Norlichexanthone purified from plant endophyte prevents postmenopausal osteoporosis by targeting ER α to inhibit RANKL signaling.

Although different types of drugs are available for postmenopausal osteoporosis, the limitations of the current therapies including drug resistances and adverse effects require identification of novel anti-osteoporosis agents. Here, we defined that norlichexanthone (NOR), a natural product, is a ligand of estrogen receptor-alpha (ERα) and revealed its therapeutic potential for postmenopausal osteoporosis. We used mammalian-one hybrid assay to screen for ERα modulators from crude extracts of several plant endophytes. As a result, NOR purified from the extract of endophyte ARL-13 was identified as a selective ERα modulator. NOR directly bound to ERα with an affinity in nanomolar range, revealing that it is a natural ligand of ERα. NOR induced osteoblast formation in MC3T3-E1 precursor cells. Conversely, NOR inhibited receptor activator of nuclear factor-kappa B ligand (RANKL)-induced osteoclast formation in both RAW264.7 macrophages and mouse primary monocytes. Mechanistically, NOR inhibited RANKL-induced association of ERα and TRAF6 to prevent ERα-mediated TRAF6 activation via Lys63-linked ubiquitination. Importantly, NOR exhibited potent anti-osteoporosis efficacy in an ovariectomized mouse model. Comparing to estrogen, NOR was of much less capability in stimulating endometrial hyperplasia and promoting mammalian cancer cell proliferation. Taken together, our study identified NOR as a natural and high affinity ligand of ERα with substantial anti-osteoporosis but less estrogenic activity.

Introduction

Estrogen receptor-alpha (ERα) belonging to nuclear receptor superfamily plays vital and variety roles in pathophysiological processes1, 2, 3, 4. Particularly, the dysfunction of ERα is highly related to breast, ovary and endometrial cancers. The activities of ERα are tightly regulated by its cognate ligands that generally bind to the ligand binding domain (LBD) of ERα. Besides the three major endogenous estrogens, i.e., estrone (E1), estradiol (E2) and estriol (E3), many natural and synthetic products were identified as ERα ligands,. Several ERα ligands including agonists and antagonists are widely used as clinical drugs. For example, tamoxifen and raloxifene are used for treating breast cancers and postmenopausal osteoporosis,.

Postmenopausal osteoporosis results from the loss of estrogen due to the functional failure of ovaries,. Bone remodeling is a dynamic process of bone resorption by osteoclast and bone formation by osteoblast. One major pathological mechanism of osteoporosis is the imbalance of the functions of osteoblast and osteoclast, leading to the abnormal bone remodeling. Estrogen is an essential factor to promote osteoblast but inhibit osteoclast formation, rendering the functional balance of osteoblast and osteoclast14, 15, 16. Although clinical drugs are available for treating postmenopausal osteoporosis, they are often associated with severe side effects17, 18, 19. Osteoclast inhibitors, such as bisphosphonates, may cause excessive suppression of bone turnover, resulting in low-trauma, atypical fractures, esophagitis and atrial fibrillation. Hormone replacement therapy (HRT), such as estrogen treatment, is effective to prevent postmenopausal osteoporosis,. However, estrogen treatment often increases the risk of endometrial, ovarian and mammary carcinoma, resulting from its strong estrogenic activity. Thus, discovering novel ERα ligands with less estrogenic activity but strong anti-osteoporosis effect will improve the benefit of ERα-targeted drugs in postmenopausal osteoporosis treatment.

Osteoclasts are fully differentiated, multi-nucleated cells originating from the hematopoietic monocyte-macrophage linage,. Receptor activator of nuclear factor-kappa B ligand (RANKL) is a key factor for inducing osteoclast formation. RANKL binds to its cognate receptor RANK on the plasma membrane to initiate downstream signaling cascade, including activation of mitogen activated protein kinases (MAPKs), nuclear factor kappa B (NF-κB) and the phosphatidylinositol 3-kinase (PI3K) pathways,. The receptor RANK lacks intrinsic kinase activity and requires the recruitment of TNF receptor associated factors (TRAFs) to transduce signaling. TRAF6-deficient mice showing defective osteoclastogenesis and severe osteopetrosis indicate the essential role of TRAF6 in osteoclast formation and in RANKL signaling,. TRAF6 is an E3 ubiquitin ligase and can catalyze itself Lys63-linked ubiquitination, which is essential for transducing RANKL signal. Estrogen has been implicated in the regulation of RANKL signaling, whereas the underlying mechanism is obscure.

Plant endophytes are an endosymbiotic group of microorganisms that colonize in plants. They are reservoirs of numerous and novel bioactive secondary metabolites with diverse structures,. Many natural active products including clinical drugs are obtained or derived from plant endophytes. In this study, we identified compound norlichexanthone (NOR) purified from a plant endophyte as an active ERα ligand. Furthermore, we unraveled that NOR is of potent therapeutic effect on postmenopausal osteoporosis by inducing osteoblast formation but inhibiting osteoclast formation. Moreover, we disclosed that NOR inhibits RANKL signaling by inhibiting ERα-mediated Lys63-linked ubiquitination of TRAF6, resulting in inhibition of osteoclast formation. Importantly, NOR is of much less estrogenic activity than E2.

Materials and methods

Reagents and antibodies

NOR (98% purity) was isolated and obtained according to the procedure depicted, dissolved in dimethyl sulfoxide (DMSO) and stored at −20 °C. 17β-Estradiol (E2), 9-cis retinoic acid (9-cis-RA), all-trans retinoic acid (ATRA), β-glycerophosphate and l-ascorbic acid were purchased from Sigma (St. Louis, MO, USA), and ICI-182780 was from MedChemExpress (Monmouth Junction, NJ, USA). Endoxifen and 4-[2-phenyl-5,7-bis(trifluoromethyl)pyrazolo[1,5-a]pyrimidin-3-yl]phenol (PHTPP) were from Target Mol (Boston, MA, USA). Recombinant murine RANKL was purchased from Abcam Systems (Cambridge, MA, USA) and macrophage colony-stimulating factor (M-CSF) was from PeproTech (Cranbury, NJ, USA). Transfection reagent linear polyethylenimine (PEI) used to transfect HEK293T cells was purchased from Polysciences (Warrington, PA, USA). Reverse transcription reagents and SYBR Green PCR Master Mix were from Yeasen (Shanghai, China). Specific primary antibodies against TRAF6, ERα, MYC, ubiquitin, ERK, JNK were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Antibodies for phospho-P44/42 MAPK (ERK1/2), phospho-SAPK/JNK (Thr183/Tyr185), phospho-P38 MAPK (Thr180/Tyr182) and P38 MAPK were obtained from Cell Signaling Technology (Danvers, MA, USA). Antibodies for β-actin and Flag were purchased from Sigma, and anti-ubiquitin (linkage-specific K63) was from Abcam. HRP-conjugated secondary antibodies against rabbit IgG and mouse IgG were purchased from Thermo Fisher Scientific (Waltham, MA, USA). The ALP activity assay kit, and cell lysis buffer for Western and IP were from Beyotime Biotechnology (Shanghai, China). The TRAP staining kit was purchased from Leagene (Beijing, China).

Cell source and culture

The MC3T3-E1 and bone marrow monocytes were cultured in minimum essential medium Eagle-alpha modification (α-MEM) supplemented with 10% fetal bovine serum (FBS) in a humidified atmosphere containing 5% CO2 at 37 °C, while RAW264.7, MCF-7, MDA-MB-231 and HEK293T cells were cultured in Dulbecco's modified Eagle medium (DMEM). Bone marrow monocytes were isolated from 6 to 8 weeks old C57BL/6 mice. In brief, the mice were sacrificed by anesthesia and immersed in 70% ethanol for 5 min, and then the femur and tibia were separated. The α-MEM medium was pipetted with a syringe to flush the bone marrow cavity, and the flushed medium was centrifuged at 1400 rpm (Centrifuge 5424 R, Eppendorf, Hamburg, Germany) for 5 min at room temperature to collect cells. Erythrocyte lysate (#555899, BD company, Franklin Lake, NJ, USA) was used to remove erythrocytes, and the cells were cultured in medium with M-CSF (50 ng/mL) overnight. The suspension cells were collected and then resuspended in α-MEM containing M-CSF (50 ng/mL) and RANKL (50 ng/mL) for osteoclast induction.

Mammalian one-hybrid assay

HEK293T cells were co-transfected with pG5-luciferase reporter (Promega, Madison, WI, USA) together with the recombinant pBIND plasmids encoding ligand binding domain (LBD) of nuclear receptors fused with the DNA-binding domain (DBD) of Gal4. One day after transfection, cells were treated with DMSO, NOR or ligands specific for each nuclear receptor. After 12 h, cells were lysed by passive lysis buffer. Firefly and Renilla luciferase activities were quantitated using the Dual-Luciferase Reporter Assay System (Promega, E1960). Renilla luciferase values were normalized to firefly luciferase activity and plotted as relative luciferase activity.

Fluorescence titration assay

The fluorescence titration assay was performed as previously described,. The ERα/LBD protein was dissolved in phosphate-buffered saline (PBS) at a concentration of 2.5 μmol/L, and the protein solution was stabilized to room temperature. Protein solution (3 mL) was transferred to the cuvette. The excitation wavelength of the instrument was set to 280 nm and the emission wavelength was set between 300 and 500 nm. The excitation and emission slit width was set to 5 nm. The compounds were then added dropwise to the cuvette, and the fluorescence spectrum of each titration was recorded. The compound concentration was the abscissa, and the change in spectroscopic signal after each titration was plotted as the ordinate. Kd value was obtained by Eq. (1) using Origin software (OriginLab, Northampton, MA, USA):in which y represents the change of fluorescence intensity after each titration, x represents compound concentration, P represents protein concentration (P is 2.5 μmol/L here), C is a constant which measure the change in signal (ΔF) changes per unit complex (ligand–protein) concentration (molar signal coefficient).

Differential scanning calorimetry (DSC)

Calorimetric measurements were performed using the VP-DSC (GE Healthcare, Fairfield, CT, USA). The VP-DSC was run without feedback and 15–30 min equilibration times at 10 °C were used before or between scans. The ERα/LBD protein samples were concentrated to 2.5 mg/mL in PBS. The proteins were scanned from 10 to 100 °C at a heating rate of 90 °C/h. A cuvette containing PBS was used as a reference. DSC data was corrected for PBS base lines and normalized for scan rate and protein concentration.

Computer-aided docking

Computational docking of E2 and NOR with ERα was initially performed in order to be used as internal control for comparative scoring purpose. The initial dockings were carried out in Maestro 11.1 (Schrödinger, New York, NY, USA). The 3D structure of ERα (PDB 1ERE) was downloaded from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB). Here, the receptor grid files were generated covering the active site region with van der Waals radius scaling of 1.0 Å to soften the non-polar region of receptor and the other atoms were left free of scaling. Ten energetically favorable conformations were selected out of per docking. Among them, the best pose in terms of significant docking score was picked.

Osteoblast formation

MC3T3-E1 cells were seeded in a 24-well plate, and treated with a mineralization-inducing solution containing 10 mmol/L β-glycerophosphate and 50 μg/mL ascorbic acid. The cells were continuously cultured for 14 days, and the medium was changed every 2 days. Differentiated cells were stained with alizarin red staining at the indicated day. Cells were gently rinsed twice with PBS, fixed with 4% paraformaldehyde for 10 min, then washed twice with double distilled water. Cells were stained with 1% alizarin red S staining solution at 37 °C for 30 min, then washed with distilled water. The alizarin red staining was then visualized and photographed. When detecting ALP activity, the cells were lysed and the supernatant was used to determine the activity by the ALP detection kit (Beyotime Biotechnology) according to the manufacturer's instructions.

Osteoclast formation

Bone marrow monocytes were cultured for 7 days in the presence of M-CSF (50 ng/mL) and RANKL (50 ng/mL) for differentiation into mature osteoclasts. Media was refreshed every 2 days. RAW264.7 cells were seeded in 6-well plates in full medium containing RANKL (50 ng/mL) for 7 days as well. For tartrate-resistant acid phosphatase (TRAP) staining, cells were fixed and stained for TRAP activity. TRAP-positive multi-nucleated cells with ≥3 nuclei were counted as osteoclasts.

Cell proliferation and viability assay

MC3T3-E1 and RAW264.7 cells were seeded respectively into 96-well plates at the density of 5 × 103 cells/well in full medium with vehicle or corresponding differentiation-inducing agents medium. MCF-7 cells were plated at the concentration of 8 × 103 cells/well followed by 48 h incubation. The plates were incubated at 37 °C for 4 h after 10 μL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, 5 mg/mL) per well was added, and then the supernatant was removed. DMSO (150 μL) was added to dissolve the crystals on the oscillator for 10 min. The absorbance at 490 nm was measured by the plate reader.

Quantitative real-time PCR

Chiefly, total RNA was isolated using Trizol reagent (Life) according to the manufacturer's recommended protocol, and then 1 μg of RNA was used for cDNA synthesis with Revert Aid First Strand cDNA Synthesis Kit (Yeasen). Real-time PCR in triplicate was performed with SYBR Green Master Mix (Yeasen) on AriaMx real-time PCR System (Agilent, Santa Clara, CA, USA). All values were reported as the mean ± standard deviation (SD) of triplicate measurements of each cDNA sample. The mRNA levels were normalized to Gapdh mRNA. PCR primers used were as followed: mouse cathepsin k, forward 5′-ACGAGAAAGCCCTGAAGA-3′, and reverse 5′-TGTAAAACTGGAAAGATGCC-3′; mouse Nfatc1, forward 5′-ACCACCTTTCCGCAACCA-3′, and reverse 5′-TTCCGTTTCCCGTTGCA-3′; and Gapdh, forward 5′-AAGAAGGTGGTGAAGCAGG-3′ and reverse 5′-GAAGGTGGAAGAGTGGGAGT-3′.

Ovariectomized mice model

C57BL/6J female mice were purchased from Shanghai SLAC laboratory animal company. Mice were housed under standard conditions (temperature, 22 °C; humidity, 40%–60%; light, 12 h of light/dark cycle; and pathogen-free-controlled environment). Food and water were available ad libitum. Mice received bilateral ovariectomy (OVX) or sham operation (Sham) at 10-week old. Two weeks after surgery, the mice were randomly divided into 4 groups with 6 mice/group: vehicle-treated sham-operated (Sham + Veh), vehicle-treated OVX (OVX + Veh), NOR (1 mg/kg)-treated OVX (OVX + NOR) and E2 (1 mg/kg)-treated OVX (OVX + E2). NOR was administered orally for 8 weeks, while sham group and OVX control group were given with 0.5% CMC-Na. Mice were euthanized and their femurs and vertebras were dissected and collected. Bones were fixed in formalin for at least 48 h. The right femurs and 5th vertebras were scanned ex vivo using a micro-computed tomography (μ-CT) system (SKYSCAN 1272, Bruker, Belgium). The parameters of the μ-CT system were set as follows: source voltage, 70 kV; source current, 142 μA; scaling image pixel size, 11.199 μm; the filter, 0.5 mm. After the 3D reconstruction by InstaRecon (Belgium), we set the first section of cross from butterfly cut where started from the distal end of the femur, and then counted 50th to 200th section. Lastly, we used CTAn software to analysis the data of femoral and vertebral cancellous bone. For quantitative assessment, the bone parameters such as bone mineral density (BMD), bone volume/tissue volume (BV/TV) and trabecular bone number (Tb.N) were calculated with scanner software according to standardized protocols.

Western blotting and co-immunoprecipitation

Cells were lysed using cell lysis buffer (Beyotime Biotechnology) and centrifuged for 15 min at 4 °C followed by collection of the supernatant. After quantified by a BCA protein assay kit (Thermo Fisher Scientific), equal amounts of protein were loaded onto and separated by (8%–10%) SDS-PAGE gels and transferred to PVDF membranes. Membranes were blocked with 5% nonfat milk for 1 h. The blots were probed with correspondent specific primary antibodies at 4 °C overnight, followed by the secondary HRP conjugated anti-mouse/rabbit antibodies at room temperature for 2 h. After washing, the bands were detected by chemiluminescence and imaged. For co-immunoprecipitation (Co-IP), antibodies were added to the supernatant of cell lysis and the mixture was incubated at 4 °C overnight, and then IgG agarose beads were added and incubated for 3 h. The beads were isolated by centrifugation with 3000 rpm (Centrifuge 5424 R, Germany) for 5 min and washed three times with prechilled PBS, followed by protein denaturation at 100 °C for 5 min and Western blotting analysis.

Cell cycle analysis

Briefly, cells were collected and washed twice with prechilled PBS, then fixed with 70% ethanol at 4 °C overnight. The cells were incubated with DAPI (sigma) in the dark at room temperature for 10 min, then analyzed using Flow Cytometers.

TRAP staining analysis of lumbar vertebra osteoclast activity

The 5th lumbar vertebra (L5) of the mice was extracted, fixed at 4% paraformaldehyde at 4 °C for 24 h, and the non-decalcified bones were embedded in methyl methacrylate. Sections (5 μm thick) were obtained by a microtome (RM2255, Leica Microsystems, Wetzlar, Germany) and Leica TC65 microtome blade (14021626379, Leica Microsystems). For static histomorphometric measures of osteoclast parameters, undecalcified sections of the vertebra were stained using TRAP staining kit (G1492, Solarbio, Beijing, China). The Osteomeasure Analysis System (OsteoMetrics, Atlanta, GA, USA) was used for bone histomorphometry using standard procedures according to the program's instruction.

Animal study approval

All animal experiments complied with the National Institutes of Health guide for the care and use of Laboratory animals and were approved by the Animal Experiment Administration Committee of Xiamen University in Fujian, China.

Statistical analysis

Data were expressed as the mean ± SD. Each assay was repeated in triplicate in three independent experiments. The statistical analyses were performed using GraphPad Prism 8 software. The statistical significance of the differences among the means of several groups was determined using the Student's t test. Level of statistical significance was determined as P < 0.05.

Results

Norlichexanthone selectively activates ERα

To explore the potential ligands of ERα from plant endophytes, we used ERα mammalian one-hybrid assay to evaluate the crude extracts of several plant endophytes. Among the 6 endophytes examined, the crude extract from endophyte ARL-13, similar as ERα agonist 1,3,5-tris (4-hydroxyphenyl)-4-propyl-1H-pyrazole (PPT), robustly transactivated Gal4/DBD-ERα/LBD fusion protein (Supporting Information Fig. S1A and S1B). The sequence of internal transcribed spacer (ITS) indicated that ARL-13 belonged to Arthrinium arundinis genus (Supporting Information Fig. S2A and S2B). The result that ARL-13 extract strongly transactivated the Gal4 chimers of ERα but not RARα, RARγ or RXRα indicated the selectivity of the extract in ERα regulation (Supporting Information Fig. S3). These results indicate that ARL-13 extract contains substances that regulate ERα transcriptional activity.

The extract of ARL-13 underwent further separation and purification by silica gel column chromatography and pre-HPLC (Supporting Information Fig. S4), and several compounds were obtained, of which compound 1 potently activated Gal4/DBD-ERα/LBD (Fig. 1A). After examination by 1H and 13C NMR spectroscopy and mass spectrometry (Supporting Information Figs. S5–S7), compound 1 was identified as norlichexanthone (NOR, Fig. 1B). We found that NOR dose-dependently activated Gal4/DBD-ERα/LBD with an EC50 of 38.81 nmol/L (Fig. 1C). Nevertheless, the efficacy of NOR on activating ERα was much weaker than 17β-estradiol (E2, Fig. 1C). As expected, NOR selectively activated ERα but not RARα, RXRα or PPARγ (Fig. 1D). Taken together, these results indicate that compound NOR from plant endophyte is a selective modulator of ERα transactivation.

Figure 1

NOR selectively activates the transcriptional activity of ERα. (A) HEK293T cells transfected with pBind-ERα/LBD and pG5-luciferase plasmids were treated with E2 (100 nmol/L) or compound 1 (100 nmol/L) for 12 h. Cells were harvested, and firefly and Renilla luciferase activities were measured. Renilla luciferase values were normalized to firefly luciferase activity and plotted as relative luciferase (Luc.) activity. (B) The chemical structure of NOR. (C) HEK293T cells transfected with pBind-ERα/LBD and pG5-luciferase plasmids were treated with E2 (10 nmol/L) or NOR (1, 10, 20, 40, 60, 80, and 100 nmol/L). Luciferase activities were measured and plotted. (D) The effects of NOR on transactivating the chimers of Gal4/DBD fused with the LBD of ERα, RARα, RXRα or PPARγ. E2, 17β-estrodiol; ATRA, all-trans retinoic acid; 9-cis-RA, 9-cis retinoic acid; RGZ, rosiglitazone. Data are expressed as the mean ± SD; ∗∗∗∗P < 0.0001 by Student's t test.

NOR binds to ERα

To further characterize NOR as an ERα modulator, we explored whether NOR directly bound to ERα. The result of differential scanning calorimetry (DSC) assay shows that NOR potently increased not only the melting temperature (Tm) but also the melting energy of ERα/LBD protein (Fig. 2A). This result suggests that NOR binds to ERα to enhance its thermostability. We further used fluorescence titration assay to determine NOR binding to ERα. We found that NOR dose-dependently reduced the fluorescence intensity of ERα/LBD but not RXRα/LBD (Fig. 2B and C), while the fluorescence of ERα/LBD was not affected by RXRα ligand K-80003 (Fig. 2D). These results indicate that NOR binds to ERα selectively. The Kd value of NOR and ERα interaction obtained by the fluorescence titration assay was 10 nmol/L (Fig. 2B), and was of the same order of magnitude with the EC50 of NOR transactivating ERα. Together, these results indicate that NOR is a natural ligand of ERα.

Figure 2

NOR binds to ERα/LBD. (A) Differential scanning calorimetry (DSC) analysis of ERα/LBD protein incubated with DMSO (black curve) or NOR (green curve). Shown are the profiles of heat capacity (Cp) vs. temperature at a scan rate of 90 °C/h for the proteins. (B) Fluorescence titration assay of NOR binding to ERα/LBD. NOR was added sequentially to the cuvette with ERα/LBD protein and the fluorescence spectra were recorded for each titration. The excitation wavelength was set to 280 nm, and the emission wavelengths were between 300 and 500 nm. The compound concentration was the abscissa, and the change in spectroscopic signal after each titration was plotted as the ordinate (left panel). The Origin software and the formula y = C (x + P + Kd)/2−C × sqrt ((x + P + Kd)2−4Px)/2 was used to calculate the Kd (right panel). (C) and (D) Fluorescence titration assays of NOR binding to RXRα/LBD (C) and K-80003 binding to ERα/LBD (D). (E) The ERα binding modes of E2 (up panel) and NOR (down panel) were mimicked by computer-aided docking approach. The structures of E2 and NOR are shown in orange and green, respectively. The docking scores of E2 and NOR are −10.607 and −9.298, respectively.

We further used computer-aided docking approaches to examine the binding mode of NOR–ERα. Our computational modeling indicated that NOR and E2 had similar ERα binding mode (Fig. 2E). The hydroxyl group of NOR at site 3 formed hydrogen bond with His524 amino acid residue in the ligand binding pocket (LBP) of ERα, while the hydroxyl group at site 6 formed hydrogen bonds with Glu353 and Arg394 residues of ERα (Fig. 2E). Thus, the hydroxyl groups at 3 and 6 sites should be essential for NOR binding to ERα. The hydrogen bond formed between hydroxyl group of NOR and Arg394 was longer than the corresponding one formed between E2 and Arg394 (Fig. 2E), which may lead to the differences of NOR and E2 in regulating ERα activities.

NOR induces osteoblastic differentiation

In view of the fact that NOR was ERα ligand, we investigated whether NOR could induce osteoblastic differentiation as estrogen does. Calcium deposits is an indication of successful osteoblast formation in vitro and can specifically be stained bright red by Alizarin Red S. Mineralization-inducing solution induced calcium deposits in MC3T3-E1 cells, which was potently enhanced by NOR and E2 at 20 nmol/L concentration (Fig. 3A). This indicates that NOR and E2 could strengthen the effect of the inducing solution on osteoblastic induction. Importantly, NOR appeared to induce more calcium deposits than E2 at the same concentration (Fig. 3A). Consistently, NOR potently increased the activity of alkaline phosphatase (ALP), a marker of osteoblast, in induced MC3T3-E1 cells (Fig. 3B). NOR treatment did not show significant effect on the proliferation of MC3T3-E1 cells (Fig. 3C), indicating that the increased calcium accumulation induced by NOR did not result from cell proliferation and most likely resulted from the osteoblastic induction.

Figure 3

NOR induces osteoblastic differentiation. (A) and (B) MC3T3-E1 cells were treated with mineralization-inducing solution including 10 mmol/L of β-glycerophosphate (β-GP) and 50 μg/mL of ascorbic acid (A.A) together with E2 or NOR. After 14 days, cells were stained with alizarin red and photographed (A, left panel), and the percentages of the red areas in the total areas were measured and plotted (A, right panel); or cells were lysed and the activities of alkaline phosphatase (ALP) were measured and plotted (B). Scale bar represents 0.3 cm. (C) MC3T3-E1 cells were incubated with mineralization-inducing solution together with E2 or NOR for 72 h. Cell viability were determined by MTT assay. Data are expressed as the mean ± SD; ∗P < 0.05 by Student's t test.

NOR inhibits osteoclast formation

We then explored whether NOR affected osteoclast formation as well. Murine macrophage RAW264.7 cells were osteoclast precursors and can be induced differentiation into osteoclast by RANKL. In the absence of RANKL treatment, RAW264.7 cells were small and round cells with single nucleus. After RANKL treatment, large multi-nucleated cells emerged and were stained red by TRAP staining, indicating the occurrence of osteoclast formation (Fig. 4A). E2 has been reported to inhibit osteoclast formation, which was corroborated in our result showing that E2 dose-dependently reduced the numbers of multi-nucleated cells (Fig. 4A). NOR also significantly reduced the numbers of multi-nucleated cells with the comparable ability as E2 (Fig. 4A). The effect of NOR on osteoclast formation was also evaluated using the primary monocytes purified from murine marrow. As shown in Fig. 4B, RANKL-induced large multi-nucleated cell formation from the primary monocytes was potently inhibited by either NOR or E2. Moreover, NOR and E2 inhibited RANKL-stimulated mRNA expression of cathepsin k and Nfatc-1, two marker genes of osteoclasts43, 44, 45 (Fig. 4C). In the absence of RANKL, NOR and E2 had no apparent effect on RAW264.7 cell viability (Fig. 4D). However, they potently inhibited RAW264.7 cell viability in a dose-dependent manner when cells were pretreated with RANKL (Fig. 4E). These data indicate that NOR inhibited the survival of RANKL-induced osteoclast but not the precursor RAW264.7 cells. In sum, these data indicate that NOR inhibits osteoclast formation from osteoclast precursors.

Figure 4

NOR attenuates RANKL-induced osteoclast formation in vitro. (A) RAW264.7 cells were treated with RANKL (50 ng/mL) together with E2 or NOR for 7 days. Cells were then fixed and the tartrate-resistant acid phosphatase (TRAP) staining was applied to identify TRAP-positive multi-nucleated cells. Images were acquired at 100 × magnification on a microscope (left panel). Black arrows indicate the induced multi-nucleated cells. Scale bar represents 20 μm. The numbers of TRAP-positive multi-nucleated cells in a 4 × microscope field of view were counted and presented (right panel). (B) Monocytes purified from bone marrows of C57BL/6 mice were stimulated with 50 ng/mL of RANKL and 50 ng/mL of M-CSF together with E2 or NOR for 7 days and then stained for TRAP (left panel). Images were acquired at 20 × magnification on a microscope (left panel). Black arrows indicate the induced multi-nucleated cells. Scale bar represents 100 μm. The numbers of TRAP-positive multi-nucleated cells in a well of a 96-well plate were counted and presented (right panel). (C) RAW264.7 cells were stimulated with RANKL (50 ng/mL) in the presence or absence of different doses of E2 or NOR for 72 h. The mRNA expression of cathepsin K and Nfatc-1 were detected by quantitative real-time PCR. (D) RAW264.7 cells were treated with NOR or E2 at the indicated concentrations for 72 h. Cell viability were determined by MTT assay. (E) RAW264.7 cells were pretreated with 50 ng/mL RANKL for 72 h, and then treated with NOR or E2 together with RANKL for 72 h. Cell viability were determined by MTT assay. Data are expressed as the mean ± SD; ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001 by Student's t test.

NOR inhibits RANKL signaling by preventing ERα-induced TRAF6 ubiquitination

Furthermore, we explored the underlying mechanism of NOR actions in osteoclastogenesis. The master transcriptional factor inducing osteoclastic differentiation is NFATc1, which is induced by transcriptional factor AP-1, a heterodimer of c-JUN and c-FOS. AP-1 is regulated by the c-JUN N-terminal kinase (JNK), P38 and extracellular signal-regulated kinase (ERK) MAPKs, downstream of RANKL signaling,. The downregulation of RANKL-induced Nfatc1 mRNA expression by NOR and E2 suggested that they may inhibit RANKL signaling (Fig. 4C). Indeed, both NOR and E2 dose-dependently inhibited RANKL-induced phosphorylation of JNK, P38 and ERK but not their expression in RAW264.7 cells (Fig. 5A). Thus, NOR and E2 were inhibitors of RANKL signaling. ICI-182780 (fulvestrant) is an antagonist of ER,. Interestingly, the inhibitory effect of NOR and E2 on RANKL-induced JNK activation was blocked by fulvestrant (Fig. 5B). To further characterize ERα or ERβ dependent effects of NOR, we used the ERα selective antagonist endoxifen and the ERβ selective antagonist PHTPP. Endoxifen but not PHTPP substantially abrogated the effect of NOR and E2 on RANKL-induced JNK activation (Fig. 5C). These data indicate the ERα-dependent effect of NOR and E2 on inhibiting RANKL signaling.

Figure 5

NOR suppresses RANKL signaling by preventing ERα-induced TRAF6 ubiquitination. (A) RAW264.7 cells were pretreated with E2 or NOR for 4 h and subsequently stimulated with RANKL (50 ng/mL) for 20 min. Cell lysates were subjected to Western blotting analysis of the phosphorylation of JNK, P38 and ERK. The bands of phosphorylated proteins and their corresponding total proteins were quantified by Image J, and the ratio was plotted. (B) and (C) RAW264.7 cells were pretreated with ICI-182780 (1 μmol/L), endoxifen (0.2 μmol/L) or PHTPP (0.2 μmol/L) for 24 h and then treated with E2 or NOR for 4 h before RANKL (50 ng/mL) stimulation for 20 min. Western blotting was applied to detect protein expression. The gray level of p-JNK bands were normalized to that of total JNK bands. (D) HEK293T cells were co-transfected with the indicated expression plasmids for 24 h and then treated with E2 or NOR for 2 h. Flag-TRAF6 was immunoprecipitated by anti-Flag antibody, and Lys63-linked and total ubiquitination of TRAF6 were examined by Western blotting using anti-K63-linked-ubiquitin and anti-ubiquitin antibodies, respectively. (E) HEK293T cells were co-transfected with Flag-ERα and MYC-TRAF6 expression plasmids for 24 h, followed by treatment with E2 or NOR for 2 h. Protein interactions were examined by immunoprecipitation using anti-Flag antibody and Western blotting using anti-MYC antibody. (F) RAW264.7 cells were pretreated with E2 or NOR for 2 h and subsequently stimulated with RANKL (50 ng/mL) for 20 min. Protein interactions were examined by immunoprecipitation using anti-ERα antibody and Western blotting using anti-TRAF6 antibody. Normal IgG was as an immunoprecipitation control. (G) Working model of ERα and NOR in RANKL signaling. Data were expressed as the mean ± SD; ns, non-significant; ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001 by Student's t test.

In the RANKL signaling pathway, RANK-triggered ubiquitination of TRAF6 initiates the activation of downstream signal cascades. The triggered auto-ubiquitination of TRAF6 via a Lys63-linked polyubiquitin chain mediates the recruitment of TAK1 binding protein 2 (TAB2), leading to TAK1 activation and in turn the phosphorylation of MKK. We found that ectopically expressed ERα potently induced the Lys63-linked ubiquitination of TRAF6 in HEK293T cells (Fig. 5D). Interestingly, ERα-induced ubiquitination of TRAF6 was inhibited by either NOR or E2 (Fig. 5D). These data suggest that NOR and E2 antagonize RANKL signaling through inhibiting ERα-induced TRAF6 ubiquitination. Moreover, the ectopically expressed ERα interacted with TRAF6, which was inhibited by NOR and E2 in a dose-dependent manner (Fig. 5E). In addition, RANKL induced the endogenous interaction of ERα and TRAF6 in RAW264.7 cells, which was also inhibited by either NOR or E2 (Fig. 5F). Together, these data suggested that ERα binds and then induces TRAF6 ubiquitination while NOR and E2 inhibit RANKL-induced interaction of ERα and TRAF6 to exert their inhibitory effect on RANKL signaling (Fig. 5G).

NOR prevents osteoporosis in a mouse postmenopausal osteoporosis model

To evaluate the potential efficacy of NOR in osteoporosis, we used an ovariectomized (OVX) mouse model, a well-established model of estrogen deficiency-mediated osteoporosis. After ovariectomized operation, mice were oral gavaged with NOR and E2 daily for 8 weeks, respectively. Right femurs were obtained and subjected to μ-CT analysis of the trabecular architecture. As shown in Fig. 6A, the structure of trabecular of distal femurs was severely impaired in OVX mice comparing to control (Sham) mice. The bone mineral density (BMD), bone volume density (BV/TV) and trabecular number (Tb.N) were largely reduced in OVX mice comparing to sham mice (Fig. 6A). These results indicate the success of our OVX model in bone loss. Notably, NOR or E2 at the dosage of 1 mg/kg potently prevented the ovariectomy-induced deterioration of trabecular architecture (Fig. 6A). The ovariectomy-induced reduction of BMD, BV/TV and Tb.N were inhibited by NOR or E2 treatment (Fig. 6A). Similar result was obtained in the fifth vertebras of mice (Fig. 6B). Moreover, the ovariectomy-induced osteoclast formation in vertebras was inhibited by NOR or E2 (Fig. 6C). Therefore, NOR showed strong efficacy in osteoporosis prevention in an OVX mouse model.

Figure 6

NOR ameliorates ovariectomy-induced bone loss in vivo. Female mice were subjected to bilateral ovariectomy (OVX) or sham operation (Sham). NOR and E2 were administered orally for 8 weeks. Mice were then sacrificed. Mouse femurs (A) and the 5th vertebras (B) were analyzed by micro-CT and the 3D reconstructions of trabecular were obtained using the software CTAn. The bone mineral density (BMD), bone volume density (BV/TV) and trabecular number (Tb.N) were determined by the scanner software. The 5th vertebras were sliced into 5-μm thick sections to perform tartrate-resistant acid phosphatase (TRAP) staining (C). Scale bar represents 200 μm. The osteoclast surface (Oc.S) and bone surface (BS) of TRAP-stained histologic sections of vertebra were respectively measured by Image J, and the fold change in osteoclast surface/bone surface (Oc.S/BS) was analyzed by GraphPad Prism 8 (C). Data are expressed as the mean ± SD, n = 6; ∗P < 0.05; ∗∗P < 0.01 by Student's t test.

NOR is of weak estrogenic activity

One of the major adverse effects of using estrogen for osteoporosis treatment is the increased risk for endometrial hyperplasia and breast cancer promotion originating from its strong estrogenic activity. One of the estrogenic effects is manifested in maintaining uterus morphology and function. In our OVX mouse model, we observed the significant shrinkage and weight reduction of uterus (Fig. 7A and Supporting Information Fig. S8A), resulting from the lack of estrogen due to the ovariectomy. Treatment with E2 at 1 mg/kg dosage could rescue OVX-induced uterine shrinkage and weight loss. Even more, the uterine weight of E2-treated OVX mice was much higher than that of sham mice (Fig. 7A and Fig. S8A), exhibiting the strong potency of E2 in uterine stimulation. However, NOR at the dosage of 1 mg/kg failed to show apparent effect on morphological change and weight increase of uterus in OVX mice (Fig. 7A and Fig. S8A), indicating the weak estrogenic potency of NOR in uterus.

Figure 7

NOR is of weak estrogenic activity. (A) Weights of uterus from Sham and OVX mice treated with E2 or NOR for 8 weeks. (B) MCF-7 cells were treated with E2 or NOR for 48 h. MTT assay was performed to examine cell proliferation. (C) MCF-7 cells were treated with E2 or NOR for 24 h. The cell cycles were detected by flow cytometry analysis. Data were expressed as the mean ± SD; ns, non-significant; ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001 by Student's t test.

Treatment with E2 strongly promoted proliferation and cell cycle progression of MCF-7 but not ERα-null MDA-MB-231 cells (Fig. 7B and C, Fig. S8B and S8C), verifying the onco-promotive effect of E2 on breast cancer in an ERα-dependent manner. However, NOR failed to show strong effect as E2 on cell cycle progression and proliferation of either MCF-7 or MDA-MB-231 cells (Fig. 7B and C, Fig. S8B and S8C), indicating the weak effect of NOR on breast cancer promotion. Therefore, these results indicate that NOR is of weak estrogenic activity.

Discussion

Discovering and dissecting novel ligands of ERα from natural products would not only provide potential drugs for ERα-related diseases, but also aid to reveal novel mechanisms of ERα activities. Here, we used ERα mammalian one-hybrid assay to screen for ERα modulators from plant endophytes and successfully identified NOR as an active ligand of ERα. Our ligand-receptor binding assay revealed the nanomolar range of ERα binding affinity of NOR. Moreover, NOR at nmol/L concentrations was able to regulate ERα transcriptional activity, osteoblast/osteoclast formation and RANKL signaling. These results indicate that NOR is a high-affinity ligand of ERα and a high-potency modulator of some biological activities, which guarantees the further exploration of ERα-dependent activities of NOR and the further optimization of NOR for drug development. The structure of NOR is distinct from the clinical ERα-targeted drugs, which lies down the foundation to develop NOR for regulating distinct ERα-related pathological processes. Moreover, NOR and its optimized derivatives may overcome the resistance or adverse effects of current clinical ERα-targeted drugs.

Our results reveal that NOR is of strong effect in osteoporosis prevention. We found that NOR not only induced osteoblast formation but also prevented osteoclast formation. Notably, the above capability of NOR was comparable to E2. As for osteoclast formation, NOR not only inhibited osteoclastic differentiation, indicating from NOR-induced decrease of osteoclast formation from precursor cells, but also inhibited osteoclast growth or survival, indicating from NOR-induced viability decrease of RANKL-pretreated RAW264.7 cells. RANKL signaling is essential for inducing osteoclastic differentiation as well as for maintaining osteoclast survival and growth. Indeed, we found that NOR potently inhibited RANKL signaling, revealing the underlying mechanism of osteoclastic inhibition by NOR. Furthermore, we revealed that ERα was able to interact with TRAF6 and induce its Lys63-linked ubiquitination, demonstrating the positive role of ERα in RANKL signaling. Importantly, NOR potently inhibited ERα-induced TRAF6 ubiquitination and RANKL-induced interaction of ERα and TRAF6, resulting in down-regulation of RANKL signaling. Thus, our study unraveled a non-genomic action of ERα in regulating RANKL signaling, which was negatively regulated by NOR.

The pivotal role of TRAF6 in transducing RANKL signaling confers it an attractive drug target for modulating RANKL signaling. Downregulation of its expression and activity will inhibit RANKL signaling and thereby inhibit RANKL-induced osteoclastogenesis. It has been reported that icaritin, a natural prenylflavonoid, inhibits RANKL signaling by inducing TRAF6 poly-ubiquitination and proteasomal degradation. We also found that NOR inhibited RANKL signaling. However, we did not observe significant effect of NOR on decreasing TRAF6 protein level. Whereas Lys48-linked poly-ubiquitination results in TRAF6 degradation, Lys63-linked poly-ubiquitination activates TRAF6 in RANKL signaling,,. Indeed, we found that NOR strongly inhibited ERα-induced Lys63-linked ubiquitination of TRAF6. Thus, instead of downregulation of TRAF6 expression, NOR was able to inhibit TRAF6 activation to antagonize RANKL signaling. Notably, we found that NOR acted through targeting ERα, which may provide an approach for interfering TRAF6 activity by small molecules targeting ERα.

E2 has been shown to downregulate RANKL expression likely through transcriptional regulation to inhibit RANKL signaling,. However, in our cell model this mechanism was excluded because we used exogenous RANKL to stimulate the signaling pathway. Thus, the genomic and non-genomic actions of E2 might both contribute to its inhibition of RANKL signaling and osteoclastogenesis. Whereas NOR possessed weaker ability than E2 in regulating ERα transactivation, it was of the comparable ability as E2 in inhibiting RANKL signaling and osteoclast formation. Thus, the regulation of non-genomic actions of ERα by some ligands, such as NOR, may produce enough benefit for osteoporosis. Development of ERα biased ligands with strong regulation of ERα non-genomic activity but weak regulation of ERα transactivation might dissociate the beneficial effect of ERα targeting from some adverse effects.

One major concern of estrogen in osteoporosis treatment is to increase the risk of endometrial, ovarian and mammary carcinoma, although clinical studies have proven that longtime and low-dose HRT has no significant risk of these carcinomas. We found that NOR was of much low ability in promoting MCF-7 proliferation and restoring the ovariectomy-induced loss of uterine weight. This indicate the much less risk of NOR than E2 in promoting estrogen-related carcinomas. The less estrogenic activity of NOR might result from its lower ability of inducing ERα transactivation or its tissue-selective activation property. However, further detailed investigation should be performed to verify the safety of NOR in osteoporosis treatment.

Conclusions

Collectively, our study demonstrates that natural product NOR, purified from plant endophyte, is a high-affinity ligand of ERα and a potential drug lead for prevention and treatment of postmenopausal osteoporosis. Moreover, we reveal the positive role and non-genomic action of ERα in RANKL signaling and the underlying mechanism of osteoclastic inhibition of NOR via targeting ERα.

Acknowledgments

This work was supported by the (Grant Nos. 31770811, 31471318 and 31271453), the (Grant No. 20720190082, China), the Regional Demonstration of Marine Economy Innovative Development Project (Grant No. 16PYY007SF17, China) and the (Grant No. 2017YZ0002-1, China).

Author contributions

This study was designed by Keqi Wang, Yongyan Chen, Ting Lin and Hu Zhou; and was conducted by Keqi Wang, Yongyan Chen, Shuo Gao, Maosi Wang, Mengmeng Ge, Qian Yang, Mingkai Liao, Lin Xu and Junjie Chen. Data were analyzed and interpreted by Keqi Wang, Yongyan Chen, Zhiping Zeng, Haifeng Chen, Xiao-kun Zhang, Ting Lin and Hu Zhou. The manuscript was written by Ting Lin and Hu Zhou; and revised by Keqi Wang, Yongyan Chen, Shuo Gao, Zhiping Zeng, Haifeng Chen, Xiao-kun Zhang, Ting Lin and Hu Zhou. All authors have approved the final article.

Conflicts of interest

The authors declare that they have no conflict of interest.