Hes1 regulates anagen initiation and hair follicle regeneration through modulation of hedgehog signaling

Abstract

Adult hair follicles undergo repeated cycling of regression (catagen), resting (telogen), and growth (anagen), which is maintained by hair follicle stem cells (HFSCs). The mechanism underlying hair growth initiation and HFSC maintenance is not fully understood. Here, by epithelial deletion of Hes1, a major Notch downstream transcriptional repressor, we found that hair growth is retarded, but the hair cycle progresses normally. Hes1 is specifically upregulated in the lower bulge/HG during anagen initiation. Accordingly, loss of Hes1 results in delayed activation of the secondary hair germ (HG) and shortened anagen phase. This developmental delay causes reduced hair shaft length but not identity changes in follicular lineages. Remarkably, Hes1 ablation results in impaired hair regeneration upon repetitive depilation. Microarray gene profiling on HFSCs indicates that Hes1 modulates Shh responsiveness in anagen initiation. Using primary keratinocyte cultures, we demonstrated that Hes1 deletion negatively influences ciliogenesis and Smoothened ciliary accumulation upon Shh treatment. Furthermore, transient application of Smoothened agonist during repetitive depilation can rescue anagen initiation and HFSC self-renewal in Hes1-deficient hair follicles. We reveal a critical function of Hes1 in potentiating Shh signaling in anagen initiation, which allows sufficient signaling strength to expand the HG and replenish HFSCs to maintain the hair cycle homeostasis.

Significance statement

The adult hair follicles cycle through regression, resting, and growth phases, which is maintained by hair follicle stem cells. During hair growth, progenitors and stem cells of the hair follicle are activated to sustain the downward growth of hair follicles. The understanding of progenitor activation and stem cell maintenance during the hair cycle is still not complete. The present study uncovers a potential link between Notch/Hes1 and Sonic Hedgehog pathways, in which Hes1 reinforces Hedgehog signaling at the onset of hair growth to expand the progenitors and replenish the stem cells to maintain the hair cycle homeostasis.

Working model: By modulating ciliary function, Hes1 potentiates Shh signaling in anagen initiation, which allows sufficient signaling strength to expand the hair germ and replenish HFSCs to maintain the hair cycle.

adult stem cells, cellular proliferation, epidermis, notch, signal transduction

Introduction

Adult stem cells maintain tissue homeostasis and regeneration throughout an animal's lifetime. The murine hair follicle (HF) provides a model system for the mechanistic study of stem cell behavior during tissue regeneration. The HF consists of three regions: the lower segment (bulb), middle segment (bulge and isthmus), and upper segment (infundibulum). After initial morphogenesis, the lower segment of HFs undergoes repeated cycles of regression (catagen), resting (telogen), and growth (anagen) phases. Underpinning this regenerative cycle is the multipotent and self-renewal capability of hair follicle stem cells (HFSCs), which reside in a specialized niche called the bulge.¹

In telogen the bulge HFSCs and secondary hair germ (HG), a small cluster of founder cells beneath the bulge, are kept quiescent through actively repressive signals coming from the niche components and extrafollicular environment.² Counteracting regulatory pathways which include activating Wnt signaling and inhibitory BMP signaling are involved in hair growth. At anagen onset, the HG becomes activated prior to bulge HFSCs by responding to BMP inhibitors and Wnt activators produced by the dermal papillae (DP), a population of mesenchymal cells that directly adjoins the HG, as well as the surrounding macroenvironment. The progeny of proliferative HG then expands downward and generates the hair matrix (Mx). The HG-derived transit-amplifying cells (TACs) in the Mx rapidly proliferate and differentiate into the hair shaft and inner root sheath (IRS) during anagen. To sustain anagen progression, TACs in early anagen secrete Shh to promote bulge HFSC proliferation and to stimulate dermal factors to support TAC expansion.³ In catagen, the hair progeny (Mx, lower ORS) undergoes apoptosis and the remaining epithelial strand retracts upward together with the DP. At the catagen/telogen transition, some slow-cycling upper ORS cells survive after catagen to become the new bulge/HG and fuel the next hair cycle.^4-6

Notch signaling involves ligand-receptor interactions between contacting cells, leading to serial proteolysis of the Notch receptor. This generates the Notch intracellular domain that translocates into the nucleus where it binds Rbpj and Mastermind to activate downstream effectors, including the Hes and Hey gene families of transcriptional repressors.⁷ Loss and gain-of-function animal studies revealed that the canonical Notch-Rbpj signaling axis acts as a commitment switch at the basal/suprabasal layer of the epidermis.⁸ Loss of Notch signaling does not affect HF patterning or hair placode formation; however, it was shown that HF terminal differentiation requires Notch activity.^{8,, 9} Whether Notch signaling plays a role in HFSC activation and HF cycling remains elusive, since ablation of Notch1 in HFs causes smaller hair bulb and premature catagen entry.^{10,, 11}

The basic helix-loop-helix gene Hes1 is an important effector mediating context-dependent functions of Notch signaling in a variety of tissue types. Hes1 maintains the stem/progenitor cells in the nervous and digestive systems by negatively regulating tissue-specific basic helix-loop-helix activators.¹² Moreover, Hes1 is expressed in spinous keratinocytes and keeps their progenitor fate during epidermal development.¹³ Interestingly, the Hes1-null epidermis developed normally when transplanted to adult mice, suggesting a restricted role of Hes1 in developmental stages. Although Hes1 is expressed at low levels in telogen HFs, its expression is increased in growing HFs.¹⁴ As a major Notch downstream effector, the role of Hes1 in HF differentiation and regenerative hair cycling remains unclear.

Hedgehog signaling is initiated by hedgehog ligands (Sonic Hedgehog, Indian Hedgehog, and Desert Hedgehog) binding to Patched receptor, which derepresses and allows accumulation of Smoothened (Smo) in the primary cilium.¹⁵ Smo activation transmits downstream signaling cascade to Gli family zinc finger transcription factors, which govern Hedgehog target gene expression. The Hedgehog signaling pathway functions in both the epithelium and mesenchyme during HF development.¹⁶ Studies in Sonic Hedgehog (Shh) conventional knockout mice reveal that Shh signaling is dispensable for HF initial morphogenesis but required for HF down-growth in the maturation phase. The smaller DP developed in Shh knockout mice also suggested that Shh is required for DP maintenance.^{17,, 18} Hedgehog signaling controls numerous developmental processes in a duration- and intensity-dependent manner.¹⁹

We have demonstrated previously that ablation of Pofut1, a critical component of Notch signaling, in HF lineages resulted in disrupted telogen-anagen transition.²⁰  Pofut1-deficient HFs turn into cysts at the second hair cycle, which prevented studying how the Notch-Hes1 axis participates in hair cycle homeostasis. In this study, we inactivate Hes1 in the skin using the K14-Cre driver and describe a novel role for Hes1 in regulating anagen initiation and HF regeneration through modulation of Shh responsiveness.

Materials And Methods

Animals

Generation of floxed Hes1 (Hes1^fx/fx) has been described previously.²¹  Hes1^fx/fx mice were in ICR background and back-crossed to C57bl/6 for 3 generations. Both Rosa26 Cre reporter and K14-Cre mice were obtained from the Jackson laboratory (Bar Harbor, ME) and maintained in C57bl/6 background. Hes1^fx/fx mice were crossed with K14-Cre mice to generate heterozygous K14-Cre^+/wt;Hes1^fx/wt mice and followed by crossing with Hes1^fx/fx mice to create Hes1^fx/fx;K14-Cre conditional knockout (Hes1eKO) mice. Age- and gender- matched littermate controls (Hes1^fx/fx or Hes1^fx/wt) were used for comparison. Genotyping was performed on tail biopsies by PCR. For depilation experiments, back skin of the anesthetized mice was shaved and depilated mechanically using the Wax Strip Kit (VIGILL Pharma. Co., Taiwan). For Smoothened agonist (SAG, Santa Cruz, Dallas, Texas) rescue experiments, mice were topically applied with 25 μL vehicle (95% acetone/5% dimethyl sulfoxide) and SAG (120 μM) at opposite sides of the dorsal skin daily for consecutive 6 days after depilation. For intradermal delivery of growth factors, Affi-Gel Blue gel beads (Bio-Rad, Hercules, California) were coated with recombinant Shh-N (2 μg/mouse, R&D, Minneapolis, MN) or 0.1% bovine serum albumin control and intradermally injected in the dorsal skin of mice (8- to 11-week-old) as previously described.²² The skins were harvested 4 days later for histological analyses.

All animal works were carried out at the research laboratory of National Health Research Institutes (NHRI) and conducted according to Taiwan COA national guidelines. All studies and procedures were performed with protocols approved by the NHRI Animal Care and Use Committee.

Histological analysis and immunostaining

Lower back skin samples were fixed with 4% paraformaldehyde for either 30 minutes on ice or 4 hours at room temperature, followed by frozen and paraffin embedding, respectively. All samples were sagittally sectioned at 6 μm. Hematoxylin and eosin staining and LacZ staining were performed using the standard procedures. To measure alkaline phosphatase activity in the DP, air-dried cryostat sections were prepared, fixed in acetone for 10 minutes, and incubated with NBT/BCIP substrate (Promega, Madison, WI) following the manufacturer's instruction.

Immunochemistry and immunofluorescence staining were performed as previously described.²⁰ Images were acquired with Olympus BX51 microscope equipped with Olympus DP71 CCD using DP controller and DP manager software or with a Leica TCS SP5 confocal microscope system with Leica Power 3D software. The sources and dilutions of primary antibodies were Hes1 (1:100, Santa Cruz or Cell Signaling, Danvers, MA), K6 (1:100, Thermo Fisher, Waltham, MA), AE15 (1:100, Santa Cruz), AE13 (1:100, Abcam, Cambridge, MA), K73 (1:150, Biorbyt, San Francisco, CA), K82 (1:100, Abnova, Taiwan), Ki67 (1:100, Thermo Fisher), CD34 (1:100, eBioscience, Waltham, MA), Sox9 (1:100, Santa Cruz), NFATc1 (1:150, Santa Cruz), β-catenin (1:100, BD, San Jose, CA), P-cadherin (1:250, R&D), p-Smad 1/5/8 (1:1000, Santa Cruz), phospho-histone H3 (1:100, Cell Signaling), Igfbp3 (1:100, R&D), Arl13b (1:200, Abcam), Pericentrin (1:500, Convance, Cambridge, MA), Smo (1:300, Abcam), K14 (1:250, Thermo Fisher), K15 (1:200, Thermo Fisher), Versican (1:50, Chemicon, Temecula, CA), and Vimentin (1:200, Abcam). Hes1 immunostaining was amplified by TSA Plus Cyanine 3 detection kit (PerkinElmer, Hopkinton, MA) following the manufacturer's protocol.

Mouse epidermal keratinocyte culture

Primary keratinocytes were isolated from the back skin of newborn mice as previously described.²³ For cilia staining, primary keratinocytes were starved 24 hours in E-media+0.1% chelexed-FBS for ciliated cell enrichment. Cells were treated with vehicle, 10 nM Shh-N (R&D), or 10 nM SAG for an additional 4 hours (immunostaining) or 16 hours (qRT-PCR).

FACS and flow cytometry

Isolation of HFSCs based on α6-integrin and CD34 were performed following the published protocol.²⁴ In brief, telogen dorsal skin with dermal adipose removed with scalpel was treated with dispase (5 U/mL, Invitrogen, Waltham, MA) in Hanks' balanced saline solution at 4°C overnight, and then transferred to Trypsin-ethylenediaminetetraacetic acid (0.25%, Invitrogen) at 37°C for 10 minutes. The resulting single cell suspension was filtered through a 70-μm cell strainer and incubated with CD49f-PE and biotinylated-CD34 antibodies followed by streptavidin-APC. Cell sorting was done on a FACS Influx cell sorter equipped with FACS Software (BD, Franklin Lakes, NJ). Keratinocytes with high forward and side scatter as well as dead cells (7-AAD+) were gated out, and the HFSCs (CD34+CD49f+) were collected. Flow cytometry were performed on a FACSCalibur analyzer (BD) and data were analyzed with the FlowJo program.

RNAscope in situ hybridization

RNAscope in situ hybridization was performed following the manufacturer's protocol (Advanced Cell Diagnostics, Newark, CA). The RNAscope probes used are Mm_Gli1 (311001), Mm_Ptch1 (402811), and Mm_Hes1 (417701). Each sample was quality controlled for RNA integrity using a positive control RNAscope probe Mm_Ppib (313911) and a negative control probe bacteria dapB (310043).

Data acquisition and statistics

For immunostaining, identical conditions of exposure and background balance for image capture were used for comparisons between control and mutant samples. Positively stained cells were counted manually in a defined area of the tissues. Image J software (NIH) was used to measure the length and pixel intensity in photos for quantification study. Statistical analyses were done using either a Student's t-test for comparing two samples or an analysis of variance followed by Tukey's multiple comparisons test for comparing multiple samples. P-value less than .05 was considered to be significant.

Results

Ablation of notch signaling effector Hes1 in the murine epidermis causes retarded hair growth

We explored the function of Hes1, a major Notch downstream target, in the epidermis and HFs using a conditional knockout study. We crossed the Hes1 fx/fx mice to K14-Cre mice, and the resulting [Hes1^fxfx;K14-Cre] mice (hereafter referred to as Hes1eKO mice) were born without any overt phenotype. We used the surrogate Rosa26-LacZ reporter mice to confirm K14-Cre-induced gene recombination in the entire postnatal epidermis (Figure 1A). Quantitative real-time PCR (qRT-PCR) of Hes1 and Hes5, two major Notch effectors in the epidermis, revealed that Hes1 gene expression is significantly decreased in the Hes1eKO epidermis whereas Hes5 is unaffected (Figure 1B).

Figure 1

Hes1 deletion using K14-Cre deleter line causes retarded hair growth. A, X-gal-stained back skin section of [K14-Cre^+/wt; Rosa-LacZ ^+/wt] mice with eosin counterstain. B, Quantitative real-time PCR (qRT-PCR) analysis of Hes1 and Hes5 expression levels in the back skin epithelium of Hes1eKO and control mice (mean ± SD, n = 4, ***P < .001). C, Illustration of the two synchronous hair cycles coordinated with age in mice after birth. D, Hematoxylin and eosin-stained back skin sections at P14 (follicular morphogenesis), P19 (first catagen), P22 (first telogen), P24 (second early anagen), P25 (second anagen), P29 (second anagen), P35 (second late-anagen/early catagen), P42 (second catagen), P56 (second telogen), and day 8 post-depilation at P56. E and F, Quantification of hair cycle stage according to morphology-based hair cycle histomorphometry²⁵ in each hair cycle phase as indicated (mean ± SD, n > 50 hair follicles (HFs) per genotype from two to three independent control and mutant pairs at each phase, n.s., nonsignificant; **P < .01; ***P < .001). G, Quantification of HF length (from the base of follicle to the follicular orifice) in each hair cycle phase as indicated (mean ± SD, n > 50 HFs per genotype from two to three independent control and mutant pairs at each phase, ***P < .001). Scale bar, 50 μm

We examined the gross phenotype of the back skin during the postnatal hair cycle and found that the anagen progression was significantly delayed in Hes1eKO mice (Figures 1C and S1A). In histological and quantitative analyses (Figure 1D, G), Hes1eKO HFs were shorter than control HFs in follicular morphogenesis (P14), but catagen induction was similar to control HFs at P19. The telogen to anagen transition was delayed in Hes1eKO HFs, since fewer HFs were in advanced anagen phase during P24 to P29 (Figure 1E).²⁵  Hes1eKO HFs were shorter than control HFs during the second anagen (P25-P35, Figure 1G). The anagen-catagen transition (P35-P42) as well as catagen-telogen transition (P42-P56) were comparable between Hes1eKO and control HFs. Plucking of telogen HFs stimulates anagen re-entry,²⁵ and the HFs of Hes1eKO mice were shorter than control mice 8 days post-depilation at P56 (Figure 1G). These data suggested that Hes1eKO HFs displayed retarded hair growth during homeostasis and depilation-induced hair regeneration.

Hes1-deficient HFs display delayed anagen initiation and shortened hair growth phase

Using in situ hybridization and immunostaining, we demonstrated that Hes1 is expressed in the bulge and enriched in the lower bulge/HG during anagen initiation. Although Hes1 expression was detected in the inner bulge and less frequent in the outer bulge layers in telogen, it was detected in both the inner and outer bulge layer during anagen initiation. Hes1 expression was absent in the HF epidermal compartment of Hes1eKO HFs, whereas that in the DP remained (Figure 2A, B). Next, we analyzed anagen initiation by immunostaining of P-cadherin (HG marker) and Ki67 (proliferative marker). We found that Hes1eKO HFs displayed decreased cell proliferation in the HG compared with control HFs at early anagen (P24), whereas no differences in the HG cell numbers were observed at telogen (P22, Figure 2C–2F). Generally, anagen activation is accompanied by nuclear translocation of β-catenin, a marker of active Wnt signaling, in the HG.²⁶ The β-catenin immunostaining revealed that Hes1eKO HFs displayed fewer nuclear β-catenin signals than control HFs (Figure 2G). Accordingly, control HFs displayed less phospho-Smad1/5/8 staining, a marker of inhibitory BMP signaling, than Hes1eKO HFs (Figure 2H). Furthermore, given the comparable immunostaining of HFSC markers CD34, Sox9, NFATc1, and K15 (Figure S1B–D) as well as the lack of TUNEL staining in control and Hes1eKO HFs (Figure S1E), we demonstrated that neither loss of HFSCs nor increased cell death in the HG accounted for the delayed anagen entry in Hes1eKO HFs.

Figure 2

Hes1 deletion results in delayed anagen initiation and shortened hair follicle (HF) growth phase. A, In situ hybridization of Hes1 (arrows) in back skin sections at P22 and P24 with hematoxylin counterstain. The dotted lines demarcate the bulge and the solid lines demarcate the boundary between DP and HG when visible. B, Back skin sections were immunostained for Hes1 at P22 and P24. C, Back skin sections were immunostained for HG marker P-Cadherin (P-Cad, arrows) at P22. D, Quantification of P-Cad + cells in the HF at P22 (mean ± SD, n > 30 HFs per genotype from three independent control and mutant pairs, n.s., nonsignificant). E, Back skin sections were double immunostained for P-Cad and proliferative marker Ki67 (arrows) at P24. F, Quantification of Ki67+ cells in the P-Cad + cells at P24 (mean ± SD, n > 30 HFs per genotype from three independent control and mutant pairs, ***P < .001). G, Back skin sections at P24 were immunostained for β-catenin. The arrows mark the nuclear β-catenin staining. The dotted lines denote the boundary between DP and HG when visible. H, Back skin sections immunostained for phospho-Smad1/5/8 (arrows). The dotted line marks the HF and solid line marks the DP. I, Double immunostaining of AE13 and KI67 in back skin sections at P29 (full anagen) and P35 (late anagen/early catagen). The dotted lines depict the line of Auber. J, Quantification of the bulb size (Ki67+ area below the line of Auber) (mean ± SD, >40 HFs from three biological replicates per genotype per stage, ***P < .001). K, Back skin sections immunostained for cell mitotic marker phospho-histone H3 (p-H3) at P29 and P35. L, Quantification of p-H3+ cells in the hair matrix (mean ± SD, >40 HFs from three biological replicates per genotype per stage, ***P < .001). DAPI counterstaining in blue. Bu, bulge; HG, hair germ; DP, derma papillae; Bb, hair bulb. M, Bright field images of club hair of four different hair types at P60. N, Quantification of club hair length of each HF type (mean ± SD, 20 club hairs for each hair types per mouse, n = 3 biological replicates per genotype, ***P < .001 determined by analysis of variance). Scale bar, 50 μm except (M), where it is 1 mm

To investigate the anagen progression defects in Hes1eKO HFs, we performed immunostaining for AE13 and Ki67 to quantify the hair bulb size (Figure 2I, J), as well as immunostaining for phospho-histone H3, a cell mitotic marker, to quantify the matrix proliferation (Figure 2K, L). Our data indicated that the hair bulbs of Hes1eKO HFs were smaller and less proliferative than that of control HFs at P29 (anagen). In late anagen-catagen transition (P35), the hair bulb size and matrix proliferation of Hes1eKO HFs did not exceed control HFs, suggesting that Hes1eKO HFs had never grown to the size as control did. We excluded increased cell death as the underlying cause for smaller hair bulbs in Hes1eKO HFs, as evidenced by TUNEL staining on samples harvested at the second hair cycle (P29-P56, Figure S2A). Smaller hair bulbs and less Mx proliferation of Notch1-deficient HFs have been attributed to paracrine Igfbp3 induced in the DP.¹¹ However, we found no discernible difference in levels of Igfbp3 protein between control and Hes1eKO DPs in both anagen and telogen phases (Figure S2B). The DP characteristics and inductive ability were examined by alkaline phosphatase activity and Versican protein expression, as well as counting the number of Versican+ cells in the DP (Figure S2C–E), and we found no difference between control and Hes1eKO HFs. Collectively, these data indicated that Hes1eKO HFs displayed delayed anagen initiation and shortened HF growth phase.

Hes1 deletion causes reduced hair shaft length but not identity changes in follicular lineages

Hes1 is expressed in the Mx, precortex, medulla, cortex, and cuticle of the hair shaft,¹⁴ which implicates its function in HF differentiation. To examine whether Hes1 deficiency causes any hair structure defect (Figure S2F), we analyzed the hair keratin markers K6, AE15, AE13, K82, and K73 at P29 (anagen) and P35 (late anagen). Although K6 staining revealed that both control and Hes1eKO HFs have comparable companion layers (Figure S2G), immunostaining of other markers revealed that Hes1eKO HFs lack the hair shaft medulla layer (AE15) and exhibited less developed hair shaft (AE13) and cuticle layers in both the IRS (K73) and hair shaft (K82) at P29. Remarkably, hair shaft AE15+ medulla layer and the AE13+, K82+, and K73+ cell layers of Hes1eKO HFs appeared to be comparable to control HFs at P35 (Figure S2H). These data indicated that Hes1 deletion caused delayed follicular lineage formation but not identity changes.

Mouse hair coat consists of four different HF types (Guard, Awl, Zigzag, Auchene) that emerge in three waves during development.²⁷ We found that Hes1eKO mice have all four HF types; however, the club hair length of four HF types is shorter in Hes1eKO mice than in control mice at P60 (Figure 2M, N). We conclude that the shortened anagen phase resulted from Hes1 deletion causes reduced hair shaft length.

Hes1 is required for HF regeneration in a sequential depilation model

To assess the function of Hes1 in regenerative hair cycle, we applied a repetitive depilation model to induce HFSC activation and monitoring the HF regeneration. Although control mice could mostly replenish the hair coat, Hes1eKO mice displayed a gradual thinning of hair coat after repetitive depilation (Figures 3A, B, and S3A, B). We observed gender difference in HF regeneration; male Hes1eKO mice displayed hair coat thinning early than female Hes1eKO mice. Immunostaining of CD34 and P-Cadherin revealed that HFSCs and HG cells were reduced in Hes1eKO mice after repetitive depilation (Figures 3C, D, and S3C, D). Using flow cytometry to quantify the HFSCs, we found a significant reduction in HFSC population in Hes1eKO mice after repetitive depilation (Figure 3E). We applied CD34 immunostaining and EdU incorporation assays to examine the HFSC activation after repetitive depilation. Although HFSC activation in control and Hes1eKO HFs were initially similar (day 2 after first depilation), HFSC activation in Hes1eKO HFs were compromised after five rounds of depilation (day 2 after fifth depilation) (Figure 3F, G). We additionally found that the hair coat of unperturbed Hes1eKO mice was thinner than control mice at about 1 year old (Figure S3E). These data indicate that Hes1eKO HFSCs cannot sustain HF regeneration after repeated hair-growth cycles.

Figure 3

Hes1 deficiency causes compromised hair follicle (HF) regeneration and hair follicle stem cell (HFSC) self-renewal after repetitive depilation. A, Sequential depilation of control littermate (Ctrl) and Hes1 conditional knockout (Hes1eKO) mice for four rounds with a three-week interval from the second telogen. Representative pictures of male mice are shown (n = 5). B, Close up of back skin at day 22 post-depilation-induced hair regeneration. C, Back skin sections from repetitive depilation (day 22 after fourth depilation) were double immunostained for CD34 and P-Cad. D, Quantification of CD34+ bulge and P-Cad + HG cells in HFs after four rounds of sequential depilation (mean ± SD, n > 50 HFs per genotype from four independent control and mutant pairs, ***P < .001). E, Quantifications of HFSCs after four rounds of sequential depilation using flow cytometry. Data are presented as percentage of Bu/CD49f + cells relative to control samples (mean ± SD, n = 4 independent pairs). F, Skin sections from sequential depilation (one and five times) were processed for CD34 immunostaining and EdU incorporation assays. Dotted lines denote the DP when visible. G, Quantification of cell proliferation within CD34+ cells and below CD34+ cells after sequential depilation (mean ± SD, n > 50 HFs per genotype from three independent control and mutant pairs for each set, ***P < .001). Bu, Bulge; HG, hair germ; SG, sebaceous glands. DAPI counterstaining in blue. Scale bar, 50 μm

Loss of Hes1 leads to compromised Shh signaling in HFSCs

To understand the molecular basis underlying the HF phenotype in Hes1eKO mice, we performed microarray gene expression profiling on FACS-purified HFSCs from control and Hes1eKO mice at P72 (telogen after depilation at P50). We identified 77 upregulated genes and 88 downregulated genes with a fold change >1.5 or <−1.5 (P < .05) in Hes1eKO vs control HFSCs (Figure S4A–C). Ingenuity pathway analysis revealed “lipid metabolism,” “cellular growth and proliferation,” and “cellular movement” among the top diseases and biological functions affected by Hes1 deletion; Acyl-CoA hydrolysis and stearate biosynthesis are among the top canonical pathways affected by Hes1 deletion (Figures 4A and S4D). Next, we performed gene set enrichment analysis on the microarray results. We found that the gene sets enriched in TGF-β superfamily signaling (BMP signaling) and apical cell adhesion are specifically upregulated in Hes1eKO HFSCs. Remarkably, gene sets enriched in Smoothened signaling regulation, mitochondrial oxidative phosphorylation, and fatty acid metabolism are specifically downregulated in Hes1eKO HFSCs (Figure 4B). qRT-PCR analyses on selected genes related to Shh signaling, top diseases and biological functions, and top canonical pathways confirmed the microarray results (Figures 4C, D, and S4E).

Figure 4

Hes1 inactivation causes compromised Shh signaling in hair follicle stem cells (HFSCs). A, Ingenuity pathway analysis result showing the top diseases and bio functions and top canonical pathways significantly affected by Hes1 deletion in HFSCs (cutoff fold change >1.5 or <−1.5, P < .05). Red and green indicate upregulated and downregulated genes in Hes1eKO HFSCs, respectively. B, Gene set enrichment analysis of microarray gene expression profiling on HFSCs from control and Hes1eKO mice (n = 2 independent pairs). C and D, qRT-PCR analysis of selected genes related to Shh signaling, top diseases and biological functions, and top canonical pathways on FACS-purified HFSCs from control and Hes1eKO mice at P72 (telogen after depilation at P50, mean ± SD, n = 4 independent pairs, *P < .05, **P < .01, ***P < .001). E, In situ hybridization of Gli1 and Ptch1 (arrows) in control and Hes1eKO back skin sections at P72 (telogen after depilation at P50). Scale bar, 50 μm

Indeed, the retarded anagen progression and hair regeneration failure observed in Hes1eKO mice closely resemble phenotypes of conditional Hedgehog component knockout mice.³ We therefore examined Shh signaling activity in control and Hes1eKO HFs at telogen, a stage when control and Hes1eKO HFs can be compared. In situ hybridization of the Shh target genes Gli1 and Ptch1 revealed decreased Shh signaling activity in Hes1eKO HFs at P72 (telogen after depilation at P50) (Figure 4E). Our data indicate a specific function for Hes1 in hair cycle control through modulation of Shh signaling.

Hes1 deletion causes primary cilia defect and influences hedgehog signaling responsiveness

Shh signaling is sensitive to the length, numbers, and architecture of primary cilia²⁸ and ciliary transport of Smo is involved in Hedgehog signaling activation. We therefore examined the cilia length in the lower HFs of control and Hes1eKO mice by double immunostaining of Arl13b (a small GTPase localized to cilia) and Pericentrin (a centrosome protein localized to cilia base). We found that the cilia in the lower Bu/HG of Hes1eKO HFs were shorter than that of control HFs at P72 (Figure 5A, B). Because the ciliary accumulation of endogenous Smo was difficult to detect in tissue sections by antibody staining, we used primary mouse epidermal keratinocyte (PMEK) cultures from control and Hes1eKO dorsal skin as an alternative system (Figure S5). We observed a decrease in both the percentage of ciliated cells and the ciliary length in Hes1eKO PMEKs when cultured in serum-starved conditions to enrich ciliated cells, as revealed by double immunostaining of Arl13b and Pericentrin (Figure 5C–E). Accordingly, qRT-PCR analysis revealed that Hes1eKO PMEKs had lower fold induction of Gli1 and Ptch1 mRNA than control PMEKs in response to Shh (Figure 5F). Additionally, we found increased gene expression of acyl-CoA thioesterase Them5 and elevated NAD/NADH ratio in Hes1eKO PMEKs, suggesting a correlation with altered cellular metabolism (Figure 5G, H).

Figure 5

Hes1 deletion causes primary cilia defect and influences Shh signaling responsiveness. A, Double immunostaining of Arl13b and Pericentrin in back skin sections of mice at P72 (telogen after depilation at P50). The dotted line box denotes the region of cilia measurement. The boxed region shows high magnification. DAPI counterstaining in blue. B, Quantification of cilia length in lower hair follicles (HFs) (dotted line box in (A)) of control and Hes1eKO mice (mean ± SD, n = 3 independent pairs; control 336 cilia; Hes1eKO 384 cilia, ***P < .01). C, Control and Hes1eKO PMEKs were immunostained for Arl13b and Pericentrin. The boxed region shows high magnification. D, Quantification of ciliated cells expressed as a percentage of total in control and Hes1eKO PMEKs (mean ± SD, n = 3 independent samples per genotype; >100 cells/microscopic field, >6 field/sample, ***P < .001). E, Quantification of cilia length in control and Hes1eKO PMEKs (mean ± SD, n = 3 independent samples per genotype; control 198 cilia; Hes1eKO 110 cilia, *P < .05). F, qRT-PCR analysis of Gli1, Ptch1, and Smo expression levels in Shh-treated relative to vehicle-treated PMEKs (mean ± SD, n = 5 independent samples per genotype, *P < .05). G, qRT-PCR analysis of Them5 on control and Hes1eKO PMEKs (mean ± SD, n = 4 independent samples per genotype, *P < .05). H, Measurement of NAD⁺/NADH ratio in control and Hes1eKO PMEKs (mean ± SD, n = 3 independent samples per genotype; *P < .05). I, Control and Hes1eKO PMEKs were serum starved and treated with vehicle, Shh (10 nM), or SAG (10 nM) and double immunostained for Arl13b and Smo. The boxed region shows high magnification. Low- and high- magnification scale bars represent 10 and 1 μm. J, Quantification of the percentage of Smo + primary cilia in control and Hes1eKO PMEKs treated with vehicle, Shh, or SAG (mean ± SD, n = 4 independent samples per genotype per condition; control >90 cilia and Hes1eKO > 110 cilia in each treatment, **P < 0.01, ***P < 0.001, n.s., nonsignificant, determined by analysis of variance). K, Normalized luciferase activity in control and Hes1eKO PMEKs transfected with reporter plasmids and treated with vehicle, SAG, or Shh (mean ± SD, n = 4 independent samples per genotype per condition, *P < .05, n.s., nonsignificant)

Next, we examined Smo ciliary accumulation in the absence or presence of Hedgehog activators by double immunostaining of Arl13b and Smo (Figure 5I). Interestingly, we observed fewer Smo + primary cilia in Hes1eKO PMEKs than control PMEKs during serum starvation. The ciliary localization of Smo was increased in control PMEKs upon Shh treatment, whereas that in Hes1eKO PMEKs remained unchanged. In contrast, ciliary localization of Smo in Hes1eKO PMEKs was increased upon SAG treatment, suggesting a regulatory mechanism upstream of Smo activation (Figure 5J). Accordingly, we found that Hes1eKO PMEKs displayed compromised Gli binding site-luciferase activity in response to Shh but not to SAG (Figure 5K). These findings indicate that Hes1 modulates Shh signaling through regulation of ciliogenesis and Smo ciliary accumulation.

Direct activation of smoothened can rescue anagen initiation and HF regeneration in Hes1eKO mice

Small molecule agonist SAG binds Smo directly and bypasses Patched receptors to activate Shh signaling. Topical application of SAG has been demonstrated to stimulate the hair regrowth in adult mouse skin.^{29,, 30} To ascertain whether direct activation of Smo can rescue HF phenotypes in Hes1eKO mice, we performed transient application of vehicle and SAG at opposite sides of the back skin during repetitive depilation (Figures 6A and S6A). Although vehicle-treated Hes1eKO HFs displayed anagen delay after sequential depilation, two rounds of depilation/SAG treatment rescued anagen initiation in Hes1eKO HFs (Figure 6B, C). In situ hybridization of Gli1 and Ptch1 demonstrated that Shh signaling activity in Hes1eKO HFs was rescued by SAG treatment (Figure 6D). After three rounds of depilation/SAG treatment, we found that both the CD34+ bulge cells and P-Cad + HG cells were increased in Hes1eKO HFs (Figure 6E, F). Additionally, the club hair length of each HF types in Hes1eKO mice was increased by three rounds of SAG treatment (Figure S6B, S6C). To demonstrate that the Shh signaling is compromised but still functional in Hes1eKO HFs, we analyzed the effect of exogenous Shh administration on the back skin of control and Hes1eKO mice. Shh and BSA-coated beads were intradermally injected in the dorsal skin of control and Hes1eKO mice. The skin sections were immunostained for P-Cad and Ki67 as well as assayed for Gli1 mRNA expression (Figure 6G–I). We observed that exogenous Shh administration can stimulate cell proliferation and Gli1 mRNA induction in the HG of both control and Hes1eKO HFs, indicating that Shh signaling is functional in both control and Hes1eKO HFs. Our results indicate that direct stimulation of Smo activity can rescue the anagen initiation and HF regeneration in Hes1eKO HFs.

Figure 6

Transient application of Smoothened agonist (SAG) can rescue anagen initiation and hair follicle (HF) regeneration in Hes1eKO mice. A, Schematics of the in vivo SAG rescue experiments. B, Histological (hematoxylin and eosin-stained) and immunostaining (Ki67/P-Cad) analyses on back skin samples harvested at early anagen (day 2 after third depilation) after two rounds of depilation/SAG treatment. C, Quantification of Ki67+ cells in HG cells after SAG rescue experiments (mean ± SD, n > 50 HFs per experimental condition from three independent pairs, ***P < .001). D, In situ hybridization of Gli1 and Ptch1 in back skin sections from mice after two rounds of depilation/SAG treatment. E, Double immunostaining of CD34 and P-Cad on back skin samples harvested at telogen (day 24 after fourth depilation) after three rounds of depilation/SAG treatment. F, Quantification of CD34+ bulge and P-Cad + HG cells in HFs after SAG rescue experiments (mean ± SD, n > 40 HFs per experimental condition from three independent pairs, ***P < .001). G, Immunostaining of P-Cad and Ki67 in skin sections of control and Hes1eKO mice intradermally injected with BSA or Shh-N coated beads. Right panels, magnified views of boxed areas. Asterisks, the injected protein-coated beads. H, Quantification of Ki67+ cells in the P-Cad + cells from intradermal injection experiments (mean ± SD, n > 15 HFs per bead injection from three independent control and mutant pairs, *P < .05, ***P < .01). I, In situ hybridization of Gli1 in back skin sections from bead injection experiments. Bu, bulge; HG, hair germ; DP, derma papillae. DAPI counterstaining in blue. Scale bar, 50 μm

Discussion

The hair cycle represents a paradigm for studying stem cell quiescence and activation, as well as progenitor cell proliferation, differentiation, and death. Here, we show that Hes1 expression is enriched in the lower bulge/HG at anagen onset. The retarded hair growth observed in Hes1-deficient HFs is resulted from a delay in anagen initiation and shortened anagen phase. Moreover, Hes1 epithelial ablation results in impaired HF regeneration after repetitive depilation. Transcriptome analysis and gene expression data indicate that Hes1 ablation compromises Shh responsiveness. Hes1 possibly influences Hedgehog signaling through regulating ciliogenesis and Smo ciliary accumulation. Therefore, direct activation of Smo can rescue anagen initiation and HFSC self-renewal in Hes1-deficient HFs. Our data suggest that Hes1 reinforces the Shh signaling during telogen-anagen transition to maintain hair cycle homeostasis (Figure 7A).

Figure 7

Working model of telogen-anagen transition and hair follicle (HF) regeneration regulated by Hes1. A, By modulating ciliary function, Hes1 potentiates Shh signaling in anagen initiation, which allows sufficient signaling strength to expand the hair germ and replenish hair follicle stem cells to maintain the hair cycle. B, Illustration of the location of Hes1 and Gli1 expression in telogen HF

A role for Notch signaling in postnatal HF development and cycling was delineated by epithelial knockout of Notch components. Smaller hair bulbs were reported at the postnatal HF morphogenesis, and premature entry into catagen was postulated to be the underlying cause.¹⁰ Similar phenotype was reported by Lee et al, in which smaller hair bulb of Notch1-deficient HFs was attributed to lower mitotic rates mediated by paracrine inhibition of IGF signaling in the Mx through DP-derived IGFBP3.¹¹ However, Hes1 expression was unaltered in Notch1-deficient HFs, nor did we observe any difference in Igfbp3 expression as well as characteristics and inductive ability between control and Hes1eKO DPs. The delayed anagen entry observed in Hes1-deficient HFs suggests a cell-autonomous role for Hes1 in stem cell/progenitor activation during anagen induction.

Notch ligands and receptors are expressed in the skin in a complex and dynamic manner.³¹ Notch downstream effectors are expressed in the hair bulb precortex and hair shaft precursors when the Mx commits terminal differentiation, suggesting a role for Notch signaling in hair shaft differentiation.¹⁴ Interestingly, we found that Hes1-deficient HFs displayed a delayed occurrence of hair shaft components without changes in hair follicular lineages, suggesting that Hes1 modulates the response of HF stem/progenitor cells to hair growth promoting signals rather than directly regulates lineage commitment.

The delayed anagen initiation could be resulted from increased expression of the cell cycle inhibitors in the bulge, since p21^Cip1, p27^Kip1, and p57^Kip2 have been identified as Hes1 downstream targets in other organs.^32-34 However, our microarray analysis showed that these cell cycle inhibitors are not affected by Hes1 deletion, but instead Hedgehog signaling is compromised. Notch signaling has been shown to shape the response of neuroepithelial cells to Shh and influences cell fate choice in spinal cord development. Notch activities seem to promote longer primary cilia and ciliary Smo accumulation by an unknown transcriptional mechanism.^{35,, 36} We found that Hes1 deletion causes shorter cilia and abolishes further Smo accumulation in the cilia upon Shh treatment, suggesting that Hes1 does not change the competence but rather the strength of Shh responsiveness during hair growth. Interestingly, Shh emanating from TACs during early anagen has been demonstrated to sustain HF growth and HFSC self-renewal.³ Therefore, our Hes1 loss-of-function studies in HFs suggest that Hes1 regulates anagen initiation and HF regeneration via modulation of Shh responsiveness. Transcriptome profiling revealed that lipid metabolism is specifically affected in Hes1eKO HFSCs. Given that lipid metabolism is closely associated with both Hedgehog signal transduction and Hedgehog ligand modification, the compromised Shh responsiveness caused by Hes1 deficiency is likely due to altered lipid metabolism that influences ciliogenesis and Smo ciliary accumulation.^{37,, 38}

In telogen HFs, Gli1 is expressed in two restricted HF epithelial compartments and in the DP. One population of Gli1+ cells, localized to the upper margin of the bulge, respond to cutaneous nerve-releasing Shh, and contribute to wound-induced epidermal regeneration. Another population of Gli1+ cells, localized to the lower portion of bulge/HG, respond to DP/HG-releasing Shh and contribute to immediate HF growth in anagen.³⁹  Hes1 expression in the lower bulge/HG during anagen initiation suggests a crosstalk between Notch and Hedgehog signaling pathways in this compartment (Figure 7B). Whether Notch signaling promotes or inhibits Hedgehog signaling or vice versa is context-dependent. Notch receptors and regulated proteolysis enzyme were found to colocalize with cilia. Elimination of primary cilia caused defects in the differentiation of embryonic epidermis, which was attributed to Notch signaling loss.⁴⁰ Normally, Notch receptor is activated by membrane-bound ligand through cell-cell interaction but not by soluble forms of ligands, so ciliogenesis is less likely to play a direct role in Notch signaling activation. There are evidences that Shh-driven stabilization of Hes1 is independent of canonical Notch signaling and Hes1 is a Hedgehog-dependent direct target of Gli2.^{19, 41, 42} In contrast, canonical Notch1/Rbpj axis has been shown to regulate Hedgehog signaling effectors Gli2/Gli3,⁴³ as well as Hes1 is shown to bind the Gli1 first intron that may inhibit its expression.⁴⁴ Therefore, we think that the crosstalk between Notch and Hedgehog pathways could be different during development, homeostasis, and carcinogenesis.

The two-step mechanism of SC activation during HF regeneration derives from the observation that HG is in close proximity to the DP and the bulge is separated from the DP by the HG.^{45,, 46} The DP activates the proliferation of primed SCs in the HG to form the TACs and sustain HF regeneration.⁴⁷ Moreover, the HG is thought to buffer the bulge from the DP to receive excess proliferating signals that will exhaust the conserved SCs. Although the two-step mode of SC activation seems to prevail as the underlying mechanism of HF regeneration, there are examples that anagen initiation and HF regeneration can occur when Shh signaling is activated in the epithelial part of the HF during telogen,^{29,, 30} suggesting that ectopic activation of Shh signaling in the bulge can substitute the signal required from the DP to activate the HG. In clinical hair medicine, whether a HF is in refractory or competent telogen⁴⁸ will greatly influence the efficacy of hair growth-promoting agents. Therefore, perhaps if we can learn more about the alternative modes of HF regeneration then the poorly effective agents can be administered more effectively. Interestingly, Jagged1-expressing regulatory T cells in the skin are shown to help HFSC activation and anagen induction,⁴⁹ which corroborates our study and suggest that manipulating Notch signaling can be used as a therapeutic strategy to gain control of the telogen stage.

Conclusion

Hedgehog signaling is one of the important pathways that governs epidermal and HF development. A Hedgehog signaling gradient established by the Patched receptors is found along the proximodistal axis of developing HFs,⁵⁰ suggesting that fine-tuning the intensity of hedgehog signaling is necessary to maintain hair cycle homeostasis. Here, we identified a critical role for Hes1 in hair cycle homeostasis. By modulating Hedgehog signaling responsiveness, the Notch-Hes1 axis facilitates signaling activity in the Shh-receiving HFSCs/HG, which is required for anagen initiation and HFSC maintenance.

Acknowledgments

We thank Dr. Ryoichiro Kageyama (Kyoto University, Japan) kindly provide Hes1 floxed allele mice. We thank Hsien-Yi Lin and Pei-Hsuan Lin for technical service. We are grateful to NHRI Core Facilities for their technical support. This work was supported by funding from NHRI intramural grants (06A1-CSPP12-014, 07A1-CSPP12-014, 08A1-CSPP12-014), central government S&T grants, Taiwan (106-0324-01-10-07, 107-0324-01-19-03, 108-0324-01-19-07), and MOST grants (103-2320-B-400-011-MY3, 106-2320-B-400-022-MY3) to L-T.Y.

Conflict Of Interest

The authors indicated no potential conflicts of interest.

Author Contributions

W.-J.S., S.-T.L.: collection and/or assembly of data, data analysis and interpretation; L.-T.Y.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

Data Availability Statement

The data that support the findings of this study are available within the article and its supplementary materials.

Author notes

Wei-Jeng Suen and Shao-Ting Li contributed equally to this work.

Funding information MOST, Taiwan, Grant/Award Numbers: 103-2320-B-400-011-MY3, 106-2320-B-400-022-MY3; Central government S&T grants, Taiwan, Grant/Award Numbers: 106-0324-01-10-07, 107-0324-01-19-03, 108-0324-01-19-07; National Health Research Institutes, Grant/Award Numbers: 06A1-CSPP12-014, 07A1-CSPP12-014, 08A1-CSPP12-014

[Correction added on 30 May 2020, after first online publication: “People's Republic of China” changed to “Taiwan, R.O.C.” in author affiliations and correspondence section]