RNase L represses hair follicle regeneration through altered innate immune signaling

Abstract

Mammalian injury responses are predominantly characterized by fibrosis and scarring rather than functional regeneration. This limited regenerative capacity in mammals could reflect a loss of proregeneration programs or active suppression by genes functioning akin to tumor suppressors. To uncover programs governing regeneration in mammals, we screened transcripts in human participants following laser rejuvenation treatment and compared them with mice with enhanced wound-induced hair neogenesis (WIHN), a rare example of mammalian organogenesis. We found that Rnasel-/- mice exhibit an increased regenerative capacity, with elevated WIHN through enhanced IL-36α. Consistent with RNase L's known role to stimulate caspase-1, we found that pharmacologic inhibition of caspases promoted regeneration in an IL-36-dependent manner in multiple epithelial tissues. We identified a negative feedback loop, where RNase L-activated caspase-1 restrains the proregenerative dsRNA-TLR3 signaling cascade through the cleavage of toll-like adaptor protein TRIF. Through integrated single-cell RNA-seq and spatial transcriptomic profiling, we confirmed OAS & IL-36 genes to be highly expressed at the site of wounding and elevated in Rnasel-/- mouse wounds. This work suggests that RNase L functions as a regeneration repressor gene, in a functional trade off that tempers immune hyperactivation during viral infection at the cost of inhibiting regeneration.

Keywords: Dermatology; Inflammation; Innate immunity; Skin.

Figure 1. Transspecies dsRNA sensing signature during skin regeneration; RNase L represses regeneration markers.

(A) 3-way Venn diagram shows a 14-gene overlap present in all of the top 200 genes in microarrays of in vivo WIHN comparing C57BL/6 × FVB × SJL mice (high regeneration strain) versus C57BL/6 mice (low regeneration strain) versus in vivo human clinical trial of participants treated with a rejuvenation laser, with in vitro human keratinocytes treated with dsRNA/Poly (I:C) as a positive control. The in vitro and in vivo human microarrays contain a total of 49,395 annotated transcripts each, and the in vivo murine microarray contains 53,145 transcripts. (B) Gene ontology analysis of each of the top 200 gene lists, highlighting the predominance of OAS family members in each data set. (C) Gene ontology terms enriched in the 14 overlapping genes from all 3 datasets include the upregulation of OAS family genes. Inset graphs show the gene fold-expression changes from the original microarray for genes present in that category; green and blue indicate mouse and human, respectively. (D) Analysis of the ribonuclease RNase L (downstream and activated by OAS), Venn diagram shows the top 200 overlapping genes in Rnasel^–/– mice after wounding (at scab detachment) and human keratinocytes treated with siRNA targeting RNase L. GO categories include multiple developmental pathways. The in vivo microarray data contain 22,206 transcripts and the in vitro RNA-seq data contain 33,264 transcripts. (E–G) Poly (I:C) treatment (10 μg/mL) of RNase L siRNA transfected human keratinocytes induces multiple morphogenesis markers (E; n = 4, 2-way ANOVA, P < 0.05), including WNT7B (green) shown by immunostaining (original magnification, × 20) (F), but inhibits differentiation markers as measured by RT-PCR with fold changes compared with siCon without stimulation (G; n = 4, 2-way ANOVA, P < 0.05).

Figure 2. RNase L–loss enhances hair follicle regeneration.

(A) Rnasel^–/– mice exhibit increased WIHN with an intact superincrease in the presence of poly (I:C) (confocal scanning laser microscopy [CSLM] and Alkaline Phosphatase [AP]staining, images; n = 12 WT, 13 WT+PIC and 12 Rnasel^–/–, 13 Rnasel^–/–+PIC, 2-way ANOVA, P < 0.0001). In each image, the dash red box indicates the area of hair follicle regeneration. (B) Rnasel^–/– mice display normal wound closure speed (n = 20 WT mice and 17 Rnasel^–/– mice). (C) Transepidermal water loss (TEWL) was measured in the center and periphery of healed skin at scab detachment day (Would Day 10 [WD10]) for both WT and Rnasel^–/– mice. Rnasel^–/– mice exhibit dramatically lower TEWL measurements, consistent with a postwounding improved barrier compared with WT mice (n=3, 2-tailed unpaired t test, P < 0.005, P = 0.001). (D) Rnasel^–/– mice have greater morphogenesis marker gene expression of Tlr3 (n = 3, P < 0.01), Il6 (n = 3, P < 0.0001), Wnt7b (n = 3, P < 0.01), and Edar (n = 3, 2-tailed unpaired t test, P < 0.01) on day of reepithelialization as measured by qRT-PCR. (E) Unwounded skin of Rnasel^–/– mice shows increased protein expression of stem cell markers KRT5 (green) and KRT15 (red) and morphogenesis marker WNT7B (green) shown by immunofluorescence (original magnification, × 20).

Figure 3. RNase L loss increases neutrophil accumulation and IL-1 production during epithelial regeneration.

(A) Gene ontology analysis of Rnasel^–/– mice during reepithelialization (approximately 10 days after wounding) show enrichment of IL-1 response, neutrophils, and wound healing pathways. Individual genes corresponding to each category are shown in green. (B) At 3 days after wounding, Rnasel^–/– mice recruit significantly more neutrophils in the wound bed. Ly6G (green) is a neutrophil marker and H4Cit (red) is a marker for citrullinated histones released from neutrophils during NETosis. Nuclear staining was performed using DAPI (blue). The white dashed line signifies the dorsal edge of the wound bed. Scale bar: 100 μm. (C) Gene ontology analysis of the top 100 genes in a microarray of high regenerating outbred WT strain mice (C57BL/6 × FVB × SJL) compared with the lower regenerating WT C57BL/6 and the top 100 proteins found in the center (high regenerating) versus the edge (low regenerating) areas of the wound show a common signature for IL-1 family member IL-36α (red) and neutrophil granule proteins (orange). (D) Heat map analyses from C show IL-1 family members are enriched in the high regeneration mice and center of the wound, particularly IL-36 family members (red). Neutrophil granule proteins (orange), known to proteolytically cleave and activate IL-336 proteins, are also enriched.

Figure 4. RNase L suppresses IL-36 expression, which is required for and promotes WIHN.

(A) Wounded tissue from Rnasel^–/– mice reveal elevated ILF16/IL-36α protein, as shown by Western blot. (B) Keratinocytes harvested and cultured from Rnasel^–/– mice actively secrete more IL-36α than WT controls, as shown by Western blot. (C) Both unwounded and wounded skin show increased expression of IL-36α (green) in Rnasel^–/– mice. IL-36α expression peaks in both WT and Rnasel^–/– mice at 3 days after wounding. Scale bar: 200 μm. (D) Injection of 50 ng of recombinant IL-36α protein underneath the scab at WD7 promotes WIHN, as shown by CSLM and AP staining (n = 6, 2-tailed unpaired t test, P = 0.0002). (E). Histology of D comparing vehicle or rmIL-36α–treated mouse skin sections. The neogenic hair follicles (purple) are shown aggregated at the center of the scar. Scale bar: 200 μm. (F) Treatment of human keratinocytes with recombinant IL-36α increases WNT7B mRNA expression (n = 4,1-way ANOVA, P < 0.05), as quantified by qRT-PCR. (G) Il36r^–/– mice fail to regenerate hair follicles and are not responsive to poly (I:C), as shown by CSLM and AP staining (n = 6, 2-way ANOVA, P = 0.0147, P < 0.0001). (H) Rnasel^–/–/Il36r^–/– mice lose the ability to regenerate hair follicles compared with Rnasel^–/– mice, as shown by CSLM (n = 3 and 7, respectively, 2-tailed unpaired t test, P < 0.0001). (I) siRNA Knockdown of IL-36α and RNase L in human keratinocytes abrogates the increases of both WNT7B and IL6 morphogenesis markers compared with RNase L siRNA alone (n = 3 each, 1-way ANOVA, P < 0.05).

Figure 5. Caspases, known downstream mediators of RNase L, restrain IL-36 release and regeneration.

(A) dsRNA activation of RNase L stimulates NLRP3 & Caspase-1 illustration. (B) QV-D-OPh induces elevation of neutrophils in unwounded skin shown by t-SNE analysis and quantification (n = 3, 2-way ANOVA, P < 0.05) (C) QV-D-OPh for 48 hours induces Il36a mRNA in cultured mouse keratinocytes (n = 6, 1-way ANOVA, ****P < 0.0001, **P = 0.0012). (D) QV-D-OPh induces IL-36α protein without inhibition of the receptor antagonist IL-36RN in whole-cell lysates of MEKs (P < 0.05 by 2-way ANOVA, n = 3). (E) siRNA knockdown of caspase-1 in mouse keratinocytes leads to elevated secretion of IL-36α. (F) Schematic of the pan-caspase inhibitor Q-VD-OPh intraperitoneal injection (1.33 mM) of mice wounded to measure WIHN. IP injections were done 1 day before and 10 days after wounding C57BL/6J mice with 1.25 × 1.25cm² square wounds. Mice were then sacrificed at wound-day 21 to measure WIHN. (G) QV-D-OPh promoted WIHN compared with vehicle, as shown by CSLM and AP staining (n = 6 versus 7, 2-tailed unpaired t test, P < 0.0001). (H) Histology of G comparing vehicle to Q-VD-OPh–treated mouse skin sections. Neogenic hair follicles (purple) are shown aggregated at the center of the scar. Scale bar: 100 μm. (I) Immunostaining of reepithelialized wounded tissue shows elevated IL-36α in Q-VD-OPh–treated mice. (J) Il36r^–/– mice do not respond to Q-VD-OPh treatment and are unable to regenerate hair follicles after wounding (n = 7, 2-tailed unpaired t test). (K) Histology of WIHN scars from J. Scale bar: 200 μm.

Figure 6. Caspase-1 cleaves Ticam 1 to restrict IL-36α and regeneration.

(A) Based on proteome of NHEKs treated with QVD (n = 3) versus DMSO (n = 3), upstream regulators are predicted to include elevated TICAM1 (TRIF) by Ingenuity analysis (B) RNA-seq transcriptome of cells treated as in A also shows elevation of TICAM1 (rows represent independent samples, color scale based on Z-score distribution) (C) Time-dependent cleavage of TICAM1 in unstimulated MEK and NHEK lysates after addition of recombinant caspase-1 protein and visualized by Western blot. (D) Western blot analysis of unstimulated Rnasel^–/– MEKs show increased TRIF and decreased cleavage. (E) qRT-PCR quantification of TRIF mRNA after CASP1 and RNASEL siRNA treatment as well as poly I:C & QVD in NHEKs (n = 6, 1-way ANOVA, P < 0.001) (F) Western blot analysis shows Caspase-1 and RNase L-dependent TRIF cleavage after siRNA treatment in NHEKs (G) qRT-PCR detects TRIF-dependent IL-3-6A expression after CASP1, RNASEL, or TRIF siRNA in NHEK with or without poly I:C addition (n = 4, 2-way ANOVA, P < 0.001) (H) Western blot analysis of IL-36α shows TRIF-dependent protein expression with Poly I:C addition (top) or QVD (bottom) (n = 3).

Figure 7. Comparative analysis of ScRNA-seq in wound center, wound edge, and nonwound regions and spatial transcriptomics in WT and RNase L–deficient models.

(A) UMAP of WT wounding dataset showing sample integration and annotation. Clusters are color coded. (B) Stacked bar plot of relative fractions of each cell type. Colors indicate annotated clusters from A. (C) Heatmap comparison of OAS gene expression in keratinocytes versus fibroblasts shows higher levels of OAS expression in wound samples. (D) Heatmap comparison of genes of interest in keratinocytes versus fibroblasts, show higher levels of IL-36 expression in the center of wound samples in keratinocytes. For C and D, sample indicated by color, rows represent independent samples, heatmap color scale based on Z-score distribution. (E) Graph-based clustering of Rnasel^–/– versus WT wound at SD0; each color-labeled dot corresponds to particular location in the tissue section (Scale bar: 1,000 μm) (F) Designated wound area shown on postxenium histology slide; wound area shown in pink, nonwound area shown in grey (Scale bar: 1,000 μm) (G) Density map of OAS transcripts in Rnasel^–/– versus WT wound tissue at SD0; color gradient indicates density of transcripts per square (density threshold, 0.050; Scale bar: 1,000 μm) (H) Heatmap comparison of OAS gene expression from merged spatial Rnasel^–/– versus WT wounding dataset (n = 3 and 3, respectively), shows higher levels of OAS expression in Rnasel^–/– wounds (rows represent independent samples (n = 3) from wound area designations, color scale based on Z-score distribution). (I) Density map of Il36a transcripts in Rnasel^–/– versus WT wound tissue at SD0; color gradient indicates density of transcripts per square (density threshold, 0.050; Scale bar: 1,000 μm) (J) Heatmap comparison of genes of interest expression from merged spatial Rnasel^–/– versus WT wounding dataset (n = 3 and 3, respectively), shows highest levels of Il36 expression in Rnasel^–/– wound keratinocytes (rows represent independent samples (n = 3) from wound area designations, color scale based on Z-score distribution).