Alantolactone inhibits cervical cancer progression by downregulating BMI1

Cervical cancer is the second most common cancer in women. Despite advances in cervical cancer therapy, tumor recurrence and metastasis remain the leading causes of mortality. High expression of BMI1 is significantly associated with poor tumor differentiation, high clinical grade, and poor prognosis of cervical cancer, and is an independent prognostic factor in cervical carcinoma. Alantolactone (AL), a sesquiterpene lactone, exhibits potent anti-inflammatory and anticancer activities. In this paper, we investigated the mechanism of AL in reducing the proliferation, migration, and invasion of HeLa and SiHa cervical cancer cells as well as its promotion of mitochondrial damage and autophagy. BMI1 silencing decreased epithelial-mesenchymal transformation-associated proteins and increased autophagy-associated proteins in HeLa cells. These effects were reversed by overexpression of BMI1 in HeLa cells. Thus, BMI1 expression is positively correlated with invasion and negatively correlated with autophagy in HeLa cells. Importantly, AL decreased the weight, volume, and BMI1 expression in HeLa xenograft tumors. Furthermore, the structure of BMI1 and target interaction of AL were virtually screened using the molecular docking program Autodock Vina; AL decreased the expression of N-cadherin, vimentin, and P62 and increased the expression of LC3B and Beclin-1 in xenograft tumors. Finally, expression of BMI1 increased the phosphorylation of STAT3, which is important for cell proliferation, survival, migration, and invasion. Therefore, we suggest that AL plays a pivotal role in inhibiting BMI1 in the tumorigenesis of cervical cancer and is a potential therapeutic agent for cervical cancer.


AL inhibits migration and invasion in vitro.
To examine the effects of AL on migration and invasion of cervical cancer cells, we treated HeLa cells with AL and determined the migration and invasion ability using 'wound healing' and transwell assays. The results of the wound healing assay showed that treatment with AL resulted in a decrease in the width of scratches among HeLa cell layers in a dose-and time-dependent manner ( Fig. 2A). The results of transwell assay showed that treatment with AL resulted in a decrease in the number of HeLa cells on the lower and upper chambers of the filters (Fig. 2B). Epithelial-mesenchymal transition (EMT) is closely related to the occurrence, invasion, and metastasis of tumors. To examine the effects of AL on EMT and ECM degradation enzymes of cervical cancer cells, we treated HeLa and SiHa cells with AL and determined the expression of EMT molecular markers (N-cadherin, β-catenin, and vimentin) and ECM degradation enzymes MMP-2, MMP-3, MMP-9, and MMP-13, which indicate invasion potential of cervical cancer cells. The results showed that treatment with AL resulted in a decrease in the N-Cadherin, Vimentin, MMP-3, and MMP-9 protein levels and MMP-2, MMP-3, MMP-9, and MMP-13 mRNA levels in HeLa cells in a dose-dependent manner (Fig. 2C,E). AL (5 μM) had the same effects in SiHa cells (Fig. 2D,F).

AL promotes mitochondrial damage, induces autophagy and inhibits BMI-1. HeLa and SiHa
cells were treated with different concentrations of AL for 24 h, stained with JC-1, and then measured by flow cytometry and fluorescence microscope. JC-1 exists in the form of polymer in the mitochondria of normal cells, revealing bright red fluorescence, while the green fluorescence in cells is very weak. On decreasing the mitochondrial membrane potential by AL treatment, JC-1 could not be stored in the mitochondrial matrix in the form of a polymer. At this time, the red fluorescence intensity in the mitochondria was significantly reduced, while the green fluorescence in the cytoplasm was enhanced (Fig. S2A,B). Flow cytometry detection revealed that the percentage of green fluorescent cells increased significantly (Fig. 3A). Mito-Tracker staining was used to visualize mitochondrial activity in HeLa and SiHa cells. DAPI staining was used to visualize nuclei (blue). The number of normally functioning mitochondria decreased significantly when treated with AL in a dosedependent manner (Fig. 3B).
Autophagy is a form of programmed cell death, which is different from cell necrosis, pyrolysis and apoptosis in terms of formation mechanism and morphology. Autophagy is distinguished by the localization of the microtubule-associated protein 1 light chain 3 (LC3) protein in the autophagosome membranes in the cytoplasm and cell organelles. The activation of polyadenosine diphosphate ribose polymerase (PAPR) and caspases are rarely found in autophagy. We detected the autophagosomes labeled with Ad-GFP-LC3B and mitochondria labeled with Mito-Tracker Red CMXRos increased the number of autophagosomes, with decrease in mitochondria (Fig. 3C), which confirmed that autophagy was activated by AL-mediated mitochondrial damage. We detected the autophagy protein P62 labeled with Ad-mCherry-P62 to confirm that autophagy was activated by AL-inhibiting BMI1. The expression of BMI1 was determined using immunocytochemistry. Also, AL reduced the expression of P62 and BMI1 in a dose-dependent manner (Fig. 3D). Western blot analysis showed that AL increased the level of Beclin-1 and LC3IIB, and decreased the level of Bcl-2, P62, and BMI1 (Fig. 3E,F).

Role of decreased BMI1 expression in inducing autophagy in cervical cancer cells and inhibiting EMT.
BMI1 is associated with self-renewal of cancer stem cells and sensitivity of tumors to chemoradiotherapy. Reducing expression of BMI1 promotes apoptosis and/or aging of tumor cells, while enhancing sensitivity of tumor cells to chemoradiotherapy 21 . Inhibiting expression of BMI1 can promote autophagic death in tumor cells 20 . To verify the effects of BMI1 in cervical cancer cells and analyze the relationship between this gene and AL-induced autophagy, BMI1 knockdown hairpin plasmid and overexpression plasmid were designed and used to transfect HeLa cells for 24 h. Then, western blot was used to detect BMI1, autophagy (Beclin-1, LC3IIB, Bcl-2, and P62) and EMT (N-cadherin, β-catenin, and vimentin) marker proteins as well as P-STAT3 signaling pathway. The results showed that both knockdown and overexpression of BMI1 gene were successful (Fig. S3). Down-regulation of BMI1 expression promoted high expression of autophagy-related genes Bec- www.nature.com/scientificreports/ lin-1 and LC3IIB, decreased the expression of substrate P62 of Bcl-2 and LC3IIB, decreased the expression of EMT-related proteins, and inhibited P-STAT3 signaling pathway (Fig. 4A). Overexpression of BMI1 showed that indicators were reversed (Fig. 4B). To detect the induction of autophagy by decreasing BMI1 expression and its association with mitochondrial damage, BMI1 gene was knocked down, the HeLa and SiHa cells were infected with Ad-GFP-LC3B for 24 h, and then mitochondria were stained with Mito-Tracker Red CMXRos. The results showed that BMI1 gene knockdown induced autophagosome formation accompanied by a reduction in mitochondria (Fig. 4C). After the expression of BMI1 gene was suppressed, HeLa and SiHa cells were infected with Ad-mcherry-P62 for 48 h, and fluorescence microscopy was used to detect the expression of P62. It was found that BMI1 gene knockdown induced a decrease in P62 expression, indicating that the occurrence of autophagy negatively correlated with BMI1 expression (Fig. 4D).

AL suppressed the tumorigenicity of HeLa cells in vivo.
To examine the effects of AL on the growth of cervical cancer tissues in vivo, we treated a Balb/C female nude mice model of cervical cancer derived from HeLa cells with AL or saline, and examined the tumor tissues, and measured the weight and volume of tumor mass. The results showed that the weight (Fig. 5C) and volume (Fig. 5A,B) of tumors in the AL treatment group were significantly less than those of the saline treatment group. The abscissa in Fig. 5B shows the process of the animal experiment. The body weight of the experimental mice did not change significantly during the treatment (Fig. 5D). At the same time, H&E detection showed that necrosis occurred in tumor tissues of the AL group (Fig. 5E), whereas the morphology of liver and kidney tissues was normal and AL did not exhibit toxicity to liver and kidney tissues (Fig. 5F). The results suggest that AL significantly inhibited the growth of cervical cancer cells in dose-dependent manner in nude mice.

Effect of AL induces autophagy and inhibits epithelial-mesenchymal transition by inhibiting BMI1 protein in vivo.
First, we found that AL could also inhibit BMI-1 expression in vivo (Fig. 6A,B) using immunohistochemistry and western blot. In silico molecular docking revealed that alantolactone was predicted to have a stronger binding affinity with 5FR6 (− 6.1 kcal/mol) than 3RPG (− 5.9 kcal/mol). The strongest binding site between alantolactone and BMI1 is shown in Fig. 6C,D. Therefore, we assume that AL inhibited EMT in cervical cancer cells to promote autophagy by inhibiting BMI-1 expression. Further detection revealed that the   www.nature.com/scientificreports/ expressions of proteins related to EMT and extracellular matrix degradation were also inhibited (Fig. 6E,F). The IHC (Fig. 6G) results showed that the protein level of Beclin-1 and LC3B increased, whereas the protein level of P62 decreased in the tumor tissues grown from HeLa cells, in the AL treatment groups compared with the control groups. We used autophagy inhibitor Ba to verify that AL promoted autophagic cell death by inhibiting BMI-1 expression (Fig. 6H).

Discussion
High expression of BMI1 is significantly associated with poor tumor differentiation, high clinical grade, lymph node metastasis, and poor prognosis of cervical cancer [22][23][24] , and is an independent prognostic factor in cervical carcinoma [23][24][25] . Aberrantly elevated BMI1 promotes cervical tumorigenicity and tumor sphere formation via enhanced transcriptional regulation of Sox2 genes as a potential oncogenic factor that participates in the carcinogenesis of cervical carcinomas 22 . The increase in the ability of tumor cells to form spheres indicates that stem cells or mesenchymal level is enhanced, and the BMI1 gene plays an important role in maintaining self-renewal www.nature.com/scientificreports/ of tumor stem cells 21,26,27 and has a pivotal role in chemotherapy sensitivity, tumor resistance, and tumor recurrence. Epithelial-mesenchymal transition (EMT) is a key developmental process that is often activated during tumorigenesis 28,29 . EMT is related to cell stemness 30,31 . BMI1 and TWIST1 function together to promote tumor dedifferentiation, metastasis, and development of drug resistance 32 . As a potential anti-inflammatory target, BMI1 regulates invasion and EMT of colorectal cancer cells via TLR4/MD-2 MyD88 complex-mediated NF-κB signaling pathway 33 .
Evidence from in vitro studies demonstrated that the main absorption mechanism of AL was passive diffusion and demonstrated good intestinal absorption 34 . The study showed poor absorption of AL in vivo 35 . High body clearance and low oral bioavailability, the highest concentration was achieved in the small intestine and feces clearance was shown to be the dominant elimination pathway of the lactones 36 . But AL has a rapid onset and does not cause significant damage to normal animal tissues and organs [35][36][37] . AL possesses multiple pharmacological activities, and its anti-tumor activity is highly attractive [38][39][40] . However, the pivotal molecules targeted by AL remain unclear. AL induces apoptosis and cell cycle arrest with low cytotoxicity and exerts an inhibitory effect on tumor cells 41 . AL selectively inhibits activation of STAT3 and exhibits potent anticancer activity in MDA-MB-231 cells 42 . By inhibiting NF-kB and its downstream target proteins, AL induces apoptosis in a dosedependent manner and selectively ablates acute myeloid leukemia stem and progenitor cells 43 . Isoalantolactone, a Does AL exert its inhibitory effects on cervical cancer through BMI1/SATA3 signaling axis? Our study indicated that AL could significantly inhibit growth of cervical cancer cells, promote apoptosis ( Fig. 1), inhibit cell migration and invasion (Fig. 2), and significantly inhibit expression of genes associated with EMT (N-cadherin, vimentin, and β-catenin) and extracellular matrix degradation (MMP-2, MMP-9, MMP-3, and MMP-13). The results of Caspase-3 activity detection and apoptosis detection with flow cytometry showed that AL did not activate Caspase-3 when promoting apoptosis in HeLa and SiHa cells (Fig.S1). However, the effects of AL on cell mitochondrial membrane potential were obvious (Figs. S2 and 3A), and the number of mitochondria also decreased, as the concentration of AL increased (Fig. 3B).
Does AL exert its inhibitory effects on cervical cancer cells by inducing autophagic cell death? The detection of autophagosomes and mitochondria showed that with the increase of AL concentration, the number of autophagosomes increased significantly while that of mitochondria decreased significantly in HeLa cells. Western blot detection showed that AL significantly promoted upregulation of expression of autophagy-related proteins Beclin-1 and LC3B and downregulation of P62 expression, inhibited expression of BMI1 protein (Fig. 3C-F). Knocking down BMI1 gene increased the number of autophagosomes and decreased the number of mitochondria (Fig. 4C), while significantly up-regulating expressions of autophagy-related genes Beclin-1 and LC3B, downregulating P62 expression, and down-regulating expressions of EMT-related proteins. By BMI1 gene overexpression, the detection of the above indicators showed opposite results (Fig. 4A,B,D). It should be noted that the expression level of p-STAT3 Tyr705 by BMI1 gene knockdown significantly decreased, and that of p-STAT3 Tyr705 by BMI1 gene overexpression significantly increased (Fig. 4A,B). It showed that BMI1 gene plays an important role in cervical cancer progression and closely correlates with autophagy of cervical cancer cells. AL may target BMI1 to induce autophagic cell death and exert potent inhibitory effects on cervical cancer. Through molecular docking simulation, it was found that AL could directly bind with BMI1 (Fig. 6C,D) 36 . Considering the pharmacokinetic characteristics and administration mode of AL, 15 mg/kg and 30 mg/kg of AL were used to treat subcutaneous xenograft tumors of HeLa cells in this study. The results revealed that AL exerted significant inhibitory effects on cervical cancer cells in vivo in a dose-dependent manner (Fig. 5A-C). Pathology section examination showed that AL group tumor tissues exhibited necrosis (Fig. 5E) whereas there was no damage to internal organs (Fig. 5F), indicating that AL is an effective cervical cancer inhibitor with low toxicity. Immunohistochemical results of tumor tissues also verified the inhibitory effects of AL on BMI1 in cervical cancer tissues (Fig. 6A,B), and its regulating effects on EMT (Fig. 6E,F) and autophagy (Fig. 6G). Previous studies have confirmed that necrosis is a new mode of caspase-independent cell death that is different from apoptosis. It relies on activation of RIPK1 and RIPK3 48 . Once active, RIPK3 then phosphorylates MLKL, which is the final stage of necrosis 49 . In ovarian cancer cells, inhibition of BMI1 gene can induce activation of RIPK1-RIPK3 complex, which phosphorylates its downstream substrate MLKL and enhances necrosis 20 . Our results confirmed that AL promoted upregulation of RIPK3 expression in cervical cancer HeLa cells and enhanced phosphorylation level of MLKL, which was reversed in the presence of bafilomycin A1 (BafA1) (Fig. 6H). BafA1 inhibited the binding of autophagosomes  www.nature.com/scientificreports/ to lysosomes. Therefore, BafA1 had no effects on LC3B expression but inhibited P62 degradation. It showed that AL could induce autophagic death in cervical cancer cells by inhibiting BMI1 expression. In summary, the inhibitory effect of AL on cervical cancer may be based on multiple factors. It was observed that AL has a significant effect on BMI1 inhibition, and BMI1 positively promotes the proliferation and metastasis of tumor cells. This may be one of the anti-tumor mechanisms of AL (Fig. 7). Therefore, AL is a potential natural chemotherapeutic compound for treating cervical cancer and its anti-tumor mechanism deserves further investigation.

Methods and materials
Chemicals and reagents. AL (purity > 98%) was purchased from Tauto   www.nature.com/scientificreports/ Determination of mitochondrial membrane potential using fluorescence microscope and flow cytometry. Cell mitochondrial membrane potential was measured using the Mitochondrial Membrane Potential Assay Kit (KeyGEN, China). JC-1 is a cationic lipid fluorescent dye that can selectively enter mitochondria. JC-1 forms a multimer in normal mitochondria and emits red fluorescence. When the mitochondrial transmembrane potential is depolarized, JC-1 is released from the mitochondria to the cytoplasm, the intensity of red light is weakened, and emits a green fluorescence. The cells were stained with JC-1 for 15 min after intervention, following the manufacturer's instructions. Fluorescence was determined by using the fluorescence microscope and flow cytometry microscopy. Immunohistochemistry staining of cancer tissues. Tumor tissues were fixed in 4% polyformaldehyde, and paraffin-embedded and sectioned into 5 μm slices. The expression of P-STAT3, BMI1, N-cadherin, Vimentin, MMP-3, MMP-9, LC3B, and P62 in tumor tissues was examined using immunohistochemistry staining. Brown-yellow staining was considered positive. The stained section was examined under a microscope and four fields were selected in the areas with clear cell staining and good background. The view was recorded in images and analyzed using Image J software (National Institutes of Health). The relative expression of P-STAT3, BMI1, N-cadherin, Vimentin, MMP-3, MMP-9, LC3B, and P62 was calculated based on the positively stained area and the total area in the views.

Molecular docking.
We performed in silico prediction of BMI1 and alantolactone binding affinity by using These two molecules represent the whole structure of BMI1. We obtained alantolactone structure (CID:72724) from the PubChem database 61 . The ligand and receptors were processed using vina tutorial in the AutoDock Tools 62 . As this was blind docking, we set the grid box to be large enough to surround the whole molecule. Partial experimental methods are elaborated in supplementary materials 1.
Statistical analysis. All data were analyzed using GraphPad Prism 7.0 (GraphPad Software, San Diego, CA) software and presented as the mean ± standard deviation. Values represent the mean ± SD from one representative experiment of three independent experiments, each performed in triplicate. Data was analyzed using the unpaired Student's t-test. Differences among test groups were analyzed by ANOVA. P values < 0.05 were considered statistically significant.