NET impact on deep vein thrombosis

Deep vein thrombosis: a major health problem

Deep vein thrombosis (DVT) is a debilitating disease that may be complicated by pulmonary embolism (PE)¹. Together DVT and PE are designated as venous thromboembolism (VTE). In the US, VTE develops in an estimated 900,000 patients each year and PE is responsible for about 300,000 deaths, which exceeds the mortality from myocardial infarction or stroke^{2, 3}. DVT complications, in addition to PE, include post-thrombotic syndrome, caused by chronic venous stasis even in the absence of active thrombosis⁴.

DVT prophylaxis and treatment include anti-coagulation by heparin, vitamin K antagonists and thrombin and Factor Xa inhibitors^{5, 6}. Surgical intervention for DVT treatment includes thrombectomy or catheter-based thrombolysis using urokinase, streptokinase or tPA⁷. The success rate of pharmacologic catheter-directed thrombolysis ranges from 59 to 100%⁷. Morbidity from VTE has not substantially changed within the last two decades¹ and contemporary prophylaxis is not always efficient⁸.

DVT is epidemiologically associated with inflammatory diseases such as infection, autoimmune disorders and cancer. Acute infections^{9, 10} and autoantibodies against prothrombin¹¹ and phospholipids¹² are risk factors for DVT and PE. Coincidence of venous thrombosis and malignancy was documented in the 19^th century¹³. Recently, the risk of VTE and death was associated with elevated neutrophil counts in cancer patients undergoing chemotherapy¹⁴. Neutrophils are the most abundant inflammatory cells and recent animal studies emphasize their importance in DVT (Figure 1)^15–17.

Pathogenesis of DVT: lessons from animal models

DVT can be triggered by disturbances in venous blood flow, activation or dysfunction of the vascular endothelium and hypercoagulability. Immobilization from long-haul flight, bed-ridden position or limb paresis may also result in flow disturbances and stagnant blood flow in veins leading to DVT⁶. Flow restriction can be modeled experimentally in mice or rats by complete¹⁸ or partial^{17, 19} ligation of the inferior vena cava (IVC). These models represent different degrees of severity of blood flow distortion, although the mechanisms of thrombus development in complete and partial IVC closure may not be identical. In both human and murine DVT, thrombi are rich in fibrin and consist of red (rich in RBCs) and white (rich in platelets) parts^{19, 20}. Similar to human DVT, venous thrombosis in mice is induced in the absence of overt endothelial denudation²⁰ and can be prevented by low molecular weight heparin¹⁷.

In contrast to arterial thrombosis, which usually results from atherosclerotic plaque erosion or rupture, the mechanisms of DVT development in intact vessels are poorly understood. Venous thrombus development in humans often starts in the valve sinus where, due to complex fluid mechanics in the valve pockets, the blood flow pattern becomes abnormal leading to endothelial dysfunction. Stagnant blood flow (stasis) in the linear section of the blood vessel may have similar consequences^{21, 22}. Stasis leads to hypoxia²³, which in turn may contribute to thrombogenesis and locally activate endothelium. We have recently demonstrated that six hours after IVC stenosis in mice, von Willebrand factor (VWF) is released from Weibel-Palade bodies (WPB) and mediates platelet recruitment to endothelium (Figure 1A)¹⁹. This represents a key step in the initiation of DVT as neither VWF-deficient nor platelet-depleted mice form a thrombus^{17, 19}. Lack of GPIbα, a platelet receptor for VWF, prevents murine venous thrombosis¹⁷ and corroborates the importance of platelet adhesion in DVT. In addition to platelets, neutrophils are crucial for experimental DVT^{16, 17, 24}. P-selectin, exposed on activated endothelium or platelets, mediates initial recruitment of neutrophils (Figure 1A). Depletion of neutrophils or deficiency in P-selectin prevents DVT in mice¹⁷. In baboons, pharmacological inhibition of P-selectin accelerated thrombolysis and restoration of blood flow in thrombosed veins²⁵.

Hypercoagulability of blood is another important mechanism which contributes to DVT. Tissue factor (TF) is considered the central coagulation-triggering molecule in DVT²⁶ (Figure 1A, B). TF initiates the extrinsic coagulation pathway and is released in inflammation on microparticles originating from activated leukocytes. In murine DVT, flow restriction-induced fibrinogen deposition in the IVC of transgenic mice expressing low levels of human TF (< 1%) and no mouse TF, was substantially impaired in comparison to mice expressing normal levels of human TF¹⁷. This finding indicates that TF is indispensable for thrombus growth. Another potential coagulation-dependent mechanism playing a role in DVT is that blood pooled in large veins (with relatively low endothelial surface-to-blood volume ratio) cannot be efficiently anti-coagulated by the vessel wall as it would be in microcirculation where surface-to-volume ratio is much higher²⁷. Thus, the mechanisms of flow disturbance-triggered DVT involve endothelial activation, pro-coagulant shift in blood and recruitment of platelets and leukocytes. We and others have recently uncovered that neutrophils are recruited and release NETs in experimental DVT (Figure 1B). NETs may be a new target for therapeutic development and their implications for DVT will be the main topic of this review.

Neutrophil extracellular traps (NETs)

Neutrophils are the first leukocytes recruited to sites of infection, where they phagocytose invading bacteria²⁸. Within the phagosome, microbes are killed by locally high concentrations of antimicrobial proteins and reactive oxygen species (ROS). NETs are produced to allow neutrophils to trap and disarm microbes in the extracellular environment²⁹. NETs are scaffolds of intact chromatin fibers with antimicrobial proteins, ideal to retain large quantities of microbes (Figure 2A). Therefore some pathogenic bacteria have evolved to express an extracellular DNase, which dismantles NETs and promotes virulence^{30, 31}.

Extracellular traps (ETs) are formed in humans, animals and even plants³² indicating that NETs provide an evolutionary conserved protective mechanism. Indeed, ET formation is not restricted to neutrophils^33–35 and different cell types employ different cellular mechanisms to release ET. One mechanism used by human neutrophils is NETosis³⁶. NETosis is a multi-step cell death program (Figure 2C). Upon activation, certain enzymes translocate from the granules to the nucleus³⁷. Histones are degraded by neutrophil elastase (NE)³⁷ and citrullinated by peptidylarginine deiminase 4 (PAD4)³⁸ to unwind chromatin. Further hallmarks are the breakdown of granular and nuclear membranes and cytolysis as the final step in NETosis³⁶. NETosis involves signaling pathways which lead to ROS production^{36, 39} and up-regulation of anti-apoptotic proteins⁴⁰. It is distinct from apoptosis, where nuclear condensation and DNA fragmentation occur, and necrosis, where the plasma membrane breaks before the nuclear envelope³⁶. In vitro, the kinetics of NETosis varies from less than 30 min to 240 min, likely depending on the type and concentration of the stimulus, isolation procedure of neutrophils and sensitivity of the detection method^{36, 41}. Alternatively, NETs may be released from viable neutrophils by ejecting mitochondrial DNA⁴², nuclear contents⁴³ or DNA-containing vesicles⁴⁴, but the underlying signaling pathways of these processes are not well defined.

NETs link innate immunity with thrombosis

NETs provide a new link between innate immunity and thrombosis. NETs stimulate platelet adhesion (Figure 2B)¹⁵ and coagulation^{17, 45} and are abundant in experimental deep vein thrombi in baboons¹⁵ and mice^{16, 17}, where they co-localize with VWF^{15, 16}. Treatment of mice with DNase1 prevents thrombus formation^{16, 17} underscoring the importance of NETs for DVT. NETs interact with endothelium, platelets, and coagulation factors and may be able to influence thrombolysis.

NETs and endothelium

Activation of endothelium and WPB release play a crucial role in the initiation of DVT. Co-cultures of activated endothelial cells and neutrophils promote NET formation, which is dependent on platelets⁴³ or interleukin-8 and ROS released from endothelium⁴⁶. NETs in turn induce endothelial cell death^{43, 46, 47}, an effect likely mediated by NET-associated proteases or cationic proteins such as defensins and histones^{47, 48}. Histones display a high affinity for phospholipids and their binding to membranes results in pore formation and an influx of ions^49–51. Interactions of histones with endothelium could promote thrombosis by exacerbating endothelial activation and WPB release through an increase of intracellular calcium levels⁵². We have observed that plasma VWF increases and DVT is aggravated in mice infused with purified histones¹⁶.

NETs and platelets

NETs can also contribute to thrombus formation through interaction with platelets (Figure 1B, 2B). NETs are very large structures and may promote thrombus stability similarly to VWF and fibrin⁵³. When perfused with blood, NETs bind platelets and support their aggregation, indicating that they are a substrate for platelet adhesion and also provide a stimulus for platelet activation¹⁵. Platelets may bind to NETs both directly and indirectly. Purified histones associate with the platelet surface in vitro⁵⁴, presumably via electrostatic interactions with phospholipids⁴⁹ or carbohydrates⁵⁵ or via Toll-like receptors on platelets⁵⁶. Platelets also bind double and single stranded DNA in vitro^{57, 58}. Interestingly, platelets from patients with systemic lupus erythematosus (SLE) have immune complexes of DNA and anti-DNA-antibodies on their surface, which can be released by incubation with DNase⁵⁹. NET degradation is impaired in SLE patients due to a reduced DNase1 activity in serum⁶⁰ and future studies may address whether the inability to degrade NETs correlates with the increased risk for venous thrombosis in these patients⁶¹. Platelet-NET interactions could also be mediated by adhesion molecules such as VWF, fibrinogen or fibronectin¹⁵. These molecules bind to NETs presumably because of their affinity for histones or DNA^62–64. Activation of platelets by NETs might be triggered by histones or neutrophil proteases in NETs. Purified histones stimulate influx of calcium into platelets and promote activation and aggregation in vitro^{54, 56}. When infused into mice, histones co-localize with platelets and induce thrombocytopenia and thrombosis^{16, 51, 54}. NETs contain enzymatically active NE and cathepsin G²⁹ and these proteases potentiate platelet aggregation through proteolytically activating platelet receptors^{65, 66}. Interactions of NETs with platelets may result in a vicious cycle of NET formation and platelet activation, because platelets prestimulated with LPS or collagen trigger neutrophils to release NETs ^{43, 45}.

NETs and red blood cells

Red thrombi are typical for DVT. But unlike platelets, the role of RBCs in thrombus formation is not well defined and they are frequently considered passively entrapped. However, RBCs may promote coagulation by exposing phosphatidylserine and altering blood viscosity⁶⁷. We found that in addition to platelets, RBCs avidly bind to NETs after perfusion of whole blood¹⁵. Activated neutrophils or platelets can also recruit RBCs at very low venous shear in vitro⁶⁸. Similar to platelets, RBC may interact with NETs directly or indirectly. DNA was eluted from the surface of isolated RBCs from cancer patients⁶⁹, indicating that RBCs can bind DNA. Interestingly, in experimental DVT in mice, NETs are predominantly found in the red, RBC-rich part of the thrombus¹⁶, suggesting that NETs could be important for RBC recruitment to venous thrombi (Figure 1B).

NETs and coagulation

Fibrin is abundantly present in venous thrombi. In vitro, NETs stimulate fibrin formation and deposition and fibrin co-localizes with NETs in blood clots^{15, 17, 45}. NETs stimulate both the extrinsic and intrinsic coagulation pathway^{17, 45}. NE is known to cleave tissue factor pathway inhibitor (TFPI) and enhance Factor Xa activity⁷⁰. NETs contain NE and bind TFPI and therefore facilitate proteolytic inactivation of TFPI by NE⁴⁵. NETs also bind Factor XII and stimulate fibrin formation via the intrinsic coagulation pathway¹⁷. DNA and histones in NETs may play an important role in stimulating coagulation as well. Nucleic acids enhance the activity of coagulation serine proteases⁷¹ and histones promote coagulation indirectly by activating platelets and stimulating release of pro-coagulant polyphosphates from platelet granules^{56, 72}. In addition, histones inhibit anticoagulants in plasma. Histones interact with thrombomodulin (TM) and protein C and inhibit TM-mediated protein C activation⁷³. As a consequence, histones dose-dependently increase plasma thrombin generation in vitro. Histones in NETs may exhibit similar functions and thus promote fibrin deposition in DVT (Figure 1B).

Implications of NETs in thrombolysis

In order to degrade and solubilize thrombi to restore blood flow, fibrin and VWF as the main scaffolds need to be proteolytically fragmented by the proteases plasmin and ADAMTS13, respectively. NETs are a newly recognized third scaffold that needs to be undone during thrombolysis (Figure 1C).

NETs were seen to co-localize with fibrin in clots¹⁵ and with VWF in venous thrombi^{15, 16}. In vitro, we could show that NETs provide a scaffold for blood clots that is resistant to tissue plasminogen activator (tPA)-induced thrombolysis¹⁵. We incubated re-calcified blood with neutrophils which were pre-stimulated to release NETs. As shown in Figure 3, after filtration, blood clots appeared in control samples and tPA- or DNase-treated blood but not in blood treated with the combination of tPA and DNase (Figure 3A). Immunostainings revealed that in the presence of tPA, blood clots lacked fibrin and were held together by a scaffold of extracellular DNA (Figure 3B).

DNase1 is the predominant nuclease in plasma. Interestingly, the plasminogen system cooperates with DNase1 during chromatin degradation⁷⁴. DNase1 has only limited activity to degrade chromatin as it preferentially degrades protein-free DNA. Plasminogen, activated by either tPA or urokinase-type plasminogen activator (uPA), degrades histones and therefore allows for degradation of DNA by DNase1⁷⁴. Monocytes/macrophages may also support the DNA degradation because their lysosomes contains DNase2, which is important for the removal of apoptotic cells (Figure 1C)⁷⁵. NETs and fibrin degradation by plasmin and DNase could result in the simultaneous release of DNA and fibrin fragments. In baboon DVT, plasma DNA increases¹⁵ with similar kinetics to the fibrin degradation product D-dimers⁷⁶. Recently, in collaboration with Thomas Wakefield’s group, we found increased levels of DNA in plasma from patients with DVT compared to healthy controls and symptomatic patients who did not have DVT. Here also plasma DNA concentrations correlated with D-dimers (Diaz JA and Fuchs TA et al., unpublished data, 2012). Therefore it is plausible that circulating DNA may reflect the degradation of NETs within a thrombus.

NETs may also promote thrombolysis. In vitro studies have shown that NE and cathepsin G can degrade fibrin⁷⁷ and these proteases are present on NETs and could enhance fibrinolysis in DVT. In addition, NETs may also recruit plasminogen from the plasma. Histone H2B can serve as a receptor for plasminogen on the surface of human monocytes/macrophages⁷⁸ and perhaps could do so in NETs.

NETs in DVT: Questions and Challenges for the future

Now that a new polymeric scaffold, NETs, has been identified in thrombi of deep veins¹⁵, what should be explored to make the best use of this observation to progress towards generation of new approaches to prevention and treatment of DVT? First, it will be important to learn more about what triggers NET formation in DVT. If the trigger(s) were identified, their generation could be inhibited. The trigger could be the interaction of neutrophils with activated cells such as platelets or endothelium^{43, 45, 46}. This could be confirmed if the neutrophil surface receptors involved were identified and their inhibition tested in DVT. Alternatively, environmental factors such as hypoxia, ROS, cytokines or possibly coagulation proteases generated early in thrombosis could induce NETosis. In vitro, ROS are a common denominator of ET formation used by different types of leukocytes^34–36. Platelets may potentiate ROS production in platelet-neutrophil complexes⁷⁹ and endothelial ROS were shown to trigger NET formation by neutrophils in vitro⁴⁶. Protease activated receptors are present on neutrophils⁸⁰ but whether or how they respond in thrombosis is an open question. Perhaps, studies should move from purified neutrophils to more complex in vitro systems, which may better model DVT and involve endothelium, platelets and neutrophils under hypoxic and other experimental conditions.

After the neutrophil is triggered to NETosis, much needs to be clarified about the cell biology of NETs formation. Again, if the cellular processes leading to nuclear dissolution and NET ejection were elucidated, their inhibition in DVT could be envisioned. At least two enzymes: neutrophil elastase³⁷ and PAD4³⁸ are implicated in chromatin decondensation and NET generation. Their sequence of action, substrates, and processes through which they access the nucleus are not clear. These enzymes’ inhibition effects on DVT should be evaluated using available inhibitors and knockout mice. Whether the final discharge of NETs (Figure 2C) from the neutrophil is produced by cellular lysis³⁶ or a secretory process^{42, 43} that could be inhibited is also an important question.

The time of onset and duration of NET formation in venous thrombogenesis needs to be determined to better understand their function in the process. The challenge will be to capture single neutrophils forming NETs in DVT by employing intravital microscopy in mice. If NETs are produced rapidly after flow restriction, they may facilitate thrombus initiation. This is supported by the absence of visible thrombi in most DNase-treated mice^{16, 17}. The observation that NETs are located at the interface between thrombus and vascular wall¹⁵ may suggest that they help anchor the thrombus to the vessel. NETs were observed in mature thrombi; but whether these were remnants or whether NETs are formed continuously during venous thrombosis is not known. In an aged thrombus, NETs may participate in thrombus remodeling, for example by recruiting endothelial progenitor cells for neovessel formation. Whether NETs may recruit tissue factor-containing microparticles that are an important component of DVT is also not known.

A very important challenge will be to determine NETs’ effect on thrombolysis. In vitro and in vivo observations indicate that chromatin, fibrin and VWF form a colocalized network within the thrombus that is similar to extracellular matrix^15–17. It is likely that each of the components will need to be cleaved by their own appropriate enzyme (Figure 1C) and also that the presence of one component may influence the degradation or stability of the other. Fibrin is crosslinked/stabilized by FXIII transglutaminase. Whether NET components provide a substrate(s) to FXIII and could be crosslinked to fibrin is not known. PAD4 is eventually secreted from neutrophils during NET formation and was shown to citrullinate fibrin in rheumatoid arthritis⁸¹. Whether this occurs in DVT and how it may affect fibrin degradation by plasmin is unknown. As part of thrombolysis monocytes are recruited to the thrombus. Interestingly, they are equipped to degrade both fibrin and DNA^{75, 78}. Whether NETs promote or interfere with monocyte recruitment to the thrombus has not been investigated.

In summary, in vitro and in vivo observations indicate that NETs could influence initiation, growth and resolution of DVT. NETs may nucleate thrombus formation by enabling locally high concentrations of platelets, RBCs and coagulation factors. NETs together with fibrin and VWF could cooperatively provide thrombus stability. Future studies should determine whether NETs formation in DVT can be prevented and most importantly from a clinical perspective whether DNase, could be developed as a new, hopefully safer, therapeutic drug for thrombolysis.