Reciprocal Modulation of Antiretroviral Drug and Steroid Receptor Function In Vitro

Millions of women are exposed simultaneously to antiretroviral drugs (ARVs) and progestin-based hormonal contraceptives. Yet the reciprocal modulation by ARVs and progestins of their intracellular functions is relatively unexplored.

ABSTRACT

Millions of women are exposed simultaneously to antiretroviral drugs (ARVs) and progestin-based hormonal contraceptives. Yet the reciprocal modulation by ARVs and progestins of their intracellular functions is relatively unexplored. We investigated the effects of tenofovir disoproxil fumarate (TDF) and dapivirine (DPV), alone and in the presence of select steroids and progestins, on cell viability, steroid-regulated immunomodulatory gene expression, activation of steroid receptors, and anti-HIV-1 activity in vitro. Both TDF and DPV modulated the transcriptional efficacy of a glucocorticoid agonist via the glucocorticoid receptor (GR) in the U2OS cell line. In TZM-bl cells, DPV induced the expression of the proinflammatory interleukin 8 (IL-8) gene while TDF significantly increased medroxyprogesterone acetate (MPA)-induced expression of the anti-inflammatory glucocorticoid-induced leucine zipper (GILZ) gene. However, peripheral blood mononuclear cell (PBMC) and ectocervical explant tissue viability and gene expression results, along with TZM-bl HIV-1 infection data, are reassuring and suggest that TDF and DPV, in combination with dexamethasone (DEX) or MPA, do not reciprocally modulate key biological effects in primary cells and tissue. We show for the first time that TDF induces progestogen-independent activation of the progesterone receptor (PR) in a cell line. The ability of TDF and DPV to influence GR and PR activity suggests that their use may be associated with steroid receptor-mediated off-target effects. This, together with cell line and individual donor gene expression responses in the primary models, raises concerns that reciprocal modulation may cause side effects in a cell- and donor-specific manner in vivo.

INTRODUCTION

Women in sub-Saharan Africa are disproportionately affected by human immunodeficiency virus type 1 (HIV-1) and face a high risk of unintended pregnancy and sexually transmitted infections (1). As a result, millions of HIV-1-positive women are concurrently taking oral antiretroviral drugs (ARVs) and progestogen-containing hormonal contraceptives. Multipurpose prevention technologies (MPTs) are in development to safely and effectively protect healthy women against HIV-1 and unwanted pregnancy. Understanding the relationship between ARVs and hormonal contraceptives is therefore essential for insight on potential short- and long-term side effects associated with their simultaneous use. Adverse effects associated with the use of ARVs and hormonal contraceptives include bone density loss, kidney and liver toxicity, cardiovascular damage, lactic acidosis, and effects on immune function (2, 3). However, studies on the reciprocal modulation of ARV and synthetic progestogen (progestin) intracellular activity are limited. We hypothesize that progestins may affect the efficacy of ARVs and that ARVs may modulate steroid receptor activity in the absence and presence of steroids, which may impact the efficacy and side effects of these drugs.

Tenofovir disoproxil fumarate (TDF), the prodrug form of tenofovir (TFV), is a nucleotide reverse transcriptase inhibitor, widely used in combination with other ARVs in the HIV-1 treatment regimen (4). Currently, TDF is approved as an oral preexposure prophylaxis (PrEP) method in combination with emtricitabine (FTC) for the prevention of HIV-1 infection in healthy individuals. It has been shown in phase I clinical trials to be safe and well tolerated as an intravaginal ring in sexually abstinent women; however, a recent phase I trial assessing the safety and pharmacokinetics of a 90-day TDF intravaginal ring in sexually active women found that the ring causes ulcerations and increases inflammatory markers in women using the ring compared to those in women receiving a placebo (5, 6). Dapivirine (DPV), a nonnucleoside reverse transcriptase inhibitor, is the most clinically advanced microbicide targeted for intravaginal delivery. In 2016, two phase III clinical trials found that a 30-day DPV intravaginal ring reduced the risk of HIV infection by approximately 30%, and modeling data from follow-up open-label extension studies have suggested a greater risk reduction of approximately 50% overall for women using the ring over a 1-year period (7–9). The DPV ring is now under regulatory review for use in developing countries with high HIV incidence.

The most commonly used progestin in sub-Saharan Africa is the 3-monthly injectable contraceptive medroxyprogesterone acetate (MPA), referred to as Depo-Provera or depot MPA intramuscular (DMPA-IM) (10). Levonorgestrel (LNG), widely used in the region as a contraceptive in different formulations and with different delivery methods, is also a leading candidate for the development of dual ARV–progestin MPTs. These progestins are synthetic steroid ligands that elicit intracellular responses by binding to and activating steroid receptors (11). The primary targets for progestins are the progesterone receptors (PR-A and PR-B); however, some progestins, such as MPA, are also known to cross talk with other steroid receptors, such as the glucocorticoid receptor (GR) and the androgen receptor (AR) (12). Through their actions as ligand-activated transcription factors, steroid receptors can activate or repress transcription to influence multiple processes in cells and tissues (13). MPA is a known full-to-partial GR agonist and induces immunomodulatory effects and increased HIV infection via GR-mediated mechanisms in vitro (13–15). Clinical epidemiological studies also suggest that DMPA-IM increases the risk of HIV-1 over that with no contraception (16) or an LNG implant (17).

Several clinical and pharmacokinetic studies have evaluated the effects of simultaneous use of some ARVs and progestins on pregnancy, HIV-1 acquisition or disease progression, and changes in serum drug concentrations. To date, efavirenz (EFV) is the only ARV shown to decrease contraceptive efficacy by influencing the metabolism of progestins in women, resulting in a decrease in progestin levels (18, 19). A study evaluating the effects of hormonal contraceptives on ARV efficacy showed that DMPA-IM, LNG, and combined oral contraceptives have no detectable effect on the efficacy of combined antiretroviral therapy containing nonnucleoside reverse transcriptase inhibitors other than DPV or protease inhibitors (20). Limited in vitro studies have shown that TDF and DPV have immunomodulatory effects (21–23), and this may indicate the involvement of steroid receptors such as the GR, a known immune regulator. A recent in vitro study showed that MPA, but not LNG or norethisterone (NET), is able to inhibit the anti-HIV-1 activities of TFV and tenofovir alafenamide in blood and tissue CD4⁺ T cells, most likely by lowering intracellular concentrations of TFV diphosphate (24). These clinical and in vitro studies suggest that drug-drug interactions do occur in vivo.

Potential off-target effects and drug-drug interactions of ARVs and progestins are likely to be dose dependent. The progestins and ARVs used for PrEP and MPTs exhibit a wide range of concentrations in the vagina and in serum or plasma (Table 1). Additional in vivo studies are required to investigate drug-drug interactions, as are in vitro studies to identify and predict as yet unidentified off-target effects and their dose dependency.

TABLE 1.

Intravaginal and plasma or serum concentrations of ARV and progestin candidates for MPTs

Producta	Concentration(s)				Reference(s)
	Released in product	In cervicovaginal fluid (μM)	In cervical tissue (μM)	In plasma or serum (nM)
ARVs
TFV gel	1% (vol/vol) gelb	41–34,000	0.7–4,800	11.83c	79, 80d
TDF ring	365 mge	88–236	9–30	1.2–5.2	5d
TDF (HAART)	300 mgh	NDf	ND	1,100–1,140	81d
TDF–FTC oral PrEP	300 mgh TDF, 200 mgh FTC	ND	79c	77–87	82g
DPV ring	25 mge	17c	2–21	0.7–0.89	7,d 9,d 54g
Progestins
DMPA-IM (injectable)	150 mge	ND	ND	21c	57,g 83,g 84d
ETG–EE ring (NuvaRing)	120 μgh ETG, 15 μgh EE	ND	0.002c (ETG), 0.0004 (EE)	3.7c (ETG), 0.05c (EE)	85g
LNG-releasing IUS (Mirena)	30 μgh ; 52 mge	100–370	2.5c	0.2,c 0.97c	34, 86d
LNG (implant, oral pill)	150 mge (implant); 1.5 mgh (oral pill)	ND	ND	0.3–28	87,g 88,d 89g
NES–EE ring	150 μgh NES, 15 μgh EE	ND	ND	0.3c	90g
MPTs
DPV + LNG ring	200 mge DPV, 320 mge LNG	ND	ND	2.1c (DPV), 5.1c (LNG)	91d
TFV + LNG ring	8–10 mgh TFV, 20 μgh LNG	34–3,484 (TFV), ND (LNG)	6.9–2,865 (TFV), ND (LNG)	8c (TFV), 0.9–1.6 (LNG)	92d

^aHAART, highly active antiretroviral therapy; ETG, etonogestrel; EE, ethinyl estradiol; IUS, intrauterine system; NES, Nestorone.

^bPericoitally applied.

^cMean concentration.

^dConcentrations in plasma.

^eSingle dose.

^fND, not determined.

^gConcentrations in serum.

^hDaily dose.

In the present study, we evaluated the abilities of select steroid ligands and ARVs to reciprocally modulate their respective intracellular functions, namely, the regulation of gene expression via activation of steroid receptors by steroid ligands and the inhibition of HIV-1 infection by TDF and DPV. Dose-dependent effects of TDF and DPV on GR, PR, and AR activity in the absence and presence of receptor agonists in in vitro cell line, peripheral blood mononuclear cell (PBMC), and ectocervical tissue explant models were investigated, and the effects of MPA, LNG, and other receptor-specific agonists on the dose-dependent inhibition of HIV-1 infection by TDF and DPV were investigated in the TZM-bl reporter cell line.

RESULTS

The concentration ranges of the ARVs used in this study were selected to reflect those that have been measured intravaginally and in blood previously (Table 1). The steroid concentrations chosen were intended to fully saturate possible cognate steroid receptors or, for dose-response analysis, to span the pharmacologically relevant concentrations of steroids (13, 25, 26). Steroid stimulation times of 24 or 48 h were used, since these have been previously established to elicit inflammatory gene responses in the models used in the study.

Effects of ARVs and steroids on cell and tissue viability.

We used 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays to investigate whether ARVs affect cell or tissue viability. In TZM-bl cells, concentrations of TDF ranging from 0.01 to 100 μM and concentrations of DPV ranging from 0.01 to 1 μM did not affect cell viability (Fig. 1A and B). However, 10 and 100 μM DPV significantly reduced cell viability (Fig. 1A and B). In PBMCs, only a 100 μM concentration of either ARV significantly reduced cell viability (Fig. 1C and D). A 5% concentration of dimethyl sulfoxide (DMSO) was used as a positive control for cytotoxicity in TZM-bl cells (Fig. 1A and B), and a range of 5 to 20% DMSO was used in PBMCs (data not shown). The viability of ectocervical explant tissues appeared unaffected by TDF and DPV concentrations ranging from 1 to 100 μM (Fig. 1E and F). Cell viability with ARVs was not altered by the presence of 100 nM steroids (dexamethasone [DEX] or MPA), alone or in combination with 1 μM ARV, in TZM-bl cells. However, in PBMCs, the presence of 100 nM DEX significantly reduced cell viability by approximately 30% (see Fig. S1A and B in the supplemental material). We next performed gene expression analysis in the various models using only ARV concentrations shown not to reduce viability in the cell/tissue models, in order to avoid confounding the interpretation of the data. Given the slight reduction in cell viability by 100 nM DEX in PBMCs, all the PBMC gene expression results were normalized for cell viability and interpreted with due caution.

FIG 1.

Effects of ARVs on the viability of TZM-bl cells, PBMCs, and cervical explant tissues. TZM-bl cells (A and B), PBMCs (C and D), and ectocervical explant tissues (E and F) were treated with different concentrations of TDF and DPV, ranging from 10⁻¹⁰ M to 10⁻⁴ M, in triplicate, for 24 h, and DMSO (0.1%, vol/vol) was used as the vehicle control. A 5% concentration of DMSO was used as the positive control in TZM-bl cells and PBMCs (data not shown). Cells were treated with MTT reagent for 2 h, and after solubilization, the absorbance was read at 595 nm. Cell viability was normalized to that for the vehicle control, which was set to 100%. Graphs show pooled results of three independent experiments for TZM-bl cells (A and B), data from four PBMC donors, with experiments for each condition carried out in triplicate and results represented as means ± standard deviations (C and D), or pooled results from four cervical explant donors (E and F). Statistical analysis was carried out using one-way ANOVA, followed by Tukey’s multiple-comparison posttest. Statistical significance in the comparison of the vehicle control to ARV treatment is indicated by asterisks (**, P < 0.01; ***, P < 0.001).

ARV modulation of endogenous steroid-regulated inflammatory genes.

To examine the ability of ARVs to modulate endogenous immunomodulatory genes, the relative mRNA expression of the established (27–29) steroid-regulated anti-inflammatory glucocorticoid-induced leucine zipper (GILZ) gene and of two proinflammatory genes, encoding interleukin 6 (IL-6) and IL-8, was assessed in TZM-bl cells, PBMCs, and tissue explants. These genes were chosen as model GR-regulated genes. GILZ has been shown previously to be upregulated in several model systems by glucocorticoids via a mechanism involving binding of the GR to glucocorticoid response elements in the promoter (30). IL-6 has been shown previously to be downregulated in several model systems by glucocorticoids and other steroids via a mechanism involving tethering of the steroid receptor to other transcription factors, such as NF-κB, recruited to the promoters (31), while effects on IL-8 are reportedly variable (32, 33). Cells and tissues were stimulated with 1 μM DPV or 10 μM TDF in the absence and presence of 100 nM DEX or MPA. In all three model systems, DEX and/or MPA was found to increase GILZ mRNA levels and to repress IL-6 mRNA levels, as expected, while no statistically significant effects were observed for IL-8. The results are summarized in Table S1 in the supplemental material.

In TZM-bl cells, neither TDF (up to 10 μM) nor DPV (up to 100 nM) affected GILZ mRNA levels on their own (see Fig.S2A and S3A and B in the supplemental material). However, TDF, more than DPV, increased the efficacy of MPA-induced GILZ expression in MPA dose-response analysis, although these effects were not statistically significant (Fig. 2A and B). TDF dose-response analysis (1 to 100 μM TDF) in the presence of 100 nM MPA revealed that TDF at 10 and 100 μM significantly increased MPA-induced GILZ expression by approximately 1.5- and 2-fold, respectively (Fig. S2A), but had no significant effect on the MPA responses of IL-6 and IL-8 in TZM-bl cells (Fig. S2B and C). Unlike TDF, DPV alone significantly induced IL-8 mRNA expression by 11-fold, and appeared to induce IL-6 mRNA expression by 10-fold, in TZM-bl cells (Fig. 2C and D). DEX and MPA significantly repressed IL-6 expression by 10-fold and 5-fold, respectively, and appeared to induce IL-8 expression by 3.4-fold and 3.5-fold, respectively (Fig. 2C and D). Costimulation with TDF did not appear to alter these DEX or MPA responses (Fig. 2C and D). However, costimulation with DPV inhibited DEX and MPA repression and returned IL-6 expression to basal levels (Fig. 2C). Interestingly, for IL-8 expression, the effect of the combination of DPV with DEX or MPA appeared to be steroid specific. DEX in combination with DPV did not alter DPV-induced IL-8 expression; however, the combination of MPA and DPV resulted in a significant 1.6-fold potentiation of the DPV-induced IL-8 response (Fig. 2D). Lower doses of TDF and DPV, in the range of 0.1 to 100 nM, had no significant effect on GILZ, IL-6, or IL-8 mRNA levels, either in the absence or in the presence of MPA (Fig. S3).

FIG 2.

Effects of TDF and DPV on inflammatory gene mRNA levels in human cervical TZM-bl cells. Gene expression was evaluated after 24 h of stimulation with the indicated concentrations of ARVs and steroids. GILZ mRNA expression with increasing doses of MPA in the absence and presence of 10 μM TDF (A) or 1 μM DPV (B). IL-6 (C) and IL-8 (D) mRNA expression in the presence of 100 nM DEX or MPA, 1 μM TDF or DPV, or combinations thereof. EtOH plus DMSO (0.1% [vol/vol] for each) was used as the vehicle control. RNA was isolated, and cDNA was synthesized. Then mRNA expression levels were determined by qRT-PCR and were normalized to GAPDH mRNA expression levels. The relative fold change in expression was determined by setting the value for the vehicle control to 1. Pooled results of three or more independent experiments are shown and are represented as means ± SEM. Panels A and B were assessed for statistical significance using two-way ANOVA with Tukey’s multiple-comparison posttest for comparison of the curves, although no statistical significance was obtained. For panels C and D, statistical significance was assessed with a nonparametric Kruskal-Wallis test, with Dunn’s multiple-comparison posttest for comparison of the conditions. Statistical significance is indicated by asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Results in PBMCs are shown for pooled effects on 13 donors (Fig. 3), as well as for subgroups of donors, in order to illustrate more clearly the diversity and frequency of donor-specific effects (see Fig.S4 to S6 in the supplemental material). In PBMCs, TDF and DPV had no detectable effect on GILZ mRNA levels for the majority of donors (Fig. 3A; also Fig. S4A). As expected, DEX treatment significantly induced GILZ mRNA expression by 8-fold, as for TZM-bl cells, but this induction remained unchanged in the presence of ARVs, in contrast to the response in TZM-bl cells. As expected, DEX also significantly repressed IL-6 mRNA levels by about 3-fold, and this repression was not influenced by combination with ARVs (Fig. 3B). DEX alone and in combination with ARVs had no detectable effect on IL-8 mRNA levels (Fig. 3C). In the majority of donors, DPV and TDF showed no detectable significant effects on IL-6 and IL-8 mRNA expression (Fig. 3B and C; also Fig.S5B and S6B). The PBMC responses in donor samples showed a high degree of variability. Subgroup analysis did, however, reveal donor-specific effects, with some donors exhibiting at least 2-fold upregulation (as for IL-6 and IL-8 in TZM-bl cells) or downregulation in the mRNA levels of some genes in response to ARVs (Fig.S5 and S6).

FIG 3.

Immunomodulatory effects of ARVs in PBMCs. Nonactivated PBMCs were stimulated with TDF and DPV at 1 μM in the absence and presence of 100 nM DEX, or with the vehicle control (EtOH plus DMSO, each at 0.1% [vol/vol]) for 48 h. RNA was isolated from PBMCs, and thereafter cDNA was synthesized. Relative changes in GILZ (A), IL-6 (B), and IL-8 (C) mRNA expression levels were determined by qRT-PCR and were normalized to GAPDH mRNA expression levels. The relative fold change in expression was determined by setting the value for the vehicle control to 1. Pooled results of 13 patients are shown for each condition (with experiments carried out in triplicate) and are represented as means ± SEM. A Kruskal-Wallis test was performed with Dunn’s multiple-comparison posttest to determine significant differences between treatments. Statistical significance is indicated by asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Experiments in explant tissue were performed with DPV and not TDF, due to the limited availability of fresh tissue. Most of the results were very similar to those obtained in PBMCs (Fig. 4A to C). DPV alone appeared to have no effect on GILZ mRNA (Fig. 4A). As observed in PBMCs and TZM-bl cells, DEX significantly induced GILZ mRNA by 8-fold, and MPA appeared to increase GILZ mRNA by 2.4-fold (Fig. 4A). MPA or DEX alone appeared to repress IL-6 by approximately 1.5- or 2-fold, respectively (Fig. 4B), but had no effect on IL-8 (Fig. 4C), as observed in PBMCs and TZM-bl cells. DPV did not appear to modulate the DEX or MPA responses for any of the genes (Fig. 4A to C). As with PBMCs, there was high donor-specific variability in responses, making it difficult to establish significance for small effects. While DPV alone appeared to have no effect on any of the three genes for the pooled samples, it appeared to have a proinflammatory effect on IL-6 and IL-8 for some of the donor samples (Fig. 4B and C), as for some PBMC donor samples and for TZM-bl cells.

FIG 4.

Immunomodulatory effects of ARVs in ectocervical explant tissue. Ectocervical explants were stimulated with 1 μM DPV or 100 nM DEX or MPA, or combinations thereof, as indicated. RNA was isolated, and thereafter, cDNA was synthesized. Relative changes in GILZ (A), IL-6 (B), and IL-8 (C) mRNA expression levels were determined by qRT-PCR and were normalized to GAPDH mRNA expression levels. The relative fold change in expression was determined by setting the value for the vehicle control to 1. Pooled results of matched incubations performed on seven donors are shown (with each condition tested in triplicate) and are represented as means ± SEM. A Kruskal-Wallis test was performed with Dunn’s multiple-comparison posttest to determine significant differences between treatments. Statistical significance is indicated by asterisks (*, P < 0.05; **, P < 0.01).

ARV effects on steroid receptor transcriptional activity.

We next investigated the effects of ARVs on the efficacy and potency of the transcriptional effects of GR, PR-B, and AR agonists by using promoter-reporter assays. Western blotting showed that steroid receptors were successfully exogenously expressed in U2OS cells (see Fig. S7 in the supplemental material). Assessment of the in vitro toxicity of TDF and DPV in U2OS cells by use of the MTT cell viability assay showed that concentrations as high as 1 μM for either ARV were not deleterious, while 10 μM DPV but not TDF resulted in a loss of viability (see Fig. S8).

TDF (1 μM) significantly increased the efficacy of DEX transcriptional activity via the GR by 33%. DPV had a similar effect, which approached significance (P = 0.0531), increasing DEX transcriptional activity by 25% (Fig. 5A and B; see also Table S2 in the supplemental material). These increases were similar to the effects of both TDF and DPV on MPA in TZM-bl cells (Fig. 2A and B; also Fig. S2A). The potency of DEX was not significantly altered in the presence of either ARV (Table S2). Neither ARV appeared to have a significant effect on the efficacy and potency of the synthetic AR agonist mibolerone (MIB) (Fig. 5C and D; see also Table S3). Surprisingly, 1 μM TDF, unlike DPV, significantly increased PR-B-mediated transcriptional activity alone and in the presence of low concentrations of LNG (0.0001 nM to 0.001 nM) (Fig. 5E). At higher concentrations of LNG, TDF did not significantly affect the efficacy of LNG via the PR. DPV also had no effect on the transcriptional efficacy and potency of LNG via the PR (Fig. 5F; see also Table S4). The potencies and efficacies for each of the dose-response curves in Fig. 5 are detailed in Tables S2 to S4.

FIG 5.

Effects of TDF and DPV on the transcriptional efficacies and potencies of GR, PR-B, and AR in the presence of their agonists. U2OS cells were seeded onto 10-cm² plates at a density of 1.5 × 10⁵ and were incubated for 24 h. Thereafter, cells were transiently transfected for 24 h with pTAT-GRE-LUC and pcDNA-3 (empty vector) or the receptor expression vector pSV-hAR, pcDNA3-hGR-WT, or pSG5-PRB. Cells were reseeded onto 96-well plates at densities of 1 × 10⁴/well for GR and AR and 5 × 10⁴/well for PR-B. The cells were then treated with increasing concentrations of the receptor agonist MIB, DEX, or LNG in the absence or presence of 1 μM TDF or DPV. EtOH plus DMSO (0.1% [vol/vol] for each) was used as the vehicle control, and cells were incubated for 24 h. Cells were lysed, and the luciferase activity was measured for GR (A and B), AR (C and D), and PR-B (E and F). Luciferase activity was normalized to the protein content per well as determined by the Bradford assay. Furthermore, luciferase activity was normalized to the plateau value of the reference ligand (DEX, MIB, or LNG), which was set to 100%, in order to obtain the relative fold induction. Pooled results from three or more independent experiments are shown, and data are represented as means ± SEM. Unpaired t tests were used to determine the statistical significances of efficacies and potencies. **, P < 0.01.

Following the observation that TDF transactivated PR-B in a progestogen-independent manner, the dose dependency of this response was similarly assessed using a promoter-reporter assay in U2OS cells. The cells were stimulated with increasing concentrations of TDF, ranging from 1 nM to 1 μM, in the absence or presence of 100 nM progesterone (P₄). TDF had a 50% effective concentration (EC₅₀) of 599 ± 33 nM for PR-B, and the potency of TDF was significantly decreased in the presence of 100 nM P₄ (Fig. 6; see also Table S5 in the supplemental material). Agonist-independent effects of TDF and DPV at concentrations ranging from 100 nM to 10 μM via the AR and GR were also assessed, but no significant induction of these receptors was observed in the absence of receptor agonists (see Fig. S9).

FIG 6.

Dose-dependent effects of TDF on ligand-independent PR-B activation. U2OS cells were seeded and incubated for 24 h. Thereafter, cells were transiently transfected for 24 h with pTAT-GRE-LUC and pcDNA-3 (empty vector) or pSG5-PRB. Cells were then reseeded and were treated with various concentrations of TDF (10⁻⁹ M to 10⁻⁶ M) in the absence or presence of 100 nM P₄ for 24 h. Cells were lysed, and luciferase activity was measured. Luciferase activity was normalized to the protein content per well as determined by the Bradford assay and to the vehicle control (0.1% EtOH, 0.1% DMSO) in order to obtain the normalized relative fold induction. Pooled results from three independent experiments are shown, and data are represented as means ± SEM. Unpaired t tests were used to determine the statistical significances of efficacies and potencies. ****, P < 0.0001.

Similarly, the induction of PR-B transcriptional activity by TDF was investigated in the MDA-MB-231 cell line, which stably and constitutively expresses PR-B (Fig. 7; see also Table S6 in the supplemental material). As expected, 1 μM TDF activated PR-B in the absence of LNG (Fig. 7) and significantly increased the transcriptional efficacy of LNG in these cells (Table S6). As shown in U2OS cells, DPV did not influence the transcriptional efficacy or potency of LNG in MDA-MB-231 cells.

FIG 7.

Effects of ARVs on endogenous PR activity in the presence of LNG in MDA-MB-231 cells. MDA-MB-231 cells stably transfected with PR-B were transiently transfected with 9 μg pTAT-GRE-LUC and thereafter were reseeded onto 96-well plates at a density of 1 × 10⁴/well. Cells were treated with increasing concentrations of LNG in the absence or presence of 1 μM TDF or DPV for 24 h. EtOH plus DMSO (0.1% [vol/vol] for each) was used as the vehicle control, and cells were incubated for 24 h. Cells were lysed, and the luciferase activity was measured for the PR. Luciferase activity was normalized to the protein content per well as determined by a Bradford assay. Furthermore, luciferase activity was normalized to the plateau of the reference ligand LNG, which was set to 100%, in order to obtain the relative fold induction. Pooled results from four independent experiments are shown, and data are represented as means ± SEM. Unpaired t tests were used to determine the statistical significances of efficacies and potencies. *, P < 0.05.

The data observed in the current study reveal that TDF, but not DPV, significantly affects the function of the steroid receptors GR and PR, while neither TDF nor DPV has a significant effect on the AR. These differential activities may result from direct interaction between the ARVs and the steroid receptors. To investigate this, in silico molecular docking was employed to assess whether TDF and DPV exhibited an affinity for the ligand-binding pockets of either the GR, PR-B, or the AR (for the latter receptor, two structures were evaluated.) To validate these calculations, binding affinities were first predicted for the crystallized agonist in complex with the receptor ligand-binding domain of each receptor. Following these test simulations, the results suggested that TDF had low affinity for the ligand-binding pockets of all three steroid receptors, in contrast to the receptor agonists (see Table S7 in the supplemental material). DPV exhibited some binding affinity for the PR and GR, although this was not reflected in transcriptional activity: DPV had no significant effect on GR or PR activity on its own. Mono- and dianionic charged states of TFV were also evaluated, and these species exhibited negligible affinity for the ligand-binding domains of the three receptors.

Modulation of ARV efficacy by steroids.

The effects of steroid ligands on the efficacies and potencies of ARVs for inhibition of HIV-1 replication were assessed using HIV-1 infection assays in TZM-bl cells. Ligands that bind to either the GR (DEX), the AR (MIB, LNG), or both the AR and GR (MPA) were chosen (25, 26, 29, 34). DEX and MPA alone appeared to increase the level of HIV-1 infection, as reported previously (15), although in this experiment, the effect was not statistically significant (Fig. 8). We found that 100 nM DEX, MPA, or LNG and 0.01 nM MIB did not significantly affect the efficacy or potency of TDF or DPV for the inhibition of HIV-1 infection (Fig. 8).

FIG 8.

Inhibition of HIV-1 infection by TDF and DPV in the absence and presence of steroid receptor agonists. TZM-bl cells were exposed to increasing concentrations of TDF or DPV in the absence or presence of 100 nM DEX or 0.01 nM MIB (A and B), or in the absence or presence of 100 nM MPA or LNG (C and D), for 24 h prior to exposure to HIV-1_BaL-Renilla in the presence of the compounds for 48 h. EtOH plus DMSO (0.1% [vol/vol] for each) was used as the vehicle control. HIV-1 infection was determined by measuring HIV-1 long terminal repeat activity, as expressed by relative luciferase units (RLU), using the Bright-Glo system (Promega), and these values were normalized to the corresponding MTT absorbance readings (RLU/MTT). Pooled results from four independent experiments are shown, where each condition was tested in triplicate, and data are represented as means ± SEM.

DISCUSSION

In the present study, we evaluated the abilities of TDF and DPV to reciprocally modulate their respective intracellular functions alone and together with steroid ligands.

In TZM-bl cells, PBMCs, and explant tissue, DEX or MPA alone exhibited anti-inflammatory effects by upregulation of GILZ and downregulation of IL-6 mRNA. These results provide confidence that the detection of expected steroid-induced changes in gene expression in these models was reproducible. Differential regulation of select immune function genes by TDF and DPV in the absence and presence of steroid ligands was observed in TZM-bl cells. However, no effects were detected for the genes assessed in pooled data from PBMCs and explant tissue. In TZM-bl cells, DPV induced proinflammatory effects on the IL-6 and IL-8 genes, unlike TDF. High concentrations of TDF and DPV were found to increase the efficacy of MPA-induced effects on GILZ in TZM-bl cells. Similar effects were observed for DEX-induced efficacy with a promoter-reporter plasmid in U2OS cells. DPV effects on the IL-6 and IL-8 genes were modulated by DEX or MPA in TZM-bl cells, and this was not seen for TDF. DEX and MPA inhibited the DPV-induced proinflammatory response of the IL-6 gene. MPA also enhanced the proinflammatory effect of DPV on IL-8 in TZM-bl cells, an effect not observed for DEX. In contrast, the results in PBMCs and explants are reassuring in that they suggest, if they occur in vivo, that there is very little reciprocal modulation of immunomodulatory gene expression for these ARVs and clinically relevant steroids. However, the cell line results provide proof of concept that reciprocal modulation could occur in some cells, including both ARV- and steroid-specific effects, but their physiological relevance is unclear.

The results for both PBMCs and explants show a high degree of interindividual donor variability, as reported previously (35, 36). Some of the individual donor responses that we observed in the primary models correspond to effects observed in TZM-bl cells and raise the possibility that for some donors, ARVs may have both proinflammatory and anti-inflammatory effects, and some reciprocal modulation between these ARVs and steroids could occur.

These donor-specific effects on gene expression in primary models may be reflected in the various responses of patients to ARV treatment, the underlying factors for which may include patient genotype, age, levels of endogenous hormones and other drugs, nutrition, and disease state (37–39). About 60% of the explant tissue donors were positive for herpes simplex virus 1 (HSV-1), and their ages ranged from 31 to 56 years, factors that may contribute to the various responses observed. Female donors of PBMCs were negative for some infections, including HIV-1 and syphilis, but were not tested for all possible infections or HSV-1 infection, and thus, it is not known whether other underlying infections influenced the immune gene responses to ARVs.

The immunomodulatory effects that we observed with TDF (0.1 nM to 100 μM) and DPV (0.1 to 1 μM) in TZM-bl cells in the present study occurred at physiologically relevant ARV doses (for TDF, 1.2 nM to 79 μM; for DPV, 0.7 nM to 21 μM) (Table 1). We did not detect responses of select genes in vitro in PBMCs with 1 μM TDF or DPV, concentrations greater than those detected (0.7 to 5 nM) in the sera of patients, suggesting that these ARVs alone are unlikely to have systemic immunomodulatory effects in vivo. However, it is possible that effects on PBMCs may be observed at lower ARV concentrations. We also did not detect responses of select genes in vitro in ectocervical tissue with 1 μM DPV alone, suggesting that inflammatory effects do not occur at this concentration. However, such effects may occur at higher concentrations in the female genital tract (FGT), as measured for the DPV vaginal ring (Table 1). In vivo clinical data for the DPV vaginal ring suggest that this is unlikely, since no serious adverse effects of the ring indicating excessive inflammation in the genital tracts of women who used it were detected in phase II and III clinical trials (9, 40). Phase I clinical trials of the TDF vaginal ring in sexually abstinent women reported previously that it was safe, with minimal adverse events (5). However, a recent phase I clinical trial showed that women using the TDF ring had higher expression of cytokines and chemokines, including IL-6 and IL-8, than women using a placebo ring (6). Further experiments are required to investigate immunomodulatory effects of high doses of DPV and TDF in FGT tissue models.

Our data in TZM-bl cells contribute to the limited in vitro studies that show that ARVs influence immune function independently of their HIV-1-inhibitory action. The immunomodulatory effects of TDF in human PBMCs have been described previously. After infection with live bacteria, followed by stimulation with Toll-like receptor (TLR) or tumor necrosis factor alpha (TNF-α), TDF at doses ranging from 12.5 to 50 μM decreased the expression of the proinflammatory cytokine IL-8 and the anti-inflammatory cytokine IL-10 and increased the expression of the proinflammatory cytokine IL-12 (41). TFV at a concentration of 3.5 mM has been shown to upregulate the expression of macrophage inflammatory protein 3α (MIP-3α), IL-8, and TNF-α in macrophages derived from blood monocytes, as well as in primary epithelial cells derived from the endometria and ectocervices of healthy women (21, 22). However, these reports (21, 22, 41) used much higher concentrations of these ARVs than are detected systemically. Additionally, in human genital epithelial cells, DPV at concentrations of ≥10 μM was shown to produce small increases in IL-8 secretion (23), in agreement with our findings in TZM-bl cells. Our results further suggest that side effects of TDF and DPV due to loss of cell viability are unlikely to occur at concentrations up to 10 μM for both ARVs in PBMCs and 100 μM in ectocervical tissue, although this is likely to be cell and tissue specific.

To our knowledge, this is the first study to investigate the effects of ARVs on GR, PR, and AR transcriptional activity in the presence of their respective agonists. Svard et al. reported previously that a panel of ARVs, including TFV and EFV, bind to the liver X receptor (LXR) and the estrogen receptor (ER) (42). In the study by Svard et al., ARVs predicted to bind to the GR were not able to bind in vitro or to induce GR transcriptional activity (42). Our results show that TDF and DPV alone do not affect the transcriptional activity of the GR (Fig. S9 in the supplemental material). We show for the first time that TDF and DPV increase the transcriptional efficacy of the GR agonist DEX in U2OS cells and that TDF significantly activates the PR in a ligand-independent manner. In contrast, TDF and DPV had no effect on the potency or efficacy of the AR agonist MIB.

These in vitro ARV effects on GR and PR activity are potentially important if they are translated in vivo. For the ubiquitously expressed GR, our results suggest that in vivo, TDF and DPV may potentiate the effects of GR ligands at times, such as when cortisol levels are high or during glucocorticoid therapy. This may be relevant for several GR-regulated physiological processes, including metabolism, bone mineral density (BMD), and cardiovascular and immune function (43). BMD loss is a known side effect of TFV use in HIV-1-positive patients and in HIV-1-negative PrEP users (44, 45). Several in vitro studies have implicated TDF in decreasing BMD (46, 47). Activation of the GR has also been implicated in decreasing BMD (48, 49). Whether the GR is involved in any of the above effects of TDF is unknown. Since the progestin MPA is a potent partial-to-full agonist of the GR (50), HIV-1-positive women using DMPA-IM for contraception and taking TDF or DPV may exhibit increased GR activity leading to increased side effects, such as loss of BMD or immunosuppression, and/or increased beneficial effects of other GR-regulated physiological functions.

A novel finding of this study is that TDF, but not DPV, significantly activates the transcriptional activity of PR-B in the absence of a progestogen and increases the efficacy of PR-B in the presence of the progestogen LNG. This has important implications that may be relevant to PR-regulated physiological processes, including reproduction, reproductive tissue cancers (51), immune function, and bone density. TDF and TFV have not been shown previously to affect reproductive functions (52, 53). BMD is regulated by both estrogen and P₄ (54–56). A dose-dependent relationship with BMD loss has also been shown for adolescent girls receiving high-dose DMPA-IM (150 mg) compared to the lower-dose subcutaneous DMPA (DMPA-SC) (104 mg) (54–58). These studies suggest that high concentrations of PR ligands such as MPA and LNG may create a hypoestrogenic environment and reduce BMD. Our results suggest a possible link between TDF use, the PR and/or the GR, and BMD loss. Since TDF used as an intravaginal ring for PrEP is set to deliver concentrations as high as 30 μM to the FGT (6), where the GR and PR are abundantly expressed (unpublished data; also reference 59), our results suggest that side effects with TDF and/or DPV via the GR or PR in the FGT may be highly relevant. PBMCs have also been shown to express GR mRNA and protein (60, 61), and some studies (60) but not others (61) show that they express detectable PR. This suggests that side effects with TDF and/or DPV via the GR or PR may also be highly relevant in a systemic context.

The mechanisms whereby TDF but not DPV affects PR-B function remain to be determined and were beyond the scope of the present study. Our in silico docking data suggest that this is not likely to be due to direct binding of TDF to the PR-B ligand-binding pocket. The finding that TDF alone does not cause similar activation of the GR or AR suggests that the mechanism is specific to PR-B signaling and is unlikely to involve general components of the transcriptional machinery. A mechanism involving a direct effect on PR-B, other than binding to the ligand-binding pocket, or an indirect effect, such as an effect on a protein that interacts specifically with PR-B, may be involved. Further studies on the transactivation and transrepression of endogenous genes and more-detailed mechanistic studies may provide further insights.

An encouraging finding of our study was that neither MPA nor LNG affected the potency or efficacy of TDF or DPV in the inhibition of HIV-1_BaL-Renilla replication in TZM-bl cells. We have also shown previously that this is the case for MPA and maraviroc (15), suggesting that these progestins, when used for contraception, are unlikely to affect the efficacy of these ARVs for HIV-1 inhibition in women. This is consistent with previous clinical studies showing that MPA does not affect ARV efficacy (20, 62). In contrast, a recent in vitro study showed that MPA, but not LNG or NET, inhibited the anti-HIV-1 activity of TFV in blood and endometrial CD4⁺ T cells, most likely via lowering of the intracellular TFV diphosphate levels (24). This suggests that the influence of MPA on ARV efficacy and potency may differ depending on the cell type. A previous clinical study has also shown that TDF levels are lowered in pregnant HIV-negative women, also suggesting that changes in hormones during pregnancy may influence ARV efficacy (63). Previous clinical studies have not shown any interactions of LNG with ARVs, and hence, LNG may be the more suitable choice for use as a hormonal contraceptive in MPTs (64, 65).

Our study had several limitations, including the limited donor size of PBMC and explant models, which affected the statistical power of the study as well as our ability to fully characterize the effects of both ARVs and the progestins on inflammatory gene expression. Furthermore, our gene expression analysis was restricted to three genes and models, and we may have observed different results for others. Since we did not investigate the effects of >1 μM concentrations of TDF and DPV in PBMCs, or >1 μM concentrations of DPV in genital tract tissue, it is possible that effects do occur in vitro under those conditions. Furthermore, we cannot draw any conclusions about in vitro effects in other genital tract compartments. The use of a transactivation model only of steroid receptor transcriptional activity is also a limiting factor, since steroid receptor activity also occurs via a transrepression model, and the effect of ARVs on transrepression is unknown. The physiological relevance of the TZM-bl cell results remains to be explored in vivo. Nevertheless, our findings provide novel insights into the mechanisms and reciprocal modulation of activities with combinatorial usage of ARVs and GR, AR, and PR ligands, which may have important implications in vivo. Our PBMC and explant gene expression results and HIV-1 infection data are largely reassuring in showing that TDF and DPV, in combination with DEX or MPA, do not reciprocally modulate key biological effects, and thus, these ARVs would be suitable for combination with MPA in MPTs. However, the TZM-bl gene expression data, individual gene expression responses for some donors in the primary models, and the effects of the ARVs on GR and PR function in vitro raise concerns that some negative effects could occur in a cell- and donor-specific manner in vivo. Further experiments are required to evaluate the dose- and time-dependent effects of ARVs and steroid receptor ligands in vitro and their physiological relevance in vivo.

MATERIALS AND METHODS

Ethics.

Ethical approval to conduct this study was granted by the Human Research Ethics Committee of the Faculty of Health Sciences of the University of Cape Town, Cape Town, South Africa (approval no. HREC 210/2011).

Test compounds.

The ARVs TDF and DPV were purchased from the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH (USA), and Selleck Chemicals (USA), respectively. The ARVs were made up to a stock concentration of 10⁻¹ M in DMSO (Sigma-Aldrich, South Africa). DMSO was chosen as the solvent due to its high maximal solubility of approximately 10⁻¹ M for both TDF and DPV (66, 67). Stock concentrations of ARVs were serially diluted 1:10 with DMSO from 10⁻² M to 10⁻⁷ M. The steroids used in the study included the progestins MPA and LNG, as well as the synthetic GR and AR agonists DEX and MIB, respectively. These were obtained from Sigma-Aldrich (South Africa), except for MIB, which was obtained from Perkin-Elmer (USA). Stock concentrations of steroids ranged from 10⁻³ M to 10⁻⁶ M in ethanol (EtOH). Ligands or a control vehicle was added to cells to give the final concentrations indicated in the figures, such that all incubation mixtures contained 0.1% (vol/vol) EtOH and 0.1% (vol/vol) DMSO. The use of EtOH and DMSO as the vehicle at these low final concentrations is common practice in the literature and was not toxic to the cells (25, 41, 68).

Cell culture.

Human cervical indicator TZM-bl cells (NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH) were used for the HIV-1 inhibition assays. U2OS human osteosarcoma cells were used for the luciferase reporter assays because they are deficient in endogenous steroid receptors (American Type Culture Collection [ATCC], USA). MDA-MB-231 cells stably transfected with PR-B were a kind gift from Valerie Lin (Nanyang Technological University, Singapore) and were also used for luciferase reporter assays. HEK293T human embryonic kidney cells (ATCC, USA) were used to generate infectious molecular clones. All cells, except the MDA-MB-231 breast cancer cells, were grown in 75-cm² flasks in full Dulbecco’s modified Eagle’s medium (DMEM) (Sigma-Aldrich, South Africa) supplemented with 1 mM sodium pyruvate (Sigma-Aldrich, South Africa), 44 mM sodium bicarbonate (Sigma-Aldrich, South Africa), 10% (vol/vol) fetal calf serum (FCS) (Thermo Scientific, South Africa), 100 IU/ml penicillin, and 100 mg/ml streptomycin (Sigma-Aldrich, South Africa). MDA-MB-231 cells were cultured in full DMEM supplemented with 7.5% (vol/vol) FCS (Thermo Scientific, South Africa) and 100 mg/ml neomycin (Sigma-Aldrich, South Africa). For experimental incubations with MDA-MB-231 cells, phenol red-free full DMEM supplemented with 5% charcoal-stripped (c-s) FCS was used. Cells were maintained at 37°C in a water-jacketed incubator (at 90% humidity under 5% CO₂).

Cell viability.

Cell viability was determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Sigma-Aldrich, South Africa) according to the manufacturer’s instructions and was measured on a spectrophotometer (Thermo Scientific, USA) at 595 nm. DMSO was used as the positive control for cytotoxicity at concentrations ranging from 5 to 20% (69, 70).

PBMC isolation.

Whole blood from anonymous healthy female donors who were negative for HIV-1, syphilis, and hepatitis B and C was obtained from the Western Cape Blood Service (South Africa) after written informed consent. PBMCs were isolated using Histopaque-1077 Hybri-Max (Sigma-Aldrich, South Africa) density centrifugation with Leucosep tubes (Greiner Bio-One, Germany) according to the manufacturer’s instructions. PBMCs were isolated as described previously (15). Cells were incubated overnight and thereafter were pelleted and washed twice by centrifugation at 250 × g in 1× phosphate-buffered saline (PBS) supplemented with 1% (vol/vol) c-s FCS.

Cervical tissue explants.

Cervical tissue was obtained from seven HIV-1-negative premenopausal women, with normal Pap smears and undergoing hysterectomies for benign reasons, after informed consent. Anonymized fresh tissue was supplied from two sites in the Western Cape, South Africa, namely, Groote Schuur Hospital and Tygerberg Hospital. The majority of the samples were positive for HSV-1 and negative for HSV-2. Cervical tissue was processed as described previously (14) between 1 and 3 h postoperation.

Stimulation with compounds, RNA isolation, and quantitative reverse transcription-PCR (qRT-PCR).

TZM-bl cells were seeded at a concentration of 1 × 10⁵/ml in 12-well plates in full DMEM. After 24 h, cells were stimulated with ARVs or steroids for 24 h and thereafter were harvested in 400 μl Tri Reagent (Sigma-Aldrich, South Africa). PBMCs were seeded into 5-ml Falcon tubes (Becton Dickinson Scientific, South Africa) at a density of 2 million cells in 2 ml full RPMI medium. Subsequently, PBMCs were stimulated with ARVs and steroids for 48 h and thereafter were pelleted by centrifugation at 250 × g for 5 min and harvested in 400 μl Tri Reagent (Sigma-Aldrich, South Africa). TZM-bl cells and PBMCs were then processed for RNA according to the manufacturer’s instructions. Cervical tissue explants were stimulated in triplicate or quadruplicate with steroid ligands and ARVs in full RPMI medium and were incubated at 37°C in a water-jacketed incubator (at 90% humidity under 5% CO₂) for 48 h. For RNA isolation, cervical explants were harvested in 800 μl QIAzol in 2-ml cryovial tubes (Nunc, Germany) and were subsequently homogenized using a hand-held homogenizer (TissueRuptor; Qiagen, The Netherlands) with disposable probes (TissueRuptor probes; Qiagen, The Netherlands). Cervical explant tissue RNA was thereafter isolated using the RNeasy Microarray Tissue minikit (Qiagen, The Netherlands) according to the manufacturer’s instructions. The times chosen for incubation of compounds in the different models were based on previous experiments in our laboratory showing that robust changes in gene expression occurred at these time points. Whether different effects occurred after different times was not investigated.

RNA (250 ng) was reverse transcribed using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, South Africa) according to the manufacturer’s instructions. Real-time qRT-PCR was performed using the FastStart Essential DNA Green Master kit (Roche) on a Rotor-Gene 3000 (Qiagen, The Netherlands) qRT-PCR machine according to the manufacturer’s instructions. The genes investigated were the anti-inflammatory GILZ gene and the proinflammatory IL-8 and IL-6 genes. A validated GILZ primer set was purchased from Qiagen South Africa. The IL-6 and IL-8 primers have been reported previously (71). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the reference gene as reported previously (32). The IL-8 and GAPDH primers were used at a concentration of 500 nM, while IL-6 primers were used at 250 nM. The qRT-PCR profiles for IL-6, IL-8, and GAPDH were established by Verhoog et al. (32).

Virus propagation and TZM-bl infection assay.

Initial viral stocks were prepared as described previously (15). TZM-bl cells were seeded at a concentration of 5 × 10⁴/ml in a 96-well flat-bottom culture plate in full DMEM. The following day, the TZM-bl cells were stimulated either with progestin or its vehicle control (0.1% [vol/vol] EtOH) or with ARV or its vehicle control (0.1% [vol/vol] DMSO), or with both, for 24 h in triplicate. Cells were then infected with 20 IU/ml HIV-1_BaL-Renilla and were harvested 48 h later with Bright-Glo luciferase lysis buffer (Promega, USA). Luminescence was determined on a luminometer (Modulus Microplate reader; Promega, USA) in which relative light units (RLU) were measured for each well. Cell viability was measured using the MTT assay by measuring absorbance at 595 nm on a spectrophotometer (Thermo Scientific, USA). Luciferase readings were normalized to MTT values (RLU/MTT). Dose-response data were analyzed relative to the maximal response generated by the vehicle (set to 100%), and a nonlinear regression model was employed with the Hill slope set to unity. Fifty percent infective concentrations (IC₅₀) were compared using an unpaired, two-tailed t test.

Plasmids and transfection.

U2OS cells were transfected with either the GR (72), the AR (73), or the PR (74) reporter system, each containing the steroid receptor plasmid, a luciferase reporter gene plasmid (pTAT-GRE-LUC [75]) containing the receptor response elements, and an empty vector (pCDNA3; Invitrogen, UK) that was used as the negative control for the steroid receptor plasmid. To ensure consistent transfection efficiency, cells were transfected in a 10-cm² dish (Greiner Bio-One International) at a density of 1.5 × 10⁶ using the X-tremeGENE 9 reagent (Roche, South Africa) according to the manufacturer’s recommendations. Thereafter, the cells were trypsinized, replated into 96-well plates, and stimulated with compounds for 24 h. The transfection conditions for each receptor were as follows: for the human GR (hGR), 10 μg pcDNA3-hGR plus 3.75 μg pTAT-GRE-LUC; for the human AR (hAR), 2.5 μg pSV-hAR plus 1.88 μg pTAT-GRE-LUC; and for human PR-B (hPR-B), 3.5 μg pSG5-hPR-B plus 1.41 μg pTAT-GRE-LUC. MDA-MB-231 cells, which are stably transfected with PR-B, were transfected with 9 μg pTAT-GRE-LUC only.

Luciferase reporter assay.

U2OS cells were incubated with the respective ARVs and/or steroid treatments for 24 h. Thereafter, the cells were washed with ice-cold PBS and were harvested in 25 μl reporter lysis buffer (Promega, Madison, WI, USA). The luciferase activity for each condition was detected in the presence of the substrate luciferin (Promega, Madison, WI, USA) using a Modulus Microplate luminometer and was normalized to the total protein concentration as determined by the Bradford assay (76).

Western blotting.

To confirm steroid receptor transfection, Western blot analysis was performed, essentially as described previously (77), with lysates of U2OS cells bulk transfected with GR, PR, or AR expression vectors or the empty vector pCDNA3 in luciferase reporter assays. Positive controls for each receptor were prepared from COS-1 cells seeded in 12-well plates at a density of 1 × 10⁵/ml and, after 24 h, transfected with 1 μg GR, PR, or AR expression vectors. Transfected U2OS cells were seeded in 12-well plates at a density of 1 × 10⁵/ml. The next day, cells were washed once in PBS, lysed with 50 μl 2× SDS sample buffer (5× SDS sample buffer is 100 mM Tris [pH 6.8], 5% [vol/vol] SDS, 20% [vol/vol] glycerol, 5% [vol/vol] β-mercaptoethanol, and 0.1% [wt/vol] bromophenol blue), and then boiled at 100°C for 10 min. The following antibodies were used: anti-AR (441; catalog no. sc-7305; Santa Cruz Biotechnology) at 1:1,000, anti-GR (G-5; sc-393232; Santa Cruz Biotechnology) at 1:5,000, anti-PR (NCL-L-PGR-312; Leica Biosystems) at 1:1,000, and anti-GAPDH (0411; sc-47724; Santa Cruz Biotechnology) at 1:15,000. A goat anti-rabbit secondary antibody (sc-2313; Santa Cruz Biotechnology) was used for the anti-AR antibody at a 1:10,000 dilution, and an anti-mouse secondary antibody [mouse IgG(κ) light chain binding protein conjugated to horseradish peroxidase (m-IgGκ BP-HRP; sc-516102; Santa Cruz Biotechnology)] was used for the antibodies to GR, PR, and GAPDH and was added at a 1:5,000 dilution.

In silico molecular docking.

All computational predictions were carried out using desktop workstations running the Scientific Linux 7.4 operating system (OS) using the Glide utility included in the Schrödinger suite, release 2017-3. The Protein Data Bank entries of the steroid receptors (1E3G, 1XQ3, 3D90, 4UDC) were prepared using the Maestro PrepWizard. Structures were completed with the addition of bond orders and missing side chains. Nonbound waters were removed, and where applicable, the B chains of dimerized structures were removed. Automated optimization protocols were then run to refine the structures. Glide docking grids of default length, centered on the native ligands, were created. Docking simulations were performed iteratively using the Glide SP setting until a plausible docking pose was found. Binding energies were calculated using the Prime molecular mechanics generalized Born surface area (MM-GBSA) minimization and the binding energy calculation package provided with the Schrödinger suite. MM-GBSA calculations were performed using the variable dielectric generalized Born solvent model. The minimization was performed with flexibility tolerated for all protein atoms within a 10 Å radius of the ligand.

Data analysis.

Results were analyzed using GraphPad Prism (version 7) software from GraphPad Software Inc. (La Jolla, CA, USA). For dose-response curves, the receptor agonists were used as reference ligands, and their values were set to 100%. Dose-response curves were fit with a nonlinear regression model using “log(agonist) versus response,” with a fixed Hill slope of 1, to obtain the best-fit maximal responses. All other curves were then plotted relative to the best-fit maximal value of the reference ligand. All the data were tested for normality, and parametric or nonparametric tests were performed accordingly (78). Unpaired t tests were performed to compare the EC₅₀ values and maximal responses of dose-response curves from different treatments. For experiments that had one condition or two different conditions, parametric one-way analysis of variance (ANOVA) or two-way ANOVA was performed with a Tukey multiple-comparison posttest (comparing all groups to the vehicle control), or a nonparametric Kruskal-Wallis test was performed with Dunn’s multiple-comparison posttest. Data were plotted as means ± standard errors of the means (SEM) on histograms; the number of replicates per condition and the number of independent biological repeats are given in each figure legend.