Decoction of HYF, a Chinese herbal medicine prescribed at the clinic, obtained from Guangdong Kangmei pharmaceutical Company, Ltd (Guangdong, China), has been used to treat POI in Guangdong Provincial Hospital of Chinese Medicine for more than 10 years. HYF consists of seven botanical drug(s), including Astragali Radix, Herba Epimedii, Dioscoreae Rhizoma, Semen Cuscutae, Rehmanniae Radix, Angelicae Sinensis Radix, Glehniae Radix, at a ratio of 5:1:1:1:1:1:1 (Table1). After soaking for half an hour, HYF had been boiled for 1.5 h using a reflux extraction device. Then the extracted liquid was collected together, the liquid was concentrated using a rotary evaporator to a final concentration of 1.1 g⋅mL-1. This concentration of liquid is administered directly to animals by intragastric administration. The drug used in vitro cell assay is Freeze-dried, the freeze-drying conditions are: 40°C refrigeration, −20°C freezing for 2 h, −10°C freezing for 16 h, 20°C drying for 36 h, 35°C secondary drying for 36 h. Freeze-dried HYF were weighed and dissolved in F12 medium to a final concentration of 100 mg/mL and centrifuged at 14,000 rpm for 10 min, the supernatant was filtered with 0.22 μL filter before use.

Batch-to-batch consistency was monitored by quantifying marker compounds via UPLC-MS/MS analysis have been reported previously (Wang et al., 2021). Briefly, the chromatographic column was Thermo Hypersil GLOD C18 (2.1 × 100 mm, 1.9 μm). The mobile phase is 0.1% formic acid water (A) and 0.1% formic acid-acetonitrile mixture (D). Gradient elution: 0–2 min, 6% D; 2–25 min, 6%–95% D; 25–30 min, 95% D; 30–32 min, 95%–5% D. Flow rate 0.3 mL/min; Column temperature 40°C; Sample tray temperature 15°C; The sample size was 5 µL.

Mass spectrum conditions: HESI electrospray ion source; Scanning mode Full MS/dd MS2; Positive ion electrospray voltage 3.5 kV, negative ion electrospray voltage 3.2 kV; The sheath gas flow rate was 35 arb and the auxiliary gas flow rate was 10 arb. Capillary temperature 320°C; Probe heater temperature 350°C; Maximum spray current 100 A; S-Lens resolution 60; Scanning range m/z 100–1,500 Da; Quality resolution 70,000 (full width at half maxima, FWHM); Secondary resolution 17,500 FWHM.

Animal studies were conducted following the guidelines approved by the Guangdong Provincial Hospital of Chinese Medicine Institutional Animal Care and Use Committee (Registration number: 2019032; Approval date: 15 July 2019). The animals used in this study were SPF-grade SD female rats provided by Guangdong Medical Laboratory Animal Center, License No. SCXK (Guangdong) 2019–0,035.

The 28 days old SD rats were randomly divided into 3 groups using random numbers generated by computer (n = 7/group): normal control group (CON), VCD group (MOD) and Huyangyangkun formula group (HYF). Rats in VCD and HYF groups received intraperitoneal injection of VCD (Sigma Aldrich Korea, lot1302946) every day for 15 consecutive days (0.146 mL/kg) (Li et al., 2023). Then, the HYF group was gavaged with HYF (0.297 g/kg) for 100 days. Rats were finally euthanized using overdosed pentobarbital sodium and serum and ovaries were collected. Smears of rat vaginal exfoliated cells were taken between 9:00–10:00 a.m. daily (Figure 1).

To determine follicle numbers and observe follicular pathological alterations continuously, ovaries were fixed, embedded and sectioned as described before (Li et al., 2023; Nam et al., 2016). The section thickness was 5 μm, and the slices were successively sliced, and one slice was taken every 9 slices for HE staining. The sections of ovarian tissue stained by HE were observed under a microscope, and follicles were graded and counted according to a unified standard (reported previously), two pathologists performed blind numbers without knowing the grouping and took the mean of their readings for registration. The follicles were divided into 5 stages: primordial follicle, primary follicle, secondary follicle, antral follicle and mature follicle. To avoid the loss and duplication of counting, only the follicles in which oocytes can be observed are counted when counting small sinus follicles, sinus follicles, and mature follicles with larger diameters, and this is not necessary when counting small primary, primary, and secondary follicles.

Blood was collected from orbital veins of rats at 30, 60, and 100 days after VCD injection for hormonal measurements. All rats were finally euthanized during the interestrous period, and blood was collected from abdominal aorta. Anti-Müllerian hormone (AMH), and estradiol (E2) were detected referring to the instructions of the ELISA kits (CSB-E11162r, CSB-E05110r, C0100090186, CUSABIO BIOTECH, Wuhan), follicle-stimulating hormone (FSH) were provided by (Elabscience, E-EL-R0391c, China).

The COV434 line of immortalized granulosa cells were obtained from Sigma. Cells were grown at 37°C in a humidified 5% CO2 and 95% air and cultured in F12 medium (C11330500BT, Gibco, United States) containing 10% FBS (16000044, Gibco, United States) and 0.5% penicillin streptomycin sulfate (15140122, Gibco, United States). The granulosa cell apoptosis model was induced by VCD (Nam et al., 2016). Lentivirus construction was performed by Obio Technology (Shanghai, China). ShRNA interference fragment for Human FTO was designed and constructed into lentiviral vector by molecular biological means (pSLenti-U6-shRNA-CMV-EGFP-F2A-Puro-WPRE). The vector can be used to transfect cells to interfere with the expression of FTO gene in cells. pSLenti-CMV-MCS-3xFLAG-PGK-Puro-WPRE GL119 as a negative control (NC). Then, the COV434 cells were infected according to the optimal MOI value (MOI = 10), followed by screening for high purity infected COV434 cells.

Total RNA in ovaries and COV434 cells was extracted by Trizol reagent (Invitrogen, 15596026, United States) following the protocol. The mRNA was purified by Dynabeads^® mRNA Purification Kit (Thermo Fisher).

The purity of mRNA is measured by a NanoDrop method and diluted into different concentrations. For the rigor of the experiment, three experimental methods (ELISA, Dot blot and LC-MS/MS) were choosing to test the change in globa.

m6A levels in mRNA. For enzyme-labeled assay use m6A RNA Methylation Quantification Kit (Cat#ab185912, Abcam, United Kingdom), the RNA was incubated with m6A antibody in wells, and m6A levels were measured at 450 nm using a microplate reader. For Dot blot analysis and LC-MS/MS, the experimental process has been reported in detail in the previous article (Li et al., 2023), so we will not repeat it here.

The methods for total RNA extraction, mRNA purification were described above, the amount and purity of total RNA were then controlled by NanoDrop ND-1000 (NanoDrop, Wilmington, DE, United States). The integrity of RNA was then detected by Bioanalyzer 2,100 (Agilent, CA, United States), and verified by agarose electrophoresis. The mRNA with PolyA was captured using the oligo (dT) magnetic beads (Dynabeads Oligo (dT) (No. 25–61005, Thermo Fisher, United States). Fragmentation was performed under high temperature conditions using the NEBNext^® Magnesium RNA Fragmentation Module (Cat#E6150S, United States) at 86°C for 7 min. Dynabeads Antibody Coupling Kit (Thermo Fisher, CA, United States) and m6A antibody (No. 202003, Synaptic Systems, Germany) was premixed in IP buffer (50 mM Tris-HCl, 750 mM NaCl and 0.5% Igepal CA-630). The IP product was synthesized into cDNA by Invitrogen SuperScript™ II Reverse Transcriptase (Cat#1896649, CA, United States). The resulting cDNA (including IP samples and Input samples) can be directly used for the next MeRIP-qPCR and conventional RT-qPCR.The sequences of the primers are shown in the (Supplementary Table 1).

The enrichment m6A of yap was expressed as the enrichment percentage relative to the input sample (%Input) = 2^−ΔΔCT (Input)-Ct (MeRIP)× Fd× 100%, where Fd is the input dilution factor (1/8).

Cells and ovaries were harvested and lysed with 1x RIPA buffer (Beyotime, P0013B, China). Protein concentration was detected by Thermo Bicinchoninic acid (BCA, 23,227, United States) kit. Equal amounts of protein were dissolved in 5×loading buffer and separated in 10%–15% SDS polyacrylamide separation gels. The protein was transferred onto PVDF membranes (Beyotime, FFN10, China), then incubated with TBST solution containing 5% BSA for 1.5 h at room temperature and incubated with the primary antibodies at 4°C overnight. The next day, the membranes were washed and incubated for 1 hour with the secondary antibody. After the membranes were covered enhanced chemiluminescence (ECL), the images were shown by ChemiDoc XRS + chemiluminescence imaging system (Bio-rad). The results were analyzed with ImageJ software. Primary antibodies (all diluted at 1:1,000):ALKBH5, ab195377, Abcam; FTO, ab92821, Abcam; pJNK,4,668, CST; P38,8690, CST; PP38,4632, CST NRF2,16396-1-AP, Proteintech; BCL2,2876, CST; BCLXL, ab32370, Abcam; MCL1, ab32087, Abcam; BAX, 2,772, CST; BAK,12,105, CST; pBIM, ab17935, Abcam; PUMA,4,976, CST; Pro-caspase9/Cleaved-caspase9,10380-1-AP, Proteintech; Caspase3,9662, CST; Cleaved-caspase3, 9,664, CST; IgG,7,074, CST.

Data analysis was processed by SPSS 21.0 software. Mean ± Standard or Median (P25∼P75) was chosen to describe the data, depending on whether the data meet the normal distribution. t-test or Mann-Whitney U test was used to compare CON group with MOD group. One-way ANOVA or Kruskal–Wallis test was used for comparison of groups. p < 0.05 was considered statistically significant. And graphs were drawn by Graphpad Prism 8 software.

Batch-to-batch consistency was monitored by quantifying marker compounds via UPLC-MS/MS analysis. Figure 2A and b display the quantification in negative and positive ion mode. UPLC-MS/MS was used to identify the active constituents of HYF. Nine major active components have been identified through the comparison of standard compounds (Table 2).

Ovarian function can be predicted by assessing the estrous cycle, follicular development, and hormonal fluctuations. Rats were periodically weighed throughout the experimental period. It was observed that the body weight of rats in the VCD and HYF groups increased at a slower rate than that of the CON group, with this difference becoming evident by the first month of VCD modeling (Day 28). The body weight of rats in the VCD group was found to be lower than that of the CON group (p < 0.01), while the HYF group exhibited higher body weight than the VCD group (p < 0.05) (Figures 3A, B). The ovarian structures of the rats are depicted in Figure 3C. Bilateral ovaries from each rat were subjected to weight measurement, and their dimensions were recorded, specifically the long diameter (L, in millimeters) and the short diameter (S, mm). The ovarian volume was subsequently calculated using formula V = L *S²/2, mm3). When compared to the control (CON) group, the VCD group demonstrated a significant reduction in ovarian volume, amounting to 46.89%. In contrast, the group not exposed to VCD exhibited a significant increase in ovarian volume by 39.28% relative to the VCD group (Figure 3D). To evaluate alterations in the estrous cycle, we performed an analysis of vaginal cell shedding at 30, 60, and 100 days following HYF administration, in addition to daily sampling over a period of 10 consecutive days. The VCD model rats exhibited significant disruptions in their estrous cycle, which intensified with prolonged VCD exposure. Similarly, the HYF group experienced disturbances in their estrous cycle; however, some individuals in the HYF group maintained regular estrous cycles by the conclusion of the experiment (Figure 3E; Supplementary Table 2). Figure 3F illustrates the maximum section of pathological sections of the ovary in three groups of rats. The VCD group exhibited significantly lower numbers of follicles at all stages compared to the CON group (p < 0.01, p < 0.01, p < 0.01, and p < 0.05, respectively). Following HYF intervention, there was a significant increase in the number of original and mature follicles compared to the VCD group (p < 0.05, p < 0.01, respectively, Figure 3G).

Serum anti-Müllerian hormone (AMH) levels were employed as a sensitive biomarker for assessing ovarian reserve, with serum samples collected on days 30, 60, and 100 following VCD injection. No significant differences in AMH levels were observed among the three groups at 30 and 60 days post-injection. However, at 100 days post-injection, the serum AMH level in the VCD group was significantly reduced compared to the control (CON) group (p < 0.01, Figure 3H). Furthermore, the HYF group demonstrated a significant upward trend in AMH levels 100 days post-VCD injection compared to the VCD group (p < 0.05, Figure 3I). Blood samples for follicle-stimulating hormone (FSH) and estradiol (E2) were collected in accordance with the estrous cycle of rats (4–5 days), and a sex hormone cycle line chart was developed to observe their respective trends. The FSH line chart indicates that the CON group is consistent with previous literature (Olvera-Juárez et al., 2020), whereas the line chart for the VCD group exhibits a completely irregular pattern (Figure 3J and Supplementary Figure 1). The E2 line chart indicates that the rats in the normal control group display periodic fluctuations, aligning with previous findings by Olvera-Juárez et al. (2020), in contrast, the E2 line chart for the VCD group exhibits a disordered pattern, characterized by the absence of wave crests or the emergence of multiple peaks. Meanwhile, the HYF group shows a slight degree of regularity, evidenced by the presence of wave crests. Furthermore, a comparison of baseline E2 levels revealed no statistically significant differences among the three groups (Figure 3J and Supplementary Figure 1).

TUNEL staining was first used to detect the apoptosis of ovarian granulosa cells in the CON, VCD and HYF groups. As shown in Figure 4A, CON showed less apoptosis signals to the naked eye, while the ovaries in the VCD-induced POI group showed a lot of apoptosis signals. On the contrary, granulosa cell apoptosis in POI rats decreased after HYF intervention. Meanwhile, according to this result and previous research basis, we detected the expression levels of various anti-apoptotic proteins associated with both extrinsic and intrinsic by WB experiment.

TUNEL staining was initially employed to assess apoptosis in ovarian granulosa cells across the CON, VCD, and HYF groups. In Figure 4A, the CON group exhibited minimal apoptotic signals visible to the naked eye, whereas the VCD-induced POI group demonstrated a substantial increase in apoptotic signals. Conversely, granulosa cell apoptosis in POI rats was reduced following HYF intervention. Based on these findings and previous research, we conducted Western blot (WB) experiments to evaluate the expression levels of various anti-apoptotic proteins associated with both extrinsic and intrinsic pathways.

The results of the rat ovarian tissue test revealed that in the VCD group, there was an increase in the expressions of pro-apoptotic proteins BAX, BAK, and pBIM (p < 0.01, p < 0.05 and p < 0.05, respectively), as well as an increase in the expression of Caspase9 protein in both the precursor Pro-caspase9 and Cleaved caspase9 forms (p < 0.05). Additionally, the expression of Cleaved caspase3 was also found to be upregulated (p < 0.01). In contrast, the HYF group exhibited an increase in the anti-apoptotic protein BCLXL in the BCL-2 family (p < 0.01), a decrease in the pro-apoptotic protein BAK (p < 0.01), and a decrease in both Cleaved-caspase9 and Cleaved-caspase3, when compared to the VCD group (p < 0.01, Figures 4B, C).

To explore the potential upstream molecular targets of HYF anti-mitochondrial apoptosis, based on the analysis of relevant literature and preliminary research, we established a potential pool of upstream signals consisting of eight potential targets, namely, JNK, pJNK, P38, pP38, NRF2, HO1, NQO1, and P53. These signals were investigated as potential regulators of the mitochondrial apoptosis pathway.

The results indicated that compared to the control group, the expressions of JNK, pJNK, and P53 were significantly increased in the VCD group (p < 0.05), while no statistically significant differences were observed for P38, pP38, NRF2, HO1, and NQO1. In comparison to the VCD group, the expression levels of JNK, pJNK, and P53 were significantly downregulated (p < 0.01, p < 0.05, and p < 0.05, respectively), while no statistically significant differences were observed for P38, pP38, NRF2, and HO1 proteins (Figures 4D, E). Our research findings indicate that HYF has a significant inhibitory effect on the expression of P53 protein under VCD stimulation. This suggests that the P53 signal may have a crucial role in the molecular mechanism of HYF. Furthermore, the anti-apoptotic effect of HYF is partially mediated through the JNK in the MAPK kinase system.

Evidence suggests that m6A modifications play a significant role in the reduction of ovarian follicular reserve. This study also explores this aspect by extracting RNA from rat ovarian tissues. The quality of the mRNA extraction was evaluated using RNA electrophoresis, as illustrated in Figure 5A. The levels of m6A-RNA modification in rat ovarian tissues were assessed utilizing ELISA, dot blot, and LC-MS/MS methodologies. The results from all three detection techniques demonstrated a consistent trend, revealing that the m6A modification in the VCD group was significantly reduced compared to the CON group (p < 0.01). Furthermore, the global m6A level in the HYF group was found to be elevated relative to the VCD group (p < 0.01, Figures 5B–D). It is noteworthy that the m6A modification of mRNA in the VCD group demonstrated a reduction of 15.36% relative to the CON group, whereas the HYF group exhibited an increase of 20.71% in m6A modification. Furthermore, to investigate the protective role of m6A involvement in HYF on ovarian granulosa cells, human ovarian granulosa cells (COV434) were employed for relevant analyses. The m6A-RNA modification level in the in vitro experiment was evaluated using Dot Blot, which revealed a significant reduction in the m6A modification level in the VCD group compared to the CON group (Figures 5E, F).

In this study, the protein expression levels of key m6A catalytic enzymes, specifically METTL3, METTL14, ALKBH5, and FTO, were analyzed to assess potential alterations in these enzymes within the ovaries of rats across three distinct groups. The expression levels of METTL3 and METTL14 exhibited no significant differences among the groups. Conversely, the expression of the demethylase ALKBH5 was significantly reduced in the VCD group (p < 0.01) and significantly elevated in the HYF group (p < 0.05). Additionally, the expression of FTO, another demethylase, was significantly increased in the VCD group and significantly decreased in the HYF group (p < 0.01, Figures 5G, H). The results indicate possible modifications in m6A-related catalytic enzymes in the ovaries of rats across various experimental groups. Western blot analysis of in vitro cells demonstrated a statistically significant difference in FTO expression between the groups, with an F-value of 8.239 and a p-value of 0.003 (Figures 5I, J). In contrast, no significant differences were observed in the expression levels of METTL14, ALKBH5, and METTL3 among the groups. This section of the study provides additional evidence that modifications in m6A are intricately linked to the fate of ovarian cells. Furthermore, it clarifies the role of the FTO/m6A axis in the regulatory effects of HYF on ovarian granulosa cells.

Previous findings suggested a significant link between FTO/m6A RNA methylation and targets like pJNK, P53, PUMA, and BAX at the protein level, both in vivo and in vitro. To explore if FTO/m6A, influenced by HYF, is related to the mitochondrial apoptosis pathway triggered by these targets, we conducted two experiments: verifying functional proteins after FTO knockdown in vitro, and examining m6A modified transcripts in ovarian samples.

As shown in Figure 6A, the expression of the FTO protein was significantly reduced in the FTO gene knockdown group compared to the blank control group. Consistent with previous reports, FTO knockdown resulted in a significant increase in the level of m6A (Figure 6B). The proliferation and apoptosis of COV434 cells were not affected by FTO knockdown (Figure 6C). Although FTO knockdown did not ultimately lead to significant changes in apoptosis phenotype, some changes occurred in certain proteins inside the cell, it was observed that the expression of JNK, pJNK, and P53 significantly decreased following FTO knockdown (p < 0.05, Figures 6D, E). Likewise, the protein levels of BAX, PUMA, and Cleaved-caspase9 were also diminished upon FTO knockdown (p < 0.01, p < 0.05, p < 0.05, respectively. Figures 6F, G).

The potential target proteins were screened and identified through both in vivo and in vitro experiments, with the results being descriptively summarized (Figure 7A). Our findings led to the identification of five potential targets: m6A-TP53, m6A-JNK, m6A-PUMA, m6A-Bax, and m6A-Bak, which are hypothesized to exert significant influence. To validate this hypothesis, we performed preliminary validation experiments on these five core candidate targets using rat ovarian tissue. These experiments specifically examined the differences between groups of total transcripts and m6A-modified transcripts.

As illustrated in Figure 7B, the mRNA expression of Tp53 demonstrated a decreasing trend in the VCD group and an increasing trend in the HYF group, which was contrary to the expression pattern observed for the P53 protein. This result is of considerable interest. Notably, the expression of m6A-Tp53 significantly increased in the VCD group and significantly decreased in the HYF group (Figure 7C, p < 0.05). This observation provides a potential explanation for the elevated levels of P53 protein in the HYF group. Moreover, Jnk expression was observed to be upregulated in the VCD group, while no statistically significant difference was detected between the HYF and VCD groups. In contrast, the expression of m6A-Jnk was downregulated in the VCD group, with no significant difference between the HYF and VCD groups (Figures 7D, E). As for Puma, Bax, and BAK, the results demonstrated minimal variation across the three groups, with inconsistent expression levels noted at both the protein and gene expression levels. Moreover, the abundance of m6A-modified transcripts did not correspond to the previously discussed proteins (Figures 7F–K). In conclusion, m6A modifications on Jnk and Tp53 (m6A-JNK, m6A-TP53) may play a contributory role in the pathogenesis of POI, with the regulation of m6A-TP53 potentially having a more significant impact on the anti-apoptotic effects of HYF.

The current study offers preliminary evidence indicating the involvement of m6A-Tp53 in the pharmacodynamic mechanism of HYF in vivo. To further investigate this, we performed immunofluorescent staining to assess P53 expression and localization in rat ovaries (Figure 7L). Our results demonstrated a reduced fluorescence intensity in the CON and HYF groups, while a significant increase was noted in the VCD group. Additionally, the fluorescence signals were primarily localized within the follicular granulosa cells, suggesting that the observed alterations in P53 and m6A-Tp53 within the ovary likely originate from these cells.

In addition, the SRAMP (http://www.cuilab.cn/sramp) and RMBase v2.0 databases (http://rna.sysu.edu.cn/rmbase) were employed to forecast the methylation locations within the mRNA sequence. The m6A modification sites pertaining to p53 were primarily located near the initiation sites of gene transcription. A total of eight methylation sites were identified in association with P53, with one discovery exhibiting a very high level of confidence, three discoveries demonstrating high confidence, four discoveries indicating moderate confidence, and five discoveries suggesting low confidence (Figure 7M; Supplementary Table 3; Supplementary Figure 2).