Cell culture materials are from Life Technologies (Cergy-Pontoise, France). Luciferin (sodium salt) was purchased from Promega (Charbonnières, France). Chemical substances used in this study are presented in Table 1 . Methyltrienolone (R1881), dexamethasone (DEX), mifepristone (RU486), aldosterone (ALDO), pregnenolone (P5), progesterone (P4), dydrogesterone (DYD), norethindrone (NET), tibolone (TIB), spironolactone (SPI), canrenone (CAN), fluticasone propionate (FT), deacetyl cortivazol/bimedrazole (DAC), cortivazol (CVZ), methylprednisolone (MPS), cortisol (CORT), dihydrotestosterone (DHT), drospirenone (DRO) and17α-hydroxyprogesterone (17-OHP) were obtained from Sigma-Aldrich (Saint-Quentin Fallavier, France). Promegestone (R5020), forbimenol, RU24782 and RU24858 are synthetic steroids non-commercially available and kindly gifts from Sanofi, Vertolaye, France.

Stock solutions of chemicals were prepared in dimethyl sulfoxide (DMSO) and stored at -20°C. Fresh solution of test chemicals in test medium were prepared before each experiment. The final DMSO concentrations during treatment did not exceed 0.1% (v/v) of the test medium.

MMTV-luciferase-SV-neo plasmid was already described (27). pSG5-hGR puromycin and pSG5-puromycin are kind gifts of H Gronemeyer (IGBMC, Illkirch-Graffenstaden, France). pSG5-zfGR puromycin plasmid was obtained by cloning zfGR (M1-L746) in the BamHI site of pSG5-puromycin.

HMLN-hGR and UMLN-zfGR clonal cell lines were already described (14, 21). Briefly, HMLN-hGR cells were obtained by stable co-transfection of GR positive human HeLa cells with a glucocorticoid responsive gene (MMTV-Luciferase) and a hGR (pSG5-hGR-puromycin) expressing plasmid. To obtain UMLN-zfGR cells, human U2OS cells that mildly expressed hGR (28, 29) were stably co-transfected with the MMTV-Luciferase and a zfGR expressing plasmids (pSG5-zfGR-puromycin). 48 h after the transfection, cells were treated with G418 (1 mg/ml) and puromycin (0.5 μg/ml). Within 21 days, G418 and puromycin resistant clones appeared. For each cell line, 10 clones were chosen for their ligand-induced luciferase expression. The clones were amplified, and luciferase expression was checked at several passages. For each cell line, the clone with the best induction of luciferase activity was selected and used for the screening of the different steroids. The basal expression of luciferase is 1% and 11% for HMLN-hGR and UMLN-zfGR cell lines, respectively, of the maximal luciferase expression obtained in presence of dexamethasone 100 nM. The stability and the inducibility of luciferase expression were checked during at least 20 passages (20 weeks). The stability of the hGR and zfGR expression by ligand binding assay was also checked at different passages.

To obtain UMLN-hGR pool cells, U2OS cells were stably co-transfected with the MMTV-Luciferase and the hGR expressing plasmids. ZFL-zfGR pool cells were obtained by stable co-transfection of zebrafish ZFL cells with the MMTV-Luciferase and the zfGR expressing plasmids. The transfected cells were treated 7 days with G418 and puromycin but not cloned.

HMLN- and UMLN-GR cells were grown in a 5% CO₂ humidified atmosphere at 37°C in Dulbecco’s Modified Eagle’s Medium: Nutrient Mixture F-12 (DMEM/F-12) containing phenol red, 1 g/L glucose, 10% fetal bovine serum (FBS), 100 units/mL of penicillin, 100 µg/mL of streptomycin, 1 mg/mL geneticin and 0.5 µg/mL puromycin (culture medium). Exposure was made in phenol red-free DMEM medium supplemented with 5% of dextran-coated charcoal FBS (DCC), 100 units/mL of penicillin and 100 µg/mL of streptomycin (HMLN hGR and UMLN GR test medium).

ZFL-zfGR cells were cultured at 28°C in humidified atmosphere with 5% CO₂ in LDF medium (50% Leibovitz 15 culture medium L15, 35% DMEM high glucose and 15% Ham’s-F12 medium) with 0.15 g/L sodium bicarbonate, 15 mM 4-(2-hydroxy-ethyl)-1-piperazineethanesulfonic acid (HEPES), 0.01 mg/mL insulin, 50 ng/mL epidermal growth factor (EGF), 50 U/mL penicillin and streptomycin antibiotics, 10% (v/v) fetal bovine serum (FBS), 1 mg/mL geneticin and 0.5 µg/mL puromycin (culture medium). Exposure was made in the same culture medium excepted that 10% (v/v) fetal bovine serum (FBS) was replaced by 5% of dextran-coated charcoal FBS (DCC-FBS). The different reporter cell lines used in this study are summarized in the Supplementary Table 1 .

In vitro transactivation assays were performed in 96-wells white opaque clear bottom culture plates (Greiner CellStar, Dutscher, Brumath, France). GR reporter cell lines were seeded at density of 5 x 10⁴ cells per well in 150 µL culture medium and incubated at 37°C and 5% CO₂ for 24 h. Then, medium was removed and cells were exposed to increasing dilutions of tested compounds in test medium (DMSO; final concentration 0.1% v/v). Cells were incubated for 16 h at 37°C and 5% CO₂.

Results of transactivation activity were expressed as percentage of the maximum luciferase activity induced by dexamethasone at 10^-7 M.

For antagonistic activity assessment, cells were exposed to different concentrations of the tested compounds and 3 nM dexamethasone. At this concentration, dexamethasone yields 60-80% of the maximal response. After the incubation period, medium was removed and replaced with 50 µL/well of test medium containing 0.3 mM luciferin. Luminescence signal was monitored in intact living cells for 2 s per well using a MicroBeta Trilux microplate scintillation and luminescence counter (PerkinElmer, Courtaboeuf, France).

The effect of the tested chemicals on cell viability was assessed with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Briefly, after luminescence detection, medium containing luciferin was removed and replaced with 100 µL/well of test medium containing 0,4 mg/ml MTT for 4 h. Colorimetric signal was monitored at 570 nM using a Pherastar microplate reader (BMG Labtech, Champigny s/Marne, france). Experiments were performed in quadruplicate and repeated three times.

For measurement of GR expression, HMLN-hGR, U2OS and UMLN-zfGR cells were seeded in 24-wells transparent plates at a density of 400,000 cells per well in culture medium. Cells were incubated for 24 h in a 5% CO₂ humidified atmosphere at 37°C. Then, medium was then removed and replaced with test medium containing 10 nM [3H]-dexamethasone (84 Cu/mmol, Perkin Elmer), in the absence or presence of 10 μM of non-radioactive dexamethasone. After 3 h, unbound material was aspirated, and cells were washed three times with cold PBS in order to remove additional unbound material. Then, 0.4 ml of lysis buffer (250 mM Tris phosphate pH 7.8, 0.1% triton X-100) was added and plates were shaked for 5 min. 0.1 ml of the total cell lysate was mixed with of 0.1 ml of LSC-cocktail (Emulsifier-Safe, Perkin Elmer) and [3H] bound radioactivity was liquid scintillation counted (MicroBeta trilux, PerkinElmer). Protein concentrations of 0.1 ml of the total cell lysate were measured by Bio-Rad protein assay (Bio-Rad, Marnes-la-Coquette, France) and used to normalize bound radioactivity values expressed in dpm. Specific binding was determined by subtracting non-specific binding from total binding and enable to determine GR expression in fentomoles of protein per mg of protein. Experiments were performed in quadruplicate and repeated three times.

For ligand competition assays, HMLN-hGR and UMLN-zfGR cells were seeded in 96-wells white opaque clear bottom culture plates (Greiner CellStar, Dutscher) at density of 10⁵ cells per well in 0.2 ml of culture medium and incubated at 37°C and 5% CO₂ for 24 h. Then, medium was removed and cells were exposed to with 1 nM [3H]-dexamethasone in the absence or presence of increasing concentrations of non-radioactive competitive compounds. After 3 h, unbound material was aspirated, and cells were washed three times with cold PBS in order to remove additional unbound material. Then, 150 μl of lysis buffer (250 mM Tris phosphate pH 7.8, 0.1% triton X-100) was added and plates were shaked for 5 min. 50 μl of the total cell lysate was mixed with of 50 μl of LSC-cocktail (Emulsifier-Safe, Perkin Elmer) and [3H] bound radioactivity was liquid scintillation counted (MicroBeta trilux, Perkin Elmer). Protein concentration of 50 μl of the total cell lysate was measured by Bio-Rad protein assay and used to normalize bound radioactivity values expressed in dpm. Results were plotted as measured dpm versus concentration of tested compound. IC50 values were defined as compound concentration required to decrease maximum [3H]-dexamethasone binding by 50%. All the experiments were performed in quadruplicates and in at least three independent experiments.

Results of transactivation activity were expressed as percentage of the maximum luciferase activity induced by dexamethasone at 10^-7 M. Dose-response curves were fitted using the sigmoidal dose-response function of a graphics and statistics software program (GraphPad Prism 8, GraphPad Software Inc.). We assumed a compound being agonist when an effect above 10% was noted as compared to the control cells, and antagonist when it decreased luciferase activity by more than 20% in the presence of the reference chemical at the concentration inducing 80% of the maximal response. Effective concentrations and inhibitory concentrations were derived from the Hill equation. For a given chemical, EC₅₀ was defined as the concentration inducing 50% of its maximal effect and IC₅₀ represented the concentration required for 50% inhibition. Lower and upper 95% confidence limits of EC₅₀s of IC₅₀s were calculated.

For ligand binding assays, dose-response curves were also fitted using GraphPad Prism. Inhibitory concentrations were derived from the Hill equation. Lower and upper 95% confidence limits of EC₅₀s of IC₅₀s were calculated.

The HMLN-hGR and UMLN-zfGR clonal cell lines were previously established in our laboratory (21). To obtain HMLN-hGR cells, we transfected HeLa cells which express endogenously hGR by a hGR expressing plasmid to overexpress hGR and the MMTV-Luc GR-responsive gene. The UMLN-zfGR cell line was established in U2OS cells which slightly express hGR (28, 29) by co-transfection of a zfGR expressing plasmid and the MMTV-Luc plasmid. As shown in Figure 1 , receptor protein level (hGR or zfGR) expressed in each cell line was estimated by saturation ligand- binding assay (LBA) with 10 nM [3H]-DEX in a ‘‘whole-cell’’ experiment. These whole-cell LBAs confirmed the endogenous expression of GR in HeLa cells (217 fentomoles/mg protein) whereas in U2OS this receptor is very slightly expressed (17 fentomoles/mg protein). Stable transfection of an hGR expression plasmid enabled to increase the expression of hGR in HMLN-hGR cells (535 fentomoles/mg protein). Finally, transfection of an zfGR expression in UMLN-zfGR cells enabled to express 324 fentomoles of zfGR per mg of protein ( Figure 1 ; Supplementary Table 2 ).

The reference glucocorticoid agonist dexamethasone was tested in transactivation assays in both reporter cell models revealing close potency between the HMLN-hGR and UMLN-zfGR cell lines (EC₅₀ of 1.38 and 2 nM for hGR and zfGR, respectively) ( Table 2 ; Figure 2A ). Almost all the chemicals belonging to the group of GCs, presented in Table 1 , showed full agonistic activity both on hGR and on zfGR in transactivation assays. Bimedrazole, cortivazol, fluticasone propionate and medrol exhibited similar potency on the two nuclear receptors ( Table 2 ) whereas cortisol was slighly more potent on zfGR than on hGR with EC₅₀ of 40 nM and 15 nM for hGR and zfGR, respectively ( Figure 2B ; Table 2 ). On the contrary, the dissociated glucocorticoids RU24782 and RU24858 which were structurally designed to distinguish between the transrepression and transactivation (10) have showed better potency on hGR than zfGR ( Table 2 ; Figures 2C, D ). Interestingly, the synthetic glucocorticoid forbimenol only activated hGR (EC50 503 nM) ( Table 2 ; Figure 2E ). Similarly, the mineralocorticoid aldosterone, which is produced naturally in human but not in fish (30), was fully agonist on hGR but partial agonist on zfGR (maximal activity of 96 and 19% for hGR and zfGR respectively) ( Table 2 ; Figure 2F ). All the other chemicals tested lacked glucocorticoid agonist activity.

RU24782, RU24858, forbimenol and aldosterone were also assessed for antagonism in UMLN-zfGR cells. Interestingly, RU24782, RU24858 partially repress dexamethasone-induced activity at sub-micromolar concentrations ( Figure 3A ). As these GCs were described as less able to recruit coactivators than dexamethasone to hGR (10), their zfGR antagonism at sub-micromolar concentrations and agonism at micromolar concentrations is probably relied to their lower efficacy for recruiting coactivators. Forbimenol and aldosterone also antagonized zfGR (IC50 3160 and 13980 nM, respectively) ( Table 3 ; Figure 3B ). Finally, the anti-progestin mifepristone was partial agonist of hGR and zfGR with higher efficacy on the last one ( Table 2 ; Figures 4A, B ). In presence of dexamethasone, it acts as partial antagonist with IC50 in the 10 to 40 nanomolar range ( Table 3 ; Figures 4A, B ). The other chemicals lacking glucocorticoid agonist activity were also tested for their antagonist activity. They all exhibited antagonist activity on hGR and zfGR. IC50 varied between 772 (promegestone) and 6101 (dihydrostestosterone) nM for hGR and 1066 (17α-hydroxyprogesterone) and 7711 (spironolactone) nM for zfGR ( Table 3 ).

Whole-cell competitive binding assays were performed with HMLN-hGR and UMLN-zfGR cells to determine whether the different potencies observed in transactivation assays for some of the chemicals reflected their abilities to differently bind to hGR and zfGR. Table 4 summarized IC₅₀ and RBA values for hGR and zfGR. IC₅₀ of DEX of 5.8 and 3.1 nM for hGR and zfGR, respectively confirmed the similar potency of this chemical for the two receptors. By contrast, cortisol bound preferentially zfGR as shown by IC₅₀s of 55.2 and 3.5 nM for hGR and zfGR, respectively ( Table 4 ) and thus explained its higher potency on the fish receptor. RU24782 bound preferentially to hGR than to zfGR but these differences in affinity did not reflect the differences in potency reinforcing the hypothesis that that this chemical is less able to recruit coactivators for zfGR than for hGR. Finally, aldosterone and forbimenol which are agonists on hGR and antagonists on zfGR bound also slightly preferentially hGR than zfGR but these differences in affinity do not explain their differences in activities for the two receptors.

Since cellular context could influence the transcriptional activity of NRs and hGR and zfGR reporter cell lines are different (Hela for hGR and U2OS for zfGR), we established UMLN-hGR pool cells. We tested in these cells the activity of RU24782, RU24858, forbimenol and aldosterone. The response profiles obtained ( Figure 5A ) are very similar to those obtained with the HMLN-hGR cell line indicating that at least for hGR these chemicals had a similar potency. The four chemicals were agonists on these cells and the EC₅₀ values for RU24782, RU24858, forbimenol and aldosterone were respectively of 3.1, 23.3, 54.5, 338 and 224 nM.

We also tested in zebrafish liver ZFL-zfGR pool cells the activity of the three chemicals. Again, the response profiles obtained ( Figure 5B ) are very similar to those obtained with the UMLN-zfGR cell line. The EC₅₀ value for DEX, RU24782 and RU24858 were respectively of 6, 1938 and 1027 nM were respectively whereas forbimenol and aldosterone only slightly activated luciferase expression ( Figure 5B ).