Oncogene (2007) 26, 3060–3068 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE The tumor suppressor effect of the glucocorticoid receptor in skin is mediated via its effect on follicular epithelial stem cells

D Chebotaev1, A Yemelyanov1, L Zhu2, RM Lavker1 and I Budunova1

1Department of Dermatology, Feinberg Medical School, Northwestern University, Chicago, IL, USA and 2Bioinformatics Core, Northwestern University, Chicago, IL, USA

Glucocorticoids are potent inhibitors of mouse skin The cellular response to glucocorticoids is mediated tumorigenesis. The glucocorticoid control of cellular through the glucocorticoid receptor (GR), a well-known functions is mediated via the glucocorticoid receptor transcription factor (Beato et al., 1995). In non- (GR), a well-known transcription factor. Recently, we stimulated cells, GR resides in the cytoplasm bound generated transgenic mice overexpressing GR under to the complex of chaperone including heat control of the keratin5 (K5) promoter, and showed that shock proteins (Hsp90, Hsp70, Hsp40)and immuno- K5.GR animals are resistant to skin . philins such as Fkbp51 (Beato et al., 1995; Pratt and Follicular epithelial stem cells (SCs), located in the bulge Toft, 2003). Following hormone binding, the GR region of the hair follicle, are believed to be one of the dissociates from the chaperones and forms homodimers, target cells for skin carcinogenesis. We found that the which enter the nucleus. Chaperons including Fkbp51 number of putative hair follicle SC detected as label- play an important role in GR-mediated signaling: retaining cells was significantly less in the K5.GR they affect GR/hormone-binding affinity, and are transgenics compared to wild type (w.t.) littermates. We involved in GR shuttling between the nucleus and also showed that GR overexpression led to a reduction in cytoplasm (Davies et al., 2002). There are two major the clonogenicity of the follicular epithelial SCs. We mechanisms of regulation by GR (Beato et al., evaluated the global effect of GR on in a 1995; De Bosscher et al., 2003; Necela and Cidlowski, population of follicular SC-enriched bulge keratinocytes 2004). The direct positive transcriptional regulation of isolated by fluorescence activated cell sorting. We found (transactivation)requires binding of the GR that GR affected the expression of numerous bulge SC homodimer to glucocorticoid-response elements (GRE) ‘signature’ genes, genes involved in the maintenance of SC in gene promoters. The indirect negative regulation and progenitor cells of non-epidermal origin and proa- (transrepression)is mediated via negative cross-talk poptotic genes. Our findings underscore the important role with other transcription factors including AP-1, NF-kB of GR signaling in the of follicular epithelial and p53. SCs, and suggest that the reduction in their number may To study the effect of GR in skin tumorigenesis, we underlie the tumor suppressor effect of GR in the skin. generated keratin5 (K5).GR transgenic animals in which Oncogene (2007) 26, 3060–3068. doi:10.1038/sj.onc.1210108; K5 promoter drives GR expression to basal keratino- published online 4 December 2006 cytes in interfollicular epidermis and in the outer root sheath of hair follicles, as well as to some other stratified Keywords: epidermis; stem cells; glucocorticoid receptor; epithelia (Perez et al., 2001). In K5.GR transgenic skin carcinogenesis animals, GR has a nuclear localization and appears to be constitutively active (Perez et al., 2001; Budunova et al., 2003). We found that K5.GR transgenic animals are highly resistant to skin tumor development induced by activated v-Ha-ras oncogene followed by chronic Introduction treatment with tumor promoter (Budunova et al., 2003). In mice with GR overexpression, skin tumors developed Glucocorticoid hormones are potent inhibitors of later, and the average number of tumors per animal as keratinocyte proliferation and effective inhibitors of well as the average tumor volume were drastically experimentally induced skin carcinogenesis in animal decreased. models especially at the stage of skin tumor promotion As a self-renewing tissue, the epidermis is governed by (reviewed in Budunova et al., 2003). stem cells (SCs). Most of SCs in mouse epidermis reside in the bulge of hair follicles – a specialized region of the outer root sheath in the permanent segment of hair Correspondence: Dr I Budunova, Department of Dermatology, follicle (Cotsarelis et al., 1990, Taylor et al., 2000; Fuchs Feinberg Medical School, Northwestern University, Ward Building et al., 2001). These follicular epithelial SCs are similar to 9-332, 303 East Chicago Avenue, Chicago, IL 60611, USA. E-mail: [email protected] SCs in other adult tissues: they are slow-cycling and Received 24 July 2006; revised 22 September 2006; accepted 28 have a high potential for self-renewal (Lavker et al., September 2006; published online 4 December 2006 2003; Blanpain et al., 2004; Blanpain and Fuchs, 2006). Effect of GR on follicular epithelial stem cells D Chebotaev et al 3061 During times of proliferative need, these cells help to Results maintain the epidermis (Lavker and Sun, 2000; Lavker et al., 2003). GR reduces the frequency of slow-cycling cells in the bulge There are several lines of evidence indicating that skin of the hair follicles tumors including both papillomas and squamous cell One of the most reliable ways to identify epithelial SCs carcinomas (SCC)originate from the follicular epithelial takes advantage of the fact that these cells are normally SCs (Morris, 2000; Owens and Watt, 2003). Impor- slow-cycling, and hence can be identified experimentally tantly, the loss of carcinogen-treated interfollicular as label-retaining cells (LRCs)(Bickenbach and Mack- keratinocytes did not affect SCC development during enzie, 1984; Morris et al., 1985). To analyse the two-stage carcinogenesis, and the number of skin distribution of putative SCs detected as LRCs in hair papillomas was decreased only twofold, suggesting a follicles of K5.GR transgenic versus wild-type (w.t.) follicular origin of all malignant and a portion of benign mice, we injected newborn pups (post-natal day 3) skin tumors in mice (Morris, 2000). subcutaneously with 5-bromo-2-deoxyuridine (BrdU) We report herein that the tumor suppressor effect of for 3 days. This protocol resulted in B100% of the GR in the skin may be mediated through a decrease in keratinocytes incorporating BrdU 24 h following the last the number of target cells (follicular epithelial SCs), injection in both w.t. and K5.GR newborn skin (Figure changes in their proliferative capacity and alterations in 1a and b). Following a 4-week chase, most of the rapidly their genetic profile that may lead to the decreased cycling transient amplifying (TA)cells in the upper part sensitivity to transformation. of outer root sheath and in interfollicular epidermis

WT K5.GR ab

cd

Figure 1 GR reduced the number of LRC in the bulge of the hair follicle. Newborn K5.GR and w.t. mice were injected s.c. with BrdU for 3 days. BrdU-positive cells were identified by immunostaining. (a and b)– initial level of BrdU incorporation 24 h after the last BrdU injection (150 Â ). Note: almost all keratinocytes in both transgenic and w.t. animals are BrdU positive (brown nuclei). (c and d) BrdU-positive LRCs in the bulge of hair follicles after the 4-week chase period (300 Â ). Note: the number of LRCs is significantly reduced in hair follicles in K5.GR skin.

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3062 divided, diluted the BrdU label and underwent terminal colonies (X4 mm in diameter)per plate was 6.3 72.3 in differentiation. At the same time, most of the quiescent w.t. and 1.570.8 in K5.GR cell cultures. BrdU-stained cells (LRCs)resided in the bulge region of the hair follicles. The number of LRCs after the 4-week chase period was significantly less in the GR transgenic Effect of GR overexpression on transcriptional profile of follicular epithelial stem cells bulge region when compared with the littermate controls: LRC represented 12.378.5% of bulge kerati- To investigate the global effect of the GR on gene nocytes in K5.GR transgenics versus 35713.3% in w.t. expression in follicular epithelial SCs, we analysed bulge keratinocytes using cDNA arrays. We performed three controls (Po0.05)(Figure 1c and d).A similar result was noted following the 8-week chase period (data not separate gene profiling experiments, and using 1.5-fold shown). or greater change and a P-value of 0.05 as criteria for gene selection identified more than 200 genes whose expression was affected by GR in all comparisons of GR reduces the clonogenicity of SC-enriched bulge follicular epithelial SCs. Following the elimination of keratinocytes unknown genes, we selected B150 genes for further Under normal conditions, SCs are slow-cycling, quies- analyses. The results of analysis for cent cells; however, culturing SCs stimulate their biological processes are presented in Supplementary proliferation. To assess whether GR overexpression Figure 2. The complete list of genes differentially affected the proliferative potential of SCs, we analysed affected by GR in SC is presented in Supplementary their behavior in culture. We isolated a follicular Tables 1 and 2. epithelial SC-rich population of bulge keratinocytes from transgenic and w.t. mice in the telogen stage of the Gene array validation hair cycle using fluorescence activated cell sorting To validate the gene array results, we selected four analysis with the recently published surface markers upregulated (including GR itself)and five downregu- CD34 and a6- (Trempus et al., 2003; Blanpain lated genes, and measured the change in their expression et al., 2004; Tumbar et al., 2004)(see Supplementary by quantitative real-time PCR (Q-RT-PCR). For Figure 1). The purity of CD34 þ /a6-integrin þ sorted B validation we used the same RNA samples that were keratinocytes was 90% as confirmed by post-sort used for the array analyses, and eight additional RNA FACS analyses and cytospin immunostaining. It is samples isolated from w.t. and K5.GR transgenic worthy to mention that the number of CD34 þ / animals. The differential expression of eight out of nine a6-integrin þ follicular SC-enriched bulge keratinocytes genes studied was confirmed by Q-RT-PCR (Table 1). (calculated as % to all sorted keratinocytes)was 1.7 fold Next, we analysed the global changes in gene higher in w.t. skin compared to K5.GR transgenics expression in w.t. a6-integrin þ /CD34 þ bulge kerati- (data not shown). This result confirmed our finding on nocytes versus w.t. a6-integrin þ /CD34À basal kerati- the decreased number of putative follicular SCs in skin nocytes isolated from the same animals. We compared of K5.GR transgenics. Further, using a colony-forming assay, we showed that the clonogenicity of the CD34 þ /a6-integrin þ SC- enriched keratinocytes isolated from K5.GR transgenic Table 1 Validation of follicular epithelial SC gene array by Q-PCR animals was decreased by 75% in comparison to w.t. Gene symbol cDNA array fold Q-PCR fold Q-PCR fold keratinocytes (Figure 2). Indeed, the number of large of change of changea of changeb Upregulated FKBP5 11 8.5 10 Tnfrsf21 3.5 4 3 a W.T.b K5.GR Cyp27 4.5 5 4 Downregulated Igfbp3 3.5 5 5 Igfbp5 4 5 4 Ankrd6 3 3.5 3 S100A4 2.5 4 7.5 Ch25h 3 1.3 1.5 GRc 2.5 3 8

RNA extracted from bulge keratinocytes enriched by follicular SCs isolated from K5.GR and w.t. mice was used for cDNA array. The change in expression induced by GR was validated for selected genes by Q-PCR using the same as in DNA array as well as different RNA samples. Abbreviations: SC, stem cells; Q-PCR, quantitative PCR. Figure 2 GR reduced the clonogenicity of SC-enriched bulge aAnalysis of RNA extracted from three K5.GR and three w.t. SC keratinocytes. Keratinocytes were harvested from 7–8 week-old isolations; the same samples were used for cDNA array. bAnalysis of animals and sorted by FACS under sterile conditions. CD34 þ / RNA extracted from four K5.GR and four w.t. additional SC a6-integrin þ keratinocytes (1000 cells/60 mm dish)were grown for isolations. cGR probes on Affymetrix DNA array are recognized by 2–5 wks on 3T3 fibroblast feeder, fixed and stained with 1% both endogenous mouse and transgenic rat GR; primers and Taqman Rhodamin B to visualize colony growth. Representative dishes probe for Q-PCR were designed to detect only transgenic GR with w.t. (a)and K5.GR ( b)keratinocyte colonies. expression.

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3063 putative follicular SC ‘signature genes’ in our array (not Bulge SC signature genes shown here)with recently published databases on the Examination of SC DNA array revealed that GR expression profiles of follicular SCs (Morris et al., 2004; affected the expression of at least 13 genes (Table 2) Tumbar et al., 2004). A high correlation was noted from the previously published follicular SC gene between our data and previously published reports. For databases (Morris et al., 2004; Tumbar et al., 2004). example, among genes that were reported to be Interestingly, GR signaling resulted in a lowered upregulated in preparations enriched in follicular expression for many of the ‘bulge-signature’ genes epithelial SC, B80% (40/55 (Tumbar et al., 2004); 21/ including TGF-b2, IGFbp5, S100A4, Pdlim3, Sardh, 25 (Morris et al., 2004)) were found on our SC/basal Pfn2 and Tnfrsf11b (Table 2)whose expression is keratinocyte array. elevated in bulge SCs in mice (Morris et al., 2004; Tumbar et al., 2004). In contrast, the expression of Ly6C, a gene whose expression is specifically inhibited in Special groups of genes affected by GR in follicular mouse bulge SCs was increased by GR. The changes epithelial SCs in expression of IGFbp5 and S100A4 were confirmed by The largest groups influenced by the GR in follicular Q-RT-PCR (Table 1). epithelial SCs were genes encoding and nucleic acid metabolism, signal transduction, , adhesion and apoptosis (Supplementary Figure 2). In Genes involved in maintenance of stem and progenitor addition, analysis of the DNA arrays revealed that GR cells of non-epidermal origin induced profound changes in the expression of genes In addition to the genes specific for follicular epithelial involved in SC and progenitor cell maintenance. The SCs, GR affected the expression of at least nine genes latter changes induced by GR in bulge SCs are unique, important for the maintenance of stem and progenitor and have not been identified in a basal keratinocyte cells of non-epidermal origin. They include genes array (data not shown), thus we will specifically focus on encoding transcription factors c-Myb, Sox6, Olig1 and these results below. Zbtb16; RNA-binding protein Elav2; Cxcl12

Table 2 Selected groups of genes affected by GR in follicular epithelial SCs Gene name/protein name Fold change P value

Follicular epithelial SC genes Igfbp5/-like binding protein 5a À4.2 0.025 Tgfb2a À2.5 0.0007 S100A4 (S100 calcium binding protein A4)a À2.5 0.02 Sardh (nSarcosine dehydrogenase)a À1.6 0.012 Pdlim3 (PDZ and LIM domain 3)a À1.6 0.05 Pfn2 (profilin 2)a À1.5 0.038 Rasa3 (RAS p21 protein activator 3) À1.5 0.006 Tgfb1i4 (transforming growth factor beta 1 induced transcript 4) À1.5 0.008 Tnfrsf11b (osteoprotegerin)a À1.5 0.0008 Ndr2 (N-myc downstream regulated gene 2)+1.7 0.003 Enpp1(ectonucleotide-pyrophosphatase/phosphodiesterase 1)+2.0 0.026 Sparc/osteonectin +2.3 0.0014 Ly6 (lymphocyte antigen 6 complex, locus C)a +2.3 0.043 Genes involved in proliferation, differentiation and maintenance of non-epidermal SC/progenitors Elav2 À4.5 0.004 Gdnf (glial cell line-derived neurotrophic factor) À2.6 0.022 Sox6 (SRY-box containing transcription factor 6) À2.3 0.001 Cxcl12 (stromal cell-derived factor 1) À2.2 0.015 Scgf (Stem cell growth factor) À1.7 0.0012 Olig1 (oligodendrocyte transcription factor 1)+1.6 0.04 Scrg1 (scrappy response gene 1) +2.0 0.0026 Zbtb16 (PLZF, promyelocytic leukemia zinc finger protein)+2.0 0.0014 c-Myb +2.1 0.031 Genes involved in apoptosis and terminal differentiation Tnfrsf11b À1.5 0.0008 Bcl2-like 11 +1.5 0.009 Filaggrin +1.5 0.01 Nrd2 (N-myc downstream regulated gene 2)+1.7 0.003 Gulp1 +1.8 0.009 Tnfrs21 +3.1 0.004

RNA extracted from bulge keratinocytes enriched by follicular SCs isolated from K5.GR and w.t. mice was used for cDNA array. Abbreviations: GR, glucocorticoid receptor; SCs, stem cells. aFollicular epithelial SC genes whose expression was affected by GR against the published gradient (Tumbar et al., 2004; Morris et al., 2004). A complete list of genes affected by GR with annotations and Affymetrix probe numbers is presented in Supplementary Tables 1 and 2 online.

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3064 and Scgf; and others (Table 2). Those genes control K5.GR w.t. proliferation, differentiation and self-renewal of dif- 12 3 45 6 positive positive control ferent SC/progenitors including embryonic SCs, pluri- kDa potent mesenchimal cells, progenitors of hematopoietic NS 57 FKBP51 cells, chondrocytes, neurons and gonadal SCs (Ward 46 et al., 2000; Ito et al., 2003; Costoya et al., 2004; Smith et al and Sigvardsson, 2004; Dumitriu ., 2006; Ligon et al., 2006). Figure 3 Effect of GR on the expression of chaperone protein Genes related to apoptosis and terminal differentiation Fkbp51. The whole-cell protein extracts were prepared from a6 of keratinocytes integrin þ keratinocytes isolated by FACS from the skin of three Glucocorticoids induce differentiation of many different w.t. and three K5.GR individual animals, and used for Western cell types; they also induce apoptosis in lymphocytes blot analysis of Fkbp51 expression (lanes 1–6). Protein extracts from 293 FT cells transiently transfected with Fkbp51 were used as (Herold et al., 2006). We found at least four apoptosis- a positive control. Probing with anti-a-actin Ab was used to control related genes on SC DNA array. Three pro-apoptotic protein loading. NS – non-specific band. Note: Fkbp51 protein is genes: Bcl2l11, Tnfrsf21 and Gulp (Daniel et al., 2001; Su expressed only in K5.GR keratinocytes. et al., 2002; Lu et al., 2006)were upregulated by GR. Overexpression of Tnfrs21 was confirmed by Q-RT- PCR (Table 1). We also found that the expression of the intermediate/late marker of keratinocyte differentiation, Discussion filaggrin (Blanpain and Fuchs, 2006)was increased in K5.GR bulge keratinocytes (Table 2). Terminal deox- Our work and those of others showed that GR signaling ynucleotidyltransferase-mediated UTP end labeling is important for the growth control of lymphoid and (TUNEL)immunostaining has not revealed a statisti- epithelial cells. Furthermore, in some tissues including cally significant increase of apoptosis in the bulge area in skin, glucocorticoids act as potent tumor suppressors K5.GR transgenics (data not shown). This may suggest (Schwartz et al., 1977; Verma et al., 1983; Yao et al., that GR signaling did not directly induce apoptosis in 2004). We found that the expression of GR is decreased follicular SCs, but decreased their survival potential. in experimentally induced skin tumors in SENCAR mice especially during the early stages of skin carcino- Increased expression of chaperone protein Fkbp51: a genesis (Budunova et al., 1997). To test the hypothesis potential mechanism of follicular epithelial SC that GR acts as a tumor suppressor in skin, we protection against GR generated K5.GR transgenic animals (Perez et al., An important group of genes whose expression was 2001). We found that K5.GR transgenic mice are indeed affected by GR in follicular epithelial SCs are genes highly resistant to chemical skin carcinogenesis: they encoding chaperone proteins and regulators of GR developed significantly fewer and smaller skin papillo- folding such as immunophilins Fkbp51, Fkbp10, mas than control animals (Budunova et al., 2003). As it Fkbp14 and heat shock protein hsp40 (dnaj6)(Supple- is well established that the follicular epithelial SCs mentary Tables 1 and 2). located in the bulge region of the hair follicle are one of FK506-binding protein Fkbp51 is the highest of the the primary targets for skin carcinogenesis, we have upregulated follicular SC genes: its expression was begun to dissect the effects that the GR has on these increased by B10-fold (Table 1). Fkbp51 encodes a cells. chaperone protein that is a part of GR-heat shock One of the major outcomes of GR overexpression was protein complex (Davies et al., 2005), and is known to the decreased number of putative follicular epithelial inhibit GR effect on gene expression in different cells via SCs (LRCs)in the bulge of K5.GR animals. The sequestration of GR in cytoplasm (Davies et al., 2005). detailed mechanisms behind this effect of GR remain To verify that Fkbp51 protein was increased in to be elucidated. However, it is clear that GR decreased K5.GR bulge keratinocytes, we used Western blot the proliferative potential of follicular epithelial SCs as analysis. As the amount of protein that we were able assessed by their clonogenic activity in vitro. The to extract from bulge keratinocytes was very limited, we expression profiling of bulge keratinocytes isolated by compared Fkbp51 expression in a6-integrin þ keratino- FACS suggested that GR may reduce the survival cytes (basal keratinocytes from the interfollicular potential of the bulge cells: among the genes upregulated epidermis and bulge)isolated by FACS from K5.GR by GR were genes directly related to apoptosis such as and w.t. animals. Fkbp51 protein was not expressed in Bcl2l11 (BCL2-like 11, also called Bim)and Tnfrsf21 a6-integrin þ cells from w.t. mice, but was significantly (death receptor 6). Bim is a key regulator of apoptosis in expressed in a6-integrin þ cells isolated from K5.GR different cells and tissues (Lu et al., 2006). Tnfrsf21 mice (Figure 3). This difference in Fkbp51 expression belongs to the TNF receptor subgroup called death could be only attributed to the input from the bulge receptors, and can induce apoptosis in different cells via K5.GR keratinocytes as DNA array and Q-RT-PCR its death domain (Daniel et al., 2001). did not show any difference in Fkbp51 mRNA expres- In addition, GR affected 13 genes that have been sion in basal interfollicular keratinocytes in K5.GR and categorized as ‘follicular SC signature genes’. This w.t. animals (data not shown). codification is based on the comparison of genes

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3065 expressed in basal keratinocytes and in enriched factor. One of the noteworthy differences was the strong populations of follicular epithelial SCs (Morris et al., induction of chaperone protein Fkbp51 by GR specifi- 2004; Tumbar et al., 2004, Blanpain et al., 2004). cally in SCs (Figure 3). Fkbp51 is a direct GR target Unexpectedly, the major mode of GR effect on the gene (Hubler and Scammell, 2004). This chaperone follicular SC ‘signature genes’ was against the gradient protein is known to inhibit nuclear import of GR and of their expression in the bulge. Many genes that other steroid receptors, and to reduce GR activity in are overexpressed in the bulge (TGF-b2, Igfbp5, S100A4 cells (Davies et al., 2005). Thus, the induction of Fkbp51 and others, Table 2)were downregulated by GR. In may represent the important mechanism (through the contrast, Ly6C, whose expression is low in bulge negative feedback loop)of self-protection of SCs against keratinocytes, was upregulated by GR. This suggests the consequences of chronically activated GR signaling. that the GR might diminish the ‘stemness’ of follicular Interestingly, the Fkbp51-based protection mechanism epithelial SCs. was specific only for the bulge keratinocytes: we did not Interestingly, some of the genes affected by GR: find the Fkbp51 induction by GR in basal keratinocytes TGF-b2, TGF-b3, Igfbp3, Igfbp5, Fzd1 and Ankrd6 (data not shown). (Supplementary Tables 1 and 2)are involved in TGF- b-, In summary, our work indicates that follicular IGF- and Wnt-dependent signaling pathways that are epithelial SCs are highly sensitive to GR-mediated particularly important for epidermal SCs. On the signaling. We found that not only the number of contrary, GR signaling did not affect the components putative follicular SCs was decreased, but also their of BMP signaling pathway that is also implicated in the proliferative potential was markedly abrogated in maintenance of follicular epithelial SCs (Blanpain et al., K5.GR skin. In addition, GR induced profound 2004; Blanpain and Fuchs, 2006). changes in the expression of genes involved in SC The cross-talk between the GR and either IGF or maintenance. This suggests a novel mechanism for the TGF-b signaling is known (Li et al., 2003; Li et al., tumor suppressive effect of GR in skin that is mediated 2005), while the effect of glucocorticoids on Wnt via a reduction in the follicular epithelial SCs, which pathway has not been well studied. However, recent lessens the potential targets for skin carcinogenesis. publications strongly suggest that Wnt pathway is clinically and developmentally an important target for the effect of GR/glucocorticoids (Takayama et al., Materials and methods 2006). Of relevance to this latter point, one of the significantly downregulated genes (expression is de- Animals creased by Bthreefold, Supplementary Table 1)is K5.GR hemizygous mice, line 285 in B6D2 (F1 C57Bl  DBA) Ankrd6, a homolog of Diego in Drosophila. Ankrd6 is genetic background have been previously described (Perez an essential component of the Wnt-signaling pathway et al., 2001). B6D2 females for breeding were obtained from that acts as a molecular switch, suppressing the The Jackson Laboratory (Bar Harbor, ME, USA). All animal canonical Wnt/beta-catenin pathway and stimulating experiments were performed in compliance with ACUC Wnt/planar cell polarity signaling (Schwarz-Romond protocols approved by the Northwestern University Animal et al., 2002). Care and Ethics Committee. In addition to the aforementioned ‘signature’ follicu- lar SC genes, GR affected the expression of numerous Detection of slow-cycling cells (LRC) genes that control proliferation, differentiation and the LRCs were detected as published previously (Taylor et al., self-renewal of non-epithelial SCs and progenitor cells: 2000). Newborn K5.GR and w.t. pups (post-natal day 3–5) embryonic SCs, hematopoietic progenitors, progenitors were injected subcutaneously (s.c.)with BrdU (50 mg/g body of neurons and others (Table 2). This finding is weight)twice daily for 3 days. Mice were killed 24 h, 4 and 8 consistent with the known role of glucocorticoids in weeks after injections, and skin samples were used for BrdU immunostaining. Quantitative analysis of LRCs is described in the regulation of SC/progenitor cells that depends on Supplementary material no. 1. the cell context. For example, glucocorticoids are key regulators of stress-induced erythroid progenitor self- renewal (Bauer et al., 1999). Conversely, glucocorticoids Isolation of keratinocyte subpopulations by FACS a6-Integrin þ /CD34 þ SC-enriched bulge keratinocytes and are required for the chondrogenic differentiation of þ À human mesenchymal SCs (Derfoul et al., 2006). Even a6-integrin /CD34 basal keratinocytes from the dorsal skin of K5.GR. and w.t. littermates at telogen stage of hair cycle (7–8 though the role of such genes in the maintenance of weeks of age)were isolated by FACS analysis as previously epidermal SCs has not been studied, one might expect published (Trempus et al., 2003; Morris et al., 2004). For that, owing to the commonalities in SC nature, some of detailed description, see Supplementary material no. 2. those genes should be important for follicular epithelial SC maintenance. Interestingly, GR ‘signature’ genes were noticeably Colony-forming assay Colony-forming assay was performed as was described different in follicular epithelial SCs and basal keratino- B (Trempus et al., 2003; Morris et al., 2004). Bulge keratinocytes cytes with only 5% overlap between SC and basal isolated by FACS were plated onto the monolayer of keratinocyte DNA arrays (basal cell array is not shown irradiated Swiss 3T3 fibroblasts (1000 keratinocytes/60 mm here). This finding suggests that unique SC micro- dish)and grown at 32 1C in William’s E medium (Gibco, environment may modify GR function as a transcription Rockville, MD, USA)containing 20% fetal bovine serum

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3066 (Hyclone, Logan, UT, USA)for 2.5 weeks. See also above. a6-Integrin þ keratinocytes consisted of basal bulge Supplementary material no. 3. and interfollicular keratinocytes were isolated by FACS. The whole-cell protein extracts were prepared using RIPA buffer as Microarray analysis of GR effect on gene expression described previously (Yemelyanov et al., 2006), resolved by For the microarray analyses, we isolated a6-integrin þ / SDS–PAGE on 10% gels and transferred to nitrocellulose CD34 þ bulge keratinocytes and a6-integrin þ /CD34- basal membranes (BioRad, Hercules, CA, USA). Membranes were keratinocytes by FACS from individual K5.GR. animals and blocked with 5% Blotto in TBS, and incubated with anti- w.t. littermates at the age of 7–8 weeks. RNA amplification, Fkbp51 polyclonal rabbit antibody (kindly provided by Dr D labeling, and hybridization were performed at the Microarray Smith, Mayo Clinic, Scottsdale, AZ, USA)overnight at 4 1C, & Genotyping core at the Center for Genetic Medicine at followed by peroxidase-conjugated anti-rabbit IgG secondary Northwestern University according to the DNA chip manu- antibodies (Cell Signaling Technology, Beverly, MA, USA). facturer’s protocols (Affymetrix, Santa Clara, CA, USA). ECL reagent (Amersham Pharmacia Biotech, Sweden)was Data analysis was performed at the Bioinformatics Core at used for the band visualization. Protein extracts from 293 FT Robert H. Lurie Cancer Center at Northwestern University. cells transiently transfected with Fkbp51 were used as a For additional information see Supplementary material no. 4. positive control. To verify equal loading and adequate transfer, the membrane was probed with anti-a-actin anti- Immunostaining bodies (Santa Cruz Biotechnology, Pasadena, CA, USA). Formalin-fixed mouse skin samples were used for BrdU staining. After Ag retrieval (5 min at 20–25 psi in citric buffer, Statistical analysis pH 6.0), the tissues were blocked with 10% horse serum in All experiments were repeated several times as indicated. Mean PBS, incubated with mouse monoclonal anti-BrdU antibody and standard error values in experiments with LRCs and (BD Biosciences, San Jose, CA, USA)followed by biotinylated colony-forming assay were calculated using Microsoft Excel secondary anti-mouse IgG from Mouse-on-mouse and software and compared using paired Student’s t-test. Statistical ABC reagent from Standard VECTASTAIN Elite ABC kit analysis of gene arrays is described in Supplementary material (Vector Laboratories, Burlingame, CA, USA). Immunostain- no. 4. ing was visualized with DAB chromogen (Vector Laboratories, Burlingame, CA, USA), and tissues were counterstained with Gill’s hematoxylin (Vector Laboratories, Burlingame, CA, USA). CD34 and a6 integrin immunostaining is described in Abbreviations Supplementary material no. 5. BrdU, 5-bromo-2-deoxyuridine; GR, glucocorticoid receptor; Quantitative real-time PCR GRE, glucocorticoid-response element; HSP, heat shock TaqMan real-time quantitative PCR to validate the gene array protein; K5, keratin5; LRCs, label-retaining cells; Q-PCR, results was performed using ABI Prism 7900 (Applied quantitative PCR; SC, stem cells; SCC, squamous cell Biosystems, Foster City, CA, USA). Reverse transcription carcinoma; s.c., subcutaneous; TA cells, transient amplifying þ þ and amplification of total RNA (1–3 ng)from a6 CD34 cells; w.t., wild type. keratinocytes was accomplished in a one-step format with TaqMan One-Step RT-PCR Master Mix Reagents Kit (Applied Biosystems, Foster City, CA, USA)and pre- Acknowledgements designed/pre-optimized gene-specific TaqMan primers and probes from TaqMan Gene Expression Assays Kits with We thank Dr C Trempus (National Institute of Environmental mouse GAPDH as endogenous control (Applied Biosystems, Health Sciences, Research Triangle Park, NC, USA)and Dr N Foster City, CA, USA). Jafari (Microarray & Genotyping cores, Center for Genetic Medicine, Northwestern University, Chicago, IL, USA)for Western blot analysis technical support. We are thankful to Dr D Smith (Mayo Keratinocytes from the dorsal skin of K5.GR. and w.t. Clinic, Scottsdale, AZ, USA)for his generous gift of anti- littermates at 7–8 weeks of age were harvested as described Fkbp51 Ab.

References

Bauer A, Tronche F, Wessely O, Kellendonk C, Blanpain C, Fuchs E. (2006). Epidermal Stem Cells of the Reichardt HM, Steinlein P et al. (1999). The glucocorticoid Skin. Annu Rev Cell Dev Biol 22: 339–373. receptor is required for stress erythropoiesis. Genes Dev 13: Budunova IV, Kang H, Carbajal S, Viaje A, Slaga TJ. (1997). 2996–3002. Altered glucocorticoid receptor expression and function Beato M, Herrlich P, Schultz G. (1995). during mouse skin carcinogenesis. Mol Carcinogen 18: receptors: many actors in search of a plot. Cell 83: 177–185. 851–857. Budunova IV, Kowalczyk D, Perez P, Yao YJ, Jorcano JL, Bickenbach JR, Mackenzie IC. (1984). Identification and Slaga TJ. (2003). Glucocorticoid receptor functions as a localization of label-retaining cells in hamster epithelia. potent suppressor of mouse skin carcinogenesis. Oncogene J Invest Dermatol 82: 618–622. 22: 3279–3287. Blanpain C, Lowry WE, Geoghegan A, Polak L, Fuchs E. Costoya JA, Hobbs RM, Barna M, Cattoretti G, Manova K, (2004). Self-renewal, multipotency, and the existence of two Sukhwani M et al. (2004). Essential role of Plzf in cell populations within an epithelial stem cell niche. Cell 118: maintenance of spermatogonial stem cells. Nat Genet 36: 635–648. 653–659.

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3067 Cotsarelis G, Sun TT, Lavker RM. (1990). Label-retaining glucocorticoid-induced apoptosis in lymphoblastic leukemia cells reside in the bulge area of pilosebaceous unit: cells. FEBS Lett 580: 3539–3544. implications for follicular stem cells, hair cycle, and skin Morris RJ, Fischer SM, Slaga TJ. (1985). Evidence that the carcinogenesis. Cell 61: 1329–1337. centrally and peripherally located cells in the murine Daniel PT, Wieder T, Sturm I, Schulze-Osthoff K. (2001). epidermal proliferative unit are two distinct cell populations. The kiss of death: promises and failures of death J Invest Dermatol 84: 277–281. receptors and ligands in cancer therapy. Leukemia 15: Morris RJ, Liu Y, Marles L, Yang Z, Trempus C, Li S et al. 1022–1032. (2004). Capturing and profiling adult hair follicle stem cells. Davies TH, Ning YM, Sanchez ER. (2002). A new first step in Nat Biotechnol 22: 411–417. activation of steroid receptors: hormone-induced switching Morris RJ. (2000). Keratinocyte stem cells: targets for of FKBP51 and FKBP52 immunophilins. J Biol Chem 277: cutaneous carcinogens. J Clin Invest 106: 3–8. 4597–4600. Necela BM, Cidlowski JA. (2004). Mechanisms of glucocorti- Davies TH, Ning YM, Sanchez ER. (2005). Differential coid receptor action in noninflammatory and inflammatory control of glucocorticoid receptor hormone-binding cells. Proc Am Thorac Soc 1: 239–246. function by tetratricopeptide repeat (TPR)proteins and Owens DM, Watt FM. (2003). Contribution of stem cells and the immunosuppressive ligand FK506. Biochemistry 44: differentiated cells to epidermal tumours. Nat Rev Cancer 3: 2030–2038. 444–451. De Bosscher K, Vanden Berghe W, Haegeman G. (2003). The Perez P, Page A, Bravo A, Del Rio M, Gimenez-Conti I, interplay between the glucocorticoid receptor and nuclear Budunova I et al. (2001). Altered skin development and factor-kappaB or activator protein-1: molecular mechanisms impaired proliferative and inflammatory responses in for gene repression. Endocr Rev 24: 488–522. transgenic mice overexpressing the glucocorticoid receptor. Derfoul A, Perkins GL, Hall DJ, Tuan RS. (2006). Gluco- FASEB J 15: 2030–2032. corticoids promote chondrogenic differentiation of adult Pratt WB, Toft DO. (2003). Regulation of signaling human mesenchymal stem cells by enhancing expression protein function and trafficking by the hsp90/hsp70-based of cartilage extracellular matrix genes. Stem Cells 24: chaperone machinery. Exp Biol Med (Maywood) 228: 1487–1495. 111–133. Dumitriu B, Patrick MR, Petschek JP, Cherukuri S, Schwartz JA, Viaje A, Slaga TJ, Yuspa SH, Hennings H, Klingmuller U, Fox PL et al. (2006). Sox6 cell-auto- Lihta U. (1977). Fluocinolone acetonide: a potent inhibitor nomously stimulates erythroid cell survival, proliferation, of mouse skin tumor promotion and epidermal DNA and terminal maturation and is thereby an important synthesis. Chem-Biol Interaction 17: 331–347. enhancer of definitive erythropoiesis during mouse develop- Schwarz-Romond T, Asbrand C, Bakkers J, Kuhl M, ment. Blood 108: 1198–1207. Schaeffer HJ, Huelsken J et al. (2002). The ankyrin Fuchs E, Merrill BJ, Jamora C, DasGupta R. (2001). At the repeat protein Diversin recruits Casein kinase Iepsilon roots of a never-ending cycle. Dev Cell 1: 13–25. to the beta-catenin degradation complex and acts in Herold MJ, McPherson KG, Reichardt HM. (2006). Gluco- both canonical Wnt and Wnt/JNK signaling. Genes Dev 16: corticoids in T cell apoptosis and function. Cell Mol Life Sci 2073–2084. 63: 60–72. Smith E, Sigvardsson M. (2004). The roles of transcription Hubler TR, Scammell JG. (2004). Intronic hormone factors in B lymphocyte commitment, development, and response elements mediate regulation of FKBP5 by transformation. J Leukoc Biol 75: 973–981. progestins and glucocorticoids. Cell Stress Chaperones 9: Su HP, Nakada-Tsukui K, Tosello-Trampont AC, Li Y, 243–252. Bu G, Henson PM et al. (2002). Interaction of CED-6/ Ito C, Sato H, Ando K, Watanabe S, Yoshiba F, Kishi K et al. GULP, an adapter protein involved in engulfment (2003). Serum stem cell growth factor for monitoring of apoptotic cells with CED-1 and CD91/low density hematopoietic recovery following stem cell transplantation. lipoprotein receptor-related protein (LRP). J Biol Chem Bone Marrow Transplant 32: 391–398. 277: 11772–11779. Lavker RM, Sun TT, Oshima H, Barrandon Y, Akiyama M, Takayama S, Rogatsky I, Schwarcz LE, Darimont BD. (2006). Ferraris C et al. (2003). Hair follicle stem cells. J Investig The Glucocorticoid receptor represses cyclin D1 by Dermatol Symp Proc 8: 28–38. targeting the Tcf-beta-catenin complex. J Biol Chem 281: Lavker RM, Sun TT. (2000). Epidermal stem cells: properties, 17856–17863. markers, and location. Proc Natl Acad Sci USA 97: Taylor G, Lehrer MS, Jensen PJ, Sun TT, Lavker RM. 13473–13475. (2000). Involvement of follicular stem cells in forming Li BG, Hasselgren PO, Fang CH. (2005). Insulin-like growth not only the follicle but also the epidermis. Cell 102: factor-I inhibits -induced proteolysis in 451–461. cultured L6 myotubes through PI3K/Akt/GSK-3beta and Trempus CS, Morris RJ, Bortner CD, Cotsarelis G, Faircloth PI3K/Akt/mTOR-dependent mechanisms. Int J Biochem RS, Reece JM et al. (2003). Enrichment for living murine Cell Biol 37: 2207–2216. keratinocytes from the hair follicle bulge with the cell surface Li G, Wang S, Gelehrter TD. (2003). Identification of marker CD34. J Invest Dermatol 120: 501–511. glucocorticoid receptor domains involved in transrepression Tumbar T, Guasch G, Greco V, Blanpain C, Lowry WE, of transforming growth factor-beta action. J Biol Chem 278: Rendl M et al. (2004). Defining the epithelial stem cell niche 41779–41788. in skin. Science 303: 359–363. Ligon KL, Kesari S, Kitada M, Sun T, Arnett HA, Alberta JA Verma AK, Garcia CT, Ashendel CL, Boutwell RK. (1983). et al. (2006). Development of NG2 neural progenitor cells Inhibition of 7-bromomethylbenz[a]anthracene-promoted requires Olig gene function. Proc Natl Acad Sci USA 103: mouse skin tumor formation by retinoic acid and dexa- 7853–7858. methasone. Cancer Res 43: 3045–3049. Lu J, Quearry B, Harada H. (2006). p38-MAP kinase Ward AC, Loeb DM, Soede-Bobok AA, Touw IP, activation followed by BIM induction is essential for Friedman AD. (2000). Regulation of granulopoiesis by

Oncogene Effect of GR on follicular epithelial stem cells D Chebotaev et al 3068 transcription factors and signals. Leukemia 14: Yemelyanov A, Gasparian A, Lindholm P, Dang L, 973–990. Pierce JW, Kisseljov F et al. (2006). Effects of IKK inhibitor Yao R, Wang Y, Lemon WJ, Lubet RA, You M. (2004). PS1145 on NF-kappaB function, proliferation, apoptosis Budesonide exerts its chemopreventive efficacy during and invasion activity in prostate carcinoma cells. Oncogene mouse lung tumorigenesis by modulating gene expressions. 25: 387–398. Oncogene 23: 7746–7752.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene