The fruits of Mangosteen were bought directly from the fruit market in Xiamen, China. The fruits were authenticated as Garcinia Mangostana. Linn by Professor Jin-zhang Zeng (School of pharmaceutical sciences and State Key Laboratory of Cellular Stress Biology, Xiamen University). The epicarp of the fruits (3 kg) was air-dried at room temperature and pulverized, then extracted with 95% ethanol (20 L) three times (36 h per time) at room temperature to yield an extract. The latter was concentrated in vacuo to yield a residue (504 g), which was suspended in H₂O and partitioned with Ethyl acetate. The solvent was evaporated under reduced pressure to afford an Ethyl acetate extract (209 g). The Ethyl acetate extract was subjected to silica gel column chromatography (200–300 mesh) with a gradient eluant system of CH₂CL₂-CH₃OH (80:1-40:1, v/v) to yield 8 fractions. TLC (thin-layer chromatography) was employed to monitor the process. Fraction 6 was subjected to silica gel column chromatography eluted with petroleum ether–Ethyl acetate (4:1–1:1, v/v) to afford 2 subfractions 6.1–6.2. Fraction 6.2 followed by semi-preparative HPLC with 94% ethanol to give compounds 1 (67 mg), 2 (72 mg), and 3 (36 mg). The chromatographic column used in this method is the capillary column of C-18ODS (20*250 mm).

The compounds were characterized by proton nuclear magnetic resonance, mass spectrometry, and elemental analysis. The compound 3 was identified as α-Mangostin according to the chromatographic behavior and mass spectral data by comparison with those of control. Spectral data: 1H-NMR (400 MHz,CDCl3):δ 1.62(6H,S, 19-CH₃,20-CH₃),1.73(3H,s, 14-CH3), 1.77(3H,s, 15-CH₃), 3.20(2H,d,J = 6.8 Hz,H-11), 3.79(3H,S, 7-OCH₃),4.09(2H,d,J = 6.8 Hz,H-16),5.16(2H,m,H-12, H-17),6.36(1H,S,H-4),6.80(1H,S,H-5),10.80(1H,brs,C-3-OH),13.78(1H,S,C-1-OH).13C-NMR (100MHz, CDCl₃):δ 17.9(C-19), 18.2(C-14), 21.4(C-11), 25.8(C-15,20), 26.6(C-16), 62.0(7-OCH₃), 93.3((C-4), 102.8(C-5),101.6(C-9a), 108.5(C-2), 112.2(C-8a), 121.5(C-12), 123.1(C-17), 133.1(C-13), 135.6(C-18), 137.0(C-8), 142.5(C-7), 154.5(C-4a), 155.3(C-10a), 155.8(C-6), 160.6(C-1), 161.6(C-3), 182.0(C-9).

The MCF-7, SKBR-3, MDA-MB-231, and HEK-293 T cells were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China), and grown in DMEM (dulbecco’s modified eagle medium) supplemented with 10% fetal bovine serum (PAN Biotech, SA) at 37°C in an incubator containing 5% CO2. The pGAL4-RXRα-LBD plasmid was obtained by inserting the RXRα ligand-binding domain (LBD) cDNA sequence (amino acids 198–462) in-frame with the GAL4 DBD coding sequence in the pBind vector.

CCND1 and RXRα siRNA was purchased from DHARMACON and was transfected into cells using RNAi- MAX reagent.

Three breast cancer cell lines were seeded at a density of 5–10 × 10⁴ cells/well in 96-well plates containing 100 μL medium (DMEM containing 10% FBS) for 18 h. Thereafter, the culture medium was removed and replaced with 100 μL medium (serum-free DMEM) containing various concentrations (40, 20, 10, 5, 2.5, and 1.2 μM) of α-Mangostin or 0.1% (v/v) dimethyl sulfoxide (DMSO) solvent only (control). The cells were grown for another 48 h. Subsequently, 10 μL MTT solution (5 g/L) was added to each well and the culture plates were incubated for 4 h to allow formazan formation. The culture medium was then removed, the formazan was solubilized by the addition of 150 μL DMSO, and the absorbance at 560 nm (A560) was measured with a microplate reader to calculate cell viability (%). The IC50(50% inhibitive concentration) value of the compound was calculated by Graphpad Prism 6.0.

The MDA-MB-231 cells were cultured at a density of 4 × 10⁵ cells/well in 6-well plates containing 1,600 μL medium for overnight culture. Then, the cells were treated with various concentrations of α-Mangostin, meanwhile, 5 μM concentration of paclitaxel acted as a control. The cells were harvested at the indicated times by trypsinization with 0.05% (w/v) trypsin in 0.5 mM ethylenediaminetetraacetic acid solution. The cells were then washed twice with cold phosphate-buffered saline (PBS), using centrifugation at 1,000 × g for 5 min at 4°C to harvest the cells each time. For apoptosis detection, the cell pellets were resuspended in 50 μL binding buffer (10 mM HEPES, pH 7.4, 140 mM NaCl, and 2.5 mM CaCl₂) and stained with 5 μL annexin V-Alexa Fluor 488 and 5 μL PI for 30 min at room temperature in the dark. For the cell cycle study, the cell pellets were fixed in 200 μL cold 70% (v/v) ethanol at −20°C overnight, harvested, and washed as described above. The washed cell pellet was then suspended in 250 μL PBS containing 0.1 g/L RNase A and incubated at 37°C for 30 min. Thereafter, it was washed as described above, resuspended in staining buffer (12.5 μL PI (1 g/L in PBS)), and incubated at room temperature for 30 min in the dark. The samples were then analyzed by flow cytometry on an FC 500 MPL cytometer that recorded 10,000 events per sample. The experiment was performed in triplicate.

293 T cells were grown in DMEM supplemented with 10% FBS. For dual-luciferase reporter assays, cells were seeded at a concentration of 5 × 10⁴ cells per well in 24-well plates and transfected with pGL5 luciferase reporter vector (40 ng/well) and pGAL4-RXRα-LBD expression vector (40 ng/well). Cells were then incubated with various concentrations of compounds for 12 h. Luciferase activities were measured using the Dual-Luciferase Assay System Kit.

Binding experiments were carried out using Biacore S200 SPR sensors with control software version 3.0 and Sensor Chip CM5 (carboxymethylated dextran surface). All assays were carried out at 25°C. For the pre-binding experiment, pure RXRα-LBD protein was dissolved with acetate buffers with different pH values (pH4.0, pH4.5, pH5.0, and pH5.5) and flowed through the surface of the chip at a rate of 5 μL/min. Then, the chip surface was activated following a standard EDC/NHS protocol with Biacore PBS-EP buffer used as the running buffer. RXRα-LBD protein at a concentration of 0.4 g/L in 10 mM phosphate buffer pH 5.0 was then injected for 12 min followed by a 7- min injection of 1 M ethanolamine to inactivate residual active groups. Typically, approximately 2000 RU receptor protein was immobilized per flow cell. By reference to the coupling steps of RXRα-LBD protein, the protein was coupled to the chip. For the preparation of the α-Mangostin solution, the insoluble residue was pelleted by centrifugation and discarded. α-Mangostin was injected into the protein channel and blank channel at seven concentrations (0.625 nM, 1.25 nM, 2.5 nM, 5 nM, 10 nM, 20 nM, and 40 nM). The supernatant (200 μL) was injected at a flow rate of 20 μL/min. The protein binding period was set to 3 min, and the dissociation period was set to 300 s. The chip was regenerated with glycine-HCl (pH 2.5, 10 mM).

Cell lysates were prepared by lysing cells with lysis buffer (RIPA) containing 50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Na-deoxycholate, 1% Triton X-100, pH 7.4 with a cocktail of proteinase inhibitors on ice for 30 min. The cells were fixed with 80 μL/well cell lysates in 12-well plates. Equal amounts of the lysates were electrophoresed on 8% SDS-PAGE gels and transferred onto polyvinylidene difluoride membranes. The membranes were blocked with 5% nonfat milk in Tris-Buffered Saline and Tween 20 [50 mmol/L Tris–HCl (pH 7.4), 150 mmol/L NaCl, and 0.1% Tween 20] for 1 h, incubated with various primary antibodies for 48 h and detected with either anti-rabbit (1:5,000) or anti-mouse (1:5,000) secondary antibodies for 1 h. The final immunoreactive products were detected using enhanced chemiluminescence (ECL).

The three-dimensional structure of human RXRα (PDB code: 3A9E) was downloaded from the Protein Data Bank. Protein preparation was performed by Chimera and autodock software and was saved as a locked PDB file for docking. Schrodinger software glide module was used to calculate the molecular docking of α-Mangostin and RXRα, list the possible binding sites, and determine the docking energy.

The quantitative data were obtained by three or more repeated experiments. Data were analyzed using an analysis of variance or Student’s t-test and were presented as the mean and standard error of the mean (±SEM).

We first assessed the anti-proliferative effects of α-Mangostin in breast cancer cells. MCF-7, MDA-MB-231, and SKBR-3 cells were treated with increasing concentrations of α-Mangostin (0, 1.2, 2.5, 5, 10, 20, and 40 μM) for 48 h α-Mangostin significantly inhibited breast cancer cell proliferation in a dose-dependent manner. (Figure 1A). The IC50 values of α-Mangostin against MCF-7, MDA-MB-231, and SKBR-3 cells were 9.69, 11.37, and 7.46 μM, respectively. We used Annexin-V and PI double staining to detect the cytotoxicity of α-Mangostin by flow cytometry. MDA-MB-231 cells were treated with α-Mangostin using paclitaxel as a positive control. α-Mangostin could induce apoptosis of MDA-MB-231 cells in a concentration-dependent manner (Figure 1B).

To further verify the apoptotic effects of α-Mangostin in breast cancer cells at the molecular level, we detected the cleavage of PARP in MCF-7, SKBR-3, MDA-MB-231 cells by western blotting. The PARP protein in these cell lines had been significantly cleaved, and the cleavage increased with the increase in α-Mangostin concentration over 20 h treatment (Figure 1C). Meanwhile, the cleavage of PARP proteins in the MCF-7 cells increased in a time-dependent manner through treatment with 10 μM α-Mangostin.

Because of potential influence on suppressing the migration and invasion of TNBC Cells in Vitro by α-Mangostin, we chose MDA-MB-231 to perform further experiments. Western blotting was used to detect cleaved caspase-3 and BCL-2 protein family expression in MDA-MB-231 cells (Figure 2). Endogenous cleaved caspase-3 mostly related to BCL-2 protein family. The expression of BAD and BAX was affected by α-Mangostin. A graded decrease in BAD expression was detected following treatment of cells with the concentration gradient of α-Mangostin for 24 h, while BAX expression increased (Figure 2B).

The PI3K/Akt signaling pathway is one of the most important survival pathways in cells and has a very close relationship with tumor development (Zhang et al., 2018). We further explored whether α-Mangostin induces apoptosis through the PI3K/Akt signaling pathway. The expression of Akt was almost unchanged with the increase in α-Mangostin concentration, while the expression of P-Akt decreased in turn (Figure 2C). After treatment with 10 μM α-Mangostin, the expression of P-Akt (activated Akt) protein in cells was significantly reduced. P-Akt was significantly suppressed compared with that of the control group. To further verify whether the induction of apoptosis of breast cancer cells by α-Mangostin depends on the PI3K/Akt signaling pathway, we added PI3K/Akt inhibitor LY294002 (20 μM) to the medium of MDA-MB-231 cells for 6 h. Then the cells were treated with α-Mangostin for 20 h. The expression of P-Akt protein in cells was then detected. LY294002 inhibited the activation of Akt (Figure 2D) and enhanced the inhibitory effect of α-Mangostin on P-Akt. Our laboratory constructed a plasmid expressing continuously activated Akt (CA-Akt) by generating a point mutation in a critical region. We transfected the plasmid into MDA-MB-231 cells. After 24 h, the cells were treated with 10 μM α-Mangostin. We observed many apoptotic cells in the control group, while the apoptosis of MDA-MB-231 cells expressing CA-Akt was much less. Western blotting demonstrated the cleavage of caspase-3 was reduced by continuous activation of Akt (Figure 2D).

α-Mangostin belongs to the xanthone group of compounds (Chen et al., 2018). Its structural formula is shown in Figure 3A. In the current study, molecular docking was further examined to explore the binding of RXRα with α-Mangostin. The aligned structures of the original X-ray and docking are shown in Figure 3B. α-Mangostin mainly depends on the hydrogen bond and π-π interaction with RXRα amino acid residues. The value of the docking energy between RXRα and α-Mangostin is −10.235 kcal/mol. The results showed that the docking effect was good and indicated that α-Mangostin may have a strong binding effect on RXRα. The binding energy values are related to the interaction between the compounds and amino acid residues of target proteins.

RXRα is usually in an inactivated state as a transcription factor bound with a repressor. If ligands combine with RXRα and block the repressor, RXRα can be bound with the specific sequence in the target promoter and promote the expression of the target gene. The Dual-luciferase assays shows the results (Figure 3C) of the competitive binding of α-Mangostin to the RXRα ligand-binding domain. The natural ligand 9-cis-RA could significantly activate the reporting gene system, but this activity was inhibited by different concentrations of α-Mangostin (2, 5, and 15 μM).

The SPR measurements show that the gradient concentrations of α-Mangostin (0.625, 1.25, 2.5, 5, 10, 20, and 40 μM) flowed through the chip channel. As can be seen from the curve of α-Mangostin binding with RXRα (Figure 3D), the signal was obvious. The results showed that the KD value was 3.897*10^–5 M and thus the binding affinity of RXRα for α-Mangostin was expecting. Moreover, the dissociation process between α-Mangostin and RXRα was slow and occurred via another process other than a common non-covalent combination, indicating that the binding of α-Mangostin and RXRα may be mainly covalent.

Based on the above experiments, we demonstrated that α-Mangostin can target binding to RXRα/tRXRα and can also induce the apoptosis of breast cancer cells through the Akt pathway. However, it is unclear as to what role RXRα/tRXRα has in the regulation of PI3K/Akt signaling. To explore this, we used siRNA technology to knockdown RXRα/truncated RXRα (Figure 4A), and then detected the effect of α-Mangostin on Akt by western blotting (Figure 4B). The results show treatment with α-Mangostin suppressed the phosphorylation of Akt and activated caspase-3 after knockdown of RXRα/truncated RXRα (Figure 4C).

Our studies have shown that α-Mangostin can significantly inhibit the proliferation of human breast cancer cells. We know that inhibition of cell proliferation is usually closely related to blockade of the cell cycle. Therefore, to comprehensively explore the molecular mechanism of the inhibitory effect of α-Mangostin, we analyzed its effect on the cell cycle of MDA-MB-231. In the flow cytometry analysis, it was found that the proportion of S phase cells increased significantly after treatment for 6 h with α-Mangostin (5 μM). and the proportion of cells in G2/M also increased after treatment for 12 h with α-Mangostin (5 μM). Meanwhile, the proportion of MDA-MB-231 cells in G0/G1 showed a decreasing trend after treatment with 5 μM α-Mangostin. This suggests that α-Mangostin can arrest MDA-MB-231 cells in S and G0/G1 phase (Figure 5A).

Cyclin D1 is a regulator of cell cycle protein-dependent kinase CDKs, which have an important effect on the blockade of the cell cycle (Body et al., 2017). The effects of siCCND1 on inhibiting MDA-MB-231 cell proliferation and blockade of the cell cycle (Figure 5B). We treated the MDA-MB-231 cells with a concentration gradient of α-Mangostin for 24 h, and then the expression of cyclin D1 was detected by western blotting. Cyclin D1 expression decreased with increased α-Mangostin concentration (Figure 5C). siRNA knockdown of RXRα resulted in obviously reduced expression of cyclin D1 (Figure 5D), indicating that RXRα was able to regulate cyclin D1. Therefore, we speculate that the blockade of the cell cycle by α-Mangostin depends on the RXRα/cyclin D1 signaling pathway.

To examine the anticancer effect of α-Mangostin on TNBC cells, we treated the highly aggressive MDA-MB-231 cells with α-Mangostin in different concentrations (2, 5, and 15 mM, respectively). Then we used wound-healing and transwell assay to determine whether α-Mangostin had the potential to inhibit breast cancer cell migration and invasion. α-Mangostin could concentration dependently slow down the wound healing process comparing to the control group (Figures 6A similar effect was observed in transwell assay, the invaded cells were decreased in α-Mangostin -treated groups when compared with the control group (Figure 6B). All these results suggested that α-Mangostin had the anticancer capability through influencing the proliferation, migration, and invasion of TNBC cells.

Clinical studies show high expression of Cyclin D1 are often found in aggressive breast cancer, especially TNBC (Figure 7A). The clinical samples were collected from The First Affiliated Hospital of Gannan Medical University.) We speculate that cyclin D1 play a major role to promote tumor cell invasion and metastasis. Therefore, the MDA-MB-231 cells were used to verify the effect of knocking down Cyclin D1 on cell migration and invasion. The results of the Wound Healing Assay, the Transwell Assay and Western Blot Analysis all showed the migrated ability and invasive ability of MDA-MB-231 cells decline after being treated with si- Cyclin D1(Figure 7C,D).