Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Nov 1.
Published in final edited form as: J Invest Dermatol. 2019 Jun 25;139(11):2281–2291. doi: 10.1016/j.jid.2019.04.021

Transcriptomic network interactions in the human skin treated with topical glucocorticoid clobetasol propionate

Loukia N Lili 1, Anna Klopot 2, Benjamin Readhead 1,3, Gleb Baida 2, Joel T Dudley 1, Irina Budunova 2,*
PMCID: PMC6814545  NIHMSID: NIHMS1532721  PMID: 31247200

Abstract

Glucocorticoids (GCs) are the most frequently used anti-inflammatory drugs in dermatology. However, the molecular signature of GCs and their receptor (GR) in human skin is largely unknown. Our validated bioinformatics analysis of human skin transcriptome (RNASeq) induced by topical glucocorticoid clobetasol propionate (CBP) in healthy volunteers identified numerous unreported GC-responsive genes, including over a thousand non-coding RNAs. We observed sexual and racial dimorphism in the CBP response including a shift towards IFNα/IFNγ and IL6/JAK/STAT3 signaling in female skin; and a larger response to CBP in African-American skin. Weighted gene co-expression network analysis unveiled a dense skin network of 41 transcription factors (TF) including circadian KLF9, and ~ 260 of their target genes enriched for functional pathways representative of the entire CBP transcriptome. Using keratinocytes with KLF9 knockdown, we revealed a feed-forward loop in GR signaling, not previously reported. Interestingly, many of the CBP-regulated TFs were involved in the control of development, metabolism, circadian clock; and 80% of them were associated with skin aging showing similarities between GC-treated and aged skin. Overall, these findings indicate that GR acts as an important regulator of gene expression in skin - both at the transcriptional and post-transcriptional level - via multiple mechanisms including regulation of non-coding RNAs and multiple core TFs.

INTRODUCTION

Glucocorticoid hormones are essential regulators of proliferation, differentiation, inflammation, and metabolism in skin (Chebotaev et al., 2007, Sarkar et al., 2017, Yemelyanov et al., 2007). Reduced endogenous glucocorticoids have been causatively linked to the development of multiple skin diseases, including psoriasis (Sarkar et al., 2017).

For that reason, synthetic glucocorticoids (GCs) are the most widely used drugs for topical therapy of inflammatory and hyperproliferative skin diseases, such as atopic dermatitis/eczema and psoriasis (Stern, 1996). Since they were introduced to the clinic in 1952, more than thirty GCs have been approved for clinical use (Gual and Abeck, 2015, Sternberg et al., 1952). Currently, topical GCs represent up to 20% of all prescriptions in dermatology (Kumar et al., 2016, Stern, 1996). Unfortunately, chronic treatment with glucocorticoids results in adverse effects including delayed wound healing, skin atrophy, pigment alteration, and increased risk of skin infections (Del Rosso and Friedlander, 2005, Kumar et al., 2016, Vandewalle et al., 2018). Therefore, alternative and safer therapies are a major unmet need in treating skin diseases (Schacke et al., 2007).

The GR/NR3C1 is a well-characterized transcription factor (Oakley and Cidlowski, 2013, Rhen and Cidlowski, 2005). In non-activated cells, it resides in the cytoplasm bound to molecular chaperones: heat shock proteins and immunophilins (Owens-Grillo et al., 1995, Tai et al., 1992). Upon GC binding, GR undergoes phosphorylation and translocates to the nucleus, where it regulates gene expression via: (i) transactivation (TA), which requires GR homodimer binding to glucocorticoid-responsive elements (GRE); and (ii) transrepression (TR) via different mechanisms including negative interaction between monomeric GR and other transcription factors (TF), such as NF-kB, AP-1, p53, STATs and SMADs (Li et al., 2003, Petta et al., 2016, Ray and Prefontaine, 1994, Yemelyanov et al., 2007, Zhang et al., 1997).

Despite the important role of GCs in skin physiology and diseases, the GC/GR molecular signature in human skin is largely unknown. Here, we present the results of RNASeq gene co-expression network analysis of the human skin transcriptome induced by topical treatment with one of the most potent glucocorticoids, clobetasol propionate (CBP). This analysis identified numerous GC-responsive genes in skin not previously described, and revealed complex multileveled mechanisms of GC/GR gene expression regulation via other cell receptors, hundreds of non-coding RNAs, and functionally diverse transcription factors.

RESULTS

Topical glucocorticoid treatment induces global changes in human skin transcriptome

To develop a comprehensive GC/GR molecular signature in human skin, first, we identified 3,524 differentially expressed genes (DEGs) in CBP-treated skin (FDR < 0.2) (Fig. 1a), which constitute approximately 17% of the entire transcriptome of those samples. Τransactivation and transrepression were almost equally presented i.e., 1,607 DEGs were up-regulated (233 by more than 2-fold), and 1,917 were down-regulated (204 by more than 2-fold) (Fig. 1a and Supplemental Table S1). About 90% of all DEGs were putative GR targets with at least one ChIP-Seq peak, 78% with at least two, and 67% with at least three ChIP-Seq peaks (GTRD database of bona fide GR targets (Yevshin et al., 2017) (Materials and Methods and Supplemental Table S1).

Figure 1. CBP molecular signature in human skin.

Figure 1

Open in a new tab

Effect of CBP on gene expression in whole thickness skin biopsies obtained from 17 healthy volunteers (17 control and CBP-treated skin sets).

a. The volcano plot representing the fold-change of expression (in log2 scale) in CBP versus Control skin, and the significance of the fold change (-log10 scale of FDR) for the entire skin transcriptome of 21,339 genes. Grey dots show the 3,524 differentially expressed genes (DEGs) with FDR < 0.2; red dots are DEGs up-regulated by > 2fold, and blue dots are DEGs down-regulated by ≤ −2fold.

b. DEGs from the different classes of non-coding RNAs. The total number of non-coding RNA DEGs was 1,471 per ENSEMBL91 annotation (more details in Supplemental Table S1).

c. Validation of DEGs by qRT-PCR. The results (normalized to Rpl27) are the means ± SD calculated for three individual RNA samples sets/DEG. *denotes statistically significant difference (t-statistics, p < 0.05) as compared to control.

Overall, the GC molecular signature in human skin reflected different facets of GR functions including its anti-inflammatory, metabolic and atrophogenic effects. The group of DEGs with the greatest expression changes included down-regulated pro-inflammatory chemokines (CCL13, CCL19 and CXCL12) along with CLDN10 (major component of tight junctions), MXRA5 (matrix remodeling protein), host of collagen-encoding genes (COL15A1, COL5A1, COL1A1), LOXL1 (responsible for crosslinks in collagen and elastin), and cell adhesion protein SELE (Fig. 1a). These changes are consistent with the anti-inflammatory GC/GR activity and its negative effects on cell junctions, skin barrier function, extracellular matrix and tissue remodeling. The metabolic effects of GCs were present in the activation of amino acid metabolism and adipogenesis regulator KLF15, glutamine synthase GLUL, adipose differentiation marker PLIN2, apolipoprotein APOB, and the central regulator of glucose metabolism PDK4 (Sugden and Holness, 2003, Takahashi et al., 2016). Among other upregulated DEGs were anti-inflammatory factor TSC22D3/GILZ, and negative regulators of mTOR/Akt, FKBP5 and DDIT4/REDD1, which play a causative role in GC-induced skin atrophy (Baida et al., 2015, Baida et al, 2018).

Although the negative effect of GCs on matrix metalloproteinase genes (MMP) genes is well known (Klopot et al., 2015, Schoepe et al., 2010) and most of MMPs were down-regulated in this data set (Supplemental Table S1), there was an unexpected activation of MMP7 (matrilysin) and MMP3 (stromelysin). This possibly occurred to counteract the negative effect of GCs on extracellular matrix at the 24 hrs of treatment.

Interestingly, 42% of all DEGs (1,471/3,524) were non-coding RNAs, including 713 long non-coding RNAs (lncRNAs) and 583 pseudogenes (Fig. 1b, Supplemental Table S1). More than 80% of those non-coding RNAs appeared to be putative GR targets [GTRD database, putative GR targets (Yevshin et al., 2017), see Materials and Methods and Supplemental Table S1]. Even though it has been already reported that GCs affect non-coding RNA expression (Austin et al., 2017, Liu et al., 2018, Sawaya et al., 2018), their effect on non-coding RNAs in skin was much less studied. Here, we show that GCs affect hundreds of non-coding RNAs in skin.

Surprisingly, the lincRNAs: ANCR, TINCR, and PRINS, which play important roles in the skin homeostasis (Hombach and Kretz, 2013), were not affected by CBP in human skin. Instead, we found other lincRNAs, including LINC01088 and LINC01091, among the most up/down- regulated DEGs (Figs. 1a and 1c). Although their role in GC-response is not known, it was reported that LINC01088 was decreased in the scalp of patients with alopecia areata (Xing et al., 2014), a disease typically treated with GCs (Gupta et al., 2017). The expression of LINC01091 was down-regulated in the psoriatic skin, and linked to the changes in bone transcriptome induced by Cushing syndrome, which in turn is induced by increased amounts of glucocorticoids (Lekva et al., 2010, Lekva et al., 2012).

These multifaceted changes in gene expression reflect the heterogeneity of whole-thickness skin biopsies used in this study. Indeed, in addition to keratinocytes, other cell types also contributed to this CBP molecular signature in skin. Cell enrichment analysis with xCell [http://xcell.ucsf.edu/; (Aran et al., 2017)] revealed significant differential enrichment between control and CBP-treated skin in: macrophages, dendritic cells, basophils, CD4+ T-lymphocytes, hematopoietic stem cells and stromal/smooth muscle cells (Supplemental Table S2).

Inflammatory, metabolic and receptor activity response orchestrate glucocorticoid molecular changes in human skin

To evaluate CBP systemic effects on human skin transcriptome, we performed GSEA pathway enrichment analysis using Hallmark gene sets and GO molecular signatures (Fig. 2, Supplemental Table S3). Consistent with the known GC effects, highly enriched processes of the 3,524 DEGs were inflammatory/immune/host defense response pathways including TNFα, STAT, and interferon signaling; stress response to UV and hypoxia; and metabolic pathways, including fatty acid metabolism and glycolysis. At the molecular function level, over-represented gene sets were receptor activity and receptor binding, including major classes of cell receptors such as ionotropic and G protein-coupled (metabotropic) receptors, as well as receptors to cytokines (Fig. 2; Supplemental Table S3).

Figure 2. Gene set enrichment analysis (GSEA) of functional changes in human skin treated with CBP.

Figure 2

Open in a new tab

Top enriched pathways (FDR < 0.05) of all 3,524 DEGs of the CBP molecular signature in human skin. The Gene Set Enrichment Analysis (GSEA) MSigDB webtool was used to query the Hallmark and GO Molecular Function gene sets (http://software.broadinstitute.org/gsea/msigdb). Note: Inflammatory, metabolic and receptor activities are among the major pathway categories.

Sexual and racial differential response to topical glucocorticoid treatment

Although skin morphology, physiology, and certain GC-treated cutaneous diseases depend on sex and race (Halder and Nootheti, 2003, Yin et al., 2014), there is limited data regarding sex- and race-biased transcriptomic changes in GC-treated skin. To address this limitation, we evaluated both CBP-induced and also basal gene expression (in control samples) for both the male/female and the African-American (AA)/Caucasian (CC) skin samples. For the sexual and racial dimorphism analysis, we used all available 10 female and 7 male samples, and all 9 AA and 8 CC samples respectively (see sample description in Supplemental Tables S1, S4 and S5).

a). Sexual dimorphism

In male/female control samples, we revealed 35 DEGs (at FDR < 0.2) not located at sex chromosomes (Supplemental Table S4). Approximately ~90% of those DEGs were up-regulated in males. The sexually biased expression of 15/35 DEGs was reported previously (Gershoni and Pietrokovski, 2017).

Sex-specific CBP response revealed 440 DEGs in males and 454 DEGs in females (FDR < 0.2, fold change > 2) (Fig. 3, Supplemental Table S4). Males and females shared 317 DEGs; 123 and 137 were unique in male and female respectively (Fig. 3b). GSEA analysis of sex-specific DEGs indicated that CBP response in females was shifted towards IFNα/IFNγ and IL6/JAK/STAT3 signaling versus IL2 STAT5 in males (Fig. 3a). IL-2 STAT5 pathway inhibits inflammatory responses and promote T regulatory cell development, while IL-6 STAT3 pathway facilitates pro-inflammatory processes (O’Shea and Plenge, 2012). A sex-biased metabolic component including glycolysis and estrogen response was also observed in male skin. Also, the chemokine CCL2 and the immune response-associated CD163L1 were less down-regulated by CBP in males suggesting that male skin may be less responsive to anti-inflammatory effects of CBP (Supplemental Table S4). We validated the difference in CBP response for several common female/male DEGs including down-regulated CCL2 and GAP43 and up-regulated IYD and MMP3 (qRT-PCR, Fig. 3c). However, due to the limited number of samples per group, the statistical power of these observed trends was limited.

Figure 3. Sexually dimorphic response to topical CBP application in skin.

Figure 3

Open in a new tab

Effect of CBP on gene expression in whole thickness skin biopsies obtained from 10 females (both racial backgrounds) and 7 males (both racial backgrounds); control and CBP-treated skin sets. a, b, c - Female samples (dark gray) and Male samples (light gray).

a. Functional sexual dimorphism in response to topical CBP application. List of top enriched pathways for the 137 female-specific and 123 male-specific DEGs obtained by GSEA analysis at FDR < 0.05 (Hallmark database). Note: Top pathways for the female-specific DEGs involved INFα/γ and IL6/JAK/STAT3, whereas for the male-specific DEGs, KRAS and IL2/STAT5.

b. Venn diagram of the 317 common DEGs between the female and male skin samples, 123 male-only DEGs, and 137 female-only DEGs.

c. qRT-PCR validation of the differences in the amplitude of response to CBP (expression fold change) between common DEGs in male and female skin. The results (normalized to Rpl27) are the means ± SD calculated for three individual RNA samples/condition. *denotes p < 0.05 for changes between male versus female response to CBP.

b. Racial dimorphism

There was a striking difference in basal gene expression between AA and CC control samples. A total of 449 DEGs in AA versus CC skin (FDR < 0.2) were highly enriched for immune responses (Supplemental Table S5). This observation associates basal gene expression with the prevalence of inflammatory skin diseases such as acne and atopic dermatitis among AA patients (Halder and Nootheti, 2003).

Racial-specific GC response showed 518 DEGs in AA and 376 DEGs in CC samples (FDR < 0.2 and fold-change > 2) indicating an overall stronger AA response to CBP (Fig. 4b). AA and CC samples shared 337 DEGs with 181 and 39 unique DEGs respectively (Fig. 4b). GSEA analysis of these DEGs suggested that AA skin response was associated with inflammation and metabolic disruptions versus cell barrier modifications of CC response (Fig. 4a). Interestingly, several GR targets such as ZBTB16, FKBP5, DDIT4 and GILZ were more activated in AA than in CC skin (Fig. 4c, Supplemental Table S5), providing additional evidence for the increased GR activity in AA skin. However, due to the limited number of samples per group, the statistical power of these interesting observed trends was limited.

Figure 4. Racially dimorphic response to topical CBP application in skin.

Figure 4

Open in a new tab

Effect of CBP on gene expression in whole thickness skin biopsies obtained from 9 African-American (AA, both sexes) and 8 Caucasian (CC, both sexes) volunteers (control and CBP-treated skin sets). a, b, c – African American skin samples (dark gray) and Caucasian skin samples (light gray)

a. Functional racial dimorphism in response to topical CBP application. List of top enriched pathways for the 181 AA-specific DEGs, and 39 CC-specific DEGs obtained by GSEA analysis at FDR < 0.05 (Hallmark database). Note: Top pathways for AA skin-specific DEGs are enriched for inflammation and metabolism.

b. Venn diagram of the 337 common DEGs between the AA and CC skin samples, 181 AA-only DEGs, and 39 CC-only DEGs.

c. qRT-PCR validation of the differences in the amplitude of response to CBP (expression fold change) between common DEGs in AA and CC skin. The results (normalized to Rpl27) are the means ± SD calculated for three individual RNA samples/condition. Note: There is a trend to increased induction of GR target genes in AA skin.

Core TF networks associated with skin aging span the molecular drivers of CBP glucocorticoid-induced response in human skin

To determine the drivers of the molecular CBP effects in skin, we employed Weighted Gene Co-expression Network Analysis (WGCNA) (Langfelder and Horvath, 2008). The initial network of all 21,339 genes in all samples grouped co-expressed genes into 30 modules/clusters, two of which significantly correlated with treatment (Fig. 5; Supplemental Table S6). Hub genes of these two modules included mostly affected validated DEGs: DDIT4, FKBP5, BEST2, DKK1, ZBTB16, TSC22D3 and LINC01088 (positively correlated with GC) as well as MXRA5, DKK2, COL15A1 and LINC01091 (negatively correlated with GC) (Fig. 5; Supplemental Table S6).

Figure 5. Transcription factor (TF) regulatory networks perturbed in human skin in response to topical glucocorticoid CBP.

Figure 5

Open in a new tab

a. Preservation statistics of all modules in the control network projected onto the CBP-treated network (after WGCNA analysis). Each node represents one module. Five modules had moderate to low preservation Zsummary (<10) and also high preservation median rank, the latter meaning that their observed preservation statistics were the highest among all modules.

b. Network representation of the 41 TFs and 260 their target genes from the 3 out of the 5 non-preserved modules. Note: These TFs and their targets were also identified as DEGs with fold change > 2 (See Supplemental Table S1). Grey squares represent TFs and orange circles represent their targets genes.

c. Functional enrichment of the genes in the five non-preserved modules using the Hallmark database. Top10 most significant pathways are shown (FDR < 0.05). Pathways identical to the major pathways identified for all DEGs in skin after CBP (Supplemental Table S3) are highlighted in bold italic.

d. Detailed network representation of the TFs and their target genes in module 3. Only the TFs are denoted in the network; red squares are significantly up-regulated TFs and blue squares are significantly down-regulated TFs in skin treated with CBP.

To identify topological changes of the transcriptional networks between control and CBP-treated samples, we constructed two separate networks using the same genes: one for control and another for CBP-treated samples (Fig. 5; Supplemental Table S6). Per WGCNA manual, the statistics showed that five modules of the control network were altered in the CBP network (modules 1–5, Fig. 5a and Supplemental Table S6). Three of these five modules included 300 DEGs identified as either TFs (41 genes) or their targets (259 genes) [TRRUST database v2 (Han et al., 2018) (Fig. 5b)]. Most of these GC-responsive TFs are known regulators of metabolism (AHR, ESR1, NRF2, RUNX1T1, RUNX2, KLF7, PPARG), circadian clock (PER1, KLF9, NR1D1, BHLHE41), development (TFCP2L1, LMO2, MEF2C, BIN1, SIX1), cell proliferation, differentiation and apoptosis (Fig. 5c). In addition, TCF7L2 and TCF4 are part of the Wnt signaling pathway, and are important for the development of stratified epidermis and skin barrier function (Mulholland et al., 2005).

More than 70% of these TFs and targets were GR putative targets (GTRD database, putative GR targets (Yevshin et al., 2017)) (Materials and Methods and Supplemental Table S1). Overall, these networks with GR-regulated TFs hubs suggest the existence of many feed-forward and feed-back loops regulated by GR itself and by downstream GR-dependent TFs. Pathway enrichment of the 300 TFs and targets provided even more compelling evidence showing that most of these pathways (12/20) were shared between the 300 TFs/targets and all 3,524 GC-dependent DEGs (Fig. 5c, bold italic). The expression changes for several TFs were validated by qRT-PCR (Fig. 6b).

Figure 6. Transcription factor involvement in GR signaling in skin: relevance to metabolism, circadian clock and aging.

Figure 6

Open in a new tab

a. The list of Transcription Factors (TFs) identified by preservation statistics of WGCNA modules. These TFs were DEGs and had fold change ≥ 2. Blue signifies down-regulation and red up-regulation of expression in CBP-treated skin samples. * Denotes association with skin aging (comparison with data from MuTHER consortium data (Supplemental Table S7).

b. Validation of DEGs by qRT-PCR. The results (normalized to Rpl27) are the means ± SD calculated for three individual RNA samples sets for each DEG. # Denotes statistically significant difference (t-statistics, p < 0.05) as compared to control.

c. KLF9 modulates GR function. KLF9 induction in immortalized human keratinocytes IHEK by fluocinolone acetonide (FA, 10–6 M) evaluated by qRT-PCR (c1) and Western blotting (c2). c3. Knockdown of KLF9 in IHEK cells by shKLF9-lentivirus. Cells infected by pGIPZ-lentivirus were used as control. c4. GR function was evaluated by GRE.Luciferase reporter test. shKLF9-IHEK and pGIPZ-IHEK control cells were infected with GRE.Luciferase lentivirus, maintained in glucocorticoid-free medium for 12 hrs, and treated with FA (10–6 M ) or vehicle for 24 hrs. The Luciferase induction is presented as fold change to pGIPZ control cells. The means +/− SD were calculated for three individual wells/group in one representative experiment. a: P<0.05 for changes compared to corresponding vehicle control. b: P<0.05 for changes in Luciferase induction in shKLF9-IHEK compared to pGIPZ-IHEK cells. Note: the effect of KLF9 on basal GRE.Luciferase activity is not statistically significant.

Our recent studies have shown that the expression of several TFs including KLF9 (Kruppel like factor 9) - a circadian TF that controls keratinocyte proliferation (Sporl et al., 2012) - is regulated by GCs in keratinocytes in vitro (Agarwal et al., 2019, Lesovaya et al., 2018). Thus, we assessed the role of KLF9 in GR signaling using IHEK immortalized human keratinocytes (Fig. 6c). We generated IHEK cells with KLF9 Knockdown (KD) using lentivirus expressing KLF9 shRNA (Fig. 6c). IHEK cells infected with pGIPZ empty lentivirus were used as control. To generate reporter cells for Luciferase assay, we infected both cell lines with GRE.Luciferase lentivirus. Interestingly, in keratinocytes with KLF9 KD, the GR transcriptional activation by FA (10–6 M x 24 hrs) was significantly (by ~ 40%) reduced compared to pGIPZ-control cells. At the same time, there was no statistically significant effect of KLF9 KD on basal already low GRE.Luciferase reporter activity in control (Fig. 6c), which is mostly driven by basal promoter in the absence of GCs (we maintained control cells in GCs-free medium). In addition, in silico analysis of TF binding sites in basal promoter of GRE.Luc reporter by Patch 1.0 tool and Transfac Public database (see Materials and Methods) has not revealed any putative KLF9 sites. Overall, these results suggest that KLF9 can modulate GR activity, and provide proof of principle that at least some of the down-stream TFs can play an important role in mediating GR transcriptional activity via feed-forward and feedback loops. In this regard, it is interesting to mention the earlier attempts to assess feed-forward and feedback loops in GR signaling mediated by GR-dependent TFs, and one of the predicted feed-forward loops in macrophages involved GR-KLF9-KLF2 (Chinenov et al., 2014).

We also show that 33 of the 41 TFs involved in CBP skin transcriptome (Fig. 6a) have been associated with skin aging using the MuTHER consortium as reference (Glass et al., 2013) (Supplemental Table S7). This aligns with the phenotype of chronically treated skin with GCs, resembling typical properties of aged skin such as atrophy of all skin compartments, increased fragility, tearing, bruising, and compromised skin barrier function (Rittie and Fisher, 2015). Among the 33 aging related TFs, there were the circadian rhythm regulators KLF9 and PER1 (circadian rhythm related Kruppel-like factor and period circadian regulator gene respectively) as well as the hematopoietic development LIM-domains LMO2, LMO3 and RUNX1/RUNX2, which are pivotal modulators of hematopoietic, hair follicle and bone stem cells and are known to be involved in GC-induced osteoporosis (Frenkel et al., 2015, Glotzer et al., 2008, Osorio et al., 2011).

DISCUSSION

Although some individual GR target genes in skin and transcriptome changes induced in human keratinocytes by GCs in vitro have been described (Agarwal et al., 2019, Ahluwalia, 1998, Baida et al., 2015, de Jongh et al., 2005, Petta et al., 2016, Stojadinovic et al., 2007, Sundahl et al., 2015), a comprehensive analysis of GC/GR molecular signature in human skin is lacking. We used high-resolution RNASeq technology and validated, integrative network biology approaches to determine the GR molecular signature and the molecular drivers of global transcriptomic alterations in GC-treated human skin.

Our results were consistent with known anti-inflammatory, metabolic and atrophogenic GC/GR effects in skin. Surprisingly however, our analysis revealed a large cohort of DEGs that either have not been previously studied in keratinocytes/skin or have not been associated with GC regulation. Some of these genes have known functions in hematopoietic (ZBTB16, IKZF1, GLUL, HOXA10), ocular (BEST2), adipose (PLIN2), and stem cells (TFCP2L1, ATOH8) (Balakrishnan-Renuka et al., 2014, Guttsches et al., 2015, Staubert et al., 2015, Ye et al., 2013, Zhang et al., 2009). Others have recently been identified as DEGs in GC-treated human keratinocytes in vitro (e.g., TFCP2L1, HOXA10, GLUL) suggesting their role in epidermis (Agarwal et al., 2019, Lesovaya et al., 2018, Stojadinovic et al., 2007). Interestingly, even though the direct overlap of DEGs induced by glucocorticoids in human skin (this study) and in primary human keratinocytes in vitro (Stojadinovic et al., 2007) was limited to 40–60 genes (Supplemental Table S8), the similarity between gene cohorts identified by GSEA was rather impressive.

Another finding that – to our knowledge - has not been previously reported was the fact that non-coding RNAs constituted almost half and lncRNAs almost 25% of all DEGs suggesting that GR effects in skin extend beyond the traditional transcriptional gene expression regulation (via TA/TR, RNA splicing, subcellular localization and stability (Hombach and Kretz, 2013, Hu et al., 2018) towards regulation at post-transcriptional level.

Analysis of potential drivers of the GC-induced molecular changes in skin identified some GC-regulated TFs (e.g., DDIT4, FKBP5, BEST2, GLI2) as pivotal modulators of the GC transcriptional network. Further analysis of gene module topology in control vs. CBP-treated samples uncovered a regulatory network of 41 TFs (most of them are putative GR targets) and ~ 260 of their target genes. This TF/target network was enriched for the majority of functional pathways of the entire CBP transcriptome. Even though the interaction between GR and other TFs on protein-protein level is known (Vandewalle et al., 2018), the extensive regulation of TF expression by GCs in skin has not been described before. Moreover, using KLF9 as an example, we obtained experimental evidence suggesting that some of these TFs may play an important role in GR signaling.

In conclusion, bioinformatics analysis of the GC-induced transcriptome unveiled complex multi-layer mechanisms of gene expression regulation by GC/GR suggesting that GR functions as an important regulator of transcription in skin. These mechanisms involved changes in binding and activity of major classes of cell receptors, changes in the expression of regulatory non-coding RNAs, and direct transcriptional control of multiple TFs. These TFs with GR co-regulate hundreds of target genes involved in physiological, therapeutic and adverse effects of GCs. Therefore, these results not only advance our understanding of GR signaling in skin, but also provide a foundation for the development of safer GR-targeted therapies in a sex- and race-dependent manner. Finally, taking into consideration the similarity between atrophic processes in aged skin and skin chronically treated with glucocorticoids as demonstrated here, our results could serve as a base for future strategies on skin protection against aging.

MATERIALS AND METHODS

Human skin biopsies

Clobetasol Propionate (CBP) ointment, USP, 0.05% (Akorn) was applied once to the volar aspect of the upper right arm skin (relatively sun-protected) of 17 healthy volunteers without a history of relevant skin disorders or exposure to topical or systemic glucocorticoids for at least 6 months. Four mm punch biopsies were taken 24 hrs after CBP application. Biopsies from the untreated left volar arm skin were used as controls. Subjects included 10 females (both racial backgrounds), 7 males (both racial backgrounds); 9 African-American (both sexes), 8 Caucasian (both sexes), ages 25–64. After biopsy, skin samples were immediately frozen in liquid nitrogen and stored at −80 C.

Informed consent was obtained in writing, following the principles established in the WMA Declaration of Helsinki and the NIH Belmont Report. All studies were approved by the Northwestern University IRB.

RNA extraction

Total RNA from whole human skin, and keratinocyte cell cultures was isolated with RiboPure kit (Ambion, Life Technologies, Grand Island, NY, USA). The RNA samples were treated with TURBOTM DNase (Ambion), checked for quality and integrity, and used for RNASeq.

Validation of RNASeq data

Validation of RNASeq data was performed using at least three paired RNA sample sets per DEG. Sets were randomly selected from the pool of 17 sets used for RNASeq. Primers were designed with NCBI Primer-BLAST, and detection was performed on the Applied Biosystems 7000 Real-Time PCR instrument (Life Technologies). qRT-PCR for each RNA sample was repeated three times (technical repeats). The qRT-PCR results were normalized to the expression of the housekeeping RPL27 gene in the same sample (de Jonge et al., 2007). The CBP effects were assessed by Student’s t-test; fold change (FC) differences (CBP to Control) were considered significant at p < 0.05.

Kruppel-like factor 9 (KLF9) effect on GR signaling in human keratinocytes

Immortalized human epidermal keratinocytes (IHEK) were generated by the infection with HPV E7-expressing lentivirus, and provided by NU SDRC skin tissue engineering Core. KLF9 was knocked down in IHEK by infection with lentivirus expressing shKLF9 targeting KLF9 3ÙTR TCATGCTGGAGTAGATGTG sequence (clone V2LHS_267048, Open Biosystems/ Dharmacon human shRNA library). Cells infected with lentivirus expressing empty pGIPZ vector (Open Biosystems/Dharmacon) were used as a control. Viral suspensions were provided by Northwestern University SDRC DNA/RNA delivery Core. Effective KLF9 knockdown was verified using Western blot with anti-KLF9 antibody (GTX129316, Genetex, 1:1000 dilution). shKLF9- and pGIPZ-IHEK cells were infected with GRE.Luc reporter lentivirus as described previously [43], maintained in glucocorticoid-free medium for 12 hrs, and treated with either FA (10–6M) or vehicle (ethanol) for 24 hrs. Luciferase activity measured using Luciferase Promega assay (Promega Corp., Madison, WI, USA) and Luminometer TD 20/20, was standardized to protein content.

In silica analysis of KLF9 binding sites (GC box sequence: GGGCGG) in basal promoter of GRE.Luc reporter generated by NU SDRC DNA/RNA delivery Core: TAGGCGTGTACGGTGGGAGGCCTATATAAGCAGAGCTCGTTTAGTGAACCGTCAGA TCGCCTGGAGACGCCATCCACGCTGTTTTGACCTCCATAGAAGACACCGGGACCGATCCAGC was performed by Patch 1.0 tool and Transfac Public database (http://gene-regulation.com).

RNASeq data analysis for differentially expressed genes (DEGs)

The RNASeq experiments were performed on the Illumina NextSeq500 platform (single-end reads of 75bp length). Per-base quality was assessed with FastQC Software v0.11.5 (Andrews, 2016). Alignment of all raw files (fastq files) to the human genome (GRCh38) was done with the STAR aligner 2.3.1z (Dobin et al., 2013) and ENSEMBL91 GTF file annotation. The raw counts were processed with HTSeq (Anders et al., 2015), and normalized with the VST method of the DESeq2 R package (Love et al., 2014). After removing genes with zero counts and selecting the genes with expression of coefficient of variation > 10% across all samples, a total of 21,339 uniquely annotated genes were used for further statistical analysis.

To identify DEGs, a linear model (adjusted for paired samples, sex, race, age and batch) was implemented (limma R package (Ritchie et al., 2015), https://www.bioconductor.org/packages/devel/bioc/vignettes/limma/inst/doc/usersguide.pdf). DEGs were considered at FDR < 0.2.

Basal expression differences between sexes and races (in control samples) were considered significant at FDR < 0.2. GC-responsive DEGs in male and female (or African-American and Caucasian) were calculated separately at FDR < 0.2 and fold-change > 2 and kept the common ones. After comparing these common GC-responsive DEGs between sexes (or races), the significant genes were determined at Student’s t-test p-value < 0.05 (Supplemental Tables S4 and S5).

Weighted Gene Co-expression Network Analysis (WGCNA)

For the WGCNA module-trait analysis, we constructed a gene network using the residuals from the limma model from all samples. The traits used were: condition (GC-treated, control), sex (female, male), race (African-American, Caucasian) and age. For the hub identification, the top 35 genes with the highest inter-modular connectivity metrics (kWithin) in each module correlated with condition were selected. More than 90% of those top35 genes had also the highest kTotal connectivity metrics (Supplemental Table S6).

For the WGCNA module preservation analysis, we analyzed and compared two separate networks, one for the control and another for the GC samples. To do so, we separated the expression data into two tables with the same genes equally normalized for the control and the GC samples as follows: after removing zero counts across all samples, we kept the residuals from the limma model and the common genes between the two tables. To calculate the preservation of the control network onto the GC network, we used the above two tables and statistics per WGCNA manual (Supplemental Table S6).

Functional pathway enrichment for DEGs

Pathway enrichment analysis was done using the online tool of GSEA (Broad Institute) utilizing the GO molecular function and the Hallmark database gene sets (http://software.broadinstitute.org/gsea/msigdb/index.jsp). All pathways assessed by GSEA were enriched using cutoff FDR < 0.05.

Cell type enrichment analysis and GTRD database information

Cell type enrichment analysis was performed with xCell, a webtool (http://xcell.ucsf.edu/) that performs cell type enrichment analysis based on the gene expression data for 64 immune and stromal cell types (Aran et al., 2017). A Wilcoxon test, suitable for comparisons that are not normally distributed, was performed to compare enrichment scores of all cell types between control and CBP-treated samples. The cell type enrichment cutoff was at FDR <0.2 between control and CBP-treated skin.

For the putative GR target genes prediction results, we used the ChIP-Seq GTRD database that combines the results of over 7,000 ChIP-Seq experiments in human cells performed in ~30 cell types, and reports site count information for ~65,000 transcribed DNA regions with ENSEMBL IDs representing coding and non-coding RNAs, pseudogenes and exons. We employed GTRDs webtool (http://gtrd.biouml.org/) and selected meta-clusters and max gene distance of 5000 for the human transcription factor NR3C1 (glucocorticoid receptor). The GTRD clusters are not tissue specific and do not contain ChIP-Seq information from skin cell types.

Supplementary Material

1

Supplemental Table S2. Evaluation of cell type enrichment between CBP-treated and Control samples from full thickness skin biopsies.

2

Supplemental Table S3. Functional enrichment of 3,524 DEGs using the Hallmark and the GO databases of GSEA MaSigDB webtool (http://software.broadinstitute.org/gsea/msigdb/index.jsp) at FDR < 0.05

3

Supplemental Table S4. Sexual dimorphism in basal gene expression (control skin) and in skin response to topical CBP (pathway enrichment FDR < 0.05).

4

Supplemental Table S5. Racial dimorphism in basal gene expression (control skin) and dimorphism in skin response to topical CBP (pathway enrichment FDR < 0.05).

5

Supplemental Table S6. Network module preservation analysis focused on Transcription Factors affected by CBP in skin.

6

Supplemental Table S7. Skin SNP-gene association statistics with age-related expression changes of significant TFs (from network preservation statistics).

7

Supplemental Table S8. Overlap in the molecular signature (genes and pathways) of glucocorticoid effect in human skin (present study) and human keratinocytes in vitro (study with GEO accession number GSE264870).

8
9

Supplemental Table S1. Information about skin samples. CBP molecular signature in human skin: protein-coding and non-coding DEGs. Results of GR binding to the regulatory sequences of DEGs.

ACKNOWLEDGEMENTS

We acknowledge Northwestern University Genomics Facility/Sequencing Core, and Northwestern University SDRC (5 P30 AR057216) skin acquisition, skin tissue engineering, and DNA/RNA delivery Cores for technical support.

Funding

The current work is supported by R01GM112945 (IB and JTD), R01AI125366 (IB and JTD), R01DK098242 (JTD)

Abbreviations:

AA

African American

CBP

clobetasol propionate

CC

Caucasian

DEG

differentially expressed gene

FA

fluocinolone acetonide

GCs

glucocorticoids

GR

glucocorticoid receptor

GSEA

gene set enrichment analysis

IHEK

immortalized human epidermal keratinocytes

KLF9

Kruppel-like factor 9

qRT-PCR

quantitative real-time reversetranscription polymerase chain reaction

RNASeq

RNA sequencing

TA

transactivation

TR

transrepression

WGCNA

weighted gene co-expression network analysis

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

CONFLICT OF INTEREST

The authors state no conflict of interest.

DATA AVAILABILITY

All raw and pre-processed RNASeq data used in this study are deposited in the GEO database with GEO accession number: GSE120783. All other data generated or analyzed during this study are included in this published article and in the Appendix Files.

Ethics approval and consent to participate

Informed consent of all human volunteers in this study was obtained in writing, following the principles established in the WMA Declaration of Helsinki and the NIH Belmont Report (http://www.hhs.gov/ohrp/humansubjects/guidance/belmont.html).

Consent for publication

All studies were approved by the Northwestern University Institutional Review Board, approval number STU00009443.

REFERENCES

  1. Agarwal S, Mirzoeva S, Readhead B, Dudley JT, Budunova I. PI3K inhibitors protect against glucocorticoid-induced skin atrophy. EBioMedicine 2019. [DOI] [PMC free article] [PubMed]
  2. Ahluwalia A Topical glucocorticoids and the skin--mechanisms of action: an update. Mediators Inflamm 1998;7(3):183–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics 2015;31(2):166–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Andrews S FastQC. Babraham Bioinformatics; 2016.
  5. Aran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol 2017;18(1):220. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Austin PJ, Tsitsiou E, Boardman C, Jones SW, Lindsay MA, Adcock IM, et al. Transcriptional profiling identifies the long noncoding RNA plasmacytoma variant translocation (PVT1) as a novel regulator of the asthmatic phenotype in human airway smooth muscle. J Allergy Clin Immunol 2017;139(3):780–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Baida G, Bhalla P, Yemelyanov A, Stechschulte LA, Shou W, Readhead B, Dudley JT, Sanchez E, Budunova I. Deletion of the glucocorticoid receptor chaperone Fkbp5 prevents glucocorticoid-induced skin atrophy. Oncotarget, 2018; 9(78):34772–34783 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Baida G, Bhalla P, Kirsanov K, Lesovaya E, Yakubovskaya M, Yuen K, et al. REDD1 functions at the crossroads between the therapeutic and adverse effects of topical glucocorticoids. EMBO Mol Med 2015;7(1):42–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Balakrishnan-Renuka A, Morosan-Puopolo G, Yusuf F, Abduelmula A, Chen J, Zoidl G, et al. ATOH8, a regulator of skeletal myogenesis in the hypaxial myotome of the trunk. Histochem Cell Biol 2014;141(3):289–300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chebotaev D, Yemelyanov A, Budunova I. The mechanisms of tumor suppressor effect of glucocorticoid receptor in skin. Mol Carcinog 2007;46(8):732–40. [DOI] [PubMed] [Google Scholar]
  11. Chinenov Y, Coppo M, Gupte R, Sacta MA, Rogatsky I. Glucocorticoid receptor coordinates transcription factor-dominated regulatory network in macrophages. BMC Genomics 2014;15:656. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. de Jonge HJ, Fehrmann RS, de Bont ES, Hofstra RM, Gerbens F, Kamps WA, et al. Evidence based selection of housekeeping genes. PLoS One 2007;2(9):e898. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. de Jongh GJ, Zeeuwen PL, Kucharekova M, Pfundt R, van der Valk PG, Blokx W, et al. High expression levels of keratinocyte antimicrobial proteins in psoriasis compared with atopic dermatitis. J Invest Dermatol 2005;125(6):1163–73. [DOI] [PubMed] [Google Scholar]
  14. Del Rosso J, Friedlander SF. Corticosteroids: options in the era of steroid-sparing therapy. J Am Acad Dermatol 2005;53(1 Suppl 1):S50–8. [DOI] [PubMed] [Google Scholar]
  15. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013;29(1):15–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Frenkel B, White W, Tuckermann J. Glucocorticoid-Induced Osteoporosis. Adv Exp Med Biol 2015;872:179–215. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gershoni M, Pietrokovski S. The landscape of sex-differential transcriptome and its consequent selection in human adults. BMC Biol 2017;15(1):7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Glass D, Vinuela A, Davies MN, Ramasamy A, Parts L, Knowles D, et al. Gene expression changes with age in skin, adipose tissue, blood and brain. Genome Biol 2013;14(7):R75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Glotzer DJ, Zelzer E, Olsen BR. Impaired skin and hair follicle development in Runx2 deficient mice. Dev Biol 2008;315(2):459–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Gual AP-CI, Abeck D. Topical Corticosteroids in Dermatology: From Chemical Development to Galenic Innovation and Therapeutic Trends. Journal of Clinical & Experimental Dermatology Research 2015;6(2):1–5. [Google Scholar]
  21. Gupta AK, Carviel J, Abramovits W. Treating Alopecia Areata: Current Practices Versus New Directions. Am J Clin Dermatol 2017;18(1):67–75. [DOI] [PubMed] [Google Scholar]
  22. Guttsches AK, Balakrishnan-Renuka A, Kley RA, Tegenthoff M, Brand-Saberi B, Vorgerd M. ATOH8: a novel marker in human muscle fiber regeneration. Histochem Cell Biol 2015;143(5):443–52. [DOI] [PubMed] [Google Scholar]
  23. Halder RM, Nootheti PK. Ethnic skin disorders overview. J Am Acad Dermatol 2003;48(6 Suppl):S143–8. [DOI] [PubMed] [Google Scholar]
  24. Han H, Cho JW, Lee S, Yun A, Kim H, Bae D, et al. TRRUST v2: an expanded reference database of human and mouse transcriptional regulatory interactions. Nucleic Acids Res 2018;46(D1):D380–D6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hombach S, Kretz M. The non-coding skin: exploring the roles of long non-coding RNAs in epidermal homeostasis and disease. Bioessays 2013;35(12):1093–100. [DOI] [PubMed] [Google Scholar]
  26. Hu G, Niu F, Humburg BA, Liao K, Bendi S, Callen S, et al. Molecular mechanisms of long noncoding RNAs and their role in disease pathogenesis. Oncotarget 2018;9(26):18648–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Klopot A, Baida G, Bhalla P, Haegeman G, Budunova I. Selective Activator of the Glucocorticoid Receptor Compound A Dissociates Therapeutic and Atrophogenic Effects of Glucocorticoid Receptor Signaling in Skin. J Cancer Prev 2015;20(4):250–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kumar S, Goyal A, Gupta YK. Abuse of topical corticosteroids in India: Concerns and the way forward. J Pharmacol Pharmacother 2016;7(1):1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 2008;9:559. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lekva T, Bollerslev J, Kristo C, Olstad OK, Ueland T, Jemtland R. The glucocorticoid-induced leucine zipper gene (GILZ) expression decreases after successful treatment of patients with endogenous Cushing’s syndrome and may play a role in glucocorticoid-induced osteoporosis. J Clin Endocrinol Metab 2010;95(1):246–55. [DOI] [PubMed] [Google Scholar]
  31. Lekva T, Ueland T, Boyum H, Evang JA, Godang K, Bollerslev J. TXNIP is highly regulated in bone biopsies from patients with endogenous Cushing’s syndrome and related to bone turnover. Eur J Endocrinol 2012;166(6):1039–48. [DOI] [PubMed] [Google Scholar]
  32. Lesovaya E, Agarwal S, Readhead B, Vinokour E, Baida G, Bhalla P, et al. Rapamycin modulates glucocorticoid receptor function, blocks atrophogene REDD1, and protects skin from steroid atrophy. J Invest Dermatol 2018. [DOI] [PMC free article] [PubMed]
  33. Li G, Wang S, Gelehrter TD. Identification of glucocorticoid receptor domains involved in transrepression of transforming growth factor-beta action. The Journal of biological chemistry 2003;278(43):41779–88. [DOI] [PubMed] [Google Scholar]
  34. Liu F, Gong R, Lv X, Li H. The expression profiling and ontology analysis of non-coding RNAs in dexamethasone induced steatosis in hepatoma cell. Gene 2018;650:19–26. [DOI] [PubMed] [Google Scholar]
  35. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15(12):550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Mulholland DJ, Dedhar S, Coetzee GA, Nelson CC. Interaction of nuclear receptors with the Wnt/beta-catenin/Tcf signaling axis: Wnt you like to know? Endocr Rev 2005;26(7):898–915. [DOI] [PubMed] [Google Scholar]
  37. O’Shea JJ, Plenge R. JAK and STAT signaling molecules in immunoregulation and immune-mediated disease. Immunity 2012;36(4):542–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Oakley RH, Cidlowski JA. The biology of the glucocorticoid receptor: new signaling mechanisms in health and disease. J Allergy Clin Immunol 2013;132(5):1033–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Osorio KM, Lilja KC, Tumbar T. Runx1 modulates adult hair follicle stem cell emergence and maintenance from distinct embryonic skin compartments. J Cell Biol 2011;193(1):235–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Owens-Grillo JK, Hoffmann K, Hutchison KA, Yem AW, Deibel MR Jr., Handschumacher RE, et al. The cyclosporin A-binding immunophilin CyP-40 and the FK506-binding immunophilin hsp56 bind to a common site on hsp90 and exist in independent cytosolic heterocomplexes with the untransformed glucocorticoid receptor. The Journal of biological chemistry 1995;270(35):20479–84. [DOI] [PubMed] [Google Scholar]
  41. Petta I, Dejager L, Ballegeer M, Lievens S, Tavernier J, De Bosscher K, et al. The Interactome of the Glucocorticoid Receptor and Its Influence on the Actions of Glucocorticoids in Combatting Inflammatory and Infectious Diseases. Microbiol Mol Biol Rev 2016;80(2):495–522. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Ray A, Prefontaine KE. Physical association and functional antagonism between the p65 subunit of transcription factor NF-kappa B and the glucocorticoid receptor. Proceedings of the National Academy of Sciences of the United States of America 1994;91(2):752–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Rhen T, Cidlowski JA. Antiinflammatory action of glucocorticoids--new mechanisms for old drugs. The New England journal of medicine 2005;353(16):1711–23. [DOI] [PubMed] [Google Scholar]
  44. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43(7):e47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Rittie L, Fisher GJ. Natural and sun-induced aging of human skin. Cold Spring Harb Perspect Med 2015;5(1):a015370. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sarkar MK, Kaplan N, Tsoi LC, Xing X, Liang Y, Swindell WR, et al. Endogenous Glucocorticoid Deficiency in Psoriasis Promotes Inflammation and Abnormal Differentiation. J Invest Dermatol 2017;137(7):1474–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sawaya AP, Pastar I, Stojadinovic O, Lazovic S, Davis SC, Gil J, et al. Topical mevastatin promotes wound healing by inhibiting the transcription factor c-Myc via the glucocorticoid receptor and the long non-coding RNA Gas5. The Journal of biological chemistry 2018;293(4):1439–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Schacke H, Berger M, Rehwinkel H, Asadullah K. Selective glucocorticoid receptor agonists (SEGRAs): novel ligands with an improved therapeutic index. Mol Cell Endocrinol 2007;275(1–2):109–17. [DOI] [PubMed] [Google Scholar]
  49. Schoepe S, Schacke H, Bernd A, Zoller N, Asadullah K. Identification of novel in vitro test systems for the determination of glucocorticoid receptor ligand-induced skin atrophy. Skin Pharmacol Physiol 2010;23(3):139–51. [DOI] [PubMed] [Google Scholar]
  50. Sporl F, Korge S, Jurchott K, Wunderskirchner M, Schellenberg K, Heins S, et al. Kruppel-like factor 9 is a circadian transcription factor in human epidermis that controls proliferation of keratinocytes. Proceedings of the National Academy of Sciences of the United States of America 2012;109(27):10903–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Staubert C, Bhuiyan H, Lindahl A, Broom OJ, Zhu Y, Islam S, et al. Rewired metabolism in drug-resistant leukemia cells: a metabolic switch hallmarked by reduced dependence on exogenous glutamine. The Journal of biological chemistry 2015;290(13):8348–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Stern RS. The pattern of topical corticosteroid prescribing in the United States, 1989–1991. J Am Acad Dermatol 1996;35(2 Pt 1):183–6. [DOI] [PubMed] [Google Scholar]
  53. Sternberg TH, Newcomer VD, Linden IH. Treatment of atopic dermatitis with cortisone. J Am Med Assoc 1952;148(11):904–7. [DOI] [PubMed] [Google Scholar]
  54. Stojadinovic O, Lee B, Vouthounis C, Vukelic S, Pastar I, Blumenberg M, et al. Novel genomic effects of glucocorticoids in epidermal keratinocytes: inhibition of apoptosis, interferon-gamma pathway, and wound healing along with promotion of terminal differentiation. The Journal of biological chemistry 2007;282(6):4021–34. [DOI] [PubMed] [Google Scholar]
  55. Sugden MC, Holness MJ. Recent advances in mechanisms regulating glucose oxidation at the level of the pyruvate dehydrogenase complex by PDKs. Am J Physiol Endocrinol Metab 2003;284(5):E855–62. [DOI] [PubMed] [Google Scholar]
  56. Sundahl N, Bridelance J, Libert C, De Bosscher K, Beck IM. Selective glucocorticoid receptor modulation: New directions with non-steroidal scaffolds. Pharmacol Ther 2015;152:28–41. [DOI] [PubMed] [Google Scholar]
  57. Tai PK, Albers MW, Chang H, Faber LE, Schreiber SL. Association of a 59-kilodalton immunophilin with the glucocorticoid receptor complex. Science 1992;256(5061):1315–8. [DOI] [PubMed] [Google Scholar]
  58. Takahashi Y, Shinoda A, Kamada H, Shimizu M, Inoue J, Sato R. Perilipin2 plays a positive role in adipocytes during lipolysis by escaping proteasomal degradation. Sci Rep 2016;6:20975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Vandewalle J, Luypaert A, De Bosscher K, Libert C. Therapeutic Mechanisms of Glucocorticoids. Trends Endocrinol Metab 2018;29(1):42–54. [DOI] [PubMed] [Google Scholar]
  60. Xing L, Dai Z, Jabbari A, Cerise JE, Higgins CA, Gong W, et al. Alopecia areata is driven by cytotoxic T lymphocytes and is reversed by JAK inhibition. Nat Med 2014;20(9):1043–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Ye S, Li P, Tong C, Ying QL. Embryonic stem cell self-renewal pathways converge on the transcription factor Tfcp2l1. EMBO J 2013;32(19):2548–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Yemelyanov A, Czwornog J, Chebotaev D, Karseladze A, Kulevitch E, Yang X, et al. Tumor suppressor activity of glucocorticoid receptor in the prostate. Oncogene 2007;26(13):1885–96. [DOI] [PubMed] [Google Scholar]
  63. Yevshin I, Sharipov R, Valeev T, Kel A, Kolpakov F. GTRD: a database of transcription factor binding sites identified by ChIP-seq experiments. Nucleic Acids Res 2017;45(D1):D61–D7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yin L, Coelho SG, Ebsen D, Smuda C, Mahns A, Miller SA, et al. Epidermal gene expression and ethnic pigmentation variations among individuals of Asian, European and African ancestry. Exp Dermatol 2014;23(10):731–5. [DOI] [PubMed] [Google Scholar]
  65. Zhang Y, Davidson BR, Stamer WD, Barton JK, Marmorstein LY, Marmorstein AD. Enhanced inflow and outflow rates despite lower IOP in bestrophin-2-deficient mice. Invest Ophthalmol Vis Sci 2009;50(2):765–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Zhang Z, Jones S, Hagood JS, Fuentes NL, Fuller GM. STAT3 acts as a co-activator of glucocorticoid receptor signaling. The Journal of biological chemistry 1997;272(49):30607–10. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

Supplemental Table S2. Evaluation of cell type enrichment between CBP-treated and Control samples from full thickness skin biopsies.

NIHMS1532721-supplement-1.xlsx (11.2KB, xlsx)
2

Supplemental Table S3. Functional enrichment of 3,524 DEGs using the Hallmark and the GO databases of GSEA MaSigDB webtool (http://software.broadinstitute.org/gsea/msigdb/index.jsp) at FDR < 0.05

NIHMS1532721-supplement-2.xlsx (12.7KB, xlsx)
3

Supplemental Table S4. Sexual dimorphism in basal gene expression (control skin) and in skin response to topical CBP (pathway enrichment FDR < 0.05).

NIHMS1532721-supplement-3.xlsx (170.7KB, xlsx)
4

Supplemental Table S5. Racial dimorphism in basal gene expression (control skin) and dimorphism in skin response to topical CBP (pathway enrichment FDR < 0.05).

NIHMS1532721-supplement-4.xlsx (119.4KB, xlsx)
5

Supplemental Table S6. Network module preservation analysis focused on Transcription Factors affected by CBP in skin.

NIHMS1532721-supplement-5.xlsx (13.4MB, xlsx)
6

Supplemental Table S7. Skin SNP-gene association statistics with age-related expression changes of significant TFs (from network preservation statistics).

NIHMS1532721-supplement-6.xlsx (4.3MB, xlsx)
7

Supplemental Table S8. Overlap in the molecular signature (genes and pathways) of glucocorticoid effect in human skin (present study) and human keratinocytes in vitro (study with GEO accession number GSE264870).

NIHMS1532721-supplement-7.xlsx (177.4KB, xlsx)
8
NIHMS1532721-supplement-8.docx (21.4KB, docx)
9

Supplemental Table S1. Information about skin samples. CBP molecular signature in human skin: protein-coding and non-coding DEGs. Results of GR binding to the regulatory sequences of DEGs.

NIHMS1532721-supplement-9.xlsx (1MB, xlsx)

RESOURCES