Electroacupuncture may improve early ovarian failure by reducing a specific RNA modification that affects cell recycling in ovarian support cells

Female C57BL/6 mice (SPF grade), aged 6–8 weeks and weighing approximately 20 ± 2 g, were sourced from the Animal Experiment Center of the First People's Hospital affiliated with Shanghai Jiao Tong University School of Medicine, animal ethics approval number: 2020AW126. The mice were maintained in barrier facilities utilizing individually ventilated cage systems under controlled conditions: temperature (22°C ± 1°C), humidity (55% ± 5%), and a 12‐h light/dark cycle. Sterilized feed and acidified water (pH 2.5–3.0) were provided ad libitum. All procedures adhered to animal ethics standards and received approval from the institutional ethics committee.

Following a 7‐day acclimatization period, the mice were randomly allocated into three groups (n = 15 per group): wild‐type (WT), POI model, and EA intervention. The POI and EA groups received a daily high‐fat diet (8 g/kg) and intragastric administration of a 30% lactulose solution (200 μL) for 8 weeks, while the WT controls were fed a standard diet and received an equal volume of saline (Liu, Lin, et al. 2021; Liu, Jing, et al. 2021; Zhu et al. 2021). Successful modeling of POI was defined by the following criteria: (1) elevated serum FSH and decreased E2 levels; (2) histopathological evidence of reduced follicular development and increased atretic follicles (Mo et al. 2024; Geng, Nie, et al. 2023).

In accordance with standardized mouse acupoint guidelines (China Association of acupuncture and Moxibustion 2025), the restrained mice in the EA group received needle insertion at the Guanyuan (CV4), bilateral Sanyinjiao (SP6), and Zusanli (ST36) acupoints. Sterile needles (0.18 mm × 13 mm) were inserted vertically to a depth of 1–2 mm and manually stimulated through lifting, thrusting, and rotation for 1 min. Subsequently, the needles were connected to an SDZ‐V electroacupuncture device, delivering a continuous wave at 2 Hz and an intensity of 0.1–1 mA for 30 min. The treatment intensity was calibrated to elicit localized tremors without causing vocalization or agitation. EA treatment was administered every other day for 4 weeks. Mice in the WT and POI groups underwent identical restraint procedures without electroacupuncture.

Fixed ovaries were dehydrated, paraffin‐embedded, sectioned, and dewaxed in xylene. Rehydration proceeded through graded ethanol to distilled water. Sections were stained with hematoxylin (Beyotime, C0105M, China) for 5 min, rinsed, differentiated in 1% acid alcohol, counterstained with eosin for 15 s, rinsed again, dehydrated through graded ethanol, and cleared in xylene. Mounted sections (Leica RM2235, Germany) were imaged under a light microscope (Nikon Eclipse Ci‐L, Japan). Follicle counting was performed in every fifth serial section to avoid double counting of the same follicle. Follicles were classified according to established morphological criteria as primordial, primary, secondary, and antral follicles. Only follicles containing a clearly visible oocyte nucleus were included in the analysis. Atretic follicles were identified based on the presence of granulosa cell disorganization, pyknotic nuclei, or oocyte degeneration and were recorded separately. Follicle numbers were quantified in a blinded manner by two independent investigators. The average number of follicles per ovary was calculated for statistical analysis.

Daily vaginal smears were collected between 08:00 and 10:00 for ten consecutive days. Using sterile saline‐moistened swabs, secretions were gently obtained from the vaginal canal and transferred to glass slides. After air‐drying, smears were stained with crystal violet for 1–2 min, rinsed with distilled water, and dried. Estrous stages were determined by examining cellular morphology under light microscopy.

Dewaxed and rehydrated sections were triple‐stained according to kit protocols (Beyotime, Shanghai, China; Cat# C0189M). Briefly: sections were stained with hematoxylin (50 μL), differentiated in acid solution (30 s), and rinsed with distilled water; treated with Ponceau‐acid fuchsin (50 μL, 10 min), differentiated in phosphomolybdic acid (2 min), and rinsed; stained with bright green solution (50 μL, 1 min), and differentiated in acid solution (1 min). Sections were subsequently dehydrated, cleared in xylene, mounted, and imaged.

Dewaxed sections were stained with Sirius Red/picric acid solution (Abcam, Cambridge, UK; Cat# ab150681) for 60 min under dark conditions, rinsed three times in absolute ethanol (10 s per rinse), cleared in xylene, and mounted. Collagen fibers were identified by bright red birefringence under light microscopy.

Serum follicle‐stimulating hormone (FSH), estradiol (E2), interleukin‐1β (IL‐1β), and tumor necrosis factor‐α (TNF‐α) levels were quantified using commercial ELISA kits (Geneme Biotechnology, Shanghai, China; Cat# JM‐02838M1, JM‐02849M2, JM‐02323M1, and JM‐02415M1, respectively). Microplates included standard, sample, and blank wells. Following reaction termination, absorbance at 450 nm was measured within 15 min using a microplate reader (BIO‐RAD 550, Hercules, CA, USA), with blank wells serving as the zero reference.

Dewaxed and rehydrated sections were processed according to kit instructions (Genomeditech, Shanghai, China; Cat# GMMS80079.2.1). Briefly: 500 μL dewaxing solution covered sections, incubated at 70°C for 5 min, and residual solution was drained. Sections were sequentially treated with 200 μL purification solution (2 min, room temperature [RT]), 200 μL clearing solution (3 min, RT), and 200 μL blocking solution (10 min, RT), with each solution carefully drained after incubation. Under dark conditions, 100 μL staining solution was applied for 30 min (RT) to bind free cholesterol. Sections were rinsed with 200 μL clearing solution, mounted, and analyzed via fluorescence microscopy for cholesterol‐positive (bright blue) signals.

Dewaxed and rehydrated sections were treated with 20 g/mL DNase‐free proteinase K at 37°C for 15 min, rinsed with PBS, and incubated with 50 μL TUNEL detection reagent (Beyotime, Cat# C1086) in the dark (37°C, 60 min). After three 5‐min washes in PBS, sections were mounted with anti‐fade medium and imaged via fluorescence microscopy to assess granulosa cell apoptosis.

Dewaxed sections underwent antigen retrieval twice, were blocked with 5% bovine serum albumin (BSA) at 37°C for 1 h, and incubated with primary antibodies overnight at 4°C. Primary antibodies included: anti‐FSHR (Proteintech, Rosemont, IL, USA; Cat# 22665‐1‐AP; 1:200). Following three 5‐min washes with Tris‐buffered saline with Tween‐20 (TBST), sections were incubated with secondary antibodies (goat anti‐rabbit IgG Alexa Fluor555, ab150078) at 37°C for 1 h. Sections were mounted with 4′,6‐diamidino‐2‐phenylindole (DAPI)‐containing medium (Beyotime, Shanghai, China; Cat# P0131) and imaged.

Ovarian tissues were preserved in MACS Tissue Storage Solution (Miltenyi Biotec, Bergisch Gladbach, Germany; Cat# 130‐100‐008) immediately post‐dissection. Single‐cell suspensions were encapsulated with gel beads in oil droplets (GEMs) using the Chromium X platform (10× Genomics, Pleasanton, CA, USA). Messenger RNA (mRNA) released from lysed cells bound barcoded primers for complementary DNA (cDNA) synthesis. Purified cDNA was amplified to construct sequencing libraries.

Data were processed using Seurat (version 4.3.0). Quality control thresholds were applied as follows: nFeature_RNA < 5000, mitochondrial gene percentage < 5%, ribosomal gene percentage > 3%, and hemoglobin gene percentage < 0.1%. After filtering (retaining 24,478 genes across 41,513 cells), data were normalized using the ScaleData function, dimensionality‐reduced via principal component analysis (PCA), and batch‐effect corrected using Harmony. Clustering (resolution = 0.3) and uniform manifold approximation and projection (UMAP) were performed for visualization. Cell types were manually annotated based on established marker genes.

Ovaries from three mice per group were fixed in 2.5% glutaraldehyde (Servicebio, Wuhan, China; Cat# G1102) at room temperature for 30 min, then stored overnight at 4°C. Tissues were rinsed in phosphate‐buffered saline (PBS), post‐fixed in 1% osmium tetroxide, dehydrated in ethanol, embedded in resin, and sectioned (40–50 nm thickness). Uranyl acetate‐stained sections were imaged using an FEI Tecnai G2 F20 transmission electron microscope (Thermo Fisher Scientific, Hillsboro, OR, USA).

Antigen‐retrieved sections were blocked with 3% hydrogen peroxide at room temperature for 10 min, followed by goat serum (Beyotime, Cat# C0265) for 10 min. Primary antibodies—anti‐LC3A/B (Affinity, Cat# AF5402), anti‐Beclin1 (Cell Signaling Technology, Danvers, MA, USA; Cat# 3738), and anti‐P62 (Cell Signaling Technology; Cat# 5114)—were applied overnight at 4°C. After PBS washes, sections were incubated with horseradish peroxidase (HRP)‐conjugated secondary antibody (Abcam, Cat# ab205718) at room temperature for 10 min, treated with streptavidin‐peroxidase for 10 min, and developed with 3,3′‐diaminobenzidine (DAB; incubation: 30 s–3 min). Sections were counterstained with hematoxylin, dehydrated, mounted, and analyzed via light microscopy.

Following protein extraction, ovarian tissue was washed with ice‐cold PBS and cut into 1 mm³ pieces. Approximately 20 mg of tissue was added to 60 μL of RIPA lysis buffer, homogenized to achieve lysis, and then centrifuged at 10000–14000 g for 3–5 min at 4°C to collect the supernatant. Protein quantification was subsequently performed using the BCA method (Epizyme Biotech Co. Ltd., Cat# ZJ102L, Shanghai, China), where the A562 value was measured using a 96‐well plate and the concentration was calculated based on a standard curve. Protein samples, after being mixed with loading buffer and boiled at 100°C for 10 min, were loaded onto self‐prepared SDS‐PAGE gels. A dual‐color pre‐stained protein marker was used to estimate molecular weights. Electrophoresis was first conducted at 80 V for 30 min, followed by protein separation at 120 V for 1.5 h. After electrophoresis, proteins were transferred onto PVDF membranes activated with methanol, using a transfer cassette assembled with WB‐specific transfer sponges, under 200 mA current for 2 h. The membranes were then blocked with a no‐protein fast blocking buffer for 10 min. Subsequently, the membranes were incubated with primary antibodies (Anti‐LC3A/B antibody, Affinity Biosciences, Cat# AF5402, USA; Anti‐Beclin 1 antibody, CST, Cat# 3738, USA; Anti‐P62 antibody, CST, Cat# 5114, USA; Recombinant Anti‐METTL3 antibody, Abcam, Cat# ab195352, UK; Anti‐Beclin 1 antibody, Affinity Biosciences, Cat# AF5128, USA) placed in a humidified chamber overnight at 4°C. This was followed by incubation with the secondary antibody (Goat Anti‐Rabbit IgG H&L (HRP), Abcam, Cat# ab205718, UK) at room temperature for 1 h. Enhanced ECL chemiluminescent substrate kit (Vazyme Biotech Co. Ltd., Cat# E411‐04, Nanjing, China) was used for detection, and band images were captured and analyzed using either a Tanon 4600 imaging system (Tanon Life Science Co. Ltd., Model Tanon 4600 SF, Shanghai, China) or a gel imager (BIO‐RAD, Model 2500, USA).

RNA extraction was first performed by grinding 10–20 mg of ovarian tissue (approximately the size of a grain of rice) in 500 μL of Lysis Buffer until complete lysis was achieved, followed by vigorous vortexing for 30 s and incubation at room temperature for 5 min, after which the lysate was centrifuged at 12,000 rpm for 5 min to collect the supernatant. The supernatant was then passed through HyperPure gDNA‐Filter Columns placed in collection tubes and centrifuged at 12,000 rpm for 5 min to remove genomic DNA and impurities. An equal volume of absolute ethanol (approximately 250 μL) was added to the supernatant, mixed thoroughly, and transferred to HyperPure RNA Columns in collection tubes. Sequential washes were performed using 500 μL of Wash Buffer A followed by two washes with 500 μL of Wash Buffer B, with each wash step involving centrifugation at 12,000 rpm for 30 s at room temperature. After a final 2‐min centrifugation of the empty column to remove residual ethanol, RNA was eluted using 50–100 μL of Elution Water. For the reverse transcription step, cDNA synthesis was carried out according to the HyperScript III RT SuperMix for qPCR kit protocol (EnzyArtisan Biological Co. Ltd., Cat# R102‐02, Shanghai, China). A 20 μL reaction mixture was prepared containing 1 ng–5 μg of total RNA, 4 μL of 5× SuperMix, and RNase‐free water. The reaction proceeded at 37°C for 15 min, followed by enzyme inactivation at 85°C for 5 s. For quantitative PCR, reactions were assembled according to the 2× S6 Universal SYBR qPCR Mix protocol (EnzyArtisan Biological Co. Ltd., Cat# Q204‐01, Shanghai, China). A 20 μL reaction mixture was prepared containing 10 μL of 2× SYBR mix, 2 μL each of 10 μM forward and reverse primers, 2 μL of cDNA template, and 4 μL of ddH₂O. Amplification was performed on a QuantStudio 6 Flex Real‐Time PCR System (Thermo Fisher Scientific, Instrument Model, USA) using the following program: initial denaturation at 95°C for 30 s (1 cycle), followed by 40–45 amplification cycles of 95°C for 3–10 s and 60°C for 10–30 s, with a final melt curve analysis step using the instrument's default settings. Relative mRNA expression levels of target genes were calculated using the 2^(−ΔΔCT) method with 18S rRNA as the internal reference gene; primer sequences for target genes are listed in Table 1.

Total m6A RNA methylation levels in mouse ovarian tissues were quantified using an m6A RNA methylation quantification kit (Abcam, Cambridge, UK; Cat# ab233491). Reagents were pre‐diluted as follows: 10× wash buffer (1:9 dilution to 1× working solution), capture antibody (1:1000), detection antibody (1:2000), and enhancer solution (1:5000). Positive control (0.5 ng/μL) and standards (0.01–0.50 ng/μL) were prepared in Tris‐EDTA (TE) buffer. For RNA binding, each well received 80 μL binding solution, 2 μL negative/positive controls, and sample RNA (100–300 ng total RNA), incubated at 37°C for 90 min, followed by three 2‐min washes. For m6A RNA capture, 50 μL capture antibody was added and incubated at room temperature for 60 min, washed; 50 μL detection antibody incubated at room temperature for 30 min, washed; and 50 μL enhancer incubated at room temperature for 30 min, washed. For colorimetric detection, 100 μL developer was added per well and incubated under dark conditions at room temperature until positive controls displayed medium blue coloration (5–10 min). Absorbance at 450 nm was measured after reaction termination.

mRNA was isolated using NEBNext Poly(A) magnetic beads (New England Biolabs, Ipswich, MA, USA), followed by enzymatic digestion and dephosphorylation. After hydrolysis at 37°C for 3 h and purification, samples were analyzed using an Agilent 6460 triple quadrupole mass spectrometer coupled to a 1260 high‐performance liquid chromatography system (Agilent Technologies, Santa Clara, CA, USA).

Total ovarian RNA was fragmented at 94°C for 5 min into 100–200 nucleotide segments. Fragmented RNA was incubated with m6A‐specific antibody (ABclonal, Woburn, MA, USA; Cat# A19841) at 4°C for 2 h with rotation. RNA‐antibody complexes were captured using Protein A/G magnetic beads, washed to remove nonspecific binding, and digested with proteinase K. Enriched m6A‐modified RNA was extracted for RT‐qPCR analysis.

Data analysis was performed using GraphPad Prism 9 and SPSS Statistics 29. Results are expressed as mean ± standard deviation (SD). Normality was assessed using Shapiro–Wilk testing: normally distributed data were compared using Student's t‐test or one‐way analysis of variance (ANOVA) with Tukey's post hoc test for multiple comparisons; non‐normally distributed data were analyzed using Mann–Whitney U test or Kruskal–Wallis test with Dunn's multiple comparisons test.

The HFHS diet‐induced POI mouse model was established, and EA was administered at Guanyuan (CV4), bilateral Sanyinjiao (SP6), and Zusanli (ST36) acupoints to evaluate therapeutic efficacy (Figure 1A). HE staining revealed a significant decrease in primordial follicles and primary + secondary follicles, as well as an obvious increase in atretic follicles in POI group ovaries compared to WT, confirming successful modeling (Figure 1B,C).

Vaginal cytology demonstrated regular 3–4 day estrous cycles in WT mice, progressing through proestrus (predominantly nucleated epithelial cells with few leukocytes), estrus (cornified epithelial cells), metestrus (mixed epithelial cells and leukocytes), and diestrus (predominantly leukocytes). POI mice predominantly remained in metestrus or diestrus without completing cycles, indicating reproductive dysfunction (Figure 1D). Masson's trichrome and Sirius Red staining showed elevated collagen deposition in POI ovarian stroma, suggesting fibrosis (Figure 1E,F). EA intervention normalized estrous cyclicity and improved ovarian morphology.

Hormonal analysis indicated significantly decreased serum E2 and elevated FSH in POI mice; both parameters were restored to near‐normal levels after EA (Figure 1G). Additionally, EA effectively reduced the increased serum IL‐1β and TNF‐α levels observed in POI mice (Figure 1H), demonstrating anti‐inflammatory effects.

Filipin staining revealed substantial cholesterol accumulation in POI granulosa cells, which was significantly reduced by EA intervention (Figure 1I). TUNEL staining indicated elevated apoptosis in POI granulosa cells, attenuated after EA treatment (Figure 1J). Immunofluorescence showed downregulated follicle‐stimulating hormone receptor (FSHR) expression in POI ovaries, which EA rescued (Figure 1K), suggesting EA may inhibit granulosa cell apoptosis through FSHR upregulation.

Collectively, EA restored ovarian structure and function in HFHS‐induced POI mice by promoting folliculogenesis, normalizing estrous cycles, suppressing granulosa cell apoptosis, rebalancing sex hormones, and attenuating ovarian inflammation.

ScRNA‐seq was performed on ovarian tissues from HFHS‐induced POI mice following the workflow illustrated in Figure 2A and Figures S1 and S2. After quality control and unsupervised clustering, ovarian cells were classified into distinct subpopulations based on established marker genes (Figure 2B). Autophagy‐related gene scores were subsequently calculated across all identified cell types, allowing stratification into high‐ and low‐autophagy activity groups (Figure 2C). Marked heterogeneity in autophagy scores was observed among ovarian cell populations, with granulosa cells exhibiting the most pronounced alterations in autophagy‐related transcriptional activity (Figure 2D). Consistently, both the number and relative proportions of ovarian cell subtypes differed significantly between the high‐ and low‐autophagy score groups (Figure 2E). GSVA correlation analysis further revealed significant positive associations between granulosa cells and multiple functional pathways, primarily involving energy and lipid metabolism, stress and inflammatory responses, apoptosis and survival regulation, hormone‐responsive signaling, as well as cell cycle and proliferation control. These findings indicate extensive functional reprogramming of granulosa cells during HFHS‐induced POI (Figure 2F). HALLMARK enrichment analysis demonstrated that inflammation‐ and stress‐related pathways, including P53_PATHWAY, TNFA_SIGNALING_VIA_NFKB, and IL6_JAK_STAT3_SIGNALING, were significantly upregulated in granulosa cells, whereas estrogen‐responsive pathways (ESTROGEN_RESPONSE_EARLY and ESTROGEN_RESPONSE_LATE) were markedly downregulated (Figure S3). KEGG pathway analysis revealed that ovarian cell subpopulations were enriched in signaling pathways such as PI3K–Akt and PPAR, both of which are closely involved in the regulation of granulosa cell autophagy. In parallel, the steroid hormone biosynthesis pathway was strongly associated with granulosa cell function, implicating its role in maintaining organelle homeostasis and hormone production during steroidogenesis. GO enrichment analysis further confirmed that granulosa cells are critically involved in female gonadal development and follicular maturation (Figure S4). Cell–cell interaction analysis indicated extensive communication among ovarian cell subpopulations, with granulosa cells exhibiting prominent interactions with multiple cell types that may cooperatively regulate the initiation and progression of autophagy (Figure S5). Pseudotime trajectory analysis further supported close functional and developmental associations among these cell populations (Figure S6). Moreover, comparison between the POI and WT groups revealed a significant difference in ovarian cell autophagy scores at the single‐cell level (Figure 2G, ****p < 0.0001), highlighting pronounced autophagy dysregulation in the POI model.

TEM revealed ultrastructural alterations in ovarian granulosa cells across groups. POI group granulosa cells exhibited abundant autophagosome accumulation and disrupted mitochondrial integrity, indicating hyperactivated autophagy with organelle damage. In contrast, EA intervention markedly reduced autophagosome numbers and restored mitochondrial structural integrity, demonstrating mitigation of autophagy‐related organellar injury (Figure 3A). IHC staining confirmed expression changes of key autophagy regulators in granulosa cells. LC3 expression was significantly elevated in POI group granulosa cells but substantially reduced by EA treatment (Figure 3B). At molecular levels, Western blot analysis showed upregulated Beclin1 protein, downregulated P62, and increased LC3‐II/LC3‐I ratio in POI ovaries. RT‐qPCR corroborated these findings with elevated Beclin1 and LC3B mRNA alongside decreased P62 mRNA. EA intervention reversed these abnormalities: Beclin1 protein/mRNA decreased, P62 protein/mRNA increased, and LC3‐II/LC3‐I ratio declined (Figure 3C,D). Collectively, EA ameliorates ovarian microenvironment by suppressing excessive granulosa cell autophagy through coordinated downregulation of Beclin1/LC3 and upregulation of P62, thereby favoring follicular development.

m6A—the most prevalent mammalian mRNA modification—dynamically regulates mRNA splicing, stability, and translational efficiency through reversible epigenetic mechanisms (Fu et al. 2014). LC–MS analysis of ovarian mRNA modifications revealed elevated levels of multiple mRNA modifications in POI mice, which EA intervention effectively normalized (Figure 4A). Colorimetric quantification further confirmed increased total m6A methylation in POI ovaries, significantly reduced by EA treatment (Figure 4B), indicating EA modulates ovarian m6A epitranscriptomic profiles.

Methyl transferase METTL3 has been extensively reported in gynecological disorders, such as influencing pregnancy outcomes by regulating placental function (Li et al. 2026) or participating in polycystic ovary syndrome (PCOS) development through modulating the Onecut2 factor (Sui et al. 2026). Based on this, the present study focused on the expression and role of METTL3 in POI. RT‐qPCR results revealed that METTL3 mRNA expression was upregulated in POI ovarian tissue, while EA intervention suppressed its expression (Figure 4C). Western blot and immunohistochemistry results further confirmed elevated METTL3 protein expression in POI, with EA treatment reversing this trend (Figure 4D,E). Collectively, these findings suggest pathological activation of METTL3 in POI, and EA may exert therapeutic effects by downregulating METTL3.

As Beclin1—a core autophagy initiator—upregulation correlates with autophagy activation (Liang et al. 1998), we investigated its role as an m6A target. Immunohistochemistry revealed increased Beclin1 protein in POI granulosa cells, suppressed by EA (Figure 4F). MeRIP‐PCR confirmed direct METTL3 binding to Beclin1 mRNA, with EA weakening this interaction. Reduced m6A methylation destabilized Beclin1 mRNA structure and diminished its translation (Figure 4G), identifying Beclin1 as a functional target of METTL3‐mediated m6A modification.

Collectively, these findings support a mechanism where METTL3‐mediated m6A hypermethylation stabilizes Beclin1 mRNA and enhances its translation, exacerbating granulosa cell autophagy in POI. EA counteracts this by downregulating METTL3 expression, attenuating m6A modification of Beclin1, and reducing Beclin1 protein levels, thereby suppressing pathological autophagy and restoring ovarian homeostasis.