Characterization of Isoorientin and Paeoniflorin as Botanical Glucocorticoid Receptor Modulators from White Peony and Chasteberry

Dexamethasone (11015), RU486 (10006317), isoorientin (26862), and IBMX (13347) were purchased from Cayman Chemical, Ann Arbor, MI, USA. Progesterone (P0130), paeoniflorin (P0038), and dimethyl sulfoxide (DMSO) (D8418) were purchased from Sigma Aldrich (St. Louis, MO, USA). All small molecules were dissolved in DMSO at a 1000× concentration to make a stock solution, ensuring a maximum of 0.1% DMSO upon treatment. For combination treatments, higher concentrations of stock solutions were prepared and used. The concentrations used throughout our work mirror doses that robustly activate steroid receptors, as determined by our assay optimization. OVCAR5 cells (American Type Culture Collection, ATCC, Manassas, VA, USA) were maintained in RPMI 1640 (10-040-CV, Corning, Corning, NY, USA) supplemented with 10% FBS, 2 mM L-glutamine, and penicillin/streptomycin (100 U/100 µg/mL, cat# 15070063, ThermoFisher, Waltham, MA, USA). Ishikawa cells stably expressing PR-B under a CMV promoter were maintained in phenol red-free DMEM/F12 media (11039–021, Invitrogen, Waltham, MA, USA) and supplemented with 5% charcoal–dextran double-stripped FBS and selected by 250 μg/mL of G418 (A1720, Sigma Aldrich, St. Louis, MO, USA) and 125 μg/mL hygromycin B (H3274, Sigma Aldrich, St. Louis, MO, USA). Ishikawa PR-B cells were generously donated by Dr. Leen Blok, Department of Obstetrics and Gynecology, Erasmus, Medical Center, Rotterdam, The Netherlands [ 30 ]. T47D cells (ATCC, Manassas, VA, USA) were maintained in RPMI 1640 (10-040-CV, Corning, Corning, NY, USA) supplemented with 10% FBS, 2 mM L-glutamine (25030081, Thermo Scientific, Waltham, MA, USA), and penicillin/streptomycin (100 U/100 µg/mL). 3T3-L1 murine preadipocyte cells (ATCC, Manassas, VA, USA) were maintained in DMEM media supplemented with 10% calf serum and penicillin/streptomycin (100 U/100 µg/mL). All cells were maintained at 37 °C in a humidified incubator and 5% CO 2 . T47D and OVCAR5 cells were maintained in steroid-free media, which lacked phenol red and had 5% dextran-coated charcoal-treated FBS, 24 h before starting the experiments involving hormone treatments. Cells were plated in phenol red-free, charcoal–dextran-stripped FBS media in 24-well plates (Corning, Corning, NY, USA) and left to recover overnight. The next day, cells were transfected with a mixture of HRE-luciferase plasmid and RSV-β-galactosidase at a ratio of 2:1 using TransIT LT1 transfection reagent (Mirus Bio, Madison, WI, USA) overnight according to the manufacturer’s protocol. Cells were then treated for 24 h with the hormones as a positive control, DMSO as the vehicle control, and the botanicals at various concentrations. Post-treatment, the cells were lysed (2% Triton-X, 0.1% DTT in GME buffer) and frozen for an hour, and then the lysate was divided into two plates for the luciferase and β-galactosidase assays as previously described [ 15 ]. Luciferase readings were normalized to the β-galactosidase activity then to the vehicle control and reported as mean fold change ± SEM of at least three biological replicates. OVCAR5 cells were seeded at 200,000 cells/well in a 6-well plate in phenol red-free and stripped media for 24 h (~70% confluency). The cells were treated with paeoniflorin and isoorientin (20 µM) for 30 min, followed by the addition of Dex (30 nM) for 6 h. RNA was extracted using TRIzol (Life Technologies, Grand Island, NY, USA) and chloroform with isopropanol precipitation, followed by ethanol washes. The RNA was reverse transcribed to cDNA using an iScript™ cDNA synthesis kit (1708891, Bio-Rad, Hercules, CA, USA) following the manufacturer’s protocol. Quantitative PCR was performed using PowerUp™ SYBR™ Green Master Mix for qPCR (1725271, Bio-Rad, Hercules, CA, USA) according to the manufacturer’s protocol on the CFX connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). Transcript expression was calculated using the ΔΔCt method and samples were normalized to the reference gene glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ). Treatment readings were normalized to the vehicle control (DMSO), set to 1, and results were reported as mean fold change ± SEM from at least 3 biological replicates. GILZ and MKP1/DUSP1 primers ( Supplementary Materials Table S1 ) were designed using the NCBI Primer-Blast tool ( https://www.ncbi.nlm.nih.gov/tools/primer-blast/ accessed on 16 September 2024, version 2.5.0) and ordered from Integrated DNA technologies (IDT, San Diego, CA, USA). Cells were seeded and allowed to adhere overnight in phenol red-free media supplemented with charcoal-stripped FBS. The cells were treated for 24 h then lysed with a lysis buffer containing RIPA, a protease inhibitor and a phosphatase inhibitor. Protein concentration was assessed via the Bradford protein assay, and the samples (30 µg of protein) were run on SDS-PAGE gels by electrophoresis, then transferred to a nitrocellulose membrane. The membrane was blocked in non-fat milk followed by primary antibody incubation at 4 °C ( Supplementary Materials Table S2 ). After 24 h, the membranes were washed and incubated with the appropriate secondary antibody and developed using SuperSignal ® West Femto Maximum sensitivity substrate (PI34096, ThermoFisher, Waltham, MA, USA) on the Azure 300 Chemiluminescent Western Blot Imager (Azure Biosystems Inc., Dublin, CA, USA) [ 15 ]. The images were quantified using ImageJ (NIH) and plotted using GraphPad Prism (Version 10.6.0 for Windows, GraphPad Software, Boston, MA, USA) after normalization. Data is presented as an average of 3 biological replicates ± SEM, and statistical analysis was done using one-way ANOVA followed by Tukey’s post hoc test, where the p -value was set to be 0.05. To initiate cell differentiation, the media was switched to DMEM supplemented with 10% FBS and penicillin/streptomycin (100 U/100 µg/mL). 3T3-L1 cells were seeded and allowed to reach 100% confluence, which marked day zero (D0). At D0 the cells were incubated with the differentiation media (DM) containing 1 µM Dex, 1 µg/mL insulin, and 0.5 mM IBMX, as previously described [ 31 ], for 3 days. Following that, the cells were incubated in maintenance media (1 µg/mL insulin) for 2 days. Afterwards, the media was changed to DMEM supplemented with FBS and pen/strep until the experiment was stopped at D10. Treatments were added to the cells starting from D0. 3T3-L1 cells were differentiated according to the protocol mentioned above for 10 days then they were washed and fixed with 4%. The cells were then washed and dehydrated with 60% isopropanol and stained with 60% ORO for 20 min. The dye was initially dissolved in 100% isopropanol then diluted to 60% in distilled water. The wells were then washed until the stain was cleared and allowed to air-dry. For quantification, the stain was redissolved in 300 µL isopropanol at RT; then, absorbance was measured at 500 nm. The data shown represents the mean ± SEM of at least three biological replicates. The images of the wells are taken by a Nikon Eclipse TE200 (Tokyo, Japan) via an AmScope microscope digital camera (Irvine, CA, USA). Affinity of paeoniflorin and isoorientin for the human GR was evaluated in a GR (h) agonist radioligand receptor binding assay performed by Eurofins Cerep (Celle l’Evescault, France; Study No. 100078167, Study ID US034-0028622) according to their protocol as described by Clark et al. [ 32 ]. Human GR was assessed in cytosolic preparations from IM-9 human B lymphoblast cells endogenously expressing GR, incubated with [ 3 H] Dex (at a final concentration of 1.5 nM, Kd 1.5 nM) for 24 h at 4 °C in the presence of the test compounds or vehicle (DMSO), with triamcinolone (10 µM) used to define non-specific binding. The bound radioligand was quantified by scintillation counting. Specific binding in the presence of paeoniflorin and isoorientin was quantified relative to control samples and used to generate a concentration–response curve for the IC 50 and K i determination. The concentration–response data for paeoniflorin and isoorientin were analyzed by non-linear regression using Cerep’s Hill software (Eurofins Discovery, Celle L’Evescault, France), validated by comparison with SigmaPlot ® 4.0 for Windows ® (SPSS Inc., Chicago, IL, USA), to determine IC 50 values. Inhibition constants (K i ) were calculated from IC 50 values using the Cheng–Prusoff equation, with ligand concentration and K d as specified above. The results were plotted using GraphPad Prism and the points represent the average of two biological replicates. The structure of isoorientin was optimized using Gaussian09 [ 33 ] with the B3LYP functional and a 6-31G(d) basis set. The structure of the antagonist form of the GR was taken from the PDB (access code 3H52) [ 34 ]. Missing residues were added to the structure of chain A using the ModLoop server [ 35 ]. The protonation state of charged residues was determined using the H++ webserver (version 4.0) at pH 7 and 0.15 μmol salinity [ 36 ]. Each small molecule was docked in the GR using the Autodock suite (version 4.2.6) [ 37 ] and the Lamarckian genetic algorithm. The docking parameters were genetic algorithm run of 30, population size of 150, and 25 million energy evaluations. Classical molecular dynamics (MD) simulations were carried out using the pmemd module of the GPU-accelerated Amber22 package [ 38 ]. The Amber ff14SB force field [ 39 ] was used for standard residues; TIP3P was used for solvent water molecules and ions. We used the Merz–Singh–Kollman scheme for RESP charge fitting of electrostatic potential generated at the B3LYP/6-31G(d) level of theory for each ligand, utilizing the generalized Amber force field (gaff) [ 40 ] to generate forcefield parameters for the substrate. The ligand–receptor complex was solvated in a square water box with a periodic boundary condition with the minimum distance between the peptide assembly and the edge of the box as 12 Å. The classical MD simulations follow four stages; minimization, heating, equilibration, and production runs. We applied 30,000 steps of energy minimization with 5.0 kcal mol −1 Å −2 restraints on the receptor and ligand. Then, the system was heated to 300 K over 10,000 steps of MD at a 1fs timstep with the same 5.0 kcal mol −1 Å −2 restraints. The system was then equilibrated over 300,000 steps of MD at a 1fs timestep, gradually releasing the harmonic restraints every 50,000 steps. Production simulations were performed at constant pressure with the Beresden barostat and 300 K (Lagevin thermostat) with a 2 fs timestep with 0.1 kcal mol −1 Å −2 restraints on the backbone atoms of the GR only. Production runs were performed for 100 ns in triplicate, employing the SHAKE algorithm for H atoms, the Particle-Mesh Ewald method [ 41 ] for long-range electrostatic effects, and an 8.0 Å cutoff for electrostatic interactions. Frames were written to the file every 5000 steps (10,000 frames per simulation). CPPTRAJ (version 6.18.1) [ 42 ] was utilized for RMSD calculations as well as interaction distances and frequencies. Clustering analysis based on the root-mean-square deviation (RMSD) of the receptor backbone was carried out using the CPPTRAJ module to identify the most populated ligand conformation in the MD simulations of all three replicas. H-bonds are specified as having a donor–acceptor atom distance of 140°. H-bond frequency is reported as the number of frames specifying H-bond criteria across the three triplicate runs of MD simulation divided by the total number of simulation frames. 3D renderings were created using PyMOL (version 3.0.0) [ 43 ]. Binding energy calculations were computed using the MPI implementation of the mmgbsa.py [ 44 ] module (version 14.0) in AmberTools22 with igb = 8 and surface tension set to 0.0072. Energetic sampling for MMGBSA was for 0.2 ns in simulation within a representative 30 ns segment. Normal mode entropy calculations were performed on 15 frames taken every 2 ns within the 30 ns simulation windows. This was done for all reported energy values except for isoorientin pose 1 (state 1 and 2), where 80 ns of simulation was used with nmode sampling every 8 ns (30 frames for nmode).

Several limitations should be acknowledged. First, all experiments were conducted in a limited number of cell lines (OVCAR5, Ishikawa PR-B, T47D, and T47D A1-2) and in murine 3T3-L1 adipocytes, which may not necessarily fully reflect GR/PR crosstalk or adipogenic responses in normal human tissues. Second, mechanistic insights are largely based on a canonical HRE-Luc reporter, a small number of GR target genes ( GILZ , DUSP1/MKP1 ) and FABP4, and thus do not capture broader transcriptomic or non-classical GR/PR signaling effects. Third, the in vitro concentrations of PFL and ISO used may exceed nutritionally achievable exposures from white peony and chasteberry preparations, and possible off-target effects on other nuclear receptors not mentioned here or kinase pathways were not assessed. Finally, the distinction between direct GR binding by ISO and putative indirect or allosteric mechanisms of paeoniflorin remains inferential.

Therapies such as herbal remedies, botanical supplements, and products derived from natural sources often have significant biological implications for health. In a study measuring the prevalence and use of complementary and alternative medicinal approaches, the highest prevalence of roughly 65% was in the Obstetrics and Gynecology or “women’s health” field [ 1 ]. Women experience multiple symptoms associated with puberty, reproductive cycling, and in the postmenopausal setting that can be modified by botanicals, including premenstrual syndrome (PMS), endometriosis, polycystic ovarian syndrome (PCOS), dysmenorrhea, and infertility [ 2 ]. Botanical dietary supplements are often used due to their ease of administration, availability, good clinical efficacy, and affordability. Progesterone is a steroid hormone secreted mainly from the gonads under the control of the pituitary–gonadal axis. It binds nuclear progesterone receptors (PR-A and PR-B) leading to receptor dimerization, nuclear internalization and interaction with progesterone response elements (PREs) to either activate or repress gene expression as part of the canonical pathway of this transcription factor [ 3 ]. Progesterone can also activate faster, non-genomic pathways to exert its activity through binding PR-C or membrane-bound receptors [ 3 ]. During the menstrual cycle, progesterone exerts pregestational effects by thickening the endometrial lining to support successful implantation. Progesterone deficiency can lead to increased myometrial contractility, higher miscarriage risk, preterm labor, decreased fertility [ 4 ] and endometriosis [ 5 ]. Moreover, the absence of progesterone during the menstrual cycle leads to unrestrained estrogen effects on the uterus, associated with a higher risk of endometrial hyperplasia and potentially endometrial cancer [ 6 ]. Glucocorticoids (GCs), primarily cortisol in humans, are steroid hormones produced by the adrenal glands that play a crucial role in regulating various physiological processes and metabolic adaptations, including stress response [ 7 ]. Glucocorticoids facilitate the early phases of preadipocyte differentiation by enhancing the adipogenic transcriptional cascade of CCAAT/enhancer-binding protein β and peroxisome proliferator-activated receptor gamma (C/EBPβ-PPARγ) induction, which subsequently leads to the induction of downstream adipocyte genes. Among these, fatty acid-binding protein 4 (FABP4/ap2) becomes highly expressed especially during terminal differentiation and in mature adipocytes, where it functions in fatty acid uptake, traffic into lipid droplets, and governing lipid accumulation and metabolic activity. Furthermore, elevated FABP4 levels can enter a feedback loop for PPARγ activity to limit additional differentiation to help stabilize the mature adipocyte pool under conditions of sustained lipid loading [ 8 , 9 ]. Increasing levels of circulating GCs have been implicated with female gynecological conditions such as PCOS [ 10 ] and stress-induced infertility [ 7 ] as well as metabolic syndrome conditions like dyslipidemia [ 11 ]. Managing the levels of GCs in the body either by decreasing the level of circulating GCs or modulating the GR can prove beneficial in these conditions. Hormone receptors exhibit complex crosstalk through multiple mechanisms at both the protein–protein interaction and chromatin-binding levels [ 12 ]. Progesterone, androgen, glucocorticoid and mineralocorticoid receptors share remarkable structural similarity, with approximately 55% sequence identity in their ligand-binding domain [ 13 ]. Moreover, these four receptors recognize nearly identical consensus DNA sequences; they all bind the same hormone response elements (HREs). Despite their architectural similarity, these receptors exhibit distinct binding patterns leading to distinct downstream physiological effects. PR and GR can bind an overlapping set of chromatin binding regions and recruit shared chaperones and cofactors, implying a cooperative or competitive behavior when multiple receptors are activated in the same cell [ 14 ]. For example, in breast cancer, Ogara et al. have found that GR antagonizes PR activity through ligand-dependent mechanisms: R5020-bound GR redistributes to a limited set of shared binding sites and forms GR-PR complexes at select enhancers, whereas Dex-bound GR occupies numerous HREs shared with PR, increasing GR-PR interaction and collectively suppressing PR-mediated transcription and proliferation [ 12 ]. Similarly, in the uterus, GR and PR exhibit opposing interaction; Austin et al. reported that baicalein from the roots of skullcap ( Scutellaria baicalensis) functions as a PR antagonist with simultaneous GR agonist activity [ 15 ]. Irilone from red clover ( Trifolium pratense L.) was shown to have a tissue-dependent PR-potentiating effect, acting through the estrogen receptor (ER) in breast cells and through GR in endometrial cells [ 2 ]. This multi-target activity underlies the therapeutic potential of botanical use in reproductive health applications. While many botanicals exhibit polypharmacological effects on hormone receptors, they also often exhibit tissue-dependent and concentration-dependent effects. White peony ( Paeonia lactiflora Pall.), also known as Bai Shao in Traditional Chinese Medicine (TCM), has been used for over two millennia in gynecological practice, primarily for its effects on hormone-responsive health conditions. Several formulations containing white peony roots are used to address menstrual irregularities, dysmenorrhea, and endometriosis [ 16 , 17 , 18 ]. Paeoniflorin (PFL), a monoterpene glucoside, is the most abundant biologically active metabolite in white peony and a marker for its identification, and has been linked to the amelioration of PCOS symptoms in an in vivo rat model where PCOS was Dehydroepiandrosterone-induced [ 19 ]. PFL has also been shown to reduce Dex-induced testosterone levels in primary murine theca cells by affecting cytochrome P450-17A1 expression levels [ 20 ]. Moreover, PFL was found to enhance the hypothalamic–pituitary–adrenal negative feedback mechanism, increasing the levels of GR in the amygdala while decreasing it in the hippocampus, indicating tissue-specific activity of PFL in the body [ 21 ]. Chasteberry extract ( Vitex agnus-castus L.) has also been used for managing premenstrual syndrome, PCOS, and endometriosis [ 22 , 23 ]. Chasteberry extracts have been shown to inhibit prolactin release in vitro, a mechanism attributed to their dopamine agonist effect on the dopamine D 2 receptor [ 24 ]. Moreover, chasteberry extracts have been shown to indirectly stimulate the secretion of luteinizing hormone (LH), leading to an increase in progesterone and 17β-estradiol [ 25 ]. Chasteberry extracts have also been reported to contain estrogen-mimicking compounds [ 26 , 27 ]. Fukahori et al. established an HPLC fingerprint for chasteberry extracts, which showed 26 characteristic peaks, and isoorientin, a C-glucosyl flavone, was identified as a pharmacopeial marker [ 28 ]. The parent aglycone of isoorientin, luteolin, exhibits potent estrogenic activity, moderate anti-progestogenic activity, and weak anti-glucocorticoid activity [ 29 ]. This study characterizes the effects of paeoniflorin and isoorientin on GR and PR signaling in vitro using an HRE-luciferase (HRE-Luc) assay across multiple cell lines and downstream RT-qPCR validation, as well as a functional metabolic assay in 3T3-L1 cell line to determine whether these botanical compounds modulate lipid accumulation in differentiated adipocytes. Overall, we identified isoorientin as a GR antagonist in multiple cell lines based on both receptor binding and computational modeling.

Both paeoniflorin and isoorientin serve as chemical markers to identify extracts from Paeoniaceae and chasteberry, respectively. Prior to functional testing, lysates from OVCAR5, T47D, T47D A1-2, and Ishikawa PR-B cells were analyzed by Western blot to confirm receptor expression. OVCAR5, Ishikawa PR-B and T47D A1-2 all express GR [ 15 ] ( Figure S1 ), whereas T47D cells express both PR isoforms (PR-A and PR-B), ER and no GR [ 12 ]. Ishikawa PR-B stably expresses PR-B under CMV control developed by the Blok Lab [ 30 ]. The HRE-Luc reporter assay was then used to define the impact of paeoniflorin and isoorientin (structure shown in Figure 1 A and 1B respectively) on GR-mediated signaling. Dex (30 nM) was used as a positive control to induce GR activity in OVCAR5 and it induced the HRE-Luc assay 22.9-fold ( p < 0.0001, Figure 1 C). Paeoniflorin alone did not activate the reporter, indicating the absence of GR agonist activity, whereas co-treatment with paeoniflorin (20 µM) and Dex reduced Dex-induced luciferase activity by 26.5% ( p < 0.005, Figure 1 C). Similarly, isoorientin alone did not activate the reporter, but co-treatment with isoorientin (20 µM) and Dex decreased the GR-driven HRE-Luc activity by 52.5% ( p < 0.0001, Figure 1 D), demonstrating that both botanicals attenuate Dex-induced GR signaling in OVCAR5 cells. To examine GR signaling in a cell line that also expresses PR, Ishikawa PR-B cells were transfected with the same HRE-Luc construct. Paeoniflorin again failed to activate the reporter, consistent with a lack of agonist activity of GR or PR in this assay. Treatment with 30 nM Dex produced a 6.9-fold induction of luciferase activity ( p < 0.0001, Figure 1 E), and co-treatment with 20 µM paeoniflorin reduced the response by 33.5% ( p = 0.0108, Figure 1 E). Isoorientin at 20 µM did not activate the reporter on its own in Ishikawa PR-B, indicating it does not activate either GR or PR, but a co-treatment with 20 µM isoorientin plus 30 nM Dex inhibited the Dex-induced signal by 58% ( p < 0.0001, Figure 1 F). To determine whether these inhibitory effects were dose-dependent, additional concentrations of paeoniflorin and isoorientin were tested in combination with Dex. Lower and higher concentrations of paeoniflorin did not block Dex activity significantly, whereas 10 µM isoorientin already produced measurable inhibition of HRE-Luc activity in OVCAR5 cells ( Figure S2 ). Ligand-bound GR can interact functionally with PR, and this crosstalk is dependent on tissue-specific receptor expression patterns, stromal context, and differential regulator recruitment and activity. To evaluate the effects of paeoniflorin and isoorientin on PR signaling in a GR-positive background, two PR/GR co-expressing models were used: the endometrial cell line Ishikawa PR-B and the breast cancer cell line T47D A1-2. Both cell lines were transfected with the HRE-Luc reporter and treated with progesterone (P 4 ) as a positive control for PR-dependent transcriptional activity. In Ishikawa PR-B cells, treatment with 50 nM P 4 induced an average 21-fold increase in luciferase activity ( p < 0.0001, Figure 2 A), and co-treatment with 20 µM of paeoniflorin with P 4 did not alter this response ( Figure 2 A). Likewise, co-treatment with 20 µM of isoorientin and P 4 had no inhibitory effects on the P 4 -induced luciferase signal ( Figure 2 B), indicating that neither compound measurably inhibits P 4 -driven activity through PR in Ishikawa PR-B. In contrast, a different pattern emerged in T47D A1-2 cells that have been engineered to express GR. The ligand P 4 alone induced the reporter 30.8-fold ( p < 0.0001, Figure 2 C). Paeoniflorin (20 µM) alone did not affect basal luciferase levels, but a combination of 20 µM paeoniflorin and 50 nM P 4 led to an inhibition of the luciferase signal by 37.3% ( p < 0.0017, Figure 2 C). Isoorientin alone at 20 µM also failed to activate the reporter by itself, yet co-treatment with 20 µM isoorientin and 50 nM P 4 inhibited the P 4 response by 47.7% ( p < 0.0001, Figure 2 D). These data suggest that GR-bound botanicals could block PR signaling. The cell line T47D A1-2 also expresses ER, raising the possibility that the inhibitory effects on P 4 -driven luciferase activity might involve ER rather than GR-PR crosstalk. To address this, the HRE-Luc assay was performed in the parental T47D line. Treatment with 10 nM of P 4 used as positive control induced a 6.3-fold increase in luciferase activity ( p < 0.0001, Figure 3 A). Paeoniflorin alone did not induce luciferase activity at 20 µM and co-treatment with 10 nM P 4 did not inhibit the P 4 response ( Figure 3 A). Similarly, 10 µM isoorientin neither activated the reporter nor reduced P 4 -induced activity when combined ( Figure 3 B), indicating that the inhibitory effects observed in T47D A1-2 require GR and are not mediated by ER in the parental T47D cell line. To further evaluate direct ER modulation, T47D cells were transfected with an estrogen response element-driven luciferase construct (ERE-Luc). Cells were treated with the vehicle, 10 nM estradiol (E 2 ) as a positive control, or E 2 in combination with the botanicals. Estradiol induced a 7.5-fold increase in luciferase activity ( p < 0.0001, Figure 3 C). Paeoniflorin (20 µM) alone did not alter ER activity; when in combination with E 2 , it also did not alter E 2 -induced activity ( Figure 3 C), and isoorientin (10 µM) likewise failed to activate ER or inhibit E 2 -driven luciferase activity ( Figure 3 D). GR is a ligand-activated transcription factor that controls a subset of genes. Glucocorticoid-induced leucine zipper ( GILZ ) and dual-specificity phosphatase 1 gene encoding for mitogen-activated protein kinase phosphatase 1 ( DUSP1/MKP1 ) are well-characterized genes regulated by GR and induced by Dex [ 45 , 46 ]. To evaluate whether paeoniflorin or isoorientin block GR transcriptional output, RT-qPCR was performed in OVCAR5 cells. Treatment with 30 nM Dex for 6 h increased GILZ mRNA by an average of 124-fold and DUSP1/MKP1 by 8.7-fold relative to vehicle control (0.1% DMSO) ( Figure 4 A and 4B, respectively). A co-treatment of Dex with 1 µM RU486 led to a decrease of 96% in GILZ and 63% in DUSP1/MKP1 mRNA abundance ( Figure S3 ). Co-treatment with 20 µM paeoniflorin and Dex attenuated these responses. GILZ expression was reduced by 37%, corresponding to a mean 0.67-fold change when normalized to Dex-treated cells ( Figure 4 C), while DUSP1/MKP1 levels decreased by 26.5%, corresponding to a 0.74-fold change ( Figure 4 D). Similarly, co-treatment with 20 µM isoorientin and Dex decreased GILZ expression by 29% (0.71-fold relative to Dex, Figure 4 E) and DUSP1/MKP1 by 15% (0.85-fold relative to Dex, Figure 4 F). These data indicate that both paeoniflorin and isoorientin partially antagonize Dex-induced GR target gene activation in OVCAR5 cells. Glucocorticoid inhibition in adipocytes reduces lipid accumulation during differentiation. To determine whether paeoniflorin or isoorientin can block GR activity in differentiating adipocytes and thereby limit lipid deposition, 3T3-L1 cells were induced to differentiate using the experimental design shown in Figure 5 A. Intracellular lipid accumulation was quantified by Oil Red O (ORO) staining followed by dye elution. Compared to differentiated control cells, both compounds produced a clear, concentration-dependent reduction in ORO signal, indicating decreased neutral lipid deposition within the lipid droplets. The treatment with paeoniflorin showed a decrease in lipid content to approximately 92%, 81% and 74%, respectively ( p < 0.05 for 50 µM and <0.01 for 100 µM, Figure 5 B). Similarly, isoorientin lowered lipid accumulation levels to 82%, 73% and 64%, respectively ( p < 0.01 for 50 µM and p < 0.001 for 100 µM, Figure 5 C). Notably, a treatment with 10 µM RU486 mimicked this reduction in Oil-Red O staining to 30% compared to differentiated control ( p < 0.001, Figure 5 D), supporting the notion that inhibition of GR signaling contributes to the anti-adipogenic effects of paeoniflorin and isoorientin in 3T3-L1 adipocytes. Moreover, differentiating 3T3-L1 adipocytes treated with paeoniflorin or isoorientin appeared smaller and contained fewer visible and measurable lipid droplets than the control differentiated cells ( Figure S4 ). Previous studies have shown that neither PFL nor ISO is toxic to 3T3-L1 cells [ 47 , 48 ]. FABP4 expression steadily increases during adipogenesis, helping to support lipid uptake and fat droplet growth, and blocking GR disrupts this sequence by disrupting the expression of FABP4 in the cells. To determine whether this phenotype reflected reduced FABP4-associated lipid accumulation, FABP4 protein abundance was assessed by Western blot in cells differentiated in the presence of 10 µM and 100 µM of paeoniflorin, 10 µM and 100 µM of isoorientin ( Figure 6 A), or 10 µM of RU486 ( Figure 6 B). Isoorientin significantly decreased FABP4 protein in a dose-dependent manner by approximately 28% at 10 µM and 50% at 100 µM ( p < 0.05, Figure 6 C), whereas paeoniflorin did not alter FABP4 levels significantly at these concentrations ( Figure 6 D). RU486 blocked GR, leading to reduced FABP4 expression by 46% ( p = 0.006, Figure 6 E), and blocked adipocyte differentiation. To test whether the compounds’ activity occurred through direct GR binding, a competitive binding assay was conducted to determine their IC 50 . Isoorientin displayed measurable binding to (h) GR with an IC 50 of 6.3 × 10 −5 M and a K i of 3.2 × 10 −5 M in the [ 3 H] Dex binding assay ( Figure 7 A). At the highest concentration (1.0 × 10 −4 M), isoorientin inhibited 67.8% of the control’s specific binding ( Figure 7 B), exceeding the 50% threshold used to define significant effects. In contrast, paeoniflorin did not achieve ≥25% inhibition of the control’s specific binding at the highest validated concentration (1.0 × 10 −4 M) ( Figure 7 B), and its IC 50 value was therefore not calculable under these assay conditions. In order to understand the mode of binding and inhibition of isoorientin to GR, we carried out a series of computational docking and molecular dynamics studies. The two most populated clusters from Autodock of isoorientin and GR-LBD were used as the starting poses for classical MD simulation. In pose 1 ( Figure 8 A), the dihydroxybenzyl group is positioned towards the exterior of the binding pocket, while in pose 2 ( Figure 8 B) the position is flipped, placing the dihydroxybenzyl group towards the interior of the binding pocket. In pose 1 simulations, two different ligand conformational states are populated, one with an average ligand RMSD of 0.68 Å ( P1 ) and another with an average RMSD 1.7 Å ( P1-rot ) ( Figure 8 C). The two different states observed during MD simulation are generated by a 180° rotation about the C–C bond connecting the fused lactone to the dihydroxybenzyl group of isoorientin. The MMGBSA calculated binding energies were −14.6 kcal/mol and −12.5 kcal/mol for P1 and P1-rot , respectively ( Figure 8 E, Supplementary Material Table S3 ). The most prominent hydrogen bonding interaction for both states was between a backbone oxygen of Leu563 and C2 hydroxyl of the glycosyl group, occurring in 53% of frames with an average distance of 2.76 Å. Additionally, isoorientin displays hydrogen bonding between the glycosyl group and Gln642, with four different hydroxyl groups registering as a hydrogen bond in 9% of frames or more in pose 1 ( Figure 8 F). The most frequent hydrogen bond was with the C4 hydroxyl, occurring in 28% of frames with an average distance of 2.73 Å. Additionally, the 4-hydroxyl of the dihydroxybenzyl group formed hydrogen bonds with Glu755 in 12% of frames with an average distance of 2.65 Å. However, due to Glu755 being in a more solvent-exposed area of the binding pocket, contacts between the dihydroxyl benzyl group and the solvent were considerably more frequent (74% of frames for the 4-hydroxyl group and 56% of frames for the 3-hydroxyl group). Free energy calculations indicate that isoorientin in pose 1 also experiences additional stabilization, primarily from van der Waals interactions with hydrophobic residues Val571 and Ile756, contributing −1.7 kcal/mol and −2.2 kcal/mol respectively ( Supplementary Materials Table S4 ). Similarly, the simulations of pose 2 yielded the two different ligand rotamer states generated from the same rotation of the dihydroxybenzyl group as pose 1. The binding energy of pose 2 rotamer state 1 ( P2, RMSD of 1.6 Å) was calculated to be −16.1 kcal/mol, and the binding energy of pose 2 rotamer state 2 ( P2-rot , RMSD of 0.8 Å, Figure 8 D) was −12.4 kcal/mol ( Figure 8 E). Pose 2 generally exhibited more persistent hydrogen bonding during MD simulation with GR-LDB than pose 1 ( Figure 8 G). The most frequent hydrogen bond was between the backbone oxygen of Met752 and the hydroxyl group of the fused lactone of the ligand, occurring in 80% of frames with an average distance of 2.8 Å. Additionally, the hydrogen bonding between Gln642 and ligand was more persistent when the dihydroxybenzyl group was placed deeper into the pocket, interacting with 4-hydroxyl and 3-hydroxyl groups for 58% of frames and 37% of frames, respectively, with both interactions at an average distance of 2.7 Å. The ion-dipole interactions with the side-chain oxygens of Glu755 were also more frequent than in pose 1, with an interaction with the C4 hydroxyl of the glycosyl group observed in 32% of MD frames, at an average distance of 2.6 Å. Pose 2 also demonstrated an interaction with Trp600 with an average oxygen-aryl distance of 3.8 Å across all MD frames, contributing −2.3 kcal/mol stabilization to the overall binding energy ( Supplementary Materials Table S5 ). This contact is characterized by an O–H/π interaction with an average distance of 3.8 Å between the hydroxyl oxygen and the center of mass of the tryptophan six-membered ring. Additionally, hydrophobic interactions with Val571 and Ile756 contributed −1.9 and 1.8 kcal/mol per residue stabilization to the overall binding energy. The relative energetic similarity between pose 1 and pose 2 indicates that the probability of isoorientin binding to the pocket is similar in both poses. Pose 2 demonstrated a greater frequency of its hydrogen bonds and noncovalent interactions ( Figure 8 B). The ligand binding mode of isoorientin does bare some similarities to the binding modes of known antagonists of the GR, mifepristone and dexamethasone. Hydrophobic interactions Leu563, Trp600, and Ile756 as well as hydrogen bonding activity of Asn564 and Gln642 in the binding mode of isoorientin are shared with the two known antagonists [ 34 , 49 ]. However, both known ligands feature hydrogen bonding with Arg611 and Gln570 found deeper in the GR binding pocket. Interactions with these deeper pocket residues were not observed with isoorientin in any of the binding poses considered, indicating the binding of isoorientin is less buried in the pocket than dexamethasone and mifepristone. Additionally, comparing the hydrogen bond occupancy of Gln642 with isoorientin to previous molecular dynamics studies on dexamethasone done using the same crystal structure of the GR [ 50 ], it was found that the occupancy of this hydrogen bond with isoorientin was much lower than that of dexamethasone (28% or 58% depending on pose vs. 76% with dexamethasone). Lastly, for isoorientin the binding energy difference resulting from flipping the position of the glycosyl and dihydroxybenzyl in the pocket resulted in a relatively small difference in the binding energy of 1.5 kcal/mol, indicating less specificity in binding. Taken together, the lack of interaction with deep lying residues, lower hydrogen bond occupancy, and small energy differences between significantly different orientations of isoorientin in the binding pocket all support the experimental findings of weaker bonding of isoorientin relative to dexamethasone ( Figure 8 H).

Women frequently use botanical preparations including white peony and chasteberry, either alone or in combination with other botanicals, in formulations for the management of various gynecological conditions like premenstrual symptoms, perimenopausal complaints and PCOS. Clinical as well as observational data support their beneficial effects in several subsets of patients [ 16 , 17 , 19 , 22 , 23 ]. However, despite their long-standing inclusion in traditional prescriptions and their increasing modern use, the molecular mechanisms by which these botanicals influence steroid hormone signaling pathways remain incompletely elucidated, particularly when crosstalk between receptors is present. This study demonstrates that paeoniflorin and isoorientin function as antagonists of GR signaling in specific cell models. Both compounds inhibited Dex-induced GR-driven reporter activity without demonstrating agonist activity on a canonical reporter construct, though they differed fundamentally in their binding mechanisms and selective effects on PR signaling. The literature on steroid receptor signaling in cancer cell lines documents substantial cell-line-specific variations in coactivator expression and preference [ 49 ]. For instance, PR and GR preferentially recruit distinct steroid receptor coactivators (SRCs); PR favors SRC-1, while GR preferentially recruits SRC-2/TIF2 [ 51 ]. This selectivity determines downstream histone acetylation patterns and chromatin remodeling capacity. Moreover, recent genomic analyses of endometrial versus breast cancer cells reveal profound differences in coactivator cofactor expression patterns and in the genomic distribution of hormone response elements. At the cofactor level, SRC-1 and p300/CBP cofactors show reduced expression in endometrial carcinoma [ 52 ]. Moreover T47D A1-2 as a breast cancer cell line expresses PR-A and PR-B; these two isoforms occupy fundamentally different genomic locations despite binding identical DNA motifs, controlling which genomic regions are accessed [ 53 ]. The failure of paeoniflorin and isoorientin to inhibit PR signaling in Ishikawa PR-B despite PR antagonism in T47D A1-2 cells expressing GR may be due to the coactivator and cofactor machinery in these two cellular contexts, likely due to interaction with GR. This is noteworthy given that numerous phytochemicals exhibit promiscuous receptor binding profiles. Co-treatment with either paeoniflorin or isoorientin partially reduced the abundance of Dex-induced transcripts GILZ and DUSP1/MKP1 , confirming that the functional antagonism observed in the reporter assays extends to native GR-responsive genes. When considered alongside the radioligand binding data—showing direct GR binding by isoorientin and not paeoniflorin—these results support a mechanistic distinction in which isoorientin primarily acts as a competitive ligand-binding antagonist, whereas paeoniflorin may operate through indirect or allosteric mechanisms that impair GR transcriptional competence without preventing ligand binding or receptor turnover. In the receptor binding assay, isoorientin exhibited measurable competitive binding to GR with an IC 50 of 6.3 × 10 −5 M, whereas paeoniflorin failed to achieve significant binding at concentrations up to 1.0 × 10 −4 M. This finding is consistent with emerging literature on selective GR modulators that demonstrate ligand affinity is not sufficient to predict the magnitude or pattern of GR antagonism [ 54 , 55 ]. The field of selective GR modulators (SGRMs) has increasingly recognized that compounds can achieve antagonistic effects through mechanisms including coactivator interference, altered helix-12 positioning, or modulation of cofactor recruitment rather than simple competitive antagonism [ 56 , 57 , 58 ]. The structural analysis of isoorientin docking revealed that the ligand’s glycosyl group and dihydroxybenzyl moiety engage multiple hydrogen bonding networks with GR-LBD residues, including Leu563, Gln642, and Glu755, with energetically favorable binding poses exhibiting favorable van der Waals interactions with Val571 and Ile756. Critically, the extended positions of isoorientin outside the binding pocket suggest a mechanism where the ligand sterically prevents coactivator binding—a phenomenon analogous to how RU486 (mifepristone) disrupts the receptor conformation required for p300/CBP recruitment [ 59 ]. This steric hindrance model aligns with mechanistic studies showing that bulky antagonists can displace helix-12 and occlude the coactivator-binding interface [ 59 , 60 ]. Nevertheless, MD simulations indicate that isoorientin adopts multiple energetically similar but relatively shallow binding orientations, with lower hydrogen bond persistence and fewer interactions with deep GR residues, collectively supporting a weaker binding mode than classical GR antagonists like RU486. For paeoniflorin, which demonstrated GR antagonism without direct binding, an allosteric or indirect mechanism could be more plausible. Prior pharmacological work on paeoniflorin reported competitive binding to GR with an inhibitor constant around 3.0 × 10 −9 M in cytosolic assays [ 61 ], implying that conditions present in live cells (including cofactor availability, lipid membrane, and post-translational modifications) may enhance its functional potency beyond that which the purified binding assay detected. This is consistent with literature findings on paeoniflorin engaging non-canonical GR pathways through MAPK modulation and GR isoform regulation, which also impairs Dex-induced GILZ and DUSP/MKP1 expression despite the absence of direct receptor binding [ 62 ]. Nonetheless, these observations remain descriptive, and other mechanistic pathways—not yet fully characterized—may also contribute to paeoniflorin’s GR-modulatory profile. Functional consequences of GR antagonism manifest in adipogenic models and link receptor modulation to metabolic outcomes. Metabolic disturbances and hormonal dysregulation are tightly interconnected, with adipose tissue now being recognized as an active endocrine organ that both responds to and shapes systemic sex steroid and glucocorticoid signaling [ 63 ]. Dysfunctional adiposity can alter local steroid metabolism, thereby modifying tissue levels of estrogen and progesterone, contributing to conditions such as anovulation, abnormal endometrial function and subfertility. Reciprocally, disturbances in ovarian steroid production or signaling, including altered crosstalk with glucocorticoid pathways, influence lipid storage and fat distribution, creating a bidirectional link between metabolic imbalance and reproductive dysfunction, and using botanicals with complex, multifaceted effects without fully understanding the underlying mechanisms may further contribute to this imbalance. Isoorientin downregulated FABP4 protein (50% at 100 µM), a key fatty acid-binding protein and PPARγ target essential for lipid storage, in differentiated adipocytes compared with paeoniflorin which had no significant effects at a similar concentration, despite comparable suppression of bulk lipid accumulation. The downregulation of FABP4 by isoorientin raises the possibility that direct GR binding may confer additional regulatory advantages beyond simple antagonism, namely, the ability of engaging alternative coactivator or corepressor complexes that specifically modulate FABP4 transcription. Alternatively, the direct GR binding by isoorientin may position the ligand–GR complex to preferentially suppress FABP4 induction relative to other GR targets, a pattern consistent with the selective GR modulator literature where ligand identity determines the set of genes whose regulation is affected [ 64 ].

In conclusion, these findings collectively demonstrate that isoorientin and paeoniflorin are functional GR modulators with distinct features that shape their downstream signaling outcomes. Both compounds attenuated GR-mediated transcriptional activation and target gene expression, with isoorientin doing so in a dose-dependent manner and acting as a weak competitive ligand-binding antagonist, whereas paeoniflorin did not directly compete with the ligand at the GR binding site and will require further studies to elucidate its mechanism of action. From a mechanistic standpoint, both compounds also showed anti-adipogenic activity in 3T3-L1 preadipocytes by reducing lipid accumulation and the late adipogenic marker FABP4, consistent with modulation of GR-associated pathways, but could also involve additional GR-independent mechanisms. Future work to establish their in vivo pharmacodynamics, receptor-selective modulation and interaction with the GR cofactor network will be essential to clarify their therapeutic potential and contextualize traditional use with modern endocrine therapeutics.