Serine inhibits granulosa cell ferroptosis to maintain ovarian function

Abstract

Cyclophosphamide (CTX) is a primary medicine for curing breast cancer which often causes premature ovarian insufficiency (POI). Our recent publication reveals that CTX induces POI by promoting the expression of SLC1A4, a transporter of serine efflux, in ovarian granulosa cells (GCs). Here, we report that there is a closed connection between the reduction of serum serine and ovarian hypofunction in the breast cancer patients treated with CTX or women of childbearing age who are suffered from the staying-up-late. Additionally, we observe that dietary serine supplementation protects mice from CTX-induced POI without altering its anti-breast cancer. Furthermore, we demonstrate that the elevated serine promotes S1P synthesis, and in turn, inhibits the nuclear translocation of Nrf2 and consequent HO-1 expression, to suppress ferroptosis in GCs. Our study reveals that the chemotherapy-induced or idiopathic POI share the same mechanisms, indicating that serine is a critical factor for maintaining ovarian function.

This study demonstrates that the reduced serum serine in patients with chemotherapy or women with poor sleep, is linked to the decline of ovarian function. Serine supplementation prevented ovarian damage in mice by inhibiting ferroptosis in ovarian granulosa cells, without affecting chemotherapy efficacy.

Introduction

Premature ovarian insufficiency (POI)/Premature ovarian failure (POF) has emerged as a significant global concern, adversely affecting family well-being and social harmony¹. According to the latest clinical guidelines, POI can broadly replace POF. Therefore, in this text, we will use POI instead of POF to demonstrate our respect for the patient. The pathogenesis of POI remains unclear, leading to a lack of effective clinical treatments to mitigate the ovarian hypofunction. Studies indicated that chemotherapy led to an incidence of appropriate 40% of POI in the patients with breast cancer, which accounts for ~5–10% of total patients with diminished ovarian function^2–5. However, the causes of the majority of patients with the idiopathic ovarian insufficiency were undetermined^6–9.

Cyclophosphamide (CTX) is a primary drug used in the treatment of breast cancer, favored for its cost-effectiveness and potent inhibitory effects. However, CTX is associated with significant reproductive toxicity, which adversely affects the reproductive health of patients undergoing chemotherapy for breast cancer^10–12. More than half of breast cancer patients experience adverse reactions such as menstrual delays or even amenorrhea following CTX treatment, with some facing severe complications, including difficulties in conception^13–15. Notably, with the increasing incidence and the trend of younger onset of breast cancer, a growing number of young patients were developing the POI after CTX chemotherapy^16–18. Meanwhile, a recent survey revealed that 50% of young female patients wish to preserve their fertility with post-chemotherapy^3,8,19,20. Therefore, it is crucial to protect ovarian function during CTX treatment in patients with breast cancer²¹. Moreover, to elucidate the mechanism and the etiology of POI induced by CTX or idiopathic POI is an urgent need for developing effective preventive and therapeutic strategies to maintain ovarian function²². Our recent study demonstrated that human urine stem cells (hUSC)-derived miRNA-27-3p selectively suppressing CTX-induced elevated expressions of SLC1A4, a serine transporter for excreting the intracellular serine, in ovarian GCs to alleviate CTX-induced POI²³. Inhibiting the expression of SLC1A4 in ovarian granulosa cells (GCs) via siRNA can directly suppress CTX-induced germ cell death. This finding indicates that the elevated expression of SLC1A4 induced by CTX is crucial for the development of POI. Building on this finding, we were eager to know whether there was a connection between the reduction of serum serine and the ovarian hypofunction in breast cancer patients treated with CTX or women of childbearing age who suffered from idiopathic POI, and whether the CTX-induced POI or idiopathic POI could be prevented and cured by dietary serine supplementation.

Serine is a conditional non-essential amino acid that functions as a vital precursor in the synthesis of proteins, phospholipids, neurotransmitters, and nucleotides. It has been reported that serine plays a central role in various physiological processes associated with multiple disease^24–28. Abnormalities of serine metabolism have been linked to neurodegenerative diseases, cardiovascular disorders, immune dysfunction, and cancer^29–34. However, the roles and the underlying mechanism of serine in female reproductive function have not been explored³⁵.

In the present study, we observed that CTX significantly reduced the levels of serum Estrogen (E2), anti-Mullerian hormone (AMH), and serine in patients with breast cancer, and there was a close connection between the reduction of serum serine and ovarian hypofunction in the women of childbearing age who suffered from staying up late. In addition, we observed that dietary serine supplementation could effectively prevent CTX-induced POI in mice. Furthermore, we demonstrated that the elevated serine promoted S1P synthesis, and in turn, inhibited the nuclear translocation of Nrf2 and consequent HO-1 expression, to suppress ferroptosis in GCs. Moreover, our study further revealed that the idiopathic and chemotherapy-induced POI shared the same mechanism, suggesting that serine is a critical factor for maintaining ovarian function. Recently, we have initiated a relevant clinical trial (ChiCTR2400079348) to evaluate the protective effects of serine against CTX-induced POI. Certainly, our study should provide insight into elucidating the mechanism of chemotherapy-induced POI or idiopathic POI, and the dietary serine supplementation might be a simple therapeutic strategy for preventing and curing ovarian hypofunction clinically.

Results

CTX reduced serum serine concentration in breast cancer patients

Our previous study has demonstrated that CTX could lead to the overexpression of SLC1A4, which subsequently results in a decrease in intracellular serine levels. To investigate the potential link between the CTX-induced reduction in serine and ovarian hypofunction, we recruited 27 breast cancer patients aged 25 and 45 years. Serum samples were collected from each patient prior to the initiation of CTX chemotherapy, and a second serum samples were obtained on day 21 after the completion of the first cycle of CTX treatment. The related parameters of ovarian function and serum serine were subsequently measured (Fig. 1a), and the results showed that the serum E₂ and AMH levels were significantly reduced in patients treated with CTX, indicating that there was a decline in ovarian function in patients who were even treated with a single course of CTX chemotherapy (Fig. 1b, c, Fig. S1a, Supplementary Table 1, and Supplementary Data 1). Surprisingly, the results showed that a single injection of CTX resulted in signs of ovarian hypofunction in 25 out of 27 patients, which was remarkably higher than the incidence of 40% in the previous reports^36,37. The results of amino acid-targeted metabolomics showed that there were significant alterations in serum amino acid levels in patients after CTX chemotherapy, and the accuracy of these findings was validated through analyses based on PCA, PLS-DA, and OPLS-DA (Fig. S1b–f). Furthermore, the differential amino acids analysis revealed that nine amino acids including serine were significantly altered after CTX chemotherapy, in which the reduction of serum serine was particularly pronounced (Fig. 1d and Fig. S1g, h). Moreover, in a total of 27 patients, 25 exhibited varying degrees of reduction in E₂ and AMH levels, which were accompanied by a concurrent decline in their serine levels (Fig. 1e and Fig. S1a). The results clearly demonstrated that the ovarian hypofunction induced by CTX was likely closely associated with the reduction in serum serine levels.

Fig. 1. CTX reduces serum serine concentration in breast cancer patients.

a Flowchart of serum collection and chemotherapy for breast cancer patients (n = 27). b E₂ test results in breast cancer patients before and after chemotherapy (n = 27, unpaired two-tailed t test). c AMH test results in breast cancer patients before and after chemotherapy (n = 27, unpaired two-tailed t test). d Serine test results in breast cancer patients before and after chemotherapy (n = 27, unpaired two-tailed t test). e Analysis of E₂ and serine trends in 27 breast cancer patients before and after chemotherapy. Graphs indicate mean ± SEM. n indicates the number of patients. *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Panel a was created with BioRender (ML2959RHUA). Created in BioRender. Gu, h. (2025) https://BioRender.com/a9q2cc1.

Idiopathic ovarian hypofunction was also closely related to the reduction of serum serine levels

In comparison to CTX-induced POI, a greater number of female patients who came to the hospital were diagnosed with idiopathic POI. Consequently, we further investigated the correlation between serum serine levels and idiopathic POI, while excluding cases of iatrogenic POI. To determine whether serine has a similar effect on POI beyond CTX chemotherapy, 124 young females with ovarian hypofunction or infertility were enrolled in our study, in which the patients were classified into two categories based on their AMH test results (Fig. 2a and Supplementary Table 2). The results showed that the serum AMH levels were remarkably reduced in the patients with POI, whereas there was an increase in serum Follicle-stimulating hormone (FSH), accompanying with elevated E2 level (Fig. 2b). In addition, the numbers of bilateral antral follicles in the ovaries of the patients with POI were fewer than 5, much less than that of patients with normal ovarian function (Fig. 2c and Fig. S2a, b). Furthermore, we randomly selected 30 serum samples from each group to conduct the amino acid metabolome assay, and the results showed that serum serine levels were generally reduced in the patients with low ovarian function (Fig. 2d and Supplementary Data 2), indicating that the reduced serum serine levels were closely correlated with the decreased AMH levels in patients with POI (Fig. 2e). Interestingly, by establishing a cutoff for serum serine at 70 μg/mL, we discovered that the POI patients with serum serine levels exceeding this threshold had significantly higher AMH levels than those with levels below 70 μg/mL (Fig. 2f). These results indicated that there was a close correlation between idiopathic POI and serum serine levels, suggesting that serine may play a critical role in both chemotherapy-induced POI and idiopathic POI.

Fig. 2. The reduced serum serine levels are closely correlated with the ovarian hypofunction of the patients.

a Flowchart of clinical research on serine and ovarian function. b Serum AMH, E2, FSH, and LH in patients (n = 50–71, unpaired two-tailed t test). c The comparison of B-ultrasound images of the ovaries in two groups. d Serum serine levels in people with normal ovarian function or with low ovarian function (n = 30, unpaired two-tailed t test). e Analysis of the correlation between serum serine and AMH in samples from people with low ovarian function (n = 27, Pearson correlation analysis). f, AMH levels in people with low ovarian function (n = 12–15, unpaired two-tailed t test). Graphs indicate mean ± SEM. n indicates the number of patients. ^nsP > 0.05, *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Panel a was created with BioRender (AD2959S608). Created in BioRender. Gu, h. (2025) https://BioRender.com/d47j7w9.

Dietary serine supplementation prevented CTX-induced POI without altering its efficacy in suppressing breast cancer

In order to further confirm our findings that CTX-induced POI was related to the reduction of the intracellular serine²³, L-serine was administered to mice through gavage or drinking water supplementation (Fig. 3a and Fig. S4a). The results showed that regardless of the forms of serine supplementation (intragastrical or drinking water administration), serine effectively preserved ovarian morphology, evidenced by restoring the size and organ rate of ovaries during CTX chemotherapy (Fig. 3b, c and Figs. S3a and S4b–d). HE staining results showed that serine also markedly prevented CTX-induced decrease of follicles and the increase in atresia. In addition, serine prevented CTX-induced alteration of ovarian structure and the reduction of ovarian surface area (Fig. 3d and Figs. S3b, c and S4e–h), and corrected CTX-induced abnormal secretions of E2, follicle-stimulating hormone (FSH), and anti-Müllerian hormone (AMH) (Fig. 3e–g and Fig. S4i, j). In the ovary, two crucial types of functional cells are present: GCs and oocytes. GCs play a vital role in the growth and development of oocytes by secreting associated hormones^38–40. Our results showed that serine supplementation effectively prevented CTX-induced death of GCs, enhanced their proliferation, and increased the expression of AMH (Fig. 3h, i and Fig. S3d, e), indicating that serine supplementation protected mice from CTX-induced POI through preventing the death of GCs, enhancing GCs proliferation, and maintaining homeostasis of sex hormone levels.

Fig. 3. Dietary serine supplementation inhibits CTX-induced POI without altering its efficacy against anti-cancers.

a Schematic diagram of animal experiments on supplementing serine to inhibit CTX-induced ovarian POI. b Representative mouse ovaries of each group at the time of sample harvesting (n = 6). c The weights of mouse ovaries in different groups were measured and calculated for ovarian coefficient (n = 4–8, one-way ANOVA with Dunnett’s multiple comparisons test). d HE staining showed the number of follicles and ovarian structure (n = 3–6). e Serum E2 levels assayed using ELISA at the time of final tissue collection (n = 4–12, one-way ANOVA with Dunnett’s multiple comparisons test). f Serum AMH levels assayed using ELISA at the time of final tissue collection (n = 4–12, one-way ANOVA with Dunnett’s multiple comparisons test). g Serum FSH levels assayed using ELISA at the time of final tissue collection (n = 4–8, one-way ANOVA with Dunnett’s multiple comparisons test). h, i The expression levels of FSHR and AMH were detected by western blot in the ovarian tissues of each group. β-actin was used as a housekeeping protein (n = 4, one-way ANOVA with Dunnett’s multiple comparisons test). j Breast cancer animal experiment diagram. k The mouse tumor volume was measured every 3 days (n = 4–8). l Representative tumor tissue from mice at the time of sample collection (n = 3–5). m HE staining results of tumor tissue (n = 5). n Serum estrogen levels (n = 3–7, one-way ANOVA with Dunnett’s multiple comparisons test). o HE staining results of mouse ovaries of breast cancer models (n = 5). p Immunofluorescence images for detecting the effects of CTX and serine on ovarian cell proliferation (n = 4). Data are presented as mean ± SEM. n indicates biological replicates (h, i) and independent animals (b–g, k–p). ^nsP > 0.05, *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Scale bars, 100 μm (m—4×, p—10×); 200 μm (d—10×, o—10×); 400 μm (d—4×, o—4×). Panels a, j were created with BioRender (NA2959U16I). Created in BioRender. Gu, h. (2025) https://BioRender.com/3s11e12.

Next, we clarified whether the protective effect of dietary serine supplementation on CTX-induced POI was able to reduce its anti-cancer effect in breast cancer. Given that the majority of patients with breast cancer predominantly exhibited POI during chemotherapy, and considering that there were differences in the internal microenvironment induced by breast cancer and the normal internal microenvironment, we simulated the dosage and duration of chemotherapy to establish a breast cancer mouse model of POI^41–43. Briefly, 1 × 10⁶ 4T1 cells were transplanted into the mouse second breast pad to generate breast cancer model, in which CTX (20 mg/kg) was administrated by injection with twice each week from third to 7th week, and serine (50 and 500 mg/kg) was supplemented by gavage with one time each week from second week to 7th week (Fig. 3j), and throughout the study, we closely tracked changes in tumor volume. Our results showed that serine did not affect the efficacy of CTX in breast cancer (Fig. 3k–m and Fig. S5a). In contrast, serine significantly improves CTX-induced reduction in the size, organ rate, and surface area of the ovaries (Fig. S5b–g). In addition, our study also revealed that serine supplementation remarkably ameliorated the dysregulation of sex hormone secretion and promoted the development of antral follicles in the ovary, as well as the proliferation of GCs (Fig. 3n–p and Fig S5h). Furthermore, three different tumor cell lines were selected for further validating whether serine affects the anti-cancer effect of CTX in vitro, and results showed that serine did not alter the therapeutic effect of CTX on multiple tumor cells (Fig. S5i–k). These results demonstrated that the protective effect of serine supplementation on CTX-induced POI was not related to its anti-cancer effect in the mouse model.

CTX induced ferroptosis in GCs but not in breast cancer cells

Next, we focused on the protective mechanism of serine in CTX-induced POI. Previous studies suggested that CTX might trigger POI by excessively activating primordial follicles^8,44,45. Additionally, other researchers observed that CTX induced POI by promoting ROS production or impairing mitochondrial function^46,47. Furthermore, some studies have reported that CTX induces apoptosis in GCs, which leads to a decline in ovarian function^48–50. Our recent report identified GCs as the primary target cells of CTX in the ovary²³. Subsequently, we conducted transcriptome sequencing on GCs treated with CTX. The sequencing results showed that there were significant alterations in gene expression in GCs after CTX treatment (Fig. 4a and Fig. S6a). We conducted a heatmap analysis of the top 50 upregulated and downregulated genes in GCs following treatment with CTX (Fig. 4b, Fig. S6b, and Supplementary Tables 3 and 4). GSEA and KEGG enrichment analyses of the upregulated genes revealed significant enrichment in the ferroptosis signaling pathway (Fig. 4c, d and Fig. S6c). A Venn diagram comparison between the upregulated differentially expressed genes and those in the ferroptosis database (FerrDb V2) identified four overlapping genes that were notably ranked among the upregulated genes (Fig. 4e). The qPCR results and the enrichment map of the ferroptosis signaling pathway indicated that CTX primarily induces the upregulation of HO-1, leading to the accumulation of ferrous ions (Fig. 4f, g). Furthermore, CTX did not induce any changes in GPX4 expression (Fig. S6d). We also observed that the elevated expression levels of nuclear receptor coactivator 4 (NCOA4), zinc transporter 14 (ZIP14), and zinc transporter 8 (ZIP8) may contribute to iron accumulation, although their upregulation was not as pronounced as that of Hmox1 (Fig. 4h). Therefore, we proposed that CTX-induced POI might not be directly related to the apoptosis or oxidative stress of GCs, but rather that it was caused by the accumulation of ferrous ions in GCs mediated by HO-1, which subsequently triggers ferroptosis.

Fig. 4. Transcriptome sequencing reveals CTX-induced ferroptosis in GCs.

a Correlation analysis of GCs samples before and after CTX treatment (n = 3). b Volcano plot sum and statistics of differential genes in GCs before and after CTX treatment. The x axis represents log2 fold change in gene abundance, and the y axis represents the negative log10 adjusted (Benjamini–Hochberg) P value. Heatmap of differential genes in GCs before and after CTX use. Hierarchical clustering was performed for the differentially enriched proteins across both conditions (n = 3, Normal and CTX, Benjamini–Hochberg). c GSEA was used to analyze the impact of CTX on the enrichment of GCs genes in ferroptosis-related pathways (n = 3, Normal and CTX). d KEGG analysis showed that differential genes were mainly enriched in the ferroptosis signaling pathway (hypergeometric test). e The intersection of upregulated genes in GCs after CTX treatment with genes in the ferroptosis database analyzed by Veen. f mRNA expressions of four ferroptosis-related genes (n = 5–6, unpaired two-tailed t test). g KEGG analysis of the significance of related differential genes in ferroptosis-related signaling pathways. h mRNA expressions of NCOA4, ZIP14 and ZIP8 before and after CTX use (n = 3–6, unpaired two-tailed t test). Graphs indicate mean ± SEM. n indicates biological replicates (a–h). *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Panels a, e, g were created with BioRender. Created in BioRender (NA2959U16I). Gu, h. (2025) https://BioRender.com/3s11e12.

Therefore, we investigated whether serine supplementation could inhibit HO-1-induced ferroptosis in the mouse POI model. Our results showed that serine suppressed CTX-induced the accumulation of the intraovarian ferric ion and ferrous ion, lipid peroxidation in a dose-dependent manner (Fig. 5a–d and Fig. S4l, m), and inhibited the CTX-activated Keap1/Nrf2/HO-1 signaling pathway, which in turn, reduced the expression of HO-1 (Fig. 5e, f and Fig. S4m). Importantly, the accumulation of MDA and ferrous ions induced by CTX was not observed in the tumor tissues of the mouse POI model with breast cancer (Fig. 5f–h). In addition, the expressions of the ferroptosis-related genes in breast cancer tissues exhibited no significant changes following the administration of CTX and the oral gavage of serine (Fig. 5i). However, the accumulation of ferric ion and ferrous ions was observed in the ovarian tissues of breast cancer model mice with POI, suggesting that CTX-induced POI could be significantly improved upon supplementation with serine (Fig. S6e). Our results indicated that the mechanisms through which CTX affects ovarian and breast cancer tissues were different, and serine supplementation should safeguard ovarian function without reducing the effectiveness of chemotherapy.

Fig. 5. Serine inhibits ovarian ferroptosis without affecting the chemotherapeutic efficacy of CTX on tumors.

a, b Prussian blue staining was used to detect the accumulation of ferrous ions in each group. The blue color after dyeing represents the accumulation of ferrous ions (n = 2–5, one-way ANOVA with Dunnett’s multiple comparisons test). c Determination of ferrous ion concentration in mouse ovarian tissues (n = 3–9, one-way ANOVA with Dunnett’s multiple comparisons test). d Determination of MDA concentration in ovarian tissue (n = 3–7, one-way ANOVA with Dunnett’s multiple comparisons test). e The expressions of the genes related to the Keap1/Nrf2/HO-1 signaling pathway were detected by western blot. β-actin was used as a housekeeping protein (n = 3–5). f Prussian Blue Staining and Semi-quantitative Analysis (n = 3–4, one-way ANOVA with Dunnett’s multiple comparisons test). g Determination of ferrous ion concentration in mouse tumor tissues (n = 5–7, one-way ANOVA with Dunnett’s multiple comparisons test). h Determination of MDA concentration in tumor tissue (n = 5–10, one-way ANOVA with Dunnett’s multiple comparisons test). i The expressions of the ferroptosis-related genes were determined by qPCR (n = 4–8, one-way ANOVA with Dunnett’s multiple comparisons test). Graphs indicate mean ± SEM. n indicates biological replicates (e, f) and independent animals (b–d, g–i). ^nsP > 0.05, *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Scale bars, 200 μm (a—10× and f—10×); 400 μm (a—4× and f—4×). Panels a, f, g were created with BioRender (QY2959V5EU). Created in BioRender. Gu, h. (2025) https://BioRender.com/m7bqasf.

Serine elevated the synthesis of S1P to inhibit ferroptosis

To further elucidate the precise molecular mechanism by which serine inhibits ferroptosis in GCs, mouse GCs were isolated following established protocols. Our in vivo results showed that serine effectively hindered the buildup of ferrous ions. To validate this observation, we employed the ferrous ion chelator deferoxamine (DFO) and the agonist Erastin as experimental controls (Fig. 6a). Our results showed that DFO or serine had a notable inhibitory effect on CTX-induced cell death in GCs, whereas Erastin exacerbated this alteration (Fig. 6b and Fig. S6f, g). Furthermore, DFO or serine effectively reduced the level of lipid peroxidation induced by CTX, while Erastin enhanced lipid peroxidation in GC (Fig. 6c). In addition, DFO or serine inhibited the accumulation of ferrous ions in GCs (Fig. 6d). Mitochondria play a crucial role in the process of ferroptosis, prompting us to investigate the morphological structure and membrane potential of mitochondria^51–54. Our results revealed that CTX or Erastin induced mitochondrial atrophy, increased membrane area, reduced ridges, and decreased membrane potential, whereas DFO or serine effectively mitigated the mitochondrial damage caused by CTX (Fig. 6e, f). Ferroptosis was characterized by the generation of ROS, and both DFO and serine notably suppressed ROS production in GCs (Fig. 6g). Furthermore, our results revealed that serine was able to counteract the elevated expressions of Nrf2 and HO-1 induced by CTX (Fig. 6h, i), and downregulated the expressions of NCOA4, ZIP14, and ZIP8, consequently decreased the accumulation of ferrous ions (Fig. S6h). Overall, our results demonstrated that serine attenuated CTX-induced iron overload and inhibited ferroptosis by inhibiting the Keap1/Nrf2/HO-1 signaling pathway.

Fig. 6. Serine-mediated elevated synthesis of S1P inhibits ferroptosis by suppressing the Keap1/Nrf2/HO-1 signaling pathway in GCs.

a Schematic of in vitro serine-mediated ferroptosis inhibition. b Morphological alterations of GCs (n = 3). c Detection of lipid peroxidation levels in GCs (n = 3). d Levels of ferrous ion accumulation within GCs (n = 3). e Transmission electron microscopy (TEM) of GCs mitochondrial morphology (n = 3). f Mitochondrial membrane potential changes in GCs under combined treatments (n = 3). g Immunofluorescence detection of ROS accumulation in GCs (n = 3). h, i The expressions of Nrf2, HO-1 and Nox4 were detected by western blot (n = 4, one-way ANOVA with Dunnett’s multiple comparisons test). j Schematic of S1P regulating the Keap1/Nrf2 signaling pathway. k Lipid peroxidation in GCs after SKI178 treatment (n = 3). l Ferrous ion levels after using SKI178 (n = 3). m TEM of mitochondrial morphology after SKI178 treatment (n = 3). n ROS levels in GCs after S1P synthesis were blocked (n = 3). o–q The expressions of Keap1, Nrf2, HO-1, and SLC1A4 were determined by western blot with the treatment of serine after S1P synthesis was blocked (n = 3, one-way ANOVA with Dunnett’s multiple comparisons test). r Immunofluorescence for Nrf2 nuclear import ratio (n = 3–4, one-way ANOVA with Dunnett’s multiple comparisons test). s Schematic of HO-1 degrading heme to produce ferrous ions. t Lipid peroxidation levels in GCs after Znpp treatment (n = 3). u Ferrous ion accumulation after Znpp treatment (n = 3). v TEM of mitochondrial morphology after Znpp treatment (n = 3). w Mitochondrial membrane potential changes under the combined effects of Znpp with CTX (n = 3). x ROS levels in GCs after Znpp used (n = 3). Graphs indicate mean ± SEM. n indicates biological replicates (a–w). ^nsP > 0.05, *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Scale bars, 500 nm (e, m, v); 20  μm (d, l, t, r, u); 100 μm (c, f, k, t, w, x); 200 μm (b, g, n). Panels a, j, s were created with BioRender (QY2959V5EU). Created in BioRender. Gu, h. (2025) https://BioRender.com/m7bqasf.

Previous studies have demonstrated that serine reacted with palmitoyl-CoA in cells, leading to the production of sphingosine through a series of reactions^30,55. Sphingosine is converted into S1P by sphingosine kinase (Sphk), and S1P might be involved in crucial processes such as ovarian growth, development, and maturation^56,57. Based on these observations, we hypothesized that serine exerts its effects through the action of S1P. To test this hypothesis, we employed the Sphk inhibitor SKI178 to suppress S1P production (Fig. 6j), and the results showed that CTX or SKI178 led to a notable reduction in S1P levels in GCs compared with the untreated GCs (Fig. S6i). Our study further revealed that SKI178 reversed the inhibitory effects of serine on CTX-induced cell death in GCs, while promoted lipid peroxidation and the accumulation of ferrous ions (Fig. 6k, l and Fig. S6j). In addition, SKI178 also reversed the restorative effect of serine on mitochondrial morphology, structure, and membrane potential and ROS generation (Fig. 6m, m and Fig. S6k). These results demonstrated that serine hindered the CTX-induced buildup of ferrous ions in GCs via S1P. Our studies further demonstrated that when S1P synthesis was inhibited, the expressions of Nrf2 and HO-1 were elevated, resulting in the accumulation of ferrous ions. Moreover, the expression of SLC1A4 was also significantly increased (Fig. 6o–q). Notably, the confocal microscopy results indicated that CTX promoted the nuclear translocation of Nrf2, leading to an increase in the expression of SLC1A4 and HO-1. Conversely, serine supplementation enhanced the production of S1P, which, in turn, reduced the nuclear import of Nrf2 and subsequently decreased the expression of HO-1 and SLC1A4 (Fig. 6r and Fig. S6l). To further clarify that serine inhibits GCs ferroptosis by converting to S1P, we conducted an S1P rescue experiment and measured key indicators of ferroptosis induced by ferrous ion accumulation. The results indicated that S1P significantly suppressed CTX-induced ferrous ion accumulation, lipid peroxidation, and ROS accumulation, effectively inhibiting GCs ferroptosis (Fig. S6m–o).

To validate that the accumulation of ferrous ions is attributed to HO-1 and to verify the regulation of the Keap1/Nrf2/HO-1 signaling pathway on ferroptosis, we employed the HO-1 activity inhibitor Znpp to inhibit HO-1 activity (Fig. 6s and Fig. S6p, q). The results showed that the HO-1 inhibitor significantly reduced CTX-induced GC death, as well as a decrease in lipid peroxidation and ferrous ion accumulation (Fig. 6t, u and Fig. S6r). In addition, our results also showed that Znpp led to a significant improvement in abnormalities in mitochondrial morphology, structure, and membrane potential induced by CTX (Fig. 6v, w), and suppressed CTX-induced production of ROS in GCs (Fig. 6x). Collectively, our data demonstrated that serine supplementation alleviated ferroptosis in GCs by inhibiting the accumulation of ferrous ions during chemotherapy, shedding light on the underlying mechanism of CTX-induced POI.

Serine deficiency induced by staying up late was the main cause of idiopathic POI in women

Although chemotherapy-induced POI accounts for ~5–10% of total patients with diminished ovarian function, the causes of the majority of the patients with idiopathic POI were undetermined^58,59, indicating that the effective treatments for these patients are not available⁶⁰. Our study indicated that there was a close relationship between idiopathic POI and the reduction of serum serine levels. Consequently, we aimed to investigate whether idiopathic POI shares the same pathological mechanisms as chemotherapy-induced POI. Existing studies indicated that staying up late might adversely affect the secretion of female sex hormones, ovarian development, and the ovarian cycle^61,62. We conducted a questionnaire survey involving 124 enrolled patients, distributing 124 questionnaires and receiving 106 completed responses. The analysis of the questionnaire results revealed that staying up late was the most significant factor contributing to idiopathic ovarian insufficiency in women, which might significantly surpass the stress and dieting (Supplementary Table 5). Further analysis of the metabolome revealed that there were significant differences in serum amino acid metabolism levels between women with normal ovarian function and those with diminished ovarian function. In addition, serum levels of AMH and serine were significantly decreased in patients who experienced sleep deprivation (Fig. 7a, b and Fig. S7a–d). The results of the questionnaire survey and medical examinations indicated that two primary factors contributing to late-night activities among women are the use of electronic devices and sleep disorders (Supplementary Table 5). Among female patients experienced with sleep disorders, their serum levels of AMH and serine were significantly decreased, which were significantly correlated with an increased frequency of late-night activities per week in this demographic (Fig. 7c–e). Our analysis indicated that there was a negative correlation between serum AMH and serine levels and the frequency of staying up late per week in patients with diminished ovarian function. These results indicated that staying up late might contribute to serine deficiency in female patients, subsequently leading to diminished ovarian function. Therefore, in our study, a strong correlation has been established between staying up late, the subsequent decrease in serine levels, and the resulting decline in ovarian function in patients with idiopathic POI (Fig. 7f–h).

Fig. 7. Sleep deprivation-induced idiopathic POI is closely associated with serine.

a Serum AMH levels between patients with and without sleep deprivation (n = 33–57, unpaired two-tailed t test). b Serum serine levels between groups (n = 20–31, unpaired two-tailed t test). c Serum AMH levels in patients without a history of staying up late, stay up late due to sleep disorders, or due to the use of electronic devices (n = 9–31, unpaired two-tailed t test). d Serum serine levels across the three patient categories (n = 8–19, unpaired two-tailed t test). e Weekly frequency of staying up late among patients with sleep disorders and those who stay up late due to electronic device usage (n = 11–33, unpaired two-tailed t test). f, g Correlation of serum AMH and serine levels with frequency of late nights, respectively (n = 21–29, Pearson correlation analysis). h Schematic of changes in serum AMH and serine caused by staying up late. i, j Flowchart illustrating the experimental design (n = 20). k Ovarian morphology of normal mice and sleep-deprived mice (n = 8). l Statistical Analysis of Ovarian Organ Rates (n = 8, unpaired two-tailed t test). m, n The levels of E2, and FSH in the serum of mice in both groups before and after sleep deprivation (n = 9–15, unpaired two-tailed t test). o Western blot and quantification of ovarian AMH protein (n = 7, unpaired two-tailed t test). p, q HE staining showing ovarian size and follicle count differences (n = 5–7, unpaired two-tailed t test). r Western blot and quantification of ovarian FSHR protein (n = 7, unpaired two-tailed t test). s Quantitative measurement results of ferrous ion concentration (n = 8–9, unpaired two-tailed t test). t qPCR analysis for detecting the expressions of SLC1A4 (n = 7–10, unpaired two-tailed t test). Graphs indicate mean ± SEM. n indicates the number of patients (a–g) and independent animals (k–t). Scale bars, 500 nm (p). *P  ≤  0.05, **P  ≤  0.01, ***P  ≤  0.001, ****P  ≤  0.0001. Panels h, i were created with BioRender (YF2959WCUZ). Created in BioRender. Gu, h. (2025) https://BioRender.com/kkcqso5.

Due to the inability to obtain ovarian tissue from the women who experienced sleep deprivation, we established a mouse model to investigate the accumulation of ferrous ions in the ovaries (Fig. 7i, j)⁶³. Following extended periods of sleep deprivation, the mice displayed a marked reduction in both survival rate and body weight. In addition, there was a significant increase in the proportion of white blood cells, neutrophils, and granulocytes, alongside heightened inflammatory levels observed in the ovaries (Fig. S8a–d). Furthermore, we observed that the mouse ovaries showed a significant shrinkage following sleep deprivation, accompanied by disordered secretion of sex hormones and a marked reduction in AMH levels within the ovaries (Fig. 7k–r). Interestingly, we also observed an increase in SLC1A4 expression in GCs, and an accumulation of iron ions in the ovaries of sleep-deprived mice (Fig. 7s, t). These findings indicated that the mechanisms underlying sleep deprivation-induced idiopathic ovarian hypofunction might largely overlap with those of CTX-induced POI, suggesting that serine might be a critical factor for maintaining ovarian function (Fig. 8).

Fig. 8. Mechanism diagram of serine inhibiting ferroptosis in GCs.

Cyclophosphamide (CTX) induces premature ovarian insufficiency (POI) by selectively elevating the expression of SLC1A4, a transporter of serine efflux, in ovarian granulosa cells (GCs). Dietary serine supplementation mitigates CTX-induced POI without altering its anti-cancer efficacy, in which the elevated serine promotes S1P synthesis, and in turn, inhibits the nuclear translocation of Nrf2 and consequent HO-1 expression, to suppress ferroptosis in GCs. There is a direct link between serum serine and ovarian function in women of childbearing age with idiopathic ovarian POI in clinical investigation, in which staying up late may be a primary contributor to POI. Mouse sleep deprivation models further demonstrate that POI induced by staying up late shares the same mechanism with clinical observation: the elevated expression of SLC1A4 and the reduced serum serine. This study highlights the specific mechanism of serine maintaining ovarian function by inhibiting ferroptosis induced by the accumulation of ferrous ions in GCs. This figure was created with BioRender (IG2959S4P4). Created in BioRender Gu, h. (2025) https://BioRender.com/ej4soz7.

Discussion

Since the causes and specific molecular mechanisms of POI remain unclear, there is no effective way for preventing and treating it clinically^13,14. In the present study, we revealed that the decline in ovarian function caused by CTX chemotherapy-induced POI and idiopathic POI was closely associated with the reduction of serum serine levels. Our further study revealed that CTX chemotherapy induced ferroptosis of ovarian GCs through increasing iron overload. Serine plays a key role in mitigating iron overload by elevating the synthesis of S1P, consequently preventing ovarian dysfunction, even POI resulting from ferroptosis in GCs. This is also similar to the findings of a recent study on ferroptosis and reproductive function⁶⁴. Our previous study demonstrated that hUSCs exerted their protective effects on CTX-induced POI by reducing the expression of SLC1A4 in GCs through paracrine following transplantation. SLC1A4 is a crucial serine transporter in mammals, and previous studies have demonstrated its influence on neural development through the regulation of serine levels^65–67. Furthermore, studies indicated that Nrf2 promoted the expression of SLC1A4²⁸. These results demonstrated a strong correlation between our findings and previous studies⁵⁵. In CTX-induced mouse POI models, SLC1A4 expression was significantly upregulated, resulting in altered serine levels in GCs²³. In vitro experiments demonstrated that both siRNA-mediated silencing of SLC1A4 in GCs and additional serine supplementation effectively inhibited CTX-induced cell death in GCs. Based on these findings, our results may provide the possibility that serine supplementation may be an alternative approach to hUSCs transplantation for protecting the ovaries. This approach may address the challenge of preserving ovarian function during chemotherapy and potentially reduce ethical concerns, and it may also have the potential for rapid translation into clinical practice.

Serine, a conditional non-essential amino acid, is an important factor for a variety of biochemical reactions in mammals^24,68. Existing research showed that serine deficiency was closely related to a variety of diseases^31,69–73. Studies have shown that dietary supplementation with serine potentially delays the onset and progression of peripheral neuropathy in db/db mice³⁰. Currently, there are over ten ongoing clinical trials for evaluating the protective effects of dietary serine supplementation in various diseases, including our own research project (ChiCTR2400079348). Our results provided a detailed explanation of the specific mechanism through which CTX leads to ovarian hypofunction, including POI. In addition, we propose the key role of serine in this process, highlighting the specific relationship between serine, ferroptosis, and ovarian dysfunction⁷⁴. Our study provides evidence that serine exerts a protective effect on the ovaries during chemotherapy. This effect is attributed to the conversion of serine into S1P, which subsequently inhibits the nuclear import of Nrf2 by modulating the Keap1/Nrf2 signaling pathway. Such regulation reduces the protein-level expression of HO-1 and SLC1A4, ultimately leading to the suppression of GCs ferroptosis²⁸. Notably, although we observed that there was a significant accumulation of ferrous ions in ovarian tissue during CTX chemotherapy, the effect was not related to its anti-cancer activity in breast cancer. Previous studies have demonstrated that serine intake promotes the growth of tumor tissues^75,76. However, our research indicates that, when chemotherapy drugs are administered, an appropriate amount of serine does not adversely affect the efficacy of chemotherapy^76–78. This discovery forms the foundation for future clinical applications of serine supplementation in a variety of diseases, suggesting that its use for ovarian protection during chemotherapy does not alter the efficacy of its anti-cancer effects. In addition, based on the elucidation of the molecular mechanism underlying ovarian damage induced by CTX and the specific protective mechanism of serine, our study holds significant translational implications. Obviously, our results suggest that dietary serine supplementation should offer a cost-effective and safe approach to safeguard ovarian function in breast cancer patients undergoing chemotherapy.

Furthermore, our study has also revealed that there was a significant association between serine levels and ovarian function in women of childbearing age, highlighting that it is necessary to conduct further clinical research on patients with idiopathic POI that are not induced by chemotherapy. Interestingly, our study confirmed that staying up late was a key trigger for idiopathic ovarian insufficiency in somehow^62,79. Through a comprehensive analysis of patients who experienced stay-up-late and the establishment of sleep deprivation animal models, we have identified that serum serine levels were also closely related to AMH levels in POI, suggesting that serine deficiency may be a common cause for various POI^62,79. In addition, we have clarified that the reduced serine levels led to the accumulation of ferrous ions in the ovaries, resulting in ferroptosis through a mechanism that is consistent in both chemotherapy-induced POI and non-chemotherapy-induced POI. These findings provide a foundation for our future clinical research on serine supplementation in idiopathic POI patients, and meanwhile, serine may also serve as a key biomarker for the early detection of ovarian function. More importantly, for patients with non-chemotherapy-induced idiopathic POI, serine supplementation could emerge as a crucial clinical strategy for achieving remission of POI. Despite the confirmed protective effect of serine supplementation on the ovaries, there are still some issues that need to be further resolved. It is essential to further investigate how SLC1A4 facilitates the transport of serine. Currently, there is limited knowledge about SLC1A4 and its mechanism of amino acid transportation across membranes in cells. The current absence of SLC1A4-related mouse models may be hindering research progress⁸⁰. As a result, we are working on constructing mice with a specifically targeted knockout of SLC1A4 in organs such as the liver, ovary, and heart. This initiative is designed to shed light on the mechanisms by which SLC1A4 facilitates serine transport across various cells/organs.

Methods

The study was conducted following the Declaration of Helsinki (as revised in 2013), and written consent for tissue donation was obtained from each patient. The two clinical protocols received approval from the Institutional Review Board of the Nanchang First People’s Hospital (approval no. K-kt2025001; approved date January 14, 2025) and the First Affiliated Hospital of Nanchang University (approval no. IIT2024-329; approved date May 11, 2024). The protocols were approved by the Animal Research Ethics Board of Nanchang University (approval no. NCULAE-20221228066; approved date July 2, 2022) following the Guidelines for the Care and Use of Animals. The study involving animals was conducted following the Basel Declaration.

Chemotherapy and serum collection in breast cancer patients

Twenty-seven breast cancer patients (Luminal A or Luminal B) aged between 25 and 40 were enrolled in the study. These patients exhibited normal ovarian function, as evidenced by regular menstrual cycles and serum levels of AMH and E2 prior to enrollment. They had not received any hormone or amino acid-based medications within one month prior to enrollment, and conditions that could potentially influence AMH values were systematically excluded. All patients received chemotherapy based on CTX (600 mg/Body surface area), which consisted of four cycles. In each cycle, CTX was administered once, followed by a 21-day interval before the subsequent cycle. No amino acid drugs were administered before or during the chemotherapy regimen. Serum samples were collected from 27 patients prior to chemotherapy to measure serum E2 levels and to conduct amino acid-targeted metabolomic sequencing, utilizing serum E2 levels as an indicator of ovarian function. Following the first course of chemotherapy, a 21-day waiting period was observed before initiating the second course. Serum samples were again collected from the same 27 patients before the second course of chemotherapy for E2 measurement and amino acid-targeted metabolomic sequencing. The participants of our study received standard-of-care treatment.

POI study participants and study design

We recruited 124 infertile patients who provided written informed consent, and the study received approval from the local institutional review board. The clinical diagnoses were conducted by experienced clinicians in the Department of Reproduction at the First Affiliated Hospital of Nanchang University. Most patients completed a questionnaire regarding their late-night activities during their initial hospital visit, and we collected ovarian B-ultrasound images and blood samples. Based on the AMH levels and the number of bilateral follicles, patients were categorized into two groups: those with normal ovarian function (Control, AMH >1.3 ng/mL) and those with POI (POI, AMH <1.3 ng/mL). We compared sex hormone levels, bilateral follicle counts, and blood amino acid levels between the two groups, and analyzed the relationships among late-night activities, ovarian function, and amino acid levels. The participants of our study received standard-of-care treatment.

Clinical examinations

All participants underwent sex hormone screening, which included measurements of AMH, E2, FSH, and Luteinizing hormone (LH). Radionuclides were utilized to detect sex hormone levels. Bilateral ovarian images of each participant were collected. The amino acid levels in each patient were analyzed using LC-MS, a process that involves relatively complex steps.

Inclusion criteria

Inclusion criteria for the experimental group comprised female participants aged between 20 and 35, with serum AMH levels below 1.3 ng/mL. In addition, imaging data, including B-ultrasound, indicated compromised ovarian function, characterized by a bilateral antral follicle count of six or fewer. Participants were also required to exhibit irregular menstruation patterns, which could manifest as significant delays, infrequent periods, or amenorrhea.

The inclusion criteria for the control group consisted of females aged between 20 and 35 who exhibited regular menstrual cycles. In addition, their imaging data, including serum AMH levels and B-ultrasound results, indicated normal ovarian function.

Clinical study exclusion criteria

Patients with conditions that affect serum AMH levels, such as polycystic ovary syndrome (PCOS) and diabetes, were ineligible to participate in this clinical study. Additionally, patients who have undergone hormone therapy or taken amino acid-based medications within one month prior to enrollment, as well as those with physical or iatrogenic ovarian function impairment, including prior ovarian surgery, were also excluded from participation. Detailed patient enrollment criteria are presented in the Supplementary Table 2.

Amino acid targeted metabolome sequencing

Samples were extracted with 400 μL of 10% formic acid in methanol-water (1:1, v/v), vortexed for 30 s, and then centrifuged at 2000×g and 4 °C for 5 min. Take an appropriate amount of supernatant and add 10% formic acid in methanol-water (1:1, v/v) to dilute 20 times and vortex for 30 s. Take 100 μL of supernatant and add 100 μL Trp-d3(20 ng/mL), vortex for 30 s. The supernatant was filtered through a 0.22-μm membrane, and the filtrate was added to the LC-MS bottle.

The samples were injected into the ACQUITY UPLC® BEH C18 Column (2.1 × 100 mm, 1.7 μm, Waters, USA), the injection volume was 5 μL, the column temperature was 40°C, and the mobile phase A-50% methanol in water (containing 0.1% formic acid), B-10% methanol in water (containing 0.1% formic acid). The gradient elution conditions were 0–6.5 min, 90–70% B; 6.5–7 min, 70–0% B; 7–14 min, 0% B; 14–14.5 min, 0–90% B; 14.5–17.5 min, 90% B. Flow rate 0–8.0 min, 0.3 mL/min; 8.0–17.5 min, 0.4 mL/min (Supplementary Table 6).

Electrospray ionization (ESI) source, positive ionization mode. The ion source temperature was 500 °C, the ion source voltage was 5500 V, the collision gas was 6 psi, the curtain gas was 30 psi, and the atomizing gas and auxiliary gas were both 50 psi. Scans were performed using multiple reaction monitoring (MRM).

Dietary serine supplementation in mice

C57BL/6 or Balb/c female mice at 8 weeks of age were purchased from Jiangsu GemPharmatech Co., Ltd (Nanjing, China). The mice were maintained on a 12-h light/dark cycle and had ad libitum freely access to food (1010083, jsxtsw, China) and water at the Laboratory Animal Center of the Institute of Translational Medicine of Nanchang University. All animal procedures described in this study were reviewed and approved by the Animal Care and Use Committee of Nanchang University.

To establish POI models, a total of 50 C57BL/6 mice were intraperitoneally injected with CTX (50 mg/kg; MCE, USA) for 15 days. For establishing the breast cancer mouse models, a total of 1 × 10⁶ 4T1 cells (CBP60352, Nanjing Cobioer Bioscience Co., Ltd.) were transplanted into the second breast pad of the mouse. The tumor volume was measured every three days following transplantation. Once the tumor volume exceeded 100 mm³, the mouse received intraperitoneal injections of CTX (20 mg/kg) every Monday and Thursday for chemotherapy for 4 weeks. Exogenous serine was supplemented into mice in two ways: (1) Supplementation with water: L-serine (100 g/L; Solarbio, China) was dissolved in sterile water for free drinking; (2) Supplementation by gavage: L-serine (50 mg/kg or 500 mg/kg) was administered by gavage every day for 5 weeks. In accordance with the animal ethics and welfare requirements established by Nanchang University, mice were euthanized through cervical dislocation following anesthesia induced by an intraperitoneal injection of Avertin.

Enzyme-linked immunosorbent assay (ELISA)

Serum was collected from blood and then allowed to clot for 4 h at room temperature and subsequently stored overnight at 4 °C. The next day, the samples were centrifuged at 180× g for 20 min, and the supernatant serum was collected. Serum E₂, FSH, and AMH levels were measured using the corresponding ELISA kits (Elabscience, China) according to the manufacturer’s protocol. To establish the standard curve, add 50 μL of the serially diluted E2, FSH, and AMH standard solution to the standard wells. In the blank wells, add 50 μL of sample diluent as a negative control to eliminate reagent background interference. Subsequently, add 50 μL of the test sample solution to each of the remaining wells. After completing the sample addition, immediately introduce 50 μL of the freshly prepared HRP-conjugated working solution into each reaction well to ensure uniformity within the reaction system. Finally, cover the 96-well plate with a specialized sealing film and incubated it in a 37 °C constant temperature incubator for 60 min. After completing the incubation step of the experiment, add 350 μL of washing buffer to each well of the microplate and allowed it to stand for 1 minute to ensure the thorough removal of unbound impurities. Following the soaking period, inverted the microplate to drain the liquid and blotted it dry using absorbent paper. This washing process must be strictly repeated five times. After completing the washing process, carefully added 90 μL of substrate solution to each well. Subsequently, covered the plate with a sterile sealing film and transfer the sealed 96-well plate to a constant temperature incubator for a 15 min incubation period. Finally, added 50 μL of stop solution to each well to terminate the reaction and measure the optical density (OD) value using a microplate reader.

Histopathology and follicle count

Ovarian tissues were collected when the experiments were completed, fixed in 4% paraformaldehyde for 24 h, subsequently dehydrated and embedded, and cut into 5-μm-thick slices for hematoxylin and eosin staining (H&E staining). Place the ovarian tissue sections under the microscope and systematically count the number of follicles at each developmental stage, including primordial follicles, primary follicles, secondary follicles, and antral follicles. Following the completion of the serine gavage, both the mouse ovarian tissue and the breast cancer tissue were collected. The collection steps of tumor tissues were identical to those described above, with the only variation being in the dehydration process for the tumor tissue.

Immunofluorescence staining

Ovarian tissues and GCs were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100, followed by incubation with the corresponding antibodies at 4 °C overnight. The antibodies used in the study were as follows: anti-PCNA (1:200, rabbit monoclonal, CST, 13110) and anti-NRF2 (1:200, rabbit monoclonal, CST, 12721 s). After washing with PBS, the cells were incubated at 37 °C for 1 h with a secondary biotinylated goat anti-rabbit IgG antibody (dilution 1:300). The cells were then washed with PBS and incubated at 37 °C for 3 min with DAPI dye liquor. The corresponding immunofluorescence staining was recorded with a laser confocal microscope (Olympus CKX41).

Ferrous ion measurement

The ovarian tissues were lysed with the corresponding lysis buffer. The appropriate amount of detection working solution and lysates were mixed according to the instructions (Solarbio, BC5415, China) and then the mixture was incubated at 37 °C for 10 min. After the incubation, 100 μL chloroform solution was added and vortexed for 5 min, and then centrifuged at 12,000× g for 10 min. The supernatant was collected and detected with a microplate reader (ThermoFisher Scientific, USA).

MDA assay

The unilateral ovarian tissues were collected, and 1 mL of extraction buffer was added for homogenization in an ice bath. Subsequently, centrifuged the mixture at 8000× g for 10 min at 4 °C, and collected the supernatant on ice for further testing. Mixed 100 µL of the sample with 300 µL of the MDA working solution, then incubated in a 100 °C water bath for 60 min, followed by cooling in an ice bath (Solarbio, BC0025, China). Finally, centrifuged at 10,000× g for 10 min at room temperature, and transferred 200 µL of the supernatant to a 96-well plate for detection using a microplate reader (ThermoFisher Scientific, USA).

GCs culture and in vitro POI model

The 3-week-old female mice were injected intraperitoneally with pregnant mare serum gonadotropin (PMSG; Solarbio, China) to stimulate follicle growth. The mice were sacrificed and dissected after 48 h of injection. The adipose tissue and connective tissue attached to the ovaries were peeled off in a pre-chilled PBS solution. The ovarian GCs were released by puncturing the follicles with a 1 mL syringe under a body vision microscope. The cells were washed with PBS three times and centrifuged at 180× g for 5 min. The collected GCs were cultured with DMEM/F12 (1:1) medium containing 15% FBS, 1% Insulin-transferrin-Se, and 100 U/ml penicillin and streptomycin at 37 °C in a humidified atmosphere with 5% CO₂. The first passage of GCs was used in all experiments, and the ovarian GCs were identified by detecting the density of immunofluorescence for the expression of follicle-stimulating hormone receptors (FSHR) since only ovarian GCs express FSHR in ovarian tissue.

In vitro POI model was established with CTX (5 mg/mL), and ferroptosis agonists and inhibitors such as DFO (50 μM; MCE, HY-B0988, USA), Erastin (5 μM; MCE, HY-15763, USA), and HO-1 inhibitor Zinc protoporphyrin (5 μM; Znpp, MCE, HY-101193, USA) were added to inhibit or promote the process of ferroptosis in GCs. The concentration of the serine group in in vitro experiments was 800 μM.

Transcriptome sequencing

Transcriptome Sequencing Library Preparation and Sequencing: Granulosa cells (GCs) were treated with 5 mg/mL CTX for 24 h prior to total RNA isolation with Trizol. An mRNA-seq library was prepared from the total RNA. Briefly, mRNA was selected using poly-T magnetic beads and fragmented. cDNA was synthesized in two steps: first-strand synthesis used random hexamers and M-MuLV Reverse Transcriptase (RNase H–), and second-strand synthesis used DNA Polymerase I and RNase H. The cDNA fragments were end-repaired, A-tailed, and ligated to adapters. Fragments sized 370–420 bp were selected and purified (APure XP system, Beckman Coulter). The library was amplified by PCR with Phusion High-Fidelity DNA Polymerase using indexed primers, followed by purification (AMPure XP system) and quality control (Agilent Bioanalyzer 2100).

Cluster Generation and Sequencing: Indexed libraries were clustered on a cBot system (Illumina) with the TruSeq PE Cluster Kit v3-cBot-HS. Sequencing was performed on an Illumina NovaSeq platform to generate 150 bp paired-end reads.

Annexin V-PI apoptosis assay

After completing each group treatment, the culture medium in the plate was collected and transferred to a 15 mL centrifuge tube. A digestion solution was then added to detach the adherent GCs, which were subsequently included in the same 15 mL centrifuge tube. Following centrifugation at 180× g for 5 min, the supernatant was discarded, and the cell pellet was retained. The pellet was resuspended in pre-cooled PBS and subjected to another centrifugation at 180× g for 5 min, after which the supernatant was again discarded. The pellet was then resuspended in 100 μL of Annexin V binding buffer, containing 5 μL of Annexin V-FITC and 5 μL of propidium iodide (PI) solution (Dojindo, AD10, Japan). After incubating at room temperature for 15 min, the cells were washed with PBS, centrifuged at 180× g for 5 min, and subsequently resuspended in 400 μL of Annexin V binding buffer. Apoptosis detection was performed and analyzed using the BD Jazz system.

Lipid peroxidation assay

To prepare the C11-BDP581/591 DMSO stock solution, add 20 μL of DMSO to the tube containing C11-BDP581/591 and mix thoroughly until fully dissolved. Subsequently, dilute the C11-BDP581/591 DMSO stock solution 1000-fold with complete GCs medium to generate the C11-BDP581/591 DMSO working solution. Add 200 μL of the C11-BDP581/591 working solution to the cells and incubate at 37 °C for 30 min (Dojindo, L267, Japan). After incubation, remove the working solution and wash the cells twice with HBSS. Finally, observe the cells using the Olympus CKX41 inverted fluorescence microscope.

Reactive oxygen species (ROS) assay

The DCFH-DA working solution (10 μmol/L; Solarbio, CA1410, China) was prepared following the manufacturer’s instructions. DCFH-DA was diluted with serum-free culture medium in a ratio ranging from 1:1000 to 1:10,000 to achieve a final concentration of 1–10 μmol/L. Subsequently, 500 μL of the DCFH-DA working solution was added to the cells and incubated at 37 °C for 20 min. The GCs were then washed three times with serum-free GCs medium to thoroughly remove any unincorporated DCFH-DA. Finally, the green fluorescence emitted by the GCs was observed under a fluorescence microscope.

S1P content assay

GCs were digested with trypsin, centrifuged at 360 g for 5 min, and the cell pellets were resuspended in pre-chilled PBS and quickly frozen in liquid nitrogen, then thawed in water at 37 °C. And this step was repeated three times to disrupt the cells, and the lysates were centrifuged at 1500× g for 15 min, the supernatant was detected according to the instructions of the S1P content kit (mlbio, m1062988, China).

Mitochondrial membrane potential assay

The JC-1 working solution (Solarbio, M8650, China) was prepared according to the manufacturer’s instructions. One mL of JC-1 working solution was added to the cell culture and incubated at 37 °C for 20 min. After the incubation, the supernatant was discarded, and the cells were washed twice with JC-1 staining buffer. Then, GCs were covered with cultured medium and monitored under a fluorescence microscope.

Intracellular ferrous ion fluorescence assay

The FerroOrange working solution (Dojindo, F374, Japan) was prepared according to the manufacturer’s instructions. GCs culture medium was then removed, and GCs were washed 3 times with serum-free GCs culture medium. FerroOrange working solution (2 μmol/L) was added into the GCs and incubated at 37 °C for 30 min in an incubator and then, the GCs were observed under a confocal microscope (Leica SP8, Germany).

Mitochondrial morphology observation

The GCs samples were fixed with prefixed electron microscopy fixative (Servicebio, China) and sent to Wuhan Servicebio Co., Ltd (Wuhan, China) to complete the electron microscope scan.

Western blot analysis

Total proteins were extracted from ovarian tissues, and the concentration of total proteins was determined by the bicinchoninic acid method (BCA; Solarbio, China). The lysates were separated by 10% denaturing SDS-PAGE gels, then transferred to nitrocellulose membranes (BioRad, USA), which were incubated with primary antibodies anti-β-actin (1:5000, mouse monoclonal, CST, 3700), anti-FSHR (1:1000, rabbit monoclonal, Protein technology, 22665-1-AP), anti-AMH (1:1000, rabbit monoclonal, Abclonal, A8535), anti-Nrf2 (1:1000, rabbit monoclonal, CST, 12721), anti-Keap1 (1:500, rabbit monoclonal, Abclonal, A17062), anti-Nox4 (1:1000, rabbit monoclonal, Abclonal, A22149), anti-HO-1 (1:1000, rabbit monoclonal, Wanleibio, WL02400) and anti-SLC1A4 (1:1000, rabbit monoclonal, Abclonal, A12507) at 4 °C overnight. After 16 h, the membrane was incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit or rabbit anti-mouse secondary antibody (Invitrogen, USA) for 1 h at room temperature. The blots were detected with enhanced chemiluminescence (ECL), and images were quantified using the Super Signal West Pico chemiluminescence detection system.

Quantitative reverse-transcription polymerase chain reaction (qRT-PCR)

Total RNAs were isolated from ovarian tissues or GCs using Trizol regent (ThermoFisher, USA) and reverse transcribed into cDNA using M-MLV reverse transcriptase (Promega, Shanghai, China). The primers for the target genes were designed as in Supplementary Table 7. Polymerase chain reactions (PCR) were carried out in a PCR thermal cycler (Thermo Hybaid, USA). Quantitative PCR was performed using the ABI-ViiA7 PCR machine. The primers for qPCR are listed in Supplementary Table 7.

Ultrasound imaging of the ovaries in infertile patients

Ultrasound images of the ovaries were obtained from infertile women early in the follicular phase of the menstrual cycle (2–5 days). The patient was supine with a moderately full bladder. Transvaginal ultrasound uses a high-frequency probe (frequency range: 5–9 MHz) to ensure optimal resolution. The ovaries were systematically scanned in multiple planes to assess ovarian volume, follicle count (including sinus follicle count, AFC), and the presence of ovarian cysts or abnormalities. All ultrasound examinations are performed by experienced sonographers according to a standardized protocol to ensure data consistency and reliability.

Sleep deprivation mouse model

Balb/c female mice were purchased from Jiangsu GemPharmatech Co., Ltd (Nanjing, China). The mice were maintained on a 12-h light/dark cycle and had ad libitum access to food and water at the Laboratory Animal Center of the Institute of Translational Medicine. Mice aged 6–8 weeks were randomly divided into an experimental group (n = 20) and a control group (n = 20). The experimental group of mice was placed in a sleep deprivation device, stay up 20 h a day, rest 4 h, for 4 days, then rest 2 days, according to this schedule for four consecutive times. The control mice ate the same diet as the experimental group, except that they were not deprived of sleep. Their mortality and weight changes were measured.

Statistical analysis

The results were presented as average value ± standard error of the mean (SEM). Student’s t test was used for analysis between the two groups. One-way analysis of variance (ANOVA) was used to compare data among three or more groups. Differences with a P value of <0.05 were considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

H.B.X. and K.Y.D. led the project and contributed to the conception and design of the study, data analysis and interpretation, and manuscript revising. H.C.G. performed the most experiments and data analysis. Y.Q.Z., Y.W.Z., Y.K.W., H.T.L., J.Q.W., S.Q.T., and X.Y.W. performed animal models and cell experiments. D.S.L., L.Q.C., Z.H.L., and Y.Y.W. were responsible for clinical sample collection and patient testing result collection. Y.Q.Z. was responsible for clinical sample metabolome sequencing and result analysis. H.C.G. and L.F.W. wrote the manuscript draft. All authors read and approved the final version of the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The RNA-seq data in this study have been deposited in the GEO database under accession number GSE314714. The amino acid-targeted metabolomics data in this study have been deposited in the MetaboLights database under accession number MTBLS13561. All other relevant source data supporting the key findings of this study are provided in this paper. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-68440-1.