Effect and Mechanism of Psoralidin on Promoting Osteogenesis And

Phytomedicine 61 (2019) 152860

Contents lists available at ScienceDirect

Phytomedicine

journal homepage: www.elsevier.com/locate/phymed

Original Article Eﬀect and mechanism of psoralidin on promoting osteogenesis and inhibiting adipogenesis T

Hui-juan Caoa, Cai-rong Lia, Lin-ying Wanga, Reihane Ziadloub, Sibylle Gradb, Yan Zhangd, ⁎ ⁎ Yan Chenge, Yu-xiao Laia, Xin-sheng Yaof, Mauro Alinib, Ling Qina,c, , Xin-luan Wanga,c, a Translational Medicine R&D Center, Institute of Biomedical and Health Engineering, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen 518057, China b AO Research Institute Davos, Clavadelerstrasse 8, Davos Platz 7270, Switzerland c Musculoskeletal Research Laboratory of Department of Orthopaedics & Traumatology and Innovative Orthopaedic Biomaterial and Drug Translational Research Laboratory of Li Ka Shing Institute of Health, The Chinese University of Hong Kong, Hong Kong, China d Spine Disease Research Institute, Longhua Hospital, Shanghai University of Traditional Chinese Medicine, Shanghai 200032, China e Department of Pharmacology, School of Pharmaceutical Sciences, Central South University, Changsha 410008, China f College of Pharmacy, Jinan University, Guangzhou 510632, China

ARTICLE INFO ABSTRACT

Keywords: Background: Psoralidin (PL), a prenylated coumestrol, is isolated from Psoralea corylifolia L. (Fabaceae), which is Psoralidin (PL) frequently used for treatment of osteoporosis. Osteogenesis Purpose: This study was designed to investigate the dual eﬀects and potential mechanism of PL on promoting Adipogenesis osteogenesis and inhibiting adipogenesis. Estrogen receptor (ER) Methods: Bone marrow mesenchymal stem cells (BMSCs) were used to investigate the eﬀect of PL on stimulating GRP30 receptor osteogenesis and inhibiting adipogenesis, while preosteoblast MC3T3-E1 cells and preadipocyte 3T3-L1 cells

were employed to explore the potential mechanisms. Estradiol (E2) and ICI 182,780 (ICI) were used as the specific agonist and antagonist of classical estrogen receptors (ER), respectively, to interfere with classical ER signaling. Meanwhile, G-1 and G-15 were introduced as the selective agonist and antagonist of G protein coupled receptor 30 (GRP30, a membrane ER) to further clarify if membrane ER involved in PL mediating osteogenesis and adipogenesis Results: PL not only promoted mineralization, but also inhibited adipocytes formation of BMSCs. In terms of osteogenesis, PL enhanced calcium nodule formation, alkaline phosphatase activity and osteocalcin levels in MC3T3-E1 cells. As for adipogenesis, PL decreased adipocyte formation in 3T3-L1 cells through down-regulating several mRNA expressions and protein synthesis of adipogenesis related factors. ICI completely blocked the effect of PL in promoting osteogenesis, but only partially suppressed its effect in inhibition of adipogenesis, while G-15 partially suppressed the effect of PL on promoting mineralization and inhibiting oil drop formation. Furthermore, during suppression of adipocyte differentiation, PL regulated protein kinase B / glycogen synthase kinase 3β / β-catenin signaling pathway. Conclusion: PL promoted osteogenesis via mediating classical ER pathway, and inhibited adipocytes formation by regulating combined classical and membrane ER pathways. PL might be a potential candidate for the treatment of postmenopausal osteoporosis by modulating the competitive relationship between osteogenesis and adipogenesis of bone marrow mesenchymal stem cells.

Abbreviations: AD, adipogenic differentiation medium; AKT, protein kinase B; ALP, alkaline phosphatase; ATCC, American Type Culture Collection; BCA, bicinchoninic acid assay; β-catenin or β-cat, beta-catenin; BSP, bone sialoprotein; BMSC, Bone marrow-derived mesenchymal stem cell; CCK-8, cell counting kit-8; cDNA, complementary deoxyribonucleic acid; CEBPα, CCAAT/enhancerbinding protein α;CO2, carbon dioxide; d, day or days; DMEM, dulbecco's modified eagle medium; DMSO, dimethylsulfoxide; E2,17β-estradiol; ER, estrogen receptor; ERE, estrogen response element; ERSE, estrogen-responsive suppressive element; FABP4, adipocyte lipid-binding protein 4; α-MEM, alpha modifications minimum essential medium ⁎ Corresponding authors. E-mail addresses: [email protected] (L. Qin), [email protected] (X.-l. Wang). https://doi.org/10.1016/j.phymed.2019.152860 Received 6 August 2018; Received in revised form 29 January 2019; Accepted 3 February 2019 0944-7113/ © 2019 Elsevier GmbH. All rights reserved. H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Introduction the molecular mechanism of PL on promoting osteogenesis and inhibiting adipogenesis. Osteoporosis is characterized by decreased bone mass and increased adipocytes in bone marrow (Schwartz, 2015). More importantly, sex steroid deficiency in postmenopausal women fails to stimulate osteo- Materials and methods blastogenesis and delays apoptosis of osteoclasts, thereby accelerating the loss of bone mass (Hughes et al., 1996). Previous studies demon- Reagents strate that the osteoblastic lineage is essential for regulating bone remodeling, while increased adipogenesis after loss of sex steroid further All reagents or antibodies listed below were commercially available, diminishes osteoblast precursors, resulting in the imbalance of bone including PL from Nanjing TCM institute of Chinese Materia (Nanjing, remodeling (Manolagas, 2000; Weinstein et al., 1997). It is well known China), purity: nesis and inhibiting adipogenesis.he Tocris bioscience (Minneapolis, MN, USA); alpha modifications minimum essential that bone marrow adipocytes and osteoblasts share common progenitor ® medium (α-MEM), fetal bovine serum (FBS), Stempro adipogenic dif- cells, bone marrow-derived mesenchymal stem cells (BMSCs), and the ® differentiation of BMSCs into adipocytes or osteoblasts is competitive ferentiation medium (AD medium) kit (Cat No. A10070-01), Stempro ff balance (Muruganandan et al., 2009). BMSCs isolated from post- osteogenic di erentiation medium (OD medium) kit (Cat No. A10072- menopausal osteoporotic patients favor adipocytes differentiation and 01), dimethylsulfoxide (DMSO), radio immunoprecipitation assay ff fl inversely, the progression of osteoblasts differentiation is suppressed in (RIPA) lysis and extraction bu er, phenylmethanesulfonyl uoride ex vitro culture (Rodriguez et al., 2000). Marrow fat as an endocrine (PMSF), phosphatase inhibitor cocktail and bicinchoninic acid assay fi ™ tissue, releases cytokines, adipokines and fatty acids, and directly af- (BCA) protein assay kit from Thermo Fisher Scienti c (Waltham, MA, fi fects the osteoblasts survival and function (Paccou et al., 2015; USA); penicillin-streptomycin (PS), trypsin and Dulbecco's modi ed Schwartz, 2015). Meanwhile, adiponectin enhances osteoclastic differ- eagle medium (DMEM) from HyClone Laboratories Inc.(Logan, UT, entiation by regulating osteoblasts, as it promotes osteoblasts expres- USA); cell counting kit-8 (CCK-8) kit from Dojindo Lab (Tokyo, Japan); β sing RANKL, which binds to RANK of preosteoclast and enhance os- E2, Oil Red O, Alizarin Red S, -glycerol phosphate and ascorbic acid teoclast differentiation (Goto et al., 2011). Hence, the changes of from Sigma-Aldrich (St. Louis, MO, USA); alkaline phosphatase (ALP) BMSCs differentiation not only affected the rate of osteoblast formation, activity test kit from Beyotime Institute of Biotechnology (Jiangsu, China); RNeasy total RNA extraction kit from Axygen (Union City, CA, but of osteoclast formation. Reduction of adipogenesis and promotion ® of osteoblastogenesis of BMSCs are therefore becoming an important USA); PrimeScript RT reagent Kit and SYBR Green PCR Master Mix target for prevention and treatment of osteoporosis after the decline of from TaKaRa Biotechnology (Dalian) Co. Ltd (Dalian, China); poly- fl sex steroid (Takada et al., 2009). Clinically, early bone loss in post- vinylidene uoride (PVDF) membrane from Bio-Rad Laboratories menopausal women with osteoporosis could be reversed by menopausal (Hercules, CA, USA); Enhanced chemiluminescence detection kit from hormone therapy (MHT) at short term (Sassarini and Lumsden, 2015), Multi Sciences Biotech Co. Ltd. (Hangzhou, China); the primers were β as MHT could rapidly decrease both bone resorption and adiposity, and purchased from BGI (Shenzhen, China); anti- -actin (E021020-01) from promote new bone formation (Limonard et al., 2015). However, long- EarthOx (San Francisco, CA, USA); and anti-BMP-2 (ab6285), anti-bone term follow-up studies reported some adverse effects after MHT treat- sialoprotein (BSP) (ab84787), anti-osterix (ab94744), anti-runx2 β ment, including breast cancer, endometrial cancer and vaginal bleeding (ab76956), anti- -catenin (ab32505), anti-adiponectin (ab22554), anti- (Gomaa et al., 2015). So, estrogen analogues with high clinical efficacy resistin (ab119501), anti-leptin (ab3583), anti-protein kinase B (AKT) and safety have attracted more attention from scientists (He et al., (ab32505), anti-p-AKT (S473) (ab66138), anti-glycogen synthase ki- β β 2017). nase 3 (GSK3 ) (ab32391), anti-p-GSK3 (S9) (ab131097), goat anti- Psoralidin (PL), 3,9-dihydroxy-2-(3-methylbut-2-enyl)-[1]benzo- rabbit lgG H&L (HRP) (ab6721) and goat anti-mouse lgG H&L (HRP) furo[3,2-c]-chromen-6-one, was isolated from Psoralea corylifolia L. (ab6789) from Abcam (Cambridge, MA, USA); osteocalcin (OCN) elisa (Fabaceae) and it is structurally similar to coumestrol, 3,9-dihydroxy-6- kit (SEA471Mu) from Cloud-Clone Corp (Katy, TX, USA). benzofurano[3,2-c]-chromanone (Fig. 1), a typical phytoestrogen ffi (Liu et al., 2014). Coumestrol has high a nity to estrogen receptor Isolation and culture of BMSCs (ER), which has shown to enhance osteogenic proteins synthesis and accelerated mineralization of osteoblast-like cells through mediating ER Sprague Dawley (SD) rats (4–5 weeks old; male) weighing signaling (Kuiper et al., 1997; Wu et al., 2009). Moreover, coumestrol 100–120 g were purchased from Guangdong Medical Laboratory was reported to inhibit oil droplet accumulation in adipocytes Animal Center (Guangzhou, China) and maintained in a special pa- (Szkudelska et al., 2002). PL is a prenylated coumestrol and recent thogen-free facility at Shenzhen Institutes of Advanced Technology, studies pointed out that PL possessed higher osteogenic activity than Chinese Academy of Sciences. As previously described (Ning et al., coumestrol, mainly attributed to the isopentenyl group (Zhai et al., 2015), bone marrow cells were harvested from femora of rats and then 2017, 2018). In vivo, PL inhibited bone loss in ovariectomized (OVX) filtered through a 70 mm cell strainer, centrifuged at 1200 rpm for rats (Zhai et al., 2018), while it promoted osteogenesis and inhibited 5 min and re-suspended in α-MEM with 20% FBS and 1% PS. Fresh osteoclasts formation and function in vitro (Zhai et al., 2017). medium was replaced every 3 days. Cells from passage 3 were used for ff Recently, increased evidences showed that the physiological e ects the following experiments. of estrogen are mediated by classical nuclear ERs and transmembrane G protein coupled receptor 30 (GPR30, a membrane ER) pathways, both of which can regulate rapid cellular activation events as well as the Osteogenic and adipogenic differentiation of rat-BMSCs long-term genomic transcriptional events (Prossnitz et al., 2008). However, the effect of PL on regulating the competitive relationship The rat-BMSCs were plated at a density of 1 × 105 cells/well in 12- between osteogenesis and adipogenesis is unclear. Thus, the present well cell culture plates in growth medium (α-MEM with 10% FBS and study was designed to use rat BMSCs to evaluate the effect of PL on 1% PS) in the incubator with 5% CO2 at 37 °C, until reaching 100% regulating osteogenic and adipogenic differentiation, and then employ confluence (designated as day 0). At day 0, the medium was replaced preosteoblast MC3T3-E1 cells and preadipocyte 3T3-L1 cells to in- with AD or OD medium for differentiation induction. Each AD and OD vestigate the underlying mechanism. In addition, we employed classical medium was refreshed every 2 and 3 days respectively. The dual-di- nuclear ER agonist estradiol (E2) and antagonist ICI182,780 (ICI), to- rectional differentiation experiment was conducted according to the gether with membrane ER agonist G-1 and antagonist G-15 to clarify design showed in Table 1.

2 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Fig. 1. Structures of psoralidin (A), the isopentenyl side chain at C2 of psoralidin is in rectangle with dotted lines; coumestrol (B), 17β-estradiol (C), ICI 182,780 (D), G-1 (E) and G-15 (F).

Table 1 MC3T3-E1 cells culture and osteogenic differentiation Study design in dual-directional differentiation experiment of BMSCs and treatment periods. MC3T3-E1 preosteoblastic cells (subclone 14, CRL-2594) were purchased from American Type Culture Collection (ATCC) (Manassas, Group aroup AD* OD E2 (0.1 m0. PL (10 ml) VA, USA). They were maintained in α-MEM with 10% FBS and 1% PS Normal 0–12 days –– – – (MC3T3-E1 cells growth medium) in the incubator with 5% CO2 at AD-8d – 0–8 days –– – fl – – – – – 37 °C and cultured until reaching 100% con uence (designated as day AD-8d/E2 0 8 days 0 8 days ff AD-8d/PL – 0–8 days –– 0–8 days 0). At that day 0, then the medium was changed to osteogenic di er- OD-12d ––0–12 days –– entiation medium (growth medium with 10 mM β-glycerol phosphate ––– – OD-12d/E2 0 12 days 0 12 days and 50 μg/ml ascorbic acid) for next 18 days. Osteogenic differentiation ––– – – OD-12d/PL 0 12 days 0 12 days medium was refreshed every 3 days. AD-1d – 0–1 days 2–12 days –– AD-2d – 0–2 days 3–12 days –– AD-4d – 0–4 days 5–12 days –– ff AD-1d/E2 – 0–1 days 2–12 days 0–12 days – 3T3-L1 cells culture and adipogenic di erentiation

AD-2d/E2 – 0–2 days 3–12 days 0–12 days – – – – – – AD-4d/E2 0 4 days 5 12 days 0 12 days 3T3-L1 preadipocytes (CL-173) were purchased from ATCC. They AD-1d/PL – 0–1 days 2–12 days – 0–12 days AD-2d/PL – 0–2 days 3–12 days – 0–12 days were maintained in DMEM with 10% FBS and 1% PS (3T3-L1 cells AD-4d/PL – 0–4 days 5–12 days – 0–12 days growth medium) in the incubator with 5% CO2 at 37 °C and cultured until reaching 100% confluence (designated as day 0). At that day 0, ⁎ AD: Adipogenic differentiation; OD: Osteogenic differentiation; d: day or cells were cultured with adipogenic differentiation medium (50% AD days; -: Null. medium and 50% 3T3-L1 cells growth medium) for 2 days. Then cells

3 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Table 2 Primer sequences used for real-time PCR.

Gene Forward primer Reverse primer

M*-gapdh 3′-CATGTTCCAGTATGACTCCACTC-5′ 3′-GGCCTCACCCCATTTGATGT-5′ M-runx2 3′-AGACTGCAAGAAGGCTCTGG-5′ 3′-TTCCTGCATGGACTGTGGTT-5′ M-osterix 3′-GTCCTCTCTGCTTGAGGAA-5′ 3′-CTTGAGAAGGGAGCTGGGT-5′ M-bsp 3′-CAGGGAGGCAGTGACTCTTC-5′ 3′-AGTGTGGAAAGTGTGGCGTT-5′ M-ocn 3′-CTGACCTCACAGATCCCAAGC-5′ 3′-TGGTCTGATAGCTCGTCACAAG-5′ M-c/ebpα 3′-GAACAGCAACGAGTACCGGGTA-5′ 3′-GCCATGGCCTTGACCAAGGAG-5′ M-pparγ 3′-CGCTGATGCACTGCCTATGA-5′ 3′-AGAGGTCCACAGAGCTGATTCC-5′ M-fabp4 3′-CATGGCCAAGCCCAACAT-5′ 3′-CGCCCAGTTTGAAGGAAATC-5′ M-lpl 3′-GGGAGT TTGGCTCCAGAGTTT-5′ 3′-TGTGTCTTCAGGGGTCCTTAG-5′

*M: Mouse. were incubated in the 3T3-L1 growth medium for other 4 days, re- of calcium deposition was quantiﬁed by de-staining with 10% ce- freshing the medium every 2 days. tylpyridinium chloride monohydrate in 10 mM sodium phosphate at room temperature for 30 min. The absorbance of the solution was Drugs treatment measured at 562 nm.

PL (10 mM), E2 (10 mM), ICI (10 mM), G-1(1 mM) and G-15 (1 mM) Oil Red O staining in 3T3-L1 cells were dissolved in DMSO and stored at 4 °C as stock solutions. Stock solutions were diluted by normal growth medium or differentiation 3T3-L1 cells (4 × 104 cells/well) were plated into 12-well plate and medium for the following experiments. Concentration of DMSO in the maintained for 2 days after reaching 100% confluence (designated as working solution was lower than 0.2%, which has been regarded as a day 0). Then, the cells were treated with PL (1 and 10 μM), E2 (0.1 μM), safe concentration (Kloverpris et al., 2010). ICI (1 μM), G-1(1 μM) or G-15 (1 μM) in adipogenic differentiation medium for 2 days. Cells were then incubated in the 3T3-L1 cells Cell toxicity growth medium with drugs up to day 6. Cells were stained with Oil Red O staining and pictures were taken by a dissecting microscope Cell viability was assessed using CCK-8 kit (Wang et al., 2013). (Olympus, Tokyo, Japan). The amount of oil droplets was quantified by BMSCs, MC3T3-E1 cells and 3T3-L1 cells were seeded at 5000 cells/ de-staining with isopropanol at room temperature for 10 min. The ab- well in 96-well plates. After 24 h of incubation, cells were treated with sorbance of the solution was measured at 520 nm.

PL (0.001, 0.01, 0.1, 1, 10 and 20 μM) or E2 (0.1 μM) in growth or differentiation medium. After 48 h, CCK-8 solution was added, and cells Quantitative real-time polymerase chain reaction (PCR) were incubated at 37 °C for 1 h. Finally, absorbance was measured on a microplate reader (PerkinElmer, Waltham, MA, USA) at 490 nm. The Total ribonucleic acid (RNA) was isolated using the RNeasy total experiment was repeated 3 times. RNA extraction kit, following the manufacturer's protocol. The total RNA (200 ng) was reverse-transcribed to complementary deoxyr- ALP activity assay ibonucleic acid (cDNA) using PrimeScript RT reagent kit with oligo dT primer and random 6 mers, following the manufacturer's protocol. The MC3T3-E1 cells were plated at a density of 3 × 104 cells/well in 12- real time PCR primers used in the experiments are shown in Table 2. well cell culture plates and cultured until reaching 100% confluence The 20 μl of the final reaction solution contained 1μl of the diluted ® (designated as day 0). At day 0, cells treated with OD medium and cDNA product, 10 μl of 2 × Power SYBR Green PCR Master Mix, 0.8 μl μ drugs, including PL (1 and 10 μM), E2 (0.1 μM) and/or ICI (1 μM). After each of forward and reverse primers and 7.4 l nuclease free water. The treatment for 3, 6, 9, 12, 15 days, the medium was removed, and the amplification conditions were: 50 °C for 2 min, 95 °C for 10 min, 40 cells were washed twice with phosphate-buffered saline (PBS). RIPA cycles of 95 °C for 15 sec, 60 °C for 1 min. The fluorescence signal was lysis buffer (100 μl) was added to each well and the plate was shaken recorded by Roche Light Cycler 480 Detection System and then the for 30 min at room temperature. The supernatant was collected after signal was converted into numerical values. The messenger RNA centrifuging at 12,000 × g for 15 min. ALP activity was then assayed (mRNA) levels of all genes were normalized using gapdh as internal using a commercial kit. 5 μl of sample was added to 195 μl of ALP control. These analyses were performed in duplicate for each sample working reagent and the mixture was incubated for 15 min. The optical using cells from three different cultures, and each experiment was re- density was measured at 405 nm. The total cell protein content was peated 4 times. measured using BCA method and the results were expressed in nano- mole of p-nitrophenol produced per min per mg of protein. Western blotting

Alizarin Red S staining MC3T3-E1 cells (7 × 104 cells/dish) and 3T3-L1 (8 × 104 cells/ dish) were grown to confluence in a 60 mm2 dish. MC3T3-E1 cells were MC3T3-E1 cells were plated at a density of 3 × 104 cells/well in 12- in osteogenic differentiation medium with drugs, including PL (1 and well cell culture plates in the growth medium until reaching 100% 10 μM), E2 (0.1 μM) and ICI (1 μM), for 3 days and 3T3-L1 cells were in confluence (designated as day 0), then, the cells were incubated in the adipogenic differentiation medium with PL (1 and 10 μM), E2 (0.1 μM) OD medium with drugs, including PL (1 and 10 μM), E2 (0.1 μM), ICI (1 and ICI (1 μM) for 30 min, respectively. Cells were lysed in RIPA lysis μM), G-1(1 μM) and G-15 (1 μM). On day 18, the cultures were fixed in buffer with 1% PMSF and 1% protease inhibitor cocktail. After treat- the plates with 3.7% (vol/vol) formaldehyde in PBS for 30 min and ment on ice for 30 min, cell lysates were clarified by centrifugation at stained with 1% Alizarin Red S for another 30 min. The stained samples 12,000 × g for 15 min at 4 °C to remove cell debris, and the protein were observed under digital camera and representative pictures were content was measured using the BCA protein assay kit. The total lysates taken by a dissecting microscope (Olympus, Tokyo, Japan). The amount were boiled in 2% sodium dodecyl sulfate polyacrylamide gel

4 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860 electrophoresis (SDS-PAGE) buffer. Aliquots of the lysates were sub- PL promotes osteogenesis and inhibits adipogenesis of rat-BMSCs jected to 10% SDS-PAGE (with 5% stacking gel) and transferred to a PVDF membrane. Immunoblotting was performed using antibodies. The As to dual-differentiation experiment, rat-BMSCs were cultured in ™ bands density was quantified using Image Lab (Bio-Rad Laboratories) AD medium for 1, 2 or 4 days, and then these cells were re-cultured in by calculating the average optical density in each field. Anti-β-actin was OD medium for 11, 9 or 7 days, respectively (Fig. 3A). After adipogenic used as a reference to normalize the differences in the amounts of differentiation for 8 days in rat-BMSCs, many lipid droplets were ob- protein among samples. served in the control group, while PL (10 μM) and E2 (0.1 μM) significantly inhibited lipids formation (Fig. 3B). Early adipogenic induction impaired the mineralization of rat-BMSCs at day 12 and showed a Statistical analysis time-dependent manner. Under these culture conditions, PL (10 μM)

and E2 (0.1 μM) significantly enhanced the calcium nodules formation All quantitative data were presented as means ± standard deviation compared to the control group (Fig. 3C). of experiments. Statistical comparisons were performed using the sta- tistic package for social science (SPSS) 19.0 software (IBM, Armonk, ff NY, USA). One-way analysis of variance (ANOVA) followed by Tukey PL promotes osteogenic di erentiation and mineralization of MC3T3-E1 post hoc test (multi-group comparison) was used to assess statistical cells via classical ER pathway significance at p < 0.05. As shown in Fig. 4A, PL (1 and 10 μM) and E2 (0.1 μM) could significantly promote mineralization of MC3T3-E1 cells at day 18 after Results treatment. ICI treatment inhibited mineralization of MC3T3-E1 cells

after induction, even the cells were exposed to E2 (0.1 μM) and PL (1 The effects of PL on cell toxicity and effective doses for anti-adipogenesis and 10 μM) (Fig. 4A). During the osteogenesis process, PL and E2 increased the ALP activity at early stage, reaching a plateau at day 9 and Rat-BMSCs, MC3T3-E1 and 3T3-L1 cells were treated with different then started to decrease after day 12 (Fig. 4B). During day 6 to day 12, concentrations of PL (0.001, 0.01, 0.1, 1, 10 and 20 μM) in normal ICI decreased the ALP activities with or without PL and E2 treatment growth medium for 2 days. Results showed that PL at the concentra- (Fig. 4B). During osteogenic differentiation and mineralization of tions of no more than 1 μM possessed better cytocompatibility than MC3T3-E1 cells, we measured the gene expression of osteogenic mar- higher doses (10 and 20 μM) groups (Fig. 2A), which exerted toxicity on kers at day 3, 6, 9, 12 and 15 after osteogenic stimulation. Compared to these three cells. However, when the cells were cultured in differ- non-induction group, the expression of runx2 did not change in all ex- entiation medium (Fig. 2B), all of the tested concentrations of PL perimental groups after osteogenic induction (Fig. 4C1); osterix levels in showed no toxicity to rat-BMSCs and MC3T3-E1 cells, and 0.001 to 10 control group were significantly up-regulated at day 3 and 6, while bsp μM of PL demonstrated good cytocompatibility on 3T3-L1 cells. More- in control group was significantly up-regulated along the whole pro- over, PL at 1 and 10 μM significantly promoted MC3T3-E1 cells pro- cedure (Fig. 4C2-3). PL (1 and 10 μM) and E2 (0.1 μM) treatment in- liferation in osteogenic differentiation medium. We further screened creased bsp levels at day 3, 6, 9 and 12, while ICI treatment diminished the effective dose(s) of E2 and PL in anti-adipogenesis (Fig. S1). Our the effect of PL and E2 (Fig. 4C3). The ocn expression had a similar results showed that E2 at 0.1 µM and PL at 1 or 10 µM exerted re- trend to ALP activity (Fig. 4C4). Elevated ocn level was observed in PL markable effect on inhibiting oil droplet formation of 3T3-L1 cells. (1 and 10 μM) groups at day 9, and it continued to increase and reached

According to above results, we chose E2 at 0.1 µM (as positive control) a peak value at day 12 in PL (10 μM) group, while the ocn expression in and PL at 1 and 10 µM for the following studies. E2 reached its peak value by day 15 after induction, which all showed

Fig. 2. Eﬀect of PL on toxicity of rat BMSCs, MC3T3-E1 and 3T3-L1 cells in normal growth medium (A) or diﬀerentiation medium (B). n =6.*p < 0.05, **p < 0.01 and ***p < 0.001, when compared to the normal group (A) or control group (B).

5 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Fig. 3. Effect of PL on promoting osteognesis and inhibiting adipogenesis of BMSCs. (A) Study design of differentiation experiment in BMSCs. (B1) Representative Oil red O staining images after induction for 8 days in BMSCs. Bar = 100 mRep (B2) Quantification assay of the oil droplets formation. n =3.*p < 0.05, when compared to Normal group; ##p < 0.01, when compared to AS-8d group. (C1) Representative Alizarin Red S staining images in BMSCs. Bar = 5 mm. (C2) Quantification assay of the calcium nodules. n =3.*p < 0.05, when compared to Normal group; #p < 0.05 and ##p < 0.01, when compared to OS-12d group; &&p < 0.01, when groups with PL or E2 compared to AS-1d, AS-2d and AS-4d groups without PL or E2, respectively. significant differences compared to the control group. At day 9 and day stimulated control group, and reached peak value at day 4, and then 12, ICI combined PL (1 and 10 μM) groups had significant lower ocn slowly decreased (Fig. 5B2). C/ebpα had a similar trend as pparγ expression when compared to that in the PL (1 and 10 μM) groups. (Fig. 5B1). When PL (1 and 10 μM) and E2 (0.1 μM) were added to the We also analyzed the protein level of bone morphogenetic protein-2 culture, we observed a significant down-regulation of pparγ and cebpα (BMP-2), Runx2, Osterix and BSP. Consistent to the observation at the expressions at day 4 after induction. This was also confirmed by the gene level, PL (1 and 10 μM) and E2 (0.1 μM) did not enhance these decreased in lpl and fabp4 expression in PL (1 and 10 μM) or E2 (0.1 protein expressions after osteogenic induction for 3 days (Fig. 4D), al- μM) groups (Fig. 5B3-4). The mRNA expressions of pparγ, cebpα, lpl and though ICI treatment decreased BSP in PL (10 μM) and E2 (0.1 μM) fabp4 at day 4 showed significant differences between drug groups with groups. The ELISA results showed that at day 12, PL (1 and 10 μM) and ICI treatment and without ICI treatment (Fig. 5B). We also measured

E2 (0.1 μM) treatment significantly increased OCN in the medium the fat-related protein markers adiponectin, resistin and leptin at 6 days compared to that in the control group, while ICI treatment diminished after differentiation and observed that E2 (0.1 μM) and PL (1 and 10 the effects of PL and E2 (Fig. 4E). μM) significantly decreased resistin and adiponectin level, while ICI treatment would significantly suppress the effect of E2 (0.1 μM) and PL (1 and 10 μM) on decreasing resistin (Fig. 5C). PL inhibits adipogenic differentiation and oil droplets formation of 3T3-L1 cells partly via classical ER pathway PL also works via membrane ER (GRP30) pathway After adipogenic differentiation for 6 days, many lipid droplets were observed in the control group. Lipids accumulation was significantly In this experiment, G-1, a specific GRP30 agonist, increased calcium inhibited by either PL (1 and 10 μM) or E2 (0.1 μM) (Fig. 5A). More- nodule formation, but its effect was weaker than E2. G-15, a specific over, ICI could decrease the effects of PL (1 and 10 μM) or E2 (0.1 μM) GRP30 antagonist, decreased mineralization of MC3T3-E1 cells when on inhibiting adipocytes formation. During adipocyte differentiation of compared to control group. Yet, G-15 did not completely inhibit the

3T3-L1 cells, we measured the gene expression of adipogenic markers at effect of E2 (0.1 μM) and PL (1 and 10 μM) in accelerating calcium day 2, 4 and 6 after adipogenic stimulation. Compared to the normal nodule formation (Fig. 6A). As for adipogenesis, G-1 significantly re- cells, pparγ increased significantly from day 2 in the adipogenic duced oil droplets formation and accumulation with similar effect to

6 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Fig. 4. Effect of classical estrogen receptor signaling pathway on promoting osteogenic differentiation and mineralization of PL in MC3T3-E1 cells. (A1) Calcium nodules after induction for 18 days stained by 1% Alizarin Red S and representative Alizarin Red S staining images. Bar = 5 mm and Bar = 100 mm and Bar = 100 a (A2) Quantification assay of the calcium nodules. n =3. (B) ALP activity of cells during differentiation process were analyzed, n = 4. (C) mRNA expression levels of runx2, osterix, bsp and ocn were normalized to Normal group using gapdh as a reference. n = 4. (D1) Representative western blots of total BMP-2, runx2, osterix, bsp, and β-actin after induction for 3 days in MC3T3-E1 cells. (D2) Quantification of the protein level of BMP-2, runx2, osterix and bsp using β-actin as references, n =3. (E) OCN protein level in medium after induction for 9 and 12 days, n =3.*p < 0.05, **p < 0.01, when compared to Normal group; #p < 0.05 and ##p < 0.01, when compared to Control group; $p < 0.05 and $$p < 0.01, comparison between the drug treatment with and without ICI.

that of E2 (0.1 μM) and PL (1 and 10 μM). In addition, G-15 completely Discussion abolished the effect of E2 (0.1 μM) and PL (1 and 10 μM) in inhibiting oil droplets formation in mature adipocytes (Fig. 6B). In this study, we found PL could promote osteogenesis and con- currently inhibit adipogenesis, but via different ER-dependent ways. Moreover, early suppression of adipocytes formation by PL would be Akt/GSK3β/β-catenin signaling is involved in PL inhibiting adipocyte beneficial for later calcium nodules formation of BMSCs. differentiation of 3T3-L1 cells In terms of osteogenesis, BMP-2 is a multifunctional growth factor and strongly induces bone formation by binding to BMP receptor Our results demonstrated that the adipocyte differentiation process (Wu and Lu, 2008). However, it has a complicated interaction with ER of 3T3-L1 cells might be modulated via ER activating cytoplasmic pathway during osteogenesis. E2 enhanced BMP-4-induced osteogenesis protein kinase cascades. So, we further investigated the effect of PL on in MC3T3-E1 cells, but not BMP-2 (Matsumoto et al., 2013). Instead, regulating the Akt/GSK3β/β-catenin pathway. As shown in Fig. 7, after the responsiveness of MC3T3-E1 cells to E2 was pronouncedly sup- induction for 30 min, the phosphorylation of Akt at Ser473 and pressed by the pretreatment with BMP-2 (Ishibashi, 2013). Our data phosphorylation of GSK3β at Ser9, as well as total β-catenin were de- showed that E2 and PL had no effect on BMP-2 expression, which was in creased. PL (1 and 10 μM) and E2 (0.1 μM) could significantly increase consist with previous studies (Matsumoto et al., 2013). Runx2 is an phosphorylation of Akt at Ser473 and phosphorylation of GSK3β at essential transcription factor for osteoblast differentiation. It directs Ser9, and further up-regulate total β-catenin level in 3T3-L1 cells pluripotent mesenchymal cells to the osteoblast lineage, triggers the compared to control group. expression of major bone matrix protein genes, but does not play a major role in the maintenance of the expression of osteocalcin in mature osteoblasts (Komori, 2010). Previous study found that runx2

7 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Fig. 5. Effect of classical estrogen receptor signaling pathway on inhibiting adipocyte formation of PL in 3T3-E1 cells. (A1) Representative Oil red O staining images after induction for 6 days in mature adipocytes. Bar = 100 mRep(A2) Quantification assay of the oil droplets formation. n =3. (B) mRNA expression levels of c/ebpα, pparγ, lpl and fabp4 were normalized to Normal control using gapdh as a reference, n = 4. (C1) Representative western blots of total resistin, leptin, adiponectin and β- actin after induction for 6 days in mature adipocytes. (C2) Quantification of the protein level of resistin, leptin and adiponectin using β-actin as references, n =3.*p < 0.05, **p < 0.01, when compared to Normal group; #p < 0.05 and ##p < 0.01, when compared to Control group; $p < 0.05 and $$p < 0.01, comparison between the drug treatment with and without ICI; &p < 0.05 and &&p < 0.01: comparison between the ICI treatment with and without a drug. weakly expressed in osteocalcin-positive mature osteoblasts bone formation both in vitro (Zhai et al., 2017) and in vivo (Zhai et al., (Maruyama et al., 2007). Our data also showed that runx2 decreased at 2018). Taken together, PL promoted osteogenesis via classical ER sig- day 9, while osteocalcin increased at that time. Further, E2 and PL naling pathway. promoted osteocalcin after day 9, and similar results was reported in With respect to adipogenesis, the effects of E2 and PL on inhibiting previous study, i.e., E2 enhanced osteocalcin mRNA expression in rat adipogenesis and adiponectin synthesis were partially blocked by ICI. It calvarial osteoblast-like cells (ROBs) (Rucinski et al., 2006). is reported that E2 could interact with estrogen-responsive suppressive In additional, PL increased ALP activity and calcium nodule for- element (ERSE) for LPL located at the LPL promoter and repress lpl mation and stimulated OCN expression. Effects of PL promoting os- expression (Homma et al., 2000). In addition, E2 inhibited PPARγ ac- teogenesis were eliminated by ICI exposure, while G-15 only partly tivity though enhancing peroxisome proliferator-activated receptor-γ reduced PL promoting mineralization. Similar results have been re- coactivator-1 (PGC-1) binding to nuclear ER but not to PPARγ and ul- ported by Zhai et al. (2017). They found that PL promoted ALP activity timately reducing the FABP4 expression (Tcherepanova et al., 2000). in rat calvarial osteoblast (ROB), which was blocked by ICI. It was Thus, PL inhibited adipogenesis partially by classical ER pathway. In proved PL was characterized as a full ER agonist, as it activated classical addition, PL and G-1 showed similar efficiency in inhibiting oil droplet ER pathway via induction of ER conformation and binding of the formation, while these effects were blocked by G-15 treatment. These psoralidin-ER complex to the EREs present in the promoter region of data hinted that membrane ER was also involved in PL reducing adi- estrogen-responsive genes. Molecular docking of PL showed that PL is pogenesis. able to mimic the binding interactions of E2, whereas its isopentenyl In this study, we proved that early adipogenic induction of rat- side chain at C2 seems to interfere with its binding to the ER ligand BMSCs suppressed late mineralization, and PL at 10 μM and E2 at 0.1 binding domain leading to weak binding affinities or transcriptional μM could restore the mineralization capability through inhibiting rat- activities (Liu et al., 2014). However, the experimental studies found BMSCs differentiation into adipocytes. Wnt signaling activation was that PL was proved to have a higher ability than coumestrol to promote involved in inhibiting adipogenesis and inducing osteoblastogenesis,

8 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Fig. 6. Effect of GRP30 receptor pathway on promoting mineralization and inhibiting oil droplets accumulation of PL. (A1) Alizarin Red S staining after osteogenic induction for 18 days, Bar = 5 mm and Bar = 100 mm and Bar = 100 t (A2) Quantification assay of calcium nodule, n = 3. (B1) Oil red O staining after adipogenic induction for 6 days in mature adipocytes, Bar = 100 mOil (B2) Quantification assay of oil droplets formation, n =3.**p < 0.01: when compared to Normal group; # ## ^^ & && p < 0.05 and p < 0.01: when compared to Control group; p < 0.01: when compared to E2 group; p < 0.05 and p < 0.01: comparison between the drug treatment with and without G-15. which mainly depended on the process of cytosolic β-catenin accumu- OVX group, while PL (10 mg/kg body weight/day) had no significant lation and translocation to the nucleus (Wang et al., 2013). When β- effect on uterine index (Zhai et al., 2018). It is suggested that PL only catenin translocated to the nucleus, it would bind to the TCF/LEF fa- have weak estrogen-like activities on uterus. However, we should still mily of transcription factors to regulate the expression of Wnt target keep keen on the estrogen-like effects on other organs for PL. genes, such as runx2, pparγ and c/ebpα (Kramer et al., 2010; Prestwich and MacDougald, 2007). As reported, β-catenin was involved in both Wnt/β-catenin and Akt/GSK-3β signaling pathways (Bhukhai et al., Conclusion

2012). Our study showed that E2 and PL activated Akt/GSK-3β/β-catenin signaling for inhibiting adipocyte differentiation. Taken together, PL as a unique herbal molecule was able to simultaneously promote our results provided the feasibility of the strategy on keeping bone mass osteogenesis and inhibit adipogenesis in ER-dependent way and worked by balancing the competitive relationship between osteogenesis and in difference mechanisms. PL might become a new candidate for adipogenesis of BMSC. treatment of postmenopausal osteoporosis by balancing the competitive As PL would be used in postmenopausal women, it is important to relationship between osteogenesis and adipogenesis. have no effect on breast cancer and endometrial cancer. Ideally, a drug used to treat osteoporosis should increase bone mineral density without increasing the risk of cancer. Although PL was characterized as a full ER Conflict of interest agonist, and PL at 10 μM was able to induce the maximum reporter gene expression corresponding to that of E2 at 1 nM treated cells, yet in The authors do not have any financial or personal relationships with an OVX model, E2 (25 μg/kg body weight/day) significantly increased other people or organizations that could influence the work described the uterine index (uterus weight/body weight) when compared to the in this manuscript.

9 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Fig. 7. GSK3β/β-catenin signaling is involved in inhibiting adipocyte differentiation of PL in 3T3-L1 cells. (A) Western blots images of total p-Aktser473, Akt, p- GSK3βser9, GSK3β, β-catenin and β-actin in 3T3-L1 cells after adipogenic medium treatment for 30 mins. (B1) Quantification of the ratio of p-Aktser473 and Akt in total cell lysates of 3T3-L1 cells. (B2) Quantifications of the ratio of p-GSK3βser9 and GSK3β in total cell lysates of 3T3-L1 cells. (B3) Quantification of the protein levels of β-catenin used β-actin as a reference in total cell lysates of 3T3-L1 cells. n =3.*p < 0.05, **p < 0.01, when compared to Normal group; #p < 0.05 and ##p < 0.01, when compared to Control group.

Acknowledgments Bonewald, L.F., Kneissel, M., 2010. Osteocyte wnt/beta-catenin signaling is required for normal bone homeostasis. Mol. Cell Biol. 30, 3071–3085. Kuiper, G.G.J.M., Carlsson, B., Grandien, K., Enmark, E., Haggblad, J., Nilsson, S., This work was supported by Sino-Swiss collaborative project from Gustafsson, J.A., 1997. Comparison of the ligand binding specificity and transcript Ministry of Science and Technology (2015DFG32200) and the Swiss tissue distribution of estrogen receptors alpha and beta. Endocrinology 138, National Science Foundation under the SSSTC program (156362) and 863–870. Limonard, E.J., Veldhuis-Vlug, A.G., van Dussen, L., Runge, J.H., Tanck, M.W., Endert, E., National Nature Science Foundation of China (81773964, 81501893, Heijboer, A.C., Fliers, E., Hollak, C.E., Akkerman, E.M., Bisschop, P.H., 2015. Short- 81302782 & 51573206), Shenzhen Science and Technology Innovation term effect of estrogen on human bone marrow fat. J. Bone Miner. Res. 30, Committee (JCYJ20160229195715386, JCYJ20160229195640212, 2058–2066. GJHS20160331171708747 & GJHS20170314161106706), Shenzhen Liu, X., Nam, J.W., Song, Y.S., Viswanath, A.N.I., Pae, A.N., Kil, Y.S., Kim, H.D., Park, J.H., Seo, E.K., Chang, M., 2014. Psoralidin, a coumestan analogue, as a novel potent Science and Technology Research Funding (20170502171625936). We estrogen receptor signaling molecule isolated from Psoralea corylifolia. Bioorg. Med. thank Ms. Keke Zhang from Dr. Yulong Sun's group for harvesting Chem. Lett. 24, 1403–1406. BMSCs from SD rats. Manolagas, S.C., 2000. Cellular and molecular mechanisms of postmenopausal osteoporosis. Serono. Symp. 134–146. Maruyama, Z., Yoshida, C.A., Furuichi, T., Amizuka, N., Ito, M., Fukuyama, R., Miyazaki, Supplementary materials T., Kitaura, H., Nakamura, K., Fujita, T., Kanatani, N., Moriishi, T., Yamana, K., Liu, W., Kawaguchi, H., Nakamura, K., Komori, T., 2007. Runx2 determines bone ma- turity and turnover rate in postnatal bone development and is involved in bone loss in Supplementary material associated with this article can be found, in estrogen deficiency. Dev. Dyn. 236, 1876–1890. the online version, at doi:10.1016/j.phymed.2019.152860. Matsumoto, Y., Otsuka, F., Takano-Narazaki, M., Katsuyama, T., Nakamura, E., Tsukamoto, N., Inagaki, K., Sada, K.E., Makino, H., 2013. Estrogen facilitates osteoblast differentiation by upregulating bone morphogenetic protein-4 signaling. References Steroids 78, 513–520. Muruganandan, S., Roman, A.A., Sinal, C.J., 2009. Adipocyte differentiation of bone Bhukhai, K., Suksen, K., Bhummaphan, N., Janjorn, K., Thongon, N., Tantikanlayaporn, marrow-derived mesenchymal stem cells: cross talk with the osteoblastogenic pro- – D., Piyachaturawat, P., Suksamrarn, A., Chairoungdua, A., 2012. A phytoestrogen gram. Cell Mol. Life Sci. 66, 236 253. diarylheptanoid mediates estrogen receptor/akt/glycogen synthase kinase 3 beta Ning, L.J., Zhang, Y.J., Zhang, Y., Qing, Q., Jiang, Y.L., Yang, J.L., Luo, J.C., Qin, T.W., protein-dependent activation of the wnt/beta-catenin signaling pathway. J. Biol. 2015. The utilization of decellularized tendon slices to provide an inductive micro- ff Chem. 287, 36168–36178. environment for the proliferation and tenogenic di erentiation of stem cells. – Gomaa, H.A., Evaniuk, D., Klachook, S., Wolfman, W.L., 2015. Novel regimen of hormone Biomaterials 52, 539 550. replacement therapy: efficacy and safety of transdermal estrogen plus intermittent Paccou, J., Hardouin, P., Cotten, A., Penel, G., Cortet, B., 2015. The role of bone marrow progesterone. Menopause 22, 1382–1383. fat in skeletal health: usefulness and perspectives for clinicians. J. Clin. Endocr. – Goto, H., Hozumi, A., Osaki, M., Fukushima, T., Sakamoto, K., Yonekura, A., Tomita, M., Metab. 100, 3613 3621. Furukawa, K., Shindo, H., Baba, H., 2011. Primary human bone marrow adipocytes Prestwich, T.C., MacDougald, O.A., 2007. Wnt/beta-catenin signaling in adipogenesis and – support TNF-alpha-induced osteoclast differentiation and function through RANKL metabolism. Curr. Opin. Cell Biol. 19, 612 617. expression. Cytokine 56, 662–668. Prossnitz, E.R., Arterburn, J.B., Smith, H.O., Oprea, T.I., Sklar, L.A., Hathaway, H.J., He, J.B., Chen, M.H., Lin, D.K., 2017. New insights into the tonifying kidney-yin herbs 2008. Estrogen signaling through the transmembrane G protein-coupled receptor – and formulas for the treatment of osteoporosis. Arch. Osteoporos. 12. GrPR30. Annu. Rev. Physiol. 70, 165 190. Homma, H., Kurachi, H., Nishio, Y., Takeda, T., Yamamoto, T., Adachi, K., Morishige, K., Rodriguez, J.P., Montecinos, L., Rios, S., Reyes, P., Martinez, J., 2000. Mesenchymal stem fi Ohmichi, M., Matsuzawa, Y., Murata, Y., 2000. Estrogen suppresses transcription of cells from osteoporotic patients produce a type I collagen-de cient extracellular ff – lipoprotein lipase gene - Existence of a unique estrogen response element on the matrix favoring adipogenic di erentiation. J. Cell Biochem. 79, 557 565. lipoprotein lipase promoter. J. Biol. Chem. 275, 11404–11411. Rucinski, M., Ziolkowska, A., Hochol, A., Pucher, A., Macchi, C., Belloni, A.S., Nussdorfer, ff Hughes, D.E., Dai, A.H., Tiffee, J.C., Li, H.H., Mundy, G.R., Boyce, B.F., 1996. Estrogen G.G., Malendowicz, L.K., 2006. Estradiol and resveratrol stimulating e ect on os- promotes apoptosis of murine osteoclasts mediated by TGF-beta. Nat. Med. 2, teocalcin, but not osteonectin and collagen-1alpha gene expression in primary culture – 1132–1136. of rat calvarial osteoblast-like cells. Int. J. Mol. Med. 18, 565 570. Ishibashi, O., 2013. Bone morphogenetic protein-2 desensitizes MC3T3-E1 osteoblastic Sassarini, J., Lumsden, M.A., 2015. Oestrogen replacement in postmenopausal women. – cells to estrogen through transcriptional downregulation of estrogen receptor 1. J. Age Ageing 44, 551 558. fi Bone. Metab. 20, 83–88. Schwartz, A.V., 2015. Marrow fat and bone: review of clinical ndings. Front. Kloverpris, H., Fomsgaard, A., Handley, A., Ackland, J., Sullivan, M., Goulder, P., 2010. Endocrinol. 6. Dimethyl sulfoxide (DMSO) exposure to human peripheral blood mononuclear cells Szkudelska, K., Szkudelski, T., Nogowski, L., 2002. Daidzein, coumestrol and zearalenone ff – (PBMCs) abolish T cell responses only in high concentrations and following coin- a ect lipogenesis and lipolysis in rat adipocytes. Phytomedicine 9, 338 345. cubation for more than two hours. J. Immunol. Methods 356, 70–78. Takada, I., Kouzmenko, A.P., Kato, S., 2009. Molecular switching of osteoblastogenesis Komori, T., 2010. Regulation of bone development and extracellular matrix protein genes versus adipogenesis: implications for targeted therapies. Expert Opin Ther Tar 13, – by RUNX2. Cell Tissue Res. 339, 189–195. 593 603. Kramer, I., Halleux, C., Keller, H., Pegurri, M., Gooi, J.H., Weber, P.B., Feng, J.Q., Tcherepanova, I., Puigserver, P., Norris, J.D., Spiegelman, B.M., McDonnell, D.P., 2000.

10 H.-j. Cao, et al. Phytomedicine 61 (2019) 152860

Modulation of estrogen receptor-alpha transcriptional activity by the coactivator differentiation in rat bone marrow stromal cells. J. Biomed. Mater. Res. B. 90B, PGC-1. J. Biol. Chem. 275, 16302–16308. 621–628. Wang, N., Wang, X.L., Cheng, W.X., Cao, H.J., Zhang, P., Qin, L., 2013. Puerarin promotes Zhai, Y., Li, Y., Wang, Y., Cui, J., Feng, K., Kong, X., Chen, L., 2017. Psoralidin, a pre- osteogenesis and inhibits adipogenesis in vitro. Chin. Med.-UK 8. nylated coumestan, as a novel anti-osteoporosis candidate to enhance bone formation Weinstein, R.S., Jilka, R.L., Parfitt, A.M., Manolagas, S.C., 1997. The effects of androgen of osteoblasts and decrease bone resorption of osteoclasts. Eur. J. Pharmacol. 801, deficiency on murine bone remodeling and bone mineral density are mediated via 62–71. cells of the osteoblastic lineage. Endocrinology 138, 4013–4021. Zhai, Y., Wang, Q., Li, Y., Cui, J., Feng, K., Kong, X., Xian, C.J., 2018. The higher os- Wu, C.J., Lu, H.K., 2008. Smad signal pathway in BMP-2-induced osteogenesis - a mini teoprotective activity of psoralidin in vivo than coumestrol is attributed by its pre- review. J. Dent. Sci. 3, 13–21. sence of an isopentenyl group and through activated PI3K/Akt axis. Biomed. Wu, X.T., Wang, B., Wei, J.N., 2009. Coumestrol promotes proliferation and osteoblastic Pharmacother. 102, 1015–1024.