“Wnt-dependent de novo hair follicle regeneration in adult mouse skin after wounding” by Mayumi Ito, Zaixin Yang1, Thomas Andl1, Chunhua Cui1, Noori Kim1, Sarah E. Millar1 & George Cotsarelis
During our studies on wound healing in mice, we noticed structures within the centre of large healing wounds that resembled early developing hair follicles. To characterize these structures, we performed timed experiments in which a 1 cm2 square of full-thickness back skin was excised from 3-week-old mice, at least 2 weeks after the last hair follicles had formed. By 10–11 days after wounding, contracture and re-epithelialization resulted in wound closure and an approximately 0.25 cm2 area composed of an epidermis and a dermis with no evidence of hair follicles (Fig. 1a). At 14–19 days after wounding, small epidermal downgrowths that resembled developing embryonic hair follicles were present (Fig. 1d–h). Older (7–8-week-old to 10-month-old) mice also showed hair follicle neogenesis after wounding, but larger wounds (2.25 cm2) were required to trigger follicle formation (Supplementary Table 1). The final rather than initial size of the wound seemed to correlate with hair follicle neogenesis because the larger wound in older mice also yielded a 0.25 cm2 area immediately after re-epithelialization.
Hair follicles consist of at least ten different epithelial and mesenchymal cell types geared towards the production of hair7. In the embryo, hair follicle development begins with the formation of a small cluster of epidermal cells (epithelial placode) that can be detected by expression of cytokeratin 17 (KRT17), an intermediate filament protein8. Placodes overlie a dermal condensate, which is identified by alkaline phosphatase activity9. Through a series of mesenchymal–epithelial interactions initiated by activation of Wnt and requiring downstream Shh signalling, placode cells proliferate, move downward and engulf the dermal condensate, eventually forming a mature follicle that cyclically produces hair throughout life10.
We discovered that hair follicle neogenesis following wounding paralleled embryonic follicle development at the molecular level. The truly nascent nature of these follicles was demonstrated by absence of expression (by PCR with reverse transcription; RT–PCR) of hair follicle differentiation markers KRT17 (ref. 8) and Lef1 (refs 10, 11) in the epidermis for several days after wound closure (Fig. lm), and their subsequent appearance coinciding with the development of hair germs and pegs (Fig. li, j, m). The newly formed hair follicles also proliferated normally (Fig. 1l) and generated hair as well as sebaceous glands (Fig. 1c, h).
We asked next whether the de novo follicles arise from hair follicle stem cells in skin bordering the wound. Hair follicle stem cells have been localized to the bulge area12, 13, and bulge cells in follicles surrounding a 4 mm wound send progeny towards the centre of the wound during re-epithelialization14, 15. However, the majority of bulge cell progeny contribute transiently to the new epidermis and do not persist beyond three weeks14. To investigate whether the nascent follicles that developed following larger wounds originated from hair follicle stem cells, we performed genetic lineage analysis using inducible Tg(Krt1-15-cre/PGR)22Cot;R26R transgenic mice (Fig. 2, Supplementary Fig. 4 and Supplementary Table 3)13, 14. These mice express CrePR1, a fusion protein consisting of Cre-recombinase and a truncated progesterone receptor that binds the progesterone antagonist RU486 under the control of the Krt1-15 promoter, which is active predominantly in bulge cells of adult mouse skin13, 16. Treatment of adult Tg(Krt1-15-cre/PGR)22Cot;R26R mice with RU486 resulted in permanent expression of lacZ in bulge cells and in all progeny of labelled bulge cells13, 14. We discovered that although bulge cell progeny migrated to the centre of the larger wounds, they did not persist. Less than 3% of the new hair follicles were labelled, suggesting that non-hair-follicle bulge cells were the primary source of regenerated follicles (Fig. 2, Supplementary Fig. 4 and Supplementary Table 3).
To investigate further the cells of origin of the regenerated follicles, we examined Krt1-15-CrePR*;R26R transgenic mice. These mice possess a mutated progesterone receptor, PR*, containing a larger portion of the progesterone binding domain compared with the PR1 clone17. The Krt1-15-CrePR* mice exhibit Cre recombinase activity during development when the Krt1-15 promoter is active in the epidermis16. In these mice, approximately 70% of bulge and 50% of non-bulge epidermal cells are labelled before wounding (Fig. 2b, Supplemental Table 3; ref. 14). After wounding, approximately half of the regenerated follicles possessed cells expressing lacZ (Fig. 2h, j, l). Regenerated follicles were chimaeric for lacZ expression (Fig. 2l), indicating that multiple progenitor cells were required for new hair follicle formation, as in development18. Thus, together with near absence of lacZ expression in regenerated follicles of induced Tg(Krt1-15-cre/PGR)22Cot;R26R skin, these data indicate that new follicles originated from cells outside of the hair follicle stem cell niche. The new follicles arose from cells in the epidermis and/or upper portion of the follicle (infundibulum). Both of theses areas are considered to possess stem cells that normally undergo epidermal rather than follicular differentiation19, 20. Our findings are the first to indicate that non-hair-follicle stem cells in genetically normal adult mice acquire competence to form hair follicles in response to wounding.
Because hair follicle stem cells are necessary for follicle survival and cycling14, we determined whether the regenerated follicles possessed a functional stem cell population. As with normal follicles, we found label-retaining cells and Krt1-15 promoter activity (Fig. 3, Supplementary Fig. 5), consistent with the re-establishment of a hair follicle stem cell population. The regenerated follicles produced hairs and cycled up to three times within 90 days after wounding (Fig. 3c–e, Supplementary Note 2), indicating the presence of functional stem cells.
The new hairs lacked pigment (Fig. 1c) and associated melanocytes (data not shown), suggesting that the melanocyte stem cell niche was not re-established or that it could not be repopulated22. These findings parallel the lack of bulge-derived epithelial cells in the regenerated hair follicles, because, in mice, melanocyte precursors localize to the bulge. Furthermore, melanocytes are not normally present in the interfollicular epidermis of adult mouse back skin22, precluding repopulation from an epidermal population.
We then asked whether overexpression of the secreted ligand, Wnt7a, in KRT14-Wnt7a (which targets the epidermis) transgenic mouse epidermis (Supplementary Fig. 6), would enhance hair follicle neogenesis following wounding. Wnt7a has been shown to maintain the hair-follicle-inducing capacity of cultured dermal papilla cells25. The overexpression of activated -catenin, an intracellular Wnt effector, in epidermis induces new hair follicles26, 27, and exogenous Wnt promotes formation of cysts with hair follicle differentiation28; however, to date, there has been no evidence that extracellular Wnt ligands can promote actual hair follicle neogenesis in adult skin. Wound closure (time to re-epithelialization) was normal in KRT14-Wnt7a mice. However, the transgenic mice developed over twice the number of hair follicles within the wounded area compared with controls (Supplementary Table 1, Fig. 4i, j, l, m). The increased hair follicle number was due to a larger area within the wound that developed follicles (18 4% in controls versus 40 15% in KRT14-Wnt7a mice, P = 0.05) at the same density as controls (Supplementary Table 2). Thus, excess Wnt in combination with wound healing potentiates regeneration of hair follicles, perhaps by altering cell fate and increasing the number of cells competent to produce hair.
A major therapeutic goal for those studying stem cells is the ability to coax these cells to differentiate into different cell types with the hope of eventually forming organs through tissue engineering29. By taking advantage of existing regenerative programmes, we demonstrated that a wound stimulus is sufficient to trigger regeneration of hair follicles from epithelial cells that do not normally form hair. This de novo formation of hair follicles in adult animals recapitulates embryogenesis at the molecular level, and provides a potential window for manipulating the number of hair follicles that form, by exposure to Wnts. This raises the possibility of treating acute wounds with modulators of the Wnt pathway to decrease scar formation, and treating hair loss by regenerating follicles through wounding and Wnt pathway activation.