REDD1 functions at the crossroads between the therapeutic and adverse effects of topical glucocorticoids

Abstract

Cutaneous atrophy is the major adverse effect of topical glucocorticoids; however, its molecular mechanisms are poorly understood. Here, we identify stress-inducible mTOR inhibitor REDD1 (regulated in development and DNA damage response 1) as a major molecular target of glucocorticoids, which mediates cutaneous atrophy. In REDD1 knockout (KO) mice, all skin compartments (epidermis, dermis, subcutaneous fat), epidermal stem, and progenitor cells were protected from atrophic effects of glucocorticoids. Moreover, REDD1 knockdown resulted in similar consequences in organotypic raft cultures of primary human keratinocytes. Expression profiling revealed that gene activation by glucocorticoids was strongly altered in REDD1 KO epidermis. In contrast, the down-regulation of genes involved in anti-inflammatory glucocorticoid response was strikingly similar in wild-type and REDD1 KO mice. Integrative bioinformatics analysis of our and published gene array data revealed similar changes of gene expression in epidermis and in muscle undergoing glucocorticoid-dependent and glucocorticoid-independent atrophy. Importantly, the lack of REDD1 did not diminish the anti-inflammatory effects of glucocorticoids in preclinical model. Our findings suggest that combining steroids with REDD1 inhibitors may yield a novel, safer glucocorticoid-based therapies.

Keywords: REDD1; glucocorticoid; glucocorticoid receptor; mTOR; skin atrophy.

Figure 1. Topical glucocorticoids induce epidermal atrophy and REDD1 mRNA expression

1. A–G B6D2 mice were treated topically with acetone (vehicle control) or glucocorticoid FA (2 μg/animal), every 72 h for 2 weeks. Human volunteers were treated with 0.05% CPB cream applied to the right arm skin once or daily for 2 weeks. Untreated skin from the left arm was used as a control. H&E staining of mouse skin (A) and human skin (E). Scale bars are 20 μm (A) and 40 μm (E). Morphometric analysis of epidermal thickness and number of basal keratinocytes in mouse skin treated with FA 1, 2, and 4 times (B), and human skin treated daily for 2 weeks (F). REDD1 mRNA expression in mouse epidermis 8 h after 1^st, 2^nd, and 4^th applications of FA (C, Q-PCR), in mouse epidermis and s.c. adipose, 24 h after FA (D, Q-PCR), and in human skin 24 h after single (G, volunteers V4 and V5) and 2-week (G, volunteers V1, V2, V3, RT–PCR) treatment with CBP. RPL27 was used as a cDNA normalization control. In human skin, the means ± SD were calculated in each individual sample compared to the untreated skin from the same individual (30 measurements/condition). In mouse skin, the means ± SD were calculated for three individual skin samples/condition in one representative experiment (30 measurements/condition) out of three experiments. Q-PCR results are the means ± SD calculated for three individual RNA samples/condition. Statistical analysis for differences between treatment and control was done by the unpaired two-tailed t-test.

Figure 2. REDD1 protein expression in epidermis is tightly regulated and correlates with inhibition of mTOR and autophagy

B6D2 mice were treated topically as in Fig1. Skin was harvested 4–24 h after single FA application or during chronic treatment (8 h after 1^st, 2^nd, and 4^th applications).

1. A Immunohistochemical staining of mouse skin treated with acetone or FA for 8 h for REDD1 (upper panels) and phosphorylated mTOR-Ser2448 (lower panels). Scale bars are 20 μm.

2. B Western blot analysis of REDD1 protein and phosphorylation of down-stream mTOR target proteins 4E-BP1 and rpS6 in murine epidermis. GAPDH is used as a normalization control.

3. C, D Glucocorticoids induce autophagy in epidermis. Western blot analysis of Beclin-1 and conversion of light chain 3 (LC3) from LC3-I to LC3-II (C). Quantification of LC3-I to LC3-II conversion and Beclin-1 expression (D). The means ± SD were calculated using Western blots from two independent experiments (each lane is whole-cell protein from three pulled individual samples of epidermis).

Figure 3. REDD1 KO mice are resistant to glucocorticoid-induced skin atrophy

REDD1 KO and isogenic wild-type B6x129 mice were treated with acetone (vehicle control) or FA (2 μg/animal) every 72 h for 2 weeks.

1. A, B H&E staining (A) and Masson's trichrome staining: Dermis/collagen fibers are blue, muscle is red, nuclei are dark red, and cytoplasm is red/pink (B). Arrows point to epidermis and brackets indicate subcutaneous adipose (A) and dermis (D). Scale bars are 20 μm.

2. C, D Morphometric analysis of epidermal thickness and dermal cellularity as described in Materials and Methods. Changes in epidermal thickness (C) are presented as % to wild-type control epidermis. Changes in dermal cellularity (D) are presented as % to corresponding control skin. The means ± SD were calculated for three individual skin samples in one representative experiment (30 measurements/condition) out of two experiments. Statistical analysis for differences between treatment and control and between control wild-type and REDD1 KO epidermal thickness was done by the unpaired two-tailed t-test.

Figure 4. Protective effect of REDD1 KO on p63⁺ progenitors and CD34⁺ follicular epithelial stem cells

REDD1 KO and wild-type animals were treated as in Fig3.

1. A–C Expression of p63 (A) and CD34 (C). Scale bars are 10 μm. Analysis of p63 staining (B). The number of p63⁺ basal keratinocytes/total number of basal keratinocytes is presented as % to the corresponding control epidermis.

2. D Similar GR expression in epidermis of wild-type and REDD1 KO mice determined by Q-PCR and Western blotting. Rpl27 and GAPDH used as a normalization controls, respectively.

Data information: The means ± SD were calculated for three individual skin samples in one representative experiment (30 measurements/condition) out of two experiments. Q-PCR results are presented as the means ± SD for three individual RNA samples/condition. Statistical analysis for differences between treatment and control was done by the unpaired two-tailed t-test.

Figure 5. Similar sensitivity of REDD1 KO and wild-type animals to the anti-inflammatory effect of glucocorticoids

Ear edema was induced by croton oil (CO) as in Materials and Methods. FA was applied 1 h before CO, and four-millimeter ear punch was weighed 9 h after CO application to assess swelling. In the additional experiment, ears were harvested 9 h after CO application and used for histological analysis.

1. A H&E staining. Scale bars are 20 μm.

2. B Ear punch weight. Results are presented as % to corresponding (wild-type or REDD1 KO) control ear weight. The means ± SD were calculated for six individual ear punches/condition in one representative experiment (out of three experiments). Statistical analysis for differences between treatment and corresponding control was done by the unpaired two-tailed t-test.

Figure 6. Down-regulation of REDD1 protects organotypic raft cultures from the hypoplastic effect of glucocorticoids

Organotypic raft cultures (ORC) made from NHEK infected with REDD1 shRNA and control pGIPZ lentiviruses were treated with glucocorticoid CBP (5 μM) or vehicle control (0.05% DMSO) for 7 days.

1. A, B H&E and BrdU staining of raft cultures (BrdU⁺ cells are indicated by arrowheads). Scale bars are 10 μm.

2. C Western blot analysis of REDD1, GR, and mTOR substrate 4E-BP1 phosphorylation. GAPDH was used as a loading control.

3. D Q-PCR analysis of REDD1 expression in rafts.

4. E Analysis of CBP effect on basal keratinocyte number (left) and keratinocyte proliferation (number of BrdU⁺ basal keratinocytes/total number of basal keratinocytes, right) is presented as % to corresponding control rafts.

Data information: The means ± SD for BrdU⁺ cells and basal keratinocytes in (E) were calculated for two individual rafts in one representative experiment (20 measurements/condition) out of two experiments. Q-PCR results in (D) are the means ± SD calculated for two individual RNA samples/condition. Statistical analysis for differences between treatment and control was done by the unpaired two-tailed t-test.

Figure 7. Global effect of REDD1 KO on the expression of glucocorticoid-responsive genes in murine epidermis

REDD1 KO and isogenic B6x129 mice were treated topically with FA (2 µg) or vehicle (acetone) for 24 h, and RNA was extracted from the epidermis and used for microarray analysis.

1. A A heatmap of gene expression as analyzed by the Mouse Whole-Genome Gene Expression BeadChips (Illumina). Columns show normalized gene expression for individual animals (red: increased, green: decreased expression).

2. B Venn diagrams show overlap of differentially expressed genes between REDD1 KO and wild-type mice (adjusted P < 0.05). Arrows connect sections of the heatmap with corresponding Venn diagrams.

3. C Most enriched gene ontology (GO) categories for biological processes and molecular functions (fold-change enrichment ≥ 3, P < 0.01), associated with transrepression and transactivation of the glucocorticoid-responsive genes in REDD1 KO and wild-type mice.

4. D Array validation by Q-PCR for genes from the different GO categories. Note: lack of overlap in GO categories between KO and wild-type mice associated with up-regulated genes and significant overlap in the function of down-regulated genes.

Data information: We used for array analysis two individual RNA samples/condition. Statistical analysis of DNA arrays is described in Materials and Methods. Statistical analysis of array validation is shown in Supplementary Fig S3.

Figure 8. REDD1 inhibition did not affect GR phosphorylation and nuclear translocation in HaCaT human keratinocytes

HaCaT human keratinocytes were infected with shREDD1 or pGIPZ (control) lentiviruses and treated with glucocorticoid FA (10⁻⁶ M) for the indicated time.

1. Western blot analysis of REDD1, GR, and phosphorylated GR-Ser211. GAPDH used as a loading control.

2. Reduced induction of Luciferase reporter in shREDD1-HaCaT cells. shREDD1- and pGIPZ-HaCaT cells were infected with GRE.Luc lentivirus and treated with FA (10⁻⁶ M) for 24 h. The Luciferase induction is presented as a fold change to corresponding vehicle-treated control. The means ± SD were calculated for three individual wells/group in one representative experiment (out of three experiments). Statistical analysis for differences between groups was done by ANOVA.

3. Immunofluorescence analysis of GR nuclear translocation and retention in shREDD1- and pGIPZ-HaCaT cells treated with FA (10⁻⁶ M). Scale bars are 10 μm.