Research Article

Hair follicle aging is driven by transepidermal elimination of stem cells via COL17A1 proteolysis

See allHide authors and affiliations

Science  05 Feb 2016:
Vol. 351, Issue 6273, aad4395
DOI: 10.1126/science.aad4395

Quiescent and aging hair follicle stem cells

Stem cells enable normal cell homeostasis, but they also exist in a quiescent state, ready to proliferate and differentiate after tissue damage. Now, two studies reveal features of stem cells in the hair follicle, an epithelial mini-organ of the skin that is responsible for hair growth and recycling (see the Perspective by Chuong and Lei). Wang et al. found that the Foxc1 transcription factor is induced in activated hair follicle stem cells, which in turn promote Nfatc1 and BMP signaling, to reinforce quiescence. Matsumura et al. analyzed hair follicle stem cells during aging. They identified type XVII collagen (COL17A1) as key to hair thinning. DNA damage-induced depletion of COL17A1 triggered cell differentiation resulting in the shedding of epidermal keratinocytes from the skin surface. These changes then caused hair follicle shrinkage and hair loss.

Science, this issue p. 559, p. 613; see also p. 10.1126/science.aad4395

Structured Abstract


During aging, most organs in mammals become smaller (miniaturize) or thinner, and their functions and regenerative capability also decline. Histologically, tissue atrophy and fibrosis are observed in many aged organs. Yet the exact mechanisms for the architectural and functional decline are unknown. Indeed, areas that are as yet underexplored include the dynamics of the constituent cells and their cellular fate, as well as determination of whether aged or damaged cells accumulate or are eliminated in tissues and organs during the aging process. Organismal aging has been explained by various theories—such as reactive oxygen species, cellular senescence, telomere erosion, and altered metabolism—but not from the viewpoint of cellular and tissue dynamics. Stem cell systems sustain cellular and tissue turnover in most mammalian organs, but it has been difficult to experimentally test the precise fate of somatic stem cells, the cellular pool for tissues and organs. This has limited our understanding of the mechanisms of aging of tissues and organs and the existence of an aging program in mammalian organs. The hair follicle (HF) is an epithelial mini-organ of the skin that sustains cyclic hair regrowth over repeated hair cycles. Hair thinning (senescent baldness) is one of the most typical signs of aging in many long-lived mammals and is often prematurely induced by genomic instability, as in progeroid syndromes. We studied the mechanism for aging of the epithelial mini-organ.


Miniaturization of HFs has long been believed to be a specific key phenomenon for male-pattern baldness (androgenic alopecia) but not for HF aging. Our study revealed that mammalian HFs do miniaturize and often disappear from the skin during aging both in mice and humans, regardless of sex. We employed in vivo stem cell fate tracing in mice during physiological aging and searched for possible links between the cell fate of aged HF stem cells (HFSCs) and the stepwise miniaturization and loss of HFs. Combining gene expression profiling of young versus aged HFSCs and conditional knockout or maintenance of gene expression in HFSCs in mice, we defined the early events and molecules that connect HF cycling, HFSC aging, and the dynamic HF aging processes, which are characterized by the stepwise miniaturization of HFs.


The fate analysis of HFSCs during aging revealed that organ aging is primed by the sustained DNA damage response against DNA damage that accumulates in renewing stem cells during aging. This now tightly links intrinsic genomic instability in stem cells to epithelial organ aging. Further, we found that stem cell aging results from proteolysis of type XVII Collagen (COL17A1/BP180) by neutrophil elastase in response to DNA damage in HFSCs and the commitment of stem cells to epidermal differentiation. Terminal differentiation of HFSCs into epidermal keratinocytes drives HF miniaturization and enables the elimination of damaged stem cells as shed corneocytes from the skin surface. The fate of aged HFSCs abrogate their commitment to follicular differentiation to grow hair. Finally, HF aging can be recapitulated by Col17a1 deficiency and can be prevented by the forced maintenance of COL17A1 in HFSCs. This demonstrates that COL17A1 in HFSCs orchestrates the stem cell–centric aging program of the epithelial mini-organ.


In vivo stem cell fate tracing of HFSCs revealed the critical role of HFSCs in the induction of aging-associated hair thinning. We identified a distinct organ aging program that is driven by transepidermal elimination of aged HFSCs through their depletion of COL17A1 via DNA damage–induced protease expression and terminal epidermal differentiation. The dynamic HF aging program is a good model of organ and tissue shrinkage and functional decline commonly seen in many different organs during aging. This paradigm could potentially open new avenues for the development of anti-aging strategies to prevent and treat aging-associated diseases.

The mechanism of HF aging and associated hair loss.

HFs sustain their cyclic regeneration through the intensive self-renewal of activated HFSCs (blue dots). The aging of HFSCs is triggered by DNA damage-induced COL17A1 proteolysis. Once aged HFSCs (red dots) are activated during the hair cycle, they leave the niche and terminally differentiate into epidermal keratinocytes and are then eliminated from the skin surface.


Hair thinning and loss are prominent aging phenotypes but have an unknown mechanism. We show that hair follicle stem cell (HFSC) aging causes the stepwise miniaturization of hair follicles and eventual hair loss in wild-type mice and in humans. In vivo fate analysis of HFSCs revealed that the DNA damage response in HFSCs causes proteolysis of type XVII collagen (COL17A1/BP180), a critical molecule for HFSC maintenance, to trigger HFSC aging, characterized by the loss of stemness signatures and by epidermal commitment. Aged HFSCs are cyclically eliminated from the skin through terminal epidermal differentiation, thereby causing hair follicle miniaturization. The aging process can be recapitulated by Col17a1 deficiency and prevented by the forced maintenance of COL17A1 in HFSCs, demonstrating that COL17A1 in HFSCs orchestrates the stem cell–centric aging program of the epithelial mini-organ.

Tissues and organs undergo structural and functional declines due to aging (1). Many different hypotheses have been proposed to explain tissue aging as well as organism aging (27). Genes involved in aging phenotypes and/or organism longevity have been reported (811), yet the exact mechanism(s) underlying tissue aging is poorly understood.

Accumulation of DNA damage has been implicated in tissue aging (12, 13). Replication errors, reactive oxygen species, eroded telomeres, and chromosome breaks represent sources of endogenous DNA damage. Typical aging phenotypes, including hair loss and graying, are accelerated by intrinsic genomic instability in individuals with premature aging syndromes (progeroid syndromes) and their mouse models (1416), as well as by extrinsic genomic instability such as by ionizing radiation (IR) (17, 18). Tissue decline due to genomic instability has been explained by cellular senescence or apoptosis (14, 19, 20). However, typical senescent cells are not readily induced in tissues simply with genomic instability (18), but appear in premalignant lesions such as carcinogen-induced papilloma and melanocytic nevi in the skin (21, 22). Aging-associated changes in stem cells (stem cell aging) are currently recognized as one of the hallmarks of aging (7); however, the fates of those aged stem cells in aging tissues, the impact of DNA damage in their fate, and the roles of stem cell aging in the tissue and organ aging process are still largely unknown.

Hair follicle stem cells (HFSCs), which reside in the bulge/sub-bulge area of the hair follicle (HF), sustain cyclic hair regrowth over repeated hair cycles (2326) (fig. S1A). It has been reported that mouse HFSCs generally do not display apparent decline (27), but with aging, hair cycle waves slow down and display imbalanced cytokine signaling in HFSCs, as well as diminished colony-forming capability in vitro (2830). On the other hand, mammals that live longer lose their hair and HFs with age (31, 32). In this article, we describe the stepwise dynamics of mini-organ aging of HFs and provide mechanistic links from DNA damage and stem cell aging to eventual hair loss phenotype, with evidence for the existence of a tissue and organ aging program.

Aging-associated thinning hair and hair loss

The HF is a mini-organ of the skin that is specialized to grow hair (fig. S1A). To understand aging-associated hair loss in genetically defined mammals, we studied HF aging in wild-type mice. Aged C57BL/6 mice show diffuse and symmetric patterns of hair thinning on the back (Fig. 1A) that typically become apparent on the dorsal-most area of the neck and the trunk, with formation of a patchy or linear hair loss pattern and distal spreading toward the flanks (Fig. 1B). Hair thinning in C57BL/6N mice is sometimes apparent as early as 16 months after birth, but usually appears at around 18 months of age and becomes conspicuous by 24 to 28 months. Those mice show a more advanced phenotype after hair depilation (18 to 31 months; n = 6 mice) (fig. S1B), indicating that cyclic hair regeneration accelerates HF aging. Notably, the areas with greatest hair loss never grow hair even with hair cycle inducers (fig. S1C), indicating that they have lost hair cycle waves. Histological analysis of the skin revealed a significant decrease of HF numbers later than 12 months of age (Fig. 1, C and D) and the miniaturization of HFs with reduction of dermal thickness (Fig. 1, C to G, and fig. S1D). We classified the chronological HF changes typically seen in affected skin areas into five different stages, judging from the level of HF miniaturization (proportional shortening of the permanent portion) and the presence or absence of dermal papillae, sebaceous glands, and the infundibulum (Fig. 1, E and F, and fig. S1E). Cyst formation was also found occasionally in the bulge/sub-bulge area of HFs, often with the loss of dermal papillae (fig. S1F). More advanced stages of HFs were found in older mice (Fig. 1G) more frequently on the dorsal-most areas with advanced hair thinning (Fig. 1, A to C, and fig. S1C). Thus, we concluded that HF aging proceeds progressively in a stepwise manner with distal advancement once homeostatic maintenance starts failing.

Fig. 1 Aging-associated hair thinning and loss is preceded by loss of the HFSC signature and HF miniaturization.

(A to G) Aging-associated hair loss and HF miniaturization in C57BL/6 wild-type mice. (A) Representative images of young (8 weeks) and aged (28 months) mice. Dashed lines indicate HL areas and surrounding H areas. (B) The hair loss is typically found on the dorsal-most area of the neck and the trunk of aged mice (>18 months). The HL areas extend distally in a diffused manner. (C) Representative histology of the dorsal skin from young (8 weeks), middle-aged (12 months) and aged (24 and 34 months) wild-type mice by hematoxylin and eosin (H&E) stain. The miniaturized HFs are indicated by arrows. (D) HF number per mm. (E) Simplified schematic of the HF structure at telogen phase. HS, hair shaft; IF, infundibulum; SG, sebaceous gland; JZ, junctional zone; Bg, bulge; SBg/HG, sub-bulge/hair germ; DP, dermal papilla. (F) Staging of chronological HF aging. Representative histological images at different HF stages are shown. Stage I, HF with standard structure and size; Stage II, shortened (miniaturized) HF with proportional size and structure, including DP and Bg; Stage III, miniaturized HF with complete loss of the DP; Stage IV, miniaturized HF retaining sebaceous glands (SG) with loss of HF bulge (Bg) and hair shaft; Stage V, complete disappearance of HFs, occasionally with mild fibrosis (FB). (G) Number of each HF stage per mm; >100 HFs from multiple samples were examined for each group. (H to K) Reduction of HFSC markers with chronological aging in mice. (H) Immunofluorescence (IF) images with the HFSC marker CD34 at 8 weeks and at 24 months of HL and H areas. (I) IF images of Stage I HFs at 8 weeks, 16 months, and 24 to 26 months for COL17A1, LHX2, and SOX9. (J) Quantitative analysis of fluorescence intensity of COL17A1, K15, and S100A6 in bulge areas of 8-week-old, 16- to 18-month-old, and 24- to 25-month-old mice; several HFs (>5) at Stage I and II from each group were examined. N = 3 mice for each group. (K) Quantitative analysis of the number of LHX2+ and SOX9+ nuclei in bulge cells of 8-week-old, 16-month-old, and 24-month-old mice. (L) FACS analyses of the CD34highITGA6highSCA-I fraction to detect HFSCs in young (8 weeks) and in aged (28 months) mouse skin. Although 28-month-old hairy skin maintains the HFSC fraction similarly to 8-week-old mouse skin, 28-month-old HL areas show a significant reduction of the CD34highITGA6highSCA-I- fraction. Error bars, mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test).

Stem cell dysregulation precedes HF miniaturization

To gain insight into the homeostatic machinery regulating HF aging, we hypothesized that changes in HFSCs, which generate all kinds of HF keratinocytes (23, 24, 33), may underlie the aging-associated structural and functional decline of the mini-organ. We examined the expression of HFSC markers, including CD34, S100A6 (34, 35), and keratin 15 (Krt1-15/K15) (33), using immunohistochemistry. Whereas bulge cells in young HFs expressed those markers, they were reduced in number in aged HFs (Fig. 1H and fig. S2A). This was prominently seen in areas of hair thinning and loss (HL), where HFs showed variable levels of miniaturization with reduced numbers of HFs, but were more sparse in the surrounding less-affected hairy (H) areas (Fig. 1, C and H, and fig. S2, A to F). Next, we tested the expression of key genes involved in HFSC maintenance, including LHX2, SOX9, and COL17A1 (3639). Analysis of the signal intensity and expression-positive cell number showed that those molecules are often decreased or even undetectable in stage I and II HFs of aged mice (24 to 26 months) (Fig. 1, I to K, and fig. S2G), indicating that the down-regulation of HFSC signature molecules precedes HF miniaturization and hair loss during physiological aging. Interestingly, the distribution of the reduced level of COL17A1 was restricted to the basement membrane zone or was undetectable in those HFs (Fig. 1I). Further histological analysis of the HF aging process indicated that individual HF components are well maintained in early stage HFs, even with some reduction of HFSC marker expression, but are diminished in a stepwise, regular, and proportional manner once HFs initiate the miniaturization process with the reduction of mesenchymal cells, including dermal papilla cells (fig. S2, B to F). Furthermore, fluorescence-activated cell sorting (FACS) analysis of the HFSC fraction (CD34highITGA6high) (40) showed that the cellular ratio is apparently diminished in HL areas but not in H areas, where reduction of CD34-expressing cells was already detectable in tissue sections (Fig. 1L and fig. S2A, H, and I). Taken together, these data demonstrate that aging-associated hair loss and the underlying HF miniaturization and loss are preceded by the dysregulation in HFSCs characterized by their loss of HFSC signatures during physiological aging.

Epidermal differentiation of HFSCs drives HF miniaturization

To characterize the mechanism(s) of the aging-associated HFSC dysregulation and depletion, we examined the expression of different cell fate markers, including those for cell death and cellular senescence, in aging tissues. However, we did not detect any significant increase in markers for apoptosis or cellular senescence (fig. S3), indicating that cell death and cellular senescence are not likely to be the major fate of HFSCs during HF aging. Next, to investigate the possibility that aged HFSCs may leave the niche and change their fate, we traced the fate of HFSCs during physiological aging using K15-CrePR;CAG-CAT-EGFP mice (Fig. 2). We treated those mice with RU486 at 7 weeks of age to genetically tag HFSC-derived cells with green fluorescent protein (GFP) and then analyzed the fate of HFSC-derived cells during the process of hair cycle progression (Fig. 2A). In young mice, GFP+ cells, which are originally localized in the bulge area of resting HFs, formed the lower permanent portion to grow each hair during hair cycle progression. In contrast, in presenile mice as early as 16 months, they became distributed in the junctional zone (JZ) located above the bulge area, then in the suprabasal epidermis, and eventually on the skin surface (Fig. 2B). This indicates that aged HFSC-derived progeny migrate up toward the skin surface through the JZ and the epidermis without renewing themselves in the niche upon induction of a hair cycle. Interestingly, those HFSC-derived GFP+ cells ectopically express the epidermal differentiation markers keratin 1 (Krt1/K1) and its partner keratin 10 (Krt10/K10), both of which are normally expressed by suprabasal cells in the interfollicular epidermis (IFE), even in the niche (Fig. 2, B and C, and fig. S4, A and B). They also expressed other epidermal differentiation markers, including involucrin, ectopically in the JZ, where they coexpressed the JZ keratinocyte marker LRIG1 (Fig. 2, D and E). Those cells are often connected to each other showing a string-like distribution mostly within the suprabasal layer of the JZ and in the epidermis. This indicates that aged HFSCs that committed to epidermal terminal differentiation concomitantly leave the niche to collectively migrate to the suprabasal layer of the epidermis upon their activation at anagen. A similar elimination of LacZ-tagged HFSCs from the niche was observed using K15-CrePR;Rosa26R mice at 16 months but not in young mice (12 weeks) (Fig. 2F). HF differentiation markers, such as AE15, were not found in keratinocytes in the bulge area of aged HFs (fig. S4C). These data demonstrate that aged HFSCs have committed to epidermal terminal differentiation and keratinization to be eliminated from the stem cell pool and the skin (Fig. 2G).

Fig. 2 Epidermal differentiation of aged HFSCs underlies the stepwise progression of HF aging.

(A) Experimental design for genetic fate analysis of HFSC-derived cells. K15-crePR;CAG-CAT-EGFP and K15-crePR;ROSA26R mice were treated with RU486 at 7 weeks to induce EGFP or LacZ expression in HFSCs. At 12 weeks or 16 months, skin samples were collected 3 days after hair cycle induction. (B to E) IF images for K1 (B), K10 (C), LRIG1 (D), and Involucrin (E) in EGFP-expressing HFSC-derived cells in K15-crePR;CAG-CAT-EGFP after hair cycle induction. K1+/K10+/LRIG1+/Involucrin+ cells were induced at 16 months in the bulge or JZ of early anagen HFs (shown by arrows). In addition, some of these K1+ cells appeared in the epidermis (B). (F) Similar experiments were performed using K15-crePR;ROSA26R mice. IF images for K1 combined with LacZ staining revealed that HFSC-derived LacZ-tagged cells are distributed to the bulge (I to III), the JZ (III to VI), and the epidermis (V) in the indicated HF stages, with the coexpression of K1. (G) Schematic for the fate of HFSC-derived cells. Young HFSCs contribute to follicular differentiation, whereas aged HFSCs contribute to epidermal differentiation, thereby causing HF miniaturization. (H) GO analysis between aged and young aHFSCs in ≥2-fold up-regulated genes. The GO terms for epidermal keratinocyte differentiation were significantly enriched in aged aHFSCs. (I) GSEA enrichment score curve of young versus aged aHFSCs with gene sets for epithelial differentiation. NES, normalized enrichment score. (J) Quantitative RT-PCR analysis of K1, c-Myc, Notch1, and Fbxw7 mRNA expression in aHFSCs (CD34highITGA6highSCA-I fraction) in aged (25- to 26-month-old) mice and in 8-week-old control mice; N = 3 to 5 mice for each group. Error bars, mean ± SEM; *P < 0.05 (Student’s t test).

To understand the mechanism of epidermal differentiation of HFSCs, we performed microarray gene expression profiling of the HFSC fraction (CD34highITGA6high) of mouse dorsal skin containing either resting (telogen) or activated (anagen) HFs (tHFSCs or aHFSCs, respectively) from young (8 weeks) and from aged (24 to 25 months) mice (fig. S5). As shown in fig. S6, A and B, gene set enrichment analysis (GSEA) of the transcriptomes of young and aged HFSCs showed that the general stemness gene expression signature, which is maintained in young mice, is significantly abrogated in both tHFSCs and aHFSCs in aged mice. Analysis of mRNA expression of individual HFSC signature genes also showed some reduction in the tHFSC fraction (fig. S6C). Then we compared the transcriptome data of young and aged HFSCs in activated HFs with those from the IFE by hierarchical clustering. The heat map shows that aHFSCs from aged mice show significant similarity to the IFE (fig. S6D). Gene Ontology (GO) analysis of the transcriptome also revealed that aged aHFSCs show significant induction of genes related to epidermal keratinocyte differentiation (Fig. 2, H and I). Microarray, quantitative reverse transcription polymerase chain reaction (QPCR), and immunohistochemical analysis all showed that NOTCH1 (41, 42) and c-MYC, key factors for epidermal differentiation (43), are induced in aged HFSCs at anagen phase (Fig. 2J and figs. S6E and S7). Moreover, down-regulation of Fbxw7, which encodes a ubiquitin ligase for c-MYC and NOTCH1 (44, 45), was also found in aged HFSCs (Fig. 2J and fig. S6E). As overexpression of c-MYC or NOTCH1 intracellular domain (NICD) in basal keratinocytes of mice induces the epidermal differentiation of stem cells and their eventual depletion (41, 42) and Fbxw7 deficiency in mice induces keratinocyte differentiation (45), the excessive induction of c-MYC and NOTCH1 in aged HFSCs is most likely to cause the epidermal differentiation of aged HFSCs upon their activation.

Sustained DDR in HFSCs by aging

To understand the underlying mechanism(s) for the commitment of aged HFSCs to epidermal differentiation, we performed GO analysis and GSEA for quiescent HFSCs in resting tHFSCs from young versus aged mice. This analysis revealed that genes involved in cellular stress responses, including DNA damage response (DDR), are enriched in aged HFSCs, whereas genes involved in DNA repair execution are conversely enriched in young HFSCs (Fig. 3, A to C). These data suggest that aged quiescent HFSCs are responding to accumulating DNA damage that has not been repaired. Indeed, comet assays, a method to evaluate DNA damages, of the sorted aged HFSCs in resting HFs show a significant comet tail moment indicative of DNA strand breaks (Fig. 3, D and E), demonstrating that DNA damage accumulates in HFSCs during aging.

Fig. 3 Accumulation of DNA damage and sustained DDR in aged HFSCs.

(A and B) Enriched GO terms related to DNA damage in the ranked lists of genes significantly up-regulated or down-regulated (≥2-fold) between aged and young tHFSCs. (C) GSEA enrichment score curve of DNA repair–related gene sets for young versus aged tHFSC illustrates efficient DNA repair in young tHFSCs. (D and E) Comet analysis of the HFSC (CD34highITGA6highSCA-I) fraction from aged (24 months) and young (8 weeks) mice. (D) Representative tail moments stained with cyber green. (E) Frequency of tail moments in aged and young HFSCs. (F) Representative IF images for γ-H2AX foci (red dots) in K15+ HFSCs (green) located in the bulge/sub-bulge (Bg/SBg) area of telogen HFs from young (8 weeks) and aged (24 months) wild-type mice. (G) The number of γ-H2AX foci per cell nucleus in 10-μm sections; N = 2 to 4 mice for each group. Error bars, mean ± SEM. γ-H2AX foci were not significantly found in any cell population [dermal cells (Der), epidermal cells (Epi), or DP cells] in young (8-week-old) HFs but were found more frequently in K15+ bulge/sub-bulge cells (Bg/SBg) of aged mouse skin. Error bars, mean ± SEM; *P < 0.05; ***P < 0.001 (Student’s t test).

Because premature hair loss can be induced by exogenous genomic stress such as exposure to IR, as well as by endogenous genomic instability in progeroid syndromes (12, 15, 4648), we first investigated whether HFSCs are responding to accumulating DNA damage in HFSCs and then whether the DDR mediates their epidermal commitment. We found that the formation of γ-H2AX foci (a marker of DDR representing sites of DNA strand breaks) was significantly increased in bulge/sub-bulge keratinocytes in resting HFs (telogen) from aged mice (Fig. 3, F and G, and fig. S8A) and was even greater in more advanced HL areas (fig. S8B). Because DDR is usually induced only transiently just after the induction of DNA damage, the sustained DDR in quiescent HFSCs is a notable event. Notably, the increased number of foci was found not only in K15+ cells but also in adjacent K15 cells in the bulge and sub-bulge area (Fig. 3F), suggesting that quiescent HFSCs lose their stem cell signature after sustained DDR. Thus, the sustained DDR in HFSCs is likely to be an early sign of HFSC aging, which precedes the loss of the stem cell signature. Furthermore, we found that hair cycling induction itself is sufficient to induce abundant DNA damage foci formation in renewing HFSCs (fig. S8C), indicating that physiological hair cycling itself generates DNA damage in HFSCs. More DNA damage foci were found in remaining bulge keratinocytes in hair loss areas on the dorsal-most area compared with surrounding less-affected areas (fig. S8B), often with reduction of HFSC marker expression in those cells. These data suggest that the accumulation of unrepaired DNA damage in tHFSCs over anagen phase causes the HFSC-aging state that is characterized by their loss of the HFSC signature with sustained DDR and their epidermal commitment.

Wnt signaling drives hair cycle induction and maintenance (49), and the persistent activation causes DDR in HFSCs (50), whereas inactivation of Wnt signaling has been implicated in male-pattern baldness [androgenic alopecia (AGA)] (51). To investigate the possible involvement of Wnt signaling in HFSC aging, we performed GSEA for tHFSCs from young versus aged mice and found that Wnt signaling is less activated in aged tHFSCs than in young tHFSCs (fig. S9A), but it was not significant with aHFSCs as well as with immunostaining (fig. S9, B to D). Therefore, we concluded that HF aging is not simply explained by altered Wnt signaling, yet Wnt signaling is likely to indirectly mediate the accumulation of DNA damage via replicative stress in HFSCs.

Genomic instability accelerates HFSC aging

Because sustained DDR is found in quiescent HFSCs during aging, extrinsic genomic instability, such as that caused by IR, or by intrinsic instability by DNA-repair–deficient, progeroid mutations may cause similar changes in HFSCs. Indeed, the loss and reduction of HFSC signatures was induced after exposure of the skin to 10-Gy (gray) irradiation (IR), which induces sustained DNA-damage foci formation in HFSCs, even in young mice (fig. S10). Genetic fate-tracing studies of HFSCs in irradiated mice (fig. S11A) showed that the expression of K1 was induced in HFSC-derived LacZ+/GFP+ keratinocytes in the permanent portion of HFs, including the bulge, isthmus JZ, and infundibulum within 1 week after hair cycle induction by 10-Gy IR (fig. S11, B and C). These data indicate that genomic stress is sufficient to induce the epidermal commitment of quiescent HFSCs. Furthermore, those K1+LacZ+ cells often showed a string-like distribution from the upper portion of HFs to the overlying epidermis, including from the suprabasal to the uppermost layer, which is desquamated at 1 week or later (fig. S11, B and C). Furthermore, excessive expression of c-MYC and NOTCH1, key drivers of epidermal differentiation, were also induced in K15+ bulge keratinocytes in irradiated mice at 3 days after hair cycle induction (fig. S12). Coupled with the epidermal differentiation of HFSCs and their migration toward the epidermis, HF miniaturization was triggered (fig. S13, A to C). These data indicate that genomic stress causes HFSC dysregulation which closely mimics physiological the “HFSC aging state” characterized by sustained DDR in HFSCs, loss of HFSC signatures, and their epidermal commitment.

We found a similar phenomenon with XPDTTD/TTD mutant mice, a progeroid mouse model for human DNA repair syndrome trichothiodystrophy (TTD) (52, 53). Those mice also show a diffuse patterned hair loss that starts from the dorsal neck area (fig. S13D) and a reduction of HFs by HF miniaturization, thinning of the skin, and cyst formation in hair loss areas (fig. S13E), as was seen in aged mouse HFs. XPDTTD/TTD mutant mice also show similar sustained formation of γ-H2AX foci and loss or reduction of HFSC signature expression (fig. S13F). These data demonstrate that HFSCs with intrinsic or extrinsic genomic instability attain the HFSC aging state after sustained DDR. Importantly, the corresponding histological changes represented by HF miniaturization/loss and skin thinning by the XPDTTD/TTD mutation show similarities to those of physiological HF aging and associated skin changes as seen in aged wild-type mice (fig. S13, G and H), which indicates the existence of a common skin aging process or program.

COL17A1 proteolysis in HFSCs triggers HFSC aging

To understand the key mechanism for HFSC aging, we compared the gene expression profiles of quiescent HFSCs from aged mice with young mice. Genes that encode hemidesmosomal components, including Col17a1, were specifically reduced in aged tHFSCs, whereas genes that encode junctional components of desmosomes and adherens junctions were not affected (fig. S14, A and B). Because hemidesmosomal collagen COL17A1 (BP180/BPAG2), which we previously reported to be critical for HFSC maintenance (39), is anchored to Laminin332 (Laminin5) with their direct association with β4 integrin (54) to stabilize the complex, we focused on the role of COL17A1 and its dysfunction in HFSCs. COL17A1 distribution was abundantly found on the whole-cell surface and the cytoplasm of young bulge keratinocytes, but there was a profound reduction of COL17A1, often with a fine linear distribution on the basement membrane zone, in a number of HFs by aging (Fig. 1I). A similar pattern of COL17A1 distribution and accelerated loss was found in bulge keratinocytes of TTD progeroid mice as well (fig. S13I). These findings suggested the possibility that the nonhemidesmosomal COL17A1 in HFSCs is preferentially degraded by proteolysis during aging and is accelerated by genomic stress.

Because COL17A1 is known to be degraded or shed by induction of several different proteases, such as ADAM9, ADAM10, ADAM17 (55, 56), MMP9 (57), and ELANE/ELA2 (58), we searched for potentially responsible proteases that are induced in aged HFSCs as identified by the microarray data (fig. S14C). That analysis showed that Elane, which encodes neutrophil elastase/elastase 2 (59), known as a cellular toxic protease when aberrantly secreted, is ectopically induced within aged HFSCs but is undetectable in young HFSCs (Fig. 4A). Immunohistochemical analysis showed that ELANE, which is normally expressed in the nonbasal layer of the mouse epidermis but not in HFs, is conspicuously and aberrantly induced in the bulge area by aging (Fig. 4, A and B, and fig. S14D) in approximately 40% of HFs at 16 months and in 60% of HFs at 25 to 27 months. The aberrant induction of ELANE in HFSCs was also detectable in vivo 12 hours after genomic stress such as 10-Gy IR (Fig. 4C). It is notable that the exogenous genomic stress, which was sufficient to induce sustained DDR in HFSCs (fig. S10), also induces COL17A1 depletion, often with linear remnant distribution at the basement membrane zone in HFSCs, even in young mice (Fig. 4C). Reduced levels of COL17A1 and sustained expression of ELANE often show a reciprocal distribution in bulge keratinocytes, which suggests that ELANE directly degrades COL17A1 in vivo. Indeed, primary keratinocytes treated with ELANE showed that both the 180-kD COL17A1 and its shed form of the 120 kD ectodomain are quickly degraded by ELANE in vitro (Fig. 4D). That degradation is completely inhibited by the protease inhibitor α1-antitrypsin (A1AT), suggesting that COL17A1 is efficiently proteolytically degraded by ELANE, which is induced within HFSCs in response to DNA damage. Some HFs restored the original distribution pattern and expression level of COL17A1 within a week after IR, whereas other refractory HFs with reduced COL17A expression in the bulge often showed a down-regulation of other key HFSC molecules (fig. S10E). This indicates that depletion of COL17A1 becomes stabilized in some HFs after the DDR-induced proteolysis of COL17A1 in HFSCs. Consistently, we found a statistically significant reduction or loss of COL17A1 in bulge keratinocytes with sustained foci formation of 53BP1 and γ-H2AX in aged mice (Fig. 4, E and F). Therefore, unrepaired DNA damage in HFSCs is most likely to cause the “HFSC aging state” characterized by their COL17A1 depletion with sustained DDR, their loss of stem cell signature, and their epidermal commitment upon their activation.

Fig. 4 DDR-induced proteolysis of COL17A1 in HFSCs induces dynamic HF miniaturization through their epidermal differentiation.

(A to C) Aging- or DDR-induced COL17A1 proteolysis by ELANE. (A) IF images for COL17A1 and ELANE in the HF bulge of young (8-week-old), presenile (12 to 16-month-old), and aged (27-month-old) mice. (B) Quantitative analysis of fluorescence intensity of ELANE in the bulge (Bg) and in the basal epidermis (Epi) of 8- to 12-week-old, 16- to 18-month-old, and 25- to 27-month-old mice; several HFs (>5) at stage I and II from each group were examined. N = 3 mice for each group. (C) IF images for COL17A1 and ELANE or SOX9 and keratin 15 (K15) in the HF bulge (8 weeks) at 12 hours, 1 day, and 7 days after 10-Gy irradiation. 7d indicates typical HFs that have lost COL17A1 expression and other HFSC markers with ELANE induction, whereas 7d R indicates typical HFs that have restored or maintained normal HFSC marker expression at 7 days after 10-Gy irradiation. (D) Dose-dependent proteolysis of COL17A1 by ELANE and its inhibition by A1AT in primary keratinocytes at 2 hours of treatment. (E) IF images of 53BP1 and γ-H2AX foci (red) in COL17A1+ bulge cells (green) from young (8-week-old) and from aged (25-month-old) wild-type mice. The foci were found in the COL17A1low area as shown with arrows. (F) Number of 53BP1 foci in the COL17A1low area and COL17A1high at 8 weeks or 25 months. 53BP1 foci were significantly found in the COL17A1low area at 25 months. (G to M) Analysis of HFSC-specific Col17a1-deficient (Col17a1 cKO) mice and control (Cont) mice after repeated depilation. (G) Hair coats on the backs of control (Cont) and Col17a1 cKO mice at 8.5 months. Hair loss and graying was induced in Col17a1 cKO but not in control mice after depilation. (H) Histological images of Cont and Col17a1 cKO mouse skin; miniaturization of HFs found in a Col17a1 cKO mouse (arrows). (I) HF number per mm of Cont (N = 3) and Col17a1 cKO (N = 6) mice. (J) Stepwise HF miniaturization by Col17a1 deficiency is morphologically similar to physiological HF aging; representative histological images are shown. (K) Number of HFs at each HF stage per mm in Cont (N = 3) and in Col17a1 cKO (N = 6) mice. Col17a1 cKO mice show the premature increase of HF miniaturization. (L and M) Induction of epidermal differentiation concomitant with the up-regulation of c-MYC and NOTCH1 expression by Col17a1 deficiency. (L) Immunostaining images for c-MYC and LHX2 or NOTCH1 and K1 in Cont and in Col17a1 cKO mice at 12 weeks old after hair cycle induction. c-MYC expression was ectopically induced in the LHX2+ bulge area of Col17a1 cKO mice. NOTCH1+K1+ cells were found in the upper bulge area of Col17a1 cKO mice. (M) Venn diagram showing overlap between ≥2-fold increased genes in aged versus young aHFSCs and ≥2-fold increased genes in Col17a1cKO versus Cont aHFSCs. GO analysis for genes commonly up-regulated in aged aHFSCs and Col17a1 cKO aHFSCs. The top GO terms enriched in the overlap are related to epidermal differentiation and to keratinization. (N) Genetic fate analysis of HFSC-derived cells in Col17a1-deficient mice with K15-crePR and ROSA26R alleles. Ectopic appearance of K1+LacZ+ HFSC-derived cells in the bulge at 7 days in the JZ and in the epidermis after hair cycle induction in Col17a1 cKO mice (shown by arrows). Error bars, mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test).

Our previous study showed that HFSCs in Col17a1-deficient mice show complete HFSC depletion after the loss of the HFSC signature in bulge cells as early as 7 to 8 weeks and show progressive hair thinning and graying (39). To investigate whether the COL17A1 deficiency in HFSCs recapitulates the “HFSC aging state” and to determine the tissue outcome, we generated HFSC-specific Col17a1-deficient mice (Col17a1 cKO) (fig. S15). We found that those mice also show thinning hair and graying hair (Fig. 4G). GSEA of tHFSCs revealed that the general stemness gene expression signature is abrogated upon Col17a1 deficiency (fig. S16, A and B). The heat map for the global transcriptome of activated HFSC fractions (aHFSCs) revealed a significantly close relationship between Col17a1 cKO HFSCs and aged HFSCs (fig. S16C). The aberrant induction of c-MYC and NOTCH1, as well as epidermal differentiation markers, was found in HFSCs upon their activation during hair cycle progression and eventual HF miniaturization and skin thinning (Fig. 4, H to L). Indeed, the most highly enriched GO terms in the overlap of up-regulated genes between Col17a1-deficient HFSCs versus control and aged HFSCs versus young HFSCs were epidermal differentiation and keratinization (Fig. 4M). Furthermore, the fate analysis of Col17a1-deficient HFSCs also showed that HFSC-derived cells coexpress K1 in the bulge area and also in the upper JZ and the epidermis during hair cycle progression (Fig. 4N), similar to that seen during physiological aging and IR (Fig. 2, A to E, and fig. S11). It is also notable that HFSC aging directly or indirectly involves various surrounding cells to drive the stepwise HF miniaturization and skin thinning. These data collectively demonstrate that the COL17A1 deficiency in HFSCs recapitulates the “HFSC aging” state and drives the HF aging process.

COL17A1 rescues HFSC aging and HF miniaturization

To address whether age-associated depletion of COL17A1 is a key for induction of the “HFSC aging state” and associated HF aging processes, we analyzed COL17A1-overexpressing mice [K14-hCOL17A1 transgenic (tg)] (60) and examined whether the forced maintenance of COL17A1 in HFSCs can rescue HFSC depletion and age-associated HF miniaturization or loss. Most of those mice showed significantly fewer miniaturized HFs and an apparent retardation of hair loss even in mice surving for 24 months (N = 3 mice) and 32 months (N = 2 mice) (Fig. 5, A to D). In contrast, all aged wild-type littermates showed HF miniaturization or loss, skin thinning, and age-associated hair loss. Consistently, the forced maintenance of COL17A1 significantly rescued the age-associated down-regulation of the HFSC signature (Fig. 5, E and F). These results demonstrate that the maintenance of COL17A1 is not only indispensable for HFSC maintenance but also effective for the protection of HFSCs against “HFSC aging” and resultant “HF aging” characterized by HF miniaturization, hair loss, and skin thinning.

Fig. 5 Forced maintenance of COL17A1 in HFSCs protects their depletion and HF aging.

(A) Hair coats on the backs of wild-type (WT) and human COL17A1 transgenic (hCOL17A1 tg) mice at 17 months, 24 months, and 32 months. The hair loss was rescued by hCOL17A1 transgene (tg) expression (n > 3 mice for each group). (B) Histological images of WT and hCOL17A1 tg mice at 24 to 25 months. Arrows indicate miniaturized HFs at stages III and IV. Arrowheads show HFs at Stage I. (C) HF number per mm in WT and in hCol17a1 tg mice. N = 3 mice for each group. (D) Number of HFs at each HF stage per mm in WT and in hCOL17A1 tg mice at 24 to 25 months. N = 3 mice for each group. (E) IF images for HFSC signature (K15 and LHX2) expression in WT and in hCOL17A1 tg mice at 24 to 25 months. The reduction of HFSC signatures with aging was rescued by hCOL17A1 tg expression. (F) Quantitative analysis of the number of LHX2+ nuclei in the bulge. The reduction of LHX2+ cells by aging was rescued in hCOL17A1 tg mice. N = 3 mice for each group. Error bars, mean ± SEM; *P < 0.05 (Student’s t test).

Human HF miniaturization with COL17A1 depletion

To examine whether human HFs also have a similar aging program that leads to senile alopecia, we analyzed the histology of healthy human scalp skin from the temporal areas of women at ages ranging from 22 to 70 years old. We found that human female scalps from the aged group (55 to 70 years old) contain significantly more miniaturized HFs (defined as HFs measuring <30 μm in diameter) compared with the younger group (35 to 45 years old) (Fig. 6, A and B, and fig. S17). To examine the changes of human HFSC signatures with aging, we performed a quantification analysis of the immunofluorescent intensities for human HFSC markers (61). We found that COL17A1, K15, and CD200 expression is significantly decreased only in miniaturized HFs but not in nonminiaturized HFs, even in aged scalp skin (Fig. 6, C and D, and fig. S18). Furthermore, the size of the K15+ bulge region is significantly decreased in miniaturized HFs (Fig. 6E), and the number of DNA damage foci is significantly greater in bulge keratinocytes from middle-aged or aged female scalp skin compared with younger scalp skin (Fig. 6, F and G). Furthermore, K15+ bulge cells that coexpress K1 were also occasionally found in aged HFs (Fig. 6H), indicating the defective renewal of HFSCs. Taken together, these data demonstrate the existence of human HFSC aging that is triggered by sustained DDR and the associated COL17A1 depletion and resultant progression of the HF aging process.

Fig. 6 Human HF miniaturization with loss of COL17A1 expression mimics mouse HF aging.

(A) H&E images of human scalps from aged (59- and 70-year-old) and from middle-aged (40-year-old) women. Arrows indicate normal HFs, and arrowheads indicate miniaturized HFs. (B) Percentage of miniaturized HFs. Miniaturized HFs were significantly increased in aged (55- to 70-year-old) HFs. (C) HFSC marker expression in young and in aged HFs. Immunostaining of human COL17A1 and K15 in miniaturized or nonminiaturized HFs in human scalps from 33-year-old and 68-year-old women. Dashed lines indicate the bulge area. (D) Quantitative analysis of fluorescence intensities for human COL17A1. Human COL17A1 expression was significantly down-regulated in aged (60- to 70-year-old) miniaturized HFs. (E) Quantitative size analysis of K15+ bulge area. The bulge size is significantly diminished in aged (60- to 70-year-old) miniaturized HFs. (F) γ-H2AX foci formation in K15+ bulge keratinocytes and in K15+ IFE of human scalps at different ages. Immunostaining for γ-H2AX and Keratin 15 is shown. Arrow indicates γ-H2AX foci. (G) Quantitative analysis of γ-H2AX foci fluorescence intensity. The fluorescence level was significantly increased in aged (45- to 70-year-old) bulge areas but not in aged interfollicular epidermal basal cells. (H) Ectopic epidermal differentiation in the human bulge (Bg) area. Representative immunostaining for K1 (green) and K15 (red) are shown. K1 expression was induced in the 59-year-old scalp. All scalps were derived from temporal regions. Error bars, mean ± SEM. n.s., no significant difference; *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test). yo, years old.

Mechanisms of mammalian HF aging

Tissues and organs generally become smaller and atrophic with aging (1). Our findings now provide evidence that an epithelial tissue aging program driven by “HFSC aging” exists and that this program is triggered by the proteolysis of COL17A1 in HFSCs and progressively advances in a stepwise manner with dynamic tissue architectural changes (Fig. 7). The stepwise dynamics have not been explained by previous stem cell aging theories (18, 62). It is notable that the “HFSC aging state” triggered by the DDR-induced deficiency of COL17A1 causes a cyclic progression of aging-associated HF changes (HF aging) characterized by HF miniaturization with profound involvement of many other types of surrounding cells in the skin and resultant hair loss. It is also notable that the aged mini-organ eventually disappears from the skin with retention of many other surrounding intact HFs. The HF aging process was delayed or rescued by repressing HFSC aging with the forced maintenance of COL17A1 in HFSCs. These results demonstrate that stem cell aging is the keystone for induction of the tissue aging program that brings about thinning hair and underline that the stabilization of COL17A1 in HFSCs is critical and also effective to prevent HFSC aging and associated skin tissue changes.

Fig. 7 Mechanisms for HFSC aging and execution of the HF-aging program.

(A and B) Schematic for the mechanism of HF aging. The HFSC pool is maintained in a long quiescent state in the HF bulges (Bg) during aging (Stage Ia). HFSCs with sustained DDR undergo COL17A1 proteolysis through induction of ELANE protease (Stage Ib). Those COL17A1low/– HFSCs lose their stem cell signature and commit to epidermal keratinocytes (Epi KC) in the niche. Those “aged” HFSCs migrate up toward the epidermis upon their cyclic activation through the JZ. They terminally differentiate into cornified keratinocytes to be eliminated from the skin surface, thereby causing the stepwise miniaturization of HFs and hair thinning and loss.

Previous studies reported that the cellular ratio of the HFSC fraction (CD34highITGA6high) is often paradoxically increased in aged mouse skin compared with young skin (2729). This is probably because the aging skin contains HFs at different HF stages and with a single bulge or multiple bulges (Figs. 1G and 7A). In any case, the complete depletion of HFSCs was apparent in more advanced skin areas of aged mice. Because the genetic background (31), gender, circadian rhythms (63), and even the housing environment affect the onset of aging-associated baldness in wild-type mice, the combination of those factors (63) may determine the onset of HFSC aging.

Miniaturization of HFs has long been believed to be a specific key phenomenon for male-pattern baldness [androgenic alopecia (AGA)] but not for HF aging (64). Our study revealed that mammalian HFs do miniaturize during aging regardless of sex. Because HFSCs are maintained at least at the onset of AGA (65), AGA-associated HF miniaturization seen in the advanced stage of AGA may represent HF aging that progresses inevitably during repeated hair follicle cycles.

Finally, what is the physiological role of the aging program? Our previous study with melanocyte stem cells (18, 66) and other studies with hematopoietic stem cells (67) reported the existence of the “stemness checkpoint” that determines whether stem cells stay in an immature state or commit to differentiation to eliminate the damaged and stressed cells from the stem cell pool. Furthermore, the role of the checkpoint as a cancer barrier also has been proposed with leukemic stem cells (68). Similarly, the elimination of stressed and damaged HFSCs from the skin surface as shed corneocytes through their terminal epidermal differentiation may also be useful to maintain the high quality of HFSCs. Therefore, the active restoration of stem cell pools by use of residual intact stem cells before the irreversible advancement of tissue architectural changes may be essential for successful tissue regeneration and anti-aging in solid tissues such as the skin.

Materials and methods


C57BL/6N mice were purchased from Sankyo Lab Service (Tokyo, Japan) and from CLEA Japan (Tokyo, Japan). XPD mutant mice (52), K15-CrePR1 (69), Rosa26R mice (70, 71), K14-CreERT2, CAG-CAT-EGFP, and Keratin14-human COL17A1 transgenic mice (72) have been described previously. For conditional knockout of the Col17a1 gene in mice, Col17a1 floxed mice whose exon2 is flanked by 2 loxP sequences were generated. All transgenic and mutant mice were backcrossed to C57BL/6J mice from Sankyo Lab Service. Mice with asymmetrical hair loss patches or whisker trimming were not used to exclude mice with excessive grooming. Animal care was in accordance with the guidance of Kanazawa University and the Tokyo Medical and Dental University for animal and recombinant DNA experiments. All animal experiments were performed following the Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Committee of Laboratory Animals. Offspring were genotyped by PCR-based assays of mouse tail DNA.

Hair cycle induction by depilation or SAG (smoothened agonist) administration

For hair cycle induction, hair on the dorsal skin of 7- to 8-week old mice was manually depilated to induce a new hair cycle only after confirmation that the skin had a light pink color, which indicates that HFs are synchronized at the telogen phase. For SAG treatment, the neck regions of 12-week-old and 19-month-old mice were shaved using an automatic hair shaver. Then, 100 μl SAG (100 μM) (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was topically applied to the shaved area once a day for 5 days.

Ionizing irradiation and hair cycle induction

Mouse skin was irradiated with x-rays at 7 to 8 weeks of age only after confirmation that the skin had a light pink color, which indicates that HFs are synchronized at the telogen phase. Irradiation was carried out by placing each mouse in a thin-walled plastic box. Low-pressure irradiation of mouse skin was performed using a Hitachi MBR-1520 (Hitachi Medical, Tokyo, Japan) operating at 50 kVp, 20 mA with a 2.0-mm Al filter (dose rate ≈ 0.4 Gy/min) or a RX-650 machine (Faxitron X ray) operating at 100 kVp with a 2.0-mm Al filter (dose rate ≈ 0.3 Gy/min).

ELANE (neutrophil elastase) and A1AT (alpha 1-antitrypsin) treatment in vitro

To test the proteolytic activity of ELANE against COL17A in primary keratinocytes (Kurabo, Osaka, Japan), primary keratinocytes were cultured for 24 hours in DermLife K Medium Complete Kit (Kurabo) containing human neutrophil elastase (ELANE) (50 to 500 ng/ml) (Millipore, Watford, UK) with or without A1AT (3.48 ug/ml) (Athens Research and Technology, GA, USA).

Western blotting

Primary keratinocytes (Kurabo) and HaCaT cells were harvested, homogenized, and sonicated with TNE buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 1 mM EDTA) containing 1X Protease inhibitor tablet (Roche, Mannheim, Germany), 1X Phosphatase inhibitor tablet (Roche), and 1% NP-40. Protein concentrations were measured using a BCA Protein Assay Kit (Pierce, Rockford, IL). The whole-cell lysates were subjected to mini-protean-tgx-precast gel (Bio-Rad, Hercules, CA, USA) electrophoresis (0.02 mA/mini-gel) for 1 hour. Using semidry transfer, the proteins were transferred to polyvinylidene fluoride (PVDF) membranes for 1 hour at 0.25 mA/mini-gel using a Transblot SD (Bio-Rad). The membranes were then immersed in Blocking One (Nacalai) for 30 min and were then incubated overnight with each primary antibody diluted in Blocking One. Primary antibodies used were mouse polyclonal antibody to human COL17A1 (Clone NC16A3, 1:100) and polyclonal antibody to αβ-tubulin (1:500) (all from Abcam, Cambridge, MA, USA). After washing three times for 5 min each with 1xTBST, the membranes were incubated for 2 hours with horseradish peroxidase (HRP) conjugated secondary antibodies diluted in Blocking One and washed three times for 5 min each with 1xTBST. After incubation with a Luminata Forte Western HRP Substrate detection system (Millipore) for a few minutes, immunoblot images were acquired using an LAS-3000 luminescence image analyzer (Fuji Photo film, Tokyo, Japan).

Preparation of paraffin and frozen sections

For paraffin sections, mouse skin or human scalp skin was fixed in 10% formalin solution at 4°C overnight and was then embedded in paraffin. Paraffin-embedded skin specimens were cut in 5-μm-thick sections (Rotary Microtome HM325, MICROM International GmbH, Walldorf, Germany). For frozen sections, mouse skin or human scalp skin was immersed in ice-cold 4% paraformaldehyde (4% PFA) in phosphate-buffered saline (PBS) (pH 7.4) and then was irradiated in a 500-W microwave oven for three 30-s cycles with intervals and then kept on ice for 20 min. The fixed skin samples were embedded in optimal cutting temperature (OCT) compound (Sakura Finetechnical Co. Ltd., Tokyo, Japan), snap-frozen in liquid nitrogen, and stored at –80°C. Frozen samples were cut in 10-μm-thick sections (Cryomicrotome CM 1850, Leica Microsystems Nussloch GmbH, Nussloch, Germany).

Hematoxylin and eosin (HE) staining

Paraffin sections were deparaffinized and then rehydrated. Frozen sections were removed from the OCT compound and then were stained with hematoxylin-eosin (Sakura Finetechnical Co. Ltd.) for analysis of tissue histology using an optical microscope.

Assessment of HF changes by aging, genomic stress, XPDTTD/TTD mutation, Col17a1 deficiency, and human COL17A1 overexpression

Five-μm-thick sections of murine skin were stained with HE. To properly assess skin sections that contain upright HFs, we chose sections that met the following criteria for further analysis. First, the section must include full-length cross sections of hair shafts in multiple HFs. Second, the section should not contain any exceptional pathological changes, including skin wounding. More than three different skin areas from the indicated number of mice were analyzed.

Staging of chronological HF aging

We categorized telogen HFs of C57BL/6N mice at different ages into five different stages by following the HF structural changes during chronological aging, as described in the legend for Fig. 1. Briefly, at Stage I, HFs show a normal structure and size. At Stage II, HFs show a slight shortening with proportional size and structure. At Stage III, HFs are miniaturized with complete loss of the dermal papillae, with retention of the hair shaft. At Stage IV, HFs are miniaturized with retention of sebaceous glands and the infundibulum. At Stage V, HFs have completely disappeared with or without retention of different degrees of fibrosis in the dermis. This staging method tends to underestimate the structural changes of large HF types ,including tylotrich follicles that grow guard hairs. Those follicles also miniaturize and become indistinguishable from other smaller HF types, including those growing short zigzag hairs (underfur) during aging. We thus categorized all HFs that are the same or larger than young (7- to 8-week-old) zigzag HFs with proportional HF structure and size as Stage I, although this underestimates the size changes of the large types of HFs that grow over hair.

Assessment of human HF miniaturization

HE images of 10-μm skin sections of at least four randomized different areas in each sample were acquired using an upright BX51 microscope (Olympus, NY, USA). The percentage of miniaturized HFs (defined as HFs measuring <30 μm in diameter) in total HFs were counted (73).

Immunofluorescence staining

Frozen sections were used for immunofluorescence analysis. For c-Myc immunostaining, antigen retrieval was performed by boiling the slides for 20 min in Dako target retrieval solution (Dako, Carpinteria, CA, USA). Nonspecific staining was blocked by incubation with PBS containing 3% skim milk (Difco, Detroit, USA) for 30 min. Tissue sections were incubated with the primary antibody at 4°C overnight and were subsequently incubated with secondary antibodies conjugated with Alexa Fluor 488, 594, or 680 (Invitrogen, Hercule, CA, USA). After washing in PBS, 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI) (Invitrogen-Molecular Probes, Eugene, OR, USA) was added for nuclear counterstaining. Coverslips were mounted onto glass slides with fluorescent mounting medium (Thermo Electron Corp., Waltham, MA, USA). All images were obtained using an upright BX51 microscope (Olympus) or an FV1000 confocal microscope system (Olympus).


Primary antibodies used: Rabbit antibody to phospho histone H2AX (Cell Signaling, Danvers, MA, USA); chicken antibody to human keratin 15 (Covance, Berkeley, CA, USA); rabbit antibody to S100A6 (Lab Vision, Fremont, CA, US); goat antibody to mouse LHX2 (Santa Cruz Biotechnology); rabbit antibody to mouse SOX9 (Santa Cruz Biotechnology); rat antibody to mouse COL17A1 (Nishimura’s laboratory); rabbit antibody to mouse keratin 1 (Covance); rabbit antibody to mouse keratin 10 (Covance); rabbit antibody to mouse filaggrin (Covance); rabbit antibody to mouse involucrin (Covance); mouse antibody to Trichohyalin (AE15) (Abcam); rabbit antibody to human TP53BP1 (Lifespan BioScience, WA, USA); rabbit antibody to mouse AXIN2 (abcam); rabbit antibody β-catenin (sigma); rabbit antibody to cleaved caspase-3 (Cell Signaling); FITC-conjugated antibody to mouse CD34 (eBioscience, San Diego, CA, USA); goat antibody to mouse NOTCH1 (Santa Cruz Biotechnology); chicken antibody to β galactosidase (Abcam); rabbit antibody mouse P16 (Santa Cruz Biotechnology); rabbit antibody to c-MYC (Abcam); goat antibody to mouse LRIG1 (R&D Systems, Abingdon, Oxfordshire, UK); rat antibody to mouse PDGFRa (eBioscience); rabbit antibody to mouse keratin 6 (Covance); mouse antibody to human COL17A1 (Clone D20, Nishimura laboratory); rat antibody to human COL17A1 (Clone 3F11A9, Nishimura laboratory); rabbit antibody to human COL17A1 (Clone NC16A3, Abcam); rabbit antibody to human ELANE (Abcam); Alexa 647 rat antibody to human CD200 (Serotec, Oxfordshire, UK); and mouse antibody to human keratin 1 (Leica microsystems).

Secondary antibodies used: Alexa Fluor 488 goat antibody to rat immunoglobulin (IgG) (Molecular Probes); Alexa Fluor 488 goat antibody to chicken IgY (Molecular Probes); Alexa Fluor 594 goat antibody to rabbit IgG (Molecular Probes); Alexa Fluor 594 goat antibody to rat IgG (Molecular Probes); Alexa Fluor 488 goat antibody to Fluorescein (Molecular Probes); Alexa Fluor 680 donkey antibody to rabbit IgG (Molecular Probes); Alexa Fluor 594 donkey antibody to goat IgG (Molecular Probes); and FITC donkey antibody to chicken IgY (Abcam).

Senescence-Associated β-Galactosidase (SA-β-Gal) staining

A Senescence Cells Histochemical Staining kit was used (Sigma-Aldrich, St Louis, MO, USA) according to the manufacturer’s instructions.

FACS analysis and isolation of HFSCs

Subcutaneous fat was removed with a scalpel, and the whole skin was placed dermis down in 0.25% trypsin (GIBCO) for 10 to 13 hours at 4°C, then was moved to 37°C for 20 min. Single-cell suspensions were obtained by scraping the skin gently. The cells were then filtered with strainers (40 μm) (BD Falcon, San Jose, CA, USA). Cell suspensions were incubated with the appropriate antibodies for 1 hour at 4°C. The following antibodies were used: CD34-FITC (eBioscience), Integrin α6-PE (BD Bioscience, San Jose, CA, USA), and ScaI-APC (MACS; Miltenyi Biotec, Bergisch Gladbach, Germany). 7-AAD (BD Bioscience) was used to exclude dead cells. Cell isolations were performed using a FACS AriaII equipped with Diva software (BD Bioscience). For RNA extraction of HFSCs, CD34highITGA6highSCA-I cells were sorted directly into Buffer RLT (Qiagen, Valencia, CA, USA).

Alkaline comet assay

Alkaline comet assays were performed with FACS-sorted HFSCs (CD34highITGA6highSCA-I) according to the manufacturer’s instructions (Trevigen, Gaithersburg, MD, USA). Briefly, 2 × 103 cells diluted in 10 μl PBS were added to 100 μl 1% Low Melting Agarose and placed onto slides. Cells were lysed in lysis solution (2.5 M NaCl, 100 mM EDTA, pH 10.0, 10 mM TrisHCl, 1% sodium lauryl sacosinate) for 60 min on ice and then were placed in alkaline lysis solution (200 mM NaOH, 1 mM EDTA, pH>13) for 60 min at room temperature. Slides were then run for 30 min at 21 V. Nuclei were stained with Cyber Green. As a positive control, DNA damage was induced by soaking in a 100 μM H2O2 solution for 10 min. Images of DNA damage comets were acquired using an upright BX51 microscope (Olympus). A percentage of Tail DNA [Tail intensity/(Tail intensity + Head intensity) × 100] and the tail moment (Tail length × a percentage of Tail DNA/100) were measured using Image J software. The tail moments were measured on 40 to 100 cells per slide.

Microarray analysis

To analyze global gene expression changes by aging both in activated and in quiescent HFSCs and IFE keratinocytes, HFSCs (CD34highITGA6highSCA-I-), IFE (ITGA6highSCA-Ihigh), and total epidermal cells were harvested from young and from aged mice. To analyze global gene expression changes by Col17A1 gene deficiency in HFSCs, HFSCs were harvested from control wild-type mice and K14-creERT2;Col17a1 flox/flox mice at 12 weeks old after intraperitoneal (i.p.) administration of Tamoxifen (TAM) (4 mg/mouse for 5 days from 7 to 8 weeks old). Dissociated cells were FACS-sorted and reserved into Buffer RLT (Qiagen) by FACS aria II. Total RNAs were isolated using a RNA easy micro Kit (Qiagen) according to the manufacturer’s instructions. cDNA samples were prepared from purified ng RNA using the Ovation Pico WTA System V2 (NuGen, San Carlos, CA, USA) according to the manufacturer’s instructions. Amplified cDNAs were labeled with Cyanine 3 (Cy3) using a SureTag DNA Labeling Kit (Agilent Technologies, Santa Clara, CA, USA) following the manufacturer's instructions. RNA and cDNA quality and cyanine incorporation were analyzed using an Agilent Bioanalyzer 2100 (Agilent) and a Nanodrop ND-1000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). For each hybridization, 1.3 μg Cy3-labeled cDNA were fragmented, and hybridized at 65°C for 17 hours to an Agilent SurePrint G3 Mouse GE v2 8x60K Microarray (Design ID: 028005). After washing, microarrays were scanned using an Agilent DNA microarray scanner. For normalization of microarray data, intensity values of each scanned feature were quantified using Agilent feature extraction software version, which performs background subtractions. We only used features that were flagged as no errors (Detected flags) and excluded features that were not positive, not significant, not uniform, not above background, saturated, or population outliers (Compromised and Not Detected flags). Normalization was performed using Agilent GeneSpring version 13.0 (per chip: normalization to 75 percentile shift; per gene: normalization to median of all samples). There are a total of 55,681 probes on the Agilent SurePrint G3 Mouse GE 8x60K Microarray (Design ID: 028005) without control probes. The altered transcripts were quantified using the comparative method. We applied more than a 2.0-fold change in signal intensity to identify significant differences of gene expression in this study.

Hierarchical clustering and Gene Ontology analysis by GeneSpring, MeV and DAVID software

Hierarchical clustering using the Pearson correlation was performed using Agilent GeneSpring version 13.0 or TIGR MeV software (version 4.8.1, To generate the Venn diagram, TIGR MeV software was used. To analyze the GO terms of ≥2.0-fold increased or decreased gene cluster and also the common gene cluster in the Venn diagram, the web-based Database for Annotation, Visualization and Integrated Discovery (DAVID) 6.7 was used (74). To extract reliable GO terms belonging to biological process, Fisher’s exact test and multiple test correction (P < 0.05) were used.

Gene set enrichment analysis

GSEA software is provided by the Broad Institute of MIT and Harvard University. To define the lists of genes involved in “DNA repair,” “Epidermal differentiation,” and “Stemness,” the Molecular signature (MSig) list was searched on Msig (Database) DB of the Broad Institute web site ( We selected and applied “HALLMARK OF DNA REPAIR” as “DNA repair” MSig, “BOSCO EPITHELIAL DIFFERENTIATION” as “epidermal differentiation” MSig, “RAMALHO STEMNESS UP” as “stemness” MSig, and “RAMALHO STEMNESS DOWN” as “dominant negative stemness” MSig. To apply the Normalized chip expression data into GSEA software, the Mouse chip annotation file was changed to the Human chip annotation file. For GSEA, the following parameter settings were used: number of permutations = 1000; collapsed data set to gene = true; permutation type = gene set; enrichment statistics = weighted; metric for ranking genes = signal to noise. To decide significant MSig change of GSEA results, the nominal P values (<0.05) and false discovery rate (FDR) q values (<0.05) were used.

Accession number

Microarray data were deposited in the Gene Expression Omnibus ( under series identifier GSE72863.

Quantitative RT-PCR

Total RNAs were purified from FACS-sorted cells by directly sorting into Buffer RLT (Qiagen) using a RNA easy micro Kit (Qiagen). cDNAs were then synthesized from equal amounts of RNA in 20-μl reaction mixtures using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA) according to standard procedures. For quantitative PCR, cDNA was added to 20 μl of the reaction mixture containing 10 μl SYBR Green qPCR Kit (Agilent Technology) and 0.5 μl 12.5 μM primers (forward and reverse). Relative levels of expression were determined by normalization to acidic ribosomal phosphoprotein PO (arbp) or hypoxanthine-guanine phosphoribosyltransferase (Hprt), using the ΔΔCt method (Biological replicate, pool of 6 to 10 mice). The reactions were run in Mx3000P Real-Time QPCR System (Agilent Technology). The primer sequences used were as follows:Arbp: 5′-ATAACCCTGAAGTGCTCGACAT-3′, 5′-GGGAAGGTGTACTCAGTCTCCA-3′.Hprt: 5′-CAACGGGGGACATAAAAGTTATTGGTGGA-3′, 5′-TGCAACCTTAACCATTTTGGGGCTGT-3′. Notch1: 5′- ACAACAACGAGTGTGAGTCC -3′, 5′- ACACGTGGCTCCTGTATATG-3′. c-myc: 5′- GCCCCTAGTGCTGCATGAG -3′, 5′-CCACAGACACCACATCAATTTCTT-3′. Keratin1: 5′- AGGATCTTGCCAGATTGCTG -3′, 5′- CTACTGCTTCCGCTCATGCT-3′. P16: 5′-GGGTTTCGCCCAACGCCCCGA-3′, 5′-TGCAGCACCACCAGCGTGTCC-3′, Fbxw7 5′-TCTTGTCTCTGGGAATGCAG-3′, 5′-CCGTCGTCTGAGCTGGTAAT-3′.

HFSC fate analysis with K15-CrePR; Rosa26R/CAG-CAT-EGFP mice

For induction of β-galactosidase or enhanced green fluorescent protein (EGFP) expression in K15-expressing cells, K15-CrePR1;CAG-CAT-EGFP mice and K15-CrePR1;Rosa26R mice were treated with RU486 (2.5 mg/day, diluted in 80% ethanol; Sigma) topically on their dorsal skin during telogen (7 weeks old) for 5 days. For hair cycle induction, telogen hairs at 7 to 8 weeks old, 16 months old, or just after 10-Gy irradiation were manually depilated under anesthesia.

Quantification of γ-H2AX foci

The images of immunostaining for γ-H2AX were acquired by FV1000 confocal microscopy (Olympus). Visible foci were counted in >100 nuclei for each cell type (bulge/sub-bulge; dermal papilla; epidermis and dermis). The number of foci per nucleus was calculated as mean ± SEM.

Measurement of fluorescence intensity and the number of positive signals in bulge nuclei

Images of immunostaining for K15, S100A6, and mouse COL17A1 in mouse bulge areas or images of immunostaining for K15, CD200, human COL17A1, and γ-H2AX in human bulge areas were acquired by confocal microscopy (Olympus). The images were then quantified using Olympus fluoview software (version 1.7) or ImageJ software (NIH image). The fluorescent intensity of the background was removed and then normalized against DAPI intensity. For counting the number of cells with positive signals (SOX9, LHX2) in K15+ bulge nuclei, the images of immunostaining for SOX9 and LHX2 were acquired by confocal microscopy (Olympus), and the number of DAPI+ cells with positive signals was measured.

Measurement of HF frequency

HE images of 10-μm skin sections were acquired by optical microscopy. The distance between two adjacent HFs was measured, and HF frequency was calculated as the number of HFs per mm of skin section.

Statistical analysis

To determine significance between two groups, comparisons were performed using an unpaired two-tailed Student t test.

Supplementary Materials

References and Notes

  1. Acknowledgments: We thank Y. Nishimori, K. Inomata, Y. Tadokoro, R. Yajima, D. Nanba, H. Shimizu, K. Yancey, I. Morita, F. Ishino, M. Kanai, M. Ohyama, S. Inui, S. Itami, and M. Amagai for their helpful support and DASS Manuscript for editing. E.K.N. has been supported by a Grant-in-Aid for Scientific Research S (26221303), a Grant-in-Aid for Scientific Research on Innovative Areas “Stem Cell Aging and Disease”(26115003), Funding Program for Next Generation World-Leading Researchers (NEXT Program) (LS042), Grant-in-Aid for Young Scientists from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and also by the Takeda Science Foundation. E.N. and H.M. are authors on a patent applied for by Tokyo Medical and Dental University that relates to drugs that modulate COL17A1 or ELANE expression or function for alopecia.
View Abstract

Navigate This Article