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 proteins 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 carcinogenesis. 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 gene regulation by GR (Beato et al., evaluated the global effect of GR on gene expression 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 genes (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 homeostasis 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 gene ontology 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-integrin (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).