The power from within – understanding the driving forces of neutrophil extracellular trap formation

ABSTRACT Neutrophil extracellular traps (NETs) are one of the most intriguing discoveries in immunological research of the past few years. After their first description in 2004, the number of research articles on how NETs affect immunodefense, and also how they contribute to an ever-growing number of diseases, has skyrocketed. However, tempting as it may seem to plunge into pharmaceutical approaches to tamper with NET formation, our understanding of this complex process is still incomplete. Important concepts such as the context-dependent dual functions of NETs, in that they are both inflammatory and anti-inflammatory, or the major intra- and extracellular forces driving NET formation, are only emerging. In this Review, we summarize key aspects of our current understanding of NET formation (also termed NETosis), emphasize biophysical aspects and focus on three key principles – rearrangement and destabilization of the plasma membrane and the cytoskeleton, alterations and disassembly of the nuclear envelope, and chromatin decondensation as a driving force of intracellular reorganization. Summary: This Review explores the forces driving the process of neutrophil extracellular trap (NET) formation, with a focus on the biophysical aspects.


Introduction
To protect themselves from intruding pathogens, many species have developed complex mechanisms of immune defense. One of the most fascinating players within the innate immune system are neutrophilic granulocytes (Amulic et al., 2012). These cells are characterized by their unique structure, including their prominent multilobulated nucleus, an arsenal of highly effective granules and a characteristic composition of nuclear envelope and cytoskeleton, allowing these neutrophils to fulfill their function within the immune defense (Fig. 1A). It has been known for many decades that they possess a broad arsenal of effective defense strategies, including the production of reactive oxygen species (ROS), phagocytosis and the release of antimicrobial substances through degranulation (Cowland and Borregaard, 2016;Nordenfelt and Tapper, 2011). Nevertheless, they were long considered to be short-lived, terminally differentiated 'unsophisticated thugs' of the immune system (Jenne et al., 2018). However, the discovery of new neutrophil functions has cast a completely new light on the complexity of neutrophil biology and has catapulted them back into the focus of immunological research (Kubes, 2018;Nicolás-Ávila et al., 2017). For instance, it is now known that they can regulate their gene expression profiles in response to certain stimuli to support cell survival, cell function and cellular interactions (Tsukahara et al., 2003;Malcolm et al., 2003). Furthermore, they do not only mediate interactions with cells that are classically assigned to the innate immune response (Jaeger et al., 2012;Breedveld et al., 2017), but also with adaptive immune cells, such as B and T cells (Costa et al., 2019;Leliefeld et al., 2015;Hampton and Chtanova, 2016). Indeed, they even express major histocompatibility complexes (MHCs), as well as co-stimulating molecules, which qualifies them as antigen-presenting cells and allows them to directly influence the adaptive immune response (Lin and Lore, 2017;Fanger et al., 1997;Sandilands et al., 2006).
Among these newly discovered functions, perhaps the most astonishing and certainly the most radical feature of neutrophil biology is the formation of so-called neutrophil extracellular traps (NETs) (Brinkmann et al., 2004), also termed NETosis (Steinberg and Grinstein, 2007). These are complex networks of decondensed chromatin, citrullinated histones and antimicrobial peptides that are released by neutrophils under a variety of physiological and pathophysiological conditions (Papayannopoulos, 2018;Sollberger et al., 2018b). Their formation and expulsion is accompanied by drastic morphological changes. After stimulation, stimulusdependent signaling cascades are activated, which induce chromatin decondensation mainly by histone degradation and/or modification. Simultaneously, the cell loses its nuclear and granular membranes, degrades its cytoskeleton, rounds up and becomes softer overall. The chromatin continuously expands until it fills the whole cell and is then expulsed through the rupture of the cell membrane (Fuchs et al., 2007;Papayannopoulos, 2018;Neubert et al., 2018;Brinkmann, 2018). In most studied scenarios, the production of NETs represents a suicide mission, leaving the neutrophil to die ('suicidal' NET formation), while the expulsed chromatin fulfills its designated function in the extracellular space by combating intruding pathogens (Brinkmann, 2018). A different, 'vital' form of NET formation that occurs within minutes after stimulation has also been described. This process does not involve the disintegration of the plasma membrane, leaving the cell alive, but still results in the sudden and explosive extrusion of (mitochondrial) DNA (Yousefi et al., 2009;Yipp et al., 2012). This process remains relatively poorly understood and will not be covered in the current manuscript, as comprehensive studies, particularly of biophysical aspects, of vital NETosis are still lacking.
In the classical (suicidal) form of NET formation, early research on NET formation focused mainly on biochemical pathways and enzymes involved, but the importance of structural changes and mechanical forces is now becoming more and more evident. These insights have shed new light on how biophysical aspects contribute to the process of NET formation and are the focus of this Review article.
An ancient mechanism, but newly discovered In 1996, Takei et al. reported for the first time that the activation of neutrophils with the mitogen phorbol-12-myristate-13-acetate (PMA) was related to peculiar changes in neutrophil morphology and subsequent cell death (Takei et al., 1996). This discovery was initially hardly acknowledged until, in 2004, Brinkmann et al. described a novel PMA-activated defense strategy of neutrophil granulocytes mediated by the release of extracellular networks from chromatin and granule proteins, which they termed the formation of neutrophil extracellular traps (NETs) (Brinkmann et al., 2004). Since then, NETs and NET-related pathologies have continued to be popular in the field of immunological research.
Although NETs were originally described as a defense strategy mediated by mammalian neutrophils, the formation of extracellular traps (ETs) is not limited to this cell type. On the contrary, many other cells, such as eosinophils (Yousefi et al., 2008), mast cells (von Köckritz-Blickwede et al., 2008, monocytes and macrophages (Wong and Jacobs, 2013;Doster et al., 2018) and even neutrophils of zebrafish (Palićet al., 2007), plant root tip cells (Wen et al., 2009), social amoeba  and coelomocytes of earthworms (Homa et al., 2016) are able to produce ETs, presumably as a mechanism to protect themselves from harmful pathogens. Therefore, although the details of the underlying mechanism might vary among different species and cell types, the formation of ETs, in general, can be seen as an evolutionarily highly conserved process, which emphasizes its importance for biology.
Ingenious defense mechanism, superfluous relic or hazard for the host?
In neutrophils, NET formation occurs as a response to a variety of stimuli, including contact with bacteria (Hoppenbrouwers et al., 2017;van der Linden et al., 2017;Kenny et al., 2017;Fuchs et al., 2007), viruses (Saitoh et al., 2012), fungi (Kenny et al., 2017;Papayannopoulos et al., 2010;Metzler et al., 2011;Branzk et al., 2014) and parasites (Abi Abdallah et al., 2012;Guimaraes-Costa et al., 2009;Ávila et al., 2016), as well as with activated platelets (Rossaint et al., 2014), crystals (van der Linden et al., 2017Schauer et al., 2014;Warnatsch et al., 2015), different cytokines (Fuchs et al., 2007;Brinkmann et al., 2004;Martinelli et al., 2004), ionophores (Hoppenbrouwers et al., 2017;Kenny et al., 2017) and mitogens such as PMA (Fuchs et al., 2007;Brinkmann et al., 2004;van der Linden et al., 2017;Kenny et al., 2017;Metzler et al., 2011;Hoffmann et al., 2016). Initially, NETs were described as a novel immune defense strategy, as they have been shown to immobilize and even kill bacteria (Brinkmann et al., 2004) (Fig. 1B). Indeed, a protective immune function has been reported for several pathogens (Kenny et al., 2017;Saitoh et al., 2012;Abi Abdallah et al., 2012;von Köckritz-Blickwede et al., 2016;Urban et al., 2006). In this context, the immobilization of pathogens and thus the prevention of their dissemination appears to be one of the main defense mechanisms of NET formation. Furthermore, the spatial   confinement and degradation of inflammatory mediators, such as cytokines by the dense aggregations of NETs, so-called 'AggNETs', may convey important anti-inflammatory effects of NETs Euler and Hoffmann, 2019), as has been suggested for monosodium urate crystal-induced NET formation in gout (Schauer et al., 2014) and confirmed for several other typical NET stimuli . However, NET-bound antimicrobial proteins, such as neutrophil elastase (NE), myeloperoxidase (MPO) and histones, which have been shown to possess direct antimicrobial properties (Hoeksema et al., 2016), are also thought to support the immune defense mechanism of NETs to some extent (Brinkmann et al., 2004;O'Donoghue et al., 2013;Parker et al., 2012). Moreover, NET-associated DNA itself has been shown to mediate pathogen killing directly by cation chelation through the phosphodiester backbone and subsequent induction of membrane destabilization and bacterial lysis (Halverson et al., 2015). Useful as NETs may be for the immune system, their production appears to come with major drawbacks for the host (Fig. 1B). NETs, particularly the histones therein, are cytotoxic and can thus cause significant tissue damage (Villanueva et al., 2011;Silvestre-Roig et al., 2019). Additionally, NETs appear to be involved in the pathogenesis of a growing number of highly relevant diseases as discussed in several recent reviews (Sollberger et al., 2018b;Brinkmann, 2018), including cancer (Cools-Lartigue et al., 2013;Albrengues et al., 2018), diabetes and impaired wound-healing (Wong et al., 2015), preeclampsia (Erpenbeck et al., 2016) and sepsis (Clark et al., 2007). NETs are also implicated in cardiovascular diseases, including arteriosclerosis, thrombosis and even myocardial infarction (Döring et al., 2017;Kimball et al., 2016;de Boer et al., 2013;Warnatsch et al., 2015;Silvestre-Roig et al., 2019;Brill et al., 2012;Martinod et al., 2017). Additionally, NET components that are deposited extracellularly, such as MPO, double-stranded DNA and citrullinated histones, appear to be a major target for the formation of autoantibodies. Owing to this, NETs have been shown to be prominently involved in the pathogenesis of autoimmune diseases, including rheumatoid arthritis (Sur Chowdhury et al., 2014;Wang et al., 2018), systemic lupus erythematosus (SLE) (Leffler et al., 2013;Lood et al., 2016) and small-vessel vasculitis (Kessenbrock et al., 2009). Pathological mechanisms leading to this immunological dysfunction include overstimulation and dysregulation of NET formation, as well as excessive agglomeration of cell debris and insufficient clearance of NETs, as reported for SLE   (Fig. 1B).
In conclusion, the role of NETs in human health and disease is a two-sided one, with pro-inflammatory and anti-inflammatory effects co-existing most likely in a context-dependent fashion ( Fig. 1). It should be considered, though, that our understanding, particularly of the potential anti-inflammatory effects of NETs, is only incomplete. However, based on current knowledge, the negative consequences of NET release seem to outweigh their potential benefits. In any case, modifying the production, release or clearance of NETs appears to be an attractive therapeutic strategy for a plethora of diseases. Owing to the emerging dual function of NETs in both inflammatory (as in autoimmune or cardiovascular diseases) and anti-inflammatory processes (as in gout or during the entrapment of bacteria), a better understanding of this complex process is highly warranted before pharmaceutical interventions can be considered. In particular, the role of biophysical processes, such as the mechanical modification of the NETotic cell, is only starting to become a focus of study (Manley et al., 2018;Neubert et al., 2018). We will summarize below the current concepts of NET formation with a focus on the biophysical machinery that drives NET release.

Setting the stageincreasing cellular vulnerability
From a biomechanical and/or biophysical point of view, the formation of NETs involves dramatic rearrangements of the cellular and nuclear content, from chromatin decondensation and cell membrane reorganization to the degradation of the cytoskeleton and the disassembly of the nuclear envelope, and final release of the NET through the plasma membrane. These processes are explained in more detail below.
Recently, we have been able to divide this complex sequence of events into two fundamentally different phases, based on chromatin morphology and other structural changes of the cell. The first 'active' phase mainly depends on biochemical processes and energy consumption, whereas the second 'passive' phase is driven by the material properties of the cell, especially the entropic swelling of the expanding chromatin (Neubert et al., 2018) (Fig. 2). While the exact duration of these two phases may change dramatically depending on the experimental setting and the stimulus used to induce NET formation (see Fig. 2 for a timeline following PMA stimulation), the general sequence of morphological changes is highly reproducible and therefore provides an excellent marker for the progression of NET formation.
In general, the signaling cascades involved in suicidal NET formation greatly vary depending on the activating stimulus and have been extensively reviewed elsewhere (Papayannopoulos, 2018). Briefly, NET formation can involve activation of protein kinase C (PKC) with subsequent Raf-MEK-ERK signaling (Hakkim et al., 2011) and ROS generation mediated by the assembly of the NADPH oxidase (Fuchs et al., 2007) or signaling pathways that are wellknown for other cellular functions, such as autophagy (Remijsen et al., 2011), necroptosis (Desai et al., 2016), mitosis (Amulic et al., 2017) and pyroptosis (Chen et al., 2018). The neutrophil enzymes MPO and NE, as well as peptidyl-arginine deiminase 4 (PAD4), have also been reported to have prominent roles in NET formation as they modify histones and thus enable chromatin decondensation (Papayannopoulos, 2018). As of today, the most-studied activators used in mechanistic studies of NET formation in vitro are PMA, lipopolysaccharide (LPS) and calcium ionophores (CaIs), while more physiological stimuli include the above-mentioned fungi, bacteria, viruses or crystals (Papayannopoulos, 2018). After successful activation, these initiate decondensation of the chromatin and subsequent release of the NET within the second passive phase. Thus, NET release can be seen as the final outcome of various biochemical activation pathways. In this context, three principles are of particular importance for the progression of NET formation, which will be discussed in detail below: (1) systematic destabilization of the cytoskeleton, (2) rearrangement and disintegration of membranes and (3) chromatin swelling.
Degradation of the actin cytoskeleton during NET formation was reported in 2014 in neutrophils activated with Candida albicans around 30 min after stimulation and was shown to be mediated by NE, which is released from the azurosome, a complex formed during NET formation on the membrane of one specific type of neutrophilic granules, the primary (azurophilic) granules (Metzler et al., 2014). More recently, early degradation of actin has been verified upon stimulation with PMA and CaIs (Neubert et al., 2018;Thiam et al., preprint) (Fig. 3). Importantly, actin degradation is required for NET formation to progress, as inhibition of actin disassembly by jasplakinolide efficiently blocked both PMA-and CaI-induced NET formation (Thiam et al., preprint;Neubert et al., 2018). Interestingly, inhibition of actin polymerization by cytochalasin D or latrunculin A also decreased NET formation rates significantly if it was applied within the first 30 min after activation (Neubert et al., 2018). Thus, although the disassembly of actin appears to be crucial for the progression of NET formation, a functional cytoskeleton is necessary during the first 15 to 30 min (while the whole process of NET formation takes ∼2 h under PMA stimulation, as defined in Fig. 2) (Neubert et al., 2018). One might speculate that, in the beginning of NET formation, the cytoskeleton is needed to transport enzymes as well as granules through the cytoplasm, for instance, to the nucleus to allow the induction of chromatin decondensation.
Similarly, the disassembly of microtubule filaments (Amulic et al., 2017;Neubert et al., 2018;Thiam et al., preprint) as well as of the intermediate filament vimentin, especially of the peripheral vimentin structure (Thiam et al., preprint), has been reported to occur during NET formation. Indeed, microtubules were shown to dissolve from their outer free plus-ends inwards to the microtubule organization center (MTOC) within 30 min after PMA and CaI stimulation. Interestingly, unlike for actin, modification of tubulin function, for example by taxanes, does not appear to have any direct functional consequences for NET formation (Neubert et al., 2018;Thiam et al., preprint) (Fig. 3).
This extensive disassembly of cytoskeletal filaments ultimately leads to mechanical destabilization of the cell, as we recently analyzed by using life-cell atomic force microscopy (AFM) measurements (Neubert et al., 2018). During the first active phase of NET formation, cells become substantially softer, and their membrane (tether) tension greatly decreases over time, indicating the destabilization of the actin cytoskeleton (Neubert et al., 2018). In this context, it is important to mention that other cell organelles within the neutrophilic cytoplasm also disappear in the early stages of NET formation. For instance, the granular membranes dissolve (Fuchs et al., 2007), and the endoplasmic reticulum starts to break up into vesicles (Thiam et al., preprint) (Fig. 4A). These structural changes most likely support the overall softening and mechanical alterations of the cell.
This cellular destabilization is accompanied by a dramatic rearrangement of the plasma membrane. Indeed, cells partly detach from their substrate and retract their cell body, thus becoming taller and rounder (Neubert et al., 2018) (Fig. 4A). During this process, cells leave a remarkable amount of lipid material (i.e. excess membrane) behind on the substrate they are located on, in form of shed membrane vesicles; this can be correlated to a reduction in membrane flexibility and, therefore, increased susceptibility of the plasma membrane to subsequently rupture ( Fig. 4A) (Neubert et al., 2018;Thiam et al., preprint). By contrast, under normal circumstances, neutrophils show high membrane flexibility, and they are able to double their cell surface area by using any excess membrane (Ting-Beall et al., 1993). The formation of membrane vesicles has been reported for both PMA-induced NET formation (Neubert et al., 2018) and for CaI-induced NET formation shortly after the start of actin disassembly (Thiam et al., 2016, preprint), and has been termed microvesicle shedding (Thiam et al., 2016, preprint). Therefore, one may speculate that cell retraction, rounding and membrane shedding occurs as a biophysical consequence of reduced cell stability owing to cytoskeleton degradation.  Interestingly, even though neutrophils appear to reduce their adhesive contact to their substrate during NET formation, initial adhesion, as well as the availability of certain surface receptors such as the integrin Mac-1 (αMβ2 integrin) appears to be a prerequisite in many, though not all, forms of suicidal NET formation (Erpenbeck et al., 2019;Neeli et al., 2009;Healy et al., 2017;Mohanty et al., 2015;O'Brien and Reichner, 2016). For instance, PMA-induced NET formation does not require adhesion, as shown by experiments on passivated surfaces (Erpenbeck et al., 2019), while for LPSinduced NET formation it is essential (Erpenbeck et al., 2019). In line with these observations, NET formation rates are increased on hydrogels with higher stiffness, which support cell adhesion (Erpenbeck et al., 2019). These results indicate that adhesion, as well as substrate stiffness, plays an important but stimuli-dependent role during NET formation. Similar to many other cells, neutrophils respond to their environment and tune their level of adhesion. For example, neutrophils behave differently when exposed to altered densities of integrin ligands such as platelet receptors (GPIbα) (Kruss et al., 2013(Kruss et al., , 2012. Therefore, the differential availability of adhesion substrates is most likely a crucial factor in the regulation of NET formation. Additionally, other external cues, such as the presence of serum albumin (Neubert et al., 2019b) or changes in the surrounding osmolarity (Wong et al., 2015;Tibrewal et al., 2014), can significantly alter NET formation. For example, the presence of serum proteins inhibits the formation of NETs by human as well as murine neutrophils in a dose-dependent manner (Neubert et al., 2019b), and the number of NETs increases with increasing osmolarity compared to what is seen in iso-osmolar conditions (Tibrewal et al., 2014).
Physiologically, all of these factors may be part of mechanisms to prevent the excessive and potentially deleterious NET formation in the blood and other body fluids, where substrate elasticity is low, contact with integrin receptors happens comparatively rarely and the concentration of plasma proteins, such as serum albumin, is high.
An important question in the field is whether the weakening and disintegration of the plasma membrane before its rupture supports the cellular instability and therefore facilitates NET release. Indeed, different stimuli permeabilize plasma membranes differently, as demonstrated by live-cell imaging with non-cell-permeable chromatin dyes. For instance, drastic membrane permeabilization was observed prior to membrane rupture for CaI stimulation (de Bont et al., 2018;Hoppenbrouwers et al., 2017). Importantly, permeabilization of the plasma membrane increases over time and allows the membrane to become leaky, so that molecules between 0.6 kDa and 70 kDa can enter the cell (Thiam et al., preprint). In contrast, during PMA-induced NET formation, permeabilization of the plasma membrane is less obvious, as cell permeable dyes are only able to enter the cell at the moment of NET release, when the membrane ruptures (de Bont et al., 2018;Hoppenbrouwers et al., 2017). Nevertheless, weakening of the cell membrane has also been described in PMA-induced NET formation (Sollberger et al., 2018a). These authors suggested NE-dependent cleavage of gasdermin D and subsequent cell membrane pore formation by its cleavage product (Sollberger et al., 2018a), similar to what Chen et al. described for the nuclear membrane (Chen et al., 2018) (discussed below) (Fig. 4A). This type of pore formation is well established during pyroptosis, a programmed lytic cell death pathway (Liu et al., 2016). How and where these pores are actually formed, as well as their possible functional consequence for NET formation, remains to be further investigated.
In summary, through these complex changes, the cell loses internal stability during NET formation and becomes highly vulnerable to mechanical forces. It appears very likely that this increased vulnerability paves the way for the final rupture of the cell.
Opening the dooralterations and disassembly of the nuclear envelope In parallel to the above-described alterations of the cytoskeleton and plasma membrane, significant alterations of chromatin structure (see below) occur during NET formation to induce decondensation (see timeline of PMA-induced NET formation in Fig. 2). To expand into the cytoplasm of the neutrophil and, later on, into the extracellular space, the chromatin has to pass the nuclear envelope as a first obstacle (Fig. 4). Based on electron microscopy images, it was initially suggested that the nuclear membrane disintegrates into vesicles, as shown at later time-points of NET formation (120 min after PMA stimulation; see Fig. 2) (Fuchs et al., 2007), and this observation was subsequently confirmed a decade later (Amulic et al., 2017) (Fig. 4B). However, more recent studies have provided us with a more-detailed picture of the nuclear envelope alterations at the earlier stages of NET formation. We now know that the nuclear envelope breaks at one or more points about halfway through NET formation, at ∼60 min after activation, meaning that certain changes occur much earlier than vesicle formation (Neubert et al., 2018;Thiam et al., preprint) (Fig. 4C). Interestingly, the lamin layer beneath the nuclear membrane in the nuclear envelope (Fig. 1A) breaks only minutes before the nuclear membrane (Thiam et al., preprint). This rupture of the nuclear envelope also marks the beginning of the 'passive' chromatin expansion into the cytoplasm and correlates with the end of the 'active' phase of NET formation (Neubert et al., 2018).
The rupture of the nuclear envelope is most likely already prepared in the active phase of NET formation, similar to the alteration of the plasma membrane. As studies addressing the alterations of the nuclear and the cytoplasmic membrane were often conducted by multiple research groups in different experimental setups, a direct comparison of all data, especially in terms of exact time points during NET formation, is difficult. However, it may be assumed that these modifications occur in parallel, as also postulated by Thiam et al. In fact, the lamin layer already shows discontinuities within minutes after stimulation with CaI and before nuclear envelope rupture (Thiam et al., preprint) (Fig. 4A). Different mechanisms have been proposed for how the lamin layer can be dissolved, including cleavage of lamins, which is also observed during apoptosis (Gruenbaum et al., 2000;Rao et al., 1996). However, this appears to not be the case during NET formation, as a recent report demonstrated that lamin B1 proteins maintain their full length (Li et al., 2019, preprint). Therefore, lamin disintegration is more likely to be mediated by another strategy, such as the phosphorylation of lamins, similar to the process of lamin disassembly that is well known from mitosis (Ottaviano and Gerace, 1985). Indeed, phosphorylation of lamin A/C (Amulic et al., 2017), as well as of lamin B1 (Li et al., 2019, preprint) has been shown for NET formation.
In this context, it is interesting to note that NET-forming neutrophils also depend on key regulators of the cell cycle, such as the cyclin-dependent kinases 4 and 6 (CDK4 and CDK6) (Amulic et al., 2017), suggesting that terminally differentiated cells are able to highjack components of the cell cycle machinery and use them for NET formation, a seemingly unrelated and, from a temporal point of view, much faster process (Amulic et al., 2017).
Another means to weaken the nuclear envelope is through the formation of gasdermin D pores, which occurs in response to intracellular infections which induce NET formation (Chen et al., 2018). Here, similar to what occurs in noncanonical pyroptosis, gasdermin D cleavage into its p30 pore-forming fragment and subsequent pore formation is induced by caspase-11 (Fig. 4A). Apart from weakening the nuclear envelope, Chen et al. postulated that these pores allow the entry of enzymes, such as caspase-11, into the nucleus. These enzymes can support chromatin decondensation for instance by histone degradation, as has been shown by Chen et al. (2018) for the cleavage of H3 by caspase-11 (Chen et al., 2018).
It is important to note that, compared to most other cells, neutrophils have a unique and weak nuclear envelope, even before any of the described modifications of the nuclear envelope take place (Manley et al., 2018). Their lamin layer consists mainly of lamin B and lamin B receptor (LBR) (Olins and Olins, 2004), and only contains a very small amount of the stabilizing lamin A/C (Yabuki et al., 1999), as well as of the linker of nucleoskeleton and cytoskeleton complex (LINC) (Olins et al., 2009). These characteristics have been correlated to changes in nuclear mechanics and decreased stiffness and, therefore, most likely contribute to the remarkable flexibility of the neutrophil nucleus (Broers et al., 2004;Lammerding et al., 2004;Thiam et al., 2016) and the fast and easy redistribution of chromatin seen in NET formation (Manley et al., 2018). Consequently, the overall more fragile nucleus of neutrophils could facilitate the drastic chromatin reorganization that occurs during NET formation. This seems to be especially true for human neutrophils. For instance, murine cells have, among other characteristics such as their 'twisted ring-shape' (Olins and Olins, 2005), overall more lamins than their human counterparts (Olins and Olins, 2005;Olins et al., 2008). Possibly, this could provide an explanation for why human neutrophils are more prone to NET formation (Neubert et al., 2019b) and this should be kept in mind when deriving general conclusions from mouse experiments.
Passing the point of no returnchromatin expansion and rupture of the outer cell membrane As mentioned in the previous section, the beginning of chromatin expansion and the subsequent break of the nuclear envelope mark the beginning of the passive phase of NET formation. During the active phase of NET formation, the cell consumes high amounts of ATP, most likely to allow the complex signaling which has been described above. In contrast, in the second phase of NET formation, the ATP levels stay constant (Neubert et al., 2018). Additionally, once nuclear rupture has taken place and chromatin expansion has begun, NET formation can no longer be inhibited by pharmacological agents and has thus passed a point of no return (Neubert et al., 2018).
To reach this point, however, chromatin decondensation (Fig. 5A) must first be actively induced, mainly by histone modification and even histone degradation, as delineated above. One of the beststudied alterations, which takes place in multiple NET formation pathways, is the degradation of histones. For instance, H4, H1 and H2B can be degraded by NE (Papayannopoulos et al., 2010;Metzler et al., 2014), and H3 is cleaved by caspase-11 as shown in the context of gasdermin D pore formation (Chen et al., 2018) (Fig. 5B). Another enzyme important for chromatin decondensation is PAD4, which causes a pan-citrullination of histones by converting arginine into citrulline through deimination in a Ca 2+ -dependent fashion (Vossenaar et al., 2004). In turn, this leads to a decrease in the number of positive charges in histones (Nakashima et al., 2002;Wang et al., 2004;Dwivedi and Radic, 2014) and is thought to weaken their electrostatic interactions with the negatively charged DNA (Leshner et al., 2012). Additionally, citrullination can interfere with the binding of histones to other chromatin-binding proteins that contribute to heterochromatin organization, for instance, the heterochromatin protein 1 (HP1) (Fig. 5B) (Leshner et al., 2012). Recently, an additional, highly interesting concept for chromatin decondensation during NET formation was introduced, that of transcriptional firing   (Fig. 5B). The authors could show that chromatin decondensation depends on the promoter-melting step of transcription, in which the two DNA strands are separated to allow RNA synthesis. During NET formation, promoter melting and, consequently, DNA opening is activated, most likely by NETosis-specific kinases, thereby promoting chromatin decondensation . Therefore, promoter melting occurs in an aberrant manner during NET formation. This finding highlights once more how NET formation makes use of cellular pathways originally described in another context, such as mitotic signaling or transcription, to allow for the extensive chromatin decondensation required.
In the end, chromatin modifications, such as histone degradation or global histone citrullination, must cross a certain threshold in order to 'liberate' the DNA, a process facilitated by several of the above-mentioned enzymes translocating to the nucleus. From the moment the chromatin has started to expand (the 'point of no return'), the fate of the cell is determined by its material properties, mainly the entropic swelling of the chromatin, as counterforces, such as the cytoskeleton, as well as the physical confinement of the nuclear envelope have, at this point, mostly been eliminated, as explained above. Comparable to a dry sponge that comes in contact with water, the chromatin, which from a biophysical perspective may be considered as a polymer, swellsdriven by a gain in entropyuntil it reaches its equilibrium state. The equilibrium radius (radius of gyration) of the 2 m long DNA (in a human nucleus) corresponds to ∼150-200 µm (Tree et al., 2013;Latulippe and Zydney, 2010). In a normal nucleus (a diameter of 2-3 µm), the attractive interactions between DNA and positive charges of the histones keep the chromatin in place and prevent swelling. However, after histone citrullination and degradation, these interactions weaken, resulting in chromatin expansion in order to reach its new equilibrium radius. Subsequently, when the expanding chromatin reaches the plasma membrane, it exerts forces until it bursts through this obstacle. Using single-cell AFM measurements, we were able to measure an intracellular pressure generation of up to 150 Pa during NET formation (Neubert et al., 2018) (Fig. 5C). Interestingly, similar forces had been previously reported to be required to burst Xenopus oocytes in inflation experiments (Kelly and Macklem, 1991). It is this force that ultimately ruptures the cellular membrane and allows the release of the (still expanding) chromatin into the extracellular space. The previous weakening of the cell membrane and the degradation of the cytoskeleton, which leaves the cell unable to counteract any internal or external forces, supports this release.
This model of chromatin expansion being the main driving force of cellular rupture is not only supported by AFM, but also by the observation that smaller cells (∼150 µm 2 ), which hold a higher entropic pressure due to less chromatin expansion, rupture faster after the chromatin has filled the entire cell compared to their larger counterparts (up to 230 µm 2 ) (Neubert et al., 2018). Interestingly, although chromatin expansion itself is an overall homogeneous, isotropic process, the initial position of the nucleus within the cell appears to determine the cellular rupture point. It appears that the membrane closest to the nucleus at the beginning of chromatin expansion will be the first to experience the force of the expanding chromatin and thus be subjected longer to the highest pressure. This membrane region will, therefore, rupture more readily than membrane areas further away from the expanding chromatin (Neubert et al., 2018).

Conclusions and perspectives
From a clinical point of view, understanding the processes underlying NET formation is of fundamental importance (Fig. 6), as aberrant NET formation appears to be a key factor in the pathogenesis of a number of diseases, most prominently autoimmune diseases and cardiovascular disease, including myocardial infarction, as well as cancer growth and metastasis. It is likely that NET formation, if caught early, can be actively inhibited. Therefore, a huge number of inhibitors interfering with the multiple signaling pathways of NET formation have been investigated, and we have gained profound insights into how different inhibitors of enzymatic activity influence the active phase of NET formation (Gupta and Kaplan, 2016 6. Strategies to intervene in NET formation. Successful initiation of NET formation depends on pre-activation and priming of the cell, and can be modified by several environmental cues, such as pH, the presence of serum proteins, temperature or substrate elasticity/adhesion. During the active phase, before the start of chromatin decondensation, enzyme inhibition of NET-related enzymes or stabilization of the cytoskeleton block NET formation. After reaching the 'point of no return', with the beginning of the passive phase, such enzyme inhibitors are no longer efficient. Possible strategies to interfere with NET formation after this point include the alteration and/or stabilization of the membrane and the chromatin structure. Additionally, certain mechanical and biophysical triggers, such as osmolarity or shear stress, could influence the further progression of NET formation within this phase. After the final NET release, the clearance of the NET can be supported, for instance, by the addition of DNase. However, at later time points, when the cell has entered the passive phase of NET formation, this active inhibition has proven to be unsuccessful. Therefore, it would be of great benefit to also interfere with the passive phase of NET formation. However, we know little about how (or if ) the passive phase can be altered. It remains to be tested whether any modifications of materials properties that would influence the expansion of the chromatin itself, are feasible and could be used to successfully target this phase. For instance, one could think about stabilizing chromatin structure in the case of ongoing NET formation or about altering neutrophil membrane composition to slow down or prevent NET release and the spread of nuclear material. Needless to say that global modifications of neutrophil functions would need to be extensively tested and carefully considered as they could result in severe immunological dysfunctions.
Another approach to alter the effects of dysregulated NET formation (including increased NET formation rates, a faster response to activating triggers or deficiency in NET-clearing enzymes such as DNase I), is to not interfere with NET formation itself, but instead target released NETs, such as the clearance of extracellular chromatin by DNase. Although this strategy has been shown to improve symptoms in vivo (Kolaczkowska et al., 2015), it has some drawbacks, as it can liberate active, NET-bound NE (Kolaczkowska et al., 2015;Podolska et al., 2019) and pathogens. Furthermore, DNase fails to remove the majority of injurious histones (see above), which are highly deleterious to vessel walls and therefore only partly protects from tissue damage (Kolaczkowska et al., 2015).
When contemplating novel therapies addressing NET formation, it is also important to consider the profound susceptibility of NET formation to external factors. This includes the presence of serum proteins (Neubert et al., 2019b), the variation of temperature (Neubert et al., 2018) and pH (Behnen et al., 2017;Naffah de Souza et al., 2017), as well as light irradiation (Neubert et al., 2019a) and changes in substrate elasticity and adhesion (Erpenbeck et al., 2019). In addition, mechanical and biophysical triggers can influence NET formation, such as changes in osmolarity (Wong et al., 2015;Tibrewal et al., 2014) or shear forces (Yu et al., 2018). For instance, it has been shown that hyperosmolar stress can increase NET rates (Tibrewal et al., 2014). Likewise, arterial shear stress enhances NET formation as shown in a model of sterile occlusive thrombosis (Yu et al., 2018). The exact action of those external factors as well as their impact on a successful treatment of NET formation warrants further investigations.
It is also important to bear in mind that, although up to now, we know a lot about how NET formation is coordinated at the singlecell level, we do not yet fully understand the collective behavior of neutrophils and their intercellular communication during NET formation. Studying these processes is therefore likely to contribute to the development of any effective interventions in NET formation.
Finally, it is fascinating to consider that, by driving the complex process of NET release, chromatin has a third function in cells that is different from storing genetic information and the antimicrobial properties of NETs (as explained in the Introduction). In this context, the strong dependency of NET formation on passive, energy-independent processes, that is the entropy-driven chromatin expansion, seems surprising, as NET formation has traditionally been considered as a mainly active, enzymatically driven process (Papayannopoulos, 2018). It is thus tempting to speculate that other classical cellular functions, such as mitosis and gene regulation, might also rely on material (active matter) properties to a higher extent than currently appreciated (Stewart et al., 2011). With this in mind, it will be an interesting task to re-evaluate classical biological processes from a biophysical perspective.