Neutrophil extracellular traps impair regeneration

Abstract Fibrosis is a major health burden across diseases and organs. To remedy this, we study wound‐induced hair follicle neogenesis (WIHN) as a model of non‐fibrotic healing that recapitulates embryogenesis for de novo hair follicle morphogenesis after wounding. We previously demonstrated that TLR3 promotes WIHN through binding wound‐associated dsRNA, the source of which is still unclear. Here, we find that multiple distinct contexts of high WIHN all show a strong neutrophil signature. Given the correlation between neutrophil infiltration and endogenous dsRNA release, we hypothesized that neutrophil extracellular traps (NETs) likely release nuclear spliceosomal U1 dsRNA and modulate WIHN. However, rather than enhance regeneration, we find mature neutrophils inhibit WIHN such that mice with mature neutrophil depletion exhibit higher WIHN. Similarly, Pad4 null mice, which are defective in NET production, show augmented WIHN. Finally, using single‐cell RNA sequencing, we identify a dramatic increase in mature and activated neutrophils in the wound beds of low regenerating Tlr3−/− mice. Taken together, these results demonstrate that although mature neutrophils are stimulated by a common pro‐regenerative cue, their presence and NETs hinder regeneration.

Inflammation and different components of the immune system have been shown to promote these and other forms of regeneration in salamanders, zebrafish and even mammals, specifically injury repair and barrier function maintenance in mucosa. [11][12][13] Although the cellular effects of macrophages and T cells are well studied, [14][15][16][17] neutrophilic effects on regeneration are not. Immediately after skin wounding, a robust inflammatory phase occurs, which allows the ingress of keratinocytes and fibroblasts to proceed afterwards. The early stage of wound healing is defined by the dramatic recruitment of mature neutrophils, which are instrumental in providing defence against microbial pathogens. [18][19][20][21][22] This is followed by an influx of macrophages (Mφ's) that continue the phagocytic processes begun by neutrophils and aid in the transition to the proliferative phase of wound healing. [23][24][25] Increasingly, macrophages have also been shown to be essential for WIHN via TNF-induced AKT/ β-catenin signalling. [15][16][17] In addition to phagocytosis and degranulation, mature neutrophils can produce extracellular traps (ET), large extracellular weblike structures composed of decondensed chromatin bound to various cytosolic and granule proteins. 21,22,[26][27][28] While originally recognized as a defence mechanism against pathogens, 21,26,27 they have also been found to mediate 'sterile' inflammatory processes. 29,30 In the absence of infection, ETs can be stimulated in sterile tissue environments through various cytokines 27,[31][32][33] and by activated platelets. 29,30 Interestingly, ETs are found within sterile wounds of mice and delay wound healing. 34 Mechanistically, ETs are formed by the rapid decondensation of the cellular chromatin, followed by the fragmentation of the nuclear membrane and mixing of the nuclear and cytoplasmic compartments, before being expelled from the cell. The ability of neutrophils to rapidly migrate to the wound site and produce ETs, coupled with the nuclear localization of some dsRNA, made us question whether neutrophils were a source of the dsRNA critical for Toll-like receptors 3 (TLR3)-dependent WIHN.
Interestingly, while there are extensive studies on the DNA components released during ET formation, 21,28,35 the RNA components are poorly understood.
Briefly, TLRs are highly conserved single-pass membranespanning receptors that recognize structurally conserved molecular components of invading microbes and activate a cascade of inflammatory signalling pathways. 36 Rather than simply recognizing pathogen-associated molecules, they can also initiate sterile inflammation upon recognizing damage-associated molecular patterns (DAMPs), which are critical to recruit immune cells and initiate wound healing. 37 TLR3 is activated by dsRNA and has primarily been studied in the context of viral infection. 38 Mounting evidence shows that TLR3 also plays an important role in wound repair. [39][40][41][42][43][44][45][46][47] Synthetic double-stranded RNA (dsRNA) polyriboinosinic-polyribocytidylic acid (poly[I:C]) treatment dramatically increases WIHN in mice. Furthermore, wound-released dsRNA activates TLR3 to promote hair follicle regeneration. 44 Notably, the dsRNA U1 spliceosomal small nuclear RNA (snRNA) may be an important endogenous RNA sensed via TLR3. 42,43,48,49 Specifically, UV damage releases U1 snRNA that stimulates cytokine production in keratinocytes and increases barrier gene transcription. 42,43 To probe how mature neutrophils influence wound regeneration and WIHN, we analysed multiple microarrays from distinct contexts of high regenerating mice and found a common neutrophil signature. Using immunofluorescence and flow cytometry, we find that neutrophils remain in the wound bed, albeit at low levels, after the acute inflammatory phase, where they produce NETs that contain the nuclear U1 dsRNA. To define how this influences regeneration, we used a neutrophil-specific diphtheria toxin ablation model to deplete mature neutrophils in the wound bed and found that-contrary to our initial hypothesis-the absence of mature neutrophils enhances WIHN. Eliminating neutrophil's ability to produce NETs by knocking out Pad4 also boosted WIHN, confirming the negative influence of mature neutrophils on regeneration. Finally, we used single-cell RNA sequencing to characterize WIHN-deficient Tlr3−/− mice and found that they too have a dramatically increased population of mature neutrophils in the re-epithelized wound bed, compared with wild-type mice, likely contributing to their diminished regenerative capacity.
These results indicate that, while important for preventing infection, mature neutrophils and their NETs negatively impact regeneration and WIHN. Although a common pro-regenerative signal might increase neutrophil infiltration, mature neutrophils instead likely contribute to fibrosis.

| Wound-induced hair neogenesis assay
All in vivo experimental surgical procedures were performed as previously characterized. 9,14,44,47,51 Briefly, after exposure to anaesthesia (Baxter, Isoflurane), the dorsal side of 3-week-old (21 days) male and female mice was shaved. Surgical scissors were used to excise 1.25 × 1.25 cm 2 of skin on wound day 0 (WD) 0, creating wounds deep into the fascia. On approximately WD21, neogenic hair follicles in the re-epithelialized skin tissue were quantified using reflectance confocal scanning laser microscopy (CSLM) as previously published. 44,47

| Neutrophil depletion
Diphtheria toxin depletion was done with PMN DTR and PMN WT littermate control mice that were IP injected with 250 ng DT (Sigma-Aldrich). The injections were primarily done one day before and after wounding, or on WD 6, 8 and 10.

| Flow cytometry
Flow cytometry was used to access neutrophil depletion. Blood was collected via retro-orbital sinus bleeds, and red blood cells were lysed RBC lysis buffer (BioLegend, 420301). Wound beds were surgically removed, and cell suspensions were prepared by digesting the tissue in a cocktail consisting of Liberase TL (Roche, 5401020001) and DNase I (Sigma, DN25) in RPMI 1640 (Gibco, 11875093). Cells were washed and then Fc blocked (BioLegend, 101320), before staining with an antibody cocktail (Table 1). Finally, cells were washed and resuspended in FACS buffer containing propidium iodide (Miltenyi, 130-093-233).
All flow cytometry experiments were performed on a BD LSR II, and downstream analysis of data was performed using FlowJo.

| Neutrophil extracellular trap measurement
Wound beds were surgically removed at WD2 and WD7, using a 6-mm biopsy punch to remove excess tissue. On the final wash, SYTOX green was added (Thermo Fisher, S7020, 1:1000). This was performed on a BD LSR II, and downstream analysis of data was performed using FlowJo.

| 3′-end single-cell RNA sequencing
The re-epithelialized wound beds (WD10) of a Tlr3−/− and a C57BL/6NJ control mouse were excised, and cell suspensions were prepared by digesting the mouse skin tissue in a cocktail consisting Seurat was then used to generate conserved genes, differentially expressed genes, feature plots, dot plots and ridge plots. Cell clusters were then defined querying conserved genes and differentially expressed genes against the ImmGen gene expression database (www. immgen.org) using the interactive tool 'My Gene Set'.

| Histology
Biopsies from mouse skin tissue were removed and fixed in 4% paraformaldehyde overnight and then transferred to 70% ethanol. Samples were then submitted to the Johns Hopkins Oncology Tissue Services Core facility where they were embedded in paraffin.
Tissue sections were obtained at 4 μm thickness and mounted onto glass slides, followed by haematoxylin and eosin (H&E) staining.

| Immunofluorescence and immunohistochemistry
Immunofluorescence (IF) microscopy was performed on de-paraffinized tissue sections that received heat-induced antigen retrieval using Target Retrieval Solution (Agilent Dako, S169984-2). After washing and permeabilization in TBS-T universal buffer (0.2% Triton X-100 in Tris-buffered saline), sections were blocked at room temperature in 5% goat, donkey or foetal bovine serum with 1% bovine serum albumin. Tissue sections were then incubated overnight at 4°C with primary antibodies (

| Microarray, RNA-seq and proteomic analysis
For both Rnasel−/− and WT mice, total RNA was isolated from mouse tissue at scab detachment (SD) day 0 (also WD10) from Proteins from the wound centre and wound edges were analysed by proteomics, as previously described. 47 Briefly, after saline washing, samples were lysed in 5% sodium deoxycholate (DOC) detergent. After sequential peptide processing (reduction, alkylation and trypsinolysation), downstream resolving and analysis were performed on a nano-ACQUITY UPLC system with a Tribrid Orbitrap-quadrupole-linear ion    Figure 1D). Finally, we analysed the gene expression changes between wounded specific pathogen-free (SPF) mice, which have increased regeneration and WIHN, when compared to germ-free (GF) mice at WD11. 52 Like the other two high regeneration models, when compared to GF, SPF mice have elevated neutrophil chemotaxis and immune response transcripts ( Figure 1E).

| Quantification and Statistical analysis
Together, these unique examples correlate neutrophil chemotaxis with high regeneration and WIHN.

| Neutrophils persist in the wound bed after the acute inflammatory phase, producing extracellular traps
Having

| Mature neutrophils inhibit wound-induced hair neogenesis
To functionally test the importance of neutrophils, we generated a transgenic mouse model for selective and inducible ablation of neutrophils upon injection of diphtheria toxin (DT). 50 Figure 3B,C). This decrease in wound bed neutrophils occurs even with the standard large size used to test for WIHN ( Figure 3C).
In line with the above findings, the reduction in mature neutrophils correlates with substantially elevated WIHN (fold = 3.23) ( Figure 3D). Later neutrophil ablation in the healing process yielded similar results (WD6, 8 and 10) ( Figure S2). Collectively, these data suggest that mature neutrophils have a detrimental effect on the regeneration of hair follicles.
Finally, given the presence of NETs late in wound healing ( Figure 2F), we sought to test the role of NETs in hair follicle neogenesis directly. We, therefore, tested WIHN in NET-deficient Pad4−/− mice, as employed in Figure 2F. In the absence of Pad4 and NETs, WIHN is enhanced (fold = 2.47, p = 0.026) ( Figure 3E). This suggests that NETs reduce the regenerative capacity of mice during the wound-healing process.

| Single-cell RNA-seq correlation of mature neutrophils with poor WIHN
We next wondered whether increased neutrophils and NETs might similarly play inhibitory roles in other contexts of poor regeneration.
Double-stranded RNA sensing, mediated by Tlr3 and downstream effector pathways Il-6/ Stat3, has been shown to be critical for WIHN. Tlr3−/− mice, though grossly normal without wounding, have substantially less regenerated hair follicles than their wild-type controls. 44 Although TLR3 dsRNA sensing has been shown to be critical for neutrophil recruitment and NET production in a model for acute lung injury (ALI) and glomerulonephritis (GN), we probed to Unsupervised clustering and UMAP non-linear dimensional reduction identified 18 cell clusters. Seurat generated conserved and differentially expressed genes, which were used to assign cluster identities ( Figure S3a). Keratinocytes and fibroblasts did have differentially expressed genes (Table S1). As anticipated from our earlier data where inhibiting mature neutrophils promotes WIHN, we find the mature neutrophil cluster in the low regenerating Tlr3−/− mice to be significantly different (p < 0.0001) and most substantially increased (fold = 6.98) relative to WT mice ( Figure 3G, Figure S3b).
There are also marked changes in the fibroblast (Fibro1) cluster (fold = 5.12), which is also significantly different (p < 0.0001) from WT mice ( Figure 3G, Figure S3c). Consistent with increased neutrophils in low regenerating TLR3 −/− mice, neutrophil-associated genes are significantly elevated in Tlr3 −/− mice ( Figure 3H), with a general elevation in gene ontology categories consistent with greater neutrophil activation in Tlr3 −/− such as ribosome-associated transcripts ( Figure 3I). Together, these data demonstrate that elevated mature and activated neutrophil levels in Tlr3−/− mice correlate with their low WIHN.

| DISCUSS ION
The wound-healing process is a careful balance of interconnected steps that weigh the benefits of quick barrier repair, which leads to fibrous scarring, and more complete regeneration, that restores function and appearance. While the role inflammation plays in regeneration and scarring is still being elucidated, increasing evidence suggests that neither excess nor lack of inflammation supports regeneration. Fgf9-producing γ-δ T cells are critical for WIHN, infiltrating into wound bed immediately before re-epithelialization and onset of hair follicle regeneration. 14 Macrophages have also been shown to be important in the process, with their ablation eliminating WIHN. [15][16][17] Additionally, the injection of the dsRNA mimic poly(I:C), as early as WD3, dramatically enhances WIHN. 44  where neutrophils play an important role in defence against bacterial pathogens but whose persistence within the wound bed too long after barrier repair hinders regeneration. In this sense, the dramatic decrease in neutrophil numbers in the course of wound healing is likely important for later WIHN. In the future, it will be interesting to see whether selectively targeting NETs (eg, PAD4 inhibitors, DNase I, N-acetylcysteine) enhance regenerative wound repair.
Our work suggests important areas of future investigation. One question is unravelling the paradox of why neutrophil infiltration signatures correlate with high WIHN, but mature neutrophils inhibit WIHN. One possible model is that a common upstream cue or factor both promotes WIHN and promotes neutrophil infiltration, but the latter serves to limit WIHN in favour of decreasing infection risk.
Defining this common upstream signal will be important for future work.
It will also be interesting to define the function of neutrophils within the re-epithelized wound since the barrier has been restored.
It is possible that the neutrophils present at later time points in the healing process are not the classical pro-inflammatory N1 subtype of neutrophils, but more recently discovered anti-inflammatory N2 neutrophils. Initially identified in cancer, N2 neutrophils have impaired anti-tumour capacity and express gene characteristic of alternatively activated M2 macrophages, such as arginase-1 (Arg1) and mannose receptor C-type 1 (Mrc1). 73 Studies in myocardial infarction show that neutrophils are temporally polarized between the two subtypes, beginning as N1 neutrophils and shifting towards N2 as inflammation recedes and tissue repair proceeds. 74 Additionally, recent evidence shows that a related subtype of immature neutrophil can promote axon regeneration post-crush injury to the optic nerve of mice. 75 For this reason, it will be important to continue to classify the subtypes of neutrophils present throughout WIHN, rather than focusing sole on mature inflammatory neutrophils.
Another question is whether nuclear RNAs released in NETs have any function besides the general theorized one for released DNA. Since NETs inhibit WIHN, nuclear dsRNA release in NETs might act differently from exogenous or other endogenous dsRNA; perhaps dsRNA in NETs is trapped and not released effectively to stimulate a dsRNA response.
In summary, we demonstrate a novel role for NETs and mature neutrophils to inhibit regeneration ( Figure S4). Future studies will be important to further understand the biology of regeneration and test the capacity for neutrophil inhibition to promote regenerative healing.

ACK N OWLED G EM ENTS
The authors also thank Conover Talbot Jr. (JHMI Deep Sequencing and Microarray Core) for assistance with microarray analysis.

CO N FLI C T O F I NTE R E S T
None of the authors have any conflict of interest regarding this manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
All gene expression data sets have been uploaded to NCBI GEO and are available at the above accession numbers listed in the Methods section.