOS) and inducible nitric oxide synthase (iNOS) activity has been demonstrated. Specifically, eNOS activity was found to be decreased compared to the upregulated iNOS activity. Overproduced NO by iNOS reacts with superoxide anion resulting in reactive nitrogen species (RNS) production, which contributes to endothelial dysfunction (38) ( Figure 1C ). Experimental and clinical studies are inconclusive regarding NO levels and their effects on the endothelium during sepsis. Some studies suggest that NO production increases during sepsis and this is associated with poor prognosis, while others support that NO levels are lower in sepsis patients compared to healthy controls (48–51). New evidence suggests that increased NO bioavailability in sepsis may be beneficial for the host without deteriorating vasodilation and subsequent hypotension, as has been proposed in previous studies (52). Overall, the dysregulation of the NO pathway during sepsis merits further investigation.
Expression of proadhesive molecules in activated endothelial cells poses a hallmark process for the progression of sepsis-induced endothelial dysfunction ( Figure 1D ). Activated endothelium interacts both with leukocytes and platelets, while platelet–leucocyte aggregates deposit on endothelium and further enhance local inflammation (23). The infection triggers immune and endothelial cells to produce reactive oxygen species (ROS) and inflammatory and anti-inflammatory cytokines. Reversely, cytokines stimulate endothelial cells to express adhesion molecules such as selectins, integrins, ICAMs, and VCAM-1 (38). These molecules interact with leukocytes, which roll, bind to the endothelium, and transmigrate through it. Furthermore, ROS released by immune cells during sepsis destroy the endothelium and result in increased permeability (38).
Accumulating evidence suggests that endothelial cells can interact with immune cells, triggering thrombus formation, a process implicated in the pathogenesis of numerous diseases, such as antiphospholipid syndrome, coronary artery disease, and sepsis (53–58). Both experimental and clinical studies, presented in Table 1 , explain how endothelial cells can lose their anticoagulant properties and lead to thrombosis when stimulated directly or indirectly by pathogens ( Figure 1E ). A large body of evidence suggests that upon sepsis-induced endothelium activation, there is increased TF expression and secretion accompanied by enhanced TF activity (26, 60, 62–64, 72). Complement, especially C5a, and the inactive terminal C complex (iTCC) formed by mixing C5b6, C7, C8, and C9, were also found to stimulate endothelial cells to produce active TF (77, 78) ( Table 2 ). Beyond TF, other important mechanisms contributing to immunothrombosis are the dysfunctional fibrinolytic and anticoagulant properties of the endothelium. Accumulating experimental data confirm that pathogen invasion is associated with increased endothelial expression of PAI-1 and decreased expression of EPCR, TM, and TFPI, as well as decreased protein C activation (25, 63, 68, 69, 72). Complement C3 activation stimulated by Shiga toxin was also shown to result in endothelial TM reduction and enhanced thrombus formation (82) ( Table 2 ). Despite decreased endothelial TM expression during sepsis, soluble TM is increased in patients with sepsis and predicts acute kidney injury, multiple organ failure, and mortality (96–99). A possible illustrative mechanism is that apoptotic endothelial cells release their intracellular proteins into circulation. Similarly, elevated TFPI plasma levels are associated with overt DIC in patients with sepsis, in contrast to the reduced TFPI expression by endothelium (100). Accordingly, increased plasma levels of PAI-1 have been correlated with DIC in patients with sepsis and predicted mortality (101). Circulating levels of proteins C and S have also been evaluated, and clinical data support their deficiency in sepsis (102). Importantly, in patients with severe sepsis, decreased plasma levels of protein C and increased syndecan-1, which characterize endothelial damage, were associated with whole blood hypocoaguability in the context of DIC, as measured by thromboelastography (103).
Endothelial cell activation, following pathogen invasion, is mediated by intracellular signaling, which includes transcription factor activation such as nuclear factor kappa B (NFκB), activator protein-1 (AP-1), interferon regulatory transcription factor-3 (IRF3), and signal transducer and activator of transcription 3 (STAT3) (38, 104) ( Figure 1A ). Specifically, activation of endothelial NFκB by pathogens has been found to mediate increased expression of procoagulant molecules such as PAI-1 and TF, while fibrinolytic and anticoagulant factors such as TM, APC, and EPCR diminished in a way that was also dependent on NFκB (72, 73). Indeed, NFκB has been associated with increased circulating d-dimers and tissue fibrin deposition in an endotoxemia mouse model (72).
Weibel–Palade bodies (WPBs) are organelles inside endothelial cells containing procoagulant proteins such as vWF, P-selectin, and factor VIII (FVIII), which can rapidly undergo exocytosis under specific circumstances (33). Lipopolysaccharide (LPS) has been shown to induce WPBs’ content secretion both in vitro and in vivo (33) ( Figure 1E ). Ultra-long vWF strings can be produced in vitro after endothelial cell activation by Shiga toxin. Furthermore, the Shiga toxin prevented cleavage of these vWF strings by inactivating ADAMTS13 (105). It has been found that in preclinical sepsis models, increased markers of endothelial activation are associated with abnormal coagulation markers, while some studies have additionally demonstrated the subsequent thrombus formation in the microcirculation. Dysregulated coagulation was indicated by markers such as high levels of circulating d-dimers, fibrin, fibrin degradation products (FDPs), thrombin-antithrombin (TAT) complexes prolonged prothrombin time (PT), activated partial thromboplastin time (aPTT), and/or decreased levels of plasma fibrinogen and platelets (44, 71, 72).
Apart from endothelial dysfunction, which facilitates immunothrombosis in sepsis, activated cells of the innate immune system, in particular neutrophils and monocytes, also play a significant role ( Figure 2 ). Neutrophils are an essential part of the host’s innate immune response and are typically the first responders to acute inflammation (28). In sepsis, PAMPs and damage-associated molecular patterns (DAMPs) can activate neutrophils by binding to PRRs. Activated neutrophils exert their antimicrobial activity mainly through three phagocytosis, degranulation, and the release of NETs (28).
Figure 2 Cell-specific immunothrombosis mechanisms. Neutrophils and monocytes can induce thrombosis through specific mechanisms, while pathogens can directly activate platelets, leading to thrombosis. NETs, neutrophil extracellular traps. ↑ (arrow pointing up) means increased. ↓ (arrow pointing down) means decreased.
A variety of different pathogens, bacterial toxins, and cytokines can induce NETosis (28, 106). NETosis can also be enhanced by activated platelets (28, 60, 107–109). More specifically, in a sepsis mouse model, it was found that LPS binding to platelet TLR4 stimulated platelets to interact with neutrophils and induce NET production in the liver sinusoids and capillaries (107). The binding of platelet P-selectin to P-selectin glycoprotein ligand 1 (PGSL1) on neutrophils was found to be a key mechanism for NETosis induction in mice (110). However, this finding was not reproduced in studies with human neutrophils, and different agonists, where an antibody against P-selectin did not affect platelet-mediated NETosis (111, 112). The interaction of platelet-derived high mobility group protein B1 (HMGB1) with the receptor for advanced glycation end products (RAGE) on neutrophils was also found to be a potential mechanism for NETosis stimulation by platelets, possibly through induction of autophagy in neutrophils (112, 113). Moreover, it was shown that LPS and Pam3CSK4, a synthetic lipopeptide, through binding to platelet TLR4 and TLR2, respectively, activated platelets to interact with neutrophils and induce NETosis through platelet GP1b-neutrophil β2-integrin (CD18) interaction. Platelet-derived vWF, platelet factor 4 (PF4), and thromboxane A2 (TXA2) were essential mediators of this process (111). Furthermore, experimental studies have shown that platelet-derived CXCL4/CCL5 heterodimers and β-defensin 1 enhanced NET production (114). Besides platelet–neutrophil interaction and subsequent NETosis induction, it was found that also complement cleavage products, in particular, C3a and C5a anaphylatoxins could stimulate NETosis in sepsis animal models or in neutrophils stimulated by LPS or platelet-rich plasma from COVID-19 patients (60, 80, 115–117) ( Table 2 ). Moreover, complement may play a role in the inhibition of NET degradation (118).
It has been shown that NETs, apart from their beneficial role in clearing the invading microorganism, are implicated in the pathogenesis of many diseases in which endothelial dysfunction is present, such as sepsis and cardiovascular diseases. Therefore, there has been a growing interest in studying the role of NETs in endothelial dysfunction in sepsis (119). In vitro studies suggest that NETs can promote endothelial dysfunction, as indicated by increased endothelial expression of ICAM-1, VCAM-1, E-selectin, TF, and vWF (60, 62, 109, 120–123). Cathepsin G is a component of NETs, capable of processing pro-IL-1α into IL-1α, which has been found to enhance VCAM-1 and ICAM-1 expression on the endothelial surface (62). Moreover, NET-induced ICAM-1 and E-selectin expression on endothelial cells was found to be dependent on the triggering receptor expressed on myeloid cell-1 (TREM-1). TREM-1 is an orphan immune receptor expressed on various types of cells, such as immune cells, platelets, and endothelial cells. Apart from mediating the NET-induced proadhesive phenotype of endothelial cells, TREM-1 potentiated NETosis triggered by LPS (121). NETs have also been associated with increased platelet adhesion on endothelial cells through enhancement of vWF expression on the endothelium (120). Both NETs and histones have been found to increase TF mRNA and TF activity in human aortic endothelial cells (HAECs) and human saphenous vein endothelial cells (HSVECs), resulting in plasma clotting acceleration (60, 62, 123). Specifically for histones, their effect on TF was mediated by TLR2 or TLR4, which in turn activated AP-1 and NFκB transcription factors (123). As one of the major components of NETs, histones have been further studied for their impact on endothelium, with results suggesting cytotoxicity in a concentration-dependent manner (124). Citrullinated histone H3 (CitH3) was found to increase endothelium permeability in vitro (125). Furthermore, images using intravital microscopy showed that CitH3 resulted in microvascular endothelial barrier dysfunction, by disrupting the adherens junctions and reorganizing the actin cytoskeleton without causing cell death (126). Accordingly, NETs have been found to decrease endothelial expression of VE-cadherin and Zonula occludens (ZO)-1 proteins, which are major components of adherens and tight junctions, respectively (109).
NETs not only contribute to endothelial dysfunction but have also been found to exert procoagulant and prothrombotic activity through various mechanisms, as shown in Table 3 . NETs could serve as a scaffold and inducer of platelet adhesion, activation, and aggregation, and this effect was largely mediated by specific NET components like histones H3 and H4 (134). Nets not only served as a scaffold for platelet adhesion but also for red blood cells (RBCs), promoting the generation of red (RBC-rich) thrombus. Plasma proteins, implicated in thrombosis, such as vWF, fibronectin, and fibrinogen could also bind to NETs, while NETs contributed to thrombin-mediated fibrin generation (134). Histones, in particular H3 and H4, could activate platelets to promote plasma thrombin generation, through binding to platelet TLR2 and TLR4 (131). Notably, this effect was enhanced when histones were in complex with DNA (131). In contrast, another study showed that the addition of deoxyribonuclease (DNase), which destroys the DNA network of NETs, enhanced histone-mediated, platelet-dependent thrombin generation, presumably due to increased exposure to histones (128). Besides NETs-induced platelet-dependent mechanisms of thrombin formation, NETs were capable of stimulating thrombin generation also in platelet-poor plasma via activation of the intrinsic coagulation pathway (128). This effect was largely attenuated with DNase administration, suggesting the crucial role of cell-free DNA (cfDNA) in that process. Accordingly, DNA–histone complexes measured in platelet poor plasma from sepsis patients were increased compared to plasma from healthy subjects, and cfDNA was positively correlated with thrombin generation (128). NETs, possibly due to their negatively charged extracellular DNA surface, could bind and promote activation of factor XII, thus triggering the contact pathway of coagulation (146). Interaction of NETs with membrane-derived microparticles released by activated neutrophils enhanced NET-mediated intrinsic coagulation pathway activation and subsequent thrombin formation (139). In support, it was demonstrated that NETs were capable of promoting pulmonary clot formation in mice with septicemia caused by LPS and heat-killed E. coli (140). However, despite robust evidence on the effect of NET components on thrombosis, the role of intact NETs may merit further investigation. Specifically, an in vitro study showed that although cfDNA and histones, separately, were capable of coagulation activation, intact NETs could not (147). One possible explanation for this is that the complex interactions between DNA and histones within the NET structure abrogate the thrombogenic negative charge of DNA and cover the histones’ binding sites. Moreover, the absence of other blood cells, endothelium, and blood flow in an in vitro model compared to an in vivo one may have contributed to the absence of the procoagulant effect of intact NETs (147).
NETs were also shown to interfere with the endogenous anticoagulant mechanisms. More specifically, extracellular nucleosomes within NETs facilitated TFPI degradation by neutrophil elastase on the surface of activated neutrophils, thus abrogating TFPI-mediated inhibition of of the extrinsic coagulation pathway (141). Neutrophil elastase bound to DNA complexes was also shown to cleave plasminogen into fragments, resulting in decreased plasmin production and impaired fibrinolysis (129). Apart from decreased plasmin generation, cfDNA was capable of binding to plasmin and fibrin at the same time, and the formed complex was shown to be responsible for decreased plasmin-mediated fibrin clot dissolution (133). Besides cfDNA, histones, specifically H3 and H4, could also interact with TM and protein C, leading to the inhibition of APC generation (132).
Therefore, experimental data support the crucial role of NETs in triggering thrombus formation. From a clinical point of view, it was shown that increased NET-forming capacity was significantly associated with thrombocytopenia, increased PT, aPTT, d-dimers, and low fibrinogen in patients with sepsis at intensive care unit (ICU) admission, while it could also predict DIC development and mortality after ICU admission (106). Noteworthy, specific components of NETs, such as circulating levels of cfDNA, neutrophil elastase, and the myeloperoxidase–DNA complex in patients with sepsis, are positively correlated with the risk of VTE (130). Increased MPO-DNA levels on days 3 and 7 during sepsis hospitalization were associated with decreased mean arterial pressure, a lower PaO2/FIO2 ratio, an increased sepsis-related organ failure assessment (SOFA) score, and 28-day mortality (148).
Regarding the production and release of TF by neutrophils, there is conflicting evidence. It was reported that neutrophils in the setting of appropriate interaction with injured endothelial cells or after stimulation with P-selectin or the peptide formyl-MetLeuPhe (Fmlp) were able to produce functional TF, initiating the extrinsic coagulation pathway (149, 150). Moreover, C3a and C5a anaphylatoxins were found to stimulate neutrophils to produce and release active TF ( Table 2 ) (60, 83, 90). However, these findings have been questioned in other studies, and it was proposed that granulocytes may acquire TF produced by monocytes (151, 152). With respect to the presence of TF in NETs, it was found that neutrophils from patients with sepsis could release large amounts of functional TF-bearing NETs, and autophagy was implicated in that process (151).
Another potential mechanism through which neutrophils may contribute to immunothrombosis is by secreting antimicrobial peptides. Specifically, α-defensins, which are secreted by neutrophils as part of the host innate immune response, have been found elevated in the plasma of septic patients (153). Recently, an in vivo study showed that α-defensins accelerate and stabilize fibrin clot formation, while also disrupting fibrinolysis (154). Consistently, circulating α-defensins have been correlated with the acceleration of clot formation and d-dimers plasma levels in COVID-19 patients (143, 155).
Monocytes and macrophages have been found to play a vital role in sepsis-induced immunothrombosis ( Figure 2 ). TF expression by monocytes seems to be the most potent pathway leading to coagulation cascade activation in sepsis (156–160). While both hematopoietic and non-hematopoietic cells produce TF and enhance coagulation in an endotoxemia mouse model, monocytes are the main source of circulating TF (161). Additionally, other blood cells have been found to acquire TF released by monocytes through microparticles (161). Numerous studies depicted in Table 4 suggest that pathogens induce monocyte-derived TF expression and release in the circulation, which in turn activates the extrinsic coagulation pathway. The binding of LPS to transmembrane receptors such as TLR4 in monocytes induces TF mRNA expression via NF-κB activation (166). Moreover, the interaction of pathogen components either with TLRs or directly with intracellular pathways in monocytes can result in inflammasome activation and subsequent TF release via pyroptosis (163, 166). Pyroptosis is a type of proinflammatory cell death induced by caspase-1 (canonical inflammasome) or caspase-11 (noncanonical inflammasome) and is characterized by gasdermin D-dependent pore formation in the cell membrane, subsequent osmotic cell lysis, and finally the release of cytosolic content. Indeed, LPS or type III secretion system (T3SS) rod proteins have been found to trigger in vivo caspase-11 and caspase-1 activation, respectively, leading to TF release (163). Gasdermin D-mediated pore formation induced by caspase-11 has also been proposed to activate TF on macrophages (162). Specifically, pore formation on the cell membrane can induce calcium influx, which triggers phosphatidylserine exposure on the membrane, followed by TF activation (162). Furthermore, sphingomyelin, another membrane lipid, is involved in the activation of TF to its procoagulant form. This process occurs on the macrophage cell membrane and is dependent on sphingomyelin hydrolysis by acid sphingomyelinase (ASMase) (167).
Tissue factor production is additionally triggered by HMGB1, which is a DAMP released by myeloid cells, platelets, and hepatocytes. HMGB1 binds LPS, enhances endocytosis of the LPS-HMGB1 complex in monocytes, and is associated with increased TF release via caspase-11-induced pyroptosis (168). Furthermore, TF expression and activity are also induced by the interaction of monocytes or their microparticles with activated platelets (158, 164). Monocyte activation by pathogens is followed by increased PSGL-1 expression and the release of TF- and PSGL-1-bearing microparticles. These microparticles can fuse in vitro with platelets, leading to increased TF activity (158). Additionally, a prospective clinical study confirms that LPS endotoxemia enhances platelet–monocyte aggregation, TF expression, and platelet activation (165). On the other hand, thrombin generated by coagulation cascade activation binds to monocytes via protease-activated receptor-1 (PAR-1), leading to increased TF expression. Accordingly, inhibition of PAR-1 by vorapaxar (a PAR-1 antagonist) has been found to attenuate sepsis-induced coagulation, as indicated by decreased plasma prothrombin fragments and TAT complexes (169).
Accumulating evidence suggests a complex interaction between platelets and different immune cells. Increased platelet activation and platelet–leukocyte aggregates have been found in a variety of thrombotic conditions, such as unstable angina, myocardial infarction, stroke, and sepsis (170–173). Multiple roles of platelets have been demonstrated in experimental and clinical studies. Platelets were suggested to contribute to tissue regeneration and myocardial recovery following acute coronary syndrome (ACS) (174–178). On the contrary, increased platelet activation and reactivity have been linked with a poor prognosis in several diseases, including ACS (179), dementia (180), diabetes (181, 182), and sepsis (183). Specifically, in sepsis, thrombocytopenia is associated with increased disease severity and an ominous prognosis (184). Decreased platelet count is attributed to excessive peripheral consumption of platelets resulting from many causes, among which are the interaction of platelets with immune cells, the microcirculation thrombosis, and the direct binding of pathogens to platelets (184). Beyond innate immune cells, which play a crucial role in immunothrombosis, pathogens can directly activate platelets, which in turn promote thrombosis (
Figure 2
). Recently, it has been found that during sepsis there is accelerated megakaryocyte maturation, triggered by PAMPs, DAMPs, TNF-α, or other cytokines. This subpopulation of megakaryocytes presents a stronger immunological profile and produces fewer platelets compared to megakaryocytes undergoing normal maturation induced by thrombopoietin (185). A variety of receptors on platelets interact with PAMPs, or pathogens coated by antibodies. Numerous pathogens have been proven to activate platelets and increase their aggregation, as presented in
Table 5
. Pathogens coated by immunoglobulins or pentraxins bind to low-affinity receptors for the constant fragment (Fc) of immunoglobulin (Ig) G (FcγRIIa) on platelets, leading to dense and α-granules secretion. FcγRIIa triggers an intracellular pathway for platelet activation by phosphorylating the tyrosine kinases Src, Syk, and phospholipase c gamma 2 (PLCγ2) (188). Glycoproteins IIb/IIIa (GP IIb/IIIa) and Iba (GPIba), expressed on the surface of activated platelets (202), constitute another receptor category that promotes platelet degranulation and aggregation after binding pathogen components (180, 188, 189, 202). Bacteria have also been found to activate platelets through TLR-2 and TLR-4 (192, 194, 203). Specifically, intracellular signaling of TLR-2 is mediated by PI3 kinase, which activates the Rap1 guanine nucleotide (192). TLR-4 activation in platelets is followed by a complicated intracellular signaling pathway that implicates NFκB. The majority of NFκB signaling proteins are present in platelets, presenting nongenomic effects and resulting in degranulation (203, 204). Moreover, LPS-induced platelet aggregation in vitro is dependent on both TLR-4 and TLR-2 platelet receptors as well as on myeloid differentiation factor 88 (MyD88) and cGMP-dependent protein kinase intracellular pathways. Accordingly, LPS accelerated in vivo thrombus formation and occlusion in FeCl3-injured carotid artery of mice, and this effect was dependent on the MyD88 signaling pathway (194).
All the aforementioned platelet receptors induce immunothrombosis through different pathways after being triggered by PAMPs. The surface of activated platelets promotes thrombin generation and, subsequently, fibrin and clot formation (205). Both dense and α-granules are secreted by pathogen-activated platelets and enhance their aggregation. Specific proteins secreted by platelets are involved in the process of immunothrombosis, such as P-selectin, PF4, vWF from α-granules, and ADP/ATP from dense granules (188). Activated platelets have also been found to release polyphosphate (PolyP), an inorganic polymer that exerts procoagulant activity. In vitro, PolyP initiates the contact pathway by FXII activation, resulting in enhanced thrombin generation and clot formation (206). The procoagulant activity of PolyP has also been confirmed in a sepsis mouse model where LPS-induced Polyp is released by platelets, resulting in increased intravascular thrombin activity (108). Protein disulfide isomerase (PDI) is found in the platelet endoplasmic reticulum and is released into circulation by endothelial cells or platelets (157, 207). The pathophysiological mechanism explaining PDI thrombotic effect in vivo is not fully elucidated. Activated platelets increase the secretion of PDI, which binds to the GPIIb/IIIa integrin and results in enhanced GPIIb/IIIa-binding affinity. An in vivo study suggests that TNF-α activated endothelium induces PDI secretion leading to platelet-neutrophil aggregation through GPIIb/IIIa and αΜβ2 receptors (207). Another possible mechanism has been proposed by an in vitro study finding that PDI enhances TF-mediated thrombin generation in peripheral blood mononuclear cells (PBMCs) triggered by LPS (157). Platelet-derived TF levels and activity have also been found to increase in vitro after activation by LPS, E. coli, or S. aureus α-toxin. TLR-4 is one of the platelet receptors mediating TF production by splicing of TF pre-mRNA (187, 191). Indeed, a clinical study confirms that platelets from septic patients present increased TF activity in contrast to healthy ones (191). Furthermore, increased expression of P-selectin and GPIIb/IIIa has been found in a sepsis mouse model and is associated with neutrophil-platelet aggregates and thrombus formation (173). Similarly, a prospective clinical trial found that COVID-19 infection amplifies platelet activation, as indicated by P-selectin, and promotes platelet aggregation (201).
The efficacy of FDA-approved drugs commonly used in clinical practice, such as heparin, P2Y12, and GPIIb/IIIa inhibitors, has been tested in patients with sepsis. AT, another FDA-approved drug indicated for the prevention of peri-operative and peripartum thromboembolic events in patients with hereditary AT deficiency, has been also assessed ( Table 6 ; Figure 3 ) (239). In addition to FDA-approved drugs, a variety of currently non-FDA-approved agents, including recombinant TM, APC, and TFPI, have been evaluated in sepsis ( Table 7 ; Figure 3 ). According to the international guidelines for sepsis and septic shock, neither anticoagulant nor antiplatelet medications are recommended for sepsis or DIC treatment, except for low molecular weight heparin (LMWH) or unfractionated heparin (UFH) for VTE prevention (13). On the contrary, AT replacement therapy and recombinant TM are recommended for sepsis-associated DIC in Japan (274). The efficacy of AT in sepsis is controversial, as derived from studies depicted in Table 6 and meta-analyses (275). Decreased AT plasma levels and DIC presence have been correlated with higher efficacy of AT treatment (221, 234, 275). Furthermore, bleeding risk in patients receiving AT is increased, especially in co-administration with heparin (275). Studies assessing the efficacy of TM have shown more promising results, as TM administration resulted in decreased mortality and was not associated with increased bleeding events in sepsis-induced coagulopathy (276, 277). APC (or drotrecogin alfa activated), despite its antithrombotic effects, is not used in clinical practice as it has not improved survival and has been associated with increased bleeding risk in sepsis (271, 278). Similarly, heparin is not recommended in sepsis management, although some studies have found a beneficial effect on 28-day mortality (279–281). Bleeding events following heparin treatment in sepsis are an important disadvantage of its use (279). Antiplatelet agents, beyond aspirin, have been tested to a lesser extent than anticoagulants, resulting in scarce data regarding their effects on the survival of sepsis patients (215–217). As for aspirin, a recent study showed no effect on mortality in septic patients (282). However, there are multiple underlying mechanisms through which aspirin could affect sepsis, specifically through its known anti-inflammatory properties. As a result, aspirin’s effects on sepsis should not be attributed exclusively to its antithrombotic properties (282). Therefore, drugs targeting immunothrombosis are not approved for clinical use in sepsis due to both inconclusive results for their efficacy and an increased risk for side effects.
Figure 3 Drugs targeting sepsis-related immunothrombosis in clinical trials. Both FDA-approved and unapproved drugs have been studied in clinical trials for their effect on sepsis-related immunothrombosis. The drugs depicted in the figure showed conflicting results regarding their effect on mortality rates and coagulation parameters in sepsis patients. DNase is the only one with no clinical results yet, as the phase 1 study (NCT05453695) using DNase intervention has recently started. CD14, cluster of differentiation 14; DNase, deoxyribonuclease; GPIIb/IIIa, glycoprotein IIb/IIIa; IL, interleukin; mAb, monoclonal antibody; PAR-1, protease-activated receptor-1.
The general failure of clinical trials to identify pharmaceutical agents with an acceptable risk-to-benefit ratio poses the need for discovering new molecules that could mitigate the deleterious effects of immunothrombosis in sepsis. A variety of such molecules have been examined in sepsis animal models, showing promising results, as depicted in Table 8 and Figure 4 . Administration of histidine-rich glycoprotein (HRG), a plasma protein found to be significantly depleted in septic conditions, attenuated immunothrombi formation in the lungs of septic mice (294). Reduced neutrophil adherence to endothelium due to shape changes and decreased NET generation, as well as reduced endothelial cell activation, all mediated by HRG, were suggested to be among the potential mechanisms responsible for this effect (294). Another study investigated the beneficial role of Annexin A1, an anti-inflammatory protein found in low levels in patients with sepsis, on LPS-induced thrombosis (287, 299). Interestingly, Annexin A1 mimetic peptide attenuated platelet aggregation and resulted in delayed and reduced microthrombi formation in the cerebral microcirculation (287).
Figure 4 Potential therapeutic targets under preclinical investigation in sepsis. Inhibitors of innate immune components, like neutrophil extracellular traps and complement components C5 and C3, drugs interfering with tissue factor formation and procoagulant activity as well as molecules with multiple actions, have shown promising results in reducing immunothrombosis in preclinical sepsis models. However, the safety, tolerability, and effectiveness of these drugs in abrogating the deleterious effects of immunothrombosis in patients with sepsis are still unknown. C5, complement 5; C3, complement 3; CD14, cluster of differentiation 14; DNase, deoxyribonuclease; HMGB1, high mobility group box 1; IFN, interferon; mAb, monocloncal antibody; NETs, neutrophil extracellular traps; NLRP3, NLR family pyrin domain containing 3; PAD4, protein arginine deiminase; SQSTM1, sequestosome 1; TF, tissue factor.
A variety of inhibitory molecules targeting pathways involved in TF expression in immune or endothelial cells or in TF procoagulant activity have also been proposed. Disrupting TLR4-mediated LPS signaling through cluster of differentiation 14 (CD14) inhibition in immune cells was one potential pathway to target. Specifically, CD14 is a protein responsible for LPS transfer to its cell surface receptor complex, TLR4/myeloid differentiation factor 2 (MD-2) (a co-receptor of TLR4) (285). Inhibitory monoclonal antibody (mAb) against CD14 led to decreased activation of the extrinsic coagulation pathway through suppression of TF expression in leukocytes while also enhancing fibrinolysis through decreasing plasma PAI-1 (285). Apart from anti-CD14 mAb, lidocaine could also inhibit LPS-induced TF expression in monocytes through upregulation of suppressor of cytokine signaling 3 (SOCS3), which in turn downregulated the TLR4-TF pathway (286). Another potential pathway that could serve as a therapeutic target for mitigating immunothrombosis was shown to be type 1 interferon (IFN-1)-induced upregulation of TF procoagulant activity. Type 1 IFNs are cytokines that increase in response to bacterial infection and are characterized by antimicrobial properties. Inhibition of the expression of IFN-1 receptors or downstream effectors (e.g., HMGB1) in macrophages reduced gram-negative bacteria-induced DIC (289). TF procoagulant activity was also found to be decreased by acid sphingomyelinase inhibitors. In response to LPS, acid sphingomyelinase translocates to the cell membrane and hydrolyzes sphingomyelin, which, in resting conditions, maintains TF in an encrypted form with minimal procoagulant activity. Therefore, inhibition of acid sphingomyelinase inhibitors could maintain sphingomyelin levels on cell membranes despite LPS stimulation, resulting in decreased TF activity and reduced TAT levels in a sepsis mouse model (167). Linagliptin, a dipeptidyl peptidase-4 inhibitor used for diabetes treatment, was also found to ameliorate LPS-induced pulmonary thrombosis in mice via decreased endothelial TF expression. This effect was shown to be mediated through linagliptin-induced activation of the Akt/eNOS pathway and subsequent NO production (59). Noteworthy, linagliptin or the glucagon-like peptide-1 (GLP-1) analog, liraglutide, were shown to have further inhibitory effects on platelet thrombin generation and platelet aggregation via the cyclic AMP (cAMP)-protein kinase A (PKA) signaling pathway (293). Among other clinically approved drugs, amitriptyline, a tricyclic antidepressant, was also capable of reducing thrombosis through TNF-α level reduction (292).
The NLRP3 inflammasome, a complex responsible for IL-1β and IL-18 release and a crucial mediator of pyroptosis, became activated in platelets in a sepsis rat model and correlated with lung and renal injury (300, 301). Blockage of NLRP3, using a selective inhibitor, led to decreased platelet activation and offered protection from organ dysfunction in rats with sepsis (288). Another potential therapeutic target was shown to be SQSTM1, a stress response protein that was released from monocytes and macrophages upon LPS stimulation (290). Interestingly, an inhibitory antibody against SQSTM1 attenuated DIC in bacteria-induced sepsis (290).
Promising results have also been demonstrated in preclinical studies examining the effect of NET inhibition on sepsis-induced thrombosis (108, 138, 296). DNase is an enzyme capable of digesting DNA. Two types of DNases have been DNase 1, which degrades free DNA, and DNase 1-like 3 (DNase1L3), which degrades DNA bound to proteins, such as histones (140). Patients with sepsis were shown to have decreased levels of DNase 1 compared to healthy controls (302). Moreover, mice with a combined deficiency of DNase1 and DNase1L3 demonstrated increased thrombosis in pulmonary vessels compared to mice capable of producing at least one of these two enzymes when both were infected either with LPS or heat-killed E. coli (140). Exogenous DNase administration promoted NET degradation and resulted in reduced intravascular thrombin activity as well as decreased microthrombus formation in liver sinusoids (108). The timing of the DNase infusion seems to affect its beneficial effect, with delayed infusion resulting in a significant decrease in TAT complexes compared to the insignificant effect observed with early infusion (296). Importantly, DNase administration, aiming to decrease LPS-induced NETosis, prior to S. aureus infection, did not negatively affect pathogen capture and clearance (138). However, the impact of DNase infusion after bacterial infection was not examined in that study. In contrast, mice treated with DNase at seven consecutive time points, beginning 1 h after cecal ligation and puncture (CLP), showed increased bacterial dissemination and higher serum IL-6 levels 6 h after CLP and higher mortality at 24 h (303). However, it should be noted that pathogens’ spread and levels of systemic inflammation 24 h after CLP, as well as overall survival, did not increase in DNase-treated mice compared to controls (303). Besides the evaluation of DNase treatment in preclinical sepsis models, a clinical trial (NCT05453695) is going to assess intravenous DNase safety and tolerability as well as its effect on organ dysfunction, duration of ICU hospitalization, mortality, and blood coagulation in patients with sepsis (304). Regarding inhibition of NET formation, reparixin, an inhibitor of interleukin-8 CXC chemokine receptor type 1/2 (CXCR1/2), resulted in decreased NET formation and subsequently reduced fibrin deposition in the lungs of septic mice (284). Similarly, infusion of platelets pretreated with stachydrine, an herbal product, attenuated NET formation, platelet–neutrophil interaction, and subsequent thrombosis (136).
Complement activation, despite being an essential component of the innate immune response, is associated with increased mortality in sepsis (305, 306), while it also contributes to immunothrombosis (307, 308). Blockade of C3a or C5a led to decreased DIC in E. coli or (CLP)-induced sepsis animal models (91, 93, 94). Noteworthy, combined inhibition of C5 activation and CD14 attenuated more robustly hypercoagulability in sepsis (295, 297).
Immunothrombosis is a key pathophysiological mechanism of sepsis, arising from the complex interplay between innate immune and endothelial cells on one side and platelets and the coagulation cascade on the other. Although initially beneficial for the host, uncontrolled and systemic activation of this process in sepsis can lead to thrombotic and bleeding complications, ranging from subclinical abnormalities in coagulation tests to severe clinical manifestations, such as DIC. Endothelial dysfunction, characterized by glycocalyx degradation, increased vascular permeability, as well as proinflammatory and procoagulant properties of endothelial cells, further facilitates immunothrombosis progression. Despite the high burden of coagulation disorders that have a negative impact on sepsis patients’ prognosis, there is no effective treatment aiming to target the immunothrombotic process. Although a variety of anticoagulants and antiplatelet drugs have been evaluated in clinical trials, their beneficial effects are inconsistent, and they are also characterized by a high rate of bleeding complications. However, there is still ongoing research assessing the effect of new molecules on thrombosis in sepsis, with agents targeting intracellular inflammatory pathways, NET formation, and complement components demonstrating promising results. Future studies are called to unravel the pathophysiologic mechanisms involved in immunothrombosis in sepsis, to identify potential prognostic biomarkers, to develop risk scores to predict the outcomes of sepsis, and to test novel therapeutics against immunothrombosis in sepsis.
EM and EA researched data for the article, contributed to the interpretation of the published studies, and drafted all display items. KSte developed the concept, discussed the content, jointly conceived the display items of the article, and edited the article. ST-C, BEV, AG, and KSta provided conceptual advice, participated in the interpretation of the data, and edited the manuscript. All authors contributed to the article and approved the submitted version.
The authors’ work is supported by the Biotechnology and Biological Sciences Research Council (BBSRC) of the UK Research and Innovation (UKRI) to KSte and AG, by a Wellcome Trust Institutional Strategic Support Fund to ST-C, by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (MODVASC, grant agreement No. 759248), and by the German Research Foundation DFG (SFB1366, project number 394046768) to KSte. This work is supported by the Health + Life Science Alliance Heidelberg Mannheim and received state funds approved by the State Parliament of Baden-Württemberg.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.