Neutrophil Extracellular Traps: A New Player in Cancer Metastasis and Therapeutic Target

As the first discovered and classical mechanism of NETs formation, the NADPH oxidase-dependent suicidal NETs formation has been reported to occur primarily at 3 to 8 hours following the activation of neutrophils by various stimuli, such as phorbol 12-myristate 13-acetate (PMA), lipopolysaccharide (LPS) and interleukin-8 (IL-8) [34]. Within this period, neutrophils undergo a series of biochemical processes, including nuclear delobulation, disassembly of the nuclear envelope, chromatin decondensation and eventually disintegration of plasma membrane, leaving the neutrophil to die, while the chromatin and granulated proteins to expulse [29].

Chromatin decondensation is a key step during NETs formation. The production of reactive oxygen species (ROS) mediated by NADPH Oxidase 2 (Nox2) have critical roles in NETosis as it triggers the downstream signalling cascade for chromatin decondensation. MPO and NE, two key enzymes stored in azurophilic granules of naïve neutrophils, are stimulated by ROS generation [35]. Activated NE then translocates into the nucleus and disrupt chromatin packaging by degradation of core histones. Subsequently, MPO binds to chromatin and cooperate with NE for further decondensation [35]. In addition, protein-arginine deiminase type 4 (PAD4), which hypercitrullinates histones, is implicated to be another downstream effector of ROS and contributes to chromatin decondensation [36]. Decondensed chromatin are then released into the cytoplasm, mixing with granule proteins and forming NETs. Finally, the plasma membrane permeabilizes, allowing for NETs expulsion while the neutrophil dies.

Intriguingly, NETs formation has also been observed in patients with chronic granulomatous disease [29], an inherited disorder in which NADPH oxidase activity is lacking and results in defective generation of ROS, raising the question of ROS necessity in NETs formation. In fact, NADPH Oxidase-independent NETosis also exists, but the ROS generation is alternatively provided by mitochondria rather than NADPH Oxidase. Some stimuli, such as immune complexes [37], calcium ionophores [38] and nicotine [39], have been implicated to trigger NETosis via mitochondrial ROS (referred as mitoNETs). More importantly, mitoNETs, in some autoimmune diseases, are enriched in oxidized mitochondrial DNA instead of nuclear chromatin which acts as potent inflammatory inducer and suggests a novel mechanism of neutrophil-mediated inflammation in autoimmune diseases [37].

Non-lytic NETs formation, still poorly understood, on the other hand, is a faster neutrophil response without loss of nuclear or plasma membrane, and can be ROS-dependent or independent. Gram-positive bacteria such as Staphylococcus aureus, has been found to trigger viable NETs formation independent of oxidant, generating NETs DNA primarily of nuclear origin [33]. Similar to the budding of extracellular vesicles, this process involves the release of nuclear DNA bound vesicles into the extracellular space where they are ruptured and the chromatin is released [33]. Another type of vital NETs formation resulting in mitochondrial DNA-containing NETs have also been observed in vitro when neutrophils are stimulated with LPS or complement factor 5a (C5a) [30]. Since previous study also reports a suicidal NETs formation with LPS treatment, the vital NETs formation induced by LPS may suggest a complementary mode of NETs formation at different stages in response to stimuli, which coordinately contribute to host defence against invaders.

One centrally important process arming the cancer cells with motility and invasiveness is the epithelial-mesenchymal transition (EMT). The ability of NETs to induce EMT is first described in vitro when the coincubation of normal endothelial cells with NETs isolated from human neutrophils stimulated with PMA leads to the loss of endothelial cell junction, the development of mesenchymal phenotype and aberrant activation of the Wnt/β-catenin signaling pathway [52]. Following this study, NETs-mediated EMT induction has been replicated in neoplastic epithelial cells as well as in murine models [53, 54], resulting in an even more significant exhibition of invasive phenotypes with high expression of EMT master regulators, including Snail, Slug and Zeb, when compared with known EMT stimulants such as TGF-β and IL-8 [55]. The ability of NETs to drive EMT both in fully normal epithelial cells and in neoplastic epithelial cells suggests that NETs may serve as one of the early contributors during the process of neoplastic transformation. More importantly, the correlation of NETs deposition with EMT activation has been confirmed in primary tumor as well as metastatic biopsy, though the mechanism of NETs-driven EMT is still not clear.

Intriguingly, Martins-Cardoso et al found that DNase treatment resulting in the destruction of NETs integrity exhibited minor impact on the invasive and migratory ability of breast cancer cells, suggesting a dispensable role of NETs integrity on EMT induction [53]. Consistently, Kajioka et al also found that degradation of NETs DNA does not affect the migration and invasion of the primary tumor whereas pharmacological inhibition of NET-associated elastase or NET-derived HMGB1 abrogates EMT induction [54], proposing the NETs proteins as the main contributor to EMT induction. Notably, degradation of NETs-DNA with DNase, though showed little influence on EMT, have been found to attenuate metastasis [54], suggesting NETs contribution to other steps of metastatic cascade, in addition to EMT.

Neoplastic cells derived from the primary site acquire the migration ability through EMT and may, sooner or later, invade into the circulation where they are termed as circulating tumor cells (CTCs) [1]. Though a surprisingly high amount of CTCs has been found in patients’ circulation, only a minority survive and form macrometastases under exposure of fluid shear forces, innate immunity and oxidative stress [1]. Therefore, early entrapment of CTCs could be one of the protective mechanism rendering their arrest in the circulation during transit. In accordance with the role of NETs in trapping microorganisms, researchers found that microvascular NETs deposition in response to cancer-associated inflammation is able to sequester CTCs. However, different from their trapping and killing pathogen, NETs-trapped cancer cells actually displayed enhanced invasiveness and subsequent metastasis potential [56] . Further investigation found that trapped cancer cells could internalize NETs via toll like receptor TLR4/9, leading to the activation of inflammatory signalling such as NFκB [56], MAPK/p38 and STAT3 pathway in cancer cells [57] and thereby facilitates metastatic outgrowth. In addition to the uptake of NETs via classical DAMP sensors on cancer cell, various mechanisms have been proposed, including the mechanical entrapment with web-like structure of NETs DNA [58], the integrin-mediated tethering [59] and the adhesion molecule-driven retention of CTCs [60]. Moreover, the accumulation of NETs in distant organs may serve as a directional chemotaxis factor for the recruitment of CTCs, as observed in a recent study where NETs DNA abundant in liver metastasis of breast cancer and colon cancer patients, acts as a chemotactic factor to attract CTCs through coiled-coil domain containing protein 25 (CCDC25), a transmembrane protein present on cancer cells [25], providing a possible mechanism to explain the site preference of metastasis and a therapeutic target in inhibiting metastasis .

After survival in the circulation and infiltration into the distant organs, newly disseminated tumor cells (DTCs) are immunologically and metabolically vulnerable to the challenging microenvironment in distant organs. The seed and soil hypothesis provides an explanation for organotropism in metastasis [8]. Indeed, the permissive and supportive pre-metastatic niche (the “soil”) are cultivated prior to the arrival of DTCs (the “seed”), thus rendering distant organs hospitable for tumor cell extravasation, proliferation, and colonize [3]. The major players involved in the initiation and formation of pre-metastatic niche includes primary tumor cells, bone marrow-derived cells and host stromal cells. In particular, emerging evidence demonstrated the recruitment of neutrophils into pre-metastatic niche such as liver [25], lung [61] or omentum [14], provides a permissive niche for the incoming cancer cells, at least in part through NETs-mediated remodelling of the local microenvironment.

Intriguingly, most DTCs that have managed to extravasate and disseminate to distant organs are destined to enter into a state of dormancy, sometimes for weeks, months, even years or decades, before forming metastasis [3]. Dormancy program confers a growth-arrested state in single DTCs or in small micrometastatic clusters, which actually reflects plasticity of DTCs to adapt to a given tissue and results in immune evasion and radiological undetectable [3]. Unfortunately, dormant cells will eventually exit the quiescent state and produce overt metastasis, raising the fundamental question of what signalling pathways or cellular events terminate the dormancy program. Albrengues [24] et al unveiled a NETs-initiated reactivation of dormant tumor cells in multiple metastasis models. Mechanistically, this effect is not via the direct contact of NETs and tumor cells. Instead, remodelling of extracellular matrix (ECM) by NETs-associated NE and MMP9 are mainly responsible for the proliferation of DTC to form metastasis, whereby the cleavage of laminin generates a new epitope which could be sensed by the integrin present on dormant cancer cells. The integrin-mediated interaction of ECM and dormant cancer cells subsequently activates the downstream effectors such as focal adhesion kinase (FAK), mitogen-activated extracellular signal-regulated kinase (MEK) and extracellular signal regulated kinase (ERK), thus leading to the dormant cancer cells re-entering the cell cycle and reinitiation of aggressive metastatic outgrowth [24].

To study the effects of NETs deficiency on tumor metastasis, conventional and neutrophil-specific PAD4 knockout mice have been generated and exhibited defects in NETosis. While neutrophil infiltration was not affected by the NETs depletion, metastatic colonization was dramatically reduced, compared with NETs-competent mice [14, 61, 71, 72]. Furthermore, several PAD4 inhibitors, including the specific inhibitor (GSK484) and the pan-pad inhibitor (Cl-amidine) have been tested in preclinical studies [14, 45] and displayed similar anti-metastatic effect as shown in PAD4-deficient mice. Given the critical role of NE and MPO in NETosis and as core protein components in NETs structure, NE-deficient mice or NE inhibitors have been utilized to study the effects of NETs depletion on metastasis. In a sepsis-associated lung carcinoma model, micrometastases was abrogated by the treatment of NE inhibitor GW311616A with markedly decreased tumor cell adhesion to hepatic and pulmonary microvasculature [58]. However, it is to note that the short serum half-life of the PAD4 and NE inhibitors currently available greatly limits their application in clinic and the development of new inhibitors with extended half-life is required. In addition, it is found that mice lacking NADPH oxidase [73] or treated with MPO inhibitor [74] fail to form NETs, which may also be utilized as a preclinical model to evaluate the effects of NETs absence in metastasis.

The context-specific pathway of NETs induction provides attractive target for interference of NETs formation under different malignancy, as exemplified by blocking IL-8 [43], Cathepsin C (CTSC) [46], Amyloid β [47] or CXCR-1, -2 [45], which are either tumor-derived pro-inflammatory factors or the receptors on neutrophils responsible for NETosis upon activation. Preventing neutrophils from receiving tumor-associated NETs stimuli can not only prevent metastatic colonization, but also synergize immunotherapy by destroying the shielding NETs of tumor cells. Importantly, compared with targeting PAD4 or NE, this strategy would specifically impede pathological NETosis without affecting other critical neutrophil functions. Therefore, a comprehensive understanding of the tumor-associated induction of NETs formation is a prerequisite for specifically inhibition of NETs generation.

Destruction of NETs structure at the primary or metastatic sites may represent another effective approach, which have the advantage of preserving antimicrobial functions of NETs in non-tumorous tissues. DNA, the backbone of the NETs structure, is therefore a feasible target for NETs degradation. Indeed, DNase I treatment is approved by FDA for the treatment of cystic fibrosis, partly attributed to the reduction of mucus viscosity caused by excessive NETs deposition during chronic inflammation [75]. Previous studies in cancer research have confirmed the potent anti-metastatic effect in preclinical models exerted by administration of recombinant DNase I [72, 76, 77], which however, requires a long-term multiple treatment due to the short half-life of DNase I in circulation [76, 77].

Recent progress have been made by crosslinking DNase I with nanoparticles to extend its half-life. As expected, treatment of DNase I-coated nanoparticles showed higher blood concentration, compared with free DNase I, but a continuous injection of the nanoparticles for two weeks is still required for the reduction of the lung metastatic burden [40]. Therefore, gene therapy may provide a feasible alternative approach to DNase I protein treatment. More importantly, the modification of gene vectors is able to achieve tissue-specific expression of transgene. Xia et al has developed a liver-specific DNase I delivery system using hepatotropic adeno-associated virus (AAV) vectors. A single administration of AAV-DNase I achieved a sustained liver expression of DNase I, which effectively inhibit CRC liver metastases along with enhanced antitumor immune response [78].

A growing body of evidence has suggested that NETs act more than a physical shield to cancer cells. Indeed the crosstalk of NETs and cancer cells promotes the invasiveness and migration ability through specific signalling. Therefore, some groups have focused on the interaction mediators present on NETs, cancer cells or both and explored the effects of inhibiting their interactions on metastasis. To date, the discovered mediator includes adhesion molecules such as integrin [79] and carcinoembryonic Ag cell adhesion molecule 1(CEACAM) [60], toll like receptor 9 [57], and CCDC25 [25]. Functional blocking these mediators using neutralizing antibody, specific antagonist, or the gene-deficient mice have exhibited reduced tumor cell migration and adhesion to the pre-metastatic niche even in the presence of widespread NETs deposition. Still it is to note that inhibition of one mediator in some cases may not completely prevent the NETs-tumor cell interaction as other proteins expressed on NETs or tumor cell can also contribute to their interactions [79], which makes this approach more complicated than direct destruction of NETs.