The heterogeneity of EVs and their different biological sources (cell cultures, blood, body fluids etc.) resulted in a corresponding variety of methods to isolate, enrich, and analyze them (see Table 1). The most frequently used approach for their isolation and enrichment has been and still is sequential centrifugation. Problems with this approach arise from non-standardized rotor types and centrifugation times, resulting in differences in yield and purity of EVs (Cvjetkovic et al., 2014). Ultracentrifugation in itself can lead to aggregation of EVs (Linares et al., 2015), but it is also evident that the type of rotor (fixed-angle or swing-out) can have a significant influence. In most studies, centrifugation is done with conventional fixed-angle rotors. However, this method has some potential drawbacks, as the horizontal forces press the fragile membrane vesicles against the wall of the centrifugation tube, where they are not only squeezed but also forced along the wall down to the site that is most distant from the rotational axis (Figure 2). Obviously, this can lead to significant mechanical stress and shear forces, which can damage the vesicles in several ways. First, they can become leaky (which we often observed in similar-sized endosomes) – and this can occur in a transient manner, as small pores in membranes have the tendency to re-seal. Thereby, they can lose their content fully or partially. Second, various components of the vesicles might be physically damaged. Third, membranous vesicles can break up completely and even re-seal in an inside-out orientation. In line with that, it has been reported for EVs that membrane proteins can occur in a reverse topology with cytosolic domains exposed to the outside (Cvjetkovic et al., 2016). Such an artificial inside-out orientation of EVs caused by their preparation would expose the inner-leaflet PS to the outside and would lead to a false-positive binding of Annexin V or lactadherin to the vesicles. The problem of the shear forces can be reduced by using swing-out rotors. Here the particles move along the axis until they reach the bottom of the tube. Yet, this approach does not avoid mechanical stress completely as the vesicles are still pressed against the bottom. The latter might be reduced by using a liquid high-density cushion, such as a sucrose-solution at the bottom of the tube. Finally, the principle of float-up gradient centrifugation (Burdon et al., 1988) can be applied, where membranous vesicles are applied to a centrifugation tube at a layer of higher density than their inherent buoyant density, which is overlaid with layers of lower density. During centrifugation in a swing-out rotor, the vesicles (but not free proteins or protein aggregates) will float up against the direction of the centrifugation force through the layers of higher density until they reach a layer, which has a lower density than their intrinsic one. There they will accumulate at the interface of the two layers, which can even have an enrichment effect (Figure 2). Such float-up gradient approaches have also been applied for EVs (Iwai et al., 2016; Li et al., 2018). An advantage of these flotation gradient techniques is that they have a significant purification effect and separate EVs from contaminating protein aggregates and lipoproteins. The pitfalls and issues of classical differential (ultra-) centrifugation using fixed-angle rotors have been described in several reports (Cvjetkovic et al., 2014; Linares et al., 2015; Livshits et al., 2015; Söderström et al., 2016). However, it still seems to be the most widely used method to prepare EV samples. Nevertheless, additional, alternative methods have been established to purify these vesicles with the aim to maintain their integrity. These include ultrafiltration and size-exclusion chromatography (Nordin et al., 2015) and are sometimes combined with polymeric precipitation using polyethylene glycol, PEG (Szatanek et al., 2015). More recently, a method has been reported, which combines PEG precipitation with Nycodenz (iohexol) gradient centrifugation followed by size-exclusion chromatography (Zhang et al., 2020), which seems to solve some of the problems that are associated with the traditional ultracentrifugation technique. Another word of caution has to be expressed with respect to potential influences of buffers, in which EVs are prepared and resuspended for further analysis. It has been demonstrated, for instance, that the number of EVs can increase with time, after the initial sampling, dependent on the buffer in which the samples are stored, with a significant increase of EV numbers in heparin- and citrate-containing buffers as compared to EDTA-buffers (Fendl et al., 2016). Moreover, it could be shown that storage of samples in PBS (in contrast to saline) can result in artefacts by creating EV-like particles, which are also positive for EV-markers in subsequent flow cytometry analyses (Xin et al., 2017).

For reasons of practicability and availability, microvesicles and other larger EVs are often investigated and analyzed by flow cytometry. However, standard cytometry equipment only allows analysis down to about 300 nm diameter particles, which excludes a high number of EVs. Specific instruments can get down to about 100 nm, which would comprise the entire size range that has classically been defined for microvesicles but would still omit most exosomes. For that reason, complementary and alternative methods are in use, such as pelleting EVs with ultracentrifugation at >100,000 g for 1 h or longer (keeping in mind the resulting limitations discussed in section 1.1), followed by biochemical analysis, such as SDS-PAGE and Western Blot testing of known markers of EVs as well as frequent contaminants (like apolipoproteins). Biochemical analysis of EVs can include degradation of presumable intraluminal cargo (like miRNAs) to check for contamination of EV preparations with inside-out vesicles. Similarly, binding of antibodies to extravesicular domains of proteins followed by cytometry or microscopy can be applied to test for correct orientation of EVs. Many groups also use electron microscopy (EM) to check for integrity and purity of EVs, as contaminating lipoprotein particles appear differently in EM. In this case, cryo-EM was reported to outperform transmission-EM, as the latter can often lead to artefacts caused by fixation and contrast agents (Cizmar and Yuana, 2017).

A technique, which found increasing application in the field, is nanoparticle tracking analysis (NTA), which basically measures the Brownian motion of particles detected by scattered light or fluorescence (Saveyn et al., 2010). This technique can determine much smaller vesicles than flow cytometry and has the advantage to measure the concentration of particles, as well as their size distribution. However, it does not differentiate between EVs and lipoprotein particles and therefore has to be combined with additional methods to prove the purity of an EV preparation. Furthermore, results can vary significantly between different machines and labs and optimized instruments settings have to be applied to obtain consistent and robust data (Vestad et al., 2017; Bachurski et al., 2019). Direct comparisons between flow cytometry and NTA for platelet-derived EVs demonstrated that only a smaller fraction of particles detected by NTA in light scatter mode are really membranous vesicles, and that only about 10% of the particles detected in fluorescence mode-NTA were PS-positive, as compared to about 60% for flow cytometry (George et al., 2021). This supports the notion that flow cytometry detects only larger EVs, which have a higher fraction of PS-positive vesicles. An overview of analysis techniques is shown in Table 2.

EVs are commonly characterized by measuring their composition. Dependent on the cellular origin and (patho-)physiological state, EVs carry a multitude of molecules inside or on their surface, including RNA, DNA, proteins, lipids and metabolites. Thorough analysis of their constituents, e.g., by next generation sequencing, Western Blot, ELISA or different omics approaches, can give insight into their pleiotropic responses. In addition, functional assays are applied to characterize EVs, which have been recently reviewed in (Coumans et al., 2017; Théry et al., 2018). A short summary of these assays and references for them are included in Table 3. Their validity is dependent on the use of specific antibodies and the development and use of reference materials and controls as well as standardized collection, handling, and isolation of EVs for data comparison.

Extracellular vesicles play crucial roles in intercellular communication and are released under physiological and pathological conditions from various cell types, including platelets, red blood cells, lymphocytes, dendritic cells, hematopoietic cells, endothelial cells or cancer cells (van Niel et al., 2018).

Cellular sources of EVs in the blood are primarily distinguished based on their surface antigens. CD41 is a common marker for platelet- or megakaryocyte-derived EVs (Flaumenhaft et al., 2009; French et al., 2020). Megakaryocytes in the bone marrow constantly release EVs into the blood stream. Therefore, these EVs dominate in healthy individuals, which is also reflected by the fact that they were initially described as “platelet-dust” (Wolf, 1967). Platelet-derived EVs (PEVs) are only released upon stimulation. They expose the platelet activation markers P-selectin and PS, which are both absent on megakaryocytic EVs. The exact concentration of PEVs in healthy plasma is still controversially discussed with concentrations specified as low as 200 up to 10⁹ EVs/µl (Arraud et al., 2014). This is most likely attributed to low sensitivity of the detection method or false co-detection of contaminants (Arraud et al., 2014; Johnsen et al., 2019). Despite this controversy, it is clear that EVs derived from platelets and megakaryocytes represent the majority in the circulation (Flaumenhaft et al., 2009; Aatonen et al., 2014; Arraud et al., 2014).

Red blood cells (RBCs) are the most abundant cell type in the blood. EVs derived from RBCs participate in physiological blood clotting by activating the intrinsic coagulation pathway via FXIIa and FXIa leading to thrombin generation (Van Der Meijden et al., 2012; Rubin et al., 2013). Conversely, RBC EVs can also interact with protein S and thereby support the anticoagulant reaction mediated by aPC (activated protein C) (Livaja Koshiar et al., 2014). Further, RBC EVs can contribute to dysregulation of hemostasis and demonstrate pro-coagulant effects in several disease states and can interact with the endothelium. EVs released from fresh or stored RBCs have pro-inflammatory properties and are closely related to the immune and inflammatory responses of blood transfusions, where they might contribute to adverse outcomes (Straat et al., 2016; Crawford et al., 2019).

Besides RBCs and platelets, also leukocytes contribute to EVs in the circulation. The main function of leukocyte derived EVs is to induce immune responses, to recognize and eliminate pathogenic or harmful substances, and to participate in inflammatory processes. Neutrophil-derived EVs can exert both pro- and anti-inflammatory effects on neighboring cells (Kolonics et al., 2020; Kolonics et al., 2021), depending on the environmental conditions during EV generation. Cell surface proteins of neutrophil-derived EVs mediate their adherence to monocytes and endothelial cells inducing downstream signaling pathways (Danesh et al., 2018). An example of an anti-inflammatory role of neutrophil-derived EVs is the potential release of TGF-β1, which inhibits the inflammatory response of macrophages to zymosan and LPS (Gasser and Schifferli, 2004). Contrary, EVs released from neutrophils upon bacterial infections in mice promote platelet activation (Letsiou et al., 2021) and enable platelet-mediated innate immune responses by inducing robust inflammation and pathogen clearance (Rossaint et al., 2016).

Endothelial cell-derived EVs (EC-EVs, markers: CD31⁺, CD144⁺) play an important role in intercellular communication under homeostatic and pathological conditions, e.g., under inflammation (Yamamoto et al., 2015), and correlate with disease severity. EC-EVs expressing tissue factor (TF) (Zwicker et al., 2011) exhibit procoagulant activity as shown in vitro and in animal studies (Holnthoner et al., 2017). EC-EVs have recently attracted attention as mediators and potential therapeutic targets especially in respiratory diseases, such as asthma or COPD. They are reviewed in more detail elsewhere (Fujimoto et al., 2020; Desideri et al., 2021).

Platelets are small anucleate cells derived from their precursors, megakaryocytes, in the bone marrow. They are primarily known for their role in hemostasis and vascular integrity. Platelets generate and release EVs upon activation by a variety of agonists. Previously, it was hypothesized that PEVs released from activated platelets or platelet-rich thrombi may reflect developing arterial thrombosis and that therefore, the role of platelets and PEVs in hemostasis may be redundant (Puhm et al., 2021). However, due to their complex set of cargo, PEVs participate in diverse biological processes such as angiogenesis, cancer, cardiovascular diseases, or inflammation. Historically, two different types of platelet-EVs were discriminated according to their size. Platelet-MVs (100–1000 nm in diameter) usually expose PS on their surface (when generated from activated platelets) and express GPIbα, αIIbβ3 and P-selectin (Heijnen et al., 1999). Smaller platelet-exosomes (40–100 nm) expose CD63, do not bind to FX and prothrombin, and are difficult to detect by flow cytometry due to their size (Heijnen et al., 1999). Accordingly, the procoagulant activity of PEVs was mainly ascribed to microvesicles. Mallat and others associated procoagulant activity of PEVs predominantly with TF (Mallat et al., 1999). Although the occurrence of this coagulation factor on platelets is still questioned (Østerud and Bouchard, 2019), it has been proposed that platelets can acquire TF from TF-bearing EVs originating from other cell types through fusion in a PSGL-1-dependent way (Del Conde et al., 2005).

Besides TF, a major procoagulant effect of PEVs is apparently the exposure of negatively charged phospholipids (e.g., PS), which create complex bonds with Ca²⁺ ions, that are further binding the (Vitamin K-dependent) γ-carboxy groups of coagulation factors. In this way, components of the coagulation cascade are concentrated on the surface of negatively charged membranes, which results in a drastic increase in the rate of proteolytic conversion and activation of coagulation factors. Thus, PS-exposing PEVs provide a surface for assembly of tenase and prothrombinase complexes, thereby being directly involved in coagulation (Owens and Mackman, 2011), which implies a causal relationship between increased PEV concentrations and thrombosis. In general, it can be stated that the release of PEVs contributes to the increase in membrane surface area together with cytoskeletal rearrangement and the formation of filopodia upon platelet activation.

The procoagulant activity of PEVs depends on the underlying trigger leading to their release from platelets (Sims et al., 1989). Highly procoagulant PEVs are formed after platelet activation with a combined stimulus of collagen and thrombin, complement C5b-9 or a calcium ionophore, whereas the procoagulant activity of PEVs is lower upon activation with ADP, epinephrin or thrombin alone (Sims et al., 1989; Zwaal et al., 1992). In line with that, PEV populations are highly heterogeneous (French et al., 2020). Further diversity arises from differences in PEV content as well as PEV quantity, which is increased in a variety of diseases associated with chronic inflammation or ongoing platelet activation, such as cardiovascular diseases, stroke, sepsis, multiple types of cancers or autoimmune diseases (such as multiple sclerosis, or rheumatic arthritis) (Abrams et al., 1990; Varon and Shai, 2009; Chiva-Blanch et al., 2017; Gasecka et al., 2017). Disease-dependent changes in the biochemical composition of PEV may be useful for differential diagnosis given the ease of sampling. While PEVs had primarily been associated with procoagulant activity, another study gave first hints that PEVs may exert both pro- and anti-coagulatory effects, as some PEVs preparations were shown to inactivate Factor Va (Tans et al., 1991) through binding of Protein S to the coagulation inhibitor protein C (Dahlbaeck et al., 1992). Besides their effects on hemostasis, PEVs can exert immune-modulatory roles like platelets themselves. It has been shown for example, that they can inhibit IL-17 and IFN-γ production by Tregs in a P-selectin dependent manner (Chimen et al., 2020).

From the observations so far, it seems likely that half-life and clearance mechanisms of (P)EVs are dependent on pathophysiological conditions, such as the presence of inflammation. It remains to be established whether increased concentrations of PEVs in thrombosis are due to increased formation, less efficient clearance, or a combination of both. An important aspect of PEVs and other EVs in contrast to platelets is that their small size allows them to cross tissue barriers, for instance through fenestrated endothelium, whose pores are too small for platelets but large enough for EVs. These pores can have diameters of about 100–200 nm and sometimes reach even 1 µm. They occur in sinusoidal vascular beds of organs with prominent filtration or transendothelial transport processes, such as kidney, spleen and liver, with the latter exhibiting the space of Disse between endothelium and hepatocytes (Krams and Bäck, 2017). It is important to note that EVs not only enter these tissues from the blood through those pores, but also can be released by the underlying tissue and find their way into the circulation. This may also be used for diagnostic purposes, for example to monitor pathological liver states (Thietart and Rautou, 2020; Marcoux et al., 2021). Furthermore, EVs can also enter other anatomical sites such as lymph (Milasan et al., 2016; Tessandier et al., 2020), bone marrow (French et al., 2020) or even synovial fluid (Boilard et al., 2010; Puhm et al., 2021). By that, they can affect many different cell types and contribute to distant cell-cell communication beyond the circulation (Puhm et al., 2021). Recently, Tessandier et al. demonstrated that platelet-derived serotonin stimulated the transit of PEVs to lymph nodes, thereby promoting rheumatoid arthritis (Tessandier et al., 2020). Besides delivering intravesicular cargo to target cells, (P)EVs sometimes contribute to intercellular communication, by transferring surface proteins (such as C3, TF, or fibrinogen) in a piggyback manner to other cells, thereby creating interactive membranes that contribute to inflammation, thrombosis and cancer (Palviainen et al., 2020).

In conclusion, it can be stated that PEVs share many of the features and functions of platelets, which explains why they had originally been designated as “platelet dust” (Wolf, 1967). Upon platelet activation, newly released PEVs contribute to the expansion of “reaction surface” in parallel to the membrane increase caused by degranulation and by cellular protrusions formed via cytoskeletal remodeling. Beyond that they contribute to the diversity of platelet function by adding another layer of complexity through their roles as intercellular communicators transferring a great variety of cargo molecules to different, even distant target cells.

EVs play a role in diseases with acute inflammation (e.g., sepsis, acute respiratory distress syndrome, ARDS) (Sanwlani and Gangoda, 2021) as well as in chronic inflammatory conditions (e.g. asthma, chronic obstructive pulmonary disease, COPD, obesity) (Campello et al., 2016a; Alashkar Alhamwe et al., 2021; Purghè et al., 2021; Sanwlani and Gangoda, 2021). An example of a crucial chronic inflammatory process is atherosclerosis, the chronic inflammation of arteries, which is driven by LDL-cholesterol and immune cell infiltration. EVs derived from blood and vascular cells participate in the progression of atherosclerosis in several ways. Together with LDL-cholesterol, they promote inflammation, vascular dysfunction, leukocyte adhesion and tissue remodeling, leading to increased thrombotic risks (Owens and Mackman, 2011). Neutrophil-derived EVs, for example, contribute to atherogenesis through delivery of miR-155 to disease-prone regions enhancing NF-kB signaling and macrophage content (Gomez et al., 2020). EVs exposing TF and PS, which are retained in the plaque, account for the procoagulant activity of the lipid core, promote coagulation, and thereby increase plaque vulnerability (Chiva-Blanch and Badimon, 2019). Furthermore, PS-positive EVs can induce apoptosis and modulate gene expression (Del Conde et al., 2005). The function of different cellular EVs in atherosclerosis was recently reviewed in greater detail (Chiva-Blanch and Badimon, 2019).

EVs of platelets, RBCs and endothelial cells mediate inflammatory responses by interaction with PBMCs (peripheral blood mononuclear cells), mostly monocytes, causing the secretion of proinflammatory chemokines and cytokines, and leading to an increased survival of PBMCs (Danesh et al., 2014). Endothelial EVs, for example, were shown to correlate directly with IL-6 in vitro and in vivo (Curtis et al., 2009). RBC-derived EVs exhibit immune-inflammatory functions by modulating the activity of T and B cells. They can inhibit for example B cell activity after LPS stimulation by inhibition of the NF-kB pathway (Gao et al., 2020). On the other hand, RBC-EVs are able to bind to monocytes, thereby inducing the release of proinflammatory cytokines (IL-1, IL-6, TNF-α) and chemokines (macrophage-derived chemokine, MDC, and macrophage inflammatory protein 1a, MIP-1a), which induce T cell proliferation (Danesh et al., 2014). Besides inducing the release of pro-inflammatory cytokines by monocytes, EVs can also contain these mediators as cargo (including IL-12, IL-15, IL-17, and IFN-γ) and transfer them to other cells at early stages of inflammation (Sanwlani and Gangoda, 2021). Similarly, PEVs carry IL-1β, lipid mediators and DAMPs (e.g. HMGB1) (Puhm et al., 2021) and can modify CRP (C-reactive protein) into its pro-inflammatory form (Braig et al., 2017). An inflammatory role has also been shown for EVs transferring CD40L, which binds to monocytic CD40 and triggers the activation of NF-kB via the alternative pathway, causing the release of inflammatory mediators from monocytes (Bei et al., 2016). On the other hand, miR-125b-5p in RBC-EVs negatively regulates IL-1β induced inflammatory gene expression by targeting the TRAF6/MAPKs/NF-kB pathway (Rasheed et al., 2019).

Beside their role in the acute or early phase of inflammation, EVs also contribute to its resolution by carrying anti-inflammatory cytokines, such IL-4 or IL-10 (Gao et al., 2019) or by providing 12-lipoxygenase to mast cells which enhances the production of lipoxin A4 (Tang et al., 2010). PEVs can further polarize macrophages to an anti-inflammatory state or regulate adaptive immunity by inducing anti-inflammatory signaling in plasmacytoid dendritic cells (Sadallah et al., 2011).

The number of EVs is increased about twofold in the blood circulation of cancer patients reaching a sum of about 4 × 10¹⁵ (Kalluri, 2016). The reason for this increased occurrence is still not clarified in detail, but it is known that ESCRT genes, as well as other proteins and lipids that play a role in EV formation, are elevated in various tumors (Kilinc et al., 2021). While the biological roles of EVs are in general very diverse and complex, some of their effects on cancer development and progression have already been elucidated. Importantly, EVs released by cancer cells can fuse with other cells in the vicinity and thereby transfer their cargo, which can then exert a variety of effects on the recipient cells. Material that can be passed on includes proteins with pro-tumorigenic roles such as growth factors or expression products of oncogenes (e.g. KRAS, or EGFR variants), but also mRNAs, as well as miRNAs, which influence the cellular program of target cells on the transcriptional and post-transcriptional level (Bebelman et al., 2018). Furthermore, EVs can influence the tumor microenvironment in a way, which fosters further tumor growth. They can for example influence the fate of neighboring cells towards cancer-associated fibroblasts, which create a favorable niche for the tumor, or they can contribute to epithelial-mesenchymal transition (EMT). EVs can also promote angiogenesis and contribute to remodeling of the extracellular matrix (ECM) in a way, which supports metastasis from the primary tumor site, as well as the generation of pre-metastatic niches at distant sites, where cancer cells can settle. Besides, EVs can have multiple effects on immune cells, with both pro- and anti-tumorigenic consequences. On one hand, tumor-derived EVs carrying peptide-loaded MHC-I molecules can activate T-cell responses either directly or by cross-presentation via dendritic cells (DCs) and they can also activate macrophages or NK cells. On the other hand, exosomes are able to suppress anti-tumor immunity and immune surveillance by various means. TGFβ-containing EVs can downregulate cytotoxic T-cell function and increase immune-response dampening regulatory T-cells. Exosomes carrying FasL or TRAIL can induce apoptosis of activated T-cells and furthermore EVs can expose PD-L1 and thereby inhibit T-cell mediated immunity against cancers. EVs have the potential to affect myeloid cells by suppressing DC differentiation and can enhance immune-suppressive functions of myeloid-derived suppressor cells (MDSCs). Moreover, it was reported that EVs released from tumor cells promoted macrophage polarization towards the M2 phenotype, which exerts rather wound healing than defense functions and is considered tumor-promoting (Kalluri, 2016; Bebelman et al., 2018; Mashouri et al., 2019; Zhang and Yu, 2019).

Apart from these direct effects of EVs on cancer development and progression, they can have important diagnostic roles for the non-invasive detection of cancer via liquid biopsies using blood, urine or other body fluids, in which for example cancer-EV specific miRNAs or mRNAs are quantified. In combination with next-generation sequencing approaches, these samples can also give information on mutational signatures of cancers with the advantage that heterogenous or multi-focal tumor tissue, which is difficult to sample by standard biopsies can be analyzed and stratified altogether in one sample. While this field seems to have a great potential, it still needs better standardization and proof via larger patient cohorts (Kalluri, 2016; Möller and Lobb, 2020).

Further potential applications of EVs lie in cancer therapy. The ability of EVs to fuse with target cells in an efficient and sometimes specific manner renders them ideal candidates for the delivery of therapeutics, where they are often more effective than liposomes or lipid-based nanoparticles. Besides chemical drugs, they are also capable of delivering siRNAs or shRNAs that target oncogenes or mutations. In that context, CD47-positive exosomes are particularly interesting, as they are less taken up by macrophages due to their “don’t eat me signal” and therefore seem superior to normal lipid particles. Interestingly, EVs can be even prepared from plant cells, which would allow an easier scale-up to larger quantities (Kalluri, 2016; Zhang and Yu, 2019; Möller and Lobb, 2020). Apart from the use as delivery vehicles, EVs are also increasingly recognized as candidates for anti-cancer vaccines, based on their ability to modulate the immune response. For that purpose, exosomes can be used to present tumor-antigens so that cytotoxic T-cells are activated, or DCs may be used to generate exosomes that express MHC I and II and thereby induce anti-tumor immunity (Kalluri, 2016; Zhang and Yu, 2019; Möller and Lobb, 2020).

Although EVs are elevated in variety of thrombotic diseases, it remains difficult to pinpoint the role of EVs in thrombosis in detail.

To analyze the role of EVs in arterial thrombosis, previous studies used a FeCl₃-induced or photochemical injury of the mouse carotid artery (Wang et al., 2009; Kretz et al., 2010). Here, thrombosis is mainly driven by TF derived from the vessel wall (Day et al., 2005), since this injury type causes disruption of the endothelium. Mice injected with TF-positive EVs derived from monocytes showed increased fibrin accumulation implying that they contribute to thrombosis. Of note, these experiments were conducted in healthy mice with low levels of circulating TF+ EVs and the amount of TF+ EVs injected in the study may exceed those found in disease states. A second, frequently used model to study arterial thrombus formation is the laser injury model, where a small arteriole, usually of the cremaster muscle, is injured with a focused laser beam. This model may be better suitable to study the role of EVs in thrombosis since the injury per se causes less injury to the vessel wall (Rosen et al., 2001). Here, a rapid accumulation of TF staining upstream of the thrombus was observed (Gross et al., 2005), indicative of TF+ EV recruitment to the thrombus (Falati et al., 2002), as well as an incorporation of exogenously added TF+ EVs into the developing thrombus. This incorporation is dependent on PSGL-1 on EVs and platelet P-selectin (Falati et al., 2003). Using low TF mice, the cellular source of EV contributing to arterial thrombosis was pinpointed to neutrophil-derived TF+ EVs binding to endothelial cells via LFA-1 ICAM-1 interactions (Darbousset et al., 2012).

While arterial thrombosis occurs mainly after rupture or erosion of atherosclerotic lesions and is mainly characterized by a platelet-rich thrombus, venous thrombosis happens predominantly in regions of blood stasis and is characterized by a fibrin-, neutrophil- and red blood cell-rich thrombus.

Ligation of the interior vena cava (IVC) is the standard model to study venous thrombosis. Similar to arterial thrombosis in the cremaster, a role for P-selectin and leukocyte-derived EVs was shown in venous thrombosis using the IVC ligation model (Myers et al., 2003) with EVs being recruited to the thrombus. While thrombus weight correlated positively with PEVs, a negative correlation was shown for leukocyte-derived EVs (Ramacciotti et al., 2009). TF-positive EVs in venous thrombosis were studied in several models (Himber et al., 2003; Myers et al., 2003; Szema et al., 2005; Hawley et al., 2010) with an overall correlation of TF with increased EV concentration and increased thrombus mass. In line with that, mice deficient for the NF-kB transcription factor p50, which have a blunted TF expression, showed reduced venous thrombosis (Li et al., 2009). In carriers of factor V Leiden mutation, circulating EVs, especially PS-positive EVs, contribute to the development of venous thrombosis (Campello et al., 2012).

Generally, the plasmatic coagulation cascade can be triggered by a so-called extrinsic and an intrinsic pathway. The first is physiologically triggered by injury of the endothelium and exposure of subendothelial TF. This initiates a complex cascade of enzyme activations, involving factors VII, VIII, IX and X that generates a significant amount of active thrombin via several amplification loops. Thrombin then converts fibrinogen to fibrin, which polymerizes and is finally cross-linked to form a stable fibrin network (Furie and Furie, 1988). The intrinsic activation pathway (or contact pathway) can be triggered by negative charged surfaces, like those exposed by neutrophil-extracellular traps (NETs) and involves factors XII and XI, and merges with the other pathway at the level of factor IX. In parallel to the plasmatic coagulation cascade, platelets adhere to the injured site, are activated, and aggregate through interactions of platelet receptors with extracellular ligands. Both the plasmatic cascade and platelet activation stimulate each other as thrombin activates platelets via PAR receptors (protease-activated receptors) and activated platelets release fibrinogen, as well as factors V and VIII via their α-granules. Finally, platelets contract to form a densely packed thrombus (Krishnaswamy et al., 1993; Swieringa et al., 2018).

Platelets interact with the coagulation system in a multitude of ways, not only during thrombus formation, but also in specific areas within a formed thrombus (Swieringa et al., 2018). In the thrombus core, procoagulant, PS-exposing platelets increase the formation of tenase and prothrombinase complexes via the negatively charged phospholipid surface. This amplifies thrombin generation and links PS-exposure to platelet activation and coagulation (Zwaal and Schroit, 1997). Multiple enforcement loops exist both in the coagulation cascade and in platelet activation leading to a fast and efficient formation of a stable fibrin- and platelet-rich thrombus (Macfarlane, 1964; Versteeg et al., 2013).

While platelets and PEVs provide the majority of negatively charged surfaces under physiological conditions, and perivascular TF is required to initiate coagulation, the situation is different under pathological conditions. In disease states such as sepsis or cancer, a significant amount of TF can be found in circulation, mostly on the surface of microvesicles, which primes the blood for coagulation (Wolf, 1967). It has been proposed that the TF on MVs is predominantly inactive, residing on the surface in a so-called encrypted conformation, but that it can be activated by a conversion of disulfide bonds, which can be catalyzed by protein disulfide isomerase (PDI), explaining the procoagulant role of this enzyme in the circulation (Versteeg and Ruf, 2011; Stopa and Zwicker, 2018).