Environmental‐relevant bisphenol A exposure promotes ovarian cancer stemness by regulating microRNA biogenesis

Abstract Bisphenol A (BPA) is a ubiquitous environmental xenobiotic impacting millions of people worldwide. BPA has long been proposed to promote ovarian carcinogenesis, but the detrimental mechanistic target remains unclear. Cancer stem cells (CSCs) are considered as the trigger of tumour initiation and progression. Here, we show for the first time that nanomolar (environmentally relevant) concentration of BPA can markedly increase the formation and expansion of ovarian CSCs concomitant. This effect is observed in both oestrogen receptor (ER)‐positive and ER‐defective ovarian cancer cells, suggesting that is independent of the classical ERs. Rather, the signal is mediated through alternative ER G‐protein‐coupled receptor 30 (GPR30), but not oestrogen‐related receptor α and γ. Moreover, we report a novel role of BPA in the regulation of Exportin‐5 that led to dysregulation of microRNA biogenesis through miR‐21. The use of GPR30 siRNA or antagonist to inhibit GPR30 expression or activity, respectively, resulted in significant inhibition of ovarian CSCs. Similarly, the CSCs phenotype can be reversed by expression of Exportin‐5 siRNA. These results identify for the first time non‐classical ER and microRNA dysregulation as novel mediators of low, physiological levels of BPA function in CSCs that may underlie its significant tumour‐promoting properties in ovarian cancer.

exposure can result in numerous ovarian abnormalities and alter oocyte meiotic events, even at doses below the reference dose. 6 There also exists a positive correlation between BPA exposure and increased incidence of polycystic ovarian syndrome and infertility. 7,8 BPA alters gene expression inducing the growth and dissemination of cultured ovarian cancer cells. [9][10][11] However, how the low BPA exposure exerts its carcinogenic action remains unclear.
Given its structural similarity to oestrogen, BPA's oestrogenic action has long been suspected. However, BPA binds to ERα at a 10,000-fold lower affinity than that of oestrogen. 12 There is also evidence that BPA can act through nonclassical ERs, including oestrogen-related receptors (ERRs) 13 and G-protein-coupled receptor 30 (GPR30). 14 For example, ERRs are frequently expressed in ovarian carcinomas and associated with poor overall survival in serous ovarian cancer patients. 15 It is reported that activated GPR30 can enhance ER-negative breast cancer cell proliferation. 16,17 It is becoming clear that a minute population of cancer stem cells (CSCs) in the bulk tumour is crucial for carcinogenesis. CSCs possess the ability to self-renew and differentiate to give rise to other subpopulations. These cells also display a greatly enhanced tumourinitiating capability and are responsible for tumour recurrence and metastasis. 18 Therefore, CSC survival selection advantage could be of critical importance to all stages of carcinogenesis. There is a need to better understand the environmental signals and regulatory pathways contributing to CSC formation or expansion to circumvent tumour development and progression.
In this study, we show for the first time that the environmentalrelevant, low doses of BPA is already adequate to accelerate ovarian CSC formation and expansion. We also provide evidence that this is via non-classical ER GPR30 through miRNA biogenesis dysregulation.
Transfection was achieved at 20 nmol/L working concentration using siLentFect (Bio-Rad) according to the instructions. In short, Opti-MEM (Gibco) and siLenFect (Bio-Rad) mixture and Opti-MEM and siRNA mixture were prepared, and incubated for 5 min. The diluted siRNA was added to the diluted siLenFect. Mix by pipetting and incubated for 20 min before dropping into well with replaced medium. Anti-miR-21 (Anti-miRNA™ miRNA inhibitor miR-21, Cat. no.: AM17000) and non-specific miR (Anti-miRNA™ inhibitor Negative Control #1, Cat. no.: AM17010) were purchased from ThermoFisher. Anti-miR-21 and NS miR were transfected into cells at 20 nM concentration using Lipofectamine RNAiMAX reagent according to the protocol.

| Reverse transcription PCR analysis
Total RNA was extracted using Trizol reagent (Invitrogen). M-MLV reverse transcriptase (Promega BioSciences) were used to reverse transcription following instructions. In brief, Oligo (dT), 1 μg mRNA, dNTPs and DEPC water were mixed in a total vol-

| Sphere formation assay
CSCs were enriched in suspension culture as previously described. 19 In brief, ovarian cancer cells were trypsinized and resuspended in non-adherent petri dish coated with 5 mL 0.5% F I G U R E 2 BPA induces cancer stem cell phenotypes. (A) HEYA8 and (B) SKOV-3 were cultured with BPA at 0 and 10 nM in non-adherent culture dish for 72 h. Total RNA was extracted and reverse transcription-PCR was performed using BMI-1, Nanog, and c-Kit sequencespecific primers. The signal intensities were quantified by densitometric analysis and the amount was normalized for the amount of β-Actin. (C) HEYA8 and (D) SKOV-3 were cultured with BPA at 0 or 10 nM in non-adherent culture dish for 72 h. After 72 h, tumour spheres were recovered, dissociated and incubated for another 72 h for secondary sphere formation. The number of tumour spheres generated was photographed, quantified by MTT assay and counted. Bar = 100 μm. Experiments were repeated three times and results are presented as the mean ± SD and were analysed using unpaired student's t test.
F I G U R E 3 BPA induces MCS formation via GPR30. (A) HEYA8 and (B) SKOV-3 transfected with nonspecific (NS) siRNA, ERRα siRNA ERRγ siRNA or GPR30 siRNA were cultured with BPA at 0 or 10 nM in non-adherent culture dish for 72 h. The number of tumour spheres generated was photographed, quantified by MTT assay and counted. Total RNA was extracted and reverse transcription-PCR was performed using ERα, ERβ, ERRα, ERRβ, ERRγ and GPR30 sequence-specific primers. Bar = 100 μm. Experiments were repeated three times and results are presented as the mean ± SD and were analysed using unpaired student's t test.

| Statistical analysis
Results are analysed with unpaired t-test (GraphPad) or two-way anova and showed as the mean ± SD. p < 0.05 was considered as statistically significant.

| BPA drives tumour-initiation and self-renewal of ovarian CSCs
The ability to form spheres in nonadherent cultures in serum-free medium is one of the characteristics of CSCs. 19 We first examined the effects of BPA on CSC sphere growth. Figure 1 shows that F I G U R E 4 BPA induces Exportin-5 and miR-21 expression. (A) HEYA8 and (B) SKOV-3 were cultured with BPA at 0 and 10 nM in non-adherent culture dish for 72 h. Total RNA was extracted and reverse transcription-PCR was performed using Drosha, Dicer, and Exportin-5 sequence-specific primers. The signal intensities were quantified by densitometric analysis and the amount was normalized for the amount of β-Actin. (C) HEYA8 and (D) SKOV-3 were cultured with BPA at 0 and 10 nM in non-adherent culture dish for 72 h. Total RNA was extracted and reverse transcription-PCR was performed using miR-21 sequence-specific primer. The signal intensities were quantified by densitometric analysis and the amount was normalized for the amount of U6. Experiments were repeated three times and results are presented as the mean ± SD and were analysed using unpaired student's t test.  was insignificant ( Figure 1A,B). DMSO was included as control (0 nM).
Furthermore, this increase was also observed at different time points with maximal increase at 72 h ( Figure 1C,D). To more precisely quantify the CSC-like phenotype, we assessed the expression of stemness markers. Using RT-PCR, our results revealed high and widespread expression of BMI-1, Nanog, and CD117 (c-Kit) in BPA-treated cells, suggesting that BPA accelerates the transition to stem-like phenotype (Figure 2A,B). Since spheroid formation capabilities in serial passages is an indirect indicator in support of stem cell self-renewal, 20 we further verified the effect of BPA in recurrence experiment. Primary spheres treated with 10 nM BPA were dissociated and replated to form secondary spheres. The growth of the secondary sphere assessed by MTT showed a more than two-fold increase in BPA-induced growth ( Figure 2C). The capacity of the secondary sphere formation was also significantly higher in BPA-treated cells compared with controls (~1.6-fold and ~2-fold, respectively) ( Figure 2D), suggesting an enhanced selfrenewal capacity.

| GPR30 is a target of BPA-mediated CSC regulation
Since SKOV-3 which has defective ERα receptor also demonstrated similar BPA-induced sphere formation as in HEYA8 which consists of functional ERα receptor, we hypothesized that BPA may induce the growth of CSCs and sphere formation in an ERindependent manner through alternative receptors, which are ERRα, ERRγ and GPR30, that have key roles in ovarian cancer progression. 21 Although ERRα, ERRγ, and GPR30 were all expressed on HEYA8 and SKOV-3, only silencing of GPR30 with siRNA abrogated BPA-mediated sphere formation and growth ( Figure 3A,B) while nonspecific siRNA had no effect. In contrast, blockade of ERRα or ERRγ, which were also present on cells, did not produce any inhibition on the BPA-mediated sphere formation ( Figure 3A,B), suggesting a key role of GPR30 in BPA-mediated CSC regulation.

| BPA increases Exportin-5 expression in CSCs
Next, we investigated the mechanism by which BPA could regulate the CSC phenotype. There is a strong correlation between Dicer/ Drosha suppression and more aggressive tumour phenotype such as stemness. 22 We, therefore, examined whether BPA may regulate Dicer/Drosha expression in ovarian CSCs. However, the expression of Dicer/Drosha remained unchanged ( Figure 4A,B). Additional mechanisms are also possible, such as the Exportin-5, which has been implicated in the nuclear export of pre-miRNA. 7 We found a significant positive correlation between BPA and the expression of Exportin-5 ( Figure 4A,B).

| Exportin-5 expression induces miR-21
To examine the consequences of the increase in Exportin-5 expression, we assessed changes in levels of mature miRNAs, which has been previously implicated in the aggressiveness of ovarian carcinoma. 23,24 BPA led to a significant increase in oncomiR, miR-21 ( Figure 4C,D). As shown in Figure 5A,B, silencing of Exportin-5 led to the repression of metastatic cancer spheroid (MCS) formation and growth resulting from BPA, as well is its target gene miR-21 ( Figure 5C,D), indicating that Exportin-5 is necessary for the BPA-mediated miRNA expression and promotion of MCS. G15 was identified as a GPR30 selective antagonist. 25,26 As shown, G15 significantly suppressed both spheroid forming and growth of HEYA8 and SKOV-3 at 20 nM ( Figure 6A,B).

| DISCUSSION
Here we show for the first time that environmental factor BPA is essential to generate and maintain the CSC subpopulation in ERnegative ovarian cancer via a miRNA biogenesis pathway at environmental relevant level. This may help to explain the significant cancer-promoting properties of BPA with low exposure. It is noteworthy that cancer cells are plastic and have been shown to dynamically switch from non-CSC to CSC. For example, such phenotypic plasticity by non-CSCs has recently been shown in melanoma and F I G U R E 6 The use of GPR30 antagonist G15 abolishes BPA-induced MCS formation. (A) HEYA8 and (B) SKOV-3 were cultured with BPA at 0 and 10 nM in non-adherent culture dish for 72 h with G15 or DMSO as the control. The number of tumour spheres generated was photographed, quantified by MTT assay and counted. (C) HEYA8 and (D) SKOV-3 were cultured with BPA at 0 and 10 nM in non-adherent culture dish for 72 h with G15 or DMSO as the control. Total RNA was extracted and reverse transcription-PCR was performed using Exportin-5 and miR-21 sequence-specific primers. The signal intensities were quantified by densitometric analysis and the amount was normalized for the amount of β-Actin or U6 respectively. Bar = 100 μm. Experiments were repeated three times and results are presented as the mean ± SD and were analysed using two-way anova. reported. 36 Several mechanisms have been described as potential regulators of Exportin-5 level including the modulation of mRNA stability through PI3K signalling. Exportin-5 expression has also been reported to be regulated by epigenetic cofactors. 37 Whether BPA uses these or other mechanisms driving this aggressive phenotype awaits further investigation. This is the first report that the endocrine disrupter BPA has the capacity to enhance the formation and expansion of CSCs in ERnegative ovarian cancer cells. Importantly, unlike some previous F I G U R E 7 Anti-miR21 inhibits CSC viability and spheroid formation ability. (A) HEYA8 and (B) SKOV3 cells were treated with anti-miR21 to a working concentration of 20 nM for 72 h with NS miR as control. The number of tumour spheres generated quantified by MTT assay and counted. Experiments were repeated three times and results are presented as the mean ± SD and were analysed using unpaired student's t test.
studies that have used high micromolar doses of BPA, we have used low nanomolar doses, which are relevant to human exposure levels. 4,5 Moreover, BPA does not appear to mediate its effect through the classical ERs in ovarian cancer cells. Rather, nonclassical GPR30 could serve as putative receptor that sensitize BPA at nanomolar level. In support, ER-independent IGF-1R, TGFβ, JAK/STAT3, MAPK/ERK and PI3K/Akt were also previously reported to mediate BPA actions in ovarian cancer cells. 38 VEGR was found to be stimulated in breast cancer. 39 AR could essentially mediate the actions of BPA in advanced prostate adenocarcinomas. 40 We also provide evidence that a new mechanism in global miRNA regulation is involved in this process, which may well explain the observation of its

CO N FI LC T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request from the authors.