Impaired Mitochondrial Biogenesis Inhibits Epithelial-Mesenchymal Transition in Villi of PCOS Patients

Abstract

Context

Polycystic ovary syndrome (PCOS) is accompanied by impaired mitochondrial biogenesis in the ovary and uterus. Whether impaired mitochondrial biogenesis exhibits in villi of PCOS, and its effect and underlying mechanism remain unclear.

Objective

This work aimed to investigate mitochondrial biogenesis status and effect on villi of PCOS patients.

Methods

Placenta RNA-sequencing data of PCOS downloaded from the GEO database was analyzed with Gene Set Enrichment Analysis (GSEA). GSEA results were validated in first-trimester villi of 8 PCOS patients with euploid miscarriage and 22 matched controls. The function and impact of mitochondrial biogenesis on trophoblast cells were investigated using human trophoblast cell lines HTR-8/SVneo and BeWo.

Results

Mitochondria-related and epithelial-mesenchymal transition (EMT) pathways were enriched in placentas of PCOS. In villi of PCOS patients with euploid miscarriage, reduced mitochondrial DNA copy number (mtDNA CN) and N-cadherin protein level, and an elevated E-cadherin protein level were detected, indicating mitochondrial biogenesis dysfunction and impaired EMT. 5 α-Dihydrotestosterone (DHT) exposure downregulated mtDNA CN via reducing mitochondrial transcription factor A (TFAM) level, a critical transcription factor of mtDNA, in HTR-8/SVneo cells. Decreased expression level of TFAM was observed in villi of PCOS. Knockdown of TFAM significantly impeded EMT, characterized by decreased levels of N-cadherin and vimentin in HTR-8/SVneo cells, and increased level of E-cadherin in BeWo cells. Reduction of reactive oxygen species (ROS) mitigated TFAM knockdown-induced impairment of EMT via increasing nuclear Yes-associated protein level in trophoblast cells.

Conclusion

The villi of PCOS patients with euploid miscarriage exhibited impaired mitochondrial biogenesis. Androgen-induced downregulation of TFAM impeded EMT via ROS/YAP axis in trophoblast cell.

Polycystic ovary syndrome (PCOS) is a common and complex endocrine disorder, occurring in 5% to 20% of reproductive-aged women (1). The characteristics of patients with PCOS include hyperandrogenism, ovarian dysfunction, and polycystic ovarian morphology. Numerous studies have shown that PCOS patients encounter elevated risk of pregnancy complications, such as miscarriage, preeclampsia, and preterm birth (2-6). Especially, PCOS patients are prone to suffer from miscarriage. The risk of miscarriage is up to 11 times higher in pregnant individuals with PCOS than in those without PCOS (5-9). Patients with PCOS have a lower risk of embryonic aneuploidy (10, 11). Even being transferred euploid embryos, patients with PCOS have a 2-fold higher risk of developing early miscarriage vs non-PCOS patients (12). Notably, hyperandrogenism, a classical phenotype of PCOS, is closely associated with an increased risk of miscarriage in patients with PCOS. It was observed that in PCOS-like mouse models, androgen excess exposure caused disrupted placental development and function (13). However, the effect and molecular mechanism of hyperandrogenism on the human placenta has not been fully explored (14-16).

Endocrine disorder in PCOS patients causes aberrant placenta structure resulting in epigenetic changes (17, 18). Disrupted placenta development is implicated in pregnancy complications of patients with PCOS. Growing evidence has revealed that in women with PCOS, placentas exhibit distinct aberrations no matter from first trimester or delivery, such as irregular morphology, fibrosis, impaired invasion of trophoblast cells; and hyperandrogenism contributes to placenta dysfunction (19-22). Likewise, in PCOS-like mice, androgen exposure led to a noticeable increase in absorbed embryos, severely compromised placenta development, and impaired invasion of trophoblast cells (13, 23). In addition to proliferation and apoptosis, the invasion of trophoblast cells is considered one of the critical biological functions in maintaining normal pregnancy. Trophoblast cells invade decidua and remodel uterine spiral arteries to ensure adequate placental perfusion (24, 25). Inadequate trophoblast invasion is associated with two-thirds of early spontaneous miscarriages (26). Various growth factors, cytokines, and transduction pathways regulate the invasion of trophoblast cells. In particular, epithelial-mesenchymal transition (EMT) plays a pivotal role in facilitating trophoblast invasion (27, 28). Abnormal EMT of trophoblast cells impedes embryo implantation and even causes miscarriage (29). However, whether impaired EMT is implicated in first-trimester villi of PCOS patients with euploid miscarriage remains to be investigated.

Trophoblast cells consume 40% of oxygen transported to the maternal-fetal interface, highlighting the active oxidative phosphorylation (OXPHOS) function in placentas (30). Mitochondrial biogenesis is essential for encoding most OXPHOS protein subunits. Most OXPHOS protein subunits are encoded via mitochondrial biogenesis. Mitochondrial biogenesis refers to the process of regulating mitochondrial quantity through transcription, replication, and translation of mitochondrial DNA (mtDNA), and import of nuclear-encoded protein. Mitochondrial biogenesis plays a critical role in maintaining abundant mitochondrial content and providing sufficient adenosine triphosphate to meet the energy demand of cells (31, 32). Nuclear genomes play a critical role in regulating mtDNA transcription via several signaling pathways (33). Abnormal mitochondrial biogenesis, characterized by the level of mtDNA, is associated with a high risk of pregnancy complications, such as miscarriage (34, 35). A conditional knockout of the gene regulating mtDNA copy number (CN) has been demonstrated to result in defects in placental development and even embryonic death (36, 37). In addition to detecting mtDNA CN to assess mitochondrial biogenesis, several measures are useful to assess mitochondria. For example, mitochondria are a critical source of reactive oxygen species (ROS) production. ROS production causes a detrimental effect on mitochondrial proteins, other organelles, and DNA. Moreover, mitochondrial membrane potential reflects the function of mitochondria. Abnormal mitochondrial membrane potential indicates a compromised electron transport chain, which leads to an increased level of ROS and disrupts the uptake of calcium (38).

Compelling evidence has shown that mitochondrial biogenesis dysfunction affects the reproductive system and peripheral blood of patients with PCOS (39). In particular, placentas of PCOS-like pregnant rats were characterized by impaired mitochondrial biogenesis and low expression levels of several nuclear genes that regulated mitochondrial biogenesis (23, 40). However, the profile and effect of mitochondrial biogenesis on the villi of PCOS patients is completely unknown. Of note, studies have demonstrated that in tumor cells, abnormal mitochondria promoted or inhibited EMT to regulate invasion capability (41-43). However, the interaction between mitochondrial biogenesis and EMT in human trophoblast cells remains uncertain. In addition, accumulating evidence has shown that androgen excess impaired mitochondrial biogenesis in different kinds of tissues and cells, including placentas of PCOS-like pregnant rats (23, 40, 44-46). The effect of hyperandrogenism on mitochondrial biogenesis in trophoblast cells needs more research.

In the present study, we investigated the role of mitochondrial biogenesis and EMT in the pathogenesis of first-trimester villi from PCOS patients with euploid miscarriage. The molecular mechanism underlying disordered mitochondrial biogenesis-induced impaired EMT was investigated in trophoblast cell lines. Our findings provide novel insight into the crosstalk between impaired mitochondrial biogenesis and EMT in villi of PCOS patients with euploid miscarriage.

Materials and Methods

Gene Set Enrichment Analysis

RNA-Seq data (GSE154274) of placentas from PCOS patients (n = 3) and non-PCOS patients (n = 3) were downloaded from the GEO database. These placentas were collected from patients who underwent elective cesarean delivery surgery during 38 to 40 weeks of gestation. Gene Set Enrichment Analysis (GSEA) was performed based on GSEA 4.2.3 software (47). The hallmark gene set was obtained from the MSigDB database as the predefined gene set. The bubble plot of signal pathways was drawn by R 4.2.0 software. The vertical axis of the bubble plot showed enrichment pathway names. The horizontal axis showed normalized enrichment scores (NES). The bubble size represented the number of enriched genes in the signal pathway. The bubble color depth indicated nominal P values. Nominal P values less than .05 and false discovery rate q-value less than .5 were considered statistically significant.

Collection of First-Trimester Villous Tissue

This study was reviewed and approved by the ethics committee of the First Affiliated Hospital of Sun Yat-sen University (No. 2021722). Informed consent forms were signed before enrollment. Patients with PCOS were diagnosed according to Rotterdam criteria (48). Exclusion criteria were as follows: 1) patients with diabetes, adrenal hyperplasia, hypertension; 2) gestational trophoblastic disease and other tumors; 3) receiving medical abortion treatment; 4) uterine anatomical defects or cervical incompetence; 5) aneuploid villi from patients with PCOS; or 6) other known causes of miscarriage. Finally, human first-trimester villous tissues were collected from PCOS patients with euploid miscarriage (PCOS with euploid miscarriage group, n = 8) and aged- and gestational week–matched healthy control women (control group, n = 22) from May 2021 to January 2023. The control group included women requiring pregnancy termination for personal reasons between 7 and 8 weeks of gestational age (according to the last menstrual period and crownrump length of the fetus). All samples were processed within 2 hours of collection. The euploidy of villi from the PCOS patients was confirmed by chromosomal microarray analysis. The PCOS with euploid miscarriage group comprised 8 samples for messenger RNA (mRNA) analysis, 7 of which were also used for DNA and protein analysis. The control group (22 samples in total) consisted of 15 samples for mRNA analysis, 6 of which were also used for protein analysis; and an additional 7 samples for DNA analysis.

Cell Culture

Human trophoblast cell lines commercially available included HTR-8/SVneo (CL-0765, Pricella) and BeWo (CL-0500, Pricella). EMT-related markers N-cadherin and vimentin proteins are exclusively expressed in HTR-8/SVneo cells rather than in BeWo cells, while E-cadherin protein is exclusively expressed in BeWo cells rather than in HTR-8/SVneo cells (49). HTR-8/SVneo cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (C11995500BT, Gibco) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. BeWo cells were grown in DMEM/F-12 (C11330500BT, Gibco) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. All cells were cultured at 37 °C in a humidified atmosphere containing 5% CO₂.

When trophoblast cells reached 50% to 70% confluence, treatments were initiated as follows: 5 α-Dihydrotestosterone (DHT, Selleck) was dissolved in dimethyl sulfoxide (DMSO) at a stock concentration of 10 μM. DHT was used at concentrations ranging from 0.01 to 1000 nM for 3, 6, 24, and 48 hours. Veterporfin (HY-B0146, MedChemexpress), a classic inhibitor of Yes-associated protein (YAP), was dissolved in DMSO and used at concentrations ranging from 10 to 1000 nM for 24 hours. Rotenone (HY-B1756, MedChemexpress), a specific inhibitor of mitochondrial electron transport chain complex I, was used as a mitochondrial ROS generation agent at a concentration of 200 nM for 24 hours. Mito-TEMPO (HY-112879, MedChemexpress) was used at concentrations ranging from 100 nM to 10 μM for 24 hours. To induce syncytialization of BeWo cells, 50 μM forskolin (HY-15371, MedChemexpress) was used to treat BeWo cells for 48 hours.

Cell Transfection

The specific short interfering RNA (siRNA) oligonucleotides and plasmids were designed and synthesized by Shanghai GenePharma Co, Ltd. Knockdown of TFAM was achieved by transfecting HTR-8/SVneo and BeWo cells with siRNA using Lipofectamine 2000 (L3000150, Invitrogen) according to the manufacturer’s instructions. TFAM overexpression was performed by transfecting HTR-8/SVneo cells with plasmid using the Lipofectamine 2000.

RNA Extraction and Real-time Quantitative Polymerase Chain Reaction

Total RNAs of trophoblast cell lines and villus were extracted using the FastPure Cell/Tissue Total RNA Isolation Kit V2 (RC112-01, Vazyme). The concentration and purity of the RNAs were quickly assessed by a spectrophotometer NanoDrop (DeNovix). Total RNAs were further reversely transcribed into complementary DNAs with a HiScript III RT SuperMix kit (R232-01, Vazyme). Real-time quantitative polymerase chain reaction (RT-qPCR) and data analysis were performed using a QuantStudio 5 instrument (Applied Biosystems) with a SYBR qPCR Master reagent kit (Q712-03, Vazyme). The qPCR reactions were as follows: 1) initial denaturation stage, activating Hot Start DNA Taq Polymerase at 95 °C for 30 seconds; 2) 40 cycles of amplification stage, consisting of denaturation at 95 °C for 5 seconds and annealing and extension at 60 °C for 15 seconds; 3) melting curve stage, consisted of 95 °C for 15 seconds, 60 °C for 60 seconds, and 95 °C for 15 seconds. The 2^−ΔΔCt method was applied to calculate the relative expression levels of target genes. RNA expression levels were normalized against ACTB mRNA. Primer sequences are listed in Table 1. The experiments were performed at least in triplicate.

DNA Extraction and Detection of Mitochondrial DNA Copy Number

Total DNAs of trophoblast cell lines and villus were extracted using a FastPure Cell/Tissue DNA Isolation Mini Kit (DC112, Vazyme) following the protocol of the manufacturer. The concentration and purity of the DNAs were assessed by a spectrophotometer NanoDrop (DeNovix). The mitochondrial biogenesis was assessed by detecting mtDNA CN. The mtDNA CN was measured by qPCR using a QuantStudio 5 instrument (Applied Biosystems), as described in other studies (50, 51). The mtDNA-specific primers included NADH dehydrogenase subunit 1 (ND1) and cytochrome c oxidase subunit 3 (COX3). The nuclear gene β-2-microglobulin (β2 M) served as a loading control. Primer sequences were listed in Table 1. mtDNA CN (ND1) and mtDNA CN (COX3) both represented mitochondrial biogenesis, respectively. The mtDNA CN was calculated using the following formulae: 1) ΔCt (β2M-ND1) = CT_β2M—CT_ND1; mtDNA CN (ND1) = 2^{ΔCt (β2M−ND1)}; and 2) ΔCt (β2M-COX3) = CT_β2M—CT_COX3; mtDNA CN (COX3) = 2^{ΔCt(β2M−COX3)}. The results were expressed as relative mtDNA CN (ND1) and (COX3). The experiments were performed at least in triplicate.

Protein Extraction and Western Blotting

The protein extraction and Western blot analysis protocol have been described in other research (40). In brief, whole proteins from trophoblast cell lines and villi were extracted using a radioimmunoprecipitation assay buffer (RIPA, PC101, Epizyme) containing 1% protease inhibitor cocktail (GRF101, Epizyme) and 1% phosphatase inhibitor (GRF102, Epizyme). Nuclear and cytoplasmic proteins from cells were extracted using the ExKine Nuclear and Cytoplasmic Protein Extraction Kit (KTP3001, Abbkine) following the manufacturer's instructions. The protein concentration was quantified using a BCA Protein Assay Kit (ZJ101, Epizyme). Equal amounts of protein were separated by 10% to 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (PG112, PG113, Epizyme) and transferred to poly (vinylidene fluoride) membranes (IPVH00010, Merck Millipore). After blocking with protein-free rapid blocking buffer (PS108, Epizyme), the membrane was incubated overnight at 4 °C with primary antibodies, including anti-GAPDH (1:50 000; Proteintech catalog No. 60004-1-Ig, RRID:AB_2107436), anti-Histone H3 (1:2000; Cell Signaling Technology catalog No. 4499, RRID:AB_10544537), anti-TFAM (1:10 000; Abcam catalog No. ab176558, RRID:AB_3676081), anti-N-Cadherin (1:1000; Cell Signaling Technology catalog No. 13116, RRID:AB_2687616), anti-E-Cadherin (1:1000; Cell Signaling Technology catalog No. 3195, RRID:AB_2291471), anti-vimentin (1:20 000; Proteintech catalog No. 60330-1-Ig, RRID:AB_2881439) anti-phospho-YAP (1:1000; Cell Signaling Technology catalog No. 13008, RRID:AB_2650553), and anti-YAP (1:1000; Cell Signaling Technology catalog No. 14074, RRID:AB_2650491). Then, the membranes were incubated for 2 hours at room temperature with horseradish peroxidase–conjugated secondary antibodies, including goat anti-rabbit immunoglobulin G (IgG) (1:5000; Proteintech catalog No. SA00001-2, RRID:AB_2722564) and goat anti-mouse IgG (1:5000; Proteintech catalog No. SA00001-1, RRID:AB_2722565). Immunoreactive protein bands were visualized using an Omni-ECL femto light chemiluminescence kit (SQ201, Epizyme) with a chemiluminescence intelligent imaging system (Life Technology). Quantitative analysis of bands was performed for gray scale values using ImageJ software. When necessary, membranes were stripped using stripping buffer (PS107, Epizyme) and then reincubated with another primary antibody.

Transmission Electron Microscopy Detection

Freshly collected villi were fixed in 2.5% glutaraldehyde (G1102, Servicebio) and post-fixed with 1% osmium tetroxide. The samples were gradient dehydrated in an ascending series of ethanol. Then, tissues were embedded in epoxy resin. Serial ultrathin sections were prepared and collected on copper grids. Sections were stained with uranyl acetate and lead citrate. The sections were examined and imaged with a transmission electron microscope (FEI spirit T12). The mitochondria structure of trophoblast cells was observed.

Immunofluorescence Assay

Sections 4 μm thick were dewaxed and dehydrated. Following antigen repair, the sections were permeabilized with 0.1% Triton X-100 for 10 minutes, and blocked with goat serum (AR0009, Boster) for 30 minutes. After blocking, sections were incubated overnight at 4 °C with diluent primary antibodies, including anti-TFAM (1:200; Abcam catalog No. ab176558, RRID:AB_3676081) and anti-hCG (1:200; Proteintech catalog No. 60334-1-Ig, RRID:AB_2881443). Then, sections were incubated for 1 hour at room temperature with secondary antibodies, including goat anti-rabbit IgG conjugated to Alexa Fluor 488 (1:500; Abcam catalog No. ab150077, RRID:AB_2630356) and goat anti-mouse IgG conjugated to Alexa Fluor 647 (1:500; Abcam catalog No. ab150115, RRID:AB_2687948). The sections were then stained with DAPI (1:10 000; Abcam catalog No. ab228549) for 5 minutes. All images were captured using a confocal laser scanning microscopy (FV3000, Olympus).

JC-1 Assay

The mitochondrial membrane potential of trophoblast cells was detected by a JC-1 assay kit (ab113850, Abcam) following the manufacturer's instructions. Briefly, HTR-8/SVneo cells were incubated with JC-1 reagent for 10 minutes at 37 °C in the dark, while the positive control group was previously incubated with FCCP (carbonylcyanide-p-trifluoromethoxyphenylhydrazone) for 4 hours. After washing with dilution buffer, changes in the mitochondrial membrane potential of cells were observed with an inverted fluorescence microscope (IX83, Olympus). The presence of green fluorescence indicated low mitochondrial membrane, while red fluorescence indicated high membrane potential.

Cell Counting Kit-8 Assay

HTR-8/SVneo cells were plated into 96-well plates at 5000 cells per well. The cell proliferation was assessed at 0, 24, 48, 72, and 96 hours using a Cell Counting Kit-8 (CCK-8) kit (Vazyme, A311-01) following the manufacturer's instructions. Absorbance at 450 nm was measured using a spectrometer (Varioskan LUX, Thermo Fisher). The experiments were performed at least in triplicate.

5-Ethynyl-2′-Deoxyuridine Assay

The proliferation of HTR-8/SVneo cells was assessed using an 5-ethynyl-2′-deoxyuridine (EdU) Imaging Kit (K1075, APExBIO) following the manufacturer's instructions. Briefly, cells seeded on glass slides were exposed to EdU reagent for 2 hours at 37 °C. Further, slides were fixed with 4% paraformaldehyde for 15 minutes and permeabilized with 0.5% Triton X-100 for 15 minutes. Then, slides were incubated with EdU click-reaction buffer for 30 minutes at room temperature in the dark. Nuclei were stained with Hoechst for 15 minutes. Images were obtained with a fluorescence microscope (BX63, Olympus).

Apoptosis Assay

The apoptosis of HTR-8/SVneo cells was assessed using an apoptosis detection kit (Biolegend, 640930) according to the manufacturer's instructions. Briefly, cells were mixed with 100 μL Annexin V Binding Buffer and then incubated with 5 μL Annexin V-APC and 5 μL 7-AAD for 15 minutes at room temperature in dark. The fluorescence intensity was detected by flow cytometry (Invitrogen, Attune Nxt). Data analysis was performed using FlowJo software (version 10.8.1), and early or late apoptosis rates were calculated. The experiments were performed at least in triplicate.

Phalloidin Staining

Cytoskeleton actin filaments were stained using an F-actin Staining Kit (Abbkine, KTC4008). Briefly, HTR-8/SVneo cells were seeded on coverslips, fixed with paraformaldehyde for 20 minutes, and permeabilized with 0.1% Triton X-100 for 10 minutes. Then, slips were incubated with 100 µL phalloidin staining solution (green fluorescence) for 30 minutes in the dark. Nuclei were stained with DAPI (ab228549, Abcam) for 5 minutes. Cells were imaged using a confocal laser scanning microscopy (FV1000, Olympus).

Mito-SOX Assay

The mitochondrial superoxide formation in HTR-8/SVneo cells was detected using MitoSOX Red dye (Thermo Fisher, M36008). Cells were incubated with 5 μM MitoSOX solution for 20 minutes at 37 °C in the dark. The fluorescence intensity was detected by flow cytometry (Invitrogen, Attune Nxt). Relative mean fluorescence intensity of mitoSOX was analyzed with FlowJo software (version 10.8.1).

Obtaining a Gene Set Containing Androgen Receptor Binding Sites and Regulating Mitochondrial DNA

Through the Gene Transcription Regulation Database (GTRD), a set of 43 963 genes containing androgen receptor binding sites was downloaded. In addition, from the MitoCarta3.0 database, a nuclear gene set containing 1136 genes implicated in mitochondrial function was acquired. Subsequently, by intersecting analysis of these two gene sets, we obtained a novel gene set containing 1062 genes that both contained androgen receptor binding sites and were involved in regulating mitochondrial function. Within this novel gene set, we used functional annotation analysis to identify 13 genes that not only regulate mitochondrial DNA but also contain androgen receptor binding sites.

Statistical Analysis

Statistical analysis was performed using SPSS software (version 26.0). Data normality was checked using the Shapiro-Wilks normality test. Normally distributed continuous variables were expressed by mean ± SD; nonnormally distributed continuous variables were represented by medians. Categorical variables were expressed as ratios or percentages. As for continuous variables with normal distributed, the t test, paired t test, or one-way analysis of variance analysis was used to compare differences between groups. For nonnormally distributed continuous variables, the Mann-Whitney U test was employed; as for categorical variables, the chi-square test was used. All P values are 2-sided with P less than .05 indicating that the difference was statistically significant. In the figures, P values are indicated by asterisks: *P less than .05; **P less than .01; and ***P less than .001.

Results

Identification of Significant Enriching Pathways in Placentas of Polycystic Ovary Syndrome Patients

Although closely associated with the occurrence of miscarriage, the pathogenesis of impaired invasion of trophoblast cells in PCOS patients remains unclear. To explore the signaling pathways enriching in the placentas of PCOS patients (n = 3) and non-PCOS individuals (n = 3), we applied GSEA to analyze RNA-Seq data downloaded from the GEO database. The results showed that a total of 10 gene sets were upregulated and 40 gene sets were downregulated in the PCOS patients, compared with the non-PCOS individuals. According to NES, the first 10 upregulated gene-enriching pathways (NES > 0) and downregulated gene-enriching pathways (NES < 0) are shown in the bubble plot (Fig. 1A). Further, in the PCOS with euploid miscarriage group, the significant enrichment pathway that upregulated genes were enriched in was the mitochondrial OXPHOS pathway, and pathways that downregulated gene sets were enriched in were the EMT pathway and APICAL_surface pathway (Fig. 1B-1D). In particular, the pathways with the highest NES were the OXPHOS pathway and the EMT pathway, respectively. These results strongly suggest that in the placentas of the PCOS patients, mitochondrial and EMT functions were different compared with the control group.

Impaired Mitochondrial Biogenesis and Abnormal Ultrastructure of Mitochondria in First-Trimester Villi From Polycystic Ovary Syndrome Patients With Euploid Miscarriage

The GSEA results indicated that in the placentas of patients with PCOS, there was a significant change in the mitochondrial OXPHOS pathway. Given the critical role of mitochondrial biogenesis in encoding OXPHOS protein subunits, we investigated mitochondrial biogenesis in villi from PCOS patients with euploid miscarriage (n = 7) and the control group (n = 7) to determine whether mitochondrial biogenesis is implicated in miscarriages of PCOS patients. As hallmarks of mitochondrial biogenesis, the level of mtDNA CN (ND1) and mtDNA CN (COX3) were detected by RT-qPCR. As shown in Fig. 2A and 2B, in villi of PCOS patients with euploid miscarriage, the mtDNA CN (ND1) and mtDNA CN (COX3) were significantly lower than those of the control group (0.80 ± 0.51 vs 1.00 ± 0.52, P = .024; 0.74 ± 0.50 vs 1.00 ± 0.56 P = .001, respectively). These results indicated impaired mitochondrial biogenesis in the villi of the PCOS with euploid miscarriage group. In addition, the ultrastructure of mitochondria was observed using the transmission electron microscopy assay (Fig. 2C). In the control group, the shape of mitochondria in villi was relatively elongated or elliptical and harbored more packed mitochondrial cristae. However, in the PCOS with euploid miscarriage group, mitochondria had a swollen shape and existed vacuolated, and reduced cristae.

DHT Impaired Mitochondrial Biogenesis and Mitochondrial Membrane Potential of HTR-8/SVneo Cells

Hyperandrogenism is a key feature of PCOS. Androgen excess is reported to regulate mtDNA CN in various cells (23, 44-46). To confirm whether androgen affected mtDNA CN in human trophoblast cells, HTR-8/SVneo cells were stimulated with DHT at concentrations ranging from 0.01 to 1000 nM for 24 hours. The results showed 0.1 nM DHT significantly inhibited levels of mtDNA CN (ND1) and (COX3) compared with the DMSO control group (0.67 vs 1.00, P = .002; 0.67 vs, 1, P = .028, respectively) (Fig. 3A and 3B). In the 10 nM DHT group, levels of mtDNA CN (ND1) and (COX3) significantly increased compared to the 0.1 nM DHT group, but were comparable to the DMSO control group (P = .028; P = .014, respectively) (Fig. 3A and 3B). To further explore the effect of DHT exposure time on mtDNA CN, HTR-8/SVneo cells were stimulated with 0.1 nM DHT for 3 hours, 6 hours, 24 hours, and 48 hours, respectively. DHT exposure for 24 hours or 48 hours reduced the levels of mtDNA CN (ND1) and (COX3) compared to the DMSO control group (P = .047; P = .009, respectively) (Fig. 3C and 3D). Thus, these findings indicate that treatment with 0.1 nM DHT for 24 to 48 hours was sufficient to inhibit mitochondrial biogenesis in HTR-8/SVneo cells. To assess whether DHT affected mitochondrial depolarization, the mitochondrial membrane potential of HTR-8/SVneo cells was detected by JC-1 staining after DHT stimulation. As shown in Fig. 3E, in the DMSO control group, aggregated JC-1 yielded red fluorescence, indicating normal mitochondrial membrane potential. In the positive control group, monomeric JC-1 yielded green fluorescence, indicating depolarized mitochondrial membrane potential. In the DHT group, JC-1 with green fluorescence and red fluorescence were detected, indicating part of the cells exhibited depolarized mitochondrial membrane potential.

Exploration and Validation of Genes With Androgen Receptor Binding Sites and Regulating Mitochondrial Biogenesis in HTR-8/SVneo Cells and Villi

To demonstrate the mechanism by which DHT suppressed mitochondrial biogenesis in trophoblast cells, the GTRD and MitoCarta3.0 database were used. Nuclear genes with androgen receptor binding sites were explored via the GTRD. Among these nuclear genes, target genes regulating mitochondrial biogenesis were further screened via the MitoCarta3.0 database. A total of 13 target genes were successfully obtained (Fig. 4A). Further, we validated the expression levels of 13 target genes in HTR-8/SVneo cells treated with 0.1 nM DHT. Strikingly, DHT significantly decreased TFAM mRNA (P < .001) and protein expression (P = .001), while it did not affect the expression of the other 12 genes (Fig. 4B-4D). TFAM is a critical factor in promoting mtDNA transcription and replication (52). The results indicated the possibility that androgen exposure inhibits mitochondrial biogenesis by suppressing the expression of TFAM.

Furthermore, we illustrated the spatial expression of TFAM in first-trimester villi from PCOS patients with euploid miscarriage using immunofluorescence staining. TFAM protein was detected in trophoblast cells with or without β–human chorionic gonadotropin expression (Fig. 4E). As shown in Fig. 4F, a decreased level of TFAM mRNA was observed in villi from PCOS patients with euploid miscarriage (n = 8) compared with the control group (n = 15). Consistently, a noticeable decrease in the protein expression of TFAM was observed in the PCOS with euploid miscarriage group (n = 7) compared with the control group (n = 6) (Fig. 4G and 4H).

Downregulation of Mitochondrial Transcription Factor A Had no Effect on the Proliferation and Apoptosis of HTR-8/SVneo Cells

To examine the role of TFAM in trophoblast cells, knockdown of TFAM gene expression was performed in HTR-8/SVneo cells (Supplementary Fig. S1) (53). The proliferation and apoptosis are critical biological functions of trophoblast cells. Mitochondria have been reported to regulate signal transductions of apoptosis and proliferation (32). Thus, we explored the effect of TFAM on proliferation using CCK-8 and EdU assays, and apoptosis using flow cytometry. The CCK-8 result showed that no statistically significant difference in optical density (OD) values was found between the siTFAM group and the negative control group (Fig. 5A). Similarly, the EdU staining result showed that TFAM knockdown had no effect on the proliferation of cells (Fig. 5B). In addition, the early apoptosis rate (3.76% ± 0.80% vs 3.41% ± 1.21%; P = .695) and late apoptosis rate (1.65% ± 0.16% vs 1.31% ± 0.38%; P = .232) of the siTFAM group were similar to the negative control group (Fig. 5C). Taken together, we found that the TFAM knockdown displayed no effect on the proliferation ability and apoptosis rate of HTR-8/SVneo cells.

Downregulation of Mitochondrial Transcription Factor A Impeded Epithelial-Mesenchymal Transition of HTR-8/SVneo Cells and BeWo Cells

In addition to proliferation and apoptosis, EMT of trophoblast cells plays a critical role in maintaining normal pregnancy (29). Consistent with our previous GSEA result, impaired EMT function in villi from PCOS patients with euploid miscarriage was detected, characterized by a notable increase in the protein level of E-cadherin and a decrease in the protein level of N-cadherin in the PCOS with euploid miscarriage group (n = 7), compared with the control group (n = 6) (Fig. 6A-6C).

Studies have proposed that mitochondrial dysfunction regulates EMT in tumor cells (41-43). To demonstrate whether the downregulation of TFAM affected EMT in trophoblast cells, we assessed EMT in two trophoblast cell lines using phalloidin staining and Western blot. HTR-8/SVneo cells established from first-trimester trophoblasts exclusively express N-cadherin and vimentin protein, while BeWo cells established from choriocarcinoma exclusively express E-cadherin protein (49). The phalloidin staining result showed that TFAM knockdown induced a striking morphologic change, characterized by curled and disordered cytoskeleton actin filament, compared with the negative control group (Fig. 6D). TFAM knockdown resulted in decreased expression of N-cadherin (P < .001) and vimentin (P < .001) proteins in HTR-8/SVneo cells (Fig. 6E-6G). To further explore the effect of TFAM on the EMT of trophoblast cells, BeWo cells were treated with forskolin to induce syncytialization. As shown in Fig. 6H and 6I, in BeWo cells treated with forskolin for 48 hours, the expression of E-cadherin protein profoundly decreased compared with the DMSO control group (P < .001), indicating that the forskolin-induced BeWo cell syncytialization model was successfully constructed. As depicted in Fig. 6J and 6K, before syncytialization, TFAM knockdown did not affect the expression of E-cadherin protein compared with the control group. When forskolin was combined with TFAM knockdown, the protein level of E-cadherin substantially increased compared with the forskolin group (P = .003). This indicates that forskolin failed to induce EMT of BeWo cells in the condition of knockdown of TFAM. Collectively, these results indicated that the downregulation of TFAM inhibited EMT in HTR-8/SVneo cells and BeWo cells after syncytialization.

Downregulation of Mitochondrial Transcription Factor A Impeded Epithelial-Mesenchymal Transition via the ROS/YAP Axis in HTR-8/SVneo Cells

Another central question is how the downregulation of TFAM reversed EMT in human trophoblast cells. The downregulation of TFAM has been reported to promote the generation of mitochondrial ROS (54). Moreover, elevated ROS is proven to suppress the nuclear translocation of YAP in tumor cells (55). ROS and YAP were both shown to participate in regulating EMT in trophoblast cells and tumor cells (56, 57). Thus, we speculated that TFAM might regulate EMT through the ROS/YAP axis in trophoblast cells. First, we assessed the effect of TFAM downregulation on mitochondrial ROS in HTR-8/SVneo cells via ROS-sensitive mitoSOX Red dye. Compared with the negative control group, the relative mean fluorescence intensity of mitoSOX in the siTFAM group was 1.53 ± 0.27 (P = .027), indicating that the TFAM knockdown increased the level of mitochondrial ROS (Fig. 7A). Moreover, we found that TFAM knockdown resulted in an increased ratio of pYAP/YAP (P = .020) (Fig. 7B and 7C). To determine whether the activity of YAP affected EMT, HTR-8/SVneo cells were treated with the YAP inhibitor verteporfin. The result showed that 100 and 1000 nM verteporfin significantly reduced the protein expression of N-cadherin compared with the DMSO control group (P = .005; P < .001) (Fig. 7D and 7E), indicating that the activity of YAP affected the EMT of trophoblast cells.

To further confirm whether TFAM regulated trophoblast EMT via the ROS-YAP axis, we used mitochondrial ROS to generate agent rotenone and scavenger Mito-TEMPO to stimulate HTR-8/SVneo cells. The administration of rotenone resulted in a notable elevation in mitochondrial ROS, indicating that the model of mitochondrial oxidative stress in HTR-8/SVneo cells was constructed (P = .019). In rotenone and 1 μM or 10 μM Mito-TEMPO cotreated cells, mitochondrial oxidative stress induced by rotenone was effectively alleviated compared with rotenone treatment alone (P = .012; P = .015) (Fig. 7F). As depicted in Fig. 7G to 7L, the cotreatment of 1 μM Mito-TEMPO and TFAM knockdown significantly upregulated the protein levels of N-cadherin (P = .002) and nuclear YAP (P = .002), accompanied by a reduction in the ratio of pYAP/YAP (P = .025) compared with TFAM knockdown alone. Furthermore, we highlighted the role of TFAM in regulating EMT by overexpression of TFAM. Western blot results revealed increased protein levels of TFAM (P < .001), N-cadherin (P = .004), and vimentin (P = .007) in HTR-8/SVneo cells with TFAM overexpression (Fig. 7M and 7O), while the ratio of pYAP/YAP remained unaltered (Fig. 7N and 7P). Taken together, these results suggest that the downregulation of TFAM may impede EMT via the ROS/YAP axis in human trophoblast cells.

Discussion

To the best of our knowledge, this is the first study to reveal that the villi of patients with PCOS who had euploid miscarriage exhibited aberrant mitochondrial biogenesis and impaired EMT. In the human trophoblast cell line, androgen exposure inhibited mtDNA CN, the hallmark of mitochondrial biogenesis, by suppressing the expression of TFAM. Furthermore, our findings demonstrated that downregulated TFAM inhibited EMT through the ROS/YAP axis (Fig. 8), but had no effect on the proliferation capability or apoptosis of trophoblast cells.

Our study provides novel evidence that levels of mtDNA CN and TFAM decreased in villus tissues of PCOS patients with euploid miscarriage, reflecting impaired mitochondrial biogenesis. As a key transcription factor of mtDNA, TFAM specifically binds to the mtDNA promoter region to form a polymer to initiate mtDNA transcription, replication, and packaging (52). Dysfunction of mitochondrial biogenesis is involved in the degeneration and fertilization processes of oocytes (58, 59), impaired implantation potential of embryos (60), and adverse pregnancy outcomes, such as miscarriage and preeclampsia (34, 35). It is well known that granulosa cells, uterus, and peripheral blood and plasma of PCOS exhibit abnormal mitochondrial biogenesis, but few studies have focused on the human placentas of PCOS patients (39). Consistent with our results, another study showed that in PCOS-like pregnant rats, placentas displaying lower mtDNA CN contents are correlated with a higher pregnancy loss rate (40). Mishra et al (23) also reported that androgen exposure caused impaired mitochondrial biogenesis in the placentas of pregnant rats and human trophoblast cells. Inconsistent with our result, Ye et al (34) reported that in non-PCOS patients with early miscarriage, mtDNA CN contents were significantly higher. This discrepancy may be due to the different patients included. It implies that too high or too low a level of mtDNA CN may be not conducive to the maintenance of a healthy pregnancy.

In line with the GSEA result, we identified impaired EMT in villi of PCOS patients undergoing euploid miscarriage. EMT plays a critical role in the invasion capability of trophoblast cells and is associated with pregnancy complications (27). Du et al (61) found that reversed EMT in placentas was correlated with the progression of preeclampsia, and impaired EMT induces embryo implantation failure and pregnancy loss (28, 29). Interestingly, we found that downregulated TFAM notably inhibited EMT in trophoblast cells, characterized by decreased levels of N-cadherin, vimentin, and an elevated level of E-cadherin. In accordance with our results, Lin et al (62) revealed that in esophageal squamous cell carcinoma cell lines, TFAM knockdown resulted in an increased expression of E-cadherin, indicating impaired EMT of cells. They proposed that the reduction of TFAM suppressed oxidative metabolism in mitochondria and further inhibited EMT function.

Furthermore, we demonstrated that downregulated TFAM inhibited EMT via the ROS/YAP axis. It has been reported that in the adipose tissue of mice, the reduction of TFAM promotes a nearly 4-fold mitochondrial membrane proton leak (54). Thus, we speculated that in trophoblast cells, inactivation of TFAM leads to an increase in ROS by remodeling mitochondrial electron transport chain function. Indeed, an elevated level of ROS was detected after the knockdown of TFAM. Interestingly, ROS is closely associated with EMT through regulating intracellular signal transduction pathways. In tumor cells, ROS is considered to play a propelling role in EMT (63). In contrast, in trophoblast cells, Lu et al (57) reported that an increased level of ROS inhibited EMT function. Consistently, our results suggested that reducing the level of mitochondrial ROS rescue impaired EMT caused by knockdown of TFAM in trophoblast cells. Furthermore, we found that the inhibition of YAP impeded EMT of trophoblast cells, and a low level of TFAM promoted the degradation of YAP. Another study summarized the role of YAP in regulating EMT (56). As a pivotal transcriptional coactivator of the Hippo pathway, nuclear YAP binds to the TWIST1 gene promoter to upregulate TWIST1 expression, then TWIST1 directly inhibits the expression of E-cadherin and promotes the expression of N-cadherin, suggesting that YAP induces EMT of cells (56). In addition, the regulation of TFAM on YAP has rarely been reported. Qi et al (64) reported that in human pluripotent stem cells, knockout of TFAM promoted YAP nuclear translocalization to block mesoderm lineage differentiation. However, in trophoblast cells, we revealed that the knockdown of TFAM decreased the level of nuclear YAP. It is inferred that the effect of TFAM on YAP differs depending on the specific cell lines. Moreover, a recent study revealed that elevated ROS suppressed nuclear translocation of YAP by activating the Hippo pathway in a human lung adenocarcinoma cell line (55). Similarly, we found that by reducing the level of mitochondrial ROS in trophoblast cells with knockdown of TFAM, the degradation of YAP was alleviated and the nuclear translocation of YAP increased. Mitochondrial oxidative damage may be a potential therapeutic target in improving EMT.

The placenta is not only a source but also a target of androgen. For example, Hsu et al (65) and Meakin et al (66) identified several androgen receptor variants in human placentas. Interestingly, a plasma member receptor mediating “nonclassic” signaling of androgen has been detected in human trophoblast cells (67). An appropriate level of androgen is critical for the maintenance of a normal pregnancy. However, androgen excess is associated with pregnancy complications. In 2024, Lu et al (13) revealed that androgen exposure resulted in placental abnormalities and an increased absorbed embryo rate. In our study, we found that androgen exposure induced mitochondrial biogenesis dysfunction in trophoblast cells, characterized by the level of mtDNA and TFAM. Similar to our results, Mishra et al (23) reported that androgen reduced mtDNA CN in cultured trophoblast cells and placentas of pregnant rats, accompanied by diminished placental and fetal weights. In placentas of rats, DHT stimulation decreased the expression of TFAM, while having no effect on mtDNA CN contents (40). However, another study reported that androgen increased mtDNA CN in pig myoblast through activated androgen receptor binding with the TFAM gene promoter region, but decreased mtDNA CN in primary preadipocytes (46). It seems that the effect of androgen on mitochondrial biogenesis varies across different tissues or cells. Nevertheless, the molecular mechanism by which androgen regulates TFAM remains poorly understood. Both classic (66) and nonclassic (67) androgen signaling receptors are detected in trophoblast cells. Further research is required to investigate the specific signaling pathway through which androgen modulates TFAM expression. Moreover, Bajpai et al (68) reported that with the guidance of a mitochondrial localization sequence, activated androgen receptor was imported into mitochondria and further decreased the expression of OXPHOS subunit proteins in cancer cells. We found that androgen excess regulated nuclear gene TFAM to indirectly decrease mtDNA in trophoblast cells. However, it is unclear whether androgen excess directly affects mtDNA via acting on the androgen receptor in mitochondria.

There are some potential limitations in our study. First, the functional study was performed in vitro using trophoblast cell lines, while the study performed in vivo in mice lacked these. Second, the effect of mtDNA heteroplasmy, another characteristic of mitochondria, is not investigated in this study. Third, we used only a few protein markers to assess the EMT of trophoblast cells, such as E-cadherin, N-cadherin, and vimentin. In this study, different model systems were used. These models have distinct characteristics and heterogeneity. For example, villi consist of not only trophoblast cells, but also matrix proteins, fibroblasts, macrophages, and blood vessels (69). HTR-8/SVeo and BeWo have been widely employed in experiments for extended periods, but these cell lines cannot fully represent placental physiology (49). It is important to consider the heterogeneity of different models when translating data across these models. Moreover, we did not include controls from euploid miscarriage patients, which means the abnormal mitochondrial biogenesis and EMT may not be specific to PCOS.

In conclusion, our study revealed that the villi of PCOS patients with euploid miscarriage were characterized by mitochondrial biogenesis dysfunction and impaired EMT. We demonstrated that androgen excess decreased mtDNA CN by inhibiting the expression of TFAM in trophoblast cells. Moreover, the reduction of TFAM damaged EMT via the ROS/YAP axis. This study provides novel insight into the mechanism through which mitochondrial biogenesis dysfunction regulates EMT of trophoblast cells in PCOS patients with euploid miscarriage.

Funding

This study was supported by the Key-Area Research and Development Program of Guangdong Province (grant No. 2023B1111020006), Guangzhou Science and Technology Planning Project (grant No. 2024B03J1252), Natural Science Foundation of Guangdong Province (grant No. 2024A1515011168), and Basic and Applied Basic Research Foundation of Guangdong Province (grant No. 2023A1515111123).

Author Contributions

Can-Quan Zhou and Qing-Yun Mai designed the study. The experiments were performed by Hui-Ying Jie. Lu Luo, Bing Cai, Yan Xu, Yuan Yuan, and Si-Min Liu contributed to tissue collection and contributed to the critical discussion. Data collection and analysis were performed by Hui-Ying Jie, Yang-Xing Wen and Ji-Fan Tan. The manuscript was drafted by Hui-Ying Jie, and was revised by Qing-Yun Mai, Can-Quan Zhou, and Ming-Hui Chen. Supervision was provided by Qing-Yun Mai and Can-Quan Zhou. All the authors have read and approved the final manuscript.

Disclosures

The authors declare that they have no conflicts of interest in this study.

Data Availability

Some or all data sets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding authors on reasonable request. The figure 8 was created in BioRender.

References

Abbreviations

 
• β2 M

  β-2-microglobulin

 
• CCK-8

  Cell Counting Kit-8

 
• CN

  copy number

 
• COX3

  cytochrome c oxidase subunit 3

 
• DHT

  5 α-dihydrotestosterone

 
• DMSO

  dimethyl sulfoxide

 
• EdU

  5-ethynyl-2′-deoxyuridine

 
• EMT

  epithelial-mesenchymal transition

 
• GSEA

  Gene Set Enrichment Analysis

 
• GTRD

  Gene Transcription Regulation Database

 
• IgG

  immunoglobulin G

 
• mRNA

  messenger RNA

 
• mtDNA

  mitochondrial DNA

 
• ND1

  NADH dehydrogenase subunit 1

 
• NES

  normalized enrichment scores

 
• OXPHOS

  oxidative phosphorylation

 
• PCOS

  polycystic ovary syndrome

 
• RNA-Seq

  RNA sequencing

 
• ROS

  reactive oxygen species

 
• RT-qPCR

  real-time quantitative polymerase chain reaction

 
• siRNA

  short interfering RNA

 
• TFAM

  mitochondrial transcription factor A

 
• YAP

  Yes-associated protein