Histone H3 clipping is a novel signature of human neutrophil extracellular traps

Article activity feed

1. 

   Version published to 10.7554/elife.68283 on eLife

   Oct 25, 2022
2. 

   eLife

   Jun 28, 2022

   Author Response

   Reviewer #1 (Public Review):

   Neutrophil extracellular traps (NETs) are defined as structures containing extracellular DNA co-localizing with granule-derived proteins, such as neutrophil elastase, and histones. While in in vitro assays a variety of protocols have been described to unambiguously detect and quantify neutrophil extracellular traps (NETs), in ex vivo tissue samples, quantification and demarcation of NETs from the remnants other forms of neutrophil cell death such as necrosis is still challenging. The current manuscript by Tilley and colleagues describes a novel tool to perform that important task. The authors have discovered that human histone H3 is processed by serine proteases at a specific cleavage site during NET formation. They created a mouse monoclonal antibody to this cleaved histone H3 and …

   Author Response

   Reviewer #1 (Public Review):

   Neutrophil extracellular traps (NETs) are defined as structures containing extracellular DNA co-localizing with granule-derived proteins, such as neutrophil elastase, and histones. While in in vitro assays a variety of protocols have been described to unambiguously detect and quantify neutrophil extracellular traps (NETs), in ex vivo tissue samples, quantification and demarcation of NETs from the remnants other forms of neutrophil cell death such as necrosis is still challenging. The current manuscript by Tilley and colleagues describes a novel tool to perform that important task. The authors have discovered that human histone H3 is processed by serine proteases at a specific cleavage site during NET formation. They created a mouse monoclonal antibody to this cleaved histone H3 and assessed its performance as a tool to detect NETs in vitro and ex vivo.

   The paper is well-structured and written, thus presenting a valuable contribution to the field. There are some open issues with the manuscript which are not clear at this point:

   1. One major point are the dynamics when this clipping occurs and if it occurs extra- or intracellularly. The authors have a used a serine protease inhibitor, AEBSF, which not only inhibits histone clipping but also NET formation and nuclear decondensation itself. I am therefore not sure if the conclusion can be drawn that histone H3 clipping is an intracellular event and "serine proteases cleave the N-terminus of H3 early during NET formation." This is also in open conflict to the study by Pieterse et al. (Ann Rheum Dis 2018) who demonstrated prevention of histone clipping by serine protease inhibitors working exclusively outside the cell.

   The reviewer raises an intriguing biological question. It will be very interesting to determine if the H3R49 clipping is happening intracellularly or extracellularly. This specific H3R49 cleavage does not appear until 120 min, however larger cleavage products appear after 30 min PMA stimulation by western blot – Fig 1A and 3A – at time points in which we know the cell membrane is not permeable as determined by sytox staining. However, we have not inferred from this that all cleavage is occurring intracellularly. Indeed, the kinetics of the appearance of the smaller 10kDa fragment, recognised by 3D9, suggests this specific cleavage likely occurs after permeabilization of the membrane which is in line with the findings of Pieterse et al, that cleavage happens after lysis or permeabilization of the membrane. However some cleavage may also happen intracellulary. A point to note is that, in their paper, they use a histone H3 antibody (#34) directed to residues-29-32, which includes a cleavage site we putatively identify in this study (Table 1): H3T32 – derived from the detection of the TGGVK peptide in Edmann degradation. This site and the epitope of the Pieterse anti-H3 (#34) are N terminal to the cleavage site H3R49. Thus, when cleaved at H3R49, a histone fragment may be released that contains this epitope. Using necrostatin, a reported inhibitor of membrane permeablisation, Pieterse et al show that the chromatin decondenses but is not externalised and this chromatin stains strongly for the histone H3 tail region (a.a. 29-32). They interpret this as the tails not being cleaved. However, we present an alternative hypothesis: the tails are cleaved, allowing chromatin decondensation, but the fragments remain concentrated in the cell as they cannot be dispersed through cell membrane permeabilization. A western blot of necrostatin treated cells may shed light on whether cleavage is really not happening in the absence of membrane permeablisation. One advantage of our cleavage site specific antibody is that it does not recognise free or cleaved tails, but the remaining histone that is still associated with the chromatin or NET. Furthermore, the epitope is not present if the histone has not been cleaved at H3R49. We hope that it will be a useful tool for the community in investigating the different molecular mechanisms at play during NET formation.

   We agree we have not tied the actual proteolytic event to the serine proteases however we can say that their activity is needed for the events leading to histone H3 cleavage. Thus, we have modified our finding to reflect this. Further research is needed to explore which proteases, serine or otherwise are mediating this cleavage event at H3R49

   Line 98-100 “This data shows that the N-terminus of H3 is cleaved early during NET formation and that this event is dependent on serine protease activity.

   1. Along these lines it is also confusing that the staining by an anti-citH3 Ab and 3D9 seems to be mutually exclusive. The authors mention this in the discussion and explain that 3D9 "may display a preference for more mature or proteolytically processed NETs." This seems to be hard to align with their claim that H3 is clipped early during NET formation. It would be important to show if citH3+ NETs progress into 3D9+ NETs. If this is not the case, that would potentially render a large part of the literature that has used citH3 staining for the detection of "NETs" useless.

   We are grateful to the reviewer for highlighting this confusion and we have made changes to the discussion to clarify our conclusions. We have now made a clearer distinction between general histone H3 cleavage and the specific cleavage site detected by 3D9 at H3R49. Histone clipping/cleavage is a process that occurs during NET formation, at various sites and time points. At later time points, the H3R49 site is cleaved and the antibody recognises this later cleavage.

   Line 267-270 Using a biochemical and proteomic approach, we determined that H3 is cleaved at multiple sites in its N-terminal tail during the course of NET formation and notably, at a novel cleavage site in its globular domain, H3R49, at ~120 min

   Line 292-294 Thus, we propose that 3D9 will allow broad detection of NETs induced by varied stimuli but that it may display a preference for more mature or proteolytically processed NETs or NETs that are citrullinated to a lesser degree or not at all.

   We have also included a time course of h3cit and 3D9 staining with PMA demonstrating that while most cells stain only for 3D9, there are a small percentage of cells staining for H3cit and an even smaller percentage that are double positive (Figure 11-figure supplement 2). This would suggest, that at least with PMA, a NOX dependent stimulus, that citrullination does not commonly precede histone cleavage although it may occur. Progression from citrullination to histone cleavage may be more common for NOX independent stimuli. Indeed, Nigericin also induces citrullination (90 min) but citrullination inhibitors do not affect the level of NET production (Kenny et al 2017, Elife. 2017, 6:e24437, Figure 6) and in the current manuscript we show that Nigercin produces NETs that are detectable with 3D9 and thus contain cleaved H3 at 2.5h. To us this suggests that the events of citrullination or histone cleavage are not mutually exclusive but that the hallmarks that remain once the NET is formed can be exclusive, specifically when looking at H3cit R2,8,19 and cleavage at H3R49. It would be every interesting to look at other citrullination markers, e.g. on H4 with a wide range of stimuli to see if the evidence of citrullination was more ‘long lived’ on a histone that was less susceptible to proteolysis but this may be examined in future research.

   1. Non-suicidal pathways of NET formation were described, where parts of the nucleus are extruded but the cell remains intact and basal cellular functions of neutrophils are still carried out. These" vital NETs" are not addressed in the manuscript.

   The phenomenon of “vital NETs” and whether histone H3 cleavage occurs during this process would be a very interesting question to explore but, unfortunately, we are not yet able to investigate this using this antibody. The occurrence of vital NETs needs to be observed with live cell imaging studies (Yipp et al., 2013, doi: 10.1182/blood-2013-04-457671). However, our antibody selection method was optimised for use with fixed or denatured human neutrophil samples only.

   We have addressed this limitation and its consequences for choice of assay in the discussion in the new manuscript.

   Line 308-319

   “In this study we detect NETs in fixed or denatured human samples from in vitro experiments and histological samples……. thus, care should be taken in the design of future assays and selection of sample when detecting cleaved H3, NETs or vital NETs under native and mild detergent conditions. In particular, 3D9 is not suitable for direct detection of NETs in complex biological fluids and a sandwich approach or colocalization with a neutrophil granule protein is critical.“

   1. The authors show that neutrophils stimulated with C. albicans released NETs not bound by 3D9, which remains unexplained.

   For this study we use the histological definition of NETs as extracellular chromatin decorated with neutrophil granule proteins. Thus, we can describe these visible structures as NETs as they stain for DNA and neutrophil elastase. It is possible that they are citrullinated NETs, as observed by Kenny et al (Elife. 2017, 6:e24437, Figure 6). However, we cannot exclude that these are remnants of other forms of neutrophil cell death and this remains a problem faced by the entire field as recently reported by Boeltz et al., 2019 (DOI:10.1038/s41418-018-0261-x).

   We have expanded on this observation in the discussion: line 282-285

   Like citrullination, not all NETs contain H3 cleaved at H3R49. With Candida albicans, some NET-like structures were not 3D9 positive (Figure 6). These may be remnants of other forms of cell death or they may be citrullinated NETs as has been shown by Kenny et al (2017).

   1. The authors suggest careful validation for cross reactivity in samples under native and mild detergent condition, e.g. in serum samples. It would be good if this validation be performed in the current study.

   This is an informative experiment for users of this antibody and we thank the reviewers for recommending to include it. In preparation for using the antibody with blood samples, we performed some preliminary experiments to address whether the antibody was suitable for a simple ELISA in the presence of plasma/serum. We have now added Figure 3-figure supplement 4 addressing this. In this experiment, we examined the ability of 3D9 to react with healthy donor plasma and serum alone by direct ELISA (Figure 3-figure supplement 4 (A) (biological replicate n=1). 3D9 strongly reacted with plasma but not with serum. We also examined detection by 3D9 of western blotted serum and plasma proteins separated by SDS page under reducing and non reducing conditions (Figure 3-figure supplement 4 (B) and found that 3D9 also detected proteins, of higher molecular weight than cleaved H3, in plasma but not in serum. We conclude that the direct detection of NETs by 3D9 in plasma containing samples and thus whole blood is not possible due to cross reaction with a plasma protein(s). Based on this we caution the use of 3D9 in serum containing samples alone as it is challenging to ensure all plasma proteins are removed and instead, we advocate for the use of sandwich and colocalization approaches.

   In the main text we have included the following lines 152-156

   However, when we performed preliminary experiments to see if the antibody had the potential to work in blood samples, we observed a strong reaction of the antibody with a plasma protein(s), but not with serum-protein(s) as determined by direct ELISA and western blot (Figure 3-figure supplement 4. Therefore, 3D9 is not suitable for direct detection of NETs in biological fluids that may contain plasma proteins.

   In the discussion Line 314-319

   Furthermore, a preliminary investigation revealed 3D9 reacts with a plasma protein(s) in ELISA and western blot - albeit of a higher molecular weight – (Figure 3-figure supplement 4A) and thus, care should be taken in the design of future assays and selection of sample when detecting cleaved H3, NETs or vital NETs under native and mild detergent conditions. In particular, 3D9 is not suitable for direct detection of NETs in complex biological fluids and a sandwich approach or colocalization with a neutrophil granule protein is critical.

   And appropriate text has been added to the methods section line 568-576

   The same approach was used to examine 3D9 interactions with immobilised serum and plasma proteins. Plasma was isolated from whole blood collected with S. Monovette sodium citrate tubes (Sarstedt). Whole blood was centrifuged at low speed to minimise cell lysis (150 xg, 20 min with no brake). Prostaglandin E1 (1 µm) was added to inhibit platelet activation and samples were further centrifuged at 650 xg (8 min) to collect cell free plasma and further centrifuged at 2000 xg (10 min) before being aliquoted and stored at -80 °C. Serum was isolated by collection of whole blood in S. Monovette serum tubes (silicate clotting activator) and incubation with gentle rotation for 30 min at RT before centrifugation at 2000 xg at 4°C (10 min) and collection of serum.

   Reviewer #2 (Public Review):

   In this study, Tilley et al. identified cleavage of histone H3 at R49 (H3R49) as a candidate marker of NETs and generated a H3R49 cleavage site monoclonal antibody (termed 3D9) as a potential tool to detect NETs in human samples. The antibody was validated using both in vitro assays and human tissues. Using human neutrophils, they demonstrated that 3D9 detects NETs induced by both ROS-dependent (i.e. PMA, heme in TNF primed neutrophils and Candida albicans) and ROS-independent stimuli (i.e. using the toxin nigericin from Streptomyces hygroscopicus). To demonstrate the specificity of 3D9, they first showed that 3D9 distinguishes NETs from other activated leucocytes in PBMCs after stimulation with PMA or nigericin. These studies also found that the anti-chromatin antibody PL2.3, broadly used to detect NETs, is not specific for NETs as it also stains nuclei of activated PBMCs. Moreover, they showed that 3D9 distinguishes NETs from other forms of neutrophil death, including spontaneous apoptosis, necroptosis induced by TNFα stimulation in the presence of a SMAC, and necrosis induced by the staphylococcal toxin α-haemolysin. Interestingly, these studies also found that the PL2.3 antibody is not specific for NETs, but also stains apoptotic cells. The detection of apoptotic cells as well as activated PBMCs by PL2.3 importantly questions previous studies in which PL2.3 has been used to specifically detect NETs. Finally, they showed that 3D9 labels neutrophils in inflamed human tissues, including tonsil, kidney, appendix and gallbladder. However, colocalization of 3D9 with other anti-NET antibodies (PL2.3 and anti-citrullinated histone H3, H3cit) is not impressive and particularly poor for H3cit. Since 3D9 detects both ROS-dependent and ROS-independent NETs, the authors concluded that histone H3 cleavage at R49 is a general feature of human NET formation. Therefore, the authors propose that the antibody 3D9 is a new tool to detect and quantify NETs in human samples.

   Some conclusions of this paper are well supported by data. However, the conclusion that this novel antibody can detect any form of human NETs is not demonstrated. The study needs a better validation of 3D9 using broader NET-inducing stimuli relevant for human diseases. In addition, this study needs to confirm the specificity of the anti-NET antibodies in tissues. Thus, for some aspects of this work, some data need to be clarified and extended.

   1. To validate that 3D9 detects all forms of NETs, the study used well-described NET inducers that generate ROS-dependent and ROS-independent NETs. PMA and nigericin are very useful in this regard because these are potent stimuli that induce NETs using either pathway (ROS-dependent and ROS-independent, respectively). However, neither PMA nor nigericin are stimuli relevant for human pathology. In particular, S. hygroscopicus (the source of nigericin) is not a human pathogen. The inclusion of NETs induced by heme in TNF primed neutrophils (a stimulus relevant for NETs in malaria) and by C. albicans is certainly an important complement for the study of 3D9 in ROS-dependent NETs associated with human diseases. However, the study is missing the analysis of ROS-independent NETs induced by stimuli associated with human illnesses.

   We thank the reviewer for the recommendation to look at additional stimuli. We apologize in advance as we have noted a mistake in the manuscript designating the disease relevant stimulus heme as NOX dependent. This has been corrected (line 302) to “both NOX dependent (PMA) and NOX independent (heme;nigericin) stimuli result in NETs”. Heme induces NETs in CGD patients lacking a functioning NADPH oxidase complex but intracellular ROS scavengers inhibit heme induced NETs (Knackstedt et al 2019 DOI: 10.1126/sciimmunol.aaw0336) -thus this stimulus is ROS dependent while being independent of the NOX induced ROS burst, with the ROS likely being induced by a chemical reaction with heme itself. For this possible confusion around the designation of ROS dependency we categorise our stimuli based on their NOX (in)dependency.

   In this manuscript we tested two NOX dependent and two NOX independent stimuli: these are the mitogen PMA and infections with Candida albicans (NOX dependent), and heme plus TNF and nigericin (NOX independent). C. albicans is a medically relevant pathogen. Heme, a product abundant in malaria, and TNF are a model for cytokine activation in Plasmodium infections. Nigericin, while not yet used in humans, is used in veterinary medicine and is being examined as an anti-tumor agent for the treatment of human colon cancer among others (DOI 10.1158/1535-7163.MCT-17-0906) . Both PMA and Nigericin are strong inducers of the two canonical pathways to NET formation – NOX dependent and independent respectively.

   In our manuscript we do not claim that 3D9 can detect all NETs. Indeed, we attempt to demonstrate, both with Candida albicans and with the tissue sections, that there can be a variety of NETs of different flavours both in vitro and ex vivo. Like H3cit, not all NETs are cleaved at H3R49. However, histone cleavage at this site is a common feature in all the stimuli we examined and thus we believe it can be broadly used to detect NETs from a wide range of stimuli – although, not all NETs generated may be detected by this antibody. We have expanded on the observation of 3D9 negative NETs in response to Candida albicans.

   line 282-285

   Like citrullination, not all NETs contain H3 cleaved at H3R49. With Candida albicans, some NET-like structures were not 3D9 positive (Figure 6). These may be remnants of other forms of cell death or they may be citrullinated NETs as has been shown by Kenny et al (2017).

   We removed the description that “histone H3 cleavage at R49 is a general feature of human NET formation” and replaced it with “histone H3 cleavage at R49 is a common feature in human NET formation” (line 304) to convey the regular occurrence in of Histone H3 cleavage in NETosis with wide ranging stimuli, rather that it being an all ways present feature all NETs.

   1. The number of diseases and stimuli associated with NETs is growing every day and it is unlikely that only two pathways defined by artificial stimuli (i.e. PMA and nigericin or calcium ionophores) can cover all mechanisms activated in humans to induce NETs. In the host, the neutrophil-pathogen interface is more complex than PMA or nigericin. For example, toxin-free S. aureus is known to induce NETs (J Cell Biol 2007, 176:231-41), but toxins released by S. aureus are also potent inducers of necrosis. Which of these stimuli may dominate during infection with S. aureus is unclear, underscoring the complexity of correlating biochemical features found in well-controlled NETs induced in vitro with changes in neutrophils found in tissues from human diseases. It is understandable that for the initial validation of 3D9, it is not possible to cover all potential inducers of NETs. However, there are diseases in which NETs have had a major impact and created new paradigms. NETs associated with malaria and C. albicans are interesting, but only cover a fraction of NET-inducing stimuli within a subset of diseases (i.e. infectious diseases). Importantly, autoimmune diseases are certainly one of the major group of diseases in which the study of NETs have had the highest impact. In some cases, NETs are considered the driving cause of these illnesses. The analysis of NETs induced by autoantibody-antigen immune complexes (specifically anti-RNP and rheumatoid factor) would be needed to increase confidence in the validation of 3D9.

   In our hands we have not been able to produce NETs using anti-RNP complexes but we accept this criticism of the limited stimuli we have used.

   We hope, through this publication, to make the antibody available to the research community to explore which types of NETs may be the aggressors or modulators in varying disease contexts and in doing so, as a community, we can assess the usefulness of 3D9 as we have done for H3cit antibodies.

   1. When comparing the specificity of antibodies used to detect NETs, the study should include a similar analysis of 3D9, PL2.3, and H3cit. This is significant to interpret their different patterns of staining in tissues. Figure 7 and Figure 8-figure supplement 1 are missing the analysis of H3cit.

   We thank the reviewers for their critical understanding of the field and their suggestion to add a comparison of staining with H3 cit to Figures 7 and 8. Including comparison stains with PL2.3 will always be useful in ensuring we are examining all areas of putative NETs, and staining for histone citrullination and/or histone H3R49 cleavage will add specificity as they detect processes that take place during NET formation. However, in this study, we did not set out to qualitatively compare H3cit as a surrogate marker of NETs to our new antibody. It is possible that other mechanisms of cell death may involve citrullination but our aim in this study was to characterise 3D9 staining behaviour in varied contexts - not H3cit behaviour, which is an ongoing area of debate in the research community - Boeltz et al., 2019 (DOI:10.1038/s41418-018-0261-x).

   Through our investigation into the precise site of the histone cleavage we determined that, by nature of the epitopes of 3D9 and the most commonly used abcam H3cit R2, R8, R17 antibody, they cannot stain the same individual histone. Thus, in theory, examining the staining patterns of 3D9 and H3cit, side by side, as we have done for PL2.3, will not provide insight as to which NETs or modes of cell death involve citrullination – only whether H3cit is present or absent under the conditions of our experiment.

   The finding, that individual neutrophils or NETs showed such ‘either or’ - H3cit or cleaved H3 – characteristics as seen in the tissue sections was surprising and intriguing and this has led us to propose that different types of NETs, specifically those that are more proteolytically processed at the H3 N-terminal, or as a reviewer has suggested, different degrees of citrullination, may be being distinguished by 3D9. However, this will need further investigation that goes beyond the scope of this manuscript and will need to examine other citrullination sites, e.g. H4cit or even pan citrullination to determine if 3D9 distinguishes NETs with more or less citrullination.

   What we have developed in this manuscript is an additional antibody, another tool in the arsenal of NET researchers, to look at NETs (that may or may not also have been citrullinated at H3 R2, 8 or 17 or at a different histone during the process of NET formation). It is another tool for detection. We examine H3cit, 3D9 and H2B co-staining in our final tissue section figures and in doing so demonstrate the diversity of NETs in tissues and highlight how it is important to follow the histological definition of NETs and not rely on a single marker. Individually each antibody has failings but researchers who make use of multiple methods to assess NETs can be confident of their assessment of NETs in their studies.

   1. Among the different forms of neutrophil death used to validate 3D9, the study should also include pyroptosis. This form of cell death shares some common effector pathways with NETs. It is therefore important to demonstrate that 3D9 can distinguish NETosis and pyroptosis.

   We thank the reviewer for this suggestion and agree that pyroptotic cell death would be interesting to examine in neutrophils with respect to distinguishing different forms of neutrophil cell death from NETosis with 3D9. However the variation in neutrophil responses to pyroptotic stimuli would warrant a deeper investigation beyond the scope of this manuscript. Unlike macrophages, human neutrophils are largely resistant to pyroptotic cell death when inflammasome pathways are activated (Chen et al 2014, Chen et al 2018, Karamakar et al 2020, Kovacs et al (Cell Rep. 2020 Jul 28; 32(4): 107967. ). Stimulation or infection of neutrophils with macrophage pyroptotic stimuli, intracellular pathogen or pathogen signals within the cytosol, elicits some of the characteristics of pyroptosis but not the typical cell morphology. In macrophages pyroptosis is preceded by inflammasome (canonical and non-canonical) activation resulting in activation of specific caspases, caspase mediated activation of Gasdermin D and assembly of a gasdermin pore in the plasma membrane resulting in disruption of osmostic regulation and rapid cell death and the concomitant release of cytokines including IL-1b/ IL-18 that are matured through caspase activity. In contrast, in neutrophils, inflammasome activation by infection with intracellular pathogens can produce diverse outcomes in terms of cell death depending on the specific intracellular pathogen or signal. While Salmonella triggers release of caspase 1 activated IL-1beta from neutrophils (Chen et al 2014, https://doi.org/10.1016/j.celrep.2014.06.028) it does not undergo pyroptosis. Infection with Citrobacter rodentium triggers release of IL-beta but goes on to produce NETs in a manner that is inflammasome driven and caspase dependent (Chen et al 2018 Science Immunology) (https://immunology.sciencemag.org/content/3/26/eaar6676.long). Most recently, Kovacs et al have shown that instead of oligermising at the plasma membrane, Gasdermin D forms pores in azurophilic granules and autophagolysome and that IL-1b release involves autophagy machinery. Thus, given the established variability in neutrophil pyroptotic-like responses and their inherent resistance to undergo typical pyroptotic cell death, it is unclear what value examining an individual pyroptotic stimulus will add in the context of characterising 3D9 staining of NETs. Indeed, in Fig 6 we use Nigericin as a stimulus to induce NETs which are detected by 3D9 and it is a known activator of the inflammasome pathway and pyroptosis in macrophages but produces NETs in neutrophils.

   To make it clear that we have not exhaustively looked at all forms of neutrophil cell death we have modified the section title, figure titles and main body of text referring to the forms of neutrophil cell death.

   line 233-234 3D9 distinguishes NETosis from apoptotic, necroptotic and necrotic cell death in neutrophils.

   Line 273-274 It distinguishes netotic neutrophils from apoptotic, necrotic and necroptotic neutrophils in vitro.

   1. There is some evidence that H3 can be citrullinated at R49 https://www.caymanchem.com/literature/methods-in-citrullination-and-analysis-of-recombinant-human-histones. This modification would likely make H3 resistant to cleavage at this site. This may explain that the detection of the H3 fragment importantly decreases at 180 mins in NETs induced by A23187 (Kenny et al, Elife. 2017, 6:e24437, Figure 7), which is a potent inducer of histone citrullination. Thus, an alternative explanation to the lack of colocalization between H3cit and 3D9 in tissues is that these antibodies are detecting different types of NETs. H3cit may stain NETs in which citrullination is dominant, making H3 resistant to cleavage. In contrast, 3D9 may detect NETs in which H3 citrullination is absent or minimal (such as NETs induced by PMA, heme in TNF primed neutrophils, C. albicans and nigericin. Elife 6. 10.7554/eLife.24437) and therefore, H3 cleavage is fully efficient.

   We agree with the reviewer that it is possible that different types of NETs are being detected by 3D9 and H3 cit. In our initial submission we proposed that 3D9 might display a preference for proteolytically processed or ‘mature’ NETs. This is not at odds, and is in fact complementary to the hypothesis presented by this reviewer, that 3D9 may detect NETs is which citrullination is minimal or absent. We have modified our discussion to reflect this.

   Line 292-294 Thus, we propose that 3D9 will allow broad detection of NETs induced by varied stimuli but that it may display a preference for more mature or proteolytically processed NETs or NETs that are citrullinated to a lesser degree or not at all.

   1. Another possibility of the lack of colocalization between 3D9 and H3cit in inflamed tissues is that analogous to PL2.3, H3cit is not specific for NETs and may be similarly detecting activated cells or some other forms of neutrophil death. Indeed, previous studies have shown that H3cit is generated during neutrophil activation and apoptosis (Sci Transl Med. 2013, 5:209ra150). If the authors show that PL2.3 and H3cit are not specific to detect NETosis, they should discuss the implications of these findings regarding all publications that have used these antibodies to mechanistically link NETs with specific human diseases.

   We thank the reviewer for this very interesting comment and the suggestion to compare H3cit and 3D9 staining in response to different stimuli in more depth. However, first we must clarify that PL2.3 has never been used by us as a specific marker of NETs. It is a general chromatin stain but when samples are minimally permeablised it binds with greater intensity to decondensed chromatin. A neutrophil marker such as anti-NE is always needed to confirm the presence of NETs if using PL2.3. Using PL2.3 alone is not good evidence of NETs.

   We are aware that H3 citrullination can also occur in other pathways of cell activation and most notably in apoptosis and this is another reason for using sandwich or colocalization approaches with a granule protein to detect NETs. There is much literature that has been presented using only H3cit, particularly during the early work in the field. However, as of late, and thanks to robust discussion in the community, H3 cit alone is rarely presented as convincing evidence of NETs. However, with this in mind, we investigated 3D9 staining using varied modes of neutrophil cell death. Its failure to stain apoptotic, necroptotic and necrotic neutrophils suggested that it was, to date, our best candidate for a NET specific marker. However, to err on the side of caution, we still advocate for the use of sandwich and colocalization approaches until the exclusive NET specificity of cleavage at H3R49 can be established using 3D9 or subsequently developed site specific cleavage antibodies for different assays.

   1. In the analysis of inflamed tissues, it is assumed that finding neutrophils only means NETs. This gives the impression that other forms of neutrophil death have disappeared in humans. To validate the anti-NET antibodies in tissues, it will be useful to include co-staining with markers of other forms of neutrophil death. This analysis will help to increase confidence that 3D9, PL2.3 or H3cit are more likely to detect NETs in tissues rather than other forms of neutrophil death. This is important because in vitro studies are not analogous to in vivo processes.

   We thank the reviewer for raising this important point and we have added to the text to clarify that other forms of neutrophil cell death will likely be occurring in inflamed tissues.

   line 246-247 Neutrophils are recruited to sites of inflammation and depending on the context or the surrounding stimuli, they may undergo varied forms of cell death.

   While not presented in our figures, there is also a DNA stain. We attempted to add a 5th marker for apoptosis using anti-cleaved caspase 3 but the results were not interpretable. However, in vitro we clarified that 3D9 staining cells were negative for cleaved caspase 3, an apoptotic marker (Figure 8-figure supplement 2).

   1. NETs are believed to be pathogenic because this process has been associated with specific pathologies, e.g. infection, autoimmunity and cancer. However, the detection of NETs in any inflamed tissue suggests that this process is driven in response to any non-specific inflammatory stimuli. To clarify this discrepancy, it will be useful to know if the inflamed tissues are from specific diseases associated with the production of NETs.

   We thank the reviewer for highlighting this. Indeed, much of the literature refers to the pathogenic nature of NETs, but, as with most actions of the immune system, there is usually a balance or threshold, a level of inflammation and possibly NET formation that is appropriate for the host defence which later requires the resolution of inflammation and repair. Unfortunately, in this paper the samples are not from diseases where NETs are considered one of the drivers of pathology – e.g. lupus, and are instead from tissues characterised as ‘inflamed’. The appendix and gall bladder tissues in this study come from patients with appendicitis (the gall bladder was extracted at the same time). The tonsil and kidney samples come from a commercial source and are only noted as ‘inflamed’. A future line of research could be to compare citrullinated histones and 3D9 staining in diseased tissues and to assess if one type of NETs is more dominant in specific diseases or disease states.

   We added further detail of the descriptions to figure legends and methods.

   line 1086 Figure 9….(A) human tonsil, denoted ‘normal’ by commercial provider but showing infiltration of neutrophils demonstrating an inflammatory event. (B & C) human kidney, denoted ‘inflamed’ by commercial provider.

   line 1089 Figure 10. Comparison of Clipped H3, H3cit & H2B staining in the gallbladder from an appendicitis patient.

   Line 1094 Figure 11. Comparison of Clipped H3, H3cit & H2B staining in the appendix of an appendicitis patient

   Methods Line 722-725 Human tonsil (denoted normal but showing neutrophil infiltration) and inflamed kidney paraffin tissue blocks were purchased from AMSbio. Inflamed tissue from a gallbladder and appendix was obtained from archived leftover paraffin embedded diagnostic appendicitis samples.

   Reviewer #3 (Public Review):

   The authors have successfully characterized a specific mechanisms that occurs during NET-formation: a NET-specififc histone H3 cleavage event. The monoclonal antibody 3D9 detects evidence of the proteolytic events that occur in NETosis -the proteolytic signature, histone cleavage at H3R49. Based on this finding they have developed a new method to detect and quantify NETs and differentiate NET-formation from apoptosis or necroptosis. The method can be used to stain mixed cell populations or also human tissue material.

   The major strength of this manuscript is that it gives mechanistical insight into NET-formation and presents at the same time a novel techique that shows several advantages compared to existing techniques.

   The methods and results are presented in detail and well controlled and presented.

   We thank the reviewer for their consideration and appraisal of our manuscript.

   Read the original source
3. 

   eLife

   Jun 28, 2022

   Evaluation Summary:

   This study describes a new antibody to identify human neutrophil extracellular traps (NETs) also in tissue samples. The paper might not only introduce an important novel tool for many areas of biomedical research, but it also touches cell biological questions of importance. The usefulness of the NET-specific antibody is impressively developed in the paper, while the mechanistic concepts are not yet fully established.

   (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 agreed to share their name with the authors.)

   Read the original source
4. 

   eLife

   Jun 28, 2022

   Reviewer #1 (Public Review):

   Neutrophil extracellular traps (NETs) are defined as structures containing extracellular DNA co-localizing with granule-derived proteins, such as neutrophil elastase, and histones. While in in vitro assays a variety of protocols have been described to unambiguously detect and quantify neutrophil extracellular traps (NETs), in ex vivo tissue samples, quantification and demarcation of NETs from the remnants other forms of neutrophil cell death such as necrosis is still challenging. The current manuscript by Tilley and colleagues describes a novel tool to perform that important task. The authors have discovered that human histone H3 is processed by serine proteases at a specific cleavage site during NET formation. They created a mouse monoclonal antibody to this cleaved histone H3 and assessed its performance …

   Reviewer #1 (Public Review):

   Neutrophil extracellular traps (NETs) are defined as structures containing extracellular DNA co-localizing with granule-derived proteins, such as neutrophil elastase, and histones. While in in vitro assays a variety of protocols have been described to unambiguously detect and quantify neutrophil extracellular traps (NETs), in ex vivo tissue samples, quantification and demarcation of NETs from the remnants other forms of neutrophil cell death such as necrosis is still challenging. The current manuscript by Tilley and colleagues describes a novel tool to perform that important task. The authors have discovered that human histone H3 is processed by serine proteases at a specific cleavage site during NET formation. They created a mouse monoclonal antibody to this cleaved histone H3 and assessed its performance as a tool to detect NETs in vitro and ex vivo.

   The paper is well-structured and written, thus presenting a valuable contribution to the field. There are some open issues with the manuscript which are not clear at this point:

   1. One major point are the dynamics when this clipping occurs and if it occurs extra- or intracellularly. The authors have a used a serine protease inhibitor, AEBSF, which not only inhibits histone clipping but also NET formation and nuclear decondensation itself. I am therefore not sure if the conclusion can be drawn that histone H3 clipping is an intracellular event and "serine proteases cleave the N-terminus of H3 early during NET formation." This is also in open conflict to the study by Pieterse et al. (Ann Rheum Dis 2018) who demonstrated prevention of histone clipping by serine protease inhibitors working exclusively outside the cell.
   2. Along these lines it is also confusing that the staining by an anti-citH3 Ab and 3D9 seems to be mutually exclusive. The authors mention this in the discussion and explain that 3D9 "may display a preference for more mature or proteolytically processed NETs." This seems to be hard to align with their claim that H3 is clipped early during NET formation. It would be important to show if citH3+ NETs progress into 3D9+ NETs. If this is not the case, that would potentially render a large part of the literature that has used citH3 staining for the detection of "NETs" useless.
   3. Non-suicidal pathways of NET formation were described, where parts of the nucleus are extruded but the cell remains intact and basal cellular functions of neutrophils are still carried out. These" vital NETs" are not addressed in the manuscript.
   4. The authors show that neutrophils stimulated with C. albicans released NETs not bound by 3D9, which remains unexplained.
   5. The authors suggest careful validation for cross reactivity in samples under native and mild detergent condition, e.g. in serum samples. It would be good if this validation be performed in the current study.

   Read the original source
5. 

   eLife

   Jun 28, 2022

   Reviewer #2 (Public Review):

   In this study, Tilley et al. identified cleavage of histone H3 at R49 (H3R49) as a candidate marker of NETs and generated a H3R49 cleavage site monoclonal antibody (termed 3D9) as a potential tool to detect NETs in human samples. The antibody was validated using both in vitro assays and human tissues. Using human neutrophils, they demonstrated that 3D9 detects NETs induced by both ROS-dependent (i.e. PMA, heme in TNF primed neutrophils and Candida albicans) and ROS-independent stimuli (i.e. using the toxin nigericin from Streptomyces hygroscopicus). To demonstrate the specificity of 3D9, they first showed that 3D9 distinguishes NETs from other activated leucocytes in PBMCs after stimulation with PMA or nigericin. These studies also found that the anti-chromatin antibody PL2.3, broadly used to detect NETs, is …

   Reviewer #2 (Public Review):

   In this study, Tilley et al. identified cleavage of histone H3 at R49 (H3R49) as a candidate marker of NETs and generated a H3R49 cleavage site monoclonal antibody (termed 3D9) as a potential tool to detect NETs in human samples. The antibody was validated using both in vitro assays and human tissues. Using human neutrophils, they demonstrated that 3D9 detects NETs induced by both ROS-dependent (i.e. PMA, heme in TNF primed neutrophils and Candida albicans) and ROS-independent stimuli (i.e. using the toxin nigericin from Streptomyces hygroscopicus). To demonstrate the specificity of 3D9, they first showed that 3D9 distinguishes NETs from other activated leucocytes in PBMCs after stimulation with PMA or nigericin. These studies also found that the anti-chromatin antibody PL2.3, broadly used to detect NETs, is not specific for NETs as it also stains nuclei of activated PBMCs. Moreover, they showed that 3D9 distinguishes NETs from other forms of neutrophil death, including spontaneous apoptosis, necroptosis induced by TNFα stimulation in the presence of a SMAC, and necrosis induced by the staphylococcal toxin α-haemolysin. Interestingly, these studies also found that the PL2.3 antibody is not specific for NETs, but also stains apoptotic cells. The detection of apoptotic cells as well as activated PBMCs by PL2.3 importantly questions previous studies in which PL2.3 has been used to specifically detect NETs. Finally, they showed that 3D9 labels neutrophils in inflamed human tissues, including tonsil, kidney, appendix and gallbladder. However, colocalization of 3D9 with other anti-NET antibodies (PL2.3 and anti-citrullinated histone H3, H3cit) is not impressive and particularly poor for H3cit. Since 3D9 detects both ROS-dependent and ROS-independent NETs, the authors concluded that histone H3 cleavage at R49 is a general feature of human NET formation. Therefore, the authors propose that the antibody 3D9 is a new tool to detect and quantify NETs in human samples.

   Some conclusions of this paper are well supported by data. However, the conclusion that this novel antibody can detect any form of human NETs is not demonstrated. The study needs a better validation of 3D9 using broader NET-inducing stimuli relevant for human diseases. In addition, this study needs to confirm the specificity of the anti-NET antibodies in tissues. Thus, for some aspects of this work, some data need to be clarified and extended.

   1. To validate that 3D9 detects all forms of NETs, the study used well-described NET inducers that generate ROS-dependent and ROS-independent NETs. PMA and nigericin are very useful in this regard because these are potent stimuli that induce NETs using either pathway (ROS-dependent and ROS-independent, respectively). However, neither PMA nor nigericin are stimuli relevant for human pathology. In particular, S. hygroscopicus (the source of nigericin) is not a human pathogen. The inclusion of NETs induced by heme in TNF primed neutrophils (a stimulus relevant for NETs in malaria) and by C. albicans is certainly an important complement for the study of 3D9 in ROS-dependent NETs associated with human diseases. However, the study is missing the analysis of ROS-independent NETs induced by stimuli associated with human illnesses.

   2. The number of diseases and stimuli associated with NETs is growing every day and it is unlikely that only two pathways defined by artificial stimuli (i.e. PMA and nigericin or calcium ionophores) can cover all mechanisms activated in humans to induce NETs. In the host, the neutrophil-pathogen interface is more complex than PMA or nigericin. For example, toxin-free S. aureus is known to induce NETs (J Cell Biol 2007, 176:231-41), but toxins released by S. aureus are also potent inducers of necrosis. Which of these stimuli may dominate during infection with S. aureus is unclear, underscoring the complexity of correlating biochemical features found in well-controlled NETs induced in vitro with changes in neutrophils found in tissues from human diseases. It is understandable that for the initial validation of 3D9, it is not possible to cover all potential inducers of NETs. However, there are diseases in which NETs have had a major impact and created new paradigms. NETs associated with malaria and C. albicans are interesting, but only cover a fraction of NET-inducing stimuli within a subset of diseases (i.e. infectious diseases). Importantly, autoimmune diseases are certainly one of the major group of diseases in which the study of NETs have had the highest impact. In some cases, NETs are considered the driving cause of these illnesses. The analysis of NETs induced by autoantibody-antigen immune complexes (specifically anti-RNP and rheumatoid factor) would be needed to increase confidence in the validation of 3D9.

   3. When comparing the specificity of antibodies used to detect NETs, the study should include a similar analysis of 3D9, PL2.3, and H3cit. This is significant to interpret their different patterns of staining in tissues. Figure 7 and Figure 8-figure supplement 1 are missing the analysis of H3cit.

   4. Among the different forms of neutrophil death used to validate 3D9, the study should also include pyroptosis. This form of cell death shares some common effector pathways with NETs. It is therefore important to demonstrate that 3D9 can distinguish NETosis and pyroptosis.

   5. There is some evidence that H3 can be citrullinated at R49 https://www.caymanchem.com/literature/methods-in-citrullination-and-analysis-of-recombinant-human-histones. This modification would likely make H3 resistant to cleavage at this site. This may explain that the detection of the H3 fragment importantly decreases at 180 mins in NETs induced by A23187 (Kenny et al, Elife. 2017, 6:e24437, Figure 7), which is a potent inducer of histone citrullination. Thus, an alternative explanation to the lack of colocalization between H3cit and 3D9 in tissues is that these antibodies are detecting different types of NETs. H3cit may stain NETs in which citrullination is dominant, making H3 resistant to cleavage. In contrast, 3D9 may detect NETs in which H3 citrullination is absent or minimal (such as NETs induced by PMA, heme in TNF primed neutrophils, C. albicans and nigericin. Elife 6. 10.7554/eLife.24437) and therefore, H3 cleavage is fully efficient.

   6. Another possibility of the lack of colocalization between 3D9 and H3cit in inflamed tissues is that analogous to PL2.3, H3cit is not specific for NETs and may be similarly detecting activated cells or some other forms of neutrophil death. Indeed, previous studies have shown that H3cit is generated during neutrophil activation and apoptosis (Sci Transl Med. 2013, 5:209ra150). If the authors show that PL2.3 and H3cit are not specific to detect NETosis, they should discuss the implications of these findings regarding all publications that have used these antibodies to mechanistically link NETs with specific human diseases.

   7. In the analysis of inflamed tissues, it is assumed that finding neutrophils only means NETs. This gives the impression that other forms of neutrophil death have disappeared in humans. To validate the anti-NET antibodies in tissues, it will be useful to include co-staining with markers of other forms of neutrophil death. This analysis will help to increase confidence that 3D9, PL2.3 or H3cit are more likely to detect NETs in tissues rather than other forms of neutrophil death. This is important because in vitro studies are not analogous to in vivo processes.

   8. NETs are believed to be pathogenic because this process has been associated with specific pathologies, e.g. infection, autoimmunity and cancer. However, the detection of NETs in any inflamed tissue suggests that this process is driven in response to any non-specific inflammatory stimuli. To clarify this discrepancy, it will be useful to know if the inflamed tissues are from specific diseases associated with the production of NETs.

   Read the original source
6. 

   eLife

   Jun 28, 2022

   Reviewer #3 (Public Review):

   The authors have successfully characterized a specific mechanisms that occurs during NET-formation: a NET-specififc histone H3 cleavage event. The monoclonal antibody 3D9 detects evidence of the proteolytic events that occur in NETosis -the proteolytic signature, histone cleavage at H3R49. Based on this finding they have developed a new method to detect and quantify NETs and differentiate NET-formation from apoptosis or necroptosis. The method can be used to stain mixed cell populations or also human tissue material.

   The major strength of this manuscript is that it gives mechanistical insight into NET-formation and presents at the same time a novel techique that shows several advantages compared to existing techniques.

   The methods and results are presented in detail and well controlled and presented.

   Read the original source
7. 

   Version published to 10.1101/2021.03.15.434949v3 on bioRxiv

   Mar 24, 2021
8. 

   Version published to 10.1101/2021.03.15.434949v2 on bioRxiv

   Mar 19, 2021
9. 

   Version published to 10.1101/2021.03.15.434949v1 on bioRxiv

   Mar 16, 2021