Receptor activator of NF-κB (RANK) stimulatesthe proliferation of epithelial cells of theepidermo-pilosebaceous unitVincent Duherona, Estelle Hessa, Monique Duvala, Marion Decossasa, Beatriz Castanedab,c, Jennifer E. Klöpperd,Leonela Amoasiia, Jean-Baptiste Barbarouxe, Ifor R. Williamsf, Hideo Yagitag, Josef Penningerh, Yongwon Choii,Frédéric Lézotb, Richard Grovese, Ralf Pausd,j, and Christopher G. Muellera,1

aCentre National de la Recherche Scientifique, Laboratory of Therapeutic Immunology and Chemistry, Unité Propre de Recherche 9021, Institut de BiologieMoléculaire et Cellulaire, University of Strasbourg, 67084 Strasbourg, France; bInstitut National de la Santé et de la Recherche Médicale, Unité Mixte deRecherche 872, Cordeliers Research Center, Team 5, Laboratory of Oral Molecular Physiopathology, F-75006 Paris, France; cUniversity of Antioquia, Medellin,Colombia; dDepartment of Dermatology, University Hospital Schleswig-Holstein, University of Lübeck, 23562 Lübeck, Germany; eSt. John’s Institute ofDermatology, Division of Genetics and Molecular Medicine, King’s College London School of Medicine, London SE1 9RT, United Kingdom; fDepartmentof Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA 30322; gDepartment of Immunology, Juntendo University Schoolof Medicine, Tokyo 113-8421, Japan; hInstitute of Molecular Biotechnology of the Austrian Academy of Sciences, 1030 Vienna, Austria; iDepartment ofPathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, PA 19104; and jSchool of Translational Medicine, Universityof Manchester, Manchester M13 9PT, United Kingdom

Edited* by Pierre Chambon, Institut de Génétique et de Biologie Moléculaire et Cellulaire, Illkirch, France, and approved February 14, 2011 (received forreview September 10, 2010)

Receptor activator of NF-κB (RANK), known for controlling bonemass, has been recognized for its role in epithelial cell activation ofthe mammary gland. Because bone and the epidermo-pilosebaceousunit of the skin share a lifelong renewal activity where similarmolecular players operate, and because mammary glands and hairfollicles are both skin appendages, we have addressed the func-tion of RANK in the hair follicle and the epidermis. Here, we showthat mice deficient in RANK ligand (RANKL) are unable to initiatea new growth phase of the hair cycle and display arrested epider-mal homeostasis. However, transgenic mice overexpressing RANKin the hair follicle or administration of recombinant RANKL bothactivate the hair cycle and epidermal growth. RANK is expressedby the hair follicle germ and bulge stem cells and the epidermalbasal cells, cell types implicated in the renewal of the epidermo-pilosebaceous unit. RANK signaling is dispensable for the forma-tion of the stem cell compartment and the inductive hair folliclemesenchyme, and the hair cycle can be rescued by Rankl knockoutskin transplantation onto nude mice. RANKL is actively transcribedby the hair follicle at initiation of its growth phase, providinga mechanism for stem cell RANK engagement and hair-cycle entry.Thus, RANK–RANKL regulates hair renewal and epidermal homeo-stasis and provides a link between these two activities.

osteoprotegerin | TRANCE | anagen | S100A8

Receptor activator of NF-κB (RANK), a member of the TNFreceptor family (TNFRSF11a), was originally identified as a

regulator of bone density. Mice deficient in Rank or Rankl displayincreased bone density owing to reduced osteoclast formation, butthe loss of the RANK ligand (RANKL) decoy receptor osteopro-tegerin (OPG) results in lower bone mass because of uncheckedRANK activation (1, 2).RANK has emerged as an important player in epithelial cell

growth and differentiation. It is required for the formation oflactating mammary glands (3, 4), thymic medullary epithelialcells (5), and intestinal microfold cells (6) and has been impli-cated in the growth and metastasis of prostate (7) and mammaryepithelial cancers (8, 9). Thus, RANK affects a great variety ofepithelial cells of different organs. The skin is the largest epi-thelial surface, and its epidermo-pilosebaceous unit comprisesthe interfollicular epidermis (IFE), the hair follicle (HF), and thesebaceous gland. The HF has the particularity of undergoingcycles of growth (anagen), regression (catagen), and relativequiescence (telogen) (10), making it an excellent system forstudying epidermal (stem) cells and organ remodeling (11–14).Each HF is composed of a permanent upper portion, which

includes the sebaceous gland and the bulge region, and a tem-porary lower cycling portion. The local balance of hair growthstimulators and inhibitors is critical for initiation of new hairgrowth (15–17). Intriguingly, many of these molecular playersalso operate in bone development and remodeling (e.g., mem-bers of the TGF-β superfamily and their antagonists, parathyroidhormones, and hedgehog proteins) (18).Because Rank and Rankl-null mice exhibit defects in mam-

mary gland development (3) and in the eruption of teeth (19),which, like HFs, are skin appendages, it raises the question of therole of the RANK–RANKL–OPG triad in HF formation andcycling. The finding that Opg is transcribed by the bulge stemcells (20, 21) supports an important role of these proteins in HFbiology. However, so far this question has not been investigated.The IFE is intrinsically linked to the HF through the outer-

root sheath, which merges distally the basal layer of IFE andproximally the HF bulge and the HF bulb (10). Like HFs, theIFE undergoes renewal (homeostasis). In this process, kerati-nocytes of the basal layer divide and terminally differentiate toreplace upper-lying cells. RANKL is expressed by the activatedIFE, and the epidermal Langerhans dendritic cells carry RANK(22, 23). RANK engagement on Langerhans cells leads to anincrease in their cell numbers and to immune response modu-lation (22, 23). However, the role of RANK in epidermal ho-meostasis has not been addressed.We have studied whether RANK plays a functionally impor-

tant role in the stimulation of epithelial cells of the HF and theIFE. Using Rankl knockout (Rankl-KO) and Rank transgenic(Rank-Tg) mice, as well as recombinant RANKL and anti-RANKL mAb, we here present previously unreported evidencethat RANK functions in HF growth and in epidermal renewal.

ResultsRANK Stimulation Is Dispensable for Normal HF Development but IsRequired for HF Anagen.Many molecular regulators of HF cyclingalso drive HF development (morphogenesis) (17). Therefore, we

Author contributions: V.D., E.H., J.E.K., J.-B.B., F.L., R.G., R.P., and C.G.M. designed re-search; V.D., E.H., M. Duval, M. Decossas, B.C., J.E.K., L.A., J.-B.B., and C.G.M. performedresearch; I.R.W., H.Y., J.P., Y.C., F.L., and R.P. contributed new reagents/analytic tools;V.D., E.H., M. Decossas, B.C., J.E.K., L.A., J.-B.B., F.L., R.P., and C.G.M. analyzed data; andR.P. and C.G.M. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1To whom correspondence should be addressed. E-mail: c.mueller@ibmc.u-strasbg.fr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1013054108/-/DCSupplemental.

5342–5347 | PNAS | March 29, 2011 | vol. 108 | no. 13 www.pnas.org/cgi/doi/10.1073/pnas.1013054108

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

first investigated the importance of RANK stimulation in HFmorphogenesis. We analyzed the expression of RANK, RANKL,and OPG mRNA by RT-PCR in total skin of wild-type (WT)mice at postnatal day 5 (P5) and P15 and found that the geneswere transcriptionally active (Fig. S1A). In situ hybridization ofRANK and RANKL mRNA at P5 showed that the genes weretranscribed in the IFE and the HF (Fig. S1B). This findingconfirms previous data of RANK and RANKL expression in theembryonic IFE and HF (24). We then compared HF morpho-genesis of Rankl−/− mice and control littermates. At P5, the backskin HFs of WT mice were in postnatal HF morphogenesis andby P15 began to enter catagen. Exactly the same observation wasmade in Rankl−/− mice (Fig. S1C), whose postnatal HF mor-phogenesis and macroscopic hair phenotype appeared in-distinguishable from those of WT littermates (25). Furthermore,all four hair types (monotrich, awl, auchene, and zigzag) werefound (Fig. S1D), indicating that RANK stimulation is dispens-able for HF morphogenesis.Next, we studied whether absent RANK stimulation affected

the HF renewal cycle with entry into the hair growth phase ataround P24 in C57BL/6 mice (26). Strikingly, Rankl−/− HFs didnot enter anagen by P28 through to P49, whereas WT littermateshad long proceeded into anagen or had already reached theirsecond catagen (Fig. 1 A and B and Fig. S2A). Also at P25, therewas no sign of anagen initiation, showing a genuine block of HFcycling rather than the occurrence of an abortive, drastically

shortened growth phase (Fig. S2B). As expected, given thatRANKL is the only ligand for RANK (27), Rank−/− mice alsofailed to undergo transition from telogen to anagen (Fig. 1C).Because hair plucking can experimentally provoke hair-cycle

activation (28), we tested this stimulus on Rankl−/− mice. Fivedays after hair plucking, 24-d-old control mice were in advancedanagen, but the Rankl-KO animals still remained in telogen (Fig.1D). Thus, in the absence of RANK stimulation by RANKL, themouse HF is arrested in first postnatal telogen and is unable toenter its anagen growth phase.

Telogen Arrest in the Absence of RANK Stimulation Does Not Resultfrom Defects of the HF Stem Cell Compartment or from Defective HFMesenchyme. At least two distinct subpopulations of HF epithe-lial cells are needed for normal anagen formation, the bulge stemcells (11, 29) and the secondary hair germ (SHG) cells (21, 30).Therefore, we looked for the presence of these cells in theRankl−/− animals.Immunofluorescence for CD34, a marker of murine bulge

stem cells (31), showed that they were preserved in the Rankl−/−

telogen HF (Fig. 2A). Their quantification by flow cytometryrevealed that their number was only slightly reduced (Fig. 2B).Immunofluorescence for P-cadherin (21) showed that the SHGwas also present (Fig. 2A). Thus, the absence of RANKL doesnot deplete these cells sufficiently to explain the HF arrestin telogen.Anagen development also requires inductive signals from the

mesenchymal follicular dermal papilla (DP) (32). Therefore, weexamined its presence by alkaline phosphatase enzyme histo-chemistry (33). The enzyme activity in the correct position be-neath the HF revealed the presence of morphologically andenzyme-histochemically normal DP in Rankl−/− mice (Fig. 2C),which shows that that the Rankl−/− telogen HF contained all ofthe cellular elements required for telogen–anagen transition.We therefore tested whether the hair-cycle arrest can be res-

cued. We transplanted shaved, 21-d-old Rankl−/− skin, whose HFswere all in telogen, onto age-matched nudemice. BecauseRANKLis produced during wound healing by neovascularization and byactivated epidermal keratinocytes (22, 34), it could be expectedthat recipient RANKL is synthesized and complements the geneticdeficiency. Hair regrowth and anagen development of the graftedskin was monitored. We observed that, by 10 d, both the WT andthe Rankl−/− transplanted skin had acquired a dark appearance,a sign of anagen, and, by 3 wk, both transplants presented full hairregrowth (Fig. 2D). Skin cross-sections confirmed that HFs of theRankl-nullmicewere in anagen.These findings established that theepithelial stem cell compartment and the inductive HF mesen-chyme of Rankl−/− HFs are functional.

RANK Is Expressed by HF Epithelial Stem Cells. To better understandhow RANK regulates the telogen–anagen transition, we de-termined the site of RANK expression in telogen HFs. Immu-nofluorescent labeling localized RANK to the bulge and theSHG (Fig. 3A). Also RANKL and the RANKL decoy receptorOPG were expressed at these sites (Fig. 3A). The immunoreac-tivity of RANK and RANKL was low, requiring signal amplifi-cation, whereas OPG could be visualized without amplification.This finding concurs with published gene profiling results thathad shown abundant transcripts of Opg in murine bulge cells (20,21). RANK, RANKL, or OPG were undetectable in the DP. RT-PCR of CD34+ and P-cadherin+ cells, which were isolated andsorted by FACS (21), confirmed transcriptional activity of Rank,Rankl, and Opg in the CD34+ bulge and the P-cadherin+ SHGepithelial cells (Fig. 3B).We then investigated whether RANKL and OPG expression

changed during the hair cycle by measuring their mRNA byquantitative RT-PCR in total skin. OPG gene expression wasmaximal in early to mid-telogen, whereas that of RANKLpeaked in early anagen (Fig. 3C). Because RANKL and OPGare soluble proteins, we measured their protein levels by ELISAin the supernatant of a 24-h skin culture. Fig. 3D shows that the

Fig. 1. RANKL is required for HF anagen. (A) At P28, Rankl−/− HFs arearrested in telogen, whereas WT littermates are in anagen. (B) Schematicrepresentation of HF morphogenesis and cycling in Rankl−/− and controls,and the failure of all Rankl−/− mice to enter anagen. (C) Arrest of Rank−/− HFsin telogen. (D) At 5 d after hair plucking at P19, HFs were in advancedanagen in WT mice but remained in telogen in Rankl−/− mice. Mock-treatedWT mice displayed early anagen. (Scale bars, 100 μm; in Insets, 50 μm.) Theparaffin-embedded sections were stained with hematoxylin/eosin.

Duheron et al. PNAS | March 29, 2011 | vol. 108 | no. 13 | 5343

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

RANKL and OPG protein levels reflected their mRNA changeswith maximal levels of OPG at telogen followed by a peak ofRANKL. To uncover the site of RANKL expression during itspeak, we performed RT-PCR on separated IFE and dermal HF-associated cells (Fig. 3E); it disclosed that RANKL expressionwas restricted to the HF-associated cells. Anti-RANKL/keratin14 immunofluorescent labeling of the whole skin revealedRANKL in the lower HF (Fig. 3F), and costaining with GATA-3,a marker for the HF inner root sheath (35), showed thatRANKL is localized to the inner root sheath and the cuticle layer(Fig. S3A). Because RANKL is a soluble protein and the de-tection at these sites could be the consequence of binding toextracellular components, we visualized its mRNA on sections byin situ hybridization. Again, we saw the RANKL gene activity inthe lower HF (Fig. S3B). There was no RANKL in the IFE ex-cept for some rare cells, in accord with earlier findings (22) (Fig.S3C). In anagen skin, RANK was seen in the bulge, the outerroot sheath, and the IFE (Fig. 3G). Double labeling for RANKwith CD34 and integrin α6, both markers for bulge cells (20, 31),validated RANK expression in the anagen bulge stem cells (Fig.S3D). Staining of OPG in anagen skin showed the protein in thebulge but not in the lower HF part (Fig. 3H). OPG can also beobserved in basal keratinocytes (Fig. S3E).Together, these findings showed that RANK, RANKL, and

OPG are expressed by bulge and SHG cells. As the hair cycleprogresses from telogen to anagen, OPG expression decreasesbut that of RANKL increases. Thus, at the telogen–anagen tran-sition, more OPG-free RANKL is available to engage RANKand drive the HF stem cells into proliferation.

RANK Stimulation Triggers Anagen Entry. To further address theimportance of RANK activation in anagen induction, we useda mouse Tg for Rank under the control of the S100A8 genepromoter. This promoter had already been noted as active in thehuman HF (36) and, recently, was found transcribed in the SHG(21). Immunofluorescent labeling of telogen HFs confirmed thatRANK was overexpressed in the SHG (Fig. S4A). EndogenousRANK was not seen because its visualization required signalamplification. Expression of the Tg RANK construct in HEK293T cells showed that RANK was localized at the cell surfaceand that it activated NF-κB signaling as expected (37) (Fig. S4B).We also monitored RANKL and OPG production in these miceby ELISA of skin cultures from weeks 3 to 5. The Tg mice dis-played an increase of RANKL, particularly during its peakproduction at anagen, but showed a reduced OPG level (Fig.S4C). Thus, with RANK overexpressed in the HF and a morefavorable RANK-activating RANKL/OPG ratio, we expected anincreased HF RANK engagement.

Indeed, we saw that the first anagen entry was advanced by 1or 2 d (Fig. 4 A and C). Moreover, two rapidly evolving pre-mature anagen cycles occurred at weeks 8 and 11, whereascontrol littermates were in telogen (Fig. 4 B and C). Three-month-old Tg mice showed mild alopecia (Fig. S4D), similar toother mutant mice with abnormally fast HF cycling (38).To test whether hair renewal can be prematurely induced in

WT mice by RANK stimulation, we injected recombinantRANKL s.c. into 8-wk-old mice in telogen. The functionality ofthe recombinant RANKL fused to the GST tag was confirmed byNF-κB signaling in RANK-expressing HEK 293T cells (Fig.S4E). Control mice received GST only. Hair-cycle analysis 6 dafter injection revealed that three of four mice responded torecombinant RANKL-GST by premature anagen (Fig. 4 D andE). These findings support the importance of RANK stimulationas a critical hair-cycle entry signal.

RANKL Regulates Epidermal Homeostasis. BecauseRank-Tg animalsexpress more soluble RANKL but less OPG in the skin, we in-vestigated whether this change would affect the IFE. First, weverified epidermal RANK expression in the Tg mice and foundthat it was expressed by the basal keratinocytes (Fig. S5A). Wecould exclude the detection of Tg RANK because the Ab is di-rected against an N-terminal peptide, which is absent from thecDNA construct (Fig. S6). It was apparent that the Rank-Tg epi-thelium was greatly thickened (Fig. 5 A and B). In contrast, theepidermal size of the Rankl-KO mice was slightly but significantlyreduced (Fig. 5A and B). We therefore determined the epidermalrenewal rate by using the thymidine analog BrdU. ProliferatingBrdU+ keratinocytes were virtually absent in the Rankl-null micebut highly abundant in the IFE of the Rank-Tg mice (Fig. 5C).Quantification of IFE BrdU+ cells by cell counting on sectionsuncovered a significant reduction for the Rankl-KO mice and ahighly significant eightfold increase for the Tg mice (Fig. 5D). Wenext analyzed Tg skin for signs of cell activation by measuring therelease of TNFα and IL-1β. As shown in Fig. S5B, there was a smallbut significant increase in TNFα and IL-1β in the Tg skin at P21and P30, respectively. However, there was no dermal CD3+ T-cellinfiltration (Fig. S5C), and culturing P21 epidermal keratinocytesresulted in three times more cell recovery after 2 wk comparedwithWT (Fig. S5D). This observation supports a cell-autonomousenhanced proliferation in the Tg mice; however, because of theincreased inflammatory cytokine production, we cannot formallyexclude their contribution to the accelerated epidermal renewal.Except for keratin 14, which extended beyond its normal re-striction in the Tg animals and is likely secondary to the highproliferation rate (39, 40), the differentiation of the IFE in bothmouse models was normal as judged by the expression of keratin

Fig. 2. The HF stem cell and the mesenchymal cell com-partment is RANK-signaling independent. (A) Identificationof CD34+ bulge (B) and P-cadherin+ SHG cells in Rankl−/− andcontrol telogen HFs by immunofluorescence. Nuclei werecolored with DAPI. (Scale bars, 20 μm.) (B Left) Flow cytom-etry of CD34+ bulge cells among keratin 14+ keratinocytes inRankl−/− and WT telogen HFs. Arrows point to CD34+ keratin14+ cells. (Right) Graph depicts the percentage of CD34+ andkeratin 14+ cells in three KO mice and littermate controls(*P < 0.05). (C) Identification of follicular DP by their alkalinephosphatase activity. (Scale bars, 20 μm.) (D) Restored Rankl−/−

hair renewal after transplantation onto nude mice at week 3and analysis at week 6. A Rankl−/− skin section shows HFs inanagen. Image is representative for five transplantation ex-periments. (Scale bar, 100 μm.) The paraffin-embedded sec-tion was stained with hematoxylin/eosin.

5344 | www.pnas.org/cgi/doi/10.1073/pnas.1013054108 Duheron et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

10, loricrin, involucrin, and filaggrin (Fig. S7A). Moreover, therewas no toluidine blue dye penetration (Fig. S7B), demonstratingnormal skin barrier function. Finally, we treated Tgmice with anti-RANKLmAb, which, as shown in Fig. 5E–G, resulted in a delay inthe hair cycle and a significantly reduced epidermal thickness.Collectively, these data show that IFE homeostasis occurs

normally during anagen when HF RANKL production is high,but that IFE hyperproliferates when RANKL is overproduced.

DiscussionWe showed that RANK regulates hair cycling by activating itsgrowth phase: (i) although dispensable for HF development,RANK stimulation was required for anagen to normally occurduring the fourth week after birth; (ii) hair plucking, known totrigger hair renewal, failed to stimulate anagen entry in Rankl-null mice; (iii) mice with Tg RANK expression in the HF and analtered skin RANKL/OPG ratio displayed advanced hair cycles;and (iv) recombinant RANKL induced anagen in WT mice. Inaddition, we showed that RANK regulates IFE renewal because(i) the proliferation rate is reduced in Rankl-KO mice but (ii) itis increased in the Rank-Tg mice and (iii) blocking RANKL inRank-Tg mice results in a delay in the hair cycle and a reducedIFE thickness.

Although RANK, RANKL, and OPG are expressed in the IFEand the HF during morphogenesis, we found no visible effect ofRANKL on HF and IFE development. This finding, togetherwith the observation that RANK signaling does not affect de-velopment of the mammary gland until pregnancy (3), suggestsa role of RANK post-embryogenesis. However, embryonic gen-eration of thymic medullary epithelial cells depends on RANKL,yet CD40L can partially compensate after birth (5), suggestingthat, if RANK is active in the developing skin, it may be maskedby other TNF receptor family members such as Edar or Troy(41–43). Interestingly, lymphotoxin-α—which, like RANKL,functions in lymphoid organ development—also plays a role inHF morphogenesis (44) and is a ligand for Troy (45).RANK or RANKL were required for the anagen phase of the

hair cycle. It is unlikely that the hair-cycle arrest is a reflection ofa sickness of these KO mice because they thrive well beyond 6 wkwithout signs of weakness. Moreover, the hair-plucking experi-ment was performed on young animals and further demonstratesa genuine block in anagen transition. It remains to be establishedwhether vibrissae hair undergo cycling. The inability to regeneratepelage hair was not the result of defects in the HF stem cell andmesenchymal cell compartment because these cellular elementswere present. Indeed, upon transplantation onto a RANKL-proficient environment, Rankl−/− skin HFs passed into anagen.The efficiency with which nude mice rescued the hair-cycle arrestmay be a reflection of the robust RANKL production by activatedepidermal keratinocytes or endothelial cells during the healingprocess (22, 34). In addition, we have found that the anagen HF isan important source of RANKL, and, although nude mice lackhair, they have a normal hair telogen–anagen transition.The expression of RANK, RANKL, and OPG in the HF bulge

and SHG supports the importance this protein triad in the controlof HF cycling. OPG localization to the bulge and the SHG cor-roborates previous studies, which have foundOpg transcription inthese cells (20, 21).Rank orRankl transcripts were not detected inthese studies, probably owing to their low abundance. Indeed,signal amplification was needed to detect RANK and RANKLprotein, and soluble RANKL levels were low during telogen. Thisfinding suggests that OPG serves as an autocrine regulator byblocking engagement of RANK by RANKL during telogen.It is tempting to speculate that, as in themammary gland (3, 46),

HF RANKL is under hormonal control because hair growth isinfluenced by the endocrine system (47). However, it is unlikelythat prolactin or parathyroid hormone related peptide (PTHrp)up-regulate RANKL in the HF because these stimuli antagonizehair renewal and induce catagen in mice (48, 49). However, theRankl promoter contains a number of cell-signaling regulatoryelements (NF-κB, Jun/AP-1, vitaminD, andRunx2) (46), allowingit to respond to amultitude of signals known to act on the hair cycle(17, 50, 51). Further work will be necessary to elucidate the impactof these signals on HF RANKL synthesis.We did not detect any RANK, RANKL, or OPG expression in

the DP, which is similar to mammary glands, where stromal cellsfailed to express RANK or RANKL even after hormonal stim-ulation (46). However, the mammary gland stromal cells con-stitutively transcribe Opg (46). it is possible that Opg is induced inthe DP in response to as-yet-unidentified stimuli.Support for the idea that RANKL and OPG synthesis is sub-

ject to changes is provided by their mRNA and protein measuresduring the hair cycle. It is unlikely that these changes are influ-enced by tissue injury after the skin excision because RNAextractions were performed immediately on the freshly collectedskin, and protein levels closely reflected the mRNA measures. Acloser investigation into the site of RANKL production at P28uncovered that it was synthesized by the lower anagen HF. RT-PCR, mRNA hybridization, and immunofluorescence point tothe proliferating matrix and bulb cells as its source, which wouldconcur with RANKL production by activated keratinocytes (22).Evidence that RANK stimulation by RANKL induces anagen

is provided by the precocious anagen phases in mice over-expressing RANK in the HF and in mice administrated with

Fig. 3. Expression of RANK, RANKL, and OPG in HF stem cells. (A) Locali-zation of RANK, RANKL, and OPG in WT telogen bulge (B) and SHG stemcells by immunofluorescence. Nuclei were colored with DAPI. (Scale bars, 20μm.) Six mice were analyzed for RANK, RANKL, and OPG. (B) RT-PCR of Rank,Rankl, Opg, and Gapdh in FACS-sorted CD34+ bulge and P-cadherin+ SHGcells. (C and D) Measures (mean ± SEM) of Rankl and Opg gene transcriptionby quantitative RT-PCR in whole skin (C) and of soluble RANKL and OPG byELISA in skin-organ cultures at the indicated ages (D). n, number of mice forthe analysis, a minimum of six mice per time point. (E) RT-PCR of Rank,Rankl, Opg, and Gapdh transcripts in the IFE and the HF-associated dermalcells at P28. (F) Immunofluorescence of RANKL and keratin 14 on a P28 skinsection. Nuclei were colored with DAPI. (G) Immunofluorescent labeling forRANK in a P28 skin section with DAPI counterstaining. (H) Immunofluores-cent labeling for OPG with DAPI coloration in P28 anagen skin. (Scale bars,50 μm.) ORS, outer root sheath; B, bulge; ep, IFE.

Duheron et al. PNAS | March 29, 2011 | vol. 108 | no. 13 | 5345

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

recombinant RANKL. The overexpression of RANK in the SHGmay be particularly effective in inducing hair cycling in view ofthe importance of these cells in anagen entry (21, 30). In addi-tion, there was an increase in RANKL levels and a reduction inOPG. The underlying reasons for the changes in RANKL andOPG synthesis are currently unclear. One possibility may be thatthe overexpression of RANK in the HF triggers RANKLthrough a positive feedback loop. Such a loop, not uncommon inthe TNF family, may be mediated by NF-κB.Because the IFE basal-layer keratinocytes also express RANK,

we studied the IFE for changes in Rankl-KO and Rank-Tg ani-mals. We found that, in the Rankl−/− mice, epidermal thicknessand cell growth were reduced, whereas the epidermis was thick-ened and cell proliferation was greatly accelerated in the Rank-Tgmice. In neither model did we find a marked defect in epidermaldifferentiation or barrier function, demonstrating that RANK haslittle impact on this process. The extended keratin 14 expressionto suprabasal layers in the Tg animals is likely owing to thehyperproliferation because suprabasal keratin 14 expression isalso seen in other mouse models of accelerated epidermal growth(39, 40). We could detect an increase in TNFα and IL-1β pro-duction from Tg skin; however, more keratinocytes were re-covered after a 15-d culture of primary cutaneous keratinocytesfrom Tg mice, which suggests a cell-autonomous enhanced ker-atinocyte proliferation. More work is needed to further clarify therole of these cytokines on keratinocyte activation.The underlying molecular mechanism for epithelial activation

of the epidermo-pilosebaceous unit by RANK is currently un-

clear. Because RANK is a potent NF-κB stimulator and becauseNF-κB has been implicated in the regulation of HF cycling (52)as well as in epidermal turnover (53), we had first postulated thatRANK triggers the NF-κB pathway in skin epithelial cells.However, we could not obtain any evidence for NF-κB activationeither in the HF or in the IFE. RANK has also been shown toengage the signaling cascades, implicating JNK, Id-2, Akt, andNFATc1 (54), but, again, we could not observe any evidence thatthese signaling flows were elicited. Moreover, mice deficient inTRAF6, a signaling module of the NF-κB and JNK pathway,display a phenotype similar to that of Eda-null mice but not ofRankl−/− animals (55). Therefore, the molecular signaling path-ways engaged by RANK in the HF and the IFE remain to beuncovered. Our findings, together with the cyclic renewal of theHF on one side and the continuous homeostasis of the IFE onthe other, provide a valuable model system to dissect the sig-naling pathways triggered by RANK in epithelial (stem) cells.The herein revealed function of RANK in murine hair-cycle

control and epidermal homeostasis further underpins the sharedmolecular controls of two seemingly distant organs, skin andbone (18), and introduces the RANK–RANKL–OPG triad as akey player in the control of epithelial stem cell biology.

Materials and MethodsAnimals. PCR primers for genotyping are listed in Table S1.

Histology/Immunofluorescence. Primary Abs are listed in Table S2.

Fig. 4. Precocious anagen entry in Rank-Tg mice and in re-sponse to recombinant RANKL. (A) At P24, WT mice HFs are inanagen II/IIIa, whereas HFs of Rank-Tg mice are already inanagen IIIc. (Scale bars, 100 μm.) (B) At P56, WT mice HFs are intelogen, whereas HFs of Rank-Tg mice are in anagen IV/V.(Scale bars, 100 μm.) (C) Schematic representation of hair mor-phogenesis and the hair-cycle telogen-to-anagen transitions inRank-Tg and control mice. Red arrow points to P24 and bluearrow points to P56, corresponding to images in A and B. (D) At8 wk, WT mice received two s.c. injections of 100 μg of RANKL-GST or control GST with a 12-h interval. Then, 6 d later, theirback skin was analyzed for anagen. Graphic representationthat three of four treated mice precociously entered anagen, incontrast to control mice remaining in telogen. (E) Image of HFsin anagen of a RANKL-GST–treated mouse, whereas they are intelogen in the GST control mouse. (Scale bars, 100 μm; in Inset,50 μm.) The paraffin-embedded sections were stained withhematoxylin/eosin.

Fig. 5. RANK regulates epidermal homeostasis. (A) Differentepidermal thickness of skin of 4-wk-old Rankl−/−, WT, andRank-Tg mice, visualized on fixed, paraffin-embedded sec-tions stained with Masson’s trichrome. (Scale bars, 20 μm.) (B)Epithelial thickness was measured on 10 randomly chosenimages from four mice of each genotype per time point. Dataare mean ± SD. (C) Identification (arrowheads) of BrdU+

proliferating IFE cells on skin sections stained with eosin.(Scale bars, 20 μm.) (D) Graph shows the number of BrdU+ IFEkeratinocytes for each mouse. Data are mean ± SD. (E) Tgmice received anti-RANKL mAb or were mock-injected (con-trol) every 3 d from P10 to P45. Skins were analyzed for HFcycle and epidermal size on sections stained with hematoxy-lin/eosin. (Scale bars, 50 μm.) (F) Schematic representation ofthe HF stage of eight treated and seven control mice. (G)Epidermal thickness measures of eight treated and sevencontrol Tg mice. (Bars = mean values.) ep, IFE. *P < 0.05; **P <0.01; ***P < 0.001.

5346 | www.pnas.org/cgi/doi/10.1073/pnas.1013054108 Duheron et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0

RT-PCR. Classic and real-time PCR primers are listed in Table S3. Detailedexperimental protocols for cell proliferation, ELISA, flow cytometry, in situhybridization, RANK signaling assay, skin cytokine production, primarykeratinocyte culture, and epidermal barrier tests are provided in SI Materialsand Methods.

ACKNOWLEDGMENTS. We thank GrahamAnderson for Rank−/− skin samplesand Raphael Doineau and Benjamin Voisin for technical help. We appreciate

critical reading of the manuscript by Viljar Jaks, Maria Kasper, Ruth Schmidt-Ullirch, and Katja Fink. This work was supported by Centre National de laRecherche Scientifique (CNRS), University of Strasbourg, Société de Rechercheen Dermatologie (C.G.M.); a grant from Deutsche Forschungsgemeinschaft(to R.P.); a German Academic Exchange Service (DAAD) short-term stipend (toV.D.); and La Ligue Contre le Cancer (E.H.). V.D. and E.H. were recipients ofCNRS fellowships Bourses de Doctorat pour Ingénieur (CNRS-BDI) and Minis-tère de la Recherche et de la Technologie (CNRS-MRT), respectively.

1. Walsh MC, Choi Y (2003) Biology of the TRANCE axis. Cytokine Growth Factor Rev 14:251–263.

2. Tanaka S, Nakamura K, Takahasi N, Suda T (2005) Role of RANKL in physiological andpathological bone resorption and therapeutics targeting the RANKL–RANK signalingsystem. Immunol Rev 208:30–49.

3. Fata JE, et al. (2000) The osteoclast differentiation factor osteoprotegerin-ligand isessential for mammary gland development. Cell 103:41–50.

4. Beleut M, et al. (2010) Two distinct mechanisms underlie progesterone-inducedproliferation in the mammary gland. Proc Natl Acad Sci USA 107:2989–2994.

5. Akiyama T, et al. (2008) The tumor necrosis factor family receptors RANK and CD40cooperatively establish the thymic medullary microenvironment and self-tolerance.Immunity 29:423–437.

6. Knoop KA, et al. (2009) RANKL is necessary and sufficient to initiate developmentof antigen-sampling M cells in the intestinal epithelium. J Immunol 183:5738–5747.

7. Luo JL, et al. (2007) Nuclear cytokine-activated IKKα controls prostate cancermetastasis by repressing Maspin. Nature 446:690–694.

8. Schramek D, et al. (2010) Osteoclast differentiation factor RANKL controlsdevelopment of progestin-driven mammary cancer. Nature 468:98–102.

9. Gonzalez-Suarez E, et al. (2010) RANK ligand mediates progestin-induced mammaryepithelial proliferation and carcinogenesis. Nature 468:103–107.

10. Paus R, Cotsarelis G (1999) The biology of hair follicles. N Engl J Med 341:491–497.11. Cotsarelis G (2006) Epithelial stem cells: A folliculocentric view. J Invest Dermatol 126:

1459–1468.12. Fuchs E (2007) Scratching the surface of skin development. Nature 445:834–842.13. Blanpain C, Horsley V, Fuchs E (2007) Epithelial stem cells: Turning over new leaves.

Cell 128:445–458.14. Watt FM, Hogan BL (2000) Out of Eden: Stem cells and their niches. Science 287:

1427–1430.15. Morgan BA (2008) A glorious revolution in stem cell biology. Nat Genet 40:

1269–1270.16. Plikus MV, Widelitz RB, Maxson R, Chuong CM (2009) Analyses of regenerative wave

patterns in adult hair follicle populations reveal macro-environmental regulation ofstem cell activity. Int J Dev Biol 53:857–868.

17. Schneider MR, Schmidt-Ullrich R, Paus R (2009) The hair follicle as a dynamicminiorgan. Curr Biol 19:R132–R142.

18. Ross FP, Christiano AM (2006) Nothing but skin and bone. J Clin Invest 116:1140–1149.19. Kong YY, et al. (1999) OPGL is a key regulator of osteoclastogenesis, lymphocyte

development and lymph-node organogenesis. Nature 397:315–323.20. Morris RJ, et al. (2004) Capturing and profiling adult hair follicle stem cells. Nat

Biotechnol 22:411–417.21. Greco V, et al. (2009) A two-step mechanism for stem cell activation during hair

regeneration. Cell Stem Cell 4:155–169.22. Loser K, et al. (2006) Epidermal RANKL controls regulatory T-cell numbers via

activation of dendritic cells. Nat Med 12:1372–1379.23. Barbaroux JB, Beleut M, Brisken C, Mueller CG, Groves RW (2008) Epidermal receptor

activator of NF-κB ligand controls Langerhans cells numbers and proliferation.J Immunol 181:1103–1108.

24. Kartsogiannis V, et al. (1999) Localization of RANKL (receptor activator of NFκBligand) mRNA and protein in skeletal and extraskeletal tissues. Bone 25:525–534.

25. Paus R, et al. (1999) A comprehensive guide for the recognition and classification ofdistinct stages of hair follicle morphogenesis. J Invest Dermatol 113:523–532.

26. Müller-Röver S, et al. (2001) A comprehensive guide for the accurate classification ofmurine hair follicles in distinct hair cycle stages. J Invest Dermatol 117:3–15.

27. Bossen C, et al. (2006) Interactions of tumor necrosis factor (TNF) and TNF receptorfamily members in the mouse and human. J Biol Chem 281:13964–13971.

28. Wilson C, et al. (1994) Cells within the bulge region of mouse hair follicle transientlyproliferate during early anagen: Heterogeneity and functional differences of varioushair cycles. Differentiation 55:127–136.

29. Panteleyev AA, Jahoda CA, Christiano AM (2001) Hair follicle predetermination. J CellSci 114:3419–3431.

30. Jaks V, et al. (2008) Lgr5 marks cycling, yet long-lived, hair follicle stem cells. NatGenet 40:1291–1299.

31. Tumbar T, et al. (2004) Defining the epithelial stem cell niche in skin. Science 303:359–363.

32. Jahoda CA, Reynolds AJ (1996) Dermal-epidermal interactions. Adult follicle-derivedcell populations and hair growth. Dermatol Clin 14:573–583.

33. Handjiski BK, Eichmüller S, Hofmann U, Czarnetzki BM, Paus R (1994) Alkalinephosphatase activity and localization during the murine hair cycle. Br J Dermatol 131:303–310.

34. Collin-Osdoby P, et al. (2001) Receptor activator of NF-κB and osteoprotegerinexpression by human microvascular endothelial cells, regulation by inflammatorycytokines, and role in human osteoclastogenesis. J Biol Chem 276:20659–20672.

35. Kaufman CK, et al. (2003) GATA-3: An unexpected regulator of cell lineagedetermination in skin. Genes Dev 17:2108–2122.

36. Schmidt M, et al. (2001) Selective expression of calcium-binding proteins S100a8 andS100a9 at distinct sites of hair follicles. J Invest Dermatol 117:748–750.

37. Anderson DM, et al. (1997) A homologue of the TNF receptor and its ligand enhanceT-cell growth and dendritic-cell function. Nature 390:175–179.

38. Mecklenburg L, Tobin DJ, Cirlan MV, Craciun C, Paus R (2005) Premature terminationof hair follicle morphogenesis and accelerated hair follicle cycling in Iasi congenitalatrichia (fzica) mice points to fuzzy as a key element of hair cycle control. ExpDermatol 14:561–570.

39. Hobbs RM, Silva-Vargas V, Groves R, Watt FM (2004) Expression of activated MEK1in differentiating epidermal cells is sufficient to generate hyperproliferative andinflammatory skin lesions. J Invest Dermatol 123:503–515.

40. Pasparakis M, et al. (2002) TNF-mediated inflammatory skin disease in mice withepidermis-specific deletion of IKK2. Nature 417:861–866.

41. Mikkola ML (2008) TNF superfamily in skin appendage development. CytokineGrowth Factor Rev 19:219–230.

42. Schmidt-Ullrich R, et al. (2006) NF-κB transmits Eda A1/EdaR signalling to activate Shhand cyclin D1 expression, and controls post-initiation hair placode down growth.Development 133:1045–1057.

43. Pispa J, Pummila M, Barker PA, Thesleff I, Mikkola ML (2008) Edar and Troy signallingpathways act redundantly to regulate initiation of hair follicle development. HumMol Genet 17:3380–3391.

44. Cui CY, et al. (2006) Ectodysplasin regulates the lymphotoxin-β pathway for hairdifferentiation. Proc Natl Acad Sci USA 103:9142–9147.

45. Hashimoto T, Schlessinger D, Cui CY (2008) Troy binding to lymphotoxin-α activatesNFκB mediated transcription. Cell Cycle 7:106–111.

46. Srivastava S, et al. (2003) Receptor activator of NF-κB ligand induction via Jak2 andStat5a in mammary epithelial cells. J Biol Chem 278:46171–46178.

47. Randall VA (2007) Hormonal regulation of hair follicles exhibits a biological paradox.Semin Cell Dev Biol 18:274–285.

48. Schilli MB, Ray S, Paus R, Obi-Tabot E, Holick MF (1997) Control of hair growth withparathyroid hormone (7-34). J Invest Dermatol 108:928–932.

49. Foitzik K, et al. (2003) Prolactin and its receptor are expressed in murine hair follicleepithelium, show hair cycle-dependent expression, and induce catagen. Am J Pathol162:1611–1621.

50. Li M, et al. (2000) Skin abnormalities generated by temporally controlled RXRαmutations in mouse epidermis. Nature 407:633–636.

51. Demay MB, et al. (2007) Role of the vitamin D receptor in hair follicle biology. JSteroid Biochem Mol Biol 103:344–346.

52. Zhang Y, et al. (2009) Reciprocal requirements for EDA/EDAR/NF-κB and Wnt/β-catenin signaling pathways in hair follicle induction. Dev Cell 17:49–61.

53. Pasparakis M (2009) Regulation of tissue homeostasis by NF-κB signalling: Implicationsfor inflammatory diseases. Nat Rev Immunol 9:778–788.

54. Wada T, Nakashima T, Hiroshi N, Penninger JM (2006) RANKL–RANK signaling inosteoclastogenesis and bone disease. Trends Mol Med 12:17–25.

55. Naito A, et al. (2002) TRAF6-deficient mice display hypohidrotic ectodermal dysplasia.Proc Natl Acad Sci USA 99:8766–8771.

Duheron et al. PNAS | March 29, 2011 | vol. 108 | no. 13 | 5347

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 14

, 202

0