Selective modulation of the glucocorticoid receptor can distinguish between transrepression of NF-κB and AP-1

Abstract

Glucocorticoids (GCs) block inflammation via interference of the liganded glucocorticoid receptor (GR) with the activity of pro-inflammatory transcription factors NF-κB and AP-1, a mechanism known as transrepression. This mechanism is believed to involve the activity of GR monomers. Here, we explored how the GR monomer-favoring Compound A (CpdA) affects AP-1 activation and activity. Our results demonstrate that non-steroidal CpdA, unlike classic steroidal GCs, blocks NF-κB- but not AP-1-driven gene expression. CpdA rather sustains AP-1-driven gene expression, a result which could mechanistically be explained by the failure of CpdA to block upstream JNK kinase activation and concomitantly also phosphorylation of c-Jun. In concordance and in contrast to DEX, CpdA maintained the expression of the activated AP-1 target gene c-jun, as well as the production of the c-Jun protein. As for the underlying mechanism, GR is a necessary intermediate in the CpdA-mediated gene expression of AP-1-regulated genes, but seems to be superfluous to CpdA-mediated JNK phosphorylation prolongation. The latter phenomenon concurs with the inability of CpdA to stimulate DUSP1 gene expression. ChIP analysis demonstrates that DEX-activated GR, but not CpdA-activated GR, is recruited to AP-1-driven promoters. Furthermore, in mice we observed that CpdA instigates a strong enhancement of TNF-induced AP-1-driven gene expression. Finally, we demonstrate that this phenomenon coincides with an increased sensitivity towards TNF lethality, and implicate again a role for JNK2. In conclusion, our data support the hypothesis that a ligand-induced differential conformation of GR yields a different transcription factor cross-talk profile.

Fig. 1

CpdA-mediated transrepression of the IL-6 gene promoter in fibroblast cells is only efficient in absence of a functional AP-1-response element. a, b Subconfluent L929sA cell monolayers stably transfected with the indicated promoter reporter gene constructs were grown in 24-well plates. The point-mutated variant is indicated by its mutated transcription factor-binding site, i.e., AP-1, between brackets. c Similar to panels (a) and (b), but with inductions performed on L929sA cells with a stably integrated recombinant (IL6κB)₃50hu.IL6P-luc+ reporter construct. Cells were left untreated or treated with 2,000 IU/ml TNF, for 5 h, preceded by a 1 h treatment with solvent (as indicated by the minus sign), DEX (1 μM) or CpdA (1 or 10 μM). At the end of the induction, cell lysates were assayed for reporter gene activities. Total solvent concentration was kept similar in all conditions. The experiments are carried out in triplicate or quadruplicate. Results are shown ± SD and are representative of two to four independent experiments. **p < 0.01, ***p < 0.001. For (a) comparisons versus the control lane are depicted by ‘#’ and comparisons versus the respective pro-inflammatory stimulus TNF are depicted by ‘§’. For b and c, comparisons were made vs. TNF

Fig. 2

Activation of IL-6 expression parameters by combined NF-κB- and AP-1-activating stimuli is repressed by DEX, but to a lesser extent by CpdA. a Subconfluent L929sA cell monolayers, stably integrated with the p1168hu.IL6P-luc+ reporter gene construct, were left untreated or treated with 2,000 IU/ml TNF, with 60 nM STS or a combination hereof, for 5 h, preceded by a 1-h treatment with solvent, DEX (1 μM) or CpdA (1 or 10 μM). At the end of the induction, cell lysates were assayed for reporter gene activities. The experiment was carried out in quadruplicate, and the results are representative of three independent experiments. ***p < 0.001. Comparisons versus the control lane are depicted by ‘#’, comparisons versus the respective pro-inflammatory stimuli are depicted by ‘§’. b L929sA cells, starved for 48 h in DMEM devoid of serum, were pre-incubated with solvent, DEX (1 μM), or CpdA (1 or 10 μM) for 1 h, before STS (60 nM) and TNF (2,000 IU/ml) were added, where indicated, for 6 h. Total RNA was isolated and subjected to RT-qPCR assaying IL-6 and two household gene mRNA levels. Specific signal for cDNA of IL-6 was normalized to the averaged household genes signal. The STS/TNF condition was set as 100 and all other conditions were recalculated accordingly to allow ratio comparisons. Total solvent concentration was kept similar in all conditions. Results are shown ± SD. Comparisons were made vs. STS/TNF. The experiment was carried out in triplicate, and the result are averages of two independent experiments. c L929sA cells were induced as in b. Medium was collected to perform a murine IL-6 ELISA. Protein levels are presented as pg per ml. Total solvent concentration was kept similar in all conditions. Results are shown ± SD. ***p < 0.001. Comparisons were made vs. STS/TNF

Fig. 3

AP-1-driven gene transcription is downregulated by DEX but not by CpdA. a, b Subconfluent L929sA cell transiently transfected with pAP-1-Luc or pCollagenase-3-Luc were untreated or treated with 2,000 IU/ml TNF, with 60 nM STS, with 50 ng/ml TPA or a combination hereof, for 5 h, preceded by a 1-h treatment with solvent, DEX (1 μM) or CpdA (1 or 10 μM). At the end of the induction, cell lysates were assayed for reporter gene activities. The experiment was carried out in triplicate, and the results are representative of at least two independent experiments. Total solvent concentration was kept similar in all conditions. Results are shown ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. Comparisons were made vs. the respective pro-inflammatory stimuli

Fig. 4

Predominantly AP-1-regulated target genes are only transrepressed by DEX- and not by CpdA-activated GR, but promoter complexity determines the final outcome. a, c L929sA cells, starved for 48 h in DMEM devoid of serum, were pre-incubated with solvent, DEX (1 μM), or CpdA (1 or 10 μM) for 1 h, before STS (60 nM) and TNF (2,000 IU/ml) were added, where indicated, for 6 h. Total RNA was isolated and subjected to RT-qPCR assaying cellular c-jun, TNFα, and β-actin and hypoxanthine–guanine phosphoribosyltransferase (HPRT) household gene mRNA levels. Specific signal for cDNA of c-jun or TNFα was normalized to the averaged household genes signal. The STS/TNF condition was set at 100 and all other conditions were recalculated accordingly to allow ratio comparisons. Total solvent concentration was kept similar in all conditions. Results are shown ± SD. The experiment was carried out at least in triplicate and the results are averages of at least two independent experiments. Results of the statistical analysis via ANOVA followed by a Tukey multiple comparison post-test are shown for particular groups of interest, in comparison to the STS/TNF group). b L929sA cells, starved for 48 h in DMEM devoid of serum, were treated with solvent, a combination of STS (60 nM) and TNF (2,000 IU/ml) for 5 h in absence or presence of a 1 h pretreatment of DEX (1 μM) or CpdA (10 μM). Total protein extracts were prepared in duplicate and subjected to Western-blot analysis to detect c-Jun protein. Detection of NF-κB p65 served as a loading control (Ctrl). The result is a representative of three independently performed experiments. d A549 cells, starved for 24 h in DMEM devoid of serum, were pretreated for 1 h either with solvent, DEX (1 μM) or CpdA (10 μM), either or not followed by a 3-h treatment with TNF (2,000 IU/ml). Gene expression levels of corresponding RNA samples were evaluated by a whole genome transcriptome Agilent array (upper panel). Genes with adjusted p values lower than 0.05 in at least one contrast and a fold change higher than 1.3 were selected as significant. Pscan analysis with a minimal statistical significance of p < 0.05, indicates enrichment for specific transcription factor binding motifs in the corresponding gene promoters. Here, we mention the identified NF-κB and AP-1 family members in bold (lower panel)

Fig. 5

Graphical illustration of differentially regulated AP-1 and NF-κB targets in the dataset. Arrows indicate experimentally confirmed promoter binding or regulation of expression. Transcription factors are depicted in red and their respective target genes are displayed in green. Only subunits with at least one unique target were kept to avoid redundancy. Promoter analysis of the six targets with no known dependency on NF-κB (b and c) revealed the presence of NF-κB motifs, indicating possible contribution of this transcription factor to their transcriptional regulation (2000–2013 Ingenuity Systems, Inc. All rights reserved)

Fig. 6

CpdA, but not DEX, blocks activated ERK in fibroblasts. a, b L929sA cells, starved for 48 h in DMEM devoid of serum, were pretreated with solvent, 1 μM DEX, or 10 μM CpdA for 1 h, followed by either or not TNF (2,000 IU/ml) (a), or TNF combined with STS (60 nM) (b), for the indicated time points (in minutes). Cell lysates were made and activated ERK was detected using the phospho-specific ERK MAPK antibody. Aspecific bands, non-phosphorylated proteins, and/or NF-κB p65 served as a loading control (indicated as load control)

Fig. 7

CpdA sustains activated JNK and c-Jun in fibroblasts. a, b L929sA cells, starved for 48 h in DMEM devoid of serum, were pretreated with solvent, 1 μM DEX, or 10 μM CpdA for 1 h, followed by either or not TNF (2,000 IU/ml) or TNF combined with STS (60 nM) for the indicated time points (in minutes). Cell lysates were made and activated JNK (a) was detected using the phospho-specific JNK MAPK antibody and activated c-Jun (b) was detected using the phospho-specific c-Jun antibody. Aspecific bands, non-phosphorylated proteins and/or NF-κB p65 served as a loading control (indicated as load control)

Fig. 8

GR is essential to mediate the gene expression modulation effect of DEX and CpdA. L929sA cells were transfected with siRNA control (siControl) or siRNA targeted at GR (siGR) and were allowed to rest for 48 h post transfection. In the 16-h period before induction or sampling, cells were starved in DMEM devoid of serum. a We controlled for the efficiency of siRNA GR targeting. Total RNA was isolated and subjected to RT-qPCR for GR mRNA levels and expression levels were normalized to housekeeping gene controls. The expression levels for GR (siControl) were set at 100 and the siGR condition was recalculated accordingly (left panel). Cell lysates were made and GR protein was visualized via Western-blot analysis (right panel). Actin served as a loading control. b SiRNA-transfected L929sA cells (siControl or siGR) were pretreated with solvent, 1 μM DEX, or 10 μM CpdA for 1 h, followed by either or not TNF (2,000 IU/ml) combined with STS (60 nM), for the indicated time points (in minutes). Cell lysates were made and activated JNK was detected using the phospho-specific JNK MAPK antibody. NF-κB p65 served as a loading control (indicated as load control). c–e SiRNA-transfected L929sA cells (siControl or siGR) were pre-incubated with solvent, DEX (1 μM), or CpdA (1 or 10 μM) for 1 h, before TNF (2,000 IU/ml) and STS (60 nM) was added, where indicated, for 6 h. Total RNA was isolated and subjected to RT-qPCR for specific target genes, and expression levels in each treatment group were normalized to housekeeping gene controls. Normalized mRNA levels for DUSP1 (c) was presented with the DEX (SiControl) set as 100 and all other conditions were recalculated accordingly. Normalized mRNA levels for IL-6 (d) and MMP13 (e) were presented with the STS/TNF condition (siControl and siGR) set at 100 and all other conditions were recalculated accordingly to allow ratio comparisons. Total solvent concentration was kept similar in all conditions. Results are shown ± SD. The qPCR was carried out at least in triplicate. Results are representative for two independent experiments

Fig. 9

Only DEX recruits activated GR to the c-jun gene promoter. a, b L929sA cells, serum-starved for 48 h in DMEM devoid of serum, were pretreated for 1 h with solvent, DEX (1 μM), or CpdA (10 μM). Ensuing the indicated stimulation with TNF (2,000 IU/ml) combined with STS (60 nM) for 30 min, cells were lysed and total cell extracts were subjected to ChIP analysis and subsequent qPCR, detecting GR protein recruitment at the c-jun or IL-6 gene promoters. qPCR signal of immunoprecipitated c-jun or IL-6 promoter fragments is presented relative to input data. Averaged results of at least two independent experiments are shown ± SD. *p < 0.05; **p < 0.01, and ***p < 0.001

Fig. 10

CpdA enhances TNF-induced AP-1-driven gene expression in vivo and JNK2-/- mice are resistant to the CpdA-mediated hypersensitivity to TNF-induced lethality. a Wild-type C57BL6/J mice were injected i.p. with solvent or TNF (25 μg) for a total of 6 h, in the presence or absence of DEX (10 mg/kg) or CpdA (8 mg/kg), which was administered 30 min before the solvent (PBS) or TNF administration. Liver mRNA was assayed for levels of MMP13 and TIMP1. b Wild-type C57BL6/J mice and JNK2^−/− mice were injected i.p. with solvent or TNF (25 μg) in the presence or absence of CpdA (8 mg/kg). Survival of wt (black lines, n = 13–16) and JNK2^−/− mice (red lines, n = 14–15) was monitored for the indicated time points. c Serum IL-6 protein levels were measured in wt and JNK2^−/− mice 6 h after the challenge with TNF (25 μg), in the presence of either PBS or CpdA (8 mg/kg). Averages of 2 independent experiments are shown for (a) and (b). ns not significant; *p < 0.05; **p < 0.01

Fig. 11

Summarizing model for the transcription factor-selective action of CpdA versus DEX. The cytoplasmic GR is kept in a ligand-receptive conformation by binding to chaperone molecules. Upon binding of either GCs or CpdA, the GR changes its specific conformation and translocates into the nucleus. GC-bound GR can form a homodimer and as such binds to a palindromic simple GRE, thus propagating its classic transactivation mechanism. Conversely, CpdA-bound GR cannot form homodimers and is therefore not able to bind a simple GRE or support transactivation. While GC-bound GR leaves p38 and ERK MAPK phosphorylations (almost equal to symbol) unaffected, it diminishes JNK MAPK phosphorylation (slanting down arrow). Stimulation with CpdA affects these MAPK phosphorylations differently, as it actually prolongs JNK MAPK phosphorylation (arrow) and sparks a decline in ERK MAPK phosphorylation (slanting down arrow). Also CpdA does not affect the level of p38 MAPK phosphorylation (almost equal to symbol). As expected from the differential MAPK phosphorylation modulations, CpdA- and GC-bound GRs also show a differentiation in transcription factor targeting. GC-bound GR is recruited onto NF-κB- and AP-1-driven gene promoters and is fully capable of transrepressing both NF-κB- and AP-1-driven gene expression. In contrast, CdpA-bound GR can only bind to NF-κB-driven gene promoters and not to AP-1-driven gene promoters. As such, CpdA-bound GR only supports transrepression of NF-κB-mediated gene transcription, and not AP-1-mediated gene transcription