Altered m6A modification is involved in up‐regulated expression of FOXO3 in luteinized granulosa cells of non‐obese polycystic ovary syndrome patients

Abstract The pathophysiology of polycystic ovary syndrome (PCOS) is characterized by granulosa cell (GC) dysfunction. m6A modification affects GC function in patients with premature ovarian insufficiency (POI), but the role of m6A modification in PCOS is unknown. The purpose of the prospective comparative study was to analyse the m6A profile of the luteinized GCs from normovulatory women and non‐obese PCOS patients following controlled ovarian hyperstimulation. RNA m6A methylation levels were measured by m6A quantification assay in the luteinized GCs of the controls and PCOS patients. Then, m6A profiles were analysed by methylated RNA immunoprecipitation sequencing (MeRIP‐seq). We reported that the m6A level was increased in the luteinized GCs of PCOS patients. Comparative analysis revealed differences between the m6A profiles from the luteinized GC of the controls and PCOS patients. We identified FOXO3 mRNA with reduced m6A modification in the luteinized GCs of PCOS patients. Selectively knocking down m6A methyltransferases or demethylases altered expression of FOXO3 in the luteinized GCs from the controls, but did not in PCOS patients. These suggested an absence of m6A‐mediated transcription of FOXO3 in the luteinized GCs of PCOS patients. Furthermore, we demonstrated that the involvement of m6A in the stability of the FOXO3 mRNA that is regulated via a putative methylation site in the 3’‐UTR only in the luteinized GCs of the controls. In summary, our findings showed that altered m6A modification was involved in up‐regulated expression of FOXO3 mRNA in the luteinized GCs from non‐obese PCOS patients following controlled ovarian hyperstimulation.


| INTRODUC TI ON
Polycystic ovary syndrome (PCOS) is a common endocrine disorder and a common cause of female infertility in women of reproductive age. 1 One of the main characteristics of the syndrome is granulosa cell (GC) dysfunction, which largely contributes to hyperandrogenism, abnormal follicle development and anti-Mullerian hormone excess. 2 Aberrant gene expression profile was found in GCs of PCOS patients. 3 Multiple differential expressed genes were high related to the pathogenesis of PCOS. However, the regulation mechanisms of these genes were largely unknown.
Forkhead Box O3 (FOXO3) plays important roles in diverse cellular processes including apoptosis, metabolism, cell proliferation and cell survival. 4 FOXO3 is regulated at several mechanistic levels, such as transcriptional activity, cellular localization, mRNA expression and protein stability. Oxidative stress induces FOXO activation and nuclear translocation by c-Jun N-terminal kinase (JNK) or mammalian Ste20-like kinase 1 (MST1) activation despite phosphorylation by protein kinase B (Akt). 5,6 After energy deprivation, the increased AMP/ ATP ratio leads to AMP-activated protein kinase (AMPK) activation. 7 AMPK activates FOXO3 activity by phosphorylation at six different residues. 8 Insulin-like growth factor-1 (IGF-I)/insulin and phosphoinositide 3-kinase (PI3K)/Akt signalling pathway inactivate FOXOs by phosphorylation resulting in FOXOs nuclear exclusion. 9 Zhao et al reported that elevated wnt family member 5A (WNT5a) activates PI3K/ Akt signalling in GCs of PCOS patients. 10 By contrast, Rice and colleagues showed metabolic insulin resistance in GCs of PCOS patients, suggesting that PI3K/Akt signalling is impaired. 11 While many studies have been focused on the regulation of FOXO3 activity by post-translational modifications, the regulation of FOXO3 expression is largely unknown. FOXO3 is one of the differential expressed genes which have higher expression in PCOS. 3 Its overexpression in GCs is associated with higher apoptosis in PCOS. 12 To date, the factors that up-regulate and activate FOXO3 in GCs of PCOS patients are still unclear. m 6 A is the most prevalent modification of mRNA in higher eukaryotes. 13 The modifications are reversible and dynamically regulated by the m 6 A modulators. m 6 A modulators consist of the 'writers', the 'erasers' and the 'readers'. Briefly, m 6 A modifications are installed by the 'writers' (Methyltransferase like 3 (METTL3) and Methyltransferase like 14 (METTL14)), removed by the 'erasers' (Fat mass and obesity-associated protein (FTO) and alkB homolog 5(ALKBH5)), and recognized by the 'readers' (YTH domain-containing proteins and Eukaryotic initiation factor 3 (eIF3)). 14-22 m 6 A modification has diverse biological functions such as nuclear RNA export, RNA splicing, protein translation regulation and RNA decay. In the YTHDF2-mediated decay pathway, mRNAs with increased m 6 A abundance in 3'-UTR are down-regulated due to reduced RNA stability. 19 Several human diseases are associated with altered m 6 A modification. m 6 A levels of sperm RNA are elevated in patients with asthenozoospermina. 23 Increased m 6 A levels in the granulosa cells were reported in patients with premature ovarian insufficiency (POI), affecting apoptosis and cell proliferation in GCs. 24 Aberrant m 6 A modification, through the effects on RNA metabolism, plays critical roles in a variety of cancers. 25 However, whether m 6 A modification plays a role in the pathogenesis of PCOS is unknown.
Given that m 6 A modification affects the GC function in patients with POI, 24 we suggested alteration of the m 6 A profile in the luteinized GCs of PCOS patients, which may account for the dysregulation of certain key genes for PCOS. Here, we showed the differences of m 6 A distribution between the luteinized GCs of normovulatory women and PCOS patients following controlled ovarian hyperstimulation. We identified FOXO3 mRNA with differential m 6 A peaks, targeted for decay by YTHDF2. Our results indicated that hypomethylated FOXO3 mRNA caused the dysregulation of FOXO3 in luteinized GCs from PCOS patients following controlled ovarian hyperstimulation.

| Measurement of hormones
Plasma levels of AMH, luteinizing hormone (LH), follicle-stimulating hormone (FSH), T and estradiol (E2) were collected and measured by chemiluminescence immunoassay (CLIA) between day 3 and day 5 of the menstrual cycle before the controlled ovarian stimulation. Serum fasting glucose levels and fasting insulin levels were measured by an oxidase-peroxidase method and a CLIA method, respectively. The inter-assay coefficients of variation were 6.1% for AMH, 7.2% for LH, 5.3% for FSH, 10.0% for T, 8.9% for E2, 4.3% for glucose and 5.4% for insulin. The intra-assay coefficients of variation were 3.8% for AMH, 5.3% for LH, 4.6% for FSH, 8.1% for T, 6.8% for E2, 2.1% for glucose and 2.9% for insulin. Insulin in follicle fluids was detected by enzyme-linked immunosorbent assay (ELISA, R&D Systems). HOMA-IR was calculated by the formula (HOMA-IR = fasting insulin (mIU/L) × fasting glucose (mmol/L)/ 22.5). 27

| Transvaginal ultrasonography
Ultrasound examination was performed on the 3rd-5th day of the menstrual cycle with a 7 MHz transvaginal transducer (LOGIC 400, General Electric Medical Systems) to calculate follicle number. The basal antral follicle count (AFC) was assessed as the sum of all follicles of 2-10 mm in diameter. The polycystic ovary was defined as 12 or more AFC in each ovary.

| RNA m 6 A quantification
Total RNA was isolated using the Total RNA Kit II Kit (Omega Bio-Tek).
Polyadenylated RNA was purified from total RNA using the GenElute mRNA Miniprep Kit (Sigma). 500 ng total RNA or polyadenylated RNA was used to determine the RNA m 6 A methylation levels using the EpiQuik m 6 A RNA Methylation Quantification Kit (Epigentek). Briefly, a standard curve was prepared according to the manufacturer's instruction. The RNA samples were coated on the strip wells, followed by incubation with capture antibody. After washes with washing buffer, detection antibody and enhance solution were added separately. Then, the signals were developed by the detection solution. The RNA m 6 A levels were quantified by the absorbance at 450 nm, and the m 6 A contents were calculated based on the standard curve. . The primers for MeRIP-qPCR were listed in Table S1.

| MeRIP-seq data analysis
Sequencing data were mapped to the reference genome (ftp:// ftp.ensem bl.org/pub/relea se-94/fasta/ homo_sapie ns/dna/) using BWA mem (version 0.7.12). m 6 A peaks were identified by peak finding algorithm in MACS2 (version 2.1.0). The threshold of enrichment was set at q < 0.05. Differential peak analysis was based on the fold enrichment of peaks between the PCOS group and the control group. When the odds ratio > 2, a differential peak was determined. The statistics of pathway enrichment of genes with differential peaks in KEGG pathway was tested using KOBAS software.

| Primary cell culture and RNA interference
The purified luteinized GCs were cultured in high-glucose Dulbecco's  Table S2.  Table S1.

| Western blot analysis
The total protein was extracted from the cultured GCs using RIPA buffer (Beyotime) containing 1mM PMSF (Beyotime). The concentration of the protein was measured using BCA protein assay

| Plasmid construction and dual-luciferase reporter assay
The proximal 3'-UTR of FOXO3 was amplified and inserted into

| Statistical analysis
Data are expressed as the means ± SD. The normality and homogeneity of variance of the data were assessed. Differences between groups were determined by one-way analysis of variance (ANOVA) followed by the Tukey multiple comparison test using SPSS 16.0 software package. Significance was set at P < 0.05.

| m 6 A levels in the luteinized GCs of the controls and PCOS patients
To investigate the role of m 6 A modification in the luteinized GCs of PCOS patients, we recruited normovulatory controls and PCOS F I G U R E 1 Quantification analysis of m 6 A modification in the luteinized GCs of the controls and PCOS patients. The m 6 A content in total RNA (A) and mRNA (B) of GCs were detected by colorimetric assay. Bars represent means ± SD, n = 5. *P < 0.05 vs the control patients in our reproductive centre (Table S3). The insulin-glucose parameters indicated that the PCOS patients had insulin resistance in the present study. We examined m 6 A levels in the total RNA of the luteinized GCs collected following ovarian hyperstimulation by RNA m 6 A quantification. m 6 A levels were twofold higher in the luteinized GCs of PCOS patients compared with the controls ( Figure 1A).
We further isolated polyadenylated RNA from the total RNA and examined the m 6 A levels. Consistently, the m 6 A levels of polyadenylated RNA were increased in the luteinized GCs of PCOS patients ( Figure 1B).

| Differential m 6 A modification in the luteinized GCs of the controls and PCOS patients
We conducted MeRIP-seq to analyse the transcriptome-wide distribution of m 6 A modification in GCs. 2764 and 3405 m 6 A peaks were identified in the controls and PCOS patients, respectively (Table S4).
We selected 8 transcripts with m 6 A peaks to verify the sequencing data by MeRIP-qPCR. The qRT-PCR results were in agreement with the sequencing data ( Figure S1). We analysed the differential peaks based on the fold enrichment of peaks of the two groups. 1719 and 2195 peaks were distinct in the controls and PCOS patients, respectively, while 996 peaks were overlapped in both groups ( Figure 2A and Table S5).
Similar to the other studies, m 6 A peaks were strongly enriched around the stop codon in the controls ( Figure 2B,D). [28][29][30] However,

| m 6 A modification targets the FOXO3 transcript in the luteinized GCs of the controls but not PCOS patients
We suggested that m 6

| m 6 A modification regulates the stability of the FOXO3 transcript via the YTHDF2-mediated decay pathway in the luteinized GCs of the controls
Our MeRIP-seq showed a differential m 6 A peak in 3'-UTR and near the stop codon of the FOXO3 transcript ( Figure 4A). We performed MeRIP-qPCR to confirm the differential m 6 A levels on the m 6 A target site ( Figure 4B). To determine how m 6 A modification regulates the FOXO3 transcript, we analysed the sequence of the m 6 A peak.
A putative m 6 A site was identified within the m 6 A peak of 3'-UTR at position + 2279 (+1 relative to the translation start site) ( Figure 4C).
As YTHDF2 is the main binding protein that account for the decay of m 6 A-modified mRNAs, we next examined the effects of YTHDF2knockdown on FOXO3 expression in human GCs. Depletion of YTHDF2 increased the amount of FOXO3 transcript, and total protein levels of FOXO3 ( Figure 4D,E). The knockdown efficiency was checked by qRT-PCR ( Figure S2). To assess the functionalities of the m 6 A site in the 3'-UTR of FOXO3 transcript, we constructed a reporter plasmid bearing FOXO3-3'-UTR with the putative m 6 A site mutated ( Figure 4F). FTO-siRNA suppress the luciferase activities of the reporter plasmid with the wild-type m 6 A site ( Figure 4G).
However, mutation of the m 6 A site reversed the suppressive effects of FTO-siRNA on the luciferase activities ( Figure 4G). The presence of the m 6 A readers in human GCs was showed by a previous study. 32 To exclude the potential of lacking the m 6 A readers in GCs of PCOS patients, we examined the expression of the m 6 A readers in PCOS.
The expression of YTHDF2 was significantly increased in the luteinized GCs of PCOS patients ( Figure S3). These results indicated that m 6 A modification regulated FOXO3 mRNA decay through the m 6 A site in 3'-UTR in the luteinized GCs of the controls.

| D ISCUSS I ON
Increasing evidence showed that m 6  The site selection of m 6 A modification is largely unknown.
miRNAs regulate m 6 A formation at corresponding target sites. 37 Altered miRNAs and other associated factors partially explain the differences of m 6 A profiles among cell types. 37 Furthermore, m 6 A profile was altered when cells exposed to heat shock, ultraviolet radiation or signalling molecules. 38 Acute stress regulates a fraction of genes with m 6 A modification in the cortex. 33 The GCs of PCOS  KEGG analysis showed that differential peaks were enriched in metabolic pathway ( Figure S4). Whether m 6 A modification is involved in the regulation of these transcripts with the differential m 6 A modification needs further investigation.
Insulin resistance is prevalent in PCOS patients and plays a cru-  In summary, the present study provided m 6 A profiles of normovulatory women and PCOS patients in luteinized GCs following controlled ovarian hyperstimulation. We demonstrated that altered m 6 A modification disturbed the regulation of FOXO3 expression in the luteinized GCs of PCOS patients. However, the site selection mechanism of m 6 A in PCOS needs to be explored in future studies.

ACK N OWLED G EM ENTS
We thank our director Prof. Xuefeng Huang for supporting the project. We are indebted to Yan Li for help with the follicular fluid collection. The work was supported by grants from the National Natural Science Foundation of China (81903293 to WN), National Key R&D Program of China (2017YFC1001604) and Natural Science Foundation of Zhejiang Province (LQ19H040003 to SZ).

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.