Molecular interactions of glucocorticoid and mineralocorticoid receptors define novel transcription and biological functions

Abstract

Glucocorticoids are primary stress hormones necessary for life that function to maintain homeostasis. These hormones and their synthetic derivatives are widely used in the clinic to combat disease but are limited by development of resistance and by severe side effects. Understanding how glucocorticoids signal is crucial for developing safer and more effective glucocorticoids. Mechanistically glucocorticoid ligands induce glucocorticoid receptor (GR) homodimerization and regulation of gene expression. Here, we show that GR and mineralocorticoid receptor (MR) form molecular complexes with distinct transcriptional responses that alter the biological roles of GR. MR inhibited GR interaction with genomic DNA and diminished glucocorticoid-regulated gene expression as well as suppressed cell apoptosis induced by GR signaling. Provocatively, multiple therapeutic glucocorticoids differentially induced the GR-MR interaction revealing unknown drug effects that are exploitable for fine-tuning glucocorticoid drug treatments. Molecular modeling of the GR-MR complex predicted an interaction interface residing in the LBD of both GR and MR. Mutation of a key amino acid in the interface of GR compromised GR-MR interaction without affecting GR activity in a gene reporter assay. Overall, our findings uncovered unique crosstalk mechanisms between distinct nuclear receptors providing a novel mechanism of diversity in the action of glucocorticoids that may contribute to context-dependent GR signaling in human health and disease.

Keywords: apoptosis; glucocorticoid receptor; mineralocorticoid receptor; protein–protein interaction; steroid hormone receptor; transcription regulation.

Figure 1

Establishing GR, MR, and MR/GR cell lines.A, Venn diagrams on heart and hippocampus gene expression change in GRKO, MRKO, and dKO mice compared to WT mice (32, 33). Much larger numbers of genes exhibited altered expressions in GR and MR double KO compared with GRKO or MRKO thus suggesting that the two receptors have functional interactions in these organs for the gene regulations. B, schematic diagram showing genealogy of cell lines used in this study. C and D, GR and MR mRNA (C) and protein (D) expression in the cell lines used in this study. Data are presented as means ± SD with individual data points of technical replicates. The data were analyzed using one-way ANOVA with Dunnett’s post hoc test. ∗p < 0.05, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. E, F, and G, GR and MR protein localizations in the cell lines were determined by immunofluorescence histochemistry. The upper panels show the immunofluorescence histochemistry of receptors, and the lower panels show the quantifications of GR and MR localizations. E, GR protein localization in GR cells, (F) MR protein localization in MR cells, and (G) GR and MR protein localization in MR/GR cells before and after treatment with DEX (1 nM, 6 h). The scale bar represents 10 μm. Quantified data in (E), (F), and (G) are presented as mean ± SD with individual data points of technical replicates. The data were analyzed using Student’s t test. ∗∗p < 0.01 and ∗∗∗p < 0.001. H, coimmunoprecipitation of GR and MR protein extracted from MR/GR cells. I, proximity ligation assay (PLA) found GR and MR exist in close range in MR/GR cells. PLA signals in red, anti-F-actin staining in green, and nuclear staining with 4′,6-diamidino-2-phenylindole in blue. The scale bar represents 10 μm. ; DEX, dexamethasone; GR, glucocorticoid receptor; GRKO, GR knockout mice; MR, mineralocorticoid receptor; MRKO, MR knockout mice.

Figure 2

RNA-seq analysis of DEX-regulated genes in U-off, GR, MR and MR/GR cells U-off, MR, GR, and MR/GR cell lines were treated with 1 nM or 10 nM DEX and their gene expressions were measured by RNA-seq analysis.A, PCA results for U-off, MR, GR, and MR/GR cell lines treated with 1 nM DEX (left panel) or 10 nM DEX (right panel). B, induced (red) or suppressed (blue) gene numbers in GR, MR, and MR/GR cells treated with 1 nM or 10 nM DEX are shown in pie charts. C, Venn diagram for DEX-regulated genes in GR, MR, and MR/GR cells. D, MR blunted DEX-induced GR-mediated gene regulation. Upper panels: Venn diagram for DEX-regulated genes in GR and MR/GR cells. Bottom panels: Experimental log ratio of genes induced or suppressed commonly in GR cells and MR/GR cells were plotted. These plots suggested MR coexpression made gene induction or suppression less sensitive to 1 nM or 10 nM DEX treatments. E, as in Figure 3D, gene expression profiles were compared between two cell lines in Venn diagrams (top: GR cell versus MR cell, bottom: MR cell versus MR/GR cell). Experimental log ratios of commonly regulated genes were plotted in the same fashion as in Figure 3D and indicated genes in MR cells respond less compared to those in GR cells or MR/GR cells. DEX, dexamethasone; GR, glucocorticoid receptor; MR, mineralocorticoid receptor; PCA, principal component analysis.

Figure 3

Pathway analyses of GR cell and MR/GR cell depicted blunting effect of MR coexpression.A, canonical pathway analysis results (1 nM DEX treatment for the top panel and 10 nM DEX treatment for the bottom panel) by ingenuity pathway analysis (IPA). Pathways which have the top 20 absolute Z-scores in GR cells were compared among GR, MR, and MR/GR cells. The pathways were plotted according to their −log10 (p value). X-axis represents −log10 (pvalue), circle size indicates number of genes, and color for the circle decodes for Z-score for each canonical pathway. B, comparisons of IPA diseases and function analysis between GR cell and MR/GR cell. Cellular functions with absolute Z-score bigger than 2 in either one of the two cell lines are shown as heat maps. DEX, dexamethasone; GR, glucocorticoid receptor; MR, mineralocorticoid receptor.

Figure 4

MR coexpression suppresses DEX-induced apoptosis in GR-expressing cells.A, heat map showing diminished modulation of DEX-regulated genes related to apoptosis in MR/GR cells compared to those in GR cells. B, DEX induced cell death in GR cell, MR cell, MR/GR cell, and MR siRNA-treated MR/GR (MR/GR MR-KD) cells. MR coexpression suppressed the DEX-induced apoptosis in GR cells and apoptosis re-emerged when MR expression was suppressed. Cells were incubated with 100 nM DEX for 48 h. C, FACS quantification of cell death data shown in (B). D, MR mRNA and protein expressions in cell lines used for cell death analyses in (B). siRNA treatment of MR/GR cells (MR/GR MR-KD) effectively suppressed MR expression compared with control siRNA treatment (MR/GR NTC). E, propidium iodide negative and annexin-V positive cells quantification of cell lines treated with DEX as in (B) suggested cell death was caused by apoptotic mechanisms. F, PARP cleavage in the cells used in (B) double confirmed apoptosis is causing cell death observed in (B). Cleaved PARP and PARP amounts were determined by Western blotting as shown in the right panel. All quantified data in (B), (D), (E), and (F) are presented as mean ± SD with individual data points of technical replicates. The data were analyzed using one-way ANOVA with Sidak’s post hoc test. ∗p < 0.05, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. DEX, dexamethasone; GR, glucocorticoid receptor; MR, mineralocorticoid receptor.

Figure 5

NanoBit assay revealed direct interaction of GR and MR in living cells.A, in combination of GR and MR with Nanobit luciferase derived fragments (LgBit and SmBit), total of 8 constructs were produced. All combinations (1, 2, 3, 4, 5, 6, 7, 8) were examined for reconstituted NanoBit luciferase activity. Only two of them with the luciferase fragments existing at the C-terminal side of the receptors (4, 8) showed robust luciferase activity by the additions of ligands (DEX and aldosterone) at 25 min after starting the measurement of the luciferase activity. B, single ligand (DEX, aldosterone, or cortisol, 100 nM) induced GR–MR interaction, and further heterodimerization was observed by DEX and aldosterone added simultaneously. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Tukey’s post hoc test. Each treatment group showed significant induction of Nanoluc luciferase activity compared to activities in control (∗∗∗∗p < 0.0001 at shown time periods). DEX + aldosterone-induced activity was higher compared to those in DEX or aldosterone single ligand treatments (####p < 0.0001 31–68 min). C–F, zinc finger mutations affect differentially GR–MR heterodimerization and GR–GR homodimerization. C, GR zinc finger DNA-binding domain and three critical amino acid mutations for DNA binding (42). D, GRE2 reporter gene activation by WT GR or the three mutants shown in (C). Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Sidak’s post hoc test. ∗∗∗∗p < 0.0001. E, the three mutants compromised GR–GR homodimerization activity. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Sidak’s post hoc test. (∗∗∗∗p < 0.0001). F, compared to WT GR, three zinc finger mutants retained partial heterodimerization activity with MR. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Sidak’s post hoc test. (∗∗∗∗p < 0.0001). G–J, therapeutic glucocorticoids induced GR–MR heterodimer. G, beclomethasone and prednisolone but not triamcinolone induced GR–MR heterodimer formation. Each drug was added (final concentration 100 nM) into the culture media and monitored Nanoluc luciferase activity time course changes. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Tukey’s post hoc test. (∗∗∗∗p < 0.0001 29–65 min compared to control. ####p < 0.0001 significant difference between beclomethasone and prednisolone during 31–65 min time points). H and I, aldosterone (100 nM) cotreatments with beclomethasone (100 nM, panel H) or triamcinolone (100 nM, panel I) synergistically induced GR–MR heterodimer. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Tukey’s post hoc test. ∗∗∗∗p < 0.0001 between double ligands and each ligand. J, cortisol (100 nM) partially suppressed prednisolone (100 nM) induced heterodimer formation. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Tukey’s post hoc test. ∗∗∗∗p < 0.0001 between cortisol and cortisol + prednisolone treatments during 36 to 65 min. DEX, dexamethasone; GR, glucocorticoid receptor; MR, mineralocorticoid receptor.

Figure 6

MR-suppressed GR interaction with genome in MR/GR cells.A–D, ChIP-seq analysis revealed attenuated GR binding to DNA elements by MR coexpression. A, heat map presentation of GR binding to DNA elements in GR cells and MR/GR cells. B, Venn diagram showing overlapping GR binding peaks of MR cells and MR/GR cells found in ChIP-seq analysis. MR coexpression suppressed GR binding to the genome. C, heat map presentation of the GR binding peaks categorized by Venn diagram shown in (B). D, examples of ChIP-seq identified peaks in KCNA5 and KLF15 genes that are in the “unique” and “common” peaks category, respectively. Locations of identified GR binding sites in chromosomes for these genes are shown at the bottom. E, qPCR quantification of KCNA5 and KLF15 mRNA in GR cell and MR/GR cell with or without DEX (100 nM, 6 h) treatment. Data are presented as mean ± SD with individual data points of technical replicates. The data were analyzed using two-way ANOVA with uncorrected Fisher’s least significant difference post hoc test. ∗∗p < 0.01, ∗∗∗∗p < 0.0001. ^###In MR/GR cell, DEX induction of KLF15 gene was nonsignificant by two-way ANOVA (0.0898). Student’s t test suggested p = 0.00412. ChIP-seq, chromatin immpunoprecipitation sequencing; DEX, dexamethasone; GR, glucocorticoid receptor; MR, mineralocorticoid receptor.

Figure 7

Molecular dynamics simulation of GR–MR LBD heterodimer.A, a representative solution structure of GR/MR heterodimer. GR residues are shown in red and white ribbons and MR in blue and cyan. Dimer-interacting surfaces of GR and MR are comprised of white and cyan ribbons and it involves the C termini of GR and MR marked in the circles with “C” inside. B, RMSDs of GR (red), MR (green), and the GR/MR dimer (black) from the three independent molecular dynamics simulations of 1.2 μs each. The trajectory represented in the bottom panel shows an unstable GR/MR heterodimer interface. C, residue RMS fluctuations of the backbone heavy atoms for each residue of GR (top) and MR (bottom) from the two molecular dynamics trajectories (red and green) with stable dimer interface. D, dynamic cross correlation matrices of GR and MR monomers. E, GR/MR dynamic cross correlations. Interacting residue ranges are marked. F, interaction energies decomposed at the residue level for GR residues using MMGBSA calculations. Labels “Sample 1” and Sample 2” are energies from the two trajectories with stable dimer interacting interfaces. Averages and error bars were calculated from the 1200 sample points collected over each trajectory. G and H, GR Trp 712 mutation attenuated GR–MR interaction in NanoBit assay. MR and WT GR (G) or MR and GR W712A (H) interaction was analyzed by NanoBit assay. Data from one 96-well plate assay were shown in two panels. Data are presented as mean ± SD. The data were analyzed using two-way ANOVA with Tukey’s post hoc test. (∗∗∗∗p < 0.0001 during 30–67 min between the luciferase activities of MR interactions with WT or W712A with the same ligand treatment). I and H, GR Trp 712 mutation did not affect GR–GR interaction in NanoBit assay. WT GR-SmBit and WT GR-LgBit (I) or WT GR-SmBit and GR W712A-LgBit (J) interaction was analyzed by NanoBit assay. K, Trp 712 mutation to Ala did not affect GR activity for GRE2 luciferase reporter gene. Data are presented as mean ± SD. The data were analyzed using one-way ANOVA with Sidak’s post hoc test. ∗∗∗∗p < 0.0001. GR, glucocorticoid receptor; LBD, ligand-binding domain; MR, mineralocorticoid receptor.

Fig S1

Fig S2

Fig S3

Fig S4

Fig S5