Hypoxic mesenchymal stem cell-Derived Exosomal circPTP4A2 improves granulosa cell mitochondrial function via YBX1

Abstract

Background

Premature ovarian insufficiency (POI) profoundly compromises female reproductive health through accelerated follicle depletion and endocrine disruption. Emerging evidence highlights the therapeutic potential of mesenchymal stem cell-derived exosomes (MSC-Exs), particularly when their function is enhanced by hypoxic preconditioning. In this study, the ability of hypoxia-preconditioned MSC-Exs (H-Exs) to ameliorate oxidative damage to granulosa cells (GCs) and restore ovarian function, was systematically evaluated, and a POI rat model was used to investigate the underlying mechanism.

Methods

CircRNAs specifically expressed in H-Exs were identified and validated. The ability of H-Exs and their corresponding circPTP4A2 to repair oxidative damage and restore mitochondrial function were evaluated by antioxidant enzyme assays, reactive oxygen species (ROS) assays, JC-1 staining, ATP level assays, oxygen consumption rate (OCR) measurements and TEM. The interaction between circPTP4A2 and YBX1 was analysed by molecular dynamics simulations, RIP, CHX assays, and MG132 assays, and the restorative effect of the circPTP4A2/YBX1 axis on ovarian function was verified.

Results

Our findings revealed that compared with normoxic MSC-Exs (N-Exs), H-Exs exerted superior protective effects, significantly attenuating oxidative stress and restoring mitochondrial bioenergetics in KGN cells. Mechanistically, circPTP4A2 was identified as a hypoxia-responsive cargo selectively enriched in H-Exs. This circular RNA stabilized Y-box binding protein 1 (YBX1) through direct interaction, increasing its antioxidative capacity and mitochondrial regulatory functions. Hypoxia-inducible factor 1-alpha (HIF-1α) was further shown to transcriptionally upregulate circPTP4A2 via direct binding to the promoter region of its host gene PTP4A2.

Conclusion

These results establish the circPTP4A2/YBX1 axis as a critical mediator of the therapeutic efficacy of H-Exs for POI, providing both mechanistic insights and a translational framework for exosome-based regenerative strategies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-025-06067-z.

Introduction

Primary ovarian insufficiency (POI) affects women under 40 years of age, causing ovarian function decline, amenorrhea, and infertility [1]. This condition affects approximately 1% of women in this age group. POI imposes heavy physical and mental burdens [2–4]. The aetiology of POI remains unclear. To date, the known factors involved include primarily idiopathic and genetic factors. Genetic factors include mainly chromosomal abnormalities and single-gene defects [5]. Notably, the pathogenesis of POI involves multiple factors, such as premature follicular exhaustion, granulosa apoptosis, oxidative stress, and disorders of the ovarian microenvironment, making effective intervention with traditional treatment methods difficult. Existing clinical treatments such as hormone replacement therapy (HRT) only relieve symptoms and do not fix the root problem in the ovaries [6–8]. In recent years, with advances in regenerative medicine, the use of mesenchymal stem cells (MSCs) and their exosomes (Exs) has shown promise for restoring ovarian function in POI patients [9]. Exosomes from hypoxic-preconditioned MSCs (H-Exs), which are rich in bioactive molecules, have better effects on cell repair [10–13]. Thus, more research on the molecular mechanisms responsible for the increased efficacy of H-Exs in POI is needed.

Mitochondrial dysfunction in ovarian GCs is a major driver of POI pathogenesis. It leads to decreased steroidogenesis and follicular atresia, which are key in ovarian ageing and infertility [14–17]. H-Exs have demonstrated considerable effectiveness in reducing mitochondrial dysfunction in different cell models, such as MODE-K cells, through the inhibition of the PI3K/AKT/mTOR pathway. They also promote the essential colocalization of mitochondria and lysosomes for cell repair [18]. Nevertheless, the precise molecular mechanisms underlying the reparative effects of H-Exs are still largely unknown.

Circular RNAs (circRNAs) are noncoding RNAs with a covalently closed loop structure that provides them with greater stability than that possessed by linear RNAs [19]. CircRNAs modulate gene expression and protein function through various mechanisms, including acting as microRNA sponges, interacting with RNA-binding proteins, and influencing mRNA stability [20–22]. Recent research has emphasized the importance of circRNAs in biological processes, including oxidative stress and mitochondrial function, both of which are crucial for POI [17, 23]. For example, in some disease models, circRNAs have been confirmed to improve mitochondrial function through miRNA spongy activity. Therefore, we speculate that the circRNAs enriched in H-Exs likely play a key role in improving the stress resistance of GCs through the targeting of mitochondrial function-related pathways. In addition, the activation of HIF-1α can lead to the enrichment of specific circRNAs in H-Exs, resulting in unique regulatory effects [24]. For example, in pancreatic cancer, after HIF-1α is activated in a hypoxic environment, the expression level of circPDK1 in exosomes is upregulated, promoting glycolysis [24].

Despite these advances, the molecular mechanisms underlying H-Exs-mediated mitochondrial repair in POI remain unclear. To address this gap, we identified circRNAs that are upregulated in H-Exs through database analysis [25]. We identified circPTP4A2 as a key molecule. Investigating the role and mechanism of action of H-Exs-secreted circPTP4A2 in POI could pave the way for the development of targeted therapies.

Materials and methods

Cell culture and transfection

293 T and KGN cells were purchased from Shanghai Jihe Biotechnology Co., Ltd. (Shanghai, China)and were cultured in DMEM (Gibco, USA) supplemented with 10% foetal bovine serum (FBS) (Gibco, USA). KGNs cells were inoculated on culture plates and allowed to grow to 80% confluence. They were transfected with si-circPTP4A2 (GenePharma, Shanghai, China), si-NC (GenePharma, Shanghai, China) and si-HIF-1α (GenePharma, Shanghai, China) using Lipofectamine 2000 (Invitrogen, Carlsbad, USA) following the manufacturer’s instructions. After 48 h of transfection, the cells were used for subsequent analyses.

The cells were treated with cycloheximide (CHX, 10 µM) (Abmole Bioscience Houston, USA) for different durations to block protein synthesis. Protein extraction was performed and protein levels were assessed using western blot analysis.

The cells were treated with 20 µM proteasome inhibitor (MG132) (Abmole Bioscience, Houston, USA) for 6 h. Proteins were extracted and protein levels were assessed using western blot analysis.

Intracellular hypoxic model construction

Exs were cultured from the bone marrow stem cells of SD rats. The femurs and tibias of the SD rats were isolated and immersed in PBS supplemented with 10% penicillin-streptomycin double antibody. The complete culture medium was then aspirated with a syringe and the marrow cavity was repeatedly flushed to collect the irrigating fluid. After resuspend the irrigating fluid, the cell precipitate was obtained and transferred to a large culture dish for culture.The cell precipitate was resuspended and transferred to large dishes for culture. Two days later, the culture medium was changed. After the cells adhered and grew well, they were passaged. Cells of the 3rd to 6th generations were selected for the subsequent experiment.

BMSCs (1.5 × 10⁵) were seeded in six-well culture plates, and CoCl₂ (Sigma-Aldrich, USA) was added to the cells to a final concentration of 100 µM. Exs were collected at 36 h. The CCK8 assay was used to evaluate cell viability after treatment with different concentrations of the drug and at different time points (Vazyme, Nanjing, China).

For physical hypoxic treatment, the cells were cultured in a 3-gas hypoxia incubator with 1% O², 5% CO², and 94% N₂.

Isolation, labelling, and quantification of exosomes

The collected normoxic and hypoxic culture supernatants were separately processed as follows. The supernatant collected from the BMSCs was centrifuged at 3000 × g for 30 min to remove dead cells and debris. Afterwards, the supernatant was concentrated using an ultracentrifugation filter at 3000 × g for 30 min. Then, total exosome separation reagent (Umibio, UR52121) was added, and the mixture was allowed to sit for 2 h. The Exs were further precipitated by ultracentrifugation at 10,000 × g for 60 min. The precipitate was subsequently washed with PBS and ultracentrifuged at 12,000 × g for 2 min again. The resulting sediment was resuspended in PBS. The particle size and form of the Exs were assessed by transmission electron microscopy (TEM). Nanoparticle tracking analysis (NTA) was used to determine the size of the Exs by using a Zeta View PMX110 system. Specific exosome surface markers were detected using anti-TSG101 antibodies (Wanleibio, Shenyang, China, 1:2000), anti-CD63 antibodies (Wanleibio, Shenyang, China, 1:2000) and anti-calnexin antibodies (Santa Cruz, USA, 1:1000). Purified exosomes were labelled with a PKH26 fluorescent labelling kit (Sigma-Aldrich, USA). The PKH26-labelled exosomes were cocultured with KGN cells. After being incubated for 24 h at 37 °C in a 5% CO₂ atmosphere, the cells were washed twice with PBS, and fixed, after which the nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Finally, a confocal laser microscope was used to observe whether the cells took up Exs. For coculture, the exosome dose was defined as the number of exosome particles per recipient cell (10⁶ particles/cell).

Western blotting

Total cellular lysates were prepared in RIPA buffer (Solarbio, Beijing, China). Equal amounts of proteins were separated by SDS-PAGE and then transferred onto PVDF membranes (Millipore, USA). After incubating them with anti-YB-1 antibodies (Abcam, USA, 1:3000), the membranes were incubated with anti-rabbit IgG secondary antibody (Biosharp, China, 1:10000) for 1.5 h, after which the strips were exposed with a gel imager. Anti-GAPDH and anti-β-actin antibodies (Proteintech, Wuhan, China, 1:10000) were used as loading controls.

Quantitative real-time polymerase chain reaction (qRT-PCR)

A TRIzol kit (Ambion, USA) was used to extract total RNA. To measure circRNA expression, reverse transcription was performed using a high-capacity cDNA reverse transcription kit (Vazyme, Nanjing, China). Real-time PCR was performed using ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China) with a quantitative real-time PCR instrument. β-actin, GAPDH or U6 was used as an endogenous reference. All primers were designed and synthesized by Shanghai Sangon Biotech. The primers used are listed in Supplementary Table 1.

Nuclear and cytoplasmic fractionation

In accordance with the manufacturer’s protocol, RNA was isolated and purified from the nuclei and cytoplasm of KGNs using a cytoplasmic and nuclear RNA purification kit (Beyotime, Shanghai, China). The abundance of circRNAs in the cytoplasmic or nuclear fractions was measured by qRT-PCR; GAPDH was used as a cytoplasmic DNA, and U6 was used as a control for nuclear DNA.

Validation of circrnas

cDNA and gDNA were amplified using divergent and convergent primers for circPTP4A2 and β-actin. Agarose gel electrophoresis was subsequently performed, followed by Sanger sequencing. For the RNase R experiment, total RNA was extracted from KGNs as described above and incubated with RNase R (Lucigen, USA) at 37 °C for 120 min. The purified RNA was reverse transcribed into cDNA and then analysed by qRT-PCR.

ROS levels measurement

The ROS-sensitive fluorescent dye dihydroethidine (Beyotime, Shanghai, China) was used to measure ROS levels in KGNs and rat ovarian tissues. KGNs or ovarian tissue were incubated with DHE (1 µM) at 37 °C for 30 min and washed with PBS 3 times. The cells were then incubated with DAPI (Solarbio, Beijing, China) for 10 min in the dark and washed 3 times with PBS. Finally, fluorescence was detected using a confocal laser scanning microscope (Leica Microsystems, Mannheim, Germany).

Mitochondrial membrane potential assay

The mitochondrial membrane potential (MMP) of KGN cells was measured using a mitochondrial membrane potential assay kit with JC-1 (Beyotime, Shanghai, China). The KGNs sections were incubated with JC-1 at 37 °C for 20 min and washed with PBS 3 times and observed under a fluorescence microscope (Leica Microsystems, Mannheim, Germany).

Senescence-associated β-galactosidase (SA-β-gal) activity assay

In accordance with the manufacturer’s instructions, a Senescence β-Galactosidase Staining Kit (Beyotime, Shanghai, China) was used to measure senescence in KGNs cells. KGNs cells (2 × 10⁶) and ovarian tissues were fixed with β-galactosidase fixing solution for 15 min at room temperature and then were washed three times. KGNs and ovarian tissues were incubated with β-galactosidase activity staining solution at 37 °C overnight to evaluate senescence.

Measurement of superoxide dismutase (SOD), malondialdehyde (MDA) and glutathione peroxidase (GPX) activity

The enzymatic activities of SOD, MDA and GPX in KGNs were measured using test kits (Beyotime, Shanghai, China) in accordance with the manufacturer’s instructions. The absorbance was measured at 450, 532–340 nm using an enzyme detector. The SOD, MDA and GPX levels were normalized to the total protein.

ATP generation assay

1.5 × 10⁵ KGNs were inoculated on a six-well culture plate. After the cells were treated with siRNA or plasmid for 48 h, lysis buffer was added, and the chemiluminescence absorbance was calculated using an ATP Assay Kit to determine the concentration of ATP in different cells (Beyotime, Shanghai, China).

Measurement of the oxygen consumption rate (OCR)

1.5 × 10⁵ KGNs were inoculated on 96-well culture plates. OCR measurements were performed using a Seahorse Bioscience XFp Extracellular Flux Analyser (Agilent Technologies AG, Basel, Switzerland). Oxygen consumption was measured every 10 min, and the following drugs were injected after every three measurements: (1) 1 mM oligomycin, (2) 1 mM FCCP and (3) 2 mM rotenone and 2 mM antimycin A.

Molecular dynamics simulation

In this study, the Amber 2020 software package was used to conduct molecular dynamics simulations of the screened receptor–ligand complexes. The AMBER19SB force field parameter was used for YBX1, and the RNA.OL3 force field parameter was used for circRNAs. The TIP3P-dominant water model was selected. The minimum distance between the atoms in the receptor and the edge of the water box was 1.0 nm. Sodium ions or chloride ions were used to neutralize the system charge on the basis of the docking results. The workflow of molecular dynamics simulation consisted of four steps: energy minimization, heating, equilibrium, and production dynamics simulation. First, the heavy atoms of the receptor (and RNA) were constrained, and the energy of the water molecule was minimized in 10,000 steps (including 5000 steps of the steepest descent method and 5000 steps of the conjugate gradient method). The constraints were subsequently lifted, and the entire system was subjected to 10,000 steps of energy minimization (including 5,000 steps of the steepest descent method and 5000 steps of the conjugate gradient method). Energy optimization involved slowly heating the system to 300 K within 50 ps. After heating was completed, the system was balanced at 50 ps under the npt ensemble. Finally, the system was subjected to a 100 ns molecular dynamics simulation under the npt ensemble. The trajectory data were saved every 10 ps, and relevant analyses were conducted using the cpptraj module. The binding free energy of ligands and receptors was calculated using the MMPBSA method.

Chromatin Immunoprecipitation (ChIP)

ChIP was performed with the SimpleChIP Plus Sonication Chromatin IP Kit (CST, USA) according to the manufacturer’s instructions. After the cells were subjected to hypoxia, they were cross-linked with 37% formaldehyde for 20 min. Glycine was used to terminate the cross-linking reaction. After the cells were collected, they were fully lysed, disrupted by ultrasonication and verified by 1.5% agarose gel electrophoresis. After the supernatant was collected, immunoprecipitation with an anti-HIF-1α antibody (Abcam, USA) and IgG antibody (BersinBio, Guangzhou, China) was performed according to the manufacturer’s instructions. The DNA in the collected samples was subsequently purified and verified by PCR.

RNA Immunoprecipitation (RIP)

The RIP assay was conducted with an RNA immunoprecipitation (RIP) kit (BersinBio, Guangzhou, China). 293 T cells (1 × 10⁷) were incubated with lysate buffers supplemented with protease and RNase inhibitors. Afterwards, the cells were lysed and incubated with magnetic beads coupled to anti-YBX1 antibodies (Abcam, USA, 4 µL) or negative control IgG (BersinBio, Guangzhou, China, 4 µL), overnight at 4 °C. The immunoprecipitated RNA was subsequently purified and analysed by qRT-PCR.

In vivo ovarian premature failure model

Female Sprague-Dawley (SD) rats were purchased from Jiangsu University Experimental Animal Center. All the animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the Medical School of Jiangsu University and performed in accordance with the laboratory guidelines for the ethical review of animal welfare. Five-week-old female SD rats were generated and intraperitoneally injected with CTX (50 mg/kg) on the first day and then continuously injected with CTX (8 mg/kg/d) for 13 days [26]. After successful modelling, all the model rats were randomly divided into 4 groups and injected with PBS (100 µL), BMSC-N-Exs (150 µg/100 µL of PBS), BMSCs-H-Exs (150 µg/100 µL of PBS) or BMSC-H-Exs-si-circPTP4A2 (150 µg/100 µL of PBS) via the tail vein every 2 days. After 7 injections, they were used for follow-up experiments.

Estrous cycles

At 8:00 a.m. every day, sterilized cotton swabs were used to obtain vaginal exfoliated cell smears from the rats, after which the estrous cycles of the rats in each group were observed. The normal estrous cycle of rats consists of four stages: preestrous, estrous, postestrous and interestrous. In the preoestrus period, many nucleated epithelial cells exist, while in the oestrous period, many anucleated keratinized epithelial cells are present. In the late postoestrus stage, keratinized epithelial cells, nucleated epithelial cells and white blood cells are mixed. In the interestrous period, mainly white blood cells are present, with a small amount of mucus.

H&E staining

The ovaries of the rats were fixed with 4% paraformaldehyde solution, embedded in paraffin, and cut into 4-um-thick sections, which were subsequently stained with haematoxylin and eosin (H&E) for routine histological examination.

Immunohistochemistry

The slices were first dewaxed, rehydrated and heated in a microwave for antigen repair. They were then incubated with primary antibody against YBX1 (Abcam, USA, 1:50) at 4 °C overnight. The slices were subsequently incubated in the dark with a secondary antibody (Biosharp, China, 1:10000) coupled to horseradish peroxidase. Finally, the sections were stained with haematoxylin (Biosharp, China) and imaged with a pathological image scanner (Pannoramic MIDI, Hungary).

Elisa

Serum concentrations of hormones, including anti-Mullerian duct hormone (AMH), follicle-stimulating hormone (FSH), luteinizing hormone (LH) and estradiol (E₂), were measured in different groups of POI model rats (n = 6) using an enzyme-linked immunosorbent assay kit (ImmunoWay, USA).

Patient group and GC acquisition

In this study, serum samples and GCs were obtained from POI patients who underwent in vitro fertilization or intracytoplasmic sperm injection embryo transfer (IVF/ICSI-ET) at the Reproductive Center of the Fourth Affiliated Hospital of Jiangsu University (Zhenjiang Maternal and Child Health Hospital) and IVF/ICSI-ET due to male and/or tubal factors in patients with normal ovarian reserve function. The collection of serum samples and GCs from patients met the ethical requirements and was approved by the Ethics Committee of the Fourth Affiliated Hospital of Jiangsu University (Zhenjiang Maternal and Child Health Hospital). All the subjects voluntarily participated and signed written informed consent forms. Individuals in the control group were matched with those in POI group in terms of age and body mass index (Supplementary Table 5). The clinical sample inclusion criteria for POI patients were as follows: ① patients who met the diagnostic criteria for POI; ② patients in whom other organic changes were ruled out. The exclusion criteria were as follows: ① use of hormones in the past three months; ② the presence of severe organic diseases such as cardiovascular, liver and kidney diseases; ③ the presence of other endocrine or autoimmune diseases. In addition, relevant genetic data were not available for some patients; however, these data were available for some patients and revealed no genetic trend.

Statistical analysis

All the values were analysed with GraphPad Prism software. Unpaired two-tailed Student’s t test was used to evaluate differences between two groups, and one-way analysis of variance was used for comparisons among three or more groups. When P was < 0.05, the results were considered statistically significant. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Results

H-Exs repair oxidative damage in CTX-KGN cells

We treated BMSCs with different concentrations of CoCl₂ and measured cell viability at different times using the CCK8 assay. Cell viability was the highest after treatment with 100 µM CoCl₂ for 36 h. At the concentration (100 µM) used in this study, the production of exosomes and the loading of cargo increased. These results effectively eliminated concerns about the observations caused by the false appearance of the CoCl₂ toxicity experiment (Supplementary Fig. 1A-D).

First, we obtained H-Exs from MSCs by inducing a hypoxic environment with CoCl₂ [27]. Characterization via TEM (Fig. 1A) and NTA (Fig. 1B) revealed that both N-Exs and H-Exs were cup-shaped structures with a size of approximately 100 nm and high expression levels of the surface markers CD63 and TSG101 (Fig. 1C). In addition, negative markers of potential external contaminants were detected by western blotting. Calnexin was not detected. These findings confirm that we effectively reduced contamination by cell debris and organelles.

Fig. 1.

H-Exs effectively improve mitochondrial function in GCs. A. Representative TEM images of N-Exs and H-Exs (scale bar = 100 nm). B. The representative particle size distribution of N-Exs and H-Exs as shown by NTA.C. Western blot analysis of N-Exs and H-Exs surface markers CD63 and TSG101.D-F. Measurement of MDA (D), SOD (E) and GPX (F) activity in cocultured cells. G. The ROS level of co-cultured cells was detected by DCFH-DA staining; green for DCFH-DA and blue for nucleus (DAPI) (scale bar = 75 μm). H. SA-β-gal staining showing senescence in co-cultured cells. Senescent cells were stained blue (scale = 200 micrometers) (scale bar = 200 μm). I. The changes of MMP in co-cultured cells were detected by JC-1 staining (scale bar = 75 μm). J. Observation of mitochondrial morphology of co-cultured cells by transmission electron microscopy. (scale bar = 500 nm). K-M. Relative level of mtDNA (K), ATP generation capacity (L) and OCR (M) of cocultured cells

To evaluate the reparative effect of H-Exs on oxidative damage to GCs, we cocultured PKH26-labelled N-Exs and H-Exs with KGNs and confirmed the successful internalization of Exs by the former (Supplementary Fig. 1E-F). Pretreatment with CTX induced oxidative stress in KGNs (Supplementary Fig. 2A-M). As illustrated in the figures, Exs transplantation significantly increased the levels of SOD and GPX, reduced the level of MDA (Fig. 1D-F), decreased ROS accumulation, and decreased the number of SA-β-gal-positive cells (Fig. 1G, H; Supplementary Fig. 3A-B). Moreover, mitochondrial function analysis demonstrated that Exs increased the MMP (JC-1 staining) (Fig. 1I; Supplementary Fig. 3 C) and maintained mitochondrial morphology (Fig. 1J), leading to a significant increase in mtDNA content (Fig. 1K) and ATP production (Fig. 1L). KGNs exhibited a relatively high OCR (Fig. 1M). Notably, the protective effects of H-Exs against oxidative damage and on mitochondrial function were clearly superior to those of N-Exs. Taken together, the above results indicate that H-Exs significantly attenuate oxidative damage in KGNs and promote the restoration of mitochondrial function.

CircPTP4A2 is enriched in H-Exs and downregulated in the GCs of patients with POI

To elucidate the key molecules involved in the reparative effects of H-Exs, we analysed the data to identify circRNAs that are specifically enriched in H-Exs. Our analysis revealed several circRNAs with high expression levels, among which hsa_circ_0007364 (circPTP4A2) exhibited the greatest degree of enrichment (Fig. 2A). QRT-PCR demonstrated that circPTP4A2 was enriched in hypoxic MSCs and exosomes (Fig. 2B, C). In addition, the expression level of circPTP4A2 showed the same trend after physical hypoxia exposure (Supplementary Fig. 1G, H). Notably, circPTP4A2 expression was downregulated in CTX-KGNs (Fig. 2D) and in GCs from patients with POI (Fig. 2E), suggesting the involvement of circPTP4A2 in POI.

Fig. 2.

CircPTP4A2 is significantly expressed in H-Exs and activated by HIF-1α under hypoxic conditions. A. Schematic diagram of the circPTP4A2 screening process. B. QRT-PCR to determine the relative expression levels of circPTP4A2 in normoxic and hypoxic MSCs (N-MSCs, H-MSCs). C. QRT-PCR to determine the relative expression levels of circPTP4A2 in exosomes from normoxic and hypoxic MSCs (N-Exs, H-Exs). D. QRT-PCR to determine the relative expression levels of circPTP4A2 in CTX-KGNs. E. QRT-PCR to determine the relative expression levels of circPTP4A2 in ovarian GCs from patients with normal ovarian function and POI. F. Schematic of circPTP4A2 genomic location and splicing, with Sanger sequencing to verify the reverse splice site. G. The existence of circPTP4A2 was verified by agarose gel electrophoresis; gDNA and cDNA were amplified respectively using divergent primers and convergent primers of circPTPA2. H. Relative expression levels of circPTP4A2 and linear PTP4A2 mRNA determined by RT-PCR after post-ActD treatment. I. Relative expression levels of circPTP4A2 and linear PTP4A2 mRNA determined by RT-PCR after treatment with or without RNase R. J. Subcellular isolation assay to detect circPTP4A2 localization.GAPDH was used as the cytoplasmic expression control, and U6 was used as the nuclear expression control. K-L. QRT-PCR to determine the relative expression levels of circPTP4A2 and linear PTP4A2 expression in N-MSCs, H-MSCs. M. QRT-PCR to determine the relative expression levels of circPTP4A2 and linear PTP4A2 expression in N-MSCs, H-MSCs following HIF-1α silencing. N. Schematic diagram of the components involved in the hypoxic response in the promoter region of PTP4A2. O. ChIP assays were performed to assess the HREs interactions with the PTP4A2 promoter region

Characterization and identification of circPTP4A2

CircPTP4A2, which is 782 nucleotides long and derives from exons 2–3 of the PTP4A2 gene, was confirmed via Sanger sequencing (Fig. 2F). The divergent primer for circPTP4A2 amplified cDNA but not genomic DNA (Fig. 2G). Compared with linear PTP4A2 mRNA, circPTP4A2 had a longer half-life and greater stability (Fig. 2H). RNase R treatment confirmed its resistance to digestion, identifying circPTP4A2 as a stable cyclized RNA (Fig. 2I). Additionally, it was predominantly localized in the cytoplasm (Fig. 2J). Overall, these findings suggest that circPTP4A2 is an abundant and stable circRNA in H-Exs that is involved in the pathogenesis of POI.

HIF-1α-mediated transcriptional activation of circPTP4A2

Under hypoxic conditions, numerous circRNAs are transcriptionally activated by HIF-1α, with their host linear genes derived from the same pre-mRNAs [24]. These findings suggest that the host gene of circPTP4A2 is activated by HIF-1α, leading to increased expression of circPTP4A2. We established a hypoxic model of MSCs via the use of CoCl2 and assessed cell viability. CircPTP4A2 and its host gene PTP4A2 were significantly overexpressed in hypoxic MSCs (Fig. 2K, L), whereas silencing of HIF-1α led to notable downregulation of both genes (Fig. 2M).

Promoter sequence analysis (via UCSC/JASPAR) revealed two potential hypoxic response elements (HREs) in the promoter region of the circPTP4A2 host gene, located at nucleic acids 1309–1314 (ACGTGA, HRE1) and 1585–1590 (ACGTCC, HRE2) (Fig. 2N). ChIP assays were conducted to explore whether HIF-1α could directly interact with these transcription start sites, which confirmed that HIF-1α bound to the PTP4A2 promoter region, specifically the HRE1 and HRE2 regions (Fig. 2O). In conclusion, these findings collectively suggest that HIF-1α binds to the PTP4A2 promoter region and facilitates transcription, thereby leading to the upregulation of circPTP4A2 expression. The enrichment of circPTP4A2 in H-Exs can, to a certain extent, be attributed to the activation of HIF-1α.

Silencing circPTP4A2 attenuates the repair of CTX-KGNs by H-Exs

Given the prominent expression of circPTP4A2 in H-Exs, we hypothesized that it plays a crucial role in mitigating oxidative stress in CTX-KGNs. To test this hypothesis, we transfected MSCs with a small interfering RNA targeting circPTP4A2 (si-circPTP4A2) and collected H-Exs-si-circPTP4A2 for coculture with CTX-treated KGNs (Fig. 3A; Supplementary Fig. 1I). Coculture experiments revealed that compared with that in the PBS group, the level of circPTP4A2 in CTX-KGNs treated with N-Exs and H-Exs was significantly elevated (Fig. 3A), especially in those treated with H-Exs, which supported its efficient transfer from Exs into KGN cells.

Fig. 3.

CircPTP4A2 released by H-Exs alleviates mitochondrial damage in GCs. A. QRT-PCR to determine the relative expression levels of circPTP4A in CTX-KGNs co-cultured with N-Exs, H-Exs, and H-Exs-si-circPTP4A2. B-D. Measurement of MDA (B), SOD (C) and GPX (D) activity in cocultured cells. E. Expression level of mtDNA of cocultured cells. F. The ROS level of co-cultured cells was detected by DCFH-DA staining; green for DCFH-DA and blue for nucleus (DAPI) (scale bar = 75 μm). G. SA-β-gal staining showing senescence in co-cultured cells. Senescent cells were stained blue (scale = 200 micrometers) (scale bar = 200 μm). H. The changes of MMP in co-cultured cells were detected by JC-1 staining (scale bar = 75 μm). I. Observation of mitochondrial morphology of co-cultured cells by transmission electron microscopy. (scale bar = 500 nm). J-K. ATP generation capacity (J) and OCR (K) of cocultured cells. L. The binding mode of the complex YBX1 protein with circPTP4A2 after 100ns MD simulation. M. The RMSD of circPTP4A2 with YBX1 protein. N. The RMSF of circPTP4A2 with YBX1 protein. O. The Rg of circPTP4A2 with YBX1 protein. P. The SASA of circPTP4A2 with YBX1 protein. Q. The hydrogen bond number of circPTP4A2 with YBX1 protein. R. RIP assay to measure circPTP4A2 expression in YBX1-overexpressing HEK 293 T cells. S-T. QRT-PCR analysis of circPTP4A2 expression after YBX1 knockdown or overexpression. U. Relative YBX1 mRNA and protein levels after circPTP4A2 knockdown. V-W. After CHX reagent treatment, WB assay was performed to detect YBX1 protein level after circPTP4A2 knockdown.Protein quantification of YBX1 (W). X. After MG132 reagent treatment, WB assay was performed to detect YBX1 protein level after circPTP4A2 knockdown

Functional analysis indicated that silencing circPTP4A2 in H-Exs significantly weakened the protective effect of H-Exs against oxidative damage in CTX-KGNs. This phenomenon was evidenced by the upregulation of ROS and MDA levels, along with the reduced activities of SOD and GPX in CTX-KGNs (Fig. 3B-D, F; Supplementary Fig.3E). Furthermore, the ability of H-Exs-si-circPTP4A2 to combat cellular ageing was reduced, as indicated by the increased number of SA-β-gal-positive cells among CTX-KGNs (Fig. 3G; Supplementary Fig. 3D).

Additionally, circPTP4A2 silencing reduced the protective effect of H-Exs on mitochondrial function, as demonstrated by an increase in the MMP (measured by JC-1) (Fig. 3H; Supplementary Fig. 3 F), a shift from a filamentous morphology to a fragmented mitochondrial morphology (Fig. 3I), and decreased mtDNA content (Fig. 3E) and ATP production in CTX-KGNs (Fig. 3J). KGNs presented a lower OCR (Fig. 3K). Overall, these results suggest that circPTP4A2 is a critical molecule that mediates the protective effects of H-Exs against oxidative damage and ameliorates mitochondrial dysfunction in CTX-KGNs.

CircPTP4A2 interacts with YBX1 and enhances its stability

Given the established role of YBX1, an RNA-binding protein, in circular RNAs [28], we predicted the binding site of circPTP4A2 with YBX1 via molecular dynamics simulation. Molecular dynamics simulation revealed that circPTP4A2 can form a stable YBX1-circPTP4A2 complex with YBX1 through salt bridging, hydrogen bonding and hydrophobic interactions. It could be seen from the root mean square deviation (RMSD) diagram that the average RMSD of the complex was less than 4 nm, and the complex essentially reached the dynamic equilibrium at approximately 40 ns. The root mean square fluctuation (RMSF) map revealed that most of the changes in the amino acid conformation of the complex were within the acceptable range. The radius of gyration (Rg), solvent accessible area (SASA) diagrams, hydrogen bond network diagrams of YBX1 and circPTP4A2, and binding free energy data revealed that the complex was tightly bound and stable. In summary, the circPTP4A2 and YBX1 proteins can form a stable complex to play an active role (Fig. 3L-Q; Supplementary Table 4). RIP experiments confirmed the binding of these molecules (Fig. 3R). Surprisingly, neither overexpressing nor silencing YBX1 affected circRNA expression (Fig. 3S-T). Moreover, circPTP4A2 silencing significantly decreased the protein level of YBX1 without affecting its mRNA expression (Fig. 3U), leading us to hypothesize that circPTP4A2 stabilizes the YBX1 protein. After CHX treatment, we observed a shorter half-life of the YBX1 protein in KGNs with downregulation of circPTP4A2, indicating the interaction of circPTP4A2 and YBX1 and the regulation of protein stability (Fig. 3V-W). Furthermore, circPTP4A2 silencing reduced YBX1 protein levels, which could be restored by the proteasome inhibitor MG132 (Fig. 3X). These results suggest that circPTP4A2 binds to YBX1 and increases its protein stability.

YBX1 counteracts the exacerbation of oxidative damage induced by circPTP4A2 Silencing

To elucidate the roles of circPTP4A2 and YBX1 in the repair of oxidative damage, we overexpressed YBX1 in circPTP4A2-knockdown KGN cells. YBX1 overexpression reduced the accumulation of ROS induced by oxidative damage, increased the activity of the antioxidant enzymes SOD and GPX, decreased the level of MDA, and alleviated cell senescence (Fig. 4A-E; Supplement Fig. 3G-H). YBX1 overexpression significantly improved mitochondrial function. Mitochondrial morphology became more normal (Fig. 4G), the membrane potential increased (Fig. 4F; Supplement Fig. 3I), the mtDNA content (Fig. 4H) and ATP production increased (Fig. 4I), and the OCR also increased (Fig. 4J), which indicated that YBX1 overexpression effectively restored and enhanced mitochondrial function after it was compromised by circPTP4A2 knockdown (Fig. 4F-J), thus highlighting the roles of YBX1 and circPTP4A2 in the repair of oxidative damage and maintenance of mitochondrial homeostasis.

Fig. 4.

Overexpression of YBX1 improved the oxidative damage induced by circPTP4A2 Silencing.A-C. Measurement of MDA (A), SOD (B) and GPX (C) activity in co-cultured cells. D. The ROS level of co-cultured cells was detected by DCFH-DA staining; green for DCFH-DA and blue for nucleus (DAPI) (scale bar = 75 μm). E. SA-β-gal staining showing senescence in co-cultured cells. Senescent cells were stained blue (scale = 200 micrometers) (scale bar = 200 μm). F. The changes of MMP in co-cultured cells were detected by JC-1 staining (scale bar = 75 μm). G. Observation of mitochondrial morphology of co-cultured cells by transmission electron microscopy. (scale bar = 500 nm). H-J. Relative level of mtDNA (H), ATP generation capacity (I) and OCR (J) of cocultured cells (&: si-NC vs. si-circPTP4A2; #: si-circPTP4A2 vs. si-circPTP4A2 + YBX1)

Effects of circPTP4A2 secretion by H-Exs on ovarian function in POI model rats

To determine whether circPTP4A2 secreted by H-Exs can relieve oxidative stress-induced ovarian dysfunction in vivo, a POI rat model was established through intraperitoneal CTX injections for two weeks, and the estrous cycle was closely monitored to confirm the validity of the model. After modelling, POI model rats received tail vein injections of PBS, N-Exs, H-Exs, or H-Exs-si-circPTP4A2 every two days for a total of seven doses (Fig. 5A). Treatment with N-Exs and H-Exs resulted in multiple beneficial changes: estrous cycles became normalized, while AMH and E₂ levels increased and FSH and LH levels decreased, resulting in improved hormonal balance (Supplementary Fig. 4A-J). Body weight and ovarian indices increased, and histological analysis revealed a shift toward more functional follicles and fewer atretic follicles. The fertility potential was increased, as indicated by increased pregnancy rates and litter sizes (Supplementary Fig. 5A-F; Fig. 5B-F). Oxidative stress in ovarian tissue was also alleviated, with reduced ROS accumulation, fewer SA-β-gal-positive cells, and better mitochondrial morphology (Fig. 5G-I). Notably, H-Exs were much more efficacious than N-Exs (Fig. 5B-I; Supplementary Fig.4B-J; Supplementary Fig. 5A-F).

Fig. 5.

Effects of circPTP4A2 Secreted by H-Exs on ovarian function in POI rats. A. Schematic diagram of POI rat model construction and treatment. B. Diagram of ovary of rats in each group after treatment. C. Ovary weight of rats in each group after treatment. D. Ovary index of rats in each group after treatment. E. HE staining of ovarian sections (scale bar = 200 μm). F. Schematic representation of pregnancy out comes in each group after 4 and 8 weeks of treatment G. DCFH-DA staining to assess ROS levels in each group (scale bar = 75 μm). H. SA-β-gal staining for senescence detection in each group (scale bar = 250 μm). I. The morphology of ovarian mitochondria was observed by transmission electron microscopy in each group(scale bar = 500 nm)

However, when circPTP4A2 was silenced in H-Exs, their therapeutic effect was severely weakened. The restoration of hormonal balance, histological improvements in the ovaries, and the enhancement of fertility were inhibited, reversing the prior beneficial effects (Fig. 5B-I; Supplementary Fig. 4B-J; Supplementary Fig. 5A-F). Compared with those in the H-Exs group, the YBX1 levels in the ovaries of rats in the H-Exs-si-circpTP4A2 group were significantly lower, and the therapeutic effect was weakened (Fig. 6A; Supplementary Fig. 5G). Overall, these findings strongly suggest that H-Exs-derived circPTP4A2 plays a crucial role in enhancing ovarian function in POI model rats, providing valuable insights for the treatment of ovarian disorders.

Fig. 6.

CircPTP4A2 Mitigates Ovarian Dysfunction in POI Rats via Interaction with YBX1. A. Immunofluorescent images stained for FSHR, YBX1 of ovaries in each group. Scale bar, 20 μm. Green: FSHR stained GCs. magenta: YBX1 staining. Blue: DAPI-labeled nucleus. B. Schematic diagram of the research mechanism

CircPTP4A2 is associated with ovarian reserve function

To investigate the correlation between circPTP4A2 expression and ovarian reserve function, we collected serum samples from 30 patients with POI and 30 patients with normal ovarian function. By performing a Pearson correlation analysis, we explored the relationships of serum circPTP4A2 levels with the concentrations of AMH, E₂, LH, and FSH in patients with POI. The results indicated that circPTP4A2 expression was positively correlated with AMH and E₂ levels and the antral follicle count (AFC) and inversely correlated with LH and FSH levels (Supplementary Fig. 6A-I). These correlations were replicated in the serum of all the subjects. Overall, these findings suggest that circPTP4A2 may be associated with ovarian reserve function.

Discussion

Hypoxic preconditioning enhances the reparative effects of MSC-Exs. Consistently, our findings indicate that H-Exs outperform N-Exs in facilitating the restoration of mitochondrial function after it is impaired by oxidative stress in GCs. Additionally, we discovered that circPTP4A2 can reduce mitochondrial dysfunction and mitigate ageing by interacting with YBX1 and inhibiting its degradation. Moreover, our research demonstrates that in vivo, H-Exs-circPTP4A2 enhanced ovarian function and increase fertility in POI model rats, indicating a role for circPTP4A2 in POI and its considerable potential as a therapeutic target.

Hypoxia triggers a series of adaptive responses in MSCs, leading to the secretion of exosomes enriched with bioactive molecules. These exosomes exhibit enhanced proliferation, migration, and antioxidant properties, which are vital for effective tissue repair [29, 30]. Notably, H-Exs demonstrate superior efficacy in mitigating mitochondrial dysfunction in models of intestinal injury, such as ulcerative colitis [18]. These exosomes help colocalize mitochondria and lysosomes, promoting autophagy, which is essential for cellular recovery. In addition, after hypoxic pretreatment, the contents of exosomes, such as BNIP3, miR-216a-5p, miR-126, miR-21-5p, circ_0000495 [29, 31–34] increase significantly, thereby effectively regulating various repair processes such as antioxidation, antiapoptotic, and proangiogenic processes. For example, after hypoxic pretreatment, the production of exosomal miR-126 can be mediated by the activation of HIF-1α. H-Exs can promote fracture healing by transmitting miR-126 and participating in the SPRED1/Ras/Erk signalling pathway Our findings suggest that H-Exs may offer a promising therapeutic avenue for enhancing mitochondrial function in GCs, further increasing ovarian reserve in a POI rat model.

The regulation of exosome content under hypoxic conditions is mediated mainly by HIF-1α [35]. Studies have shown that HIF-1α can bind to the promoter of RRAGB and promote transcription to drive tumour progression [36]. These findings indicate a potential feedback loop involving mTORC1 that may modulate HIF-1α translation. Moreover, this regulatory pattern encompassing HIF-1α and circRNAs suggests a broader network, as evidenced by the interaction of HIF-1α with other genes, such as PKD1, ZNF91 and PLEKHM1 [24, 37, 38], which enhances the transcriptional activity of circRNAs. The findings presented here in demonstrate that HIF-1α activates circPTP4A2 by binding to HREs on host genes, revealing a shared mechanism influencing circRNA expression.

CircPTP4A2 plays a crucial role in various cellular functions, particularly in relation to ageing-related disorders. A previous study demonstrated that circPTP4A2 stabilizes mitochondrial metabolism in MSCs under hypoxic conditions by promoting PDK2 expression through the sponging of miR-330-5p [39]. Unlike other studies highlighting the role of circPTP4A2 in promoting proliferation and invasion in cancers, such as cervical and non-small cell lung cancer, our research emphasizes its protective functions in noncancerous contexts, specifically in ageing [40, 41]. While previous investigations have explored the effects of circPTP4A2 on immune responses and microglial polarization [41–43], the present study reveals that circPTP4A2 as a key cargo of H-Exs, coordinates mitochondrial function with cell survival in ovarian granulosa cells by specifically binding to the transcription factor YBX1. Future studies exploring its role within exosomes may offer additional insights into its therapeutic potential for age-related diseases.

Our study reveals that the reparative effect of H-Exs on CTX-KGNs was weakened after circPTP4A2 silencing, supporting the presence of other influential factors. Hypoxic environments can alter the transcription and translation processes of cells, and these changes may be reflected in the protein and RNA composition of secretory exosomes. We speculate that hypoxic exosomes may be rich in other miRNAs and proteins, which can enhance the function of ovarian mitochondria and may even contain circRNAs that work in synergy with circPTP4A2. For instance, circDennd2a can alleviate granulosa cell damage induced by oxidative stress [44]. In the future, by comparing and analysing proteomics and RNA sequencing data for hypoxia-exposed exosomes and normal exosomes, we will help identify these potential cofactors and provide deeper insights to achieve a comprehensive understanding of the secretome of hypoxia-exposed exosomes.

YBX1 belongs to the RNA-binding protein (RBP) family and is a pleiotropic protein that contains a conserved cold shock domain (CSD) [45]. YBX1 can regulate gene expression by regulating transcription, translation, posttranslational modification, and RNA stability [46–49]. Previous studies have demonstrated that noncoding RNAs can modulate the stability, location and activity of YBX1. For instance, circRNA-SORE has been shown to bind to YBX1 via the Y-box sequence, preventing its degradation by the E3 ubiquitin ligase PRP19 and thereby promoting cell survival in hepatocellular carcinoma [50]. CircIPO7 binds to YBX1 and promotes its phosphorylation at serine 102 (p-YBX1S102) by the kinase AKT, thereby promoting its nuclear localization and activating FGFR1, TNC, and NTRK1 transcription [51]. Similarly, the lncRNA FGD5-AS1 stabilizes YBX1 through specific recognition motifs (UAAUCCCA, ACCAGCCU and CAGUGAGC), leading to increased cell proliferation and reduced oxidative stress [52]. These findings suggest that YBX1 may be a key hub protein for the function of circRNAs. On the basis of the abnormal expression of both YBX1 and circRNA in POI that we observed, we strongly speculate that there is a functional interaction between the two. First, we revealed that the reparative effect of H-Exs increased and that the expression of circPTP4A2 significantly increased. Second, we investigated RNA-binding proteins (RBPs) that interact with RNAs, such as ZAR1/2 [53], PCBP2 [54], MSI1 [55]and YBX1, which are related to ovarian reserve function, especially granulosa cell function. Finally, molecular dynamics simulations revealed that the docking effect of YBX1 was the best, and the accuracy of this prediction was verified through experiments. Molecular dynamics simulations revealed that circPTP4A2 (U-331, A-326, A-146, G-144, G-145, G-333, C-147, U-327, and U-143) can interact with YBX1 (ARG-69, LYS-137, ASP-83, ALA-120, LYS-118, ASN-70, TRP-65, TYR-138, and LYS-52).The results of this investigation elucidated the interaction between circPTP4A2 and YBX1, revealing that circPTP4A2 not only binds to YBX1 but also increases its stability by protecting it from proteasomal degradation. This protective mechanism was evidenced by the observed decrease in YBX1 protein levels upon circPTP4A2 silencing, as well as the restoration of YBX1 levels in the presence of the proteasome inhibitor MG132. Additionally, functional assays revealed that circPTP4A2 plays a pivotal role in improving mitochondrial function through the stabilization of YBX1, thereby promoting the repair of oxidative damage in GCs.

We believe that the circPTP4A2-YBX1 axis is highly specific to POI and GCs. First, the normal function of GCs is strongly dependent on mitochondrial energy metabolism to maintain steroid hormone synthesis and follicular development. Second, oxidative stress and hormonal imbalance in the pathological microenvironment of POI may create a unique molecular background that enables the same circRNA to ultimately perform cellular protective functions through different downstream pathways rather than promoting lesion formation.

CoCl_2, as a chemical used to stimulate hypoxia, can replace ferrous ions at the catalytic sites of proline hydroxylase and aspartate hydroxylase, thereby stabilizing HIF-1α. Therefore, the treatment of cells with CoCl₂ can increase the expression of HIF1α, which simulates changes in intracellular bioenergetics due to hypoxia.

Although CoCl₂ has the advantages of simple operation and cost-effectiveness in inducing the stability of HIF-1α, it cannot fully reproduce the complex physiological hypoxic microenvironment. In addition, as a chemical stressor, CoCl₂ may produce cytotoxic effects at higher concentrations or after prolonged exposure. The CoCl₂ concentration and treatment time (100 µM, 36 h) used in this study were based on a complete concentration–time study of the specific primary cells used in the research. Future research on physical hypoxia will be highly valuable, as it may verify and expand our findings in an environment that more closely mimics physiological conditions.

Although our results theoretically confirm that circPTP4A2 is a potential target for the treatment of POI, we must acknowledge the limitations of the in vivo experimental methods used and that the assembly of circRNAs into H-Exs has not been studied in depth. We verified the therapeutic effect of circPTP4A2 through the use of H-Exs rather than directly and specifically targeting circPTP4A2. Therefore, the potential of circPTP4A2 as a direct target for POI treatment requires future research, such as through the enhancement of its endogenous expression level by engineering of circRNA expression vectors. In addition, how to selectively assemble circRNAs and others into H-Exs under hypoxic conditions requires further investigation.

Although we did not directly evaluate the off-target effect of circPTP4A2 in nonovarian tissues, the use of exosomes as vectors offers a potential advantage because compared with synthetic viral vectors or nanoparticles, exosomes have the inherent ability to home to damaged tissues and exhibit lower immunogenicity. Studies have shown that N-Exs can accumulate preferentially in the ovaries, especially under conditions of inflammation or injury, which may partially alleviate concerns about systemic exposure [17].

Before clinical translation, it is necessary to conduct comprehensive biodistribution and toxicological studies in animal models and to comprehensively assess the accumulation of exosomes rich in circPTP4A2 in major organs and their potential side effects. There are significant obstacles to the production of clinical-grade H-Exs, including standardization under hypoxic conditions, large-scale MSC culture, repeatable exosome isolation, and strict quality control.

Conclusion

In conclusion, H-Exs significantly outperformed N-Exs in alleviating oxidative stress and enhancing mitochondrial function in GCs, which was facilitated by the interaction of circPTP4A2 with YBX1. This interaction highlights the crucial role of circPTP4A2 in mitigating cellular senescence and maintaining mitochondrial integrity. HIF-1α transcriptionally bound to the PTP4A2 promoter region, promoting its transcription and leading to the specific enrichment of circPTP4A2 in H-Exs. Furthermore, the downregulation of circPTP4A2 expression in GCs and serum from patients with POI correlated with impaired ovarian reserve dysfunction. While this regulatory pathway has been established, gaps in knowledge regarding the intricate molecular mechanisms linking circPTP4A2 with YBX1 have been identified, necessitating further exploration. Overall, these findings underscore the feasibility of circPTP4A2 as a novel therapeutic target for POI, suggesting promising implications for future clinical applications in fertility restoration.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

POI

premature ovarian insufficiency

MSCs

mesenchymal stem cells

Exs

exosomes

MSC-Exs

mesenchymal stem cell-derived exosomes

H-Exs

exosomes from hypoxic-preconditioned MSCs

GCs

granulosa cells

N-Exs

normoxic MSC-Exs

YBX1

Y-box binding protein 1

HIF-1α

hypoxia-inducible factor 1-alpha

HRT

hormone replacement therapy

circRNAs

Circular RNA

TEM

transmission electron microscopy

NTA

Nanoparticle tracking analysis

DAPI

4′,6-diamidino-2-phenylindole

qRT-PCR

Quantitative real-time polymerase chain reaction

MMP

Mitochondrial membrane potential

SA-β-gal

Senescence-Associated β-Galactosidase Activity

SOD

superoxide dismutase

MDA

malondialdehyde

GPX

Glutathione peroxidase

OCR

Oxygen consumption rate

CHIP

Chromatin immunoprecipitation

RIP

RNA Immunoprecipitation

SD

Sprague-Dawley

CHX

cycloheximide

MG132

proteasome inhibitor

IVF/ICSI-ET

in vitro fertilization or intracytoplasmic sperm injection embryo transfer

FSH

follicle stimulating hormone

E₂

estradiol

AMH

anti-mullerian hormone

LH

luteinizing hormone

AFC

antral follicle count

Author contributions

Xiaolan Zhu obtained the funding and designed this study; Xuyan Shi, Jingjing Lu, Wenxin Li, Yueqin Liu, Lin Jiang and Yanting Lv performed the clinical studies; Xuyan Shi, Jingjing Lu performed the experiments; Xuyan Shi wrote the manuscript. Xiaolan Zhu revised the manuscript. All authors read, revised, and approved the final manuscript.

Funding

This research was supported by grants from the National Natural Science Foundation of China (Grant No. 82172838), the 333 Project Excellent Young Talents Project of Jiangsu Province, the Natural Science Foundation of Jiangsu Province (Grant No. BK20241860), the Key Medical Research Projects of Jiangsu Provincial Health Commission (Grant No. K2023078).

Data availability

No datasets were generated during the current study.

Declarations

Consent for publication

Not applicable.

Moral statement

The collection of patients’ GCs and serum samples was approved by the Ethics Committee of The Fourth Affiliated Hospital of Jiangsu University (Zhenjiang Maternal and Child Health Hospital), which was in accordance with the ethical requirements, and all subjects participated voluntarily and signed a written informed consent.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.