Neutrophil extracellular traps have auto-catabolic activity and produce mononucleosome-associated circulating DNA

Sufficient evidence now exists regarding NET as one (previously unappreciated) source of cirDNA to indicate their strong potential as a source of biomarkers for various inflammatory disorders and other pathologies. However, no study to date has directly proven that either chromatin or NET DNA fibers degrade into mono-N DNA which could be released into the blood circulation without implying cell death.

The gHMW DNA can be degraded in serum into mononucleosomes

For more than two decades, the prominence of mononucleosomes in cirDNA extract encouraged the postulation that apoptosis was the main mechanism of release into blood circulation [4, 7, 41]. Previous efforts to study genomic DNA degradation, however, offered conflicting conclusions regarding the role of apoptosis and apoptotic nucleases in internucleosomal DNA fragmentation and the production of mononucleosomes. On the one hand, certain studies report that cirDNA originates in the apoptotic DNA fragments generated by caspase-activated DNase (CAD), because the cfDNA fragment sizes generally approximate to a mononucleosomal size [42]. Koyama et al. [14] propose that cirDNA can originate from both apoptotic and necrotic cells and that CAD, DNase γ, and DNase I are responsible for the generation of DNA fragments. On the other hand, a different study showed that apoptosis in CAD-deficient mice caused DNA fragmentation equivalent to that in WT mice; thus concluding that apoptosis-associated CAD is therefore not essential for DNA degradation in vivo, they suggested that phagocytes are rather as the main actor in DNA fragmentation [43]. Finally, yet another study in mice has shown that, in apoptosis, CAD is insufficient for degrading chromatin and requires DNase1L3 to degrade it into mononucleosomes [44]. Given these contradictory data and the uncertain role of apoptosis in DNA fragmentation in cell culture or in vivo in mice, we preferred to focus on the general ability of human blood to fragment DNA regardless of the mechanisms of cell death [7]. Our current study first investigated DNA degradation by adding gHMW DNA from a cell lysate to the serum or plasma of a HI under physiological conditions meaning that no cells (and thus no apoptosis or any other cell death events) could be involved in DNA degradation. Our study design implies the use of gHMW DNA and blood donors of opposite genders, and explores the gHMW DNA topology under degradation by examining the SRY sequence. An accumulation of DNA fragments associated with mononucleosomes showed that, in the absence of apoptosis, exposure to serum nucleases alone is sufficient for the effective degradation of gHMW DNA down to mono-N DNA fragments. Data also confirmed mono-N as the most stabilized structure in blood with which cirDNA can associate [31, 45].

In previous studies, we and other groups have scrutinized the size distribution of cirDNA in the plasma of HI and cancer patients [31, 45]. Among those findings, a sWGS assay revealed the predominance of chromatosome/mononucleosome-associated cirDNA fragments and the presence in the plasma of HI and cancer patients of a minor population corresponding to dinucleosome-associated cirDNA. A combination of double- and single-strand DNA library preparations (DSP and SSP) and nested qPCR (N-qPCR) enabled the estimation of the wide size distribution of cirDNA fragments inserted in mononucleosomes, di-nucleosomes, and chromatin of higher molecular size (> 1000 bp) as 67.5–80%, 9.4–11.5%, and 8.5–21.0%, respectively [31]. We found a striking reproducibility of the cirDNA size profiles in HI [31], in that the maximal size and frequency of the major peak and all subpeaks varied by merely 1 or 2 bp and by 2–20%, respectively.

In the present study, we observed a characteristic chromatin pattern after gHMW DNA degradation by serum nucleases, in which the most prominent peak is of mono-N cfDNA, and the less prominent peak of di-N cfDNA (Fig. 2B) as observed for cirDNA in vivo. We observed a very rapid degradation and the depletion of long DNA fragments resulted in the formation of oligonucleosomes, making them the main substrate for DNase attacks, and leading to the subsequent accumulation of mono-N and di-N (after 2 h incubation). This also led to the accumulation of short DNA fragments in the 40–160 bp range, indicating the dynamic action of serum nucleases on the internucleosomal DNA, the DNA linked to the histone H1, and the 14 DNA base pairs exposed at the surface of the nucleosome core particle at a ~ 10 bp periodicity [31, 45, 46].

As determined by sWGS, the analysis of the DNA fragment fractions corresponding to mono-N, di-N, and tri-N confirms these conclusions and adds to our quantitative assessment of the kinetics of gHMW DNA degradation in serum. When comparing the proportions of mono-N, di-N, or tri-N DNA, data revealed the higher resistance of mononucleosomes to DNase attacks, as compared to oligonucleosomes. We therefore conclude that in normal physiological conditions blood might be capable of degrading gHMW DNA, and especially nicking and degrading the internucleosomal DNA; this leads to the accumulation of mononucleosomes/chromatasomes.

Activated neutrophils produce HMW DNA fragments ranging from 1500 to 30,000 bp

Although there are several reports based on the visualization of NET in vitro by fluorescent microscopy, which demonstrates that NET consist of the DNA fibers, no studies have analyzed the structure and length of DNA released by activated neutrophils and its association with cirDNA production.

In our present study, we describe the accumulation of HMW DNA fragments and the increase of NET markers in the supernatant of ex vivo activated neutrophils. This would indicate the formation of these DNA fibers decorated with NET-specific proteins, such as NE and MPO. Our data confirms these results by revealing the significant correlations of the DNA markers (based on size and amount) with MPO and NE. These results show the association of extracellular DNA with NET and suggest that NET are the source of the dynamic degradation undergone by HMW DNA in blood.

The MNR negatively correlated with the concentration of NE and MPO and decreased with the activation of the neutrophils (Fig. 3C and Additional file 1: Fig. S3). In previous studies, we have already described the decrease of MNR [26, 35, 47] and the increase of NE and MPO in cancer plasma, as compared to healthy individuals [26]. We believe this to be linked to the activation of neutrophils and the release of NET and could be explained by the structural differences between nuclear and mitochondrial DNA. The release of mitochondrial DNA (mtDNA) by activated neutrophils has been detected in several studies [48, 49]. Being devoid of an association with histones (and the protection which results from that), mtDNA is more subject to fragmentation than nuclear DNA, when not inside the circulating mitochondria [50]. We can therefore suggest that the rapid and massive production of both nuclear and mtDNA, combined with the faster degradation of mtDNA compared to the relatively stable nDNA (protected by the mononucleosome structure), leads to an imbalance in the proportions of cir-nDNA and cir-mtDNA, and thus to a decrease of MNR.

The in vitro degradation in serum of the long DNA fragments originating from ex vivo generated NET leads to the accumulation of the mononucleosome-associated fraction of cfDNA

qPCR quantification revealed the progressive decrease of the total DNA and the long DNA fragment concentrations (Fig. 4A), as well as the synchronous decrease of the DII (Additional file 1: Fig. S2), confirming our postulate that DNA from NET could be degraded by serum nucleases down to mono-N fragments. Analysis of the same DNA extracts by capillary electrophoresis demonstrated the shortening and dynamic reduction of the 1–30 kb DNA fraction during incubation in serum (Fig. 4B). That observation agrees with the qPCR results. The analysis of DNA by capillary electrophoresis does not allow us to distinguish male and female DNA using the sequence of the Y chromosome and therefore to directly quantify their respective amount. However, given the 30-fold excess of DNA from NET (0.36 ng/μL) compared to cirDNA from serum (0.012 ng/μL), we can assume that more than 95% of the DNA fragments detected by capillary electrophoresis originate from NET.

The NET production was associated with the accumulation of di-N and tri-N DNA. Conversely, in the course of the degradation of NET DNA in serum we observed an inversion of all those proportions, with the progressive disappearance of di-N and tri-N associated DNA fragments, and the accumulation of mono-N DNA fragments (Fig. 4D). These results confirm our hypothesis that, when expelled to the extracellular milieu (in this case, the serum), the long fragments of DNA originating from NET are predominantly degraded to mononucleosomes, given that (of all NET byproducts) these constitute the most stabilized structure with which cirDNA can associate.

The rate of DNA degradation of NET and gHMW DNA can in no way be compared to previous observations about the half-life of cirDNA in the circulation, where pharmacological aspects such as clearance, mononucleosome phagocytosis, or metabolic pathways are heavily involved. Our main objective is to show that the degradation of NET or chromatin leads to mono-N cirDNA. In physiological conditions, detectable cirDNA results from the balance between extracellular DNA release and elimination, as above mentioned. Extracellular release may result from various mechanisms. Of these, NET could be a significant source, along with other mechanisms such as apoptosis, necrosis, active cellular release, or extracellular traps generated from other blood cell types [25].

Chromatin organization along the genome is not random and may be specific to cell gene regulation and cell types [51, 52]. CirDNA cell-of-origin was recently investigated using WGS and fragmentomics. Analytical signals from the cirDNA fragmentome include non-random fragmentation pattern, transcription factor occupancy, GC contents, and end motif pattern, and by extension methylation. Studies from Y. Dor’s team based on cirDNA methylome analysis [53, 54] appear to indicate that the main cirDNA cells-of-origin are megakaryocytic and lymphocytic (especially neutrophilic). More recent reports confirmed that among a great variety of cells-of-origin, the neutrophil origin appears to be one of the most important [55]. Interestingly, this report and others also found that the cell-of-origin proportion may vary when comparing plasma of HI and non-HI, particularly cancer patients. In addition, cirDNA fragmentation analysis in open chromatin regions [34] and nucleosome occupancy also informs about tissue of origin. Thus, Shendure’s milestone study [46] revealed that lymphoid or myeloid origins have the largest proportions consistent with hematopoietic cells as the dominant source of cirDNA in HI, in contrast to cancer patients. This opens up a wide range of opportunities for diagnosis and the interrogation of physiological and pathological processes [53, 56].

NE affects the cirDNA-associated structure in vivo by enhancing DNA fragmentation

NE was found to be involved in DNA decondensation in the nuclei of activated neutrophils [27]. It could therefore be suggested that NE contributes to DNA fragmentation. In our in vivo study of cirDNA deriving from the plasma of WT, NE KO, and AAT KO mice, we demonstrated that the cirDNA size profiles revealed by sWGS differ according to NE activity. The absence of NE (in NE KO mice) was associated with the increase of the di-N associated cirDNA fragments and a significant increase of the average fragment size, which would indicate impaired DNA fragmentation, as compared to WT mice. The cirDNA fragment size profile of WT and AAT KO mice were visually similar, despite the increased activity of NE in the AAT KO mice. Capillary electrophoresis analysis revealed a decrease in the 1–10 k bp cirDNA fragment fraction in AAT KO mice, suggesting that NE facilitates or participates in the degradation of DNA fragments of sizes equivalent to those generated from ex vivo PMA-stimulated neutrophils. Taken together, these results permit the conclusion that NE affects the cirDNA structure and, consequently, the fragmentation features of cirDNA which originates from NET.

NE and MPO synergistically catabolize NET

The mechanism of cirDNA generation, in particular its enzymatic activity, remains insufficiently understood. Theoretically, the catabolism of DNA which leads to cirDNA production may be enabled by nucleases either intracellularly (DNase1L1, DNA fragmentation factor B or DFFB, DFFB/CAD and endonuclease G) or extracellularly (DNase1 and DNase1L3 or DNase γ). The paradigm in which, historically, this mechanism has been understood involved mainly apoptosis as well as necrosis [4, 7]. This is due to initial observations of laddered patterns of oligonucleosomal-size pieces (being the hallmarks of canonical apoptosis) by gel electrophoresis of cirDNA extract, and due also to the more recent characterization of the mononucleosome structure, principally by sequencing and fragmentomics [31, 45, 46, 57]. The main nuclease responsible for the internucleosomal double-strand (ds) cleavage of genomic DNA is DFFB/CAD; it may also be responsible, however, for single-strand breaks (ssB) during apoptosis [58]. Watanabe et al. observed that DNase1L3 cooperated with caspase-activated DNase (CAD) in the case of anti-Fas-mediated hepatocyte apoptosis, while DNase1L3 was responsible only for the generation of cirDNA in acetaminophen-induced hepatocyte necrosis. However, they were unable to detect cirDNA in CAD − / − DNase1L3 − / − mice, indicating that DNase1 is not sufficient for the generation of cirDNA following apoptosis [44]. This supported the observations made by D. Lo’s group [59, 60], who reported similar cirDNA levels in both DNase1 − / − and WT mice in steady-state condition. Although the mechanism of DNase activity is not fully understood, it shows correlation with inflammation; for instance, its induction mechanism may be associated with macrophage activation [61]. Collating various observations [14, 44, 60, 62] provides clues that CAD plays a key role in the initiation of cell death, by producing dsDNA breaks intracellularly, while DNase1L3 and DNase1 (both induced by apoptosis and necrosis) pursue chromatin catabolism extracellularly, such as in blood circulation. DNase1 is expressed by non-hematopoietic tissues and preferentially cleaves protein-free DNA [63, 64]. DNase1L3 (also known as DNase gamma) is secreted by immune cells and targets DNA–protein complexes such as nucleosomes [61, 64]. Note, several reports have postulated the role of active release in the production of DNA associated to extracellular vesicles [65]. The literature showed discrepancies about DNA being incorporated onto or into microvesicles. DNA encapsulated in vesicles would be less sensitive to nuclease attack, potentially modifying DNA fragmentation rate and process. Conversely, it was demonstrated that nuclease DNA1L3 could degrade DNA in microvesicles [61]. Regardless of these considerations, once released from vesicles its fragmentation process should be similar to that of cirDNA not associated with extracellular vesicles.

Over the last 5 years, we [3] and several other authors [44, 66, 67] have speculated about the role of NETosis in the generation of cirDNA. There are conflicting observations as to the nature of the nuclease-derived enzymatic activity involved in the degradation of NET to cirDNA. Both DNase1 and DNase1L3 have been shown to degrade NET in blood circulation [67, 68]. Jimenez et al. [67] showed that the toleration of chronic neutrophilia (as well as the prevention of vascular occlusion by NET clots during that process) requires both enzymes.

Data suggests that DNA degradation is increased when NE and/or MPO is added to serum nucleases and that DNA degradation is increased to a much larger extent in plasma when nuclease activity is greatly inhibited. Data strikingly showed that, in serum and in plasma, the DNA degradation rate due to NE was nearly equivalent, as it was (to a higher extent) due to MPO activities, irrespective of nuclease inhibition by EDTA as a chelating agent. Note, NE activity was found not dependent of Ca and Mg ions and determined in plasma [69]. Whereas MPO is dependent on Ca ions, its activity was still present in plasma [70]. This was the first direct observation of NE and MPO DNA degradation activity. Overall, our study consequently suggests that NE and MPO contribute to DNA catabolism. In addition, our data clearly demonstrated, for the first time, that NE and MPO contribute to and favor the DNA degradation process by releasing oligonucleosomes, mainly mono-N from NET. In addition, we suggest that, following neutrophil stimulation principally by oxidative explosion and its translocation to the nucleus, NE is continuously released from extracellular NET-associated granules, and continuously decondenses the chromatin. It has been established that histones are cleaved by NE derived from granulation and are citrullinated by peptidyl arginine deiminase 4 (PAD4) during intracellular NET formation. In the initial phase of this process, NE degrades the linker protein H1 and then the core histone proteins, in a process which is known to be enhanced by MPO [27]. MPO produces oxidizing agents such as hypochlorous acid (HOCl) from hydrogen peroxide (H2O2) and chloride anion (Cl −). Thus, we might postulate another possible hypothesis: oxidizers damage DNA, thus making it more susceptible to degradation by nucleases [71]. Drugs producing oxidizers might therefore be conceived as NET inhibitors, in pathologies where NET formation is uncontrolled [25].

In addition, our study first revealed that the synergistic association of NE and MPO contributes to the degradation of in vitro gHMW DNA, leading to the release of oligo- and principally mononucleosomes in serum containing medium. However, our study is limited since it cannot fully establish whether NE/MPO contribute to NET degradation either directly (such as having nuclease activity) or indirectly (such as chromatin unpacking, facilitating nuclease attack). Nonetheless, taking these observations as a whole, we suggest that the combined action of NE and MPO may have a dual role, firstly in intracellularly initiating NET formation and secondly in participating in NET degradation in the extracellular milieu (i.e., blood), which would point to a potential NET autocatabolism.

Conversely, other NET constituents may be involved in protecting DNA from nucleases such as the antimicrobial peptide LL-37 [72]. Alternatively, thrombin may bind to NET, thus conferring mutual protection against nuclease and protease degradation [73], illustrating the complex interplay between coagulation and NET formation [74].

Note, we found low but significant levels of both enzymes in the circulation of healthy individuals. We believe that the homeostasis of both enzymes and of nucleases such as DNase1L3 or DNase1 is critical in controlling NET formation on the one hand and cirDNA plasma concentration on the other. We speculate that genetic or epigenetic alterations impacting the expression of these entities may mitigate or exacerbate both the damage-associated molecular patterns (DAMPs) of cirDNA [3, 75] and NET formation, leading to chronic or acute disorders.

NET association with several disorders

NET formation is an efficient, innate response strategy for counteracting intrusive microorganisms. However, exacerbated NET formation in the host can be detrimental, on account of the toxicity of its exposed compounds to endothelial cells and to parenchymal tissue [56], whose pathological consequences can include thrombosis and fibrinolysis disorders. For instance, NET components (DNA, histones, and granule proteins such as MPO or NE) may trigger an inflammatory process. Thus, the release of NET byproducts as a result of dysregulated NET formation, therefore, can be implicated in both autoimmune and non-autoimmune diseases [76,77,78].

Intense investigation has been done on the deleterious role of NET in autoimmune diseases, particularly in the pathogenesis of systemic lupus erythematosus (SLE). The dysregulation of NET formation has been repeatedly demonstrated in SLE [79, 80]; patients with SLE, notably, have increased levels of NET in their tissues and circulation. Thus, NET lead to the exposition of many autoantigens within the extracellular space, in particular double-strand DNA, and MPO. As a result, autoimmune complexes containing nucleic acids associated with various proteins (including antibodies, the chromatin-associated protein HMGB1, the antimicrobial peptide LL39, and ribonucleoproteins) are found in the sera of SLE patients [80]. Evidence has also been presented that the ability to degrade NET is impaired in patients with SLE [81], leading to a continuous release of interferon. Since this key cytokine in the SLE pathophysiology is also responsible for sensitizing neutrophils to NET formation, the overall result is the establishment of an amplifier loop which (with other factors) enables the autoimmune reaction to be sustained, and the disease to progress. Given all of the above, NET byproducts are now used as biomarkers for SLE clinical diagnostics [82]. In our study, we found a strong association of cirDNA with NE and MPO, as well as a negative correlation of MNR with NET-related biomarkers in the plasma of SLE patients [83]. These results are similar to those found in the supernatant of ex vivo activated neutrophils (Fig. 3), suggesting NET involvement in the pathogenesis of SLE.

Several recent reports have also described the involvement of NET formation in cancer progression [84,85,86,87,88]. Some authors even postulate that NET are a seedbed for metastasis [88]. We recently demonstrated the association of MPO and NE concentrations with the amounts of cirDNA and anti-cardiolipin auto-antibodies in a large cohort (N = 217) of newly diagnosed metastatic colorectal cancer (mCRC) patients [

留言 (0)

沒有登入
gif