Ribosome dysregulation and intervention in age-related infertility.

J.L. designed the experiments, conducted the major experiments, analyzed the data, and prepared the manuscript. C.L. conducted important mouse experiments and revised the manuscript. H.C. and H.Z. collected the clinical materials and revised the manuscript. Linlin Liu, X.Y., G.F., and Y.W. conducted part of the experiments and discussed the experiments and data analysis. Q.L. provided clinical materials. T.Y. provided the clinical material. X.B. and W.S. provided clinical materials and relevant guidance for clinical diagnosis, discussed the experiments, and revised the manuscript. H.W., P.Z., and X.W. conducted the clinical translational experiments on rapamycin and revised the manuscript. D.L.K. advised the project and revised the manuscript. Lin Liu conceived the project, designed the experiments, and revised the manuscript.

We obtained 71 oocyte and 114 CC samples from female donors aged 23–48 years who consented to the experiments and multi-omics analysis. With increasing age, the antral follicle count (AFC), the number of follicles >12 mm, the number of oocytes retrieved, and E2 levels decreased, whereas the levels of FSH (follicle-stimulating hormone) and FSH/LH (luteinizing hormone) increased ( Figure S1 A), which was consistent with the decreasing ovarian reserve with increasing age. 2 , 26 , 27 cDNA libraries were constructed, and RNA was sequenced from single oocyte and CC samples via a previously established method 28 ( Figures S1 B–1E; Table S1 ). We identified 19,098 genes expressed in oocytes and 28,761 genes expressed in CCs. The RNA sequencing (RNA-seq) data of the oocyte and CC samples were normalized and subjected to Pearson’s correlation analysis. We attempted various age-correlation coefficient cutoffs, based on the principle described 29 , 30 (also see method details section), and showed that the cutoff of |r| ≥ 0.4 (for oocytes) or |r| ≥ 0.5 (for CCs) produced significant differences (adjusted p value < 0.05) between the age and gene expression levels ( Tables S2 and S3 ). Notably, a distinct shift in oocyte gene expression profiles emerged around the age of 34, and this transition involved 517 downregulated and 253 upregulated genes in older compared to younger oocytes ( Figure 1 A). Analysis of Kyoto Encyclopedia of Genes and Genomes pathways and Gene Ontology (GO) for genes upregulated with age revealed several key pathways, such as Parkinson’s disease, Huntington’s disease, oxidative phosphorylation, and ribosome-associated genes (e.g., RPS15 , RPL36 , and RPS5 ) ( Figures 1 A and 1B). Additionally, we took advantage of an independent transcriptome dataset of a large cohort of oocytes from women of varying reproductive age 15 and validated the noticeable transition of gene expression profiles around the age of 34 by the age-correlation coefficient cutoff of |r| ≥ 0.4 (adjusted p value < 0.05) ( Figure S1 F). Moreover, the independent dataset also revealed the enrichment of ribosome signaling with age and upregulated expression of ribosome-related genes in older oocytes ( Figures S1 F and S1G), consistent with our findings. The robust replication across the datasets strengthens the association between ribosome dysregulation and oocyte aging. Figure 1 Age-related transcriptome changes in single oocytes (A) Heatmap showing the expression signatures of specifically expressed genes in oocytes with increasing age, and aging-specific genes were identified based on age-gene expression correlation, with an absolute correlation coefficient cutoff of |r| ≥ 0.4 and statistical significance (adjusted p value < 0.05). The value for each gene is the row-scaled Z score, and the color key from blue to red indicates the relative gene expression level from low to high, respectively. The index number and age of each sample are marked at the bottom of the heatmap; n = 48 for the RNA-seq data. Representative enriched terms of aging-related genes are marked on the right. (B) Pearson correlations between age and the expression levels of genes related to ribosomes in oocytes. The size of the circles represents the expression levels of genes. The gene expression level was quantified as transcripts per million (TPM). (C) Gene set enrichment analysis (GSEA) indicating that upregulated genes in oocytes >34 years (including 34) were highly enriched in ribosome-related gene sets. Red, upregulated genes in oocytes >34 years; NES, normalized enrichment score; FDR, false discovery rate. (D) Violin plot displaying the expression of ribosomal subunit-related genes in oocytes from donors >34 years old compared with those <34 years old; genes were obtained from gene sets including cytosolic large and small ribosomal subunits via GSEA. (E) Heatmap showing the expression levels of key genes related to ribosomes in oocytes from donors 34 (including 34) years old. The bar plot shows −log10 ( p value), computed in DESeq2 via the Wald test; the black dotted line represents −log10 (0.05). (F) Gene sets related to the attachment of spindle microtubules to kinetochores and to kinetochores themselves were enriched with downregulated genes in oocytes >34 years old. (G) Correlations between female age and the expression (TPM) of kinetochore-related genes, such as CENPU and CENPQ , and the correlation coefficient were computed via the Pearson method. Statistical significance is defined as p < 0.05. For (D), violin plot shows the kernel density of the data with box in middle representing the median, the 25% and 75% interquartile range (IQR), and the ±1.5 IQR; p value was calculated by two-tailed unpaired t test; (G) Pearson correlation. ∗∗∗∗ p < 0.0001. Age-related transcriptome changes in single oocytes (A) Heatmap showing the expression signatures of specifically expressed genes in oocytes with increasing age, and aging-specific genes were identified based on age-gene expression correlation, with an absolute correlation coefficient cutoff of |r| ≥ 0.4 and statistical significance (adjusted p value < 0.05). The value for each gene is the row-scaled Z score, and the color key from blue to red indicates the relative gene expression level from low to high, respectively. The index number and age of each sample are marked at the bottom of the heatmap; n = 48 for the RNA-seq data. Representative enriched terms of aging-related genes are marked on the right. (B) Pearson correlations between age and the expression levels of genes related to ribosomes in oocytes. The size of the circles represents the expression levels of genes. The gene expression level was quantified as transcripts per million (TPM). (C) Gene set enrichment analysis (GSEA) indicating that upregulated genes in oocytes >34 years (including 34) were highly enriched in ribosome-related gene sets. Red, upregulated genes in oocytes >34 years; NES, normalized enrichment score; FDR, false discovery rate. (D) Violin plot displaying the expression of ribosomal subunit-related genes in oocytes from donors >34 years old compared with those <34 years old; genes were obtained from gene sets including cytosolic large and small ribosomal subunits via GSEA. (E) Heatmap showing the expression levels of key genes related to ribosomes in oocytes from donors 34 (including 34) years old. The bar plot shows −log10 ( p value), computed in DESeq2 via the Wald test; the black dotted line represents −log10 (0.05). (F) Gene sets related to the attachment of spindle microtubules to kinetochores and to kinetochores themselves were enriched with downregulated genes in oocytes >34 years old. (G) Correlations between female age and the expression (TPM) of kinetochore-related genes, such as CENPU and CENPQ , and the correlation coefficient were computed via the Pearson method. Statistical significance is defined as p < 0.05. For (D), violin plot shows the kernel density of the data with box in middle representing the median, the 25% and 75% interquartile range (IQR), and the ±1.5 IQR; p value was calculated by two-tailed unpaired t test; (G) Pearson correlation. ∗∗∗∗ p < 0.0001. Gene set enrichment analysis (GSEA) further demonstrated the upregulation of ribosome genes in oocytes from women older than 34 years ( Figures 1 C–1E). The downregulated genes in older oocytes beyond the age of 34 were enriched for important terms related to meiosis, including centromere kinetochores ( CENPs ), microtubule-organizing centers, the actin cytoskeleton, sister chromatid cohesion, as well as DNA repair ( Figures 1 A and 1F). Meiotic deficiency with increasing age has been reported previously, 31 , 32 , 33 validating the single-cell RNA-seq analysis reported here. Correlation analysis revealed that key genes, such as CENPU and CENPQ , were negatively correlated with age ( Figure 1 G). F-Actin nucleates on chromosomes and coordinates their capture by microtubules in oocyte meiosis. 34 CENPU constitutes the centromere and facilitates the kinetochore-microtubule attachment necessary for the correct separation of chromosomes. 35 , 36 Prematurely separated chromosomes, associated with loss of cohesion with increasing maternal age, are prone to incorrect segregation during anaphase and contribute to embryo aneuploidy. 4 , 37 Aging-specific transcriptional changes in CCs are more pronounced than those in oocytes, including 2,181 genes whose expression is downregulated and 396 genes whose expression is upregulated with maternal age ( Figure 2 A). The signaling pathways enriched with the downregulated genes included the lysosome (e.g., CLTA , HEXB , and GLB1 ) ( Figure 2 B), the proteasome, and metabolic pathways (e.g., purine metabolism, pyrimidine metabolism, and fatty acid metabolism). The upregulated genes were also enriched for the ribosome, in addition to the glycolysis/gluconeogenesis and transforming growth factor β signaling pathways ( Figure 2 A). Moreover, GSEA revealed that ribosome gene sets were upregulated and lysosome gene sets were downregulated with age ( Figures 2 C and 2D). Additionally, genes involved in the response to oxidative stress were downregulated with age ( Figure 2 E). The expression of GPX4 , an antioxidase that strongly inhibits lipid oxidation, was decreased in aging primate ovaries 14 and noticeably decreased with maternal age in human CCs ( Figure 2 E). Figure 2 Age-induced changes in the transcriptome of CCs (A) Heatmap showing the expression signatures of genes specifically expressed in CCs with age, and aging-specific genes were identified based on age-gene expression correlation, with an absolute correlation coefficient cutoff of |r| ≥ 0.5 and statistical significance (adjusted p value < 0.05). n = 30 RNA-seq data points. Representative enriched terms of aging-related genes are marked on the right. (B) Pearson correlations between age and the expression levels of genes related to lysosomes in CCs. (C and D) GSEA highlighting that upregulated and downregulated genes in CCs >34 years old were enriched in gene sets related to ribosomes and lysosomes, respectively. (E) Response to oxidative stress was significantly enriched in downregulated genes in CCs from individuals >34 years old. Key genes are shown in a bar plot, and the color key indicates the correlation coefficient, which was calculated via the Pearson method on the basis of female age and gene expression levels. The vertical axis represents −log10 ( p value); the black dotted line indicates −log10 (0.05). (F and G) Integration of translation- and ribosome-related GO enrichment and genes by Cytoscape in female oocytes (F) and CCs (G). Rectangles represent GO enrichment, and ovals represent genes associated with GO enrichment. The color key from yellow to orange indicates the age coefficient from low to high, respectively. The age coefficient was calculated via Pearson correlation between the donor age and gene expression level (TPM) via “cor” in R software. The width of the edge indicates betweenness, reflecting the amount of control that this node exerts over the interactions of other nodes in the network. Age-induced changes in the transcriptome of CCs (A) Heatmap showing the expression signatures of genes specifically expressed in CCs with age, and aging-specific genes were identified based on age-gene expression correlation, with an absolute correlation coefficient cutoff of |r| ≥ 0.5 and statistical significance (adjusted p value < 0.05). n = 30 RNA-seq data points. Representative enriched terms of aging-related genes are marked on the right. (B) Pearson correlations between age and the expression levels of genes related to lysosomes in CCs. (C and D) GSEA highlighting that upregulated and downregulated genes in CCs >34 years old were enriched in gene sets related to ribosomes and lysosomes, respectively. (E) Response to oxidative stress was significantly enriched in downregulated genes in CCs from individuals >34 years old. Key genes are shown in a bar plot, and the color key indicates the correlation coefficient, which was calculated via the Pearson method on the basis of female age and gene expression levels. The vertical axis represents −log10 ( p value); the black dotted line indicates −log10 (0.05). (F and G) Integration of translation- and ribosome-related GO enrichment and genes by Cytoscape in female oocytes (F) and CCs (G). Rectangles represent GO enrichment, and ovals represent genes associated with GO enrichment. The color key from yellow to orange indicates the age coefficient from low to high, respectively. The age coefficient was calculated via Pearson correlation between the donor age and gene expression level (TPM) via “cor” in R software. The width of the edge indicates betweenness, reflecting the amount of control that this node exerts over the interactions of other nodes in the network. Integration of translation and ribosome-related GO enrichment and genes by Cytoscape revealed that in both oocytes and CCs, genes related to ribosomes and translation increased in expression with age ( Figures 2 F and 2G). Gene sets for both the cytosolic large and small ribosomal subunits were significantly enriched for upregulated genes in aging oocytes and CCs ( Figures 1 C and 2 D). Additionally, oocytes expressed genes encoding mitochondrial ribosomal proteins, including MRPL17 , MRPL53 , MRPS25 , MRPL34 , MRPL43 , MRPS24 , and MRPL49 , which are involved in the composition of mitochondrial ribosomes and protein synthesis within the mitochondrion ( Figure 2 F). Together, oocytes and the surrounding CCs display distinct transcriptome changes around the mid-thirties, when age-related fertility loss accelerates. Increased expression of ribosome genes has not been reported previously in association with oocyte or ovarian aging. Furthermore, when we compared RNA-seq data between relatively older and younger women under the age of 34, the vast majority of genes related to cytosolic large and small ribosomal subunits presented increased expression in both CCs and oocytes from older women ( Figures 1 D and 3 A). Nucleoli are the nuclear structures responsible for ribosome biogenesis, and an enlarged nuclear area and a reduced number of nucleoli are associated with premature aging. 38 We evaluated the nucleolar area as defined by nucleolin staining in CCs. An increase in total nucleolar area ( Figures 3 B and 3C) and a decrease in nucleolar number were observed in aged CCs ( Figure 3 D). Additionally, we quantified the levels of 28S and 18S ribosomal RNAs (rRNAs). The content of 18S and 28S rRNAs increased with age in CCs ( Figure 3 E). Consistently, oocytes from reproductively aged mice displayed increased ribosome numbers associated with altered nucleolar architecture. 39 Taken together, these results suggest that ribosome components and associated translation could increase in CCs with age. Figure 3 Ribosomes and lysosomes in aging human CCs (A) Violin plot showing the expression of ribosomal subunit-related genes in CC samples from donors >34 years (including 34) compared with those from donors <34 years old. Genes were obtained from gene sets, including large and small cytosolic ribosomal subunits, via GSEA. (B) Representative immunofluorescence image of the nucleolar protein nucleolin (red) in CC samples of different ages from four independent experiments. Scale bar, 10 μm. (C) Comparison of nucleolar area between the 34-year-old groups. The total area of nucleolar cross-sections per CC was determined by boundaries of nucleolin immunofluorescence. n = 600 cells, and each CC sample contained 100 cells. A total of 6 CC samples from 6 women in each age group were collected (5 independent experiments). (D) Comparison of the percentages of the number of nucleoli between the 34-year-old groups. The calculated data are shown as percentages (%), n = 600 cells. The color depth of the histogram represents the number of nucleoli. (E) Abundance of 18S and 28S rRNAs determined by qPCR in CC samples from donors of varying ages ( n = 22; 24–44 years), normalized to GAPDH mRNA. The coefficient was calculated via Pearson’s correlation, and the p values for 18S and 28S rRNA abundance changes with age are shown. (F) SA-β-gal staining of female CCs (CCs-56, 26 years and CCs-60, 37 years) treated with or without rapamycin (−1: 0.25 μΜ and −2: 0.50 μΜ) or CHX (cycloheximide, 0.30 μΜ) for 3 days. Scale bar, 100 μm. n = 3 (for each independent experiment, 20 visual fields were randomly selected for statistical analysis). (G) Violin plot showing the expression of lysosome-related genes (17 genes) in CC samples from donors >34 years compared with those <34 years. Lysosome-related genes were selected from lysosomes via Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis and met the criterion of an age coefficient of at least −0.5. (H) Representative immunofluorescence images of CC samples from donors of varying ages stained with LysoTracker (red, acidic organelles; lysosomes) and Hoechst (blue, nuclei). The cells were treated with the lysosome inhibitor bafilomycin A (BafA, 700 nM) for 6 h as a control. Scale bar, 10 μm. Right: the integrated intensity of LysoTracker fluorescence was estimated by ImageJ; n = 480; each sample contained 80 cells, with a total of 6 independent samples per group (5 independent experiments). (I) Representative immunofluorescence images of CC samples from donors of varying ages stained with Proteostat (red, protein aggregates) and Hoechst (blue, nuclei). The cells were treated with the proteasome inhibitor MG132 (10 μM) for 12 h as a control. Scale bar, 10 μm. Right: integrated intensity of Proteostat fluorescence estimated with ImageJ; n = 480; each sample contained 80 cells, with a total of 6 independent samples per group (5 independent experiments). (C, F, H, and I) Data are represented as mean ± SEM. (C, H, and I) Mann-Whitney test; (F) ANOVA tests for multiple comparisons; (A and G) violin plot shows the kernel density of the data with box in middle representing the median, the 25% and 75% interquartile range (IQR), and the ±1.5 IQR; p value was calculated by two-tailed unpaired t test; (D) chi-squared test; (E) Pearson correlation. ∗ p < 0.05, ∗∗∗ p < 0.001, and ∗∗∗∗ p 34 years (including 34) compared with those from donors <34 years old. Genes were obtained from gene sets, including large and small cytosolic ribosomal subunits, via GSEA. (B) Representative immunofluorescence image of the nucleolar protein nucleolin (red) in CC samples of different ages from four independent experiments. Scale bar, 10 μm. (C) Comparison of nucleolar area between the 34-year-old groups. The total area of nucleolar cross-sections per CC was determined by boundaries of nucleolin immunofluorescence. n = 600 cells, and each CC sample contained 100 cells. A total of 6 CC samples from 6 women in each age group were collected (5 independent experiments). (D) Comparison of the percentages of the number of nucleoli between the 34-year-old groups. The calculated data are shown as percentages (%), n = 600 cells. The color depth of the histogram represents the number of nucleoli. (E) Abundance of 18S and 28S rRNAs determined by qPCR in CC samples from donors of varying ages ( n = 22; 24–44 years), normalized to GAPDH mRNA. The coefficient was calculated via Pearson’s correlation, and the p values for 18S and 28S rRNA abundance changes with age are shown. (F) SA-β-gal staining of female CCs (CCs-56, 26 years and CCs-60, 37 years) treated with or without rapamycin (−1: 0.25 μΜ and −2: 0.50 μΜ) or CHX (cycloheximide, 0.30 μΜ) for 3 days. Scale bar, 100 μm. n = 3 (for each independent experiment, 20 visual fields were randomly selected for statistical analysis). (G) Violin plot showing the expression of lysosome-related genes (17 genes) in CC samples from donors >34 years compared with those <34 years. Lysosome-related genes were selected from lysosomes via Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis and met the criterion of an age coefficient of at least −0.5. (H) Representative immunofluorescence images of CC samples from donors of varying ages stained with LysoTracker (red, acidic organelles; lysosomes) and Hoechst (blue, nuclei). The cells were treated with the lysosome inhibitor bafilomycin A (BafA, 700 nM) for 6 h as a control. Scale bar, 10 μm. Right: the integrated intensity of LysoTracker fluorescence was estimated by ImageJ; n = 480; each sample contained 80 cells, with a total of 6 independent samples per group (5 independent experiments). (I) Representative immunofluorescence images of CC samples from donors of varying ages stained with Proteostat (red, protein aggregates) and Hoechst (blue, nuclei). The cells were treated with the proteasome inhibitor MG132 (10 μM) for 12 h as a control. Scale bar, 10 μm. Right: integrated intensity of Proteostat fluorescence estimated with ImageJ; n = 480; each sample contained 80 cells, with a total of 6 independent samples per group (5 independent experiments). (C, F, H, and I) Data are represented as mean ± SEM. (C, H, and I) Mann-Whitney test; (F) ANOVA tests for multiple comparisons; (A and G) violin plot shows the kernel density of the data with box in middle representing the median, the 25% and 75% interquartile range (IQR), and the ±1.5 IQR; p value was calculated by two-tailed unpaired t test; (D) chi-squared test; (E) Pearson correlation. ∗ p < 0.05, ∗∗∗ p < 0.001, and ∗∗∗∗ p < 0.0001. Mechanistic target of rapamycin (mTOR) tightly regulates protein synthesis and autophagy by phosphorylating substrates. 40 The aging of CCs was accompanied by an increase in p-mTOR protein and a decrease in the autophagy-related protein LC3-II ( Figure S2 A). L-homopropargylglycine incorporation for protein synthesis assays further confirmed that protein synthesis increased significantly in aging CCs ( Figure S2 B). To further verify ribosome and protein synthesis dysregulation, rapamycin was utilized to inhibit ribosome biogenesis and protein synthesis via mTOR. 41 Additionally, cycloheximide is commonly used to inhibit protein synthesis. Notably, 0.5 μM rapamycin effectively attenuated the increase in protein synthesis induced by aging CCs ( Figure S2 C). Moreover, senescence-associated β-galactosidase activity was suppressed in aged CCs after rapamycin or cycloheximide treatment ( Figure 3 F). Additionally, RNA-seq revealed decreased expression of lysosome- and proteasome-related genes with age ( Figures 2 A and 3 G). We used LysoTracker to label lysosomes and detected that the activity of lysosomes indeed dramatically decreased with age ( Figures 3 H and S2 D). Aberrant expression of ribosomal and lysosome-related genes with age may disrupt protein homeostasis. To examine proteostasis, we performed fluorescence analysis with Proteostat dye in CCs of different ages. Proteostat becomes highly fluorescent upon binding to the amyloid-type β sheet tertiary structure of protein aggregates. 42 Compared with young CCs, aging CCs displayed increased fluorescence of Proteostat ( Figures 3 I and S2 E). The protein aggregates were only slightly increased in oocytes with increasing age ( Figure S2 F). The accumulation of protein aggregates in aging CCs may imply increased protein synthesis and/or decreased protein degradation, which could be linked to the upregulation of ribosome-related genes and the downregulation of lysosome-related genes, disrupting protein homeostasis. The epigenetic clock resulting from DNA methylation underlies somatic aging. 20 , 21 , 43 , 44 To understand the molecular mechanisms underlying the changes in the transcriptome with age, we performed low-input post-bisulfite adaptor tagging DNA methylation sequencing to delineate the DNA methylome of oocytes or CCs. A cohort of females was recruited ( Table S1 ) and divided into old (>34 years) and young (<34 years) groups on the basis of transcriptome analysis. Two independent biological replicates per age group were assessed, and robust data were obtained for all the oocyte samples, with reproducible replicates ( Figure S3 A). A total of 112,510,677 reads per group and approximately 11,259,424 CpGs in the human genome were covered in each sample ( Table S4 ). In general, aging led to a slight increase in DNA methylation in oocytes ( Figure S3 B), and old oocytes contained a slightly greater proportion of highly methylated sequences than did young oocytes ( Figures S3 C–S3G). We observed a peak of differentially methylated regions (DMRs) located immediately upstream and downstream of transcription start sites ( Figure S3 H). The majority of DMRs were located in transposable element (TE) and intergenic and intronic regions, whereas the promoter and exon regions contained 409 and 373, respectively, of the DMRs, with far more hypermethylated than hypomethylated DMRs ( Figure S3 I). The methylation levels in older oocytes were greater than those in younger oocytes in terms of genomic features, including 3′ UTRs, exons, intergenic regions, introns, promoters, and TEs ( Figure S3 J). Moreover, the DMRs of genes were enriched mainly in important signaling pathways, such as the cAMP, oxytocin, and calcium signaling pathways ( Figure S3 K). Furthermore, association analysis of the transcriptome and methylation revealed several important genes potentially regulated by methylation, such as RPS15 , which were hypomethylated and upregulated in old oocytes compared with young oocytes ( Figure S3 L). Conversely, CUL4B was hypermethylated and downregulated in old oocytes compared with young oocytes ( Figure S3 L). CUL4B , a component of the cullin-ring finger ligase-4 complex, is involved in germ cell development and maternal reprogramming, and depletion of CUL4B markedly decreases endogenous 5-hydroxymethylcytosine levels. 45 Overall, there were no or only minimal changes in DNA methylation in CC samples with age ( Figures S4 A–S4C), which differed from what was observed in oocytes. Moreover, CCs presented greater DNA methylation than did oocytes ( Figures S4 B and S4C versus Figures S3 B and S3C). The DMRs in CC samples between the young and old groups were far less numerous than those in oocytes ( Figures S4 D versus S3 D), contained 2,398 DMRs hypermethylated and 2,227 DMRs hypomethylated in old CC samples ( Figure S4 E), and were enriched in several signaling pathways, including the mitogen-activated protein kinase and gonadotropin-releasing hormone (GnRH) signaling pathways ( Figure S4 F). Furthermore, the methylation levels of the promoters and gene bodies of ribosome-related genes decreased with age ( Figure S4 G). Many DMRs are located on TEs, and TE expression in both oocytes and CCs changes with increasing age ( Figures S5 A–S5E), in particular, long interspersed nuclear element 1 (L1) is markedly upregulated in aged CCs in association with decreased DNA methylation of L1 ( Figures S5 F and S5G). Quantitative reverse-transcription PCR analysis further confirmed that L1 RNA levels in CCs increased with age and that the azidothymidine (AZT, L1 inhibitor) inhibited this increase ( Figure S5 H). Moreover, the suppression of L1 expression by AZT reduced DNA damage, as evidenced by decreased 53BP1 foci ( Figure S5 I), supporting the notion that DNA damage represents one of the phenotypes of aging CCs. Notably, many L1 retrotransposons are located very close to ribosome-related genes ( RPS5 and RPL36A ) ( Figure S5 J). We also analyzed previously published RNA-seq data from mouse embryonic stem cells 46 and reported that LINE1 RNA knockdown resulted in decreased expression of ribosome-related genes ( Figure S5 K), which were upregulated in aging CCs. These data suggest a link between L1 and several ribosome-related genes. In short, the methylation of CCs also changes with age. The elevated expression of some ribosome genes and L1 is associated with decreased DNA methylation. The level of heterochromatin also changes with age. 47 We examined changes in the heterochromatin of CCs, which are typically marked by H3K9me3 and H3K27me3. H3K9me3 immunofluorescence in CCs decreased with increasing maternal age ( Figure 4 A). Loss of heterochromatin has also been reported in human prophase I-arrested oocytes with age. 48 However, H3K27me3 immunofluorescence intensity did not differ between young and old CCs ( Figure S6 ). Figure 4 H3K9me3 abundance on ribosome genes decreases with age (A) Immunostaining of H3K9me3 (red) in CCs (Passage 2) from young women and women of advanced reproductive age. Scale bars, 10 μm. n = 200 cells (from three young donors aged 28, 31, and 31 years or three older donors aged 38, 39, and 43 years). (B) Violin plots displaying the normalized signal of H3K9me3 at peaks identified by CUT&Tag in CCs collected from two repeated experiments, including paired young and relatively old women (30 vs. 41 or 31 vs. 39 years old). (C) H3K9me3 enrichment around young-specific and old-specific H3K9me3 peaks. (D) Functional enrichment analysis of genes marked by young-specific (upper) and old-specific (lower) H3K9me3 peaks. (E) Normalized signals of H3K9me3 peaks identified from genes associated with ribosome biogenesis included preribosomes, small subunit precursors, positive regulation of rRNA processing, and regulation of ribonucleoprotein complex localization in young (30 and 31 years) and old (39 and 41 years) CCs. The center indicates the center of the peaks. Right, the boxplot showing the signal of H3K9me3 enrichment. (F) Normalized signals of H3K9me3 peaks identified from genes related to the negative regulation of cytoplasmic translation in young and old CCs. (G) Normalized signal of H3K9me3 peaks in young and old CCs identified from promoters (±1,000 bp around the TSS) and gene bodies (from the TSSs to the TESs) associated with upregulated ribosome-related genes identified via RNA-seq. (A) Data are represented as mean ± SEM. (B) Violin plot shows the kernel density of the data with box in middle representing the median, the 25% and 75% interquartile range (IQR), and the ±1.5 IQR. (E–G) Box represents the median, the 25% and 75% IQR, and the ±1.5 IQR. (A, B, and E–G) Two-tailed unpaired t test. ∗∗∗ p < 0.001 and ∗∗∗∗ p < 0.0001. H3K9me3 abundance on ribosome genes decreases with age (A) Immunostaining of H3K9me3 (red) in CCs (Passage 2) from young women and women of advanced reproductive age. Scale bars, 10 μm. n = 200 cells (from three young donors aged 28, 31, and 31 years or three older donors aged 38, 39, and 43 years). (B) Violin plots displaying the normalized signal of H3K9me3 at peaks identified by CUT&Tag in CCs collected from two repeated experiments, including paired young and relatively old women (30 vs. 41 or 31 vs. 39 years old). (C) H3K9me3 enrichment around young-specific and old-specific H3K9me3 peaks. (D) Functional enrichment analysis of genes marked by young-specific (upper) and old-specific (lower) H3K9me3 peaks. (E) Normalized signals of H3K9me3 peaks identified from genes associated with ribosome biogenesis included preribosomes, small subunit precursors, positive regulation of rRNA processing, and regulation of ribonucleoprotein complex localization in young (30 and 31 years) and old (39 and 41 years) CCs. The center indicates the center of the peaks. Right, the boxplot showing the signal of H3K9me3 enrichment. (F) Normalized signals of H3K9me3 peaks identified from genes related to the negative regulation of cytoplasmic translation in young and old CCs. (G) Normalized signal of H3K9me3 peaks in young and old CCs identified from promoters (±1,000 bp around the TSS) and gene bodies (from the TSSs to the TESs) associated with upregulated ribosome-related genes identified via RNA-seq. (A) Data are represented as mean ± SEM. (B) Violin plot shows the kernel density of the data with box in middle representing the median, the 25% and 75% interquartile range (IQR), and the ±1.5 IQR. (E–G) Box represents the median, the 25% and 75% IQR, and the ±1.5 IQR. (A, B, and E–G) Two-tailed unpaired t test. ∗∗∗ p < 0.001 and ∗∗∗∗ p < 0.0001. To reveal genome-wide enrichment of H3K9me3, we performed CUT&Tag of H3K9me3 in CCs from young and older women. The overall abundance of H3K9me3 in the genome decreased with age ( Figure 4 B). H3K9me3 enrichment at promoters also decreased with maternal age ( Figure S7 A). H3K9me3 was primarily enriched in intergenic regions, with its enrichment at distal intergenic region slightly decreasing with age ( Figure S7 B). Moreover, H3K9me3 peaks were found with a specific distribution with age, including 9,999 young-specific H3K9me3 peaks that decreased with age and 17,208 old-specific H3K9me3 peaks that increased with age ( Figure 4 C). Through correlation analysis, these peaks were highly correlated within the group but differed between young and older CCs ( Figure S7 C). The young-specific H3K9me3 peaks were detected in genes related to misfolded protein binding, preribosome, regulation of ribonucleoproteins, and rRNA processing, which are involved in ribosome biogenesis, and the H3K9me3 enrichment on these genes was lower in old CCs than in young CCs ( Figures 4 D and 4E). In contrast, genes marked by old-specific H3K9me3 peaks were associated with protein localization to microtubules, negative regulation of cytoplasmic translation, regulation of cell proliferation, and oxidoreductase activity ( Figure 4 D). The old-specific H3K9me3 peaks enriched in genes related to the negative regulation of cytoplasmic translation were reduced in young CCs ( Figure 4 F). Furthermore, H3K9me3 enrichment at the promoter and gene body of ribosome-related genes was reduced in the old CCs ( Figure 4 G). The altered H3K9me3 enrichment in these terms implied that the transcriptional inhibition of H3K9me3 on ribosomal biogenesis- and translation-related genes may be abrogated during aging, corroborating the increased transcription of ribosomal genes and potential promotion of translation. Werner syndrome is a premature aging disorder caused by WRN protein deficiency. We analyzed the H3K9me3 chromatin immunoprecipitation sequencing data of WRN −/− hMSCs (human mesenchymal stem cells) published previously. 49 Consistently, H3K9me3 enrichment on ribosome-related genes was reduced in WRN −/− hMSCs ( Figure S7 D), accompanied by increased expression of ribosome-related genes ( Figure S7 E). Furthermore, CCs treated with chaetocin (H3K9me3 inhibitor) 50 exhibited upregulated expression levels of ribosome-related genes, such as RPS19 , RPL22 , and RPL36A ( Figure S7 F). Combined with the aforementioned results, the changes in the expression of ribosome-related genes during aging can be regulated by alterations in heterochromatic H3K9me3. The lysosome-related genes presented increased H3K9me3 enrichment in gene body regions, whereas promoter regions did not change with age ( Figure S7 G), which was correlated with downregulation of lysosome-related genes. In addition, genes related to oxidoreductase activity were enriched with the old-specific H3K9me3 peak ( Figures 4 D and S7 H) and were downregulated with age ( Figures S7 I and S7J). These data suggest that H3K9me3 enrichment at specific loci with increasing age may influence the expression of genes related to lysosomes and oxidoreductase activity in CCs. Upregulation of ribosome-related entries was identified in old mouse oocytes by comparative transcriptome analysis with young oocytes, suggesting that ribosome dysregulation was conserved between human and mouse oocytes during aging ( Figures S8 A and S8B). We identified 53 ribosome-related genes upregulated in aged mouse oocytes ( Figure S8 C) and five ribosome-related genes, similar to those in human oocytes, including Rps5 , Mrps24 , Mrpl53 , Rpl36 , and Rpl5 ( Figure S8 D). We took advantage of rapamycin to inhibit mTOR and translation to explore its potential function in delaying aging. Oocytes and CCs were collected from the ovaries of young (2-month-old) and old (10-month-old) mice ( Figure S9 A). Rapamycin at 0.5 μM blocked the increased phosphorylation of 4E-BP1 and S6 in old CCs, which regulated translation and ribosome biogenesis ( Figures S9 B and S9C). Furthermore, rapamycin inhibited the activity of SA-β-gal and the increase in reactive oxygen species (ROS) levels in old CCs ( Figures S9 D and S9E). The elevated p-S6 protein level in aged oocytes during in vitro maturation (IVM) was also decreased by rapamycin ( Figure S9 F), whereas the S6 protein level did not change ( Figure S9 G). Moreover, rapamycin promoted the rate of IVM ( Figure S9 H) and reduced ROS levels in aging oocytes ( Figure S9 I). Additionally, the rates of disrupted spindles and chromosome misalignment were elevated in aging oocytes, and rapamycin alleviated these abnormalities ( Figure S9 J). Hence, rapamycin treatment effectively delays the aging of oocytes and CCs. We tested whether rapamycin can improve oocyte quality, subsequent embryo development, and clinical pregnancy. A randomized controlled clinical trial was conducted. A total of 122 women underwent screening, and 100 met the eligibility criteria. The 100 women who provided informed consent were randomly assigned to the control group or rapamycin group, with 50 patients in each group ( Figure 5 A). After randomization, patients in the control group received a standardized GnRH agonist long protocol, whereas patients in the rapamycin group received a standardized GnRH agonist long protocol plus rapamycin treatment for 21–28 days from the day of endogenous hormone downregulation until the oocyte retrieval day. Previous clinical trials of rapamycin and aging used doses of approximately 0.5–2.0 mg for 8 or 12 weeks. 51 , 52 In our trial, patients in the rapamycin group received 1 mg rapamycin orally daily for 21–28 days. Three patients in the control group and five patients in the rapamycin group did not have oocytes after controlled ovarian hyperstimulation according to the protocol. Figure 5 Rapamycin improves fertility in human IVF clinics (A) Enrollment and outcomes. (B) Poor-quality (arrested or fragmented) and good-quality embryos on day 3 (upper); poor-quality and top-quality blastocysts on day 5 (lower); scale bars, 100 μm. (C) Outcomes of controlled ovarian hyperstimulation in the control and rapamycin groups. The plus-minus values are the means ± SDs. The asterisks indicate the medians (Q1, Q3). Differences between the two groups were calculated via the Wilcoxon test. Embryos were scored according to morphological criteria, with “good” defined as grade I: a cell number of 7–9, even cell size, less than 10% fragmentation, and no multinucleation. Day-5/6 blastocyst cultures were performed in cases of poor embryo quality or at the request of a patient. 53 , 54 A top-quality blastocyst-stage embryo is at least in the expanded blastocyst stage (BL3) and has a type A inner cell mass (in which cells are tightly packed, with many cells present) and a trophoectoderm that is either type A (with many cells forming a cohesive epithelium) or type B (with few cells forming a loose epithelium). 55 , 56 (D) Outcomes of pregnant patients with or without rapamycin treatment (control). p values were calculated via the chi-square test. Clinical pregnancy was defined as the observation of a gestational sac via ultrasonography. (C and D) Continuous data are represented as the mean ± SD or medians (Q1, Q3); categorical data are represented as frequencies and percentages. (C) Wilcoxon test; (D) chi-square test. Rapamycin improves fertility in human IVF clinics (A) Enrollment and outcomes. (B) Poor-quality (arrested or fragmented) and good-quality embryos on day 3 (upper); poor-quality and top-quality blastocysts on day 5 (lower); scale bars, 100 μm. (C) Outcomes of controlled ovarian hyperstimulation in the control and rapamycin groups. The plus-minus values are the means ± SDs. The asterisks indicate the medians (Q1, Q3). Differences between the two groups were calculated via the Wilcoxon test. Embryos were scored according to morphological criteria, with “good” defined as grade I: a cell number of 7–9, even cell size, less than 10% fragmentation, and no multinucleation. Day-5/6 blastocyst cultures were performed in cases of poor embryo quality or at the request of a patient. 53 , 54 A top-quality blastocyst-stage embryo is at least in the expanded blastocyst stage (BL3) and has a type A inner cell mass (in which cells are tightly packed, with many cells present) and a trophoectoderm that is either type A (with many cells forming a cohesive epithelium) or type B (with few cells forming a loose epithelium). 55 , 56 (D) Outcomes of pregnant patients with or without rapamycin treatment (control). p values were calculated via the chi-square test. Clinical pregnancy was defined as the observation of a gestational sac via ultrasonography. (C and D) Continuous data are represented as the mean ± SD or medians (Q1, Q3); categorical data are represented as frequencies and percentages. (C) Wilcoxon test; (D) chi-square test. Among the remaining patients, 19 patients underwent IVF, and 28 patients underwent intracytoplasmic sperm injection (ICSI) in the control group. In the rapamycin group, 12 patients underwent IVF, and 33 patients underwent ICSI. There were no significant differences in baseline variables between patients undergoing IVF or ICSI in either the control or rapamycin groups ( Figures S10 A and S10B), especially in indicators related to embryologic characteristics such as the number of zygotes and embryos. Therefore, the effect, if any, of different fertilization procedures was minimal for our subsequent analysis. Notably, significantly more zygotes, embryos, and good-quality embryos were obtained in the rapamycin group than in the control group (median value, zygotes: 3.0 vs. 2.0, p = 0.012; embryos: 2.0 vs. 1.0, p = 0.001; and good-quality embryos: 2.0 vs. 1.0, p < 0.001) ( Figures 5 B and 5C). The number of metaphase II oocytes retrieved and baseline variables, including anti-mullerian hormone (AMH), FSH, and AFC, did not differ between the control and rapamycin groups ( Figure 5 C). The number and top quality of blastocysts in the rapamycin group were significantly greater than those in the control group (median value, blastocysts: 2.0 vs. 1.0, p = 0.012; top-quality blastocysts 55 : 1.0 vs. 0, p = 0.043) ( Figures 5 B and 5C). These results indicate that the addition of rapamycin to patients can effectively improve oocyte quality and subsequent embryo development following the fertilization of retrieved oocytes. To date, 40 of 45 patients (88.9%) in the rapamycin group and 39 of 47 patients (83.0%) in the control group completed embryo transfer ( Figure 5 A), and 20 of 40 patients in the rapamycin group and 28 of 39 patients in the control group opted to undergo transfer of day 3 embryos, whereas 20 of 40 patients in the rapamycin group and 11 of 39 patients in the control group opted to undergo transfer of day 5–6 blastocysts ( Figure 5 D). The decision for embryo transfer largely depended on embryo quality on day 3, when more good-quality 4- to 8-cell embryos were available, and the embryos were subjected to further culture to the blastocyst stage. After embryo transfer, the rate of clinical pregnancy in the rapamycin group was significantly greater than that in the control group (50.0% vs. 28.2%), with a rate ratio of 1.77 (95% confidence interval [CI], 0.98 to 3.19; p = 0.047) ( Figure 5 D). The pregnancy rate might differ between blastocyst-stage (day 5–6) embryo transfer and cleavage-stage (day 3) embryo transfer. Hence, we separately compared the pregnancy rates for patients who underwent day 3 embryo transfer and those who underwent day 5–6 embryo transfer. The clinical pregnancy rate of the patients who underwent day 5–6 blastocyst transfer in the rapamycin group was significantly greater ( p = 0.021) than that of the control group (27.5% vs. 7.7%; rate ratio, 3.58; 95% CI, 1.08 to 11.85) ( Figure 5 D). These findings suggest that blastocyst transfer in patients treated with rapamycin can be superior to day 3 embryo transfer in terms of clinical pregnancy. The transfer of top-quality blastocysts to patients receiving rapamycin improved the clinical pregnancy rate. Furthermore, we tracked the live birth rate, and thus far, 10 of the 14 patients (71.4%) in the rapamycin group who were followed up to their due date had delivered live-born infants, and 6 of the 8 patients (75.0%) in the control group had delivered live-born infants ( Tables S5 and S6 ). Some patients had not yet reached the appropriate date (rapamycin, 6 patients; control, 3 patients), and the current live birth rate did not differ between the rapamycin and control groups. These results, nevertheless, indicate that rapamycin does not negatively impact the live birth rate. The results of clinical trials thus far indicate that rapamycin noticeably increases oocyte and embryo quality and the clinical pregnancy rate. Moreover, the pregnant patients had successful live birth delivery rates at least comparable to those of the controls who did not receive rapamycin, further confirming the safety of rapamycin. We also examined the protein levels of S6 and p-S6 in CCs collected from patients who received rapamycin compared with those collected from controls who were not treated with rapamycin. Rapamycin indeed reduced p-S6/S6 levels ( Figure S10 C). Further refinement of the rapamycin dosage and time and duration of treatment may further increase the efficacy of pregnancy and fertility.

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Lin Liu ( [email protected] ). This study did not generate new, unique reagents. • All the data are available in the main text or the supplemental information . The raw RNA-seq, MethylC-Seq, and CUT&Tag sequencing data generated in this study have been deposited in the Genome Sequence Archive (GSA) of the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences, under accession number HRA005257 (BioProject accession: PRJCA018848), which can be accessed at https://ngdc.cncb.ac.cn/gsa-human/ . These data are available under restricted access, as individual genomic sequencing data are protected owing to patient privacy and Regulations on the Management of Human Genetics Resources of China. The raw data can be requested via the GSA-Human System and can be authorized for downloading by the Data Access Committee for research and noncommercial use only. The RNA-seq data for the oocytes and CCs of donors (N8, N13, and N12) were obtained from our previous publication, GSE155489 ( GSM4705187 – GSM4705192 ; GSM4705199 – GSM4705202 ). Mouse oocyte RNA-seq data were downloaded from our previous publication ( GSE184637 ). • This paper does not report original code. • Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. All the data are available in the main text or the supplemental information . The raw RNA-seq, MethylC-Seq, and CUT&Tag sequencing data generated in this study have been deposited in the Genome Sequence Archive (GSA) of the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences, under accession number HRA005257 (BioProject accession: PRJCA018848), which can be accessed at https://ngdc.cncb.ac.cn/gsa-human/ . These data are available under restricted access, as individual genomic sequencing data are protected owing to patient privacy and Regulations on the Management of Human Genetics Resources of China. The raw data can be requested via the GSA-Human System and can be authorized for downloading by the Data Access Committee for research and noncommercial use only. The RNA-seq data for the oocytes and CCs of donors (N8, N13, and N12) were obtained from our previous publication, GSE155489 ( GSM4705187 – GSM4705192 ; GSM4705199 – GSM4705202 ). Mouse oocyte RNA-seq data were downloaded from our previous publication ( GSE184637 ). This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

We propose that elevated transcription of ribosome genes and components and aberrant proteostasis impair oocytes and their surrounding CCs, providing a mechanism to explain age-related poor embryo development and infertility. Transcriptome analysis of single oocytes and CCs from aged women undergoing IVF revealed aberrant upregulation of ribosome genes by the mid-thirties, and this consistent characteristic could be explained by the close interactions and communication between oocytes and CCs. The potential alterations in the state of these two cell types are likely to affect each other, thereby affecting folliculogenesis and oogenesis. 57 , 58 Mechanistically, the elevated transcription of ribosome genes could be attributable to the loss or reduction of heterochromatin and altered methylation. Indeed, inhibition of H3K9me3 notably increases the expression levels of ribosome-related genes, such as RPS19 , RPL22 , and RPL36A . RPL22 is a driver of cellular senescence, promotes the degradation of the heterochromatin proteins HP1γ and KAP1, and stimulates the transcription of rRNA. 59 Interestingly, TORC1 promotes the transcription of rRNA and ribosome-related genes by preventing heterochromatin formation. 60 Aged hematopoietic stem cells and fibroblasts from patients with Hutchinson-Gilford progeria syndrome (premature aging disorders) also exhibit increased expression of ribosomal genes or proteins and rRNA hypomethylation. 38 , 61 Thus, these previous studies and our findings support the notion that ribosome dysregulation is tightly linked to heterochromatin and DNA methylation during aging. Our data also confirmed previous findings in oocytes that meiosis, the microtubule and actin cytoskeleton, and cohesion gene expression decrease with maternal age. 4 Defective spindle and actin assembly and chromosome misalignment are linked to high aneuploidy levels in older eggs, contributing to age-related infertility. 62 , 63 , 64 , 65 Additionally, the expression of lysosomal genes in CCs decreases with increasing maternal age, and aberrant protein aggregates accumulate. Like in mouse oocytes, 66 human oocytes do not show a significant increase in protein aggregation with age. Protein degradation mechanisms in oocytes might differ from those in CCs, and this warrants further investigation. Our findings that increased transcription of ribosome-related genes is accompanied by increased protein aggregation and disrupted proteostasis in aging CCs are consistent with findings that increased ribosome pausing and collision result in accumulated protein aggregation and impaired proteostasis during aging in C. elegans and yeast models. 67 Aging-induced ribosome-related quality control overload may require increased expression of ribosome-related genes, further impairing proteostasis. Ribosomes are maintained in a dormant state and translationally repressed in vertebrate eggs (zebrafish and Xenopus laevis ) and transition to an active state after fertilization. 68 Notably, ribosome activity is downregulated in the human population with exceptional longevity from Hainan Province, China. 69 Similarly, we show that inhibiting ribosome biogenesis and protein synthesis could abrogate oocyte and CC senescence, improving embryo quality and development and thus pregnancy. Taken together, increased transcription of ribosome genes is implicated as a potential player in the deterioration of oocyte quality and unexplained infertility with age and can serve as a potential target to improve fertility. Rapamycin, which is commonly employed as an immunosuppressant for organ transplantation and as an anti-aging strategy, may play a role in the treatment of infertility in older women. Some clinical trials on rapamycin and aging have confirmed that oral rapamycin can affect erythrocyte parameters and senescence-associated secretory phenotype-related factors in elderly individuals, 51 , 52 which reflects the systemic effect of rapamycin on patients. However, in our clinical trial, whether rapamycin improves clinical pregnancy in infertile women is direct or based on systemic effects, and larger and more comprehensive clinical trials are warranted. The clinical implications and translational potential of our findings necessitate further validation in larger patient cohorts. Another limitation is the sample size of the sequencing; however, through multiple sequencing methods, systematic experimental validation, and critical intervention experiments, we have demonstrated that ribosome represents a promising therapeutic target for female infertility.

The age-related decline in female fertility is attributable primarily to oocyte quality, 1 which impairs embryo development and causes repeated in vitro fertilization (IVF) failures. In human oocytes, chromosome segregation errors increase with age, especially after the mid-thirties. Meiotic defects leading to embryonic aneuploidy are the prime driver of reproductive aging in women. 2 , 3 , 4 , 5 , 6 Factors contributing to aneuploidy include precocious separation of sister chromatids, chromosome misalignment, spindle disruption, and deficient spindle checkpoints. 3 , 4 , 5 , 6 , 7 , 8 Embryos resulting from the fertilization of aneuploid oocytes typically arrest during development or miscarry. 9 The rate of chromosome segregation errors accelerates from 35 years of age onward. 5 In addition to chromosome aneuploidy, the DNA damage response, oxidative stress, mitochondrial dysfunction, telomere attrition, autophagy, inflammation, and fibrosis also contribute to ovarian aging. 10 , 11 , 12 , 13 , 14 , 15 The human ovary contains several cell types, particularly oocytes and surrounding cumulus cells. 16 , 17 Cumulus and granulosa cell connections enable bidirectional communication with oocytes. These proteins involved in bidirectional communication are essential for normal folliculogenesis and germ cell development. 18 , 19 DNA methylation provides another hallmark of cellular aging (epigenetic aging clocks), enabling estimation of biological age for most tissues across the entire life course. 20 , 21 Epigenetic aging clocks used to accurately predict biological age have been identified in somatic cells but are more complex in the context of oocyte aging. 22 Previous single-cell sequencing studies arbitrarily grouped the age as young or old to explore the impact of aging on female oocytes or ovaries. 23 , 24 , 25 Through integrated multi-omics analysis, including transcriptome analysis via correlation analysis with age, the methylome, and the histone modification, we identified molecular changes in aging oocytes and cumulus cells (CCs). Notably, both oocytes and CCs exhibit striking transcriptional changes by age 34 without artificial grouping by age in our study, providing a potential molecular basis for the declining fecundity in women. Both oocytes and CCs display consistently upregulated ribosome genes, which promote protein translation. Notably, the inhibition of translation by rapamycin, which has been approved for routine clinical use as an immunosuppressant, greatly improves fertility in women who previously failed assisted reproduction.

The authors declare no competing interests.

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies H3K9me3 Abcam Cat#ab8898; RRID: AB_306848 H3K27me3 Millipore Cat#07–449; RRID: AB_310624 53BP1 Abcam Cat#ab36823; RRID: AB_722497 LINE-1 ORF1p Millipore Cat#MABC1152; RRID: AB_2941775 Rabbit monoclonal anti-Phospho-S6 Ribosomal Protein (Ser235/S236) Cell Signaling Technology Cat#4858S; RRID: AB_916156 Rabbit monoclonal anti-S6 Ribosomal Protein (5G10) Cell Signaling Technology Cat#2217S; RRID: AB_331355 anti-β-Actin ABclonal Cat#AC026; RRID: AB_2768234 4E-BP1 Cell Signaling Technology Cat#9644T; RRID: AB_2097841 Phospho-4E-BP1 (Ser65) Cell Signaling Technology Cat#9451T; RRID: AB_330947 mTOR Cell Signaling Technology Cat#2983S; RRID: AB_2105622 Phospho-mTOR (Ser2448) Cell Signaling Technology Cat#5536T; RRID: AB_10691552 alpha-Tubulin-FITC Sigma-Aldrich Cat#F2168; RRID: AB_476967 Nucleolin Cell Signaling Technology Cat#14574S; RRID: AB_2798519 Donkey anti-rabbit IgG Alexa Fluor 594 antibody Thermo Scientific Cat#A-21207; RRID: AB_141637 Donkey anti-mouse IgG Alexa Fluor 488 antibody Thermo Scientific Cat#A-21202; RRID: AB_141607 Biological samples Female oocytes Patient donators in this study N/A Female cumulus cells Patient donators in this study N/A Chemicals, peptides, and recombinant proteins Hyaluronidase Sigma-Aldrich Cat#H6254-500MG Polyvinylpyrrolidone Sigma-Aldrich Cat#P0930-50G Tyrode’s Acidic Solution Sigma-Aldrich Cat#T1788-100ML DMEM NUTRIENT MIX F12 Thermo Scientific Cat#11330032 Hoechst 33342 Thermo Scientific Cat#H3570 MG132 MCE Cat#HY-13259 Bafilomycin A Selleck Cat#S1413 Chaetocin MCE Cat#HY-N2019 LysoTracker red Beyotime Cat#C1046 Azidothymidine Sigma-Aldrich Cat#A2169 Chemiluminescent HRP substrate Millipore Cat#WBKLS0500 PVDF membranes Millipore Cat#ISEQ00010 FastStart Universal SYBR Green Master Mix Roche Cat#4913914001 M-MLV Reverse Transcriptase Invitrogen Cat#28025-021 Agencourt Ampure XP beads Beckman Cat#A63881 M-280 Streptavidin Dynabeads Life Technologies Cat#65001 KAPA HiFi HotStart DNA Polymerase KAPA Biosystems Cat#KK2801 Spermidine MCE Cat#HY-B1776 Roche Complete Protease Inhibitor EDTA-Free tablets Sigma-Aldrich Cat#5056489001 Critical commercial assays Senescence-associated β-galactosidase staining kit Beyotime Cat#C0602 RNeasy RNA Micro Kit QIAGEN Cat#74034 PROTEOSTAT® Aggresome detection kit Enzo Life Sciences Cat#ENZ-51035-K25 TruePrep DNA Library Prep Kit V2 for Illumina® Vazyme Biotech Cat#TD503-02 QIAamp® DNA Micro Kit QIAGEN Cat#56304 BeyoClick™ HPG-594 Protein Synthesis Assay Kit Beyotime Cat#P1209S Deposited data The raw RNA-seq data for female oocytes and cumulus cells This paper NGDC: HRA005257 The raw MethylC-Seq data for female oocytes and cumulus cells This paper NGDC: HRA005257 The raw CUT&Tag data for female cumulus cells This paper NGDC: HRA005257 Oligonucleotides Primers for qRT-PCR, see Table S7 This paper N/A Software and algorithms Trimmomatic (v0.38) Bolger et al. 70 http://www.usadellab.org/cms/index.php?page=trimmomatic hisat2 (v2.1.0) Kim et al. 71 https://ccb.jhu.edu/software/hisat2/index.shtml featureCounts (v1.6.3) Liao et al. 72 http://subread.sourceforge.net/ R (v4.4.1) CRAN https://www.r-project.org/ DAVID (v6.8) Huang et al. 73 https://david.ncifcrf.gov Cytoscape (v3.9.1) Shannon et al. 74 https://cytoscape.org GSEA (v4.3.3) Subramanian et al. 75 https://www.gsea-msigdb.org/gsea/index.jsp Bismark (v0.19.1) Krueger et al. 76 http://www.bioinformatics.babraham.ac.uk/projects/bismark/ MethylKit (1.30.0) Akalin et al. 77 https://www.bioconductor.org/packages/release/bioc/html/methylKit.html DESeq2 (v1.44.0) Love et al. 78 https://www.bioconductor.org/packages/release/bioc/html/DESeq2.html Bowtie2 (v2.3.4) Langmead et al. 79 http://bowtie-bio.sourceforge.net/bowtie2/index.shtml macs2 (v2.1.2) Zhang et al. 80 https://pypi.org/project/MACS2/ ChIPseeker (v1.28.3) Yu et al. 81 https://guangchuangyu.github.io/software/ChIPseeker/ deepTools (v3.2.0) Ramirez et al. 82 https://deeptools.readthedocs.io/en/develop/ SPSS Statistics (v29.0.0) IBM https://www.ibm.com/spss ImageJ NIH https://imagej.nih.gov/ij/ Prism (v8.4.3) GraphPad Software https://www.graphpad.com/ Young (2-month-old) and old (10-month-old) female C57BL/6 background mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. All the mice were cared for in individually ventilated cages (IVCs) on a standard 12 h:12 h dark cycle in the sterile animal facility at the College of Life Sciences. The mice used in this study were approved by the Nankai University Animal Care and Use Committee, and all mouse experiments were carried out in accordance with the guidelines and relevant regulations. Isolation of mouse cumulus cells (mCCs) and culture were performed as previously described. 83 Briefly, PMSG was injected into the abdominal cavity of mice 46 h before the mCCs were isolated. The mice were humanely sacrificed, and the ovaries were dissected. Insulin syringes were used to puncture visible follicles on the surface of the ovaries under a stereomicroscope to release mCCs into the culture medium, avoiding isolation of mCCs from small follicles. In addition, the oocytes were filtered out via a 40-μm cell strainer (Falcon). The isolated mCCs were subsequently washed three times and seeded in culture medium. This study was approved by the Ethics Committee of TianJin Medical University General Hospital (No: IRB2018-102-01), the Sixth Medical Center of Chinese People’s Liberation Army (PLA) General Hospital (No: HZKY-PJ-2021-33), Shanxi Medical University (No. 2022SJL75) and Amcare Women’s & Children’s Hospital (No: AM2020-001-01; AM-2024001) and was conducted in accordance with approved institutional guidelines. Written informed consent was obtained from the donors. The isolated oocytes and surrounding CC samples were obtained from 100 female donors ranging in age from 23 to 48 years. The inclusion criteria for donors undergoing IVF/ICSI with their own oocytes included maternal age and male factor infertility, and donors were excluded from the study if they had endometriosis, cancer, chronic infections, or autoimmune or genetic diseases. The mean ovarian reserve was 18 (range 3–55), which was represented by the antral follicle count (AFC) obtained from donors whose clinical samples were used for RNA-seq and MethylC-seq. The clinical characteristics of all donors, including cell type and age, were collected and are summarized in Table S1 . We conducted a randomized, controlled trial at the Children’s Hospital of Shanxi and Women’s Health Center of Shanxi. The study was approved by the Ethics Committee of Shanxi Medical University (ethics number: 2022SJL75) and conducted according to the Declaration of Helsinki 2013, and the trial was registered at the Chinese Clinical Trial Registry ( https://www.chictr.org.cn/ ) with registration number (ChiCTR2300069828). Independent data and safety monitoring boards were established to oversee the study. All patients provided written informed consent before participation. From April 2023, a total of 122 patients underwent screening, and 100 patients met the eligibility criteria and were enrolled in the study. The study included infertile women who had a history of more than one IVF cycle failure and who provided informed consent. The average age of all the women was approximately 36 years. The exclusion criteria included patients with infertility due to male factors such as oligozoospermia or azoospermia, and patients with sexually transmitted diseases, genetic diseases, uterine malformations and a history of endometriotic cyst surgery. Patients with allergies to rapamycin and its derivatives and other serious systemic diseases were excluded. Referring to a published clinical trial article, 84 a priori power analysis was conducted to determine the sample size required to detect statistically significant changes associated with the clinical pregnancy rate. The study was designed to have a power of 80% at a two-sided significance level (α) of 0.05 to detect an absolute difference of 30 percentage points in the clinical pregnancy rate between the two groups (clinical pregnancy rate of 30% for the control group and 60% for the rapamycin group) by means of Pearson’s chi-square test. At least 42 patients per study group were needed, a number that we increased to 50 to allow for a dropout rate of 10% (Details are provided in Data S1 ). The patients were randomly assigned to one of the two study groups (the control and rapamycin groups) at a 1:1 ratio, and the statisticians used SPSS statistical software to generate random numbers. The odd numbers were assigned to the rapamycin group, and the even numbers were assigned to the control group. The random numbers and grouping information were prepacked in the envelope, which was unknown to the clinical investigators. When the subjects were enrolled in the group, the envelopes were opened in sequence and grouped according to the groups in the envelopes. All patients received a standardized long GnRH agonist protocol, oocyte retrieval, fertilization, and planned embryo transfer. In brief, in the mid-luteal phase of the previous menstrual cycle, 0.1 mg of triptorelin acetate (triptorelin) was used for downregulation for 14–16 days. On the 2nd to 5th days of the menstrual cycle, the serum hormone levels (FSH, LH, and E2) and ultrasonography results were monitored. Gonadotropin (Gn) was used after the downregulation standard was reached (FSH ≤5 mIU/mL, LH ≤ 5 mIU/mL, E2 ≤ 50 pg/mL). Gonadotropin (Gn) was given at 75–300 IU. The Gn dose was adjusted according to the growth of the follicle and the hormone levels. When more than one follicle with a diameter greater than 18 mm appeared, intramuscular injection of human chorionic gonadotropin (hCG) 6000–10000 IU was given. Oocyte retrieval was performed 36 h later. The obtained oocytes were inseminated approximately 4–6 h by a conventional method or intracytoplasmic sperm injection according to the sperm quality, and a fertilization check was then performed 16 to 18 h after insemination. The embryos were scored according to the morphological criteria. 54 The quality of the blastocysts was assessed according to the criteria of Gardner and Schoolcraft. 56 High-quality embryos on day 3 or blastocysts on days 5–6 were selected for fresh transfer or cryopreserved via vitrification and frozen-embryo transfer. Transfer on day 3 or days 5–6 depends on the quality of the embryo and the patient’s wishes. When more embryos were high quality on day 3 and the probability of achieving high-quality blastocysts at days 5–6 was high, the patients were persuaded to choose continuous cultures to obtain 5–6 blastocysts for embryo transfer. Otherwise, the patients were suggested to choose embryo transfer immediately without attempting further culture to obtain blastocysts. Luteal-phase support was administered before embryo transfer and continued until 10 weeks of gestation. Biochemical pregnancy was defined as a human chorionic gonadotropin level of more than 10 mIU per milliliter, as measured at 10 days after embryo transfer. Clinical pregnancy was defined as the presence of a gestational sac in the uterine cavity at 30 days after embryo transfer, as detected by ultrasonography. Live birth was defined as the delivery of a live-born infant around the due date of pregnancy. 84 The dosage and duration of rapamycin were determined according to previous articles and registered clinical trials. 51 , 52 For patients in the rapamycin group, oral rapamycin (sirolimus, Rapamune; Pfizer) at a daily dose of 1 mg was administered for 21–28 days beginning on the day of endogenous hormone downregulation until oocyte retrieval, which was defined as short-term use. The primary outcomes were oocyte number and embryo number. The secondary outcome was the rate of clinical pregnancy. All donors underwent controlled ovarian stimulation. An ultrasound scan and serum estradiol assays were performed to monitor follicular size, which ensured that the cumulus–oocyte complexes obtained from comparable follicles of the same developmental stage. When two or more follicles were at least 12 mm in diameter, 10,000 IU human chorionic gonadotropin (hCG) was administered 36 h before oocyte retrieval. The cumulus‒oocyte complex (COC) was isolated via ultrasound-guided vaginal puncture and classified according to the oocyte nuclear maturation stage: GV (germinal vesicle), MI (metaphase I) and MII (metaphase II). We collected only GV-stage oocytes and surrounding CCs for this study, whereas MII-stage oocytes were used for clinical fertilization. The CCs were collected as previously described. 85 Briefly, CCs were mechanically stripped from oocytes under stereomicroscopy, and then, the isolated CCs were dispersed into single cells with 0.03% hyaluronidase (H6254-500MG, Sigma‒Aldrich) and resuspended three times in PBS. The separated CCs were counted as up to 500 cells and placed in the lysate. Tyrode’s acidic solution (T1788-100ML, Sigma‒Aldrich) was used to facilitate stripping of the zona pellucida to produce naked oocytes. Oocytes were observed under a microscope to ensure the absence of contamination with CCs. Naked oocytes were carefully washed three times with PBS containing 0.1% polyvinylpyrrolidone (PVP, P0930-50G, Sigma‒Aldrich) to prevent them from adhering to tools or dishes and then placed in lysis buffer. Human cumulus cells were dissected from the GV-stage cumulus–oocyte complex (COC) and purified from Amcare Women’s & Children’s Hospital. Briefly, cumulus cells were mechanically stripped from oocytes under stereomicroscopy and then dispersed into single cells with 0.03% hyaluronidase. Next, the cells were suspended in 1 mL of medium containing DMEM/F12 supplemented with 10% FBS and 1% penicillin–streptomycin and centrifuged again. Next, the supernatant was decanted, and the cell pellet was resuspended in fresh medium and plated. The cells were cultured for three days at 37°C in a 5% CO 2 incubator. Young and old CC samples were taken from the same day and cultured for the same number of days in each batch of experiments. At 2 and 10 months of age, female mice were humanely sacrificed after PMSG injection at 44–46 h intervals. After the ovaries were dissected, fully grown germinal vesicle (GV) oocytes were collected under a microscope by pricking the follicles in IVM medium (α-MEM with 5% fetal bovine serum (FBS), 0.24 mM sodium pyruvate, 1 IU/mL PMSG, and 1.5 IU/mL hCG) via an insulin syringe. The obtained GV oocytes were divided into two equal parts and then placed in IVM medium (with or without rapamycin). Oocytes were matured in IVM medium for 17–18 h at 37°C. MII oocytes were determined by extrusion of the first polar body. 2′,7′-Dichlorodihydrofluorescein diacetate (H2DCFDA, HY-D0940, MCE) was utilized to estimate the ROS levels via a previously described method. 86 Briefly, denuded mouse oocytes were incubated in HEPES-buffered KSOM (HKSOM) medium containing 5 μM H2DCFDA in the dark for 30 min at 37°C and then washed three times in HKSOM medium prior to being mounted on a glass slide and imaged with a fluorescence microscope. Mouse cumulus cells were incubated with 5 μM H2DCFDA solution in PBS in the dark for 30 min at 37°C, washed three times with PBS, and fluorescence was detected and imaged with an Axio-Imager Z2 fluorescence microscope (Carl Zeiss). In accordance with a previous method, 87 the spindles and chromatin of mouse oocytes were stained and observed via immunofluorescence microscopy. Mouse oocytes were fixed in fixative (MTSB XF) at 37°C for at least 30 min and then washed four times with washing buffer (phosphate-buffered saline supplemented with 0.02% NaN3, 0.01% Triton X-100, 0.2% nonfat dry milk, 2% goat serum, 2% bovine serum albumin and 0.1 M glycine). Afterward, the oocytes were left in washing buffer for 2 h at 37°C for blocking. Oocytes were incubated with FITC-α-tubulin (1:100, F2168, Sigma) overnight at 4°C. Next, the samples were washed and stained with DAPI to label the DNA. For the expression of p-S6 and S6, fixed oocytes were left in washing buffer for 2 h at 37°C for blocking and then incubated with p-S6 (1:100, 4858S, Cell Signaling Technology) and S6 (1:100, 2217S, Cell Signaling Technology) antibodies overnight at 4°C. Oocytes were washed and incubated with secondary donkey anti-rabbit IgG Alexa Fluor 594 antibody (1:200, A-21207, Thermo Scientific) at 37°C for 2 h and stained with DAPI to label DNA. Oocytes were mounted on glass slides, sealed with nail polish, and imaged with an Axio-Imager Z2 fluorescence microscope (Carl Zeiss). ImageJ was used for relative fluorescence quantification. The cells were washed twice with PBS, fixed with fresh 3.7% paraformaldehyde for 30 min at 4°C, permeabilized with 0.1% Triton X-100 in blocking buffer (3% goat serum plus 0.1% BSA in PBS) for 20 min at room temperature (RT), incubated with blocking buffer for 1 h at RT, and stained with primary antibodies overnight at 4°C. The cells were subsequently incubated with fluorescence-labeled secondary antibodies for 2 h at RT. Hoechst 33342 (Thermo Scientific, H3570) was used to stain the nuclear DNA. The antibodies used in this experiment were as follows: H3K9me3 (1:200, ab8898, Abcam), H3K27me3 (1:200, 07–449, Millipore), LINE-1 ORF1p (1:100, MABC1152, Millipore), 53BP1 (1:300, ab36823, Abcam), donkey anti-rabbit IgG Alexa Fluor 594 (1:200, Thermo Scientific, A-21207), and donkey anti-mouse IgG Alexa Fluor 488 (1:200, Thermo Scientific, A-21202). The fluorescence was detected and imaged via an Axio-Imager Z2 fluorescence microscope (Carl Zeiss). The integrated fluorescence intensity was estimated via ImageJ software. For LysoTracker red staining, the cells were seeded on glass bottom cell culture dishes (NEST, 801002, TC-treated). To stain the cells, 75 nM LysoTracker red (Beyotime, C1046) was added to the CC medium for 30 min at 37°C, and Hoechst 33342 was used to label the nuclear DNA. The cells were then washed once with CC medium, which was replaced with fresh CC medium, followed by confocal imaging. The lysosome inhibitor bafilomycin A (700 nM, Selleck, S1413) was added to CC medium for 6 h as a control. 88 Proteostat staining was performed according to the instruction manual of the PROTEOSTAT Aggresome detection kit (Enzo Life Sciences; ENZ-51035-K25). Briefly, the cells were washed twice with PBS, fixed with 4% formaldehyde for 30 min at room temperature, treated with permeabilization solution (0.5% Triton X-100, 3 mM EDTA, pH 8.0) on ice, gently shaken for 30 min, washed twice with PBS, and stained with a 1:500 Proteostat in 1× Assay buffer for 2 h at room temperature and with Hoechst 33342 for nuclei. Afterward, the samples were washed twice with PBS and placed under coverslips on slides. The stained cells were imaged via confocal microscopy with a standard rhodamine filter set for cell aggresome signaling and a DAPI filter set for imaging the nuclear signal. The cells were treated with the proteasome inhibitor MG132 (10 μM, MCE, HY-13259) for 12 h as a control. 89 For the LysoTracker and Proteostat staining experiments, we completed a total of five batches of independent experiments, and the same settings of confocal microscopy were applied to each batch of all experimental conditions. For the quantification of the LysoTracker + and Proteostat + areas per cell via ImageJ, the images were converted to black and white, and thresholds were set to outline the positive areas. The same threshold values were used for all images across all conditions in each batch of independent experiments. For the proteostat staining of oocytes, refer to a previously published article. 66 Human GV oocytes were fixed with 4% formaldehyde for 1 h at room temperature and then washed and permeabilized with shaking on ice for 1 h. After brief washing, the oocytes were incubated with 1× assay buffer containing 0.1% BSA and a 1:500 Proteostat at room temperature. The oocytes were washed three times with 1×Assay Buffer containing 0.1% BSA, after which the fluorescence was imaged. Three batches of independent experiments were conducted. Protein synthesis was performed according to the instructions of the BeyoClick HPG-594 Protein Synthesis Assay Kit (P1209S, Beyotime). Briefly, cumulus cells were incubated in 1× HPG working solution with methionine-free and serum-free culture medium for 30 min at 37°C, fixed at room temperature for 15 min, and permeabilized at room temperature for 15 min. Click reaction solution was prepared according to the instructions and incubated at room temperature in the dark for 30 min. Nuclear DNA was stained with Hoechst 33342 solution, and then fluorescence was imaged. Human cumulus cells were treated with rapamycin (0.50 μM) or CHX (0.30 μM) for three days. To inhibit L1 reverse transcriptase in cumulus cells, the collected cumulus cells were seeded into plates and then treated with 60 μM AZT (azidothymidine, Sigma‒Aldrich, A2169) for 4 days. The administration and doses of AZT used in this study were based on previous reports. 90 Moreover, we set up a series of concentration gradients and used immunofluorescence to further determine the appropriate concentration for detecting significantly reduced L1-ORF1 expression. After AZT treatment, cumulus cells were fixed and subjected to immunofluorescence and qRT‒PCR. The administration and doses of rapamycin were described in a previous article. 91 Rapamycin at 0.50 μM was added during IVM of the oocytes, and then, immunofluorescence and ROS detection were performed. Cumulus cells were seeded into plates and then treated with rapamycin (0.25 μM and 0.5 μM) for three days. Treating CCs with chaetocin (50 nM) for 24 h inhibited H3K9me3. 50 The cells were washed twice in PBS, lysed in cell lysis buffer on ice for 30 min and then sonicated for 1 min at an amplitude of 60 at 2 s intervals. After centrifugation at 10,000 × g for 10 min at 4°C, the supernatant was transferred into new tubes. The protein concentration of each sample was measured via bicinchoninic acid, and the protein samples were boiled in SDS sample buffer at 100°C for 10 min. The protein of each cell extract was resolved via 10% Acr-Bis SDS‒PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore). The membrane was blocked with 5% skim milk in TBST at room temperature for 2 h and then incubated with primary antibodies overnight at 4°C. β-actin served as a loading control. The immunoreactive bands were then probed for 2 h at RT with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies. The protein bands were detected with a chemiluminescent HRP substrate (WBKLS0500, Millipore). The antibodies used for western blotting were as follows: p -mTOR (5536T, Cell Signaling Technology), mTOR (2983S, Cell Signaling Technology), p-S6 (4858S, Cell Signaling Technology), S6 (2217S, Cell Signaling Technology), p-4E-BP1 (9451T, Cell Signaling Technology), 4E-BP1 (9644T, Cell Signaling Technology), LC3 (14600-1-AP, Proteintech) and β-actin (AC026, ABclonal). The senescence assay of cumulus cells was performed with a senescence-associated β-galactosidase staining kit (Beyotime, China, C0602) according to the manufacturer’s instructions and a previous method. 92 Briefly, cumulus cells were washed three times with PBS and then fixed with 4% paraformaldehyde for 15 min at room temperature. After being washed three times with PBS, the cells were incubated overnight at 37°C in darkness with working solution containing 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-gal). Detection of 18S and 28S abundance was conducted according to a previously published method. 38 Total RNA was extracted from cumulus cells via an RNeasy RNA Micro Kit (74034, QIAGEN) according to the manufacturer’s instructions. Reverse transcription was performed on purified total RNA to generate cDNA via M-MLV reverse transcriptase (Invitrogen) and random hexamer primers (18S and 28S rRNAs) or Oligo (dT)18 primers according to the manufacturer’s instructions. qPCR was performed with FastStart Universal SYBR Green Master Mix (4913914001, Roche) on an iCycler MyiQ2 detection system (Bio-Rad). Each sample was set up in duplicate and normalized to GAPDH. qPCR data were analyzed via the ΔΔCt method. qPCR primers ( Table S7 ) were confirmed for their specificity via dissociation curves, and primer design for L1 was performed via a previously published method. 90 The isolation of gDNA and mRNA from individual cells was performed as previously described, 93 and separated mRNAs and gDNA were used for transcriptome analysis and other experiments (such as telomere length measurement), respectively. The RNA-seq library was constructed according to the Smart-seq2 protocol. 28 Briefly, single oocyte or CC samples were quickly placed in lysis buffer, and then reverse transcription, template switching and preamplification were performed to obtain cDNA. Next, RNA-Seq libraries were constructed with a TruePrep DNA Library Prep Kit V2 for Illumina (TD503-02, Vazyme Biotech) according to the instruction manual. The quality of the cDNA library was checked via qPCR analysis of the housekeeping gene GAPDH . For the accuracy and repeatability of the RNA-seq data, we performed a duplicate when we constructed a library for every single oocyte, and two samples of CCs (from the same donor as the oocytes) were collected for RNA-seq library construction. The final indexed libraries were pooled and sequenced on an Illumina HiSeq X10 platform with a 150-bp paired-end read length. DNA methylation libraries of oocyte and CC samples were constructed according to a previously reported method with minor modifications. 94 For oocytes, 10 oocytes from 3 to 5 donors of similar age were collected in one tube as a sample. For cumulus cells, we collected CC samples from 6 donors of different ages and then performed DNA extraction in strict accordance with the instructions of the QIAamp DNA Micro Kit (56304, Qiagen). Detailed sample information, including age and sample size, is listed in Table S1 . Bisulfite conversion was performed on cell lysates with the following steps: incubation at 98°C for 10 min and 64°C for 120 min. DNA was eluted in 10 mM Tris-Cl (pH 8.5) and combined with 10 mM dNTPs, 5 μM BioPEA_N4_37 (5′-biotin-ACACTCTTTCCCTACACGACGCTCTTCCGATCTNNN N -3′), and 10x NEBuffer 2 (E7645S, NEB, Ipswich, England) before incubation at 95°C for 5 min, followed by a 4°C pause for 2 min. Then, 75 U of Klenow Fragment (M0212M, NEB) was added, and the samples were incubated at 4°C for 5 min, +1°C/15 s to 37°C, and 37°C for 30 min. Samples were incubated at 95°C for 1 min and transferred immediately to ice prior to the addition of fresh 1 mM dNTPs, 10 nM BioPEA_N4_37, 10x NEBuffer 2, and 75 U Klenow Fragment in a total volume of 2.75 μL. The samples were incubated at 4°C for 5 min and then at +1°C/15 s to 37°C for 30 min. This random priming and extension were repeated a further three times (five rounds in total). The samples were then incubated with 40 U of exonuclease I (M0293V, NEB) for 1 h at 37°C before the DNA was purified via 1× Agencourt Ampure XP beads (A63881, Beckman) according to the manufacturer’s guidelines. The samples were eluted in 10 mM Tris-Cl (pH 8.5) and incubated with washed M-280 streptavidin Dynabeads (65001, Life Technologies) for 30 min with rotation at room temperature. The beads were washed twice with 0.1 N NaOH and twice with 10 mM Tris-Cl (pH 8.5), and resuspended in 48 μL reaction mixture: 10 mM dNTPs, 10x NEBuffer 2, and 10 μM Primer 2.0 (5′-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTNNNN-3′). The samples were incubated at 95°C for 45 s and then transferred immediately to ice before the addition of 100 U of Klenow fragment and incubated at 4°C for 5 min, +1°C/15 s to 37°C, and 37°C for 90 min. Washed beads with 10 mM Tris-Cl (pH 8.5) and resuspended in 50 μL reaction mixture: 1 U of KAPA HiFi HotStart DNA Polymerase (KK2801, KAPA Biosystems), 10 μM Primer 1.0 (5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT-3’), and 10 μM Index. Libraries were then amplified via PCR as follows: 98°C for 45 s; eight repeats of (98°C for 15 s; 65°C for 30 s; 72°C for 30 s); 72°C for 1 min; and a hold at 4°C. Amplified libraries were purified via 0.8 × Agencourt Ampure XP beads. The samples were eluted in another 27 μL reaction mixture: 1 U KAPA HiFi HotStart DNA Polymerase, 10 μM Primer 1.0 forward primer, and 10 μM Index, and amplified by PCR for an additional eight repeats. Amplified libraries were assessed for quality and quantity via high-sensitivity DNA chips on an Agilent Bioanalyzer. The DNA methylation libraries were sequenced by Annoroad via a HiSeq X10 platform with a 150-bp paired-end read length. As described previously, 95 female CC samples were harvested and centrifuged for 3 min at 600 × g at RT. The cells were washed with 1 mL of wash buffer (20 mM HEPES pH 7.5, 150 mM NaCl, 0.5 mM spermidine, and 1 × protease inhibitor cocktail) by gentle pipetting and centrifuged for 3 min at 600 × g at RT. Concanavalin A-coated magnetic beads (Bangs Laboratories, BP531) were washed three times with binding buffer (20 mM HEPES pH 7.5, 10 mM KCl, 1 mM MnCl 2 and 1 mM CaCl 2 ) and then resuspended in the original volume of binding buffer. Ten microliters of activated beads were added per sample, incubated at RT for 10 min and collected with a magnet stand. The bead-bound cells and 2 μL of H3K9me3 antibody (Abcam, ab8898) were resuspended in 100 μL of antibody buffer containing 2 mM EDTA and 0.1% BSA in 2 mL of Dig-wash buffer (0.05% digitonin in Wash buffer) and rotated at RT for 2 h. The mixture was collected by a magnet stand, the supernatant was discarded, and the mixture was then incubated with a secondary antibody (1:100) diluted in Dig-wash buffer at RT for 1 h. The mixture was washed three times with Dig-wash buffer by using a magnet stand. The pG-Tn5 adaptor complex (∼0.04 μM) was prepared in Dig-300 buffer (20 mM HEPES pH 7.5, 300 mM NaCl, 0.5 mM spermidine, 1 × protease inhibitor cocktail and 0.05% digitonin) at a 1:100 dilution, and 100 μL was added to the cells with the liquid removed and gently rotated at RT for 1 h. The mixture was collected by a magnet stand and washed three times in Dig-300 buffer. Next, the cells were resuspended in 300 μL of tagmentation buffer (10 mM MgCl 2 in Dig-300 buffer), and the mixture was gently mixed and incubated at 37°C for 1 h. To stop the reaction, 10 μL of 0.5 M EDTA, 3 μL of 10% SDS, and 2.5 μL of 20 mg/mL proteinase K were added to the sample, and the mixture was incubated at 37°C overnight. The fragmented DNA in the mixture was extracted via a standard phenol‒chloroform extraction procedure. To amplify the libraries, 24 μL of DNA was mixed with 5 μL of ddH2O, 10 μL of 5 × TAB (TD503-02, Vazyme Biotech), 1 μL of TAE (TD503-02, Vazyme Biotech), and 5 μL each of uniquely barcoded P5 and P7 primers (TD204-207, Vazyme Biotech), and each sample contained a different barcode. The amplification reaction was performed with the following cycling conditions: 72°C for 3 min; 98°C for 30 s; 15 cycles of 98°C for 15 s, 60°C for 30 s, and 72°C for 3 min; a final extension at 72°C for 5 min; and a hold at 4°C. Purified DNA libraries were tested for high quality prior to high-throughput sequencing. The libraries were sequenced on an Illumina NovaSeq 6000 platform with a 150-bp paired-end read length by Novogene. We collected two batches of CC samples, and in each batch of experiments, the young and older samples were collected on the same day. However, owing to the limited number of samples, the number of cells in the two batches was not consistent (10,000 and 30,000 cells), but the number of cells in different age groups in the same batch was consistent (30 years versus 41 years & 31 years versus 39 years). The raw RNA-seq data with low-quality bases and adapters were trimmed via Trimmomatic to obtain clean reads. 70 Next, the trimmed clean reads were aligned to the UCSC human hg19 reference genome via HISAT2 with the default settings. 71 featureCounts was further used to calculate read counts for each annotated gene using the -M parameter. 72 The gene expression level in a sample was quantified as the transcripts per million (TPM), which was calculated according to the following formula: T P M i j = C i j / length of gene i ∑ i C i j / length of gene i × 10 6 , where C ij was the count value of gene i in sample j . For all sequenced cells, we counted the number of genes detected in each cell, and cells with fewer than 10,000 genes or 1,000,000 mapped reads were filtered out. After the critical filtering process, 48 sets of RNA-seq data from oocytes and 30 sets of RNA-seq data from CC samples were retained for downstream analysis. To ensure the accuracy of the gene expression levels, only genes with TPM >1 in at least ten oocyte or CC samples were analyzed. The total mapped reads and mapped ratios are shown ( Figures S1 C–S1E) and the median number of genes detected was 16,105 in each oocyte or 21,255 in CC samples. The read counts were loaded into RStudio (v4.1.0), and the critical filtering process was performed to filter out genes with low expression. Aging-specific genes were defined via correlation analysis. Next, we computed Pearson’s correlations between donor age and the expression levels of genes (TPM) via ‘cor’ in R language and identified aging-specific genes via Pearson’s coefficient (defined as the age coefficient). For oocytes, aging-specific genes were defined as those with an age coefficient threshold of 0.4 (≥0.4, upregulated genes with age) or −0.4 (≤−0.4, downregulated genes with age), after testing the significant differences of gene expression with age using various age-correlation coefficient cutoffs, based on the principle described. 29 , 30 For the CC samples, aging-specific genes were defined as those with an age coefficient threshold of 0.5 (≥0.5, upregulated genes with age) or −0.5 (≤−0.5, downregulated genes with age), after testing various cut-offs as above. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of aging-specific genes were performed via DAVID (v6.8), 73 and only enriched pathways whose p value was <0.05 were considered significantly enriched. The interactions between translation- and ribosome-related GO enrichments and related genes were generated via Cytoscape (v3.9.1). 74 GSEA was conducted to identify enrichments associated with aging in oocytes and cumulus cells, and only gene sets with an FDR <0.05 were considered significantly enriched. 75 For transposable element analysis, clean reads were aligned to the UCSC human hg19 reference genome by STAR with the parameters ‘-winAnchorMultimapNmax 100’ and ‘-outFilterMultimapNmax 100’. 96 Referring to a previously published study, 97 only the TEs whose distributions in intergenic regions were mapped were considered, excluding the locations between the transcription start sites and transcription end sites of genes. TEs annotated via the UCSC Genome Browser (RepeatMasker) were counted via featureCounts. The median mapped rates were 11.8% for the oocytes and 16.8% for the CC samples, and the median mapped reads were 612,781 for the oocytes and 1,905,441 for the CC samples. TE expression was evaluated as counts per million (CPM). To ensure the accuracy of TE expression, only TEs with CPM >1 in at least ten oocytes and CPM >2 in at least ten CC samples were analyzed. After critical filtering to remove TEs with low expression, aging-specific TEs were identified via correlation analysis. Specifically, Pearson’s correlations between donor age and the expression levels of TEs (CPMs) were computed via ‘cor’ in R language, and then, aging-specific TEs were determined via Pearson’s coefficient (defined as the age coefficient) with a threshold of 0.4. Age coefficients ≥0.4 were defined as TEs whose expression increased with age, whereas those ≤ −0.4 were defined as TEs whose expression decreased with age. Adaptors and low-quality bases of bisulfite sequencing reads were first trimmed by Trimmomatic ( http://www.usadellab.org/cms/index.php?page=trimmomatic ) with default parameters. Next, reads that passed quality control were mapped to the human reference genome (hg19) via Bismark (version 0.19.1) 76 in paired-end alignment mode. Only reads with a unique mapping location in the genome were retained for further analysis. After alignment, the reads were further deduplicated via Picard ( http://broadinstitute.github.io/picard/ ). The bisulfite conversion rate was estimated by the spike-in of unmethylated lambda DNA. Methylation calls were extracted via the Bismark methylation extractor. The annotations of exons, introns, CGIs, TSSs and transcription end sites (TESs) were downloaded from the UCSC Genome Browser (hg19). All repetitive element annotations were downloaded from RepeatMasker (hg19) via the UCSC Genome Browser. Promoter regions were defined as the 1 kb upstream to 1 kb downstream of transcription start sites (TSSs). 98 The methylation levels of the CpGs were quantified according to previously published methods. 99 For each CpG, the DNA methylation level was determined by the ratio of the number of reads supporting C (methylated) to the total number of reads (methylated and unmethylated). Metaplots of CpG methylation levels were generated by calculating the degree of methylation within each RefSeq gene and ten 1 kb windows of flanking sequences. Initially, the human genome was divided into 1 kb tiles, and tiles with at least 3 CpGs covered by both age groups in a comparison were considered the background to find DMRs. For tiles that passed the criteria, the methylation level of each retained tile was determined as the ratio of the number of alignments with C (methylated) to the sum of the alignments with C and T for all the CpGs in the tile. When identifying DMRs in oocyte or CC samples of different age groups, we used the calculateDiffMeth function from the methylKit R package with the following criteria: difference in methylation >20% and adjusted p value <0.05, which were adjusted via the SLIM method. 77 , 98 , 100 Hierarchical clustering via Euclidean distance metrics derived from methylation levels per DMR for each age group was conducted via the clusterSamples function from the methylKit R package with the ‘ward’ agglomeration method. The distance of DMRs from the TSS was calculated on the basis of a previously published method. 101 Briefly, the distance was reported by subtracting the mean DMR genomic localization from the TSS positions, grouping the distances into 500-bp clusters from position −10 kb to position +10 kb, and quantifying the DMRs in each cluster. DMRs with a distance of more than 10 kb from the respective TSS were excluded and not counted. Enrichment analysis of genes with differentially methylated regions was performed via DAVID 6.8 ( https://david.ncifcrf.gov ). Enriched pathways with a p value less than 0.05 were considered statistically significant. The differentially methylated regions used for enrichment analysis contained only gene bodies, promoters and CGIs, and regions located on intergenic and transposable elements were excluded. To correlate the epigenetic and transcriptome datasets, DMRs were annotated to the nearest transcription start site via the annotatePeaks.pl function in HOMER (v4.11.1). 102 Only DMRs located in the gene body, promoter, and CGI regions were used for association analysis. DESeq2 78 was used to obtain the statistical significance of the DEGs of different groups, and only the genes with a fold change greater than 1.2 and adjusted p value <0.05 from the DEseq2 results were considered to be differentially expressed. Adjusted p values were computed in DESeq2 via the Wald test and adjusted for multiple testing via the procedure of Benjamini and Hochberg. 103 The raw data were processed according to the following pipeline. The sequencing adapters were trimmed, and read pairs with low quality or low complexity were filtered from the raw data through TrimGalore (v0.5.0). The trimmed read pairs were aligned to the human reference genome hg19 via Bowtie2 (v2.3.4) 79 with default parameters. For peak calling, the parameters used were macs2 callpeak “-q 0.05 -g hs –keep-dup all”. 80 , 95 For annotation and enrichment analysis of the genomic regions, we referred to previously published methods. 104 First, the ‘‘annotatePeak’’ function in the R package ChIPseeker (v1.28.3) 81 was utilized to profile the distribution of genomic regions. Second, the functional enrichment of genes that are marked by or near H3K9me3 peaks was analyzed via Genomic Regions Enrichment of Annotations Tool (GREAT) analysis with default settings via the R package rGREAT (v1.24.0). 105 We used the ‘macs2 bdgdiff’ function to identify young-specific and old-specific H3K9me3 peaks. We utilized the ‘‘intersect’’ subcommand in bedtools (v2.27.1) to identify peaks related to ribosome- and lysosome-related genes, including promoters and gene bodies, and these ribosome- and lysosome-related genes were defined via RNA-seq data via correlation analysis. For data visualization, bigwig files were generated via bamCoverage and merged via bigWigMerge, and H3K9me3 modification signals around peaks were visualized via deepTools (v3.2.0). 82 Promoters were defined as the regions 1 kb upstream and 1 kb downstream of TSSs. Gene bodies were defined as the regions from the TSSs to the TESs. Statistical analyses were performed via an unpaired two-tailed Student’s t test in PRISM software (GraphPad 8 Software) to compare the differences between the treatment and control groups, assuming equal variance. The Mann‒Whitney test was used to determine the significance of differences between data without a normal distribution. One-way or two-way ANOVA with Tukey’s test was used for multiple comparisons. The chi-square test was used to test differences between two groups for categorical variables. ∗ , ∗∗ , ∗∗∗ and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001 and p < 0.0001, respectively. NS indicates not significant. The bioinformatics data were statistically analyzed via a two-tailed t test with R software, and p values are indicated in each figure. Correlations between clinical indicators and age and the corresponding statistical significance were calculated on the basis of Pearson’s rank correlation coefficient (r). For correlation analysis between the age and gene expression levels, p values were adjusted using the Benjamini–Hochberg (BH) method. Selection criteria for age-correlation coefficient cut-offs included, (1) statistically significant correlations (BH-adjusted p < 0.05), and (2) at least moderate correlation between age and gene expression levels (|r| ≥ 0.4). 29 , 30 For the clinical trials of rapamycin therapy, categorical data are presented as frequencies and percentages, and between-group comparisons were conducted with the chi-square test. Continuous data are expressed as the means (±SD) or medians (Q1, Q3), with a Wilcoxon rank-sum test for between-group differences. The detailed clinical information of the control group and the rapamycin group is presented in Tables S5 and S6 . The study has been registered on the Chinese Clinical Trial Registry ( https://www.chictr.org.cn/ ) with registration number (ChiCTR2300069828).

This work was supported by the China National Key R&D Program ( 2022YFA1103800 ), the 10.13039/501100001809 National Natural Science Foundation of China ( 91749129 , 32030033 , 82230052 , and 31970667 ), and the Innovation Plan of Medical Science and Technology of “ Four in a batch ” ( 2020TD19 ). We thank Mo Gou for assisting with the MethylC-Seq library construction experiments, Jiao Yang for helping with some of the experiments, and Zhengmao Zhu for helping with statistics analysis.