Differential off-target glucocorticoid activity of progestins used in endocrine therapy.

Competitive whole cell binding assays were conducted in COS-1 cells exogenously expressing human GR to investigate the RBA of the selected progestins for the GR ( Fig. 1 ). RBAs expressed as a % DEX (reference agonist (100%)) revealed that all the progestins exhibited significantly different RBAs with the rank order DEX > MPA > ETG > NES > LNG > NET ( Table 1 ). No statistically significant differences were observed between the IC 50 values for the synthetic GR agonist DEX versus MPA and ETG, while NES, LNG and NET have significantly lower IC 50 values. However, NES, LNG and NET exhibited significantly different affinities when compared to each other. Experiments in the absence of expressed human GR showed no binding to the endogenous GR ( Supplementary Fig. 1 ). We investigated the relative agonist efficacies (maximal responses) and potencies (EC 50 values) for transactivation ( Fig. 2 , Table 2 ) of a synthetic GRE-containing promoter-reporter construct (pGRE) and the endogenous GILZ gene of the progestins via the GR. Incubation with 10 μM progestin confirmed that LNG and NET exhibited no significant activity on the synthetic GRE promoter via the GR above the background ( Supplementary Fig. 2 ) and they were thus excluded from further promoter-reporter dose-response analysis. The results show that MPA, ETG and NES are partial GR agonists for transactivation on the promoter-reporter construct ( Fig. 2A ). Relative to the reference agonist DEX, MPA is significantly more efficacious than NES and ETG, while NES and ETG display statistically similar efficacies. While no significant differences in potency were detected between MPA, ETG and NES, MPA appears to be less potent than DEX on the promoter-reporter construct. Similar results were obtained on the endogenous GILZ gene in human PBMCs ( Table 2 ). ETG and NES have a lower efficacy and are weaker agonists than DEX, with MPA displaying statistically similar efficacy as DEX ( Fig. 2B ). ETG is significantly less efficacious than NES, with NET appearing to be the least efficacious ligand ( Fig. 2B ). Non-linear regression found the LNG curve to be unstable and thus the potency and efficacy were not obtained. Stimulation with 100 nM MPA or NES significantly increased mRNA levels of GILZ ( Supplementary Fig. 3A ). Co-treatment with the GR/PR antagonist RU486 prevented this upregulation of GILZ ( Supplementary Fig. 3A ). The relative agonist efficacies and potencies of the progestins for transrepression via the GR on AP-1 ( Fig. 2C ) and NFκB-containing ( Fig. 2D ) promoter-reporter constructs were investigated in COS-1 cells ( Table 3 ). Treatment with 10 ng/mL PMA resulted in a ~7.94-fold and ~13.41-fold induction, respectively ( Fig. 2C and D ). Like DEX, MPA, ETG and NES are GR agonists via AP-1 ( Fig. 2C ) and NFκB ( Fig. 2D ), while both NET and LNG are weak partial GR agonists. Although all progestins investigated are less potent than DEX via the AP-1 promoter, MPA is more potent than NET, while significant differences were not detected between potencies of LNG, ETG and NES. However, on the NFκB promoter ( Fig. 2D ), no significant difference was observed in potency for DEX versus MPA, NET and LNG via the GR. No significant differences in potencies were observed between NET, LNG, ETG and NES via both the AP-1 and NFκB promoter-reporter constructs, while MPA is significantly more potent than ETG via the NFκB promoter-reporter construct. Broadly consistent with the reporter gene results, Fig. 2E shows that all the progestins, except NET, are agonists for transrepression of the endogenous IL6 gene in PBMCs ( Table 3 ). The result for NET was inconclusive due to large error in the data and the resulting instability of the curve. Although no significant differences were detected, the progestins appear less potent than DEX, with NES appearing to display higher potency than MPA and ETG. Stimulation with 100 nM MPA or NES significantly decreased the mRNA levels of IL6 ( Supplementary Fig. 3B ). Co-treatment with the GR/PR antagonist RU486 prevented this repression of IL6 ( Supplementary Fig. 3B ). Selective regulation of the GR may exhibit varying therapeutic effects depending on the molecular mechanism involved in the regulation of inflammation [ 78 ]. Ligand-selective GR phosphorylation at S226 and S211 has previously been shown to be indicative of transcriptionally active GR [ 76 , 77 ]. Having shown that the GR is most likely required for the transcriptional regulation of GILZ and IL6 by MPA and NES in PBMCs, next we wanted to determine if the progestins could phosphorylate the GR in PBMCs. We found that stimulation with MPA and NES, but not NET, LNG or ETG, results in GR phosphorylation at S226 and S211, similar to the effect of DEX, albeit to a slightly lower level for S211 ( Fig. 3 ).

We report for the first time the RBAs, potencies, and efficacies for transactivation and transrepression for this panel of progestins determined in parallel on synthetic and endogenous genes. The synthetic promoters include AP-1 and NFκB cis -elements involved in inhibition of immune function [ 2 ]. Upregulation and downregulation of the GILZ and IL6 endogenous genes, respectively, are associated with inhibition of inflammation and immune function [ 12 ]. GILZ modulates important inflammatory signaling pathways through its interaction with transcription factors [ 79 , 80 ]. Consequently, GILZ expression protects against damage caused by neuroinflammation, allergy and heart disease, amongst other conditions, as it plays a role in the anti-inflammatory activities of GCs [ 79 , 81 ]. Our results show that while the progestins all bind to the GR at high concentrations, MPA, ETG and NES exhibit much greater affinity for the GR than LNG and NET. Relative to DEX (100%), our RBA results for LNG (3%) and NES (6%) are consistent with animal model binding data [ 2 , 61 , 62 ], unlike our finding of substantial GR binding for ETG (30%). LNG and NET exhibit no or relatively very little agonist activity on synthetic or endogenous promoters transactivated by the GR, consistent with other reporter gene and rat data for LNG [ 82 , 83 ]. Despite a low RBA, NES exhibited potent and substantial partial agonist activity on synthetic and endogenous promoters transactivated by the GR, unlike the lack of activity observed in rat bioassays, possibly due to inactivation of NES in the bioassay [ 82 ]. Consistent with our binding data, ETG exhibited potent and substantial partial agonist activity on synthetic and endogenous promoters transactivated by the GR, unlike other reports using reporter genes [ 61 ] or rat bioassays [ 82 ]. ETG and NES are agonists for transrepression of synthetic and endogenous genes by the GR but partial agonists for transactivation. Both LNG and NET exhibit substantial activity at high concentrations for transrepression of synthetic and endogenous genes by the GR, although they are much less efficacious than MPA at peak serum concentrations used in contraception ( Supplementary Table 1 ) [ 46 ]. We show for the first time that MPA is the most and NET the least efficacious ligand compared to NES, ETG and LNG for both transactivation and transrepression of synthetic promoters via the GR. Taken together, our novel results in PBMCs are consistent with previously reported data showing MPA, unlike NET, is a potent and efficacious GR ligand for transactivation and transrepression [ 21 , 60 ]. We now also show for the first time that relative to DEX and MPA, NES and ETG both bind to the GR with a RBA greater than LNG and NET and exhibit significant GR activity with variable potencies and efficacies for transactivation and transrepression, in a promoter- and context-specific manner. Several lines of evidence support our conclusion that the biological activities measured in the COS-1 cells and in PBMCs occur via the GR. Our binding and synthetic promoter assays, performed in the absence and presence of expressed GR, as well as our comparison with results for the GR agonist DEX, show that the ligands bind to the GR, which is required for transcriptional regulation. The results obtained on the synthetic promoters compared to the endogenous genes in PBMCs for both transactivation and transrepression are remarkably similar, with only a few small differences. Additionally, as for DEX, the GR/PR antagonist RU486 inhibited MPA- and NES-induced transcriptional regulation of GILZ and IL6 in PBMCs. Both MPA and NES were shown to ligand-selectively phosphorylate the GR at Serine 226 (S226) and Serine 211 (S211) in a manner similar to DEX, a property only observed with GR agonists or partial agonists. Taken together with the observation that the GR is the only SR detected in PBMCs by western blotting under our conditions [ 72 ], our results strongly suggest that the responses in PBMCs are due to the GR. Our findings suggest that binding of MPA to the GR is more likely to be physiologically relevant than for the other progestins, based on whether their IC 50 or EC 50 values fall within the ranges of serum concentrations for contraceptive users ( Supplementary Table 1 ). However, since potencies and efficacies can increase when GR levels are increased [ 84 ], it is possible that both ETG and NES could exert physiologically relevant biological effects in select environments. Our findings suggest that MPA, ETG and NES, but less so for LNG and NET, have the potential to exert side-effects on multiple GR-mediated functions. While the effects of high levels of chronic exposure to potent GCs are relatively obvious, the effects of lower levels of potent GCs or less potent or less efficacious GR agonists are less predictable. It is emerging that the mechanisms of GR regulation are very complex, exhibit differential sensitivity to GC concentrations and are cell-, gene-, locus- and signal-specific [ 21 , 26 , 27 , 30 , 85 ]. Thus, genes vary in their sensitivity to GR ligand concentrations, depending on multiple factors, including GR expression levels. Our findings are consistent with this complexity and further show that progestin activity via the GR for transrepression of transcription is more potent than for transactivation. These data provide a plausible potential biological mechanism for some of the observed differential GC-like side-effects of progestins. Moreover, awareness is raised of the need to use the lowest possible concentrations of progestins in endocrine therapy required for therapeutic efficacy and for more robust and directed clinical studies to investigate potential side-effects on specific physiological processes that may be affected by the GR.

The COS-1 ( RRID:CVCL_0223) monkey kidney cell line was purchased from ATCC, USA, maintained as previously described [ 64 ], and was mycoplasma negative for the duration of the study. Sources of reagents were as follows: Dexamethasone (DEX), MPA, NET, LNG, ETG, NES, RU486 (mifepristone) and phorbol 12-myristate13-acetate (PMA) (Sigma-Aldrich, RSA); [ 3 H]-DEX (78 Ci/mmol) (AEC Amersham, RSA); The human GR expression plasmid, pcDNA3.1-GR (pGR) (D.W. Ray, University of Manchester, UK) (Ray et al., 1999); empty vector pcDNA3.1 (Invitrogen, USA); pTAT-GRE-E1b-luciferase plasmid (pGRE) (G. Jenster, Erasmus University of Rotterdam, The Netherlands) [ 65 ]; 5x Nuclear Factor kB-luciferase (pNFκB) and 7x Activator Protein-1-luciferase (pAP-1) promoter-reporter plasmids (Stratagene, Houston, USA); antibodies to GR (H-300, sc-8992, RRID: 2155784 ) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (0411; sc-47724, RRID: 627678 ) and secondary anti-rabbit (sc-2313 , RRID: 641181 ) (Santa Cruz Biotechnology, USA); anti-P-S226 or anti-P-S211 specific antibody (Dr. M.J. Garabedian, New York University, USA) [ 66 , 67 ]. Although multiple methods are available for the determination of binding affinities [ 68 , 69 ], we conducted competitive whole cell binding assays as described [ 21 , 70 ] with minor modifications. COS-1 cells were transiently transfected with either 5 μg of pGR or the pcDNA3.1 empty vector, using X-tremeGene9 (Roche, RSA) as per manufacturer’s instructions. Counts per minute were measured and normalised to total protein content per well as determined by the Bradford assay [ 71 ]. Total binding ([ 3 H] DEX in the absence of unlabelled competitor) was set at 100%. Specific bound [ 3 H] DEX was calculated as the difference between total and nonspecific binding ([ 3 H] DEX plus 10 μM unlabelled DEX). RBAs (%) were calculated as follows: [IC 50 value of DEX (M) / IC 50 value for each ligand (M)] x 100, with DEX IC 50 value set as 100%. Reporter assays were performed essentially as previously described [ 13 , 14 , 21 ] with a few modifications. For transactivation assays, COS-1 cells were seeded into 10 cm dishes (Griener, Germany) at a density of 1.5×10 6 cells and transiently transfected with 10 μg pGR or pcDNA3.1 empty vector and 3.75 μg pGRE using X-tremeGene9. Twenty-four hours later, the transfected cells were plated into 96-well plates at a density of 1×10 4 cells/well. The next day cells were stimulated and treated for 24 hours with varying concentrations of the ligands. For transrepression assays, COS-1 cells were seeded into 10 cm dishes at a density of 2 × 10 6 cells. After 24 hours, the cells were transiently transfected with 3.75 μg pGR and 7.5 μg pAP-1 or pNFκB using X-tremeGene9. The next day, the transfected cells were plated into 96-well plates at a density of 1 × 10 4 cells/well and treated for 24 hours with 10 ng/mL PMA in the absence and presence of varying concentrations of the test compounds. Luciferase activity was measured in relative light units and normalised to total protein content per well as determined by the Bradford assay [ 71 ]. PBMCs were isolated, cultured and stimulated as previously described [ 22 ]. Briefly, isolated PBMCs were washed, counted, and cultured in RPMI at 2 ×10 6 cells/ml. For endogenous gene analysis 4 ×10 6 PBMCs were seeded in round-bottom tubes and stimulated with respective ligands, as indicated in the figure legends, for 48 hours. Thereafter the PBMCs were centrifuged at 1 200xg for 5 minutes and the supernatant was discarded. Pelleted cells were used for RNA isolation. For western blot analysis, 1 ×10 6 PBMCs were seeded in round-bottom tubes and stimulated with 100 nM ligand or vehicle (ctrl 0.1% v/v EtOH), for 30 minutes. Thereafter the PBMCs were centrifuged at 1 200xg for 5 minutes, the supernatant was discarded and pelleted cells were used for western blotting. RNA was isolated from PBMCs stimulated with ligands as previously described [ 22 , 72 ], using Tri-reagent (Sigma Aldrich, RSA) as per the manufacturer’s instructions. RNA (250 ng) was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosciences, ThermoFisherScientific). qPCR was performed using FastStart PCR Master kit containing SYBR green (Sigma Aldrich, RSA) and Rotor-gene, RG-3004 (Corbett Research, RSA). Primer sets were as follows: Glucocorticoid-induced leucine zipper (GILZ) (cat#QIA249900-QT00091035, Qiagen, RSA); Interleukin 6 (IL6) (forward) 5’-TCTCCACAAGCGCCTTCG-3’ and (reverse) 5’-CTCAGGGCTGAGATGCCG-3’ and GAPDH (forward) 5’TGAACGGGAAGCTCACTGG-3’ and (reverse) 5’CCACCACCCTGTTGCTGTA-3’ [ 73 , 74 ]. The relative expression of GILZ and IL6 was calculated using the Pfaffl method and normalized to relative GAPDH expression levels [ 75 ]. Whole cell lysates were prepared using a N-[Tris(hydroxymethyl)-methyl]-3-aminopropanesulfonic acid (TAPS) buffer (0.1 M TAPS, pH 9.5) on ice as described. Western blot analysis was performed as previously described [ 76 ]. Briefly, equal amounts of cell lysate were loaded onto 8% SDS-polyacrylamide gels and separated by electrophoresis at 75 V for 20 minutes then 120 V for 1 hour in 1X running buffer (25 mM TRIS-HCl, 250 mM glycine and 0.1% ( v/v ) SDS pH 8.4) using a Bio-Rad Mini Protean II electrophoresis system (Bio-Rad, South Africa). Proteins were then blotted onto Hybond-ECL nitrocellulose membrane (AEC-Amersham, South Africa) for 1 hour at 180 mA in cold 1X transfer buffer (25 mM TRIS, 200 mM glycine, 20% ( v/v ) methanol). Membranes were subsequently blocked for 1 hour at room temperature by shaking in 4% ( w/v ) ECL advance blocking powder (AEC-Amersham, South Africa) in 1X TRIS-buffered saline (50 mM TRIS, 150 mM NaCl; TBS) containing 0.1% ( v/v ) Tween (TBS-Tween; TBST). Primary antibodies were diluted in 4% ECL-TBST and incubated on membranes overnight with shaking at 4°C. Membranes were then washed three times in 1X TBST for 5 minutes and incubated with secondary antibodies diluted in 5% (w/v) skim milk powder in 1X TBST for 1 hour at room temperature with shaking. After three 5-minute washes in 1X TBST, membranes were placed in 1X TBS prior to a 1-minute incubation with Pierce ECL-chemiluminescent western blotting substrate (Thermo Scientific, USA). Proteins were visualized by autoradiography using Amersham Hyperfilm™ MP high performance autoradiography film (AEC-Amersham, South Africa) and quantified using ImageJ. For the phosphorylation assays, membranes were first incubated using P-S226- or P-S211-specific antibodies, with their respective secondary antibodies. After developing and autoradiography, the membranes were stripped as described in [ 77 ] and re-probed for total GR using the H-300 GR antibody. Statistical and data analysis were performed using GraphPad Prism™ software version 9. All data were first tested for normality, before parametric or non-parametric tests were performed. Competitive binding data were analysed using non-linear regression and one site – fit logIC 50 options and the Newman-Keuls post-test was used for statistical analysis. Dose-response analysis of the luciferase reporter assays, as well as endogenous genes was plotted as described. Non-linear regression and sigmoidal dose-response were used for which the slope was set to +1 for transactivation and −1 for transrepression and one-way ANOVA and the Tukey (compares all pairs of columns) post-test were used for statistical analysis of the efficacies and potencies. For dose-response analysis, all curves are shown as “best-fit” curves that are not ambiguous, unless stated as “unstable” in the respective figure legends. These “unstable” curves should be interpreted with caution, as a wide range of values would essentially lead to the same curve. For 100 nM ligand responses on endogenous genes, a non-parametric Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test was performed when comparing all samples to vehicle (control). A Mann-Whitney t-test was performed when comparing between different ligands. GR phosphorylation data was analysed using a one-way ANOVA, with a Dunnett’s multiple comparisons post-test, comparing each ligand to control. Statistically significant differences are indicated by different letters or symbols (* or #), as indicated in the figure legends. All data, except for the representative western blot, were expressed as means, where the errors bars represent the standard error of the mean (SEM) with n values given in each figure legend.

Synthetic progestogens or progestins are widely used in women for hormonal contraception, hormone replacement therapy (HRT) and treatment of endometriosis [ 1 – 3 ] Progestins exert their therapeutic actions by mimicking the actions of progesterone via binding to the progesterone receptor (PR) [ 2 ]. However, use of many progestins has been associated with a number of side-effects, including increased risk of breast cancer, cardiovascular disease, weight gain, loss of bone density and possibly increased susceptibility to infections such as HIV-1[ 1 , 4 – 12 ]. While some side-effects of progestins may be mediated via the PR which is activated by progestins, others may be mediated via other steroid receptors (SRs) [ 12 – 15 ]. Particular concerns have been raised regarding side-effects of the progestin medroxyprogesterone acetate (MPA), used in HRT and in the contraceptive DMPA [ 4 , 5 , 12 , 16 – 19 ]. We and others have proposed and provided plausible in vitro evidence that at least some of the side-effects of DMPA may be due to its glucocorticoid (GC)-like effects, mediated via the glucocorticoid receptor (GR) [ 2 , 12 , 20 – 25 ]. GCs regulate a wide range of physiological functions via the ubiquitous GR, including development, homeostasis, metabolism, immune function, reproduction and the stress response [ 26 – 30 ]. Exogenous GCs are widely used to treat acute and chronic and inflammatory diseases due to their potent anti-inflammatory and immunosuppressive properties. Long-term exposure to GCs is associated with multiple adverse health outcomes including weight gain, elevated glucose levels, hypertension, glucose intolerance, insulin resistance, osteoporosis, and psychological effects leading to depression and mood changes [ 28 , 29 ] and could lead to Cushing’s syndrome, a pathological condition characterized by muscle wasting, fat accumulation, and susceptibility to infections [ 31 ]. Neither the use of DMPA nor any other progestin used in endocrine therapy has been associated with broad systemic immunosuppression or the degree of symptoms found in Cushings syndrome patients. However, several of the reported side-effects of DMPA, such as select changes in immune function, weight gain and increased susceptibility to HIV-1, are consistent with select GC activity, although the clinical data are inconsistent not consistent [ 11 , 12 , 32 – 41 ]. The extent to which these side-effects are due to MPA acting via the GR is unknown, and very difficult to determine clinically. In general, only some, but not all possible side-effects that would be consistent with select and weak GR activity have been investigated in clinical studies for widely-used progestins [ 1 , 7 , 42 – 44 ] A key outstanding question is whether other progestins used in endocrine therapy exert any side-effects via the GR. Long-acting progestin-only injectable contraceptives like DMPA and the two-monthly intramuscular injectable, Nuristerate, containing norethisterone enanthate (NET-EN) are widely used in the developing world and are highly effective, discreet, and reversible contraceptive methods [ 1 , 2 , 12 ]. Etonogestrel (ETG), levonorgestrel (LNG) and nestorone (NES) are increasingly being used in contraceptives as long-term alternatives to injectables [ 45 ]. ETG is used in combined oral contraception, administered as subdermal implants and used in intravaginal rings [ 46 ], while LNG is administered in intrauterine devices, subdermal implants, vaginal tablets or gels, as well as for progestin-only or combined oral contraception[ 12 , 46 , 47 ]. NES is used in contraception in vaginal rings in combination with ethinyl estradiol (EE), in subcutaneous capsules, or in implants [ 12 , 46 ]. ETG has been associated with mild insulin resistance and could affect bone density and increase body weight [ 42 , 48 – 51 ], while NET has been associated with an increase in blood glucose levels and an unfavourable lipid profile, but is not associated with increased HIV-1 acquisition relative to no contraception or infrequent condom use [ 11 , 52 , 53 ]. In contrast, LNG and NES have been associated with relatively few side-effects [ 42 , 44 , 54 , 55 ] for those that have been investigated. However, it is not possible to exclude the possibility that some select gene- and/or cell-specific and biologically significant GR-mediated side-effects occur for any of these contraceptives, based on the available clinical data. Given the difficulty in establishing whether the GR is involved in side-effects of progestins used in endocrine therapy clinically, another approach is to establish the potential for such side-effects as proof-of-concept, by determining the relative affinity, potency and efficacy of progestins for regulation of gene expression via the GR in vitro . The GR is a ligand-activated transcription factor that exert their biological responses by binding to target gene promotors to either increase (transactivation) or decrease (transrepression) transcription [ 12 ]. We and collaborators have previously focussed on determination of the relative binding affinity and transcriptional activity of MPA and NET via other SRs, including the GR [ 13 – 15 , 21 , 56 – 58 ]. While side-effects of progestins that may be mediated via the PR are likely to be similar [ 59 ], others mediated via the GR may differ between progestins. We have established that MPA, but not NET, exhibits relatively potent and efficacious GC-activity via the GR [ 2 , 13 , 21 , 60 ]. However, very little information is available about the relative activity via the GR for LNG, ETG and NES, compared to each other and to MPA and NET. Limited data from animal models suggest that ETG, LNG and NES bind with low affinity to the GR [ 2 , 61 , 62 ]. In vivo bioassays in rats found that, like NET, both LNG and NES exhibit no GR agonist activity. In contrast, ETG was shown to have weak GR agonist activity in a cell line transactivation reporter assay [ 61 ] and in in vivo bioassays in rats [ 12 , 63 ]. However, most of these data are likely to be confounded by multiple factors including comparisons between non-parallel investigations, species-specific effects, and the presence of competing SRs. Given the potential for the ubiquitous GR to regulate multiple genes and the resulting potential for side-effects of progestins via the GR, more comparative data on the activity of progestins via the GR are urgently required. This may be particularly relevant to HIV-1 acquisition, since both in vitro and clinical data suggest that DMPA, but not NET-EN, may increase HIV-1 infection [ 12 ]. For the first time, this study directly compared the relative binding affinities (RBAs) and transcriptional activities for both transactivation and transrepression by the GR, using dose-response analysis, for MPA, NET, LNG, ETG and NES in parallel, in a cell line on synthetic promoters and in a primary cell model on endogenous genes, as well as the ability of these progestins to result in phosphorylation and activation the GR.