The Downregulation of miR-509–3p Expression by Collagen Type XI Alpha 1-Regulated Hypermethylation Facilitates Cancer Progression and Chemoresistance via the DNA Methyltransferase 1/Small Ubiquitin-like Modifier-3 Axis in Ovarian Cancer Cells

Background MicroRNAs are a group of small non-coding RNAs that are involved in development and diseases such as cancer. Previously, we demonstrated that miR-335 is crucial for preventing collagen type XI alpha 1 (COL11A1)-mediated epithelial ovarian cancer (EOC) progression and chemoresistance. Here, we examined the role of miR-509–3p in EOC. Methods The patients with EOC who underwent primary cytoreductive surgery and postoperative platinum-based chemotherapy were recruited. Their clinic-pathologic characteristics were collected, and disease-related survivals were determined. The COL11A1 and miR-509–3p mRNA expression levels of 161 ovarian tumors were determined by real-time reverse transcription-polymerase chain reaction. Additionally, miR-509–3p hypermethylation was evaluated by sequencing in these tumors. The A2780CP70 and OVCAR-8 cells transfected with miR-509–3p mimic, while the A2780 and OVCAR-3 cells transfected with miR-509–3p inhibitor. The A2780CP70 cells transfected with a small interference RNA of COL11A1, and the A2780 cells transfected with a COL11A1 expression plasmid. Site-directed mutagenesis, luciferase, and chromatin immunoprecipitation assays were performed in this study. Results Low miR-509–3p levels were correlated with disease progression, a poor survival, and high COL11A1 expression levels. In vivo studies reinforced these findings and indicated that the occurrence of invasive EOC cell phenotypes and resistance to cisplatin are decreased by miR-509–3p. The miR-509–3p promoter region (p278) is important for miR-509–3p transcription regulation via methylation. The miR-509–3p hypermethylation frequency was significantly higher in EOC tumors with a low miR-509–3p expression than in those with a high miR-509–3p expression. The patients with miR-509–3p hypermethylation had a significantly shorter overall survival (OS) than those without miR-509–3p hypermethylation. Mechanistic studies further indicated that miR-509–3p transcription was downregulated by COL11A1 through a DNA methyltransferase 1 (DNMT1) phosphorylation and stability increase. Moreover, miR-509–3p targets small ubiquitin-like modifier (SUMO)-3 to regulate EOC cell growth, invasiveness, and chemosensitivity. Conclusion The miR-509–3p/DNMT1/SUMO-3 axis may be an ovarian cancer treatment target.


Results
A low miR-509-3p level correlates with a poor clinical outcome To investigate the expression level and signi cance of miR-509-3p in ovarian cancer, we rst performed quantitative real-time PCR to evaluate the miR-509-3p expression level in 161 ovarian cancer and 23 noncancerous tissues. Low miR-509-3p mRNA levels in patients with EOC were signi cantly associated with old age (P = 0.005), an advanced stage (P < 0.001), serous histology (P = 0.002), and cancer death (P = 0.010) ( Table 1). However, no correlation was found between the miR-509-3p levels and the response to chemotherapy (P = 0.188) or the progression-free interval (P = 0.286). The miR-509-3p levels in cancerous tissues tended to be lower than those in non-cancerous tissues (Supplementary Table 1). Furthermore, low miR-509-3p levels were signi cantly related to high COL11A1-expressing tumors (P < 0.001). Patients with low miR-509-3p mRNA levels have signi cantly shorter OS (Fig. 1A, P = 0.013) and PFS ( Fig. 1B, P = 0.029) than those with high miR-509-3p levels. The invasive phenotypes of EOC cells are regulated by miR-509-3p To determine the impact of miR-509-3p on the aggressive traits of EOC cells, A2780CP70 and OVCAR-8 cells were transfected with miR-509-3p mimics to increase the expression of miR-509-3p ( Figs Hypermethylation of the p278 promoter region regulates miR-509-3p transcription Several studies have shown that certain miRNA genes are silenced in human tumors via aberrant CpG island hypermethylation [25]. To understand whether miR-509-3p is downregulated by hypermethylation in ovarian cancer tumorigenesis, A2780CP70 cells were treated with the demethylation reagent 5-aza. The expression of miR-509-3p was restored after 5-aza treatment (Fig. 3A). The promoter region information of the miR-509-3p gene was obtained from FANTOM database [26], and we identi ed two CpG sites in this region (Fig. 3B, upper panel). To examine whether this region could attenuate transcription upon hypermethylation, p278 (chr 146,413,629 to 146,413,907) was methylated by SssI methylase and transfected into A2780CP70 cells. Luciferase activity was almost null in these cells (Fig. 3B, lower panel). These results suggest that the methylation of the two CpG sites in miR-509-3p is responsible for promoter downregulation.
To further con rm that the promoter hypermethylation of miR-509-3p was the main mechanism responsible for the decrease in the expression of miR-509-3p, four ovarian cell lines were sequenced.
Mammalian DNA methylation is essential for development and is controlled by various factors, including 3 active DNA cytosine methyltransferases (DNMT1, DNMT3A, and DNMT3B), while DNMT1 is the main enzyme responsible for maintaining the methylation patterns [27]. An aberrant DNA methylation of CpGisland-containing promoters leads to gene silencing in both physiological and pathological contexts, especially in cancer cells [27]. Therefore, we hypothesized that the downregulation of miR-509-3p transcription by COL11A1 might be regulated by DNMT activation. First, we examined the expression of the three DNMTs in ovarian cancer cells with different COL11A1 expression statuses. COL11A1 depletion via RNA interference reduced the DNMT1 expression level in high COL11A1-expressing A2780CP70 cells.
Conversely, the overexpression of COL11A1 increased DNMT1 expression in low COL11A1-expressing A2780 cells (Fig. 4C). However, the expression of DNMT3A and DNMT3B did not change. Furthermore, the binding of DNMT1 to the miR-509-3p promoter increased in COL11A1-overexpressing A2780 cells and decreased in COL11A1-knockdown A2780CP70 cells (Fig. 4D). The activity of the binding of DNMT1 to the miR-509-3p promoter was inhibited by 5-aza (Fig. 4E). There results indicate that COL11A1 downregulates miR-509-3p expression in ovarian cancer cells by inhibiting the binding of DNMT1 to the miR-509-3p promoter.
Increased phosphorylation of DNMT1 has been shown to in uence DNMT1 stability [28]. We have previously demonstrated that COL11A1 confers chemoresistance to ovarian cancer cells through the activation and phosphorylation of Akt and phosphoinositide-dependent kinase-1 (PDK1) stabilization [19]. These results led us to hypothesize that the downregulation of miR-509-3p by COL11A1-mediated DNMT1 activation can be caused by an increased DNMT1 phosphorylation. Our results show that the phosphorylation of Akt and DNMT1 increased in COL11A1-overexpressing A2780 cells and decreased in COL11A1-knockdown A2780 cells (Fig. 4F). It has been shown that silencing DNMT1 lead to p16 activation [29]. The p16 decreased in COL11A1-overexpressing A2780 cells and increased in COL11A1knockdown A2780 cells (Fig. 4F), indicating p16 is another target gene in uenced by DNMT1. After MG132 treatment, DNMT1 ubiquitination was more extensive in A2780/V cells than in A2780/COL11A1 cells, and it could be rescued by COL11A1 overexpression. Conversely, COL11A1 silencing facilitated the ubiquitination of DNMT1 in A2780CP70/shCOL11A1 cells (Fig. 4G). Sequencing analysis revealed that the miR-509-3p methylation pattern was obverse in A2780/COL11A1 and A2780CP70/shV cells, which showed a low miR-509-3p expression. Inversely, miR-509-3p methylation was not observed in A2780/V and A2780CP70/shCOL11A1 cells, which had a high miR-509-3p expression (Fig. 4H).

miR-509-3p targets SUMO-3 in ovarian cancer
Based on miRNA target analysis algorithms (miRanda and TargetScan), SUMO-3 is a potential target mRNA of miR-509-3p (Fig. 5A, upper panel). We demonstrated that the co-expression of miR-509-3p signi cantly inhibited the re y luciferase reporter activity of the wild-type SUMO-3 3′-UTR, but not that of the mutant 3′-UTR (Fig. 5A, lower panel), using a dual-luciferase reporter system, indicating that SUMO-3 is a direct target of miR-509-3p.
We further explored whether SUMO-3 was critical for the effect induced by miR-509-3p. The expression of SUMO-3 was inhibited by miR-509-3p, which could be rescued by the transfection of A2780CP70 cells with SUMO-3 ( Figure S1). The MTT and invasion experiment results implied that the re-expression of SUMO-3 abolished the inhibiting in uences of miR-509-3p on A2780CP70 cell growth (Fig. 5D) and invasion ability (Fig. 5E). SUMO-3 increased the cell resistance to cisplatin treatment (Fig. 5F). The increased cell sensitivity to cisplatin by miR-509-3p mimics was reduced by the re-expression of SUMO-3 (IC 50 value: from 8.99 to 32.98 µM, P < 0.01, Fig. 5F). Therefore, miR-509-3p inhibited the aggressive phenotype of EOC cells by targeting SUMO-3.
Discussion miRNA dysregulation, which can result from aberrant DNA methylation, contributes to cancer tumorigenesis [30,31]. In this study, we found that miR-509-3p expression was reduced in ovarian cancer, was mediated by the hypermethylation of the promoter region of the miR-509-3p gene, and signi cantly associated with a poor prognosis and COL11A1 expression. Mechanistic studies indicated that miR-509-3p transcription, which was downregulated by COL11A1 through an increased DNMT1 stability, was achieved by the binding of DNMT1 to the miR-509-3p promoter. miR-509-3p inhibited ovarian cancer cell growth, invasion ability, and enhanced their chemosensitivity by downregulating SUMO-3. The reinforced expression of SUMO-3 reversed the suppressive effects of miR-509-3p. These ndings support the important roles of miR-509-3p in suppressing the tumorigenesis of ovarian cancer (Fig. 5G).
miR-509-3p is expressed from a genomic cluster of miRNAs of around 100 kb on chrXq27.3. The members of the Xq27.3 miR cluster have been reported to be strongly anti-correlated to a multi-cancer 'metastasis-associated broblast' gene signature [32]. Furthermore, the expression of these miR cluster members has been associated with the cancer stage and the survival of patients with ovarian cancer and is lower in omental metastases than in primary tumors [33][34][35][36][37][38][39][40][41]. miR-506-3p, a member of this miR cluster, is downregulated in pancreatic cancer due to the hypermethylation of its promoter region [42]. Therefore, we hypothesized whether the expression of other miRNAs in the cluster may be regulated by the same mechanism. In this study, we rst performed sequencing to evaluate the detailed methylation patterns of two CpG islands in the promoter region of miR-509-3p in EOC cell lines and tumor samples and found that promoter hypermethylation by DNMT1 resulted in miR-509-3p silencing in ovarian cancer and that miR-509-3p hypermethylation was associated with a decreased survival time. miR-509-3p has been shown to target CDK2 and to in uence the cell cycle, colony formation, and migration of human lung and cervical cancer cell lines [43]. miR-509-3p directly targets XIAP to inhibit proliferation [12], and targets XIAP, Golgi phosphoprotein-3 (GOLPH3), and Wnt ligand secretion mediator (WLS) [10] to regulate platinum sensitivity in chemoresistant ovarian cancer cells. Another study showed that YAP1 is both a major effector of miR509-3p-mediated attenuation of migration and invasion, and spheroid formation in ovarian cancer cells [12]. Here, we provide evidence that a novel miR-509-3p epigenetic silencing alternation and SUMO-3 targeting reduce ovarian cancer cell aggressiveness and chemoresistance.
SUMOs are a group of ubiquitin-like proteins that are attached to substrate proteins via a reversible posttranslational protein modi cation termed SUMOylation. To date, ve isoforms of SUMO have been identi ed in the human genome [44]. Dysregulated mRNA and protein levels of the SUMO machinery components have been implicated with certain prognosticators, such as a higher histological grade, a more advanced cancer stage, the presence of metastases, and a poor prognosis [44]. Further investigation is required to explore the exact mechanisms underlying the regulation of EOC cell aggressiveness and chemosensitivity by SUMO-3.
Our study suggests that a low miR-509-3p or miR-509-3p hypermethylation level is associated with a poor patient survival. Of the 161 EOC patients, the tumor miR-335 expressions of 137 patients have been described previously [24]. It is worth mentioning that the patients with low miR-509-3p and miR-335 levels had the shortest survival. Compared to patients with high miR-509-3p and miR-335 levels, the patients with low miR-509-3p and miR-335 levels tended to have shorter OS ( Figure S2A, P < 0.001) and PFS ( Figure S2B, P < 0.001). Similar ndings were observed among patients with serous histology; compared to those with high miR-509-3p and miR-335 levels, those with low miR-509-3p and miR-335 levels had shorter OS ( Figure S2C, P < 0.028) and PFS ( Figure S2D, P = 0.058). Altogether, a low miR-509-3p level, especially when combined with a low miR-335 level, could help identify the patients with ovarian cancer at the highest risk of showing a poor clinical outcome. Therefore, miR-509-3p may serve as a prognostic biomarker of ovarian cancer.
We also provide evidence that miR-509-3p promoter hypermethylation, especially at position 1, may play a predictive role on the EOC prognosis. However, the occurrence rates of hypermethylation at position 1 or  phosphorylation has been shown to affect its methyltransferase activity [50] and its interaction with PCNA and UHRF1 [51], as well as DNMT1 stability [27]. Our previous report indicated that COL11A1 activates Akt phosphorylation [19]. Here, we showed that COL11A1 upregulates DNMT1 phosphorylation (Figs. 4E and 4H), and subsequently enhances DNMT1 protein stability (Figs. 4F and 4I). These results suggest that COL11A1 might regulate DNMT1 phosphorylation through Akt, and then induce the binding of DNMT to the miR-509-3p promoter ( Fig. 4C and 4J). Our results also indicated that 5-aza treatment restored miR-509-3p expression (Fig. 3A) and reduced the cell invasion ability of EOC cells (Fig. 3D).
Several epigenetic regulators, including DNMT inhibitors and histone deacetylase (HDAC) inhibitors, are currently under clinical investigation or have been approved for clinical use [52]. Our study provides new insight into an epigenetic therapeutic target for ovarian cancer treatment. miRNA isolation. Ovarian cancer and non-cancerous (control) specimens were collected and de-identi ed.

Materials And Methods
Total RNA was extracted from the specimens (100 mg with a tumor cellularity of 70% or greater), using a miRNeasy kit (Qiagen, Germany). The extracted RNA was quanti ed using a NanoDrop 1000 spectrophotometer (Thermo Scienti c). In addition, the integrity of the extracted RNA was determined using an Agilent 2100 Bioanalyzer.
Quanti cation of miR-509-3p. To obtain the miRNA distribution pro le and quantify miR-509-3p in the specimens, we performed a quantitative real-time polymerase chain reaction (qPCR) according to the manufacturer's instructions (Qiagen). In brief, 100 ng of total extracted RNA was collected and pooled from the samples, and cDNA was synthesized using a miScript Reverse Transcription kit (Qiagen). The cDNA sample was mixed with the qPCR master reagent (Human miScript Assay 384 set v10.1 [Qiagen]) using a Matrix Hydra eDrop (Thermo Scienti c). Only wells with single melting-temperature values were included in further analysis. miRNAs were normalized with reference to the global miRNA mean and expression was calculated using the comparative C t . method. Statistical analysis was performed using the Student's t-test. P-values < 0.05 were considered statistically signi cant.
Quantitative reverse transcriptase PCR (RT-PCR). The RNA obtained (5 µg) was used as the template for the cDNA synthesis reactions, together with random primers and superscript III reverse transcriptase (Applied Biosystems). The resultant cDNA solution (at a 1:20 dilution) was used to detect the level of the target gene mRNA using quantitative PCR (qPCR). An accurate quantitation was achieved based on standard curves, which were drawn by serially diluting a known amount of RNA obtained through an in vitro transcription reaction and by performing TaqMan qPCR using these dilutions, in addition to using the patient samples. Quantitative analysis of the mRNA expression was performed using the Light Cycler® 2.0 System (Roche Diagnostics GmbH). The primers and TaqMan probes used for the analyses were designed using the manufacturer's software, Primer Express. The following primers were used: COL11A1 (HS01097664) and GAPDH (HS99999905). No-reverse-transcription (no-RT) control reactions were performed using 100 ng of the total RNA derived from each individual sample as a template to ensure that the ampli cation did not occur due to DNA contamination. No signal corresponding to the no-RT controls was detected. The target gene mRNA expression was assessed using real-time RT-PCR. GAPDH was used as an internal control for RNA quality. All quantitative analyses were performed in duplicate to assess the consistency of the results. The relative expression levels of the target gene, which were normalized to those of GAPDH, were calculated as follows: ΔC t = C t (target) -C t (GAPDH). The ratio of the number of copies of the target gene mRNA to the number of copies of GAPDH was then calculated: After 48 h of incubation, the in vitro cytotoxic effects of these treatments were determined using an MTT assay (at 570 nm) and the cell viability was expressed as the percentage of control (untreated) cells (% of control). The MTT analysis was conducted as previously reported [19].
Transwell invasion assay. The Transwell cell invasion assay was performed using polycarbonate membranes with 8 µm pores (Costar, Cambridge, MA, USA). Cells (5 × 10 4 ) were seeded on the membrane of the upper chamber of the Transwell pre-coated with rat collagen I (60 µg/Transwell). Fibronectin in medium (0.6 mL) was added to the lower chamber as the chemoattractant in a 24 h assay at 37°C under 5% CO 2 . The remaining cells in the upper chamber that did not migrate were removed using a cotton swab. The lters were xed in 95% ethanol and stained with 0.005% crystal violet for 1 h. Migrated cells were counted using a phase-contrast microscope (Olympus, Lake Success, NY, USA). The mean of 10 contiguous elds represented the cell number. Each experimental condition was assayed in triplicate.
used. The invasive capacity of cells was normalized to that of each corresponding control. One-sample unpaired Student's t-test was conducted to analyze the differences between the normalized invasive capacities obtained from the three independent experiments and the hypothetical value (which was set to 1).
Chromatin immunoprecipitation (ChIP) assays. Native protein-DNA complexes were cross-linked via treatment with 1% formaldehyde for 15 min, and ChIP assays were performed as previously reported [20].
Statistical analysis. Data were analyzed using SPSS statistical software (version 21.0, IBM Corp., Armonk, NY, USA). Categorical variables are presented as frequencies and percentages and were analyzed using Chi-square test or Fisher's exact test. Continuous variables are expressed as the mean ± standard deviation or as the median ± interquartile range. Interval variables were analyzed using Student's t-test or Mann-Whitney U test. The cut-off values obtained based on the receiver operating characteristic curve for miR-509-3p, COL11A1 and miR-335 were optimized for their diagnostic sensitivity and speci city in predicting cancer progression or death. Survival was estimated using the Kaplan-Meier method and was compared using the log-rank test. Two-sided P-values < 0.05 were considered statistically signi cant. Cox proportional hazards models were implemented to estimate the hazard ratios (HRs) and 95% con dence intervals (CIs).

Declarations
Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Figure 1
Ten-year overall survival (OS)(A) and progression-free survival (PFS) (B). Kaplan-Meier curves strati ed by the miR-509-3p mRNA level and analyzed using a log-rank test.