Epidermal β-catenin activation remodels the dermis via paracrine signalling to distinct fibroblast lineages

Abstract

Sustained epidermal Wnt/β-catenin signalling expands the stem cell compartment and induces ectopic hair follicles (EFs). This is accompanied by extensive fibroblast proliferation and extracellular matrix (ECM) remodelling in the underlying dermis. Here we show that epidermal Hedgehog (Hh) and Transforming growth factor-beta (TGF-β) signalling mediate the dermal changes. Pharmacological inhibition or genetic deletion of these pathways prevents β-catenin-induced dermal reprogramming and EF formation. Epidermal Shh stimulates proliferation of the papillary fibroblast lineage, whereas TGF-β2 controls proliferation, differentiation and ECM production by reticular fibroblasts. Hh inhibitors do not affect TGF-β target gene expression in reticular fibroblasts, and TGF-β inhibition does not prevent Hh target gene induction in papillary fibroblasts. However, when Hh signalling is inhibited the reticular dermis does not respond to epidermal β-catenin activation. We conclude that the dermal response to epidermal Wnt/β-catenin signalling depends on distinct fibroblast lineages responding to different paracrine signals.

Figure 1. Reprogrammed fibroblasts retain increased proliferative potential in culture.

(a) Outline of experimental procedure for preparing and repopulating de-epidermized dermis (DED) from murine skin. (b,c) Sections of P2 DEDs after 2 weeks of culture stained with antibodies to PDGFRα (green) and collagen 3 (red), counterstained with 4,6-diamidino-2-phenylindole (DAPI; blue). DEDs were unseeded (b) or seeded with 2 × 10⁵ neonatal fibroblasts (c). Scale bars, 50 μm (d) Quantification of fibroblasts isolated from 2-day-old mice and seeded onto P2 or adult telogen (TELO) DEDs. Fibroblasts were cultured in DMEM/10% FCS for 2–3 weeks. *P≤0.05; **P≤0.005; ***P≤0.0005 between cells seeded on neonatal or adult DEDs; ^#P≤0.05; ^##P≤0.005; compared with 1 × 10⁵ seeded fibroblasts; by a one-way analysis of variance (ANOVA). (e) Quantification of fibroblasts isolated from neonatal (P2), telogen skin (Telo) or skin containing ectopic follicles (EF). (e) *P≤0.05; between cells seeded on neonatal or adult DEDs; ^#P≤0.05; ^##P≤0.005; compared with telogen; by a one-way ANOVA. MF, microscopic field. (d,e) Data show means±s.e.m. of triplicate samples in a representative experiment (n=3 independent experiments). (f-i) Bar graphs showing percentage of proliferating (f,g) and apoptotic (h,i) fibroblasts isolated from neonatal (P2), telogen (Telo), anagen (ANA) and reprogrammed (EF) skin quantified 48 h after seeding flow-sorted fibroblasts onto culture dishes (g,i) or 3 weeks after seeding flow-sorted fibroblasts onto P2 DEDs (f,h). (f–i) Data represent mean±s.e.m. (n=3). *P≤0.05, **P≤0.005 by a one-way ANOVA, compared with telogen; ^#P≤0.05, compared with P2.

Figure 2. Pharmacological inhibition of FGF, TGF-β or Hh signalling impairs ectopic follicle (EF) formation in vivo.

(a) Heat map (blue: low expression; red: high expression) showing differential expression of growth factors in epidermis from neonatal, adult telogen (resting phase), anagen (active phase) and reprogrammed (EF) skin. (b) Relative expression of the indicated growth factors in flow-sorted epidermal cells measured by qRT–PCR. Data represent mean±s.e.m. (n=3–4). *P≤0.05; **P≤0.005; compared to telogen; ^#P≤0.05; ^##P≤0.005; compared to anagen; ^§P≤0.05; compared to P2; by a one-way analysis of variance (ANOVA). c,d) Relative expression of growth factors in epidermal cells (c) and growth factor target genes in fibroblasts (d) isolated from K14ΔNβ-cateninER transgenic mice treated seven times with vehicle alone (0x) or with 4OHT once or up to seven times. Data represent mean±s.e.m. (n=2). *P≤0.05; **P≤0.005; by Student's T test. (e) Percentage of proliferating (EdU+) fibroblasts 24 or 48 h after treatment with the indicated growth factors or inhibitors. Green and red dotted lines represent the mean percentage of vehicle-treated cells after 24 h (red) and 48 h (green). Cells were pulsed with EdU for 2 h. Cells were pooled from three technical replicates and the results are representative of n=3 independent experiments. (f) Bar graphs showing relative expression of the indicated genes in fibroblasts isolated from adult skin and treated with Shh, TGF-β2, FGF2 or vehicle in vitro. Data represent mean±s.e.m. (n=3). *P≤0.05; **P≤0.005 by a one-way analysis of variance. (g) Treatment scheme. Topical application of 4OHT induces Wnt/β-catenin activation in the epidermis of K14ΔNβ-cateninER transgenic mice. Inhibitors were applied by subcutaneous injection (RepSox, PD173074) or oral gavage (IPI4182) every other day. (h–j) Inhibition of FGF (h), TGF-β (i) Hh (j) signalling impairs or blocks induction of anagen and EFs in K14ΔNβ-cateninER transgenic mice. Representative haematoxylin and eosin stainings are shown. AD, adipocytes; ANA, anagen hair follicle; HF, hair follicle; RD, reticular dermis. Scale bars, 200 μm.

Figure 3. Effects of pharmacological Hh and TGF-β inhibitors on fibroblast proliferation, ECM remodelling and adipogenesis.

(a,b) Immunofluorescent staining of skin sections from telogen and reprogrammed skin from K14ΔNβ-cateninER/ PDGFRαH2BeGFP double transgenic mice showing an increase in dermal fibroblasts upon epidermal Wnt/β-catenin activation. Keratin 14 is expressed in epidermal cells. All fibroblasts express H2BeGFP in their nuclei. Nuclei were detected with 4,6-diamidino-2-phenylindole (DAPI). (c–e) Quantification of proliferating fibroblasts in K14ΔNβ-cateninER/ PDGFRαH2BeGFP double transgenic mice treated with up to six doses of 4OHT (c) or treated with Hh (d) and TGF-β inhibitors (e). Quantification was performed by FACS (c,e) or in immunolabelled skin sections (d). Mice were injected with EdU 3 (c,e) or 4 h (d) before tissue collection. Data represent mean±s.e.m. (n=3–4). *P≤0.05, inhibitor-treated compared with vehicle-treated mice; ^#P≤0.05, EF skin compared with telogen skin, by a one-way analysis of variance (ANOVA). (f,g) Herovici staining of skin sections from telogen and reprogrammed (EF) skin treated with Hh (f) or TGF-β inhibitors (g). Immature collagen fibres appear blue, whereas mature collagen is stained pink. Red box highlights a patch of immature collagen in the lower dermis (f). (h,i) Relative expression of the indicated genes in fibroblasts isolated from telogen and reprogrammed (EF) skin treated with IPI4182, RepSox or vehicle. Data represent mean±s.e.m. (n=4–5). (j) Number of adipocytes per microscopic field (MF) quantified in H&E-stained sections. Data represent mean±s.e.m. (n=3). (h–j) *P≤0.05; **P≤0.005, inhibitor-treated compared with vehicle-treated mice; ^#P≤0.05, ^##P≤0.005 compared with telogen skin by a one-way ANOVA. (k,l) Immunofluorescent staining of skin sections with antibodies to Caveolin and K14 or Lipidtox, counterstained with DAPI. Scale bars in a,b,f,g,k and l represent 200 μm.

Figure 4. Hh and TGF-β inhibitors affect different fibroblast lineages.

(a,b) Labelling of the reticular dermal lineage using Dlk1 expression. (a) Experimental strategy; (b) adult telogen skin. (c–e) Labelling the papillary dermal lineage using Lrig1 expression. (c) Experimental strategy; (d,e) adult telogen (d) and anagen (e) skin. (f–h) Labelling Lrig1+ cells in adult skin. (f) Experimental strategy; (g,h) K14ΔNβ-cateninER (h) and wild-type littermate control (g) skin following six doses of 4OHT. (b,d,e,g,h) Sections were stained for tdTomato and nuclei were detected with 4,6-diamidino-2-phenylindole (DAPI). Scale bars represent 200 μm (b,d,e) and 100 μm (g,h). (i) Section of 4OHT-treated K14ΔNβ-cateninER skin stained for Lrig1 and GFP. Red arrow heads show Lrig1+ fibroblasts associated with ectopic HF. (j–l) Flow cytometry plots showing fibroblast populations isolated from telogen (j), anagen (k) and EF (l) skin. PDGFRαeGFP was used to gate on fibroblasts. Plots are representative of n=3-4 mice per condition. (m–o) Bar graphs showing flow cytometric quantification of proliferating Lrig1+ and Sca1+ (m,n) and total Sca1+ fibroblasts (o). Data represent mean±s.e.m. (n=3–4). (m,n) Cells were labelled for 4 h with EdU. (p–v) Representative flow cytometry plots (p,s) and summary of flow cytometric analysis of fibroblast populations after gating on PDGFRαH2BeGFP+ cells (q,r,t–v) isolated from telogen and reprogrammed (EF) skin treated with the Hh inhibitor IPI4182 (p–r), RepSox (s–v) or vehicle. Data represent mean±s.e.m. (n=3–4). *^,#P≤0.05; **P≤0.005 by a one-way analysis of variance, compared with telogen or EF skin, respectively.

Figure 5. Deletion of Smoothened in fibroblasts affects EF formation.

K14ΔNβ-cateninER mice were examined in telogen or following 4OHT treatment to induce ectopic hair follicles (EFs) in combination with heterozygous Smoothened deletion via Dermo1Cre (SmoΔDermo/+) or homozygous Smoothened deletion via PdgfraCreER (SmoΔPdgfrα). (a–c) H&E (a) and immunofluorescent staining (b,c) of dorsal skin sections. In b,c, sections were labelled with antibodies detecting GFP and K14. 4,6-Diamidino-2-phenylindole (DAPI) was used to visualize nuclei. (c) Higher magnification images of b. (d) Representative flow cytometry plots of Lrig1+ fibroblasts after gating on PDGFRαH2BeGFP+ cells. (e–g) Summary of flow cytometric analysis. Data represent mean±s.e.m. (n=3). *^,#P≤0.05; **P≤0.005 by a one-way analysis of variance, compared with telogen or EF skin, respectively. (h) Immunofluorescent staining with antibodies detecting Caveolin and K14, counterstained with DAPI. (i) Herovici staining. Scale bars, 200 μm.

Figure 6. Epidermal β-catenin activation induces Hh and TGF-β signalling in distinct fibroblast lineages

(a,b) Immunolabelling (a) and Herovici staining (b) of skin from K14ΔNβ-cateninER mice in telogen, following 4OHT treatment to induce ectopic hair follicles (EFs) or in combination with heterozygous Smoothened deletion via Blimp1Cre (SmoΔBlimp1/+). Scale bars, 200 μm. (c) Representative flow cytometry plots of Sca1+ and Lrig1+ fibroblasts after gating on PDGFRαH2BeGFP+ cells. (d–f) Summary of flow cytometric analysis. Data represent mean±s.e.m. (n=3). *^,#P≤0.05; ***^,###P≤0.0005 by a one-way analysis of variance, compared with telogen or EF skin, respectively. (g–i) Labelling Lrig1+ cells. (g) Experimental strategy; (h,i) skin sections from adult mice treated with vehicle (h) or RepSox (i). Insets are higher magnification images. (j–n) Labelling Dlk1+ cells. (j) Experimental strategy; (k–n) skin sections from adult mice treated with vehicle (k,m) or RepSox (l,n). (m,n) Grey scale images with higher magnification of the red channel in e,f, respectively. AD, adipocyte layer; PC, panniculus carnosus. Dotted lines denote thickness of adipocyte layer. Scale bars, 200 μm. (o,p) Quantification of tdTomato+ cells within the interfollicular dermis (IFD) and AD of skin treated with RepSox or vehicle in Lrig1CreER (o) or Dlk1CreER (p) transgenic mice shown in g–n. Data represent mean±s.e.m. (n=3; 3 fields quantified per biological sample). *P≤0.05, ***P≤0.0005 by Student's t-test. (q–s) Relative expression of TGF-β (q) and Hh target genes (s) and TGF-β receptors (TGFBR; r) in Sca1− and Sca1+ (q) or Lrig1+ and Lrig1− (r,s) fibroblasts in adult dermis. Data represent mean±s.e.m. (n=3–4). Data were normalized to Sca1− or Lrig1+ cells in telogen. *P≤0.05, **P≤0.005 by Student's t-test. (t,u) Relative expression of Hh target genes (t) and TGF-β and FGF target genes (u) in Sca1+ and Sca1− fibroblasts isolated from K14ΔNβ-cateninER/PDGFRαH2BeGFP double transgenic mice treated with 4OHT once or up to seven times. Data represent mean±s.e.m. (n=2; pooled from 3 to 5 mice).