The Retinoid-Related Orphan Receptor RORα Promotes Keratinocyte Differentiation via FOXN1

RORα is a retinoid-related orphan nuclear receptor that regulates inflammation, lipid metabolism, and cellular differentiation of several non-epithelial tissues. In spite of its high expression in skin epithelium, its functions in this tissue remain unclear. Using gain- and loss-of-function approaches to alter RORα gene expression in human keratinocytes (HKCs), we have found that this transcription factor functions as a regulator of epidermal differentiation. Among the 4 RORα isoforms, RORα4 is prominently expressed by keratinocytes in a manner that increases with differentiation. In contrast, RORα levels are significantly lower in skin squamous cell carcinoma tumors (SCCs) and cell lines. Increasing the levels of RORα4 in HKCs enhanced the expression of structural proteins associated with early and late differentiation, as well as genes involved in lipid barrier formation. Gene silencing of RORα impaired the ability of keratinocytes to differentiate in an in vivo epidermal cyst model. The pro-differentiation function of RORα is mediated at least in part by FOXN1, a well-known pro-differentiation transcription factor that we establish as a novel direct target of RORα in keratinocytes. Our results point to RORα as a novel node in the keratinocyte differentiation network and further suggest that the identification of RORα ligands may prove useful for treating skin disorders that are associated with abnormal keratinocyte differentiation, including cancer.


Introduction
The retinoid related orphan receptor RORa belongs to the steroid nuclear hormone receptor superfamily and functions as a transcription factor by binding as a monomer to the RORa responsive elements (ROREs) in the regulatory region of target genes [1,2,3]. In humans, there are four RORa isoforms (RORa 1-4), which differ only in their amino-terminal A/B domain and are generated by alternative promoter usage and exon splicing [1,4]. RORa is expressed in a temporal and spatial-dependent manner during embryonic development, and is critical for lipid metabolism [5,6,7], inflammation, and differentiation of Purkinje neuronal cells, adipocytes, lymphocytes, osteoblasts, as reviewed in [2,3].
RORa is widely expressed in normal epithelial tissues [8], including the epidermis [9,10], and is often down modulated in epithelium-derived tumors [8]. Yet, very little is known about the functions of RORa in epithelial cell biology. In mice, RORa transcripts are highly expressed in the skin, specifically in suprabasal epidermal cells, sebaceous glands, and hair follicles [9]. Mice with homozygous disruption of the RORa gene exhibit sparse pelage, with a slow rate of hair re-growth after shaving, pointing to a possible role of RORa in epidermal differentiation [9].
The epidermis is a stratified epithelial tissue structure composed of a single basal layer of proliferating keratinocytes that become progressively more differentiated as they migrate upwards and transit through the overlying spinous, granular, and cornified stratum layers [11]. The keratinocyte life cycle is associated with well-characterized molecular and biochemical changes. Keratinocytes at each stage express signature structural and signaling molecules [12]. For example, keratinocytes switch from expressing the keratin 5/14 pair in the basal layer to keratins 1/10, followed by involucrin, in the suprabasal layers. At later stages of differentiation in the granular layer, keratinocytes begin to express filaggrin and loricrin and produce lipid granules containing cholesterol, free fatty acids, and ceramides [13].
A well-controlled transcription network is critical for the balanced transition of keratinocytes from one stage to the next and the long-term maintenance of skin homeostasis [11,14]. The initial switch from basal to spinous differentiation is controlled by Notch and p63 signaling pathways, which interplay with each other and with other transcription factors, such as interferon regulatory factor 6 (IRF6), AP2, and Kruppel-like factor 4 (KLF4) [15,16,17,18,19,20] The early stage of differentiation is also regulated by FOXN1, a member of forkhead transcription factor family, via the interplay with the PKC and Akt signaling pathways [21,22].
Epidermal keratinocytes provide an excellent experimental system to study intrinsic growth/differentiation controls of epithelial cells. Well defined culture conditions make it possible to study primary keratinocytes of human and mouse origin in vitro [23]. Experimental systems, such as organotypic culture and grafting of cultured keratinocytes, recapitulate the complex cellcell interactions that occur in the epidermis in vivo [17,24,25]. Applying gain-and loss-of function approaches to the above experimental system, we show that RORa is a key integral element of the pro-differentiation network in epidermal keratinocytes, functioning as an upstream regulator of the FOXN1 gene and genes involved in epidermal lipid/barrier functions.

1) RORa4
Isoform Expression is up-regulated during Differentiation in Human Keratinocytes, and Downregulated in Keratinocyte-derived Skin Cancer To determine which RORa isoforms are expressed in keratinocytes, conventional RT-PCR analysis was performed using primers specific for the 4 individual isoforms. We found that RORa4 was the predominant isoform present in human primary keratinocytes (HKCs) cultured under growing conditions, and upon induction of differentiation by high density or suspension cultures [28] (Fig. 1A). While RORa1 and RORa3 were detectable in none of these conditions, low expression of RORa2 mRNA was detected only in cells induced to differentiate by 48-hour culture in suspension. Real time qRT-PCR analysis using a set of primers common for all RORa isoforms, as well as one specific for RORa4, showed a similar induction of RORa mRNA expression upon differentiation (Fig. 1B).
The RORa2 and RORa4 proteins have different size and electrophoretic mobility. The two proteins were clearly distinguished by immunoblot analysis of HKC extracts infected with retroviruses expressing RORa2 or RORa4 cDNAs. Consistent with the mRNA results, endogenous RORa2 protein was barely detectable, while RORa4 expression was most prominent and increased with the differentiation marker keratin 1 (Fig. 1C). Immunofluorescence staining of intact human skin showed that RORa was present in both basal and suprabasal layers (Fig. 1D,  Fig. S1). Nevertheless, the fluorescence intensity of RORa signal in the suprabasal layers was significantly stronger than in the keratin 14-positive cells of the basal layer (Fig. 1D). In addition, RORa staining was mainly concentrated in the nucleus, consistent with its function as transcription factor. SCC development is associated with altered keratinocyte differentiation [14,29]. RORa4 mRNA and protein levels were reduced in a panel of skin squamous cell carcinoma (SCC) cell lines when compared with a panel of normal HKCs (Fig. 1E, F). Real time qRT-PCR analysis also showed a significant downmodulation of RORa expression in a set of clinically occurring cutaneous SCCs when compared with normal epidermis (Fig. 1G). Immunofluorescence staining of SCC specimens revealed a decrement of RORa expression, when compared with surrounding relatively normal epidermis (Fig. S2). These results further indicate that RORa may play an important role in maintaining skin homeostasis.

2) RORa Promotes Differentiation of Human Keratinocytes
To determine whether RORa4 plays a role in keratinocyte differentiation, we overexpressed RORa4 in HKC at a level similar to that of the endogenous protein induced upon differentiation (Fig. 1C). At 72 h after infection, increased RORa4 expression significantly induced the expression of early and intermediate differentiation markers (keratin 1/10 and involucrin), as well as granular layer markers (loricrin and filaggrin) at both mRNA and protein levels ( Fig. 2A).
The late stage of keratinocyte differentiation is associated with changes in cellular metabolism, in particular lipid-related, that are essential for epidermal outer layer formation [13]. A number of epidermal barrier-related genes, including ALOXE3, ABCA12, ADRP (adipose-differentiation related protein), and AQP3 (aquaporin 3) [30,31,32,33], were up regulated in HKCs induced to differentiate by suspension culture (Fig. 2B), as well as by increased RORa4 expression (Fig. 2C), implicating RORa in this aspect of the keratinocyte differentiation program.
Increased RORa4 expression exerted no significant effect on markers of the basal proliferative compartment (Fig. 2D). However, Alamar blue density assay showed that increased RORa4 expression caused growth inhibition in HKCs from day 6 after transduction (Fig. 2E). The fraction of keratinocytes with clonal growth capability was also significantly reduced after prolonged culture (Fig. 2F). These results suggest that increased expression of RORa4 is sufficient to promote the differentiation program while reducing the long term growth potential of HKCs.

3) RORa is Essential for Keratinocyte Differentiation
We next investigated whether RORa is required for normal differentiation, using a loss of function gene silencing approach. Two separate siRNAs and a lenti-viral shRNA that target all RORa isoforms were able to efficiently knock down RORa expression at both mRNA and protein levels ( Fig. 3A, B). Although RORa silencing had no effect on the expression of basal layer markers like integrin b4 and keratin 14, it significantly reduced the expression of outer differentiation markers at mRNA and protein levels (Fig. 3A, B). For better evaluation of late differentiation, HKCs at growing density (72 h after siRNA transfection) were forced to differentiate by culture in suspension (Fig. 3C, D). RORa silencing decreased significantly the expression of loricrin, filaggrin (Fig. 3C) and epidermal barrierrelated genes (Fig. 3D), indicating that RORa is broadly required for keratinocyte differentiation.
To determine whether RORa plays a similar regulatory function in vivo, HKCs infected with lenti-viruses expressing control or RORa shRNA were injected at the dermal-epidermal junction of immune-deficient NOD/SCID mice. As previously reported [17,34], 8 days after injection, control cells formed well differentiated epidermal cysts, with pronounced granular and squamous layer formation. In contrast, RORa silencing led to structures without ordered layer formation and cornification (Fig. 4A). RORa expression, as assessed by immunofluorescence analysis, was much stronger in control cysts than in those formed by shRNA expressing keratinocytes, confirming the knockdown efficiency in vivo (Fig. 4B). Keratin 10 was strongly expressed in the well stratified and differentiated layers of control cysts, while weaker expression and association with loosely connected cells were found after RORa depletion (Fig. 4C, Fig. S3A). Loricrin was also strongly expressed in control cysts, while it was undetectable in the cysts formed by RORa knockdown keratinocytes (Fig. 4C, Fig. S3B). Oil red staining revealed that neutral lipids were accumulated in the stratum corneum of control cysts, but not of RORa knockdown cysts (Fig. 4D). Therefore, the in vivo cyst assays support the in vitro findings that RORa is required for normal keratinocyte differentiation, including the lipid layer formation program.

4) The Pro-differentiation Effects of RORa are Partially Mediated by FOXN1
Keratinocyte differentiation relies on an integrated transcriptional network, involving increased expression and activity of the FOXN1 and Notch1 genes. Real time qRT-PCR analysis showed (D) Immunofluorescence analysis of RORa expression in human skin. Frozen sections (8 mm) of normal human skin were co-stained with antibodies against RORa (green) and keratin 14 (red). DNA was with counterstained with Hoechst (blue). Images are representatives of several independent that mRNA expression of FOXN1 and Notch1 was significantly induced by increased expression of RORa4 (Fig. 5A), and decreased by RORa silencing (Fig. 5B). In contrast, RORa level did not affect the expression of other transcription factors (p53, cmyc, and NF-kB) (Fig. S4). Mat-Inspector software (Genomatrix) was used to analyze the transcription regulatory region of these genes, spanning 6 kb of upstream and 2 kb of downstream sequence from the transcription start sites (TSS). Multiple consensus RORa response elements (ROREs) were found in both FOXN1 and Notch1 regulatory regions. Chromatin Immunoprecipitation (ChIP) analysis of human epidermis by real time RT-PCR showed binding of endogenous RORa to an upstream region of the FOXN1 gene containing a predicted RORa binding site (-4.8 kb), but not to a downstream region containing another such site (+1.6 kb) (Fig. 5C). Despite the presence of predicted ROREs, ChIP assays failed to detect any significant binding of RORa to the Notch1 promoter region. Consistent with FOXN1 functioning as a direct RORa target, expression of the primary FOXN1 transcript, as detected by the primers corresponding to the first intron/exon junction, was similarly induced or blocked by modulation of RORa as the mature transcript (Fig. 5D).
To assess whether FOXN1 functions as a mediator RORa in differentiation, we tested the impact of increased RORa4 expression in HKCs plus/minus FOXN1 silencing (Fig. 6). Induction of early differentiation markers keratin 1/10 and Notch1 by increased RORa expression was counteracted to a large extent by siRNA-mediated FOXN1 knockdown (Fig. 6A). FOXN1 silencing showed opposite enhancing effects on expression of filaggrin (Fig. 6B), consistent with previous findings that FOXN1 functions as repressor rather inducer of late keratinocyte differentiation markers [22]. Selective effects of FOXN1 knockdown were also observed with the epidermal barrier-related genes. Silencing of FOXN1 blocked the ability of RORa to induce ADFP and AQP3 expression, while causing no repression or even slight fields (see more in Fig. S1), bar = 50 mm Fluorescence intensity of RORa/cell was quantified in 100 cells (from 3 human skin samples) in K14 positive basal versus negative (suprabasal) keratinocytes. Data are presented as mean fold-change of fluorescence signal/cell over signal/cell in basal layer 6 S.E.M., ***, p,0.001, N = 3 human skin samples. (E-F) Real time qRT-PCR (E) and western blot analysis (F) of RORa in HKCs (N = 5 batches), in parallel with keratinocyte-derived skin SCC12 and SCC13, as well oral SCCO12, SCCO22, and SCCO28 cell lines. Primers specific for RORa4 or all RORa isoforms were used for the RT-PCR analysis. Values are normalized to 36B4, and presented as mean fold-change over HKCs 6 S.E.M. **, p,0.05 ***, p,0.001, N = 3. (G) Real time qRT-PCR analysis of RORa expression in clinically occurring skin SCC tumors versus normal epidermis (NS), ****, p,0.0001. doi:10.1371/journal.pone.0070392.g001  up-regulation of ALOXE3 and ABCA12 (Fig. 6C). These results establish FOXN1 as a direct RORa target and selective mediator of its function in differentiation.

Discussion
The nuclear orphan receptor RORa plays a key role in embryonic development and a variety of physiological processes, such as lipid metabolism, bone formation, inflammation and T H 17 cell differentiation [6,35] [3]. In parallel with its inverse expression in normal keratinocytes and epidermis versus SCC cells and tumors, we have shown here that RORa plays an essential positive role in keratinocyte differentiation, with FOXN1 as a direct target and mediator. RORa4 is the major RORa isoform expressed in keratinocytes, and its expression is further induced upon differentiation at both mRNA and protein levels. In contrast to Notch1 and FOXN1, which promote early stages of keratinocyte differentiation and suppress the later ones [21,36,37], RORa4 is a positive determinant of both, with a role that extends to control of a group of genes that are involved in lipid synthesis, lipid delivery to lamellar bodies, lipid deposition/release from droplets, and water/glycerol transport [33,38,39,40,41,42].
FOXN1 is a key player in the epidermal differentiation program [21,22], hair follicle development [43], and skin pigmentation [44]. Moreover, we have previously reported that it functions as a determinant of benign versus malignant keratinocyte tumor development [27]. Little is known of the upstream factors that regulate FOXN1 expression in keratinocyte differentiation, except that it appears to be a direct negative target of the c-Jun transcription factor, and under opposite control of EGFR/ERK versus FGFR3 signaling [27,37]. We have shown here that RORa can also directly bind to the regulatory region of the FOXN1 gene and induce its transcription. The findings are of functional importance, as silencing of FOXN1 prevented the ability of RORa4 to induce expression of a number of differentiation marker genes, as well as a few of the genes involved in epidermal/ lipid barrier formation. Importantly, the ability of RORa to induce expression of Notch1, another key regulator of keratinocyte differentiation [29], was also prevented by the FOXN1 knockdown, consistent with a previous finding that FOXN1 is required to maintain Notch1 expression in the hair follicle matrix of mice [45]. In contrast to these genes, induction of filaggrin by RORa4 was enhanced rather than prevented by knocking down FOXN1, consistent with the already mentioned negative impact of this gene on late stages of differentiation [21,37]. FOXN1 knockdown did not affect RORa4-induced expression of specific genes connected with the lipid barrier function like ALOXE3 and ABCA12. This indicates that RORa functions as inducer of keratinocyte  differentiation through both FOXN1-dependent and -independent mechanisms. Such a conclusion is consistent with a recent report that filaggrin gene expression is positively controlled by RORa through an as yet undefined AP1-dependent mechanism [46].
A number of lipid products have been identified as natural or synthetic ligands or agonists/antagonists of RORa [47], including cholesterol sulfate [48,49,50], which is formed during squamous differentiation and may be involved in induction of RORa activity in the skin [46,50]. In general, nuclear receptors function as intracellular sensors of various lipids (Evans 2004) and, in addition to being part of the structural component, lipids can serve as signaling molecules for nuclear receptors involved in establishment of the epidermal barrier [51]. Thus, by inducing lipid metabolizing genes, enhanced RORa expression has the potential of modulating lipid molecules involved in its own regulation and/or regulation of other nuclear hormone receptors with a role in the skin. In particular, ALOXE3, a target gene of RORa, encodes an epidermal specific lipoxygenase eLOX3, which acts as a hydroperoxide isomerase and generates specific types of epoxyalchols (hepoxilins) [52,53]. Hypoxylins have been shown to bind and activate PPARa [38], and thus may mediate a cross talk with RORa function.
As reported in other epithelial tumors [8], we have found that RORa expression is down-regulated in keratinocyte-derived SCC cell lines and tumors. However, the causes of RORa low expression in SCCs are still unknown. One attractive possibility is that p53, often lost or mutated in skin SCCs, is positively controlling RORa, as already reported in colon cancer cell lines [54]. In this context, RORa acts as a positive feedback loop on p53 stability and pro-apoptotic functions [54,55]. In our own work, we have also obtained evidence that RORa expression can be positively controlled by p53, even though this was not observed in all tested conditions and with different human keratinocyte strains (unpublished results).
Besides genetic heterogeneity of response, another level of complexity that still has to be deciphered in further studies is the connection between RORa and control of the circadian cycle. In fact, RORa is also known to play a role in the circadian cycle [56], and a cross-talk between this and cell cycle control is currently emerging as an important element of cancer susceptibility [57]. It has also been reported that the circadian clock temporally fine- . Human epidermis was processed for ChIP assays utilizing rabbit antibodies specific for RORa or non-immune IgG control followed by PCR amplification of the indicated regulatory regions of the FOXN1 gene. The relative amount of precipitated DNA was calculated after normalization for total input chromatin, according to the following formula [59]: % total = 2 DCt 65, where DCt = Ct (input) -Ct (immunoprecipitation), Ct, cycle threshold. Statistical significance of the results was determined by unpaired Student's t-test, comparing the ratio RORa/IgG signal for each binding site relative to the one for the binding site at the RORE negative region. **, p,0.01, N = 3. (D) HKCs with increased or knocked down RORa expression were analyzed by qRT-PCR for primary FOXN1 transcript levels, using a primer corresponding to the first intron/exon junction. Samples were the same as described in (A-B). Values are presented as mean fold-change over control 6 S.E.M., ***, p,0.001, N = 3. doi:10.1371/journal.pone.0070392.g005 tunes the self renewal potential of epidermal stem cells [58]. Thus, besides its clear role in keratinocyte differentiation, RORa may play additional functions in the skin, depending on growth/ differentiation stages of keratinocytes and/or in response to multiple exogenous signals.

Ethics Statement
The animal study (protocol #: 2004N000170) was specifically approved by the Subcommittee on Research Animal Care (SRAC), which serves as the Institutional Animal Care and Use Committee (IACUC) in Massachusetts General Hospital. The animal study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All surgery was performed under sodium pentobarbital anesthesia, and all efforts were made to minimize suffering.

Cell Culture and Human Specimens
Primary human keratinocytes were isolated and cultured in serum-free keratinocyte-SFM medium (Gibco) supplemented with 30 mg/ml bovine pituitary extract (BPE) and 0.2 ng/ml rEGF on the collagen coated plates. Suspension induced differentiation was achieved by culturing HKCs on the poly (2-hydroxyethylmethacrylate) [poly-HEMA] (Sigma) coated Petri dishes for 24 or 48 hours. SCC13 and SCC12 cell lines were provided by Dr. J. Rheinwald (Brigham and Women's Hospital), and the SCC O12, O22 and O28 cells were provided by Dr. J. Rocco (Massachusetts General Hospital). The information of skin SCC samples is described previously [17].

Plasmids and Viruses
The retroviral pinco-Flag-RORa2 and pinco-Flag-RORa4 plasmids were generated by cloning the Flag-tagged full-length cDNAs of the two human RORa isoforms into the BamHI/EcoRI sites of the pinco-GFP vector. The cDNAs with restriction enzyme sites were generated from PCR using pcDNA-RORa2 and pCR-BluntII-TOPO-RORa4 plasmids (Open Biosystems) as templates, as well as the following primers 1) Forward for RORa4:59-GATTCCGGATCCGCCACCATGGACTACAAGGACGAC-GATGACAAGATGATGTATTTTGTGATCGCAGCG; 2) Forward for RORa2:59-GATTCCGGATCCGCCAC-CATG-GACTACAAGGACGACGATGACAAGATGAAT-GAGGGGGCCCCAGGAGAC; 3) Reverse for both RORa2 and RORa4:59-GCTGCTGAATTCCTATTACCCATCAA-TTTGCATTGCTGG. The lentiviral MISSION shRNA against all RORa isoforms is obtained from Sigma (TRCN0000022154). Conditions for retro-and lenti-virus production and infection were as previously reported [26]. For mRNA analysis, 500 mg of total RNA, isolated with RNeasy Mini QIAcube kit (Qiagen), was reversely transcribed into cDNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). The PCR procedure and primers for RORa 1-4 isoforms were as described previously (Pozo et al. 2004). qRT-PCR with SYBR Green detection (Roche Applied Science, Indianapolis, IN, USA) was performed on the Light Cycler 480 Real Time PCR instrument (Roche Applied Science), according to manufacturer's instructions. Each sample was tested in triplicate, and results were normalized with the expression of the housekeeping 36b4 gene. The list of gene-specific primers for qRT-PCR is provided in Table S1.

Western Blot Analysis
Conditions for immuno-blotting were as previously described, with the same antibodies listed for fluorescence microscopy, except rabbit anti-RORa (sc-28612, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA). As loading controls, membranes blotted with rabbit antibodies were incubated with blocking buffer containing 0.2% sodium azide and were re-probed with mouse anti-c-tubulin (GTU-88, Sigma) antibody.

Alamar Blue Assay
Cell proliferation was measured by the Alamar Blue assay (Invitrogen). HKCs were plated in triplicate on the 96-well collagen coated plates at a density of 1000 cells/well in 100 ml medium. At the specific time points, 5 ml of Alarma Blue reagent was added to the medium for 1 hour at 37uC. Fluorescence was monitored at 530-560 nm excitation wavelength and 590 nm emission wavelength on the Victor TM X3 Multilabel Plate Reader (Perkin Elmer, Salem, MA, USA).

Clonogenicity Assay
HKCs were plated at a density of 1000 cells/well in the 6-well plates without collagen coating. Nine days later, cells were fixed with 100% ethanol for 10 min at RT, followed by staining with 0.1% crystal violet in 10% ethanol at RT for 1 hour. After washes in water, the plates were dried at RT overnight. Colonies were counted with the Image J software.

Cyst Assay
For in vivo cyst assays, HKCs infected with lenti-virus expressing control or RORa shRNAs were selected by puromycin 2 days after infection. After selection, cells were collected and admixed with matrigel (4:1), followed by intra-dermal injection (2610 6 cells per spot) into the back skin of NOD/SCID mice (Taconic Farms Inc. Germantown, NY, USA), as previously described [17]. To minimize the individual animal variations, HKCs plus/minus RORa knockdown were injected in parallel in the right and left flank of the same mice. Mice were sacrificed 1 week after injection and the nodules formed from HKCs were processed to make frozen blocks with OCT (Fisher Scientific, Hanover Park, IL, USA).

Chromatin Immunoprecipitation (ChIP)
Human epidermis was separated from the underlying dermis by a brief heat treatment [27]. The finely minced tissue samples were then cross-linked with 1% formaldehyde/PBS at RT for 10 min, followed by addition of 125 mM glycine. After washes in PBS, the tissue pellets were processed for chromatin immunoprecipitation (ChIP) assays as described in [26], using the ChIP assay kit (Millipore) and the rabbit anti-RORa antibody (ab60134, Abcam), in parallel with the affinity-purified non-immune rabbit IgG. The relative amount of precipitated DNA was analyzed by qRT-PCR using primers against the RORE-containing or RORE-negative regions, and calculated after normalization to total input chromatin, according to the formula: % total = 2 DCt 65, where DCt = Ct (input) -Ct (immunoprecipitation), Ct, cycle threshold.

Statistics
All statistical evaluations were carried out using GraphPad Prism 5.0. All analyses are unpaired two-tailed Student's t-test. Real-time RT-PCR samples were tested in triplicate, and repeated at least three times. After normalization to the housekeeping gene 36b4, combined data was represented as mean-fold over control 6 S.E.M. P-values,0.05 were considered significant. Figure S1 Immunofluorescence analysis of RORa in human skin. Frozen sections (8 mm) of normal human skin were co-stained with antibodies against RORa (green) and keratin 14 (red). DNA was counterstained with Hoechst (blue). Images are representatives of independent fields from 2 skin samples, derived from different patients, as in Fig. 1D, bar = 50 mm.