20(S)-Rg3 blocked epithelial-mesenchymal transition through DNMT3A/miR-145/FSCN1 in ovarian cancer

Epithelial-mesenchymal transition (EMT) is one of the key mechanisms mediating cancer progression. MicroRNAs (miRs) are essential regulators of gene expression by suppressing translation or causing degradation of target mRNA. Growing evidence illustrates the crucial roles of miRs dysregulation in cancer development and progression. Here, we have found for the first time that the ginsenoside 20(S)Rg3, a pharmacologically active component of Panax ginseng, potently increases miR-145 expression by downregulating methyltransferase DNMT3A to attenuate the hypermethylation of the promoter region in the miR-145 precursor gene. Restoration of DNMT3A reverses the inhibitory effect of 20(S)-Rg3 on EMT. FSCN1 is verified as the target of miR-145 to suppress EMT in human ovarian cancer cells. The results from nude mouse xenograft models further demonstrate the suppressive effect of miR-145 on malignant progression of ovarian cancer. Taken together, our results show that 20(S)-Rg3 blocks EMT by targeting DNMT3A/miR-145/FSCN1 pathway in ovarian cancer cells, highlighting the potentiality of 20(S)-Rg3 to be used as a therapeutic agent for ovarian cancer.


INTRODUCTION
Ovarian cancer is the most lethal gynecological tumor, existing predominantly in the form of epithelial ovarian cancer (EOC) [1,2]. The prognosis of EOC patients remains poor, and cancer metastasis and recurrence are the major causes of death. Emerging evidence suggests that the epithelial-mesenchymal transition (EMT), the conversation of epithelial cells to fibroblast-like cells mainly characterized by loss of epithelial molecules, acquisition of mesenchymal markers, enhancement of cell mobility and invasion, plays a crucial role in the progression of EOC by increasing cancer cell invasion and metastasis [3]. Thus, targeting EMT process for novel anticancer drug discovery is highly significant for the clinical benefits of ovarian cancer patients.
Many EMT players such as E-cadherin transcription repressors including snail, zeb and Twist, and microRNAs have been identified so far. MicroRNAs are small noncoding RNAs that modulate gene expression at the posttranscriptional level through complementary binding to 3'UTR of target mRNAs [10]. Some microRNAs including miR-145, miR-200, and miR-29b, to name a few, have been suggested as EMT repressors in various types of cancer [11][12][13][14]. miR-145 has been downregulated in many cancers [15][16][17][18][19], functioning as tumor suppressor to inhibit tumor cell growth and survival, induce cell apoptosis and cell cycle arrest, and attenuate tumor cell migration and invasion via targeting various molecules [16][17][18][20][21][22][23]. Reduction of miR level in cancers can be associated with aberrant epigenetic regulation such as DNA hypermethylation [24]. DNMT3A and DNMT3B are two major de novo DNA methyltransferases in mammals, while DNMT1 is in charge of maintenance of DNA methylation [25]. DNMT3A and DNMT3B have been reported upregulated in ovarian cancers [26]. Both of them have been able to hypermethylate the promoter region in microRNA precursor genes and thus inversely regulate microRNA transcription.
In the present study, we discovered that 20(S)-Rg3 enhanced miR-145 expression by downmodulating DNMT3A to attenuate the methylation level in the promoter region of miR-145 precursor gene. The promotion of miR-145 by 20(S)-Rg3 directly targeted FSCN1 to reverse EMT in vitro and vivo. These results not only uncovered the novel anti-cancer mechanism of 20(S)-Rg3, but also revealed the regulatory pathway for miR-145 expression.

20(S)-Rg3 upregulated miR-145 via suppressing DNMT3A to demethylate pre-miR-145 gene
Since DNA hypermethylation has been connected to miR-145 deregulation in prostate cancer [27], we compared the methylation status of miR-145 precursor gene in 20(S)-Rg3-treated cells relative to non-treated cells. The initial assessment of potential CpG islands in the 2000 bp upstream genomic sequence encoding pre-miR-145 found no CpG islands but some CG sites. The MSP results illustrated that the methylation level in the promoter region of pre-miR-145 promoter was decreased in 20(S)-Rg3-treated SKOV3 and 3AO cells ( Figure  2A). We thus detected the influence of 20(S)-Rg3 on the expression of DNMT family members. Western blot results showed that DNMT3A, the DNA methyltransferase responsible for de novo methylation, was obviously decreased while DNMT1 and DNMT3B were remained unchanged in 20(S)-Rg3-treated SKOV3 and 3AO cells ( Figure 2B). DNMT3A plasmid was transfected into 20(S)-Rg3-treated cells to restore the expression of DNMT3A ( Figure 2C), which reversed the 20(S)-Rg3triggered upregulation of miR-145 ( Figure 2D). In parallel, MSP results showed that overexpression of DNMT3A increased the methylation level in the promoter region of pre-miR-145 in 20(S)-Rg3-treated SKOV3 and 3AO cells ( Figure 2E). These results indicated that 20(S)-Rg3 enhanced miR-145 via inhibiting DNMT3A expression and thus relieved the methylation restrain on pre-miR-145 transcription.
We next assessed the effects of DNMT3A on the anti-EMT activity of 20(S)-Rg3. DNMT3A overexpression reversed 20(S)-Rg3-triggered E-cadherin upregulation, N-cadherin and vimentin downregulation ( Figure 2F). And the attenuation of migration and invasion of 20(S)-Rg3treated cells was largely reversed by ectopic expression of DNMT3A ( Figure 2G). These results showed that 20(S)-Rg3 blocked EMT through downregulating DNMT3A in ovarian cancer cells.

miR-145 inhibited DNMT3A-promoted EMT
We further studied the role of miR-145 in DNMT3A-promoted EMT. Ectopic expression of miR-145 in DNMT3A-overexpressed cells ( Figure 3A) did not in turn affect DNMT3A expression ( Figure 3B), but blunted www.impactjournals.com/oncotarget the EMT-inducible activity of DNMT3A as shown by restoration of E-cadherin, loss of N-cadherin and vimentin ( Figure 3C), and attenuation of cell motility and invasion ( Figure 3D). These results showed that DNMT3A induced EMT via inhibiting miR-145.

miR-145 directly targeted FSCNI to inhibit EMT
Next we explored the target of miR-145 to block EMT. We first predicted putative target genes of miR-145 by searching the TargetScan database (release 5.1, http://www.targetscan.org/), and FSCN1 was chosen to be experimentally verified. 3'-UTRs of FSCN1 containing the wild-type or mutant putative miR-145 binding site were cloned into a luciferase reporter plasmid, respectively. Specifically, although 4 conserved miR-145 seeding sequences were predicted to exist in the 3'-UTR of FSCN1, only one of them was experimentally confirmed as an actual binding site which was cloned into a luciferase reporter plasmid in our study [28]. The luciferase reporter assay showed that luciferase activity was significantly inhibited in cells co-transfected with miR-145 mimic and FSCN1 WT-3' UTR vector, while no changes of luciferase activity were detected in cells transfected with miR-145 mimic and luciferase reporter plasmids containing the mutant seeding sequence ( Figure  4A). Additionally, western blot analysis showed that miR-145 overexpression diminished FSCN1 expression level ( Figure 4B). These data indicated the direct suppressive effect of miR-145 on FSCN1. Since the role of FSCN1 in EMT was not fully explored, we investigated whether FSCN1 mediated the inhibitory effect of miR-145 on EMT. In cells overexpressed both FSCN1 and miR-145, the anti-EMT effect of miR-145 as evidenced by   upregulation of E-cadherin, downregulation of N-cadherin and vimentin ( Figure 4C), and attenuation of migration and invasion ( Figure 4D) was largely reversed by ectopic expression of FSCN1. Meanwhile, EMT was reversed in SKOV3 and 3AO cells by downregulation of FSCN1, as shown by increased expression of E-cadherin, decreased expression of N-cadherin and vimentin ( Figure 4E), and demotion of migration and invasion ( Figure 4F). Taken together, our data suggested that miR-145 blocked EMT by targeting FSCN1 in ovarian cancer cells.

miR-145 inhibited ovarian cancer EMT in vivo
To determine the effect of miR-145 on ovarian cancer progression in vivo, the primary tumor growth in nude mice were examined. miR-145-up SKOV3 cells (transfected by miR-145-expressing lentivirus) or negative control cells (NC) were subcutaneously inoculated into nude mice. All of the mice developed subcutaneous xenografts. Although no differences in body weight were detected between NC and miR-145-up groups ( Figure  5A), the average tumor volume of miR-145-up group was reduced by nearly 40% at week 4 compared with the NC group ( Figure 5B). Immunohistochemistry analysis of subcutaneous tumors showed that E-cadherin was higher, vimentin and FSCN1 were lower in miR-145-up tumors compared to NC tumors ( Figure 5C).
We then examined the inhibitory effect of miR-145 on the intraperitoneal dissemination of ovarian cancer. NC or miR-145-up SKOV3 cells were inoculated into the abdomen of nude mice. During the 28-day observation, the average body weigh was higher in miR-145-up group than NC group, which became significant since day 22 ( Figure 5D). And less metastatic tumor nodules and 3'UTR sequences used for construction of luciferase report vectors. Luciferase reporter assay showed that, compared to negative control, miR-145 significantly diminished luciferase activity of wild type FSCN1 3'-UTR, while had no effect on luciferase activity of the mutated ascites were developed in miR-145-up group ( Figure  5E). Intraperitoneal tumor nodules presented throughout the mice's abdominal cavity. Tumors in spleen fascia and diaphragm of miR-145-up mice were substantially smaller in size than those in NC mice, and immunohistochemistry analysis of metastatic tumor showed the tumors were from the intraperitoneal dissemination ( Figure 5F). These data reproduced the suppressive action of miR-145 on ovarian cancer cell in vivo.

DISCUSSION
20(S)-Rg3 inhibited cancer cell motility and invasiveness, and the mechanism could be partly attributed to its anti-EMT effect [29][30][31]. Here, we first reported that the pathway composed of DNMT3A, miR-145, and FSCN1 was implicated in the anti-EMT mechanism of 20(S)-Rg3 (Figure 6), and that the methylation repression of miR-145 by DNMT3A played an important role in EMT occurrence.
Ginseng, is one of the oldest herbal medicines and induces a variety of physiological and pharmacological effects. Ginseng contains saponins called ginsenosides, which are considered as the biologically active ingredients in ginseng. Many researches have proposed that ginsenoside blocks EMT in cancers. Zhang and colleagues have recently shown that ginsenoside 25-OCH3-PPD, isolated from Panax notoginseng and belonged to protopanaxadiols group as 20(S)-Rg3, reduces EMT markers in normoxically cultured breast cancer cells [32]. Xie et al. have reported that ginsenoside Rg1, a major active component also isolated from Panax notoginseng but belonged to protopanaxatriol group, blocks TGFβ1-induced EMT in rat renal tubular epithelial cells [33]. Collectively, these findings, together with our data, shed new light on the anti-metastasis mechanism of ginsenosides. Nevertheless, the pathway relative to the entrance of the ginsenosides into cells and the direct targets of the ginsenosides have been yet to be determined. It is worthwhile to identify the specific mechanism mediating the anti-tumor role of ginsenosides.
DNA methylation is a critical epigenetic signature that is involved in transcriptional regulation, genomic imprinting, and silencing of repetitive DNA elements [34]. Abnormal hypermethylation correlates with the transcriptional repression of multiple miRNAs. This can lead to the upregulation of oncogenic targets of microRNAs and constitutive activation of signaling pathway that can increase invasion and migration activities [35]. However, to our knowledge, there have been no reports about the methylation regulation of miR-145 in ovarian cancer. In this study, we found that miR-145 was under control of DNMT3A-mediated DNA methylation, and 20(S)-Rg3 inhibited DNMT3A expression to demethylate pre-miR-145 and thus increase miR-145 expression.
DNMT3A together with DNMT3B and DNMT1 are catalytically active DNMTs responsible for genome methylation [36]. DNMT3A and DNMT3B are de novo methyltransferase with different targets [37]. DNMT1 is a maintenance DNA methyltransferase for retaining methylation pattern, with inefficient de novo methylation ability. Increasing evidence shows that these DNMTs work together to maintain a normal methylation pattern, and deregulation of either one could promote malignancies [38]. More recently, DNMT3A has been reported as contributor of EMT [39,40]. In the present study, we showed that DNMT3A induced EMT via decreasing miR-145 expression.
Decreased expression and anti-tumor function of miR-145 have been observed in several cancers [41,42]. The mechanistic studies about miR-145 inhibition on cancer progression have initially focused on its roles in cell apoptosis, cell cycle and cell proliferation [43,44]. Lately, miR-145 has been inversely connected to cancer cell motility and invasiveness, and the mechanism may partly be attributed to its anti-EMT effect [45,46]. FSCN1 [47] is a newly identified miR-145 target in a few cancers. FSCN1 is an actin-bundling protein functioning in cell movement under physiological or pathological conditions [48]. Overexpression of FSCN1 promotes migration and invasion of cancer cells [7,[49][50][51][52][53], and is associated with clinically unfavorable phenotypes in human epithelial cancers including EOC [48][49][50][51][52][53][54]. Nevertheless, the correlation of miR-145-FSCN1 with EMT has not been compellingly proven in the published data. Here we provided the evidence that overexpression of FSCN1 was sufficient to confer EMT to ovarian cancer cells, and its aberrant elevation in ovarian cancer tissues was possibly benefited from miR-145 diminution. We provided evidence that miR-145 inhibited EMT by suppressing FSCN1, however, the detailed mechanism about function of FSCN1 in EMT occurrence still necessitated further elucidation.
In conclusion, we found that 20(S)-Rg3 reversed EMT to inhibit ovarian cancer cells migration and invasion via antagonizing DNMT3A-mediated methylation of pre-miR-145 to promote inhibition of miR-145 on FSCN1. 20(S)-Rg3 is an efficient, low-toxicity and multi-target natural agent, and it is expected to become an important agent in the comprehensive treatment of ovarian cancer, which may improve the current strategy of ovarian cancer therapy.

Reagents and antibodies
Ginsenoside 20(S)-Rg3 was obtained from Tasly Pharmaceutical Company (Tianjin, China) and dissolved at a concentration of 4 mg/ml in DMSO as a stock solution (stored at −20 °C). It was then further diluted in cell culture medium to create working concentrations. The maximum final concentration of DMSO was less than 0.1% for each treatment, and was also used as a control. Antibodies including DNMT3A, DNMT3B, E-cadherin, N-cadherin, vimentin, and β-actin were from Cell Signaling Technology (Beverly, MA), FSCN1 was from Abcam (Cambridge, MA, USA), and DNMT1 was from Active Motif (Carlsbad, CA, USA).

Cell culture and 20(S)-Rg3 treatment
The human ovarian cancer cell line SKOV3 was obtained from the Shanghai Cell Bank of Chinese Academy of Sciences (Shanghai, China), 3AO was from the Shandong Academy of Medical Sciences (Jinan, China). Cells were maintained in RPMI 1640 medium (Gibco-BRL, Gaithersburg, MD, USA) supplemented with 10% (v/v) fetal bovine serum at 37°C under a humidified 5% CO 2 atmosphere. Cells were incubated with 80 μg/ ml (for SKOV3) or 160 μg/ml (for 3AO) of 20(S)-Rg3 for 24 h.

Western blot
Total protein was collected from cells by RIPA lysis buffer containing protease inhibitors (Roche, Indianapolis, IN, USA) and 1 mM PMSF on ice. Protein concentration was measured using the BCA-200 Protein Assay kit (Pierce, Rockford, IL, USA). After heat denaturation at 100 °C for 5 min, proteins were separated by electrophoresis on 10 % SDS-PAGE gels and then transferred onto nitrocellulose membranes (Pall Life Science, NY, USA). The membranes were blocked with 5 % non-fat milk at room temperature for 1 h, and then incubated overnight at 4 °C with rabbit anti-

siRNA and transient transfection
Human FSCN1 siRNA was purchased from GenePharma (Shanghai, China). Ovarian cancer cells were seeded into 6-well plates until they reached 40 %-50 % confluency. FSCN1 siRNA (GCAGCCTGAAGAAGAAGCA) was transiently transfected 100nM per well using the X-treme GENE siRNA Transfection Reagent (Roche, Indianapolis, IN, USA). After 48 hours transfection, the cells were harvested for further studies.

Cell migration and invasion assay
After treated, cells were trypsinized and counted. A total of 1×10 5 cells (for migration assay) or 5×10 5 cells (for invasion assay) in 100 μl serum-free medium were added into millicells (Millipore Co., Bedford, MA, USA) without (for migration assay) or with (for invasion assay) Matrigel (Becton Dickinson Labware, Bedford, MA, USA) coated. 500 μl of 1640 medium containing 20 % newborn bovine serum was added to the bottom chambers as the chemotactic factor. After incubation for 24 h (for migration assay) or 48 h (for invasion assay) at 37 °C, cells remaining on the upper surface of the filter were removed using cotton swabs. Then the migrated cells were fixed using methyl alcohol and stained using 0.1 % crystal violet. Migratory (or invasive) cells were counted and averaged from images of five random fields (original magnification ×200) captured using an inverted light microscope. The mean values of three duplicate assays were used for statistical analysis.

Luciferase reporter plasmid construction
The wild-type 3'-UTR sequence of the target gene carrying a putative miR-145 binding site was amplified by PCR. To generate mutant 3'-UTR fragment of miR-145 target gene, we adopted the two-step PCR method as reported previously. Briefly, two fragments of 3'-UTR were amplified by two sets of overlapped primers in which mutated seeding sequences of miR-145 were introduced. The wild-type and mutant PCR products were digested with Hind III and Sac I enzymes, inserted into pMIR-REPORT TM Luciferase vectors (Ambion, Austin, TX, USA) and verified by DNA sequencing. The following primers were used: FSCN1-WT 3'UTR forward:

Luciferase reporter assay
Luciferase reporter assays were carried out in both SKOV3 and 3AO cells. Cells were seeded into 24-well plate. When reached 80 %-90 % confluency, cells were co-transfected with pRL-TK vector (20 ng), wild-type (WT-3' UTR) or mutant (MUT-3' UTR) reporter vectors (180 ng), along with miR-145 mimic or negative control at a final concentration of 20 nM using the X-treme GENE siRNA Transfection Reagent. Transfections were performed in triplicate. 24 h after transfection, the relative firefly luciferase activity (normalized to Renilla luciferase activity) was measured using a dual-luciferase reporter gene assay system (Promega, Madison, WI, USA), and results were depicted as the percentage change over the respective control. Each experiment was performed in triplicate.

Animal study
The care and use of experimental animals were approved by the ethical committee of the First Affiliated Hospital of Xi'an Jiaotong University, and were adherent to the institutional guidelines and ethical standards. SKOV3 cells were infected with miR-145expressing lentivirus GV209-miR145 (GENECHEM, Shanghai, China) or negative control lentivirus GV209 (GENECHEM, Shanghai, China). For subcutaneous xenograft experiment, cells were trypsinized and resuspended in PBS at a final concentration of 2×10 7 cells/ ml. 2×10 6 cells were injected subcutaneously into the flank of six-week old BALB/c nude mice. Animal body weights and tumor perpendicular diameters were recorded every two days. Tumor volumes were calculated according to the formula of V=0.5236×(L×W 2 )(V: tumor volume, L: length, W: width). The mice were killed after 28 days on the experimental treatments. Tumor samples were fixed in 4% polyformaldehyde for 24h at room temperature, paraffin-embedded, and sectioned for immunohistochemical analysis. For intraperitoneal xenograft model, SKOV3 cells were trypsinized and resuspended in PBS at a concentration of 1×10 8 /ml. Mice were inoculated with 1×10 7 cells in the abdominal cavity at day 0. The mice were observed daily and sacrificed after 28 days. Necropsies were performed, disseminated lesions were resected from mice and pathologically confirmed using HE staining. Ascitic fluid was collected by 1mL syringe and measured by pipet. The number and weight of tumor nodules on the surface of the spleen fascia, diaphragm, bowel and omentum, liver and bladder were counted, measured and statistically analyzed.

Immunohistochemistry analysis
Paraffin-embedded xenograft tissue sections on poly-1-lysine-coated slides were deparaffinized and rinsed with 10 mM Tris-HCl (pH 7.4) and 150 mM sodium chloride. Paroxidase was quenched with methanol and 3 % hydrogen peroxide. Slides were then placed in 10 mM citrate buffer (pH 6.0) at 100°C for 20 min in a pressurized heating chamber. Detection of antigens was carried out using incubation with the primary antibodies for 2 hours at room temperature, followed by incubation with HRPlabeled secondary antibody (MaxVision TM HRP-Polymer anti-Mouse/Rabbit IHC Kit) at room temperature for 30 minutes and color development with DAB. Negative control specimens were incubated in PBS without the primary antibody under the same conditions. Slides were counterstained with hematoxylin, dehydrated in an ascending alcohol series, and mounted for analysis. Digital images were acquired on an Olympus BH-2 microscope (Tokyo, Japan) installed with the DeltaPix Camera and software (Maalov, Denmark).

Statistical analysis
All experiments were performed at least in triplicate, and each experiment was independently performed at least 3 times. The graphical presentations were performed using GraphPad Prism 5.0. Data were presented as the means±SE and were analyzed using SPSS 22.0 software (Chicago, IL, USA). Statistical differences were tested by Chi-square test or two-tailed t-test, and Fisher exact test was performed to analysis MSP results. Differences were considered significant at P <0.05 (*) or highly significant at P <0.001 (**).