A tissue injury sensing and repair pathway distinct from host pathogen defense

Abstract

Pathogen infection and tissue injury are universal insults that disrupt homeostasis. Innate immunity senses microbial infections and induces cytokines/chemokines to activate resistance mechanisms. Here, we show that, in contrast to most pathogen-induced cytokines, interleukin-24 (IL-24) is predominately induced by barrier epithelial progenitors after tissue injury and is independent of microbiome or adaptive immunity. Moreover, Il24 ablation in mice impedes not only epidermal proliferation and re-epithelialization but also capillary and fibroblast regeneration within the dermal wound bed. Conversely, ectopic IL-24 induction in the homeostatic epidermis triggers global epithelial-mesenchymal tissue repair responses. Mechanistically, Il24 expression depends upon both epithelial IL24-receptor/STAT3 signaling and hypoxia-stabilized HIF1α, which converge following injury to trigger autocrine and paracrine signaling involving IL-24-mediated receptor signaling and metabolic regulation. Thus, parallel to innate immune sensing of pathogens to resolve infections, epithelial stem cells sense injury signals to orchestrate IL-24-mediated tissue repair.

Keywords: STAT3; angiogenesis; coordinated tissue repair; epithelial stem cells; hypoxia; innate immune signaling; interferons; interleukin-24; microbiome-independent responses; tissue injury.

Figure 1.. IL-24 is specifically produced by epithelial stem cells near the wound site

(A) Schematic of the wound repair process in mouse skin. (B) Sagittal sections of homeostatic skin, and wounds (days indicated) immunolabeled for p-STAT3 at Tyr705 (n = 5 mice). (C) qRT-PCR for putative STAT3-targeting cytokines in homeostatic skin and day-1 wound. $\mathit{\text{Il1β}}$ served as a positive control. Values were normalized to Ppib (n = 3 mice). (D) qRT-PCR of Il24 mRNA in FACS-purified cell populations isolated from homeostatic and wounded skin (n = 3 mice). (E) IL-10 cytokine family expression from RNA-seq performed on FACS-purified EpdSCs from homeostatic and wounded skin. TPM, transcripts per kilobase million (n = 3 mice). (F) PLISH (proximity-ligation-based in situ hybridization) images of sagittal sections of homeostatic and wounded skin, probed for Il24 and Krt14 mRNA. Serial skin sections of Il24 PLISH and immunolabeling of integrin-$\text{α}5$ in day-3 wounds. The red-boxed region was magnified and shown at the right to highlight the Il24 PLISH signal in the re-epithelializing (migrating) epidermis. Asterisk (*) denotes autofluorescence of hair shaft and stratum corneum (n = 3 mice). Experiments were performed R ≥3×. White dotted lines, epidermal-dermal border; wound site, red dotted line; epidermal migration direction, red arrow. DAPI, nuclei; scale bars, 100 μm. Data in (D) and (E) are presented as mean ± SEM. N.D., not detected. See also Figure S1 and Tables S1 and S2.

Figure 2.. Injury-induced IL-24 signaling resembles infection-induced interferon signaling

(A) qRT-PCR of Il24 mRNA in FACS-purified EpdSCs from homeostatic and wounded skin from specific-pathogen-free (SPF) vs. germ-free (GF) C57BL/6J WT mice (SPF, n = 5–6, GF, n = 5–9 mice). (B) qRT-PCR of Il24 mRNA in epidermis microdissected from homeostatic and wounded skin from WT vs. Myd88^−/−Trif^−/− mice (n = 3 mice per genotype; representative of 3 independent experiments). (C) qRT-PCR analysis of Il24 mRNA in EpdSCs FACS-purified from homeostatic and wounded skin from WT vs. Rag2/IL2rg DKO mice. Note that Rag2/IL2rg DKO mice lack all functional lymphocytes (n = 5–7 mice per genotype). (D) Diagram depicting our central hypothesis that parallel but distinct signaling pathways are used for responding to and resolving pathogen infection and tissue injury. Steps tackled in current study are highlighted by question marks. Data in (A)–(C) are presented as mean ± SEM. Statistical significance was determined using two-tailed unpaired Student’s t tests; ns; not significant; N.D.; not detected. Dots in the graphs indicate data from individual mice. See also Table S3.

Figure 3.. Epithelial-expressed IL-24 coordinates dermal repair and re-epithelialization

(A) Schematic of two C57BL/6J Il24^−/− mouse strains generated by CRISPR-Cas9-mediated frameshift deletions within Il24 exon 2. Impairments of wound repair were indistinguishable between two loss-of-Il24-function strains, used interchangeably for experiments. (B) Sagittal sections of day-3 wounds from wild-type (WT) vs. Il24 null mice immunolabeled for p-STAT3. Note that p-STAT3 is still seen in Il24 null wounded epidermis (asterisk). Graphs show quantifications of the percentage of EpdSCs expressing p-STAT3 (upper), and the thickness of keratin 14 (KRT14⁺) progenitor layers (lower) (n = 5 mice per genotype). (C) Il20rb RNA-seq of FACS-purified cell populations from homeostatic skin and day-5 wounds (note: immune cells were only from day-5 wounds). TPM, transcripts per kilobase million (n = 5 mice). (D) Sagittal sections of day-5 wounds immunolabeled for KRT14 (epidermis), CD31 (endothelial cells), and labeled with 5-ethynyl-2′-deoxyuridine (EdU) (proliferation). Boxed regions are magnified in insets to better visualize EdU incorporation of S-phase cells (scale bars, 10 μm). Graphs show quantifications of percentage of EdU⁺ cells in epidermis and dermis. For epidermis, quantifications were performed separately for the cells in the migrating zone (to the right of the wound site) and behind the migrating zone (to the left of the wound site) (n = 5 mice per genotype). (E) Left: quantifications of the percentages of migrating epidermis displaying adjacent CD31⁺ endothelial cells (top) and the percentages of the wound beds at day-5 and −7 post wounding that were repopulated with sprouting blood vessels (CD31⁺ cells) (middle and bottom). Mouse genotypes are as indicated (see STAR Methods). Top and middle: WT: n = 5, Il24 Het: n = 6, Il24^−/−: n = 9 mice, one-way ANOVA, Tukey’s multiple comparisons test; bottom, WT: n = 5, Il20rb^−/−:n =6 mice, two-tailed unpaired t test; dots in the graphs indicate data from individual mice. Right: Images of whole-mount immunofluorescence microscopy and 3D image reconstruction performed on day-5 wounds from WT vs. Il24 null mice (scale bars, 50 μm. Immunolabeling was for KRT14 [epidermis] and endomucin [blood vessels]) (n = 3 mice per genotype). (F) Sagittal sections of day-5 wounds immunolabeled for CD31 and $\text{PDGFRα}$ (left), or for $\text{PDGFRα}$, collagen-I, and KRT14 (right). Asterisk (*) denotes a paucity of fibroblasts (${\text{PDGFRα}}^{+}$) and their deposition of collagen-I ECM in the dermis of Il24^−/− skin. The boxed region magnified in the color-coded insets shows additional Ki67 immunolabeling (Scale bars, 20 μm). Yellow arrows denote Ki67⁺ proliferating fibroblasts (Ki67⁺PDGFR⁺). Quantifications are of fibroblast amount ($\text{PDGFRα}$ intensity, upper) and collagen deposition (lower) (n = 5 per genotype). (G) Sagittal sections of day-5 wounds immunolabeled for p-STAT3 and KRT14. Percentage and number/area of p-STAT3⁺ dermal cells beneath the wound bed are quantified (n = 3 mice per genotype). (H) Left: sleeping beauty system used to generate epidermal-specific Il24 mRNA knockdown mice. Middle top: qRT-PCR of Il24 mRNA in FACS-purified EpdSCs from homeostatic and day-1 wounded skins from control (Ctrl) vs. shIl24 mice (n = 5–6 mice for each genotype). Right: sagittal sections of day-5 wounds from control (Ctrl) vs. shIl24 mice immunolabeled for CD31, KRT14 and labeled with EdU. Percentage of migrating epidermis adjacent to CD31⁺ capillaries is quantified in middle bottom panel (n = 6 mice per genotype). White dotted lines: epidermal-dermal border; wound site, red dotted line; epidermal migration direction, red arrow. DAPI, nuclei; scale bars except for boxed regions and whole mount: 100 μm. Data in (B)–(H) are presented as mean ± SEM. Dots in the graphs (E) and (H) indicate data from individual mice. Statistical significance was determined using two-tailed unpaired Student’s t tests in (D), (E; bottom panel), (F), (G), and (H); and using one-way ANOVA, Tukey’s multiple comparisons test in (B) and (E; top two panels); **** p < 0.0001; *** p < 0.001; ** p < 0.01; * p < 0.05; and ns, not significant. See also Figures S2–S5.

Figure 4.. Ectopic IL-24 induction in homeostatic skin epithelium elicits a wound-like response without injury

(A) Schematic of the generation of TRE-IL-24 mice. Selective targeting to skin EpdSCs was achieved by packaging the transgene in a lentivirus and in utero injection into the amniotic sacs of E9.5 mouse embryos genetic for the Krt14-rtTA doxycycline inducible transcriptional activator. The lentivirus also contained a constitutively expressed Pgk-H2BGFP to monitor integration efficiency. Skins were harvested after mice were fed Dox food for 2, 3, or 4 days. (B) Left: images of mice at postnatal days 1 and 4. Note flaky skin phenotype, evident by day-4. Right: Images of hematoxylin and eosin (H&E) staining and trichrome staining performed on sagittal sections of homeostatic skins from Dox-fed WT and Tre-Il24 mice. Quantifications are of epidermal thickness and intensity of trichrome staining to evaluate dermal collagen deposition (n = 3 mice per genotype). (C) Sagittal sections of homeostatic skins from WT and Tre-Il24 mice immunolabeled for Ki67, GFP and CD31. Quantifications are of percentages of proliferating (Ki67⁺) EpdSCs (top), and underlying endothelial cells (Ki67⁺CD31⁺) (middle) and non-endothelial dermal cells (Ki67⁺CD31⁻) (bottom) (n = 3 mice per genotype). (D) Sagittal sections of homeostatic skins from WT and Tre-Il24 mice immunolabeled for GFP and CD31. Quantifications are of percentage of interfollicular epidermis close to CD31⁺ endothelial cells (top), and the distance (mm) between epidermis and CD31⁺ vasculature (bottom) (n = 3 mice per genotype per time point). (E) Sagittal sections of homeostatic skins from WT and Tre-Il24 mice were immunolabeled for GFP and p-STAT3. Prior to collecting skins, mice were given Dox food for 2 days. Quantifications are of percentage of p-STAT3⁺ epidermal, endothelial, and fibroblast cells. Quantifications of dermal cell types were made by performing similar immunofluorescence as for epidermis, but using antibodies against CD31 and PDGFα, respectively (n = 3 mice per genotype). White dotted lines: epidermal-dermal border. DAPI, nuclei; scale bars, 100 μm. Data in (B)–(E) are presented as mean ± SEM. Experiments were performed R ≥3×. Statistical significance was determined using two-tailed unpaired Student’s t tests; **** p < 0.0001; *** p < 0.001; ** p < 0.01; and * p < 0.05.

Figure 5.. Tissue-damage-associated hypoxia and HIF1α in wounds are important for robust Il24 expression

(A) Sagittal section of day-3 wound harvested just after pimonidazole injection to label tissue hypoxia (n = 5 mice). (B) Sagittal section of day-3 wound immunolabeled for CD31 and $\text{HIF1α}$. The distance (μm) from ${\text{HIF1α}}^{\text{Low}}$ vs. ${\text{HIF1α}}^{\text{High}}$ EpdSCs to the nearest CD31⁺ blood vessels is quantified (n = 5 mice). (C) Sagittal sections of day-5 wounds from WT and Il24 null mice immunolabeled for CD31 and $\text{HIF1α}$. Boxed regions of the migrating epidermal tongue are magnified at right (scale bars, 20 μm). b, basal EpdSCs; sb, suprabasal epidermal cells (n = 5 mice per genotype). (D) Schematic of the experiment and qRT-PCR of Il24 and Vegfa mRNA in YFP⁻ ($\mathit{\text{Hif1α}}$ WT) or YFP⁺ ($\mathit{\text{Hif1α}}$Δexon2) FACS-purified EpdSCs from homeostatic skin and from day-1 wounds of Krt14CreER; ${\mathit{\text{Hif1α}}}^{\mathit{\text{fl/fl}}}$; RosaYFP^+/fl mice treated with topical 4OH-Tam (n = 5 mice). White dotted lines: epidermal-dermal border; wound site, red dotted line; epidermal migration direction, red arrow. DAPI, nuclei; scale bars except for the boxed regions: 100 μm. Data in (B) and (D) are presented as mean ± SEM. Experiments were performed R ≥ 3×. Statistical significance was determined using two-tailed unpaired Student’s t tests; **** p < 0.0001; * p < 0.05. See also Figures S6.

Figure 6.. Critical roles for both hypoxia/HIF1α and STAT3 in governing robust Il24 expression

(A) qRT-PCR of Il24 mRNA in keratinocytes with GFP or IL-24-receptor reconstitution cultured under different oxygen, nutrient, substrate, glycolytic product, and oxidative stress conditions for 48 h. AA, amino acid; Leu, leucine. Note that the native IL-24-receptor, robustly expressed by EpdSCs in their native niche in vivo, is silenced under the culture conditions in vitro. (B) EpdSCs were isolated from skins of Krt14CreER; ${\mathit{\text{Hif1α}}}^{\mathit{\text{fl/fl}}}$ mice, reconstituted with either GFP or IL-24-receptor, and cultured in normoxic (21% O₂) or hypoxic (1% O₂) conditions. 4OH-Tam was used to replace the endogenous $\text{HIF1α}$ with $\text{HIF1α}$ lacking the bHLH DNA binding domain ($\mathit{\text{Hif1α}}$Δexon2). Cells were then immunoblotted for $\text{HIF1α}$, LDHA (lactate dehydrogenase A; encoded by a classical hypoxia-sensitive gene), p-STAT3, STAT3, and vinculin as the loading control. (C) qRT-PCR of Il24 mRNA in the cells described in (B). (D) Il24 expression from RNA-seq data performed on FACS-purified EpdSCs from homeostatic skin and day-1 wounds from WT and Krt14Cre; Stat3^fl/fl (Stat3 cKO)mice treated with 4OH-Tam. TPM, transcripts per kilobase million (n = 3 mice for each genotype). (E) Normalized peaks of RNA-seq, assay for transposase-accessible chromatin sequencing (ATAC-seq), and Cut&Run-seq (with IgG control or antibodies against at $\text{HIF1α}$ or STAT3) at the Il24 locus. Red boxes indicate the 5 chromatin regions at the Il24 locus that opened upon wounding (ATAC) and have both $\text{HIF1α}$ and STAT3 binding peaks (Cut&Run). Peaks from the same experiments are indicated on the same scale. Data in (A), (C), and (D) are presented as mean ± SEM. Sequencing experiments were in duplicates; others were performed ≥3×. Statistical significance was determined using two-tailed unpaired Student’s t tests; *** p < 0.001; and ** p < 0.01. See also Figure S7.

Figure 7.. IL-24 signaling promotes epithelial glucose uptake and influences dermal repair

(A) GLUT1 expression is dependent upon both hypoxia and IL-24-receptor-signaling. qRT-PCR and immunoblot analyses showing that both events are essential for optimal GLUT1 expression. (B) Glucose transport family expression from RNA-seq performed on EpdSCs that were FACS-purified from homeostatic skin (unwd_Epi) and day-5 wound (5d_migrating Epi). TPM, transcripts per kilobase million. (C) Sagittal sections of day-3 wounds from WT vs. Il20rb null skins immunolabeled for GLUT1. Graphs show quantifications of the thickness of GLUT1-expressing epidermis (n = 3 mice per genotype). (D) Glut1 expression depends upon STAT3. Left: Slc2a1 mRNA TPM value from RNA-seq of FACS-purified EpdSCs from homeostatic and wounded skin in WT vs. Krt14-Cre; Stat3flfl (Stat3 cKO) mice. Right: sagittal sections of day-3 wounds from WT vs. Krt14-Cre; Stat3flfl;Yfp+/fl (Stat3 cKO) immunolabeled with GLUT1 and YFP (n = 3 mice per genotype). (E) Graphs show relative rates of glucose consumption (left) and lactate production (right) by keratinocytes with GFP or IL-24-receptor reconstitution under normoxic vs. hypoxic conditions. Note that under conditions of hypoxia and IL-24-receptor reconstitution, both measurements are the most elevated. (F) Sagittal sections of day-3 wounds from WT vs. Krt14Cre; Glut1^fl/fl mice treated with topical 4OH-Tam. Sections were immunolabeled for GLUT1 and CD31 (left), or for GLUT1 and $\text{PDGFRα}$ (right). Asterisk (*) in the right images denotes a paucity of fibroblasts (${\text{PDGFRα}}^{+}$) in the dermis of Glut1 cKO skin. Quantifications at right (n = 6 mice per genotype). (G) Model depicting the similarities between evolutionarily conserved pathogen-induced IFN signaling for defense and injury-induced IL-24 signaling for repair. In contrast to pathogens, which lead to induction of IFN and p-STAT1/2, tissue damage causes hypoxia, leading to $\text{HIF1α}$, IL-24, and p-STAT3. Specifically, EpdSCssense wound hypoxia caused by severed blood vessels, and induce IL-24 and receptor signaling, which subsequently activates STAT3 and further fuels Il24 expression to promote a coordinated dermal repair and re-epithelialization. The autocrine and paracrine mechanisms underlying wound-induced IL-24-signaling in tissue repair are parallel and functionally analogous to pathogen-induced IFN signaling in pathogen defense, and the two pathways share multiple levels of homology. White dotted lines, epidermal-dermal border; wound site, red dotted line; epidermal migration direction, red arrow. DAPI, nuclei; scale bars, 100 μm. Data in and (C)–(F) are presented as mean ± SEM. Experiments were performed ≥3×. Statistical significance was determined using two-way ANOVA and Tukey’s multiple comparisons tests in (A) and (E), using one-way ANOVA and Tukey’s multiple comparisons tests in figure (C), and using two-tailed unpaired Student’s t tests in (F); **** p < 0.0001; *** p < 0.001; ** p < 0.01; * p < 0.05; and ns, not significant.