METTL3-mediated m6A modification of SIRT1 mRNA inhibits progression of endometriosis by cellular senescence enhancing

Thirty cases of fresh ovarian ectopic endometrial tissues (EC) and 30 matched cases of eutopic endometrial specimens (EU) were collected from patients with ovarian endometriosis. Thirty samples of normal control endometrial tissues (NC) were collected from patients with cervical intraepithelial neoplasia, excluding endometriosis. All specimens were collected during the proliferative phase of the menstrual cycle. Patients who did not receive hormonal treatments for approximately three months underwent surgery at the Second Affiliate Hospital of Harbin Medical University with confirmation by postoperative pathology. Written informed consent was obtained from all patients. This research was administrated and approved by the Second Affiliate Hospital of Harbin Medical University (Ethics number: KY2022-115).

Relative m6A levels among total RNAs were evaluated using a m6A RNA Methylation Quantification Kit (Colorimetric) (P-9005, Epigentek, USA). Total RNA was extracted using TRIzol extraction reagent (Ambion, USA), strips were treated with 200 µg/μl total RNA. According to the instructions, RNA was allowed to bind to the strip upon incubation in high binding solution at 37 °C for 90 min. As reported in a previous study, capture and detection antibodies were added to detect the m6A colorimetric levels by reading the absorbance at a wavelength of 450 nm [19]. The relative absorbance was quantified for five experimental replicates for each reaction.

m6A dot blotting was performed as previously described [20]. Total RNA was harvested as previously described with RNA quantity monitoring. RNA was denatured by heating at 95 °C for 10 min and then preserved on ice. Then, mRNAs were blotted onto an N⁺ nylon membrane (FFN10, Beyotime, China). Afterwards, the membrane was treated with ultraviolet cross-linking and blocked with 5% nonfat milk in PBST for one hour at room temperature. The membranes were incubated with a m6A-specific antibody (1:2000, 202003, Synaptic Systems) for reactivation at 4 °C overnight. After washing, the membranes were incubated with a goat anti-rabbit IgG HRP antibody (1:5000, bs-40295G-HRP, Bioss) with gentle shaking for 1 h at room temperature. After washing, the membranes were incubated with an ECL detection reagent (MA0186, Meilunbio, China) and visualized using the detection system.

Total RNA was extracted from all samples using TRIzol reagent (Ambion, USA) and reverse transcribed to cDNA by a PrimeScript™ RT Reagent Kit with gDNA Eraser (RR047A, Takara, Japan). RT‒qPCR was performed using TB Green^® Premix Ex Taq™ (RR820A, Takara, Japan). The RT‒qPCR program was as follows: 40 cycles at 37 °C for 15 min, 60 °C for 5 s, and 72 °C for 30 s. All relative mRNA expression levels were analysed using the 2−ΔΔCt method. All experiments were performed according to the instructions with three repeats.

The primers used are listed in Additional file 1: Table S1.

After rinsing in PBS, all samples were treated with RIPA buffer (R0010, Solarbio, China) for complete lysis. Then, 1% PMSF (P0100, Solarbio, China) was added, and samples were incubated on ice for 30 min. After centrifugation, we measured the protein concentration with a BCA protein assay kit (MA0082, Meilunbio, China). Protein was separated by SDS‒PAGE (P0015L, Beyotime, China) and transferred to PVDF membranes (Roche, USA). After blocking in quick block buffer for 30 min, the membranes were incubated with primary antibody. Samples were incubated in secondary goat anti-rabbit IgG HRP antibodies (1:5000, bs-40295G-HRP, Bioss) for one h at room temperature. An ECL detection reagent (MA0186, Meilunbio, China) was added to the blots, and imaging was performed on a ChemiDocMP (Bio-Rad) imager. The membranes were incubated with the following primary antibodies at 4 °C overnight: anti-METTL3 (1:2000, ab195352, Abcam), anti-YTHDF2 (1:5000, 24744-1-AP, Proteintech), anti-p53 (1:5000, 10442-1-AP, Proteintech), anti-p21 (1:1000, 10355-1-AP, Proteintech), anti-Lamin b1 (1:2000, 12987-1-AP, Proteintech), anti-FOXO3a (1:2000, 28755-1-AP, Proteintech), anti-SIRT1 (1:2000, 13161-1-AP, Proteintech) and GAPDH (1:5000, 10494-1-AP, Proteintech).

Endometrial stromal cells were isolated from ectopic, eutopic, and normal control endometrial tissues and were named ESCs, EuSCs, and NESCs, respectively, as described in a previous study [21]. Samples were washed twice with phosphate-buffered saline (PBS), minced, and coated with 1 mg/ml collagenase type IV (BS165, Biosharp, The Netherlands). Samples were incubated at 37 °C with mild shaking for two hours in a culture flask. Next, a 40-mm sieve was used to remove debris and other cells, such as endometrial epithelial cells, and the solution that passed through was centrifuged for 10 min at 1000 r/min. Cell precipitates were collected by removing the supernatant. The cells were grown in a 5% CO₂ humidified atmosphere at 37 °C in DMEM/F12 (MA0214, Meilunbio, China) supplemented with 10% foetal bovine serum (BI, Israel).

Small interfering RNAs (siRNA) explicitly targeting METTL3 (si-METTL3), SIRT1 (si-SIRT1), YTHDF1 (si-YTHDF1), YTHDF2 (si-YTHDF2), YTHDF3 (si-YTHDF3) and negative control (si-NC) were obtained from Hanbio (Shanghai, China) and used as previously described [22]. The METTL3 (oe-METTL3), YTHDF2 (oe-YTHDF2), and negative cintriol (oe-NC) were constructed and designed on the overexpression plasmid and synthesized by Hanbio (Shanghai, China). The siRNAs were transfected into the cells using RNA Fit (Hanbio, Shanghai, China) according to the protocol, and plasmid transfection was achieved using LipoFiter^TM3.0 Liposomal Transfection Reagent (Hanbio, Shanghai, China) as previously described [22]. siRNA sequences are listed in Additional file 1: Table S2.

Wild-type ESCs and ESCs transfected with the oe-METTL3 plasmid were analysed by Aksomics Company (Shanghai, China) using a human m6A epitranscriptomic microarray and mRNA microarray at Arraystar m6A single-base resolution [23]. Total RNA was treated with anti-N6-methyadenosine (m6A) for immunoprecipitation. After merging immunoprecipitated magnetic beads and recovered supernatant, Arraystar RNA was used as labelled RNA and hybridized with Arraystar Human mRNA & lncRNA Epitranscriptomic Microarray (8 × 60 K, Arraystar). Arrays were scanned with Agilent Scanner G2505C in the two-colour channel, and credit analysis was performed. Genes with |FC|≥ 2 and p value < 0.05 were chosen as differentially m6A-methylated target RNAs.

One hundred microlitres of culture medium containing 3 × 10⁴ cells was cultured in 96-well plates for 24 h. Ten microlitres of Cell Counting Kit-8 (CCK-8) reagent (MA0218-2, Meilunbio, China) was added to each well. After incubating for an additional two hours in a humidified atmosphere at 37 °C with 5% CO₂, cellular absorbance at 450 nm was assessed using a spectrophotometer. Each reaction was repeated thrice.

Briefly, 2 × 10⁵ ESCs from each well were cultured in a 24-well plate. One microlitre of EdU reagent (MA0425, Meilunbio, China) was added to 1 ml of culture medium. After incubation in a humidified atmosphere at 37 °C for three hours, the cells were immobilized with 4% PFA, permeabilized with 0.3% Triton X-100 in PBS, and stained with the click reaction solution, which is comprised of 555 azide and Hoechst 33342 stain. The results were imaged by fluorescence microscopy. Specifically, red fluorescence was detected at 555/567 nm, and blue fluorescence was detected at 346/460 nm. Three images were randomly obtained for each reaction.

A total of 1 × 10⁶ cells from each well were cultured in a 6-well plate until 80–90% confluent. Scratches were created using a sterile 200-μl pipette tip in a straight and cautious manner. Photographs were taken at 0 and 24 h, and three replicates were performed for each experimental condition.

Transwell chambers that were precoated with Matrigel (BD Biosciences, USA) were used to separate ESCs. Specifically, the upper section contained DMEM/F12 with 0.1% FBS, and the bottom chamber contained DMEM/F12 with 20% FBS. After incubation in a humidified atmosphere at 37 °C and 5% CO₂ for 24 h, cells that did not invade were removed. Cells invading the surface of the bottom chamber were fixed and stained with 0.1% crystal violet (G1062, Solarbio, China). After air drying, the invading cells were photographed using a microscope and quantified by counting five random fields.

SA-β-Gal staining was performed as previously described [24] using a Senescence β-Galactosidase Staining Kit (C0602, Beyotime, China). Briefly, cultured cells were rinsed twice with PBS and fixed at room temperature for 15 min in β-galactosidase staining fixative. Fixed cells were stained with SA-β-Gal staining solution A, B, C, and X-gal as per the instructions at 37 °C overnight, and images were captured using a microscope camera. Three experimental replicates were performed.

The RNA Immunoprecipitation Kit (P0102, Geneseed, China) was used to conduct the RIP experiment according to the manufacturer’s protocol [25]. ESCs were lysed in an IP lysis buffer, and the resulting cell lysate was separated into anti-METTL3, anti-YTHDF2, anti-IgG, and input samples. Protein A/G beads coated with identical amounts (5 g) of specific antibodies were applied to incubate cell lysates. After washing, the lysates were digested with protease and RNase inhibitors for purification. RT‒qPCR was performed to measure the target RNA levels.

RNA isolation was performed as previously described. A riboMeRIP m6A Transcriptome Profiling Kit (C11051-1, Ribobio, China) was used for the MeRIP experiment as previously described [26]. Briefly, total RNAs were sheared into approximately 100- to 150-nucleotide fragments using the RNA Fragmentation Buffer at 94 °C for 3 min; we preserved a small amount and marked them as the input RNAs. Prewashed magnetic beads A/G were precoated with m6A-specific antibody for 30 min at room temperature. Then, the beads were mixed with the fragmented RNA for immunoprecipitation. Relative gene expression was determined by qRT‒PCR as previously described.

After transfection, ESCs were treated with 5 μg/ml actinomycin D (M4881, Abmole, USA) for different time periods: 0, 3.0 h, and 6.0 h. Then, the cells were collected for RNA extraction, and RT‒PCR analysis was performed as previously described [27].

Immunohistochemistry was performed as previously described [28]. Briefly, human and mouse sections were dewaxed and rehydrated followed by antigen retrieval using Tris antigen-retrieval buffer. Then, the sections were incubated with primary and secondary antibodies. Immunofluorescence staining was performed as previously described. After incubation with primary antibodies, the cultures were incubated with Alexa Fluor 488 (1:200, ab150077, Abcam) or Cy3 (1:500, ab6939, Abcam) conjugated secondary antibodies, and images were captured with experiments repeated thrice.

For all animal studies, animals were randomly distributed and showed no size or appearance differences at the onset of the experiments. A Cre/Lox system was used to generate a tissue-specific knockout model. Endometriosis CKO mouse models were constructed as previously described [29, 30]. PRcre mice (C001035, Cyagen, China) were crossed with METTL3fl/fl homozygous mice (TOS171205 WZ1, Cyagen, China) to generate PRCre/ + METTL3 −/− biogenic mice in which METTL3 expression is abrogated in PR-expressing cells. Mice were genotyped to confirm METTL3fl/fl homozygosity together with the PgrCre/ + transgene using PgrCre/ + -specific primers (F: 5′-GCGCTAAGGATGACTCTGGTC-3′ and R: 5′-CCCTTCTCATGGAGATCT GTC-3′) and METTL3fl/fl-specific primers (F: 5′-TCCAAGAGTCTAATATCC ACCAGCAC-3′ and R: 5′-TGATCAGCAAATGATGGTCCCAG-3′). Mice were randomly divided into donor and recipient groups in both the PRCre/ + METTL3 −/− group (CKO-METTL3 group) and METTL3f/f group (Control group). Donor mice were injected with oestradiol (E2) (1 μg/ml) every three days at six weeks of age for two weeks. After 2 weeks, the uterine horns of the donor mice were surgically removed and dissected in saline to completely expose the endometrial surface, transplanted to the intraperitoneal cavity of the recipient mice to generate endometriosis. Mice were sacrificed on Days 3, 5, and 7 after receiving an intraperitoneal injection, and endometrial lesions were recorded through observation. Lesion volumes were computed using the prolate ellipsoid geometric model: (length × width)²/2. All animal experiments were approved by the Ethics Committee of the Second Affiliate Hospital of Harbin Medical University (ethics number: SYDW2022-080).

Female BALB/c nude mice purchased from the Charles River Laboratory were maintained in SPF conditions for one week before use. The experiments were approved by the Ethics Committee of the Second Affiliate Hospital of Harbin Medical University. For in vivo studies, the shMETTL3 sequence was subcloned into adeno-associated virus (AAV) construct 9 (AAV9). Recombinant AAV9 was manufactured by GenePharma (Shanghai, China). The ectopic endometrium was obtained from a 28-year-old woman who had undergone an operation for ovarian endometriosis. The tissue sample was washed twice with PBS and cut into three 3- to 5-mm pieces, and suspended by PBS. Briefly, 200 μl of suspension was injected under the bilateral axilla in each mouse as previously described [31]. The animals were administered intraperitoneal injections of 30 mg/kg 17-oestradiol every three days after the endometrial injections. Xenografts were generated under bilateral axillae after 14 days with tumour volumes of approximately 10 mm³.

Subsequently, nude mice with lesions were randomly divided into two groups. In the study, seven mice were administered intratumor injections of METTL3 shRNA AAV, whereas the other seven mice received control shRNA AAV. Mice were given injections with 5 μL of AAV daily for 10 days [32]. Every five days, the tumour was measured using a slide calliper, and the volume was determined using the method previously described. Twenty days after virus injection, all mice were anaesthetized and euthanized by cervical dislocation, and tumours were removed and measured.

All the data were analysed as the mean ± SD, and two-tailed Student’s t test and one-way ANOVA by SPSS (v.16.0) were performed using GraphPad software (v.9.0.0) to determine statistical significance. All experiments were performed at least thrice. Here, ns indicates not significant (p ≥ 0.5); *p < 0.05 denotes a moderate statistically significant result; **p < 0.01 denotes a statistically significant result; and ***p < 0.001 indicates a highly statistically significant result.

To elucidate the relevance of m6A modification in EMs, the degree of m6A modification was evaluated in samples from the ectopic (EC) group (n = 30), eutopic (EU) group (n = 30), and normal control (NM) group (n = 30). The results indicated that m6A modification levels were decreased in EC samples compared to EU and NM endometrium samples (Fig. 1A, B). Additionally, we detected familiar m6A regulators, including writers (METTL3, METTL14, and WTAP), erasers (FTO and ALKBH5), and readers (YTHDF1 and YTHDF2), in EC, EU, and NM endometrial samples. RT‒qPCR analysis results showed that METTL3 mRNA levels were the most significantly reduced in the EC group. However, no statistically significant alterations in METTL14, YTHDF1, and ALKBH5 were noted in the ectopic group compared with the eutopic or normal group (Fig. 1C). Additionally, METTL3 protein levels were significantly decreased in EC samples compared to the EU and NM groups as assessed by western blotting assay (Fig. 1D). To properly explore METTL3 activation in EMs, we performed immunohistochemistry (IHC) staining to examine METTL3 expression in ectopic, eutopic, and normal endometrial specimens (Fig. 1E). We found the same expression trend among the three groups in our study. We subsequently investigated whether METTL3 expression correlated with the clinicopathological characteristics of EMs. METTL3 expression was independent of dysmenorrhea and age. Endometriosis patients at stages III + IV (according to the revised American Fertility Society classification, r-AFS) or without deeply infiltrating endometriosis and cyst size ≥ 3 cm in diameter showed reduced METTL3 expression (Table 1). Furthermore, we assessed METTL3 expression in the following primary cell groups: ESCs (ectopic endometrial stromal cells), EuSCs (eutopic endometrial stromal cells), and NESCs (normal control endometrial stromal cells). METTL3 mRNA and protein levels were reduced in ESCs compared with EuSCs and NESCs as demonstrated qPCR and western blotting analyses (Fig. 1F, G). Taken together, these studies revealed that METTL3 expression levels and m6A RNA modifications were reduced in EMs.

Based on the above findings, METTL3 was downregulated in EMs samples and ESCs. We hypothesized that METTL3 serves as a disease inhibitor in EMs. To explore whether METTL3 regulates RNA m6A modification in ESCs, si-METTL3 transfection efficiency was measured, as shown in Fig. 2A. Then, total RNA m6A modification levels were measured, and the results demonstrate that METTL3 deletion decreased m6A modification levels (Fig. 2B). To determine the function of METTL3 in EMs, METTL3 was knocked down and overexpressed in ESCs. The role of METTL3 in cell viability, invasion, proliferation, migration as well as and the level of cellular senescence was assessed. As expected, METTL3 depletion markedly enhanced the viability, proliferation, and migration ability of ESCs (Fig. 2C–F). In addition, METTL3 expression increased the modification level (Fig. 2G, H) and significantly decreased the viability, proliferation, and migration ability of ESCs (Fig. 2I–L). The results of SA-β-gal staining showed that METTL3 knock down effectively rescued METTL3-induced promotion of cellular senescence in ESCs. The senescent cells positivity rate was significantly augmented and the cell size was enlarged after METTL3 overexpression (Fig. 2M). To further detect the effect of METTL3 expression on ESCs senescence, p53, p21, and Lamin b1 were chosen as markers in our research (Fig. 2N). The expression of the p53 and p21 proteins was significantly enhanced with METTL3 overexpression, while the expression of Lamin b1 protein decreased, indicating a trend towards cellular senescence. In contrast, ESCs with METTL3 knocked down showed the opposite trend.

To confirm that METTL3 drives the formation of EMs in vivo, nude mouse models with subcutaneous xenografts were employed. In these experiments, Day 0 was two weeks after the establishment of subcutaneous xenografts.

After AAV injection for five days, we recorded the volumes of the xenografts every 5 days (Days 5, 10, 15, and 20). As shown, AAV-shMETTL3 intervention robustly enhanced the expansion of EMs xenografts in nude mice (Fig. 2O, P). All EMs xenografts were delicately extracted and weighed on “Day 20”. Xenografts from the AAV-shMETTL3 group were significantly larger and heavier than those from the AAV-shNC group (Fig. 2Q, R). Whether AAV-shMETTL3 caused a similar cellular senescence signalling shift in vivo was investigated. As shown by immunohistochemistry, a dramatic increase in the cellular senescence inhibitor (Lamin b1) and an decrease in cellular senescence promoters (p53 and p21) were observed in EMs xenograft tissues injected with AAV-shMETTL3 (Fig. 2S).

To further investigate the function of METTL3 in EMs, matched oe-NC ESCs and oe-METTL3 ESCs were treated, an RNA-seq (Fig. 3A) and human m6A epitranscriptomic microarrays and mRNA microarrays (Fig. 3B) were used to explore the potentially methylated target mRNAs of METTL3. We focused on pinpointing METTL3 targets, which were identified as abnormally expressed transcripts in RNA-seq and differentially methylated transcripts in m6A epitranscriptomic microarrays. As shown in the Venn diagram, potential downstream mRNAs of METTL3 were obtained, and SIRT1 (Fig. 3C), a vital FoxO signal pathway that induces transcription factors involved in cellular senescence, autophagy, and pyroptosis processes, was identified as an important gene in this analysis. KEGG pathway analysis revealed that these differentially expressed genes were significantly enriched in the cell cycle, ubiquitin-mediated proteolysis, proteoglycans in cancer, autophagy, protein processing in the endoplasmic reticulum, FoxO signalling pathway, adherens junction, and cellular senescence (Fig. 3D). Notably, SIRT1 was highly enriched in the FoxO signalling pathway and cellular senescence, which ranked as the target enriched signalling pathways. The most prominent m6A motif, GGAC, was enriched in the m6A peaks (Fig. 3E). In particular, the m6A peaks were enriched proximal to the stop codon and 3′ untranslated region (3′ UTR) compared with peaks in the 5′ UTR and CDS region (Fig. 3F). RNA-seq data indicated SIRT1 downregulation after oe-METTL3 transfection in ESCs (Fig. 3G). m6A epitranscriptomic microarray data also showed enrichment of METTL3 methylation and the frequency of m6A after oe-METTL3 transfection in ESCs (Fig. 3H). We consider SIRT1, which was selected as a potential target, to be downstream of METTL3.

To assess the regulatory roles of METTL3 and SIRT1, first, total RNA was extracted, and SIRT1 levels were determined. SIRT1 expression levels were assessed by RT‒qPCR, which showed increased expression in ectopic tissues (Fig. 3I). As shown in Fig. 3J, consistent with previous studies, ectopic endometrial lesions showed reduced METTL3 expression and increased SIRT1 expression. In contrast, normal endometrial tissues with increased METTL3 expression exhibited lower SIRT1 levels as demonstrated by immunofluorescence (IF) staining.

To confirm METTL3-mediated m6A modification of SIRT1, we assessed SIRT1 mRNA levels in METTL3-overexpressing and METTL3-knockdown ESCs. METTL3 overexpression reduced SIRT1 transcript levels in ESCs (Fig. 3K), suggesting the negative regulation of SIRT1 by METTL3. Then, we investigated the role of m6A modification in the regulation of SIRT1 by METTL3. Quantitative RIP assays indicated that SIRT1 mRNA interacts with METTL3 in ESCs. Furthermore, SIRT1 enrichment was markedly reduced in ESCs upon METTL3 knockdown (Fig. 3L). To verify that m6A modification of SIRT1 was regulated by METTL3, we performed MeRIP-qPCR experiments in ESCs with METTL3 knockdown. METTL3 knockdown attenuated the enrichment in SIRT1 levels (Fig. 3M).

To further verify that METTL3-mediated regulation of EMs was dependent on SIRT1, we constructed an endometriosis model of METTL3 deletion using a CKO mouse endometriosis model, and the entire workflow is shown in Fig. 3N. Donor mouse endometrium was grafted into the recipient peritoneal cavity (labelled Day 0). Then, we sacrificed the mice on Days 3, 5, and 7 and observed lesions in the peritoneal cavity or visceral peritoneum. The EMs lesion showed apparent enlargement in the CKO-METTL3 group compared to the control group on Days 3, 5 and 7 (Fig. 3O). Furthermore, all grafting lesions were carefully removed on Day 7. We assessed METTL3 expression by RT‒qPCR. A reduction in METTL3 expression in the grafting lesions was noted in the CKO-METTL3 group (Fig. 3P), which is consistent with the aforementioned observation (Fig. 3Q). As shown by immunohistochemistry (Fig. 3R), SIRT1 expression was dramatically increased in grafted tissues from the CKO-METTL3 group, which confirms that the negative coordination of METTL3 and SIRT1 promotes the progression of EMs.

As depicted in Fig. 4, SIRT1 knock down considerably repressed cell viability, proliferation, and migration, promoting cellular senescence as expected. These results showed that SIRT1 is a potential promoter in EMs progression. The above results revealed that METTL3 and SIRT1 perform beneficial reverse functions in ESCs. Moreover, we showed that reduced SIRT1 expression exacerbated the repression of METTL3-induced limitations in viability, proliferation, and migration and promoted cellular senescence escape in ESCs, providing further evidence of the mutual interaction. In summary, METTL3 acts as an EMs inhibitor by modulating SIRT1 expression in ESCs.

m6A modification depends on the functional roles of reader proteins in biological processes. To identify the readers of SIRT1, we measured SIRT1 mRNA levels after reducing YTHDF1, YTHDF2, and YTHDF3 expression. si-YTHDF2 showed the most significantly enhanced ability (Fig. 5A). A quantitative RIP assay was performed to assess the levels of SIRT1 mRNA bound to YTHDF2 in ESCs. SIRT1 enrichment was drastically reduced in ESCs upon the elimination of YTHDF2 expression (Fig. 5B). Then, we found that YTHDF2 knockdown substantially increased SIRT1 expression, and overexpression of YTHDF2 showed a decreased effect (Fig. 5C). Based on previous research, we considered YTHDF2 to act as a m6A reader in the EMs model. YTHDF2 (YTH domain family 2), which is recognized as a m6A reader protein, affects the stability of transcripts. We tested whether m6A modification affects the mRNA stability of SIRT1. We treated ESCs with the transcription inhibitor actinomycin D (Act D), and the half-lives of SIRT1 transcripts were significantly decreased in oe-METTL3 ESCs (Fig. 5D) and oe-YTHDF2 ESCs (Fig. 5E) compared with the control group. Further to confirm the involvement of m6A reader YTHDF2 in the stability regulation of SIRT1 transcripts, we modulated METTL3 and YTHDF2 expression simultaneously in ESCs and examined the SIRT1 mRNA decay rate after Act D treated (Fig. 5F), which given evidence that YTHDF2 influenced the stability of SIRT1 transcripts through m6A modification independently.

However, silencing YTHDF2 led to increased SIRT1 expression, and this effect could be partially repressed by METTL3 overexpression (Fig. 5G). In contrast, the reduction in m6A modification levels induced by the reduction of METTL3 expression led to increased expression of SIRT1, and this effect cannot be remedied in ESCs by upregulation of YTHDF2 expression (Fig. 5H). Our findings indicate that METTL3-mediated m6A modification maintains SIRT1 expression in a m6A-YTHDF2-dependent manner by regulating SIRT1 mRNA stability.

Previous studies have shown that the SIRT1-FoxO3a signalling pathway is critical in regulating cellular senescence. In the above m6A microarray, the KEGG results showed that an activated FoxO signalling pathway with SIRT1 enrichment may be downstream of METTL3 modification in EMs. Our study demonstrated that SIRT1 is negatively regulated by METTL3 expression. To further explore whether the SIRT1/FOXO3a pathway played a role in METTL3-induced ESC senescence escape in EMs, we pretreated cells with si-METTL3 and oe-METTL3. Western blotting analysis showed that silencing METTL3 expression indeed restored the reductions in SIRT1 and FOXO3a protein levels (Fig. 5I). In addition, a si-SIRT1 ESC model was employed to investigate the positive role of SIRT1 in FOXO3a expression (Fig. 5J). These results suggest that METTL3 reduces m6A modification levels and cellular senescence in ESCs by inactivating the SIRT1/FOXO3a pathway. Finally, we propose a model for the critical link between METTL3-m6A-SIRT1-YTHDF2 in EMs progression (Fig. 6).