Neuroendocrine characteristics of induced pluripotent stem cells from polycystic ovary syndrome women

Polycystic ovary syndrome (PCOS) is a common female reproductive endocrinopathy that afflicts up to 10%–15% of women in reproductive age worldwide (Nestler, 2016). Women with PCOS exhibit hyperandrogenism, intermittent/ absent menstrual cycles, and polycystic ovaries on ultrasound (Rotterdam, 2004). The pathophysiology of PCOS extends beyond infertility and hirsutism to hypothalamic neuroendocrine dysfunction (Goodarzi et al., 2011). Most women with PCOS exhibit increased luteinizing hormone (LH) levels, resulting from high-frequency gonadotropin-releasing hormone (GnRH) secretion (Cimino et al., 2016). Prenatal testosterone (T) treatment in sheep results in disrupted steroid feedback on gonadotropin release, which increases pituitary sensitivity to GnRH and subsequently leads to LH hypersecretion (Sullivan and Moenter, 2004; Cardoso et al., 2016). A recent study shows that GnRHdependent LH pulsatility and secretion are elevated by antiMüllerian hormone (AMH) in PCOS disease. The increased prenatal AMH reprograms fetus and induces PCOS in adults (Tata et al., 2018). Furthermore, the androgen receptor (AR) plays a role in hyperandrogenism and ovarian folliculogenesis in PCOS (Wang et al., 2015; Abbott, 2017). However, the disease mechanism behind PCOS remains unclear, and current management focuses on treating the symptoms but not the mechanism (Chen et al., 2016; Shi et al., 2018). A further understanding of this disease is necessary to uncover the pathology of PCOS and develop new potential therapeutic avenues and drugs. To establish a cell model for investigating PCOS disease, we first generated iPSCs from the skin tissues of three PCOSand three non-PCOS patients. The primary fibroblasts of the patients were transduced with lenti-viral vectors expressing OCT4, SOX2, C-MYC and KLF4 according to a direct reprogramming protocol (Okita et al., 2011) (Fig. S1A). The generated iPSCs displayed typical pluripotent stem cell morphology (Fig. 1B), expressed pluripotency markers of OCT4, NANOG and SOX2 (Fig. 1C), and displayed normal karyotypes (Fig. S1B). These results showed successful derivation of iPSCs from patients. Next, the total RNA was extracted from PCOSand nonPCOS-derived iPSCs for RNA microarray analysis. The global transcriptional genes of PCOS-derived iPSCs were identified and enriched by gene ontology (GO). Filtering by P value less than 0.01 and fold-changes of more than 2, 2,904 differentially expressed genes (DEGs) were collected between PCOSand non-PCOS-derived iPSCs. Among these DEGs, we were interested in genes related to neuroendocrine and metabolic processes in PCOS. The enriched top 40 upand down-regulated genes were shown in heatmaps (Fig. 1D). The GO enrichment revealed that downregulated transcripts were associated with neurogenesis, enteroendocrine cell differentiation and LDL particle binding process. Up-regulated transcripts were enriched in neural crest cell development, progesterone receptor pathway and cholesterol storage process (Fig. 1E). We focused on genes of GABA receptor, CYP family, TGF-β pathway and estrogen receptor pathway in neuroendocrine processes (Fig. 1F). Then seven significantly modulated genes (FBP1, PYGL, GAPDH, KDM1A, STAT5, GPI and UGP2) were verified by RT-qPCR, according to functional annotation and their fold change (Fig. 1G). Moreover, the protein levels of above genes were verified by Western blot in PCOS-derived iPSCs, demonstrating the similar changes with mRNA levels (Fig. 1H). These genes (FBP1, PYGL, GAPDH, GPI and UGP2) involved in glycolysis process were expressed abnormally, indicating deregulation of glucose metabolism (glycolysis and gluconeogenesis) in PCOS (Fig. 1J). In addition, we measured the expression levels of testosterone (T) and estradiol (E2) in the cultures of iPSCs via ELISA (enzyme-linked immunosorbent assay), to verify the neuroendocrine characteristics in PCOS. The results showed that PCOS-derived iPSCs secreted greater T significantly than non-PCOS-derived iPSCs, while there were no significant differences in E2 level between PCOS and non-PCOSderived iPSCs (Fig. 1H). The change of T level was consistent with the clinical hyperandrogenic feature of PCOS (Haouzi et al., 2012). In order to compare the differences of metabolic function after reprogramming, we measured the mitochondrial respiration ability of ovarian granulosa cells


Generation and culture of PCOS-derived iPSCs
Human fibroblasts were derived from skin cells of patients. The iPSC clones were reprogrammed as described previously. Briefly, epithelial cells were transduced with OCT4-, SOX2-, KLF4-and C-MYC-expressing lentiviral vectors in MEF (mitomycin-C-treated mouse embryonic fibroblasts) medium without serum. On day 5, the medium was replaced with iPSC medium [DMEM/F12 supplemented with 20% (v/v) knock out serum replacer (Knockout SR), 2 mM L-glutamine, 2 mM non-essential amino acids, and 0.1 mM β-mercaptoethanol (Invitrogen) with no additional bFGF]. Putative PCOS-derived iPSC colonies were emerged within 21 days after transduction. Generated colonies were mechanically dissociated for passage.

Neural stem cell differentiation from iPSCs
NSC differentiation was performed as described previously. iPSC were picked from the MEF feeder and suspended cultured for 4 days in iPSC medium without bFGF, to induce embryoid bodies (EB). For NSC differentiation, RA (Retinoic acid) was added at final concentration of 1-2 µM after EB formation. After 4 days as a floating culture, EBs were collected and plated onto matrigel coated dishes cultured in EB medium without RA.

Microarray analysis
Microarray hybridization was carried out at CapitalBio (Beijing, China). Total RNA (100 ng) was used to prepare twice-amplified and labelled RNA for hybridization with HG-UI33 plus 2.0 arrays (n=3).

Quantitative real-time PCR
According to the manufacturer's protocol, total RNA was extracted using Trizol reagent, and cDNA was synthesized by a ReverAid First Strand cDNA Synthesis Kit (Invitrogen). The PCR products were amplified using the SYBR Green mix kit by an QuantStudio3 system (Applied Biosystems). Fold-change by RT-PCR was measured and calculated by normalizing to the housekeeping gene (β-actin) and calculating the

Immunofluorescence staining
Cells were fixed in 4% (w/v) paraformaldehyde in phosphate-buffered saline (PBS) for 20 min and then blocked with PBS containing Triton X-100 (Sigma-Aldrich) for 30 min at room temperature. After blocking, the cells were incubated with primary antibodies overnight at 4 °C, then incubated with secondary antibodies 1hr at room temperature. The primary antibodies (all at a 1:250 dilution, Abcam) were used to detect OCT4, SOX2, and NANOG expression of iPSC. NSC was stained by primary antibody of SOX2 and NESTIN (proteintech). We visualized antigen localization using goat anti-mouse/rabbit Alexa Fluor 488, 555 and 594. The nuclei were stained with Hoechst (Invitrogen).

ELISA
The culture medium of iPSC were collected and centrifuged with 12,000 rpm in 4 ℃ before assays. The testosterone and estrodiol levels were detected using ELISA kit (R&D system).

Mitochondrial oxygen consumption detection
The mitochondrial respiration was detected using the XF24 extracellular flux analyzer from Seahorse Bioscience (Billerica). A classical Mito stress test was performed based on the following procedure: (1) the basal respiration was measured before adding chemicals; (2) oligomycin (2.0 μM) was added to inhibit ATP production; (3) The maximal respiration was measured by adding the uncoupler carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP); and (4) rotenone and antimycin A (0.5 μM) were applied in combination to block respiration. The final results were normalized to cell number (10 5 cells per well).

Statistical analysis
The microarray data were analysed to identify statistically significantly different expression. The list of identified genes (fold change, FC>2; FDR<0.05) was submitted to AmiGO2 and DAVID database to identify the Gene Ontology (GO) terms associated with biological functions and pathways. Data are presented as the mean±SD and were analysed using GraphPad Prism 5 program (GraphPad Software) and R language package.