Progenitor-derived endothelin controls dermal sheath contraction for hair follicle regression

Abstract

Substantial follicle remodelling during the regression phase of the hair growth cycle is coordinated by the contraction of the dermal sheath smooth muscle, but how dermal-sheath-generated forces are regulated is unclear. Here, we identify spatiotemporally controlled endothelin signalling-a potent vasoconstriction-regulating pathway-as the key activating mechanism of dermal sheath contraction. Pharmacological blocking or genetic ablation of both endothelin receptors, ET_A and ET_B, impedes dermal sheath contraction and halts follicle regression. Epithelial progenitors at the club hair-epithelial strand bottleneck produce the endothelin ligand ET-1, which is required for follicle regression. ET signalling in dermal sheath cells and downstream contraction is dynamically regulated by cytoplasmic Ca²⁺ levels through cell membrane and sarcoplasmic reticulum calcium channels. Together, these findings illuminate an epithelial-mesenchymal interaction paradigm in which progenitors-destined to undergo programmed cell death-control the contraction of the surrounding sheath smooth muscle to orchestrate homeostatic tissue regression and reorganization for the next stem cell activation and regeneration cycle.

Extended Data Fig. 1. Validation of DS, DP and DF isolation strategy during the growth-to-regression transition

a, Immunofluorescence for pH3 and CASP3* on P16 Tbx18^H2BGFP;Crabp1-GFP; Lef1-RFP reporter mouse back skin (n = 2 mice). P13 anagen back skin served as a positive control for pH3. In the hair bulb, proliferation is decreasing and apoptotic cells are beginning to appear, indicating the growth-to-regression transition. Scale bars, 50 μm. b, Immunofluorescence for aSMA, PDGFRA and ITGA8 on P16 Tbx18^H2BGFP;Crabp1-GFP;Lef1-RFP reporter mouse back skin (n = 2 mice). Scale bars, 50 μm. c, Immunofluorescence for ITGA8, aSMA, and GFP on P16 Tbx18^H2BGFP;Crabp1-GFP;Lef1-RFP reporter mouse back skin, demonstrating colocalization in the dermal sheath (n = 2 mice). Scale bars, 50 μm. d, Immunofluorescence for GFP and RFP on fresh frozen and fixed P16 Tbx18^H2BGFP;Crabp1-GFP;Lef1-RFP reporter mouse back skin, both demonstrating colocalization in the DP (n = 2 mice). Scale bars, 50 μm.

Extended Data Fig. 2. Analysis of DP, DS, and DF transcriptomes at the growth-to-regression transition

a, Principal components analysis of P16 DS, DP, and DF bulk RNA-sequencing data. b, DF, DP and DS RPKM expression levels of Wif1, Acta2 and Lum at regression onset (n = 2, data are mean with individual data points). c, Gene ontology terms enriched in the DS population. “Muscle contraction” is the top enriched GO term for the DS. P < 0.05, Fisher’s exact test. d, Transcription factors, adhesion/ECM molecules, ligands, receptors, and enzymes that are part of the gene signatures of the DS, DP and DF populations. e, Expression of endothelin receptors across cell types from previously published transcriptome data at P5^,(n = 2, data are mean with individual data points).

Extended Data Fig. 3. Endothelin signaling contracts DS cells in a dose-dependent manner

a, Endothelin-1 (ET-1) triggers DS contraction and surface area reduction in DS cells grown on Matrigel in a dose-dependent manner. Scale bars, 50μm. b, Quantification of cell surface areas during contraction time course (0, 30, and 60 minutes). 24, 20, 21, 18 and 19 cells for control, 100 pM, 1 nM, 10 nM, 100 nM and 1uM ET-1, respectively; n = 3 independent experiments. **P = 0.001 (30 min) and 0.002 (60 min) for 1 nM ET-1. ***P = 0.00027 (30 min) and 8.65 x 10⁻⁵ (60 min) for 10 nM ET-1, 1.24 x 10⁻⁵ (30 min) and 2.5 x 10⁻⁵ (60 min) for 100 nM ET-1, and 2.53 x 10⁻⁵ (30 min) and 1.14 x 10⁻⁵ (60 min) for 1 μM ET-1. Data are mean ± s.d. and statistical significance was determined by one-way ANOVA with post-hoc Tukey’s HSD for multiple comparison testing.

Extended Data Fig. 4. Endothelin signaling functionally contracts DS cells independent of proliferation

a, Endothelin-1 (ET-1) triggers contraction and surface area reduction in DS cells grown on matrigel, both in the absence or presence of proliferation inhibition with Mitomycin-C. Scale bars, 50 μm. b, Quantification of cell surface areas during contraction time course. N = 19, 20, 19 cells for control, ET-1, Mitomycin C + ET-1, respectively, from two independent experiments. **P = 0.0015 (5 min) for Mitomycin C + ET-1; ***P = 0.0002 (5 min), 0.0001 (10 min), 1.42 x 10⁻⁵ (20 mins) and 1.12 x 10⁻⁵ (30 mins) for ET-1; 0.0002 (10 min), 3.59 x 10⁻⁵ (20 min) and 61.90 x 10⁻⁵ (30 min) for Mitomycin C + ET-1. Data are mean ± s.d. and statistical significance was determined by one-way ANOVA with post-hoc Tukey’s HSD for multiple comparison testing.

Extended Data Fig. 5. Topical application of endothelin receptor antagonists impedes HF regression with concentrated and local effects

a, Schematic of the experimental design for topical application of endothelin receptor antagonists BQ123 and BQ788 (“BQ”) or vehicle control (DMSO), harvesting of the back skins, and imaging strategy. b, Whole-mount immunofluorescence for K14 from the center of the application area (Zone 1). Most follicles failed to regress in regions treated with BQ123 + BQ788. Scale bars, 50 μm. c, Whole-mount immunofluorescence for K14, LEF1, and DAPI showing a stalled follicle from Zone 1 and a regressed follicle from Zone 3 of BQ123 + BQ788 treated back skin. Scale bars, 50 μm. d, Whole-mount immunofluorescence for K14 at the edge (Zone 2) and just outside of (Zone 3) the application area, demonstrating a progressive decline in stalled follicles towards the periphery. Scale bars, 50μm. e, Quantification of stalled HFs from each of the three zones. 1200 follicles per zone for vehicle control; 938, 841, and 1087 follicles for Zone 1, 2, and 3, respectively, in BQ123 + BQ788 treated regions; n = 4 mice). *P = 0.0228 (Zone 3), ***P = 1.69 x 10⁻¹⁰ (Zone 1) and 2.42 x 10⁻⁷ (Zone 2), unpaired two-tailed Student’s t-test. Data are mean ± s.d. with individual data points.

Extended Data Fig. 6. Topical application of endothelin receptor antagonists impedes follicle regression during the second hair cycle

a, Schematic of the experimental design for topical application of endothelin receptor antagonists BQ123 and BQ788 (“BQ”) or vehicle control (DMSO), harvesting of the back skins, and imaging strategy. b, Whole-mount immunofluorescence for K14 and DAPI in DMSO (vehicle control) and BQ123 + BQ788 treated back skin. Scale bars, 200 μm. c, Quantification of stalled HFs observed in DMSO (vehicle control) and BQ123 + BQ788 treated back skins. 924 follicles for DMSO and 544 follicles for BQ123 + BQ788, n = 3 mice. **P = 0.0031, unpaired two-tailed Student’s t-test. Data are mean ± s.d. with individual data points. d, DS-specific Ednra ablation at P17. Low magnification image of the follicle from the high magnification image in Fig. 3b. e, DS-specific Ednrb ablation at P17. Low magnification image of the follicle from the high magnification image in Fig. 3E. Scale bars, 50 μm.

Extended Data Fig. 7. Stalling follicles, but not regressing follicles, in dcKO back skin lack both ET_A and ET_B receptors

a, Schematic of DS-specific Ednra and Ednrb double conditional genetic ablation (dcKO) in the skin. b, Immunofluorescence for K14 and both ET_A/ET_B in control P17 back skins. Endothelin receptors are expressed in the DS. c, Immunofluorescence for K14 and both ET_A/ET_B in DS-specific dcKO P17 back skin. Only stalling follicles lack both receptors ET_A and ET_B in the DS. Scale bars, 50 μm. n = 4 mice. Green arrows = presence of ET_A and/or ET_B in the DS. White arrowheads = lack of both ET_A and ET_B in the DS.

Extended Data Fig. 8. Patterns of endothelin ligand expression during regression

a, Edn3 mRNA expression across cell types from previously published transcriptome data at P5^,(n = 2, data are mean with individual data points). b, Edn1 smFISH in a follicle transitioning from early-to-mid regression (n = 3 mice). While only very diffuse expression of Edn1 is detectable throughout most of the ORS, the progenitors located in the bottleneck region of the follicle exhibited very high focal expression of Edn1, corresponding to the site of known DS contraction. Scale bars, 50 μm. c, Fire LUT conversion of ET-1 immunofluorescence signal from Fig. 5d highlights the high focal expression of ET-1 in the bottleneck region of regressing follicles. White arrows indicate the bottleneck region. d, Schematic of ET-1 expression during different stages of regression.

Extended Data Fig. 9. ET-1 ablation in ORS progenitors by mid-regression

Immunofluorescence for K14, ET-1 and CASP3* in P17 back skins from Edn1 cKO and control (n = 4 mice per condition). A marked reduction in both ET-1 and CASP3* in the K14⁺ ORS progenitors of the HFs in cKO back skins was observed compared to control back skins. Scale bars, 50 μm. b, Insets showing K14 and ET-1 colocalization in regressing HF from control back skin, with high focal expression of ET-1 in the bottleneck region. In Edn1 cKO back skins, some HFs exhibit complete ET-1 ablation while other HFs only exhibit partial ET-1 ablation in the K14⁺ ORS progenitors. Scale bars, 50 μm.

Extended Data Fig. 10. ET-1-induced DS contraction depends on dynamically regulated cytoplasmic Ca²⁺

a, Images of Fluo8 levels in DS cells from Fig. 7b including intermediate timepoints (5, 10, 20 mins) during the 30-minute ET-1 or vehicle control (PBS) exposure; n = 2 independent experiments. Scale bars, 50 μm. b, Images of tdT-labeled DS cells at 0, 5, 10, 20 and 30 minutes of ET-1 or vehicle control (PBS) exposure following a 1 h preincubation in calcium channel blockers (50μM NNC 55-0396 + 50 μM Diltiazem + 10 μM Ryanodine + 10 μM Xestospongin C), ML7 (200 μM), or vehicle control (DMSO); n = 2 independent experiments. Endothelin-mediated DS contraction is abrogated by either blocking of calcium channels or by inhibition of MLCK activity. Scale bars, 50 μm.

Fig. 1.. The dermal sheath expresses receptors of the endothelin signaling smooth muscle contraction pathway.

a, Schematic of regressing hair follicle illustrating apoptosis, contraction and movement processes during the destruction phase of the hair growth cycle. The epithelial progenitor pool (blue) is reduced through apoptosis in the epithelial strand between hair shaft (black) and dermal papilla (DP, red). Dermal sheath (DS) contraction at the bottleneck between epithelial strand and club hair pushes the shaft up and pulls the DP up. As shaft/strand/DP move up towards the stem cell reservoir new progenitors enter the strand keeping it at a constant length. The molecular mechanisms controlling DS contraction and centripetal constriction are unknown. b, Labeling of DS (arrowheads) and DP (arrow) with Tbx18^H2BGFP;Crabp1-GFP;Lef1-RFP transgenic reporters. Scale bar, 50 μm. c, Isolation of DS and DP from early-stage regressing follicles (P16) by flow cell sorting of transgenic reporter back skin after immunofluorescence for PDGFRA and ITGA8. DF were isolated for comparison. d, Hierarchical clustering of DS, DP, DF with differentially expressed genes. N = 2, both replicates are shown. e, Venn diagram of gene signatures. f, DS signature genes in key gene categories, identifying both endothelin receptors, Ednra and Ednrb. g, Expression levels of receptors of the major smooth muscle contraction-regulating signaling pathways represented as reads per kilobase per million mapped reads (RPKM). N = 2, data are mean with individual data points. h, DF, DP and DS expression levels of Ednra and Ednrb at regression onset. N = 2, data are mean with individual data points. i, Immunofluorescence for endothelin receptors ET_A and ET_B (n = 3 mice). ITGA8 marks the DS in regressing P17 hair follicles. Scale bar, 50 μm.

Fig. 2.. Endothelin signaling activation contracts DS cells and pharmacological inhibition impairs DS contraction and follicle regression.

a, Schematic of DS contraction in vitro assay. Short-term cultured DS cells are genetically labeled with tdTomato (tdT) and incubated with Endothelin-1 (ET-1). b, ET-1 triggers DS contraction and surface area reduction in DS cells grown on matrigel. Scale bar, 50 μm. c, Quantification of cell surface areas during contraction time course. N = 24, 21, 19 cells for control, 1 nM ET-1, 100 nM ET-1, respectively from three independent experiments. **P = 0.001 (30 min) and 0.002 (60 min) for 1 nM ET-1. ***P = 1.24 x 10⁻⁵ (30 min) and 2.5 x 10⁻⁵ (60 min) for 100 nM ET-1. Data are mean ± s.d. and statistical significance was determined by one-way ANOVA with post-hoc Tukey’s HSD for multiple comparison testing. d, Schematic of endothelin receptor antagonist BQ123/BQ788 or DMSO topical application to back skin during regression. e, Whole-mount immunofluorescence for K14 shows complete regression of follicles into the resting phase of the hair cycle in control-treated back skins, but impaired regression in BQ123/BQ788-treated back skins. Scale bar, 50 μm. f, Quantification of the percentage of stalled follicles in the center of the application area (1200 control follicles; 938 BQ123/BQ788-treated follicles; n = 4 mice). ***P = 1.68 x 10⁻¹⁰, unpaired two-tailed Student’s t-test. Data are mean ± s.d. with individual data points.

Fig. 3.. Genetic ablation of either endothelin receptor does not impair contraction and regression.

a, Schematic of DS-specific Ednra conditional knockout (cKO) in the skin. b, ET_A ablation by mid-catagen (P17). Immunofluorescence for ET_A imaged in the bottleneck region hair follicles during the regression phase. All follicles in control back skins express ET_A on the surface of ITGA8⁺ DS. Follicles in Ednra ablated back skin lack ET_A in ITGA8-marked DS. Scale bar, 50 μm. c, Whole-mount immunofluorescence for K14 shows normal and complete regression of follicles in both control and Ednra ablated back skins. Scale bar, 50 μm. d, Quantification of fully regressed hair follicles in DS Ednra cKO and control back skins (2500 follicles each for control and cKO; n = 10 mice). e, Schematic of DS-specific Ednrb cKO. f, ET_B ablation by mid-catagen (P17). Immunofluorescence for ITGA8 and ET_B at the follicle bottleneck. All control follicles express ET_B in ITGA8⁺ DS. Follicles in Ednrb ablated back skin lack ET_B in ITGA8⁺ DS. Scale bar, 50 μm. g, Whole-mount immunofluorescence for K14 shows complete regression of follicles in Ednrb ablated back skins. Scale bar, 50μm. h, Quantification of fully regressed hair follicles in DS Ednrb cKO and control back skins (~2500 follicles each for control and cKO; n = 10 mice).

Fig. 4.. Genetic dual ablation of both endothelin receptors in the DS impairs follicle regression.

a, Schematic of DS-specific Ednra and Ednrb double conditional knockout (dcKO). b, Whole-mount immunofluorescence for K14 with complete follicle regression in control back skins. In dcKO back skins many HFs failed to regress. c, Stalled HFs retained their hair shafts down to the follicle base with DP positioned at the proximal end. Scale bars, 50 μm. d, Quantification of stalled follicles in dcKO back skins (3,592 control and 3,731 dcKO follicles; n = 12 mice). ***P = 0.00035, unpaired two-tailed Student’s t-test. Data are mean ± s.d. with individual data points. e, Immunofluorescence for ITGA8 of Ednra;Ednrb dcKO back skins at P20 (n = 2 mice). APM, arrector pili muscle. Scale bars, 50 μm. f, Immunofluorescence for ITGA9 in dcKO back skin (n = 2 mice). Scale bars, 50 μm. g, Immunofluorescence for K14 and apoptosis marker CASP3* or proliferation markers Ki67 and pH3 in dcKO back skin (n = 2 mice). K14⁺ epithelial cells of stalled follicles lack apoptosis and proliferation. Regressing (P17) and growth-phase (P12) follicles served as positive controls. Scale bars, 50 μm. h, Immunofluorescence for K14 and ET_A/ET_B in dcKO back skin at P17 demonstrates correlation of stalling phenotype penetrance with ablation efficiency. Only stalled follicles lack both ET_A and ET_B (n = 4 mice). Green arrows and white arrowheads highlight the presence and absence, respectively, of ET_A/ET_B in the DS. Scale bars, 50 μm. i, Quantification of ET_A and ET_B ablation efficiency in regressing and stalling follicles (30 control follicles, n = 3 mice; 39 dcKO regressing follicles and 20 dcKO stalling follicles, n = 4 mice). ***P = 0.00028, unpaired two-tailed Student’s t-test. Data are mean ± s.d. with individual data points.

Fig. 5.. Spatiotemporal endothelin expression and DS contraction during follicle regression.

a, Schematic illustrating potential sources of endothelin ligand(s) that may induce endothelin receptor activation and DS contraction during regression. b, RPKM expression levels of endothelin ligands Edn1 and Edn2 from P5 transcriptome data and combined analysis^,. The ORS progenitor population has enriched Edn1 expression. N=2, data are mean with individual data points. c, Single-molecule fluorescence in situ hybridization for Edn1 mRNA during follicle regression phase. Focal expression of Edn1 is detectable in the bottleneck – corresponding to the known site of DS contraction (n = 3 mice). White arrows indicate the bottleneck region of the regressing follicles. d, Immunofluorescence for ET-1 protein in K14^H2BGFP P17 back skins during all stages of the regression phase of the hair cycle, starting from the transition from growth through early, mid and late regression. H2BGFP marks all epithelial cells. ITGA6 marks basal epithelial cells during all regression stages. (n = 3 mice). Low ET-1 expression is seen during the transition (“trans”) and early regression. Focal, strong ET-1 upregulation in the bottleneck of mid-regressing follicles. ET-1 expression decreases in the latest stage when the follicle is largely regressed just prior to the rest phase. Scale bars, 50 μm. e, High magnification of ET-1 and ITGA6 localization in the bottleneck region of P17 regressing follicle (n = 3 mice). Scale bar, 50 μm. f, phospho-Myl9 (activated) immunofluorescence shows enrichment in the DS surrounding and below the bottleneck relative to DS above the bottleneck (n = 2 mice). g, Schematic representation of spatially regulated DS contraction by progenitor-derived endothelin.

Fig. 6.. Endothelin-1 from ORS progenitors is required for follicle regression.

a, Schematic of epithelial-specific Edn1 cKO during the first hair cycle. b, Whole-mount immunofluorescence for K14 reveals massive stalling and failed follicle regression in Edn1 cKO back skins by P20. Scale bar, 50 μm. c, Quantification of stalled follicles in cKO back skins (1500 follicles in control and 1340 in dcKO, n = 6 mice). **P = 0.0018, unpaired two-tailed Student’s t-test. Data are mean ± s.d. with individual data points. d, Schematic of epithelial-specific Edn1 cKO during the second hair growth cycle through administration of 4-hydroxy tamoxifen (4-OHT) either by topical application or intradermal injection. e, Whole-mount immunofluorescence for K14. Widespread stalling and failed follicle regression in 4-OHT-induced Edn1 cKO back skins during the second hair growth cycle (n = 2 mice). Scale bars, 100 μm.

Fig. 7.. ET-1 regulates DS contraction through cytoplasmic Ca²⁺ signaling.

a, Schematic of Fluo8 calcium indicator assay to detect cytoplasmic Ca²⁺ in short-term cultured DS cells with ET-1 (1 μM) or vehicle control and preincubations of endothelin receptor antagonists BQ123 (1 μM) and BQ788 (1 μM). b, Fluo8 fluorescence in DS cells at 0 and 30 min of ET-1 or vehicle (PBS) with or without preincubation in BQ123, BQ788 or both. Fluo8 levels increased with ET-1, which was partially or fully blocked by individual or combined antagonists, respectively. Scale bars, 50 μm. c, Quantification of Fluo8 fluorescence levels. N = 17, 15, 16, 15, 17 cells for control, ET-1, ET-1+BQ123, ET-1+BQ788, ET-1+BQ123+BQ788, respectively, from two independent experiments. **P < 0.01, ***P < 0.001 (P values are listed in methods), one-way ANOVA with post-hoc Tukey’s HSD for multiple comparison testing. Data are mean ± s.d. d, Schematic of calcium-dependent ET_A/ET_B signaling pathways and small molecule inhibition of calcium channels. e, Fluorescence of tdT-marked DS cells at 0 and 30 min of ET-1 (1μM) or vehicle (PBS) exposure following 1h preincubation in calcium channel blockers, MLCK inhibitor ML7, or DMSO control. Scale bars, 50 μm. f, Quantification of DS cell surface area over time. N = 19, 18, 22, 22 cells for control, ET-1, ET-1+Ca²⁺ channel inhibitors, ET-1+ML-7, respectively, from two independent experiments. ***P < 0.001 (P values are listed in methods), one-way ANOVA with post-hoc Tukey’s HSD. Data are mean ± s.d. g, Schematic of spatiotemporal contraction regulation and molecular mechanism of endothelin signaling. High ET-1 at the bottleneck between epithelial strand and club hair activates endothelin signaling and contraction in neighboring DS. ET-1 binding to endothelin receptors activates Ca²⁺ influx and Ca²⁺/calmodulin/MLCK signaling resulting in DS cell contraction. h, “Rolling wave” model: As the hair shaft gets pushed upwards, the bottleneck and epithelial strand structures move up. ORS cells previously located at the bottleneck become situated in the epithelial strand and undergo apoptosis, while cells previously above the bottleneck now join the bottleneck, produce high ET-1 and regulate DS contraction.