Thioredoxin-1 inhibits granulosa cell ferroptosis to rescue ovarian aging through mitophagy-dependent activation of BNIP3L

19 Ovarian aging is closely associated with a decline in fertility and an increase in 20 reproductive dysfunction. Ovarian granulosa cells (GCs) support oocyte homeostasis 21 and development, yet insight into GC dysfunction during aging is limited. Here, we 22 show that aged GC s of humans and mice have indications of elevated ferroptosis, 23 including increased ferroptosis-related metabolites, lipid peroxidation, and iron 24 accumulation. The ferroptosis inhibitor Ferrostatin-1 reversed ovarian impairment and 25 fertility of aged mice in vivo. We show that the age-related reduction in the expression 26 of TXN (thioredoxin) leads to ferroptosis in human and mouse GCs by blocking 27 BNIP3L-dependent mitophagy. Exogenous activation of TXN could promote 28 mitophagy, thereby clearing excessive ROS and inhibiting ferroptosis. These results 29 suggest that anti -ferroptosis-related treatments may assist in treating aging -related 30 reproductive disorders. 31 32

Ferroptosis, mitophagy, ovarian aging, TXN, BNIP3L, granulosa cells 33 34 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 2 35 Key points: 36 • Ferroptosis signatures are upregulated in aged GCs from human and mouse ovaries 37 • TXN is deregulated in aged GCs, leading to mitochondrial and ROS metabolic dysfunction 38 • TXN binds to DNA to regulate autophagy and mitophagy genes, including BNIP3L 39 • Inhibition of ferroptosis can ameliorate GC dysfunction 40 41

42 Ovarian aging contributes to the progressive decline in female reproductive capability 43 and fertility by decreasing the quality and quantity of oocytes 1. Although the 44 contribution of endocrine, genetic, and metabolic factors that cause decreased oocyte 45 quality in ovarian aging has been explored2, practical strategies to improve or delay 46 ovarian aging remain limited, and new methods focusing on improving ovarian function 47 are required. In both mice and humans, multiple studies have described the cellular and 48 molecular changes that are associated with ovarian aging 1,3-6. Aging transforms the 49 ovary from a youthful, functional state to a senescent one through multiple pathways 7. 50 Extensive work has focused on oocyte development quality 8-10, but less work has been 51 paid to other cells in the ovary. Particularly important are the granulosa cells (GCs) that 52 protect and communicate with oocytes to manage follicle development, steroidogenesis, 53 and endocrine regulation 11,12. Dysfunction of GCs causes follicle atresia, premature 54 ovarian failure, and female subfertility or infertility 11,12, and affects oocyte quality 13-55 15. Multiple pathways are deregulated in the aging ovarian GCs, particularly increased 56 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 3 inflammation 16, DNA damage and oxidation 17, lipid peroxidation, and reactive oxygen 57 species (ROS) that can serve as biomarkers for ovarian function 18. However, clinical 58 interventions based on these pathways remain limited. 59 Ferroptosis is a type of programmed cell death caused by iron -dependent 60 accumulation of lipid peroxides and ROS 19-23. Evidence suggests that ferroptosis is 61 associated with aging and aging -related pathologies, including increased oxidative 62 stress and cell mortality 24. Ferroptosis is potentially related to ovarian aging due to 63 high ROS and abnormal lipid metabolism 21,25 which disrupts the normally low ROS 64 levels in oocytes and dormant follicles 26. Iron metabolism and ROS are dysregulated 65 in polycystic ovary syndrome (PCOS) 27,28, and miR-93-5p drives NF-kB apoptosis and 66 ferroptosis in PCOS GCs 29. Premature ovarian insufficiency (POI) can be caused by 67 deregulation of the BNC1-NF2-YAP pathway that induces ferroptosis in oocytes 30. 68 Ferroptosis is also widely observed in ovarian cancer 31,32. However, a relationship 69 between ferroptosis and redox metabolism in aging and GC dysfunction has not been 70 established. 71 Mitochondrial dysfunction can have a significant impact on aging GCs by 72 disrupting energy homeostasis and redox balance and impairing steroidogenesis and 73 follicular development. Mitophagy, which is the selective degradation of mitochondria 74 by autophagy , is crucial for maintaining cellular homeostasis and is frequently 75 deregulated in aging 33. Mitochondrial deregulation is observed in aged oocytes 8,34, and 76 GCs 35, suggesting that proper mitophagy is essential for maintaining mitochondrial 77 integrity during follicular development. Spermidine supplementation can improve 78 mitochondrial quality by enhancing mitophagy and likely contributes to improved 79 fertility in aged mice 36. However, how mitophagy is mechanistically regulated in the 80 aging ovary remains unclear. Potentially, t he accumulation of damaged mitochondria 81 could increase the susceptibility to ferroptosis due to excessive lipid peroxidation and 82 production of ROS 37,38. Thioredoxin (TXN) combats ROS-induced oxidative stress and 83 has an inhibitory role in ferroptosis by detoxifying lipid ROS by activating GPX4 84 (glutathione peroxidase 4) 39,40. However, whether TXN functions in ovarian aging and 85 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 4 reproductive capability remains unclear. Crosstalk between mitophagy, ROS response 86 (via TXN and GPX4), and ferroptosis could have implications for understanding the 87 complex mechanisms underlying ovarian function decline and age-related pathologies. 88 In this study, we propose a link between ferroptosis, mitophagy, and antioxidant 89 regulation in ovarian aging in GCs. Our results show that human GCs from advanced 90 maternal age women (>40 years) have increased signs of ferroptosis and decreased 91 mitochondrial function. This is accompanied by decreased TXN activity, which leads 92 to impaired BNIP3L-dependent mitophagy, lipid peroxidation, and ROS accumulation 93 that feeds back to cause increased ferroptosis. Inhibition of ferroptosis or activation of 94 TXN improved the quality of aging GCs in mouse and human models and improved 95 fertility in old mice and premature ovarian failure (POF) mouse models. 96 97

98 Age-associated senescence and ferroptosis are upregulated in human and mouse 99 ovarian granulosa cells 100 A lot of attention has been paid to understanding aging in the context of zygote 101 development and implantation 9,10,41, however, less effort has been expended on the 102 ovary, particularly on GC support cells 11. Hence, we explored signs of senescence that 103 may explain GC dysfunction. Patients with PCOS, premature ovarian insufficiency 104 (POI), and other ovarian hypofunctions caused by tumors or endocrine diseases were 105 excluded from this study. Only patients requiring assisted reproductive treatment (ART) 106 between 22-46 years old were included. GCs were purified from young (36 years old) by 108 excluding CD45+ mononuclear cells by FACS (Supplementary Figure 1a). GC purity 109 was confirmed by immunostaining with FSHR (follicle-stimulating hormone receptor) 110 (Supplementary Figure 1b). Interestingly, senescence-associated SA-β-galactosidase 111 (SA-β-gal) and p21CDKN1A were both elevated in GCs from older patients (Figure 1a-112 c). Whilst LMNB1 (Lamin B1), a negative senescence marker, declined with age 113 (Figure 1b and c). We reanalyzed scRNA-seq (single-cell RNA-seq) data from human 114 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 5 ovaries from young and old patients and isolated the GCs (Figure 1 d and 115 Supplementary Figures 1c) 42. As previously reported, there were three main clusters 116 of GCs: cluster 1 comprised young GCs, cluster 3 was mainly old, and cluster 2 was 117 intermediate ( Supplementary Figure 1 d and e). Aged GCs had elevated levels of 118 ferroptosis-driver genes and decreased expression of ferroptosis -suppressors (Figure 119 1d-f and Supplementary Figure 1 f, g) 43, and increased senescence genes 120 (Supplementary Figure 1 h). Interestingly, based on the scRNA -seq, elevated 121 ferroptosis was mainly restricted to GCs and monocytes, and other ovary tissues were 122 relatively unaffected (Supplementary Figure 1i). In support of enhanced ferroptosis 123 in GCs, the ferroptosis-related metabolite MDA (Malondialdehyde), an end-product of 124 lipid peroxidation 22, was significantly increased from young to old GCs (Figure 1g). 125 Conversely, GSH (reduced glutathione) , which is the substrate required by GPX4 to 126 reduce lipid peroxides and protect against ferroptosis 40,44, was significantly decreased, 127 suggesting GPX4 activity is impaired (Figure 1h). Additionally, old GCs had increased 128 ferrous (Fe 2+) ion deposition, as measured by FerroOrange ( Figure 1 i). Further, 129 suggestive of ferroptosis, mitochondrial membrane potential was reduced in old GCs 130 (Figure 1j). These data support elevated ferroptosis -related pathways and proteins in 131 GCs of older maternal-age women. 132 To explore ferroptosis-related metabolic changes in more detail 45, we performed 133 metabolomic mass spectrometry on purified human GCs from young and old women 134 (Supplementary Figure 2a). Overall, 1198 metabolites were detected , with 102 135 metabolites downregulated in young and 146 metabolites up -regulated in old GCs 136 (Figure 1k, Supplementary Figure 2 a-c, and Supplementary Table 1 ). Metabolic 137 Ontology analysis indicated the up-regulation of ferroptosis-related metabolites in old 138 GCs (Figure 1l and m), particularly GSH, and the downregulation of arachidonic acid 139 (AA; Figure 1 l-n). The ratio of GSH/GSSG was lower in old GCs ( Figure 1 o), 140 supporting impaired GPX4 catalytic activity. Lipid peroxidation is both a driver and 141 marker for ferroptosis 46, hence we used BODIPY-C11 staining to determine the level 142 of lipid peroxidation, which was indeed up-regulated in the old GCs (Figure 1p and q). 143 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 6 These results show that human GCs from old patients have increased markers of 144 senescence and ferroptosis. 145 To explore ferroptosis in aged GCs, we utilized the human ovarian GC KGN cell 146 line and treated the cells with increasing concentrations of H2O2. This led to a decrease 147 in LMNB1 levels, and a corresponding increase in the senescence markers p16 CDKN2A 148 and p21CDKN1A were matched by increases in SA-β-gal (Supplementary Figure 3a and 149 b), mimicking the phenotype in primary human GCs. As expected, cell viability was 150 significantly reduced in the H 2O2-treated cells compared to the control group 151 (Supplementary Figure 3c). The cell viability could be partially rescued when KGN 152 cells were treated with the ferroptosis inhibitor Fer-1 (Ferrostatin-1) (Supplementary 153 Figure 3c). Fer-1 treatment also led to a decrease in the ferroptosis metabolite MDA 154 and an increase in GSH ( Supplementary Figures 3d and e). Lipid ROS levels were 155 significantly increased in H2O2-treated KGN cells compared with control cells, and this 156 effect could be reversed by treatment with Fer -1 (Supplementary Figure 3 f and g). 157 Ferrous (Fe2+) ion levels were increased, and mitochondrial membrane potential was 158 decreased in H 2O2-treated KGN cells compared with the control ( Supplementary 159 Figure 3h-j). Together, increased lipid ROS and MDA , and decreased GSH and 160 mitochondrial membrane potential in H 2O2-treated KGN cells suggest ferroptosis is 161 induced in aged GCs and may be related to the dysfunction of ovarian GCs. 162 Aging impairs reproductive capacity in mice and leads to reduced litter size and 163 disrupted estrous cycles. GCs from old (52 weeks) mice, as in humans, showed a similar 164 pattern of aging -related senescence markers, including increased p21 CDNK1A and 165 decreased LMNB1 in old mice (Supplementary Figure 4 a). The ovarian follicle 166 counts of old mice were significantly lower than young mice (8 weeks) 167 (Supplementary Figure 4b), and the number of oocytes was significantly reduced in 168 old mice (Supplementary Figure 4c). Furthermore, ovarian tissues showed increased 169 MDA and decreased GSH that correlated with the increasing age of the mice, especially 170 after 48 weeks ( Supplementary Figure 4 d and e). Fe 2+ was increased, and 171 mitochondrial membrane potential was decreased in GCs of old mice compared with 172 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 7 young mice (Supplementary Figure 4f and g), suggesting that GCs from old mice also 173 have increased levels of ferroptosis. 174 175 Inhibition of ferroptosis improves ovarian function and fertility in old mice 176 To define the damage caused by ferroptosis to female fertility in vivo , we built a 177 ferroptosis inhibition and activation mouse model by repeated intraperitoneal injection 178 of the ferroptosis inhibitor Fer-1 in old mice, and ferroptosis initiator Erastin in young 179 mice (Figure 2a). We also generated a premature ovarian failure (POF) mouse model 180 induced by busulfan and cyclophosphamide treatment (Figure 2a) 47. Ovarian size 181 weight (ovarian weight/body weight) of 12-week-old mice treated with Erastin was 182 decreased in the POF mice (Figure 2b and c). Notably, ovarian weight was rescued by 183 Fer-1 injection in both 52-week-old and POF mice (Figure 2b and c). All follicle types 184 were significantly decreased in the Erastin-treated and POF groups, and several follicle 185 types were significantly up-regulated in the old or POF mice treated with Fer-1 (Figure 186 2d). This effect extended to litter size, as the average litter size was significantly smaller 187 in both Erastin-treated and POF mice compared to their respective controls (Figure 2e 188 and f), and, remarkably, 52-week-old mice treated with Fer -1 gave birth (Figure 2e 189 and f). Erastin-treated and old mice had low hormone levels that could be improved 190 with Fer-1 treatment (Figure 2g). Fer-1 and Erastin cause systemic effects; hence, to 191 remove potential confounding effect of the mouse endometrium on implantation rates, 192 we performed in vitro fertilization of POF and aged mice oocytes and monitored 193 blastocyst development in vitro. Quantification of cleavage rates showed a si gnificant 194 increase in the 2-cell (2C) cleavage rate in embryos from old mice treated with Fer-1, 195 and a significant reduction in 2C -cleavage in young mice that had been treated with 196 Erastin (Figure 2h and i). Fer-1 also significantly improved 2C-cleavage in the POF 197 model (Figure 2h and i). This shows that inhibition of ferroptosis in mice improved 198 oocyte quality and embryonic developmental potential. 199 200 The TXN antioxidant system is deregulated in aged GCs 201 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 8 The data from the mouse models suggest that a consequence of increased ferroptosis is 202 a depleted number of follicles and ovarian function. Hence, we reanalyzed an RNA-seq 203 dataset 4 of GCs from six patients with diminished (DOR) or normal ovarian reserves 204 (NOR). In the DOR GCs, 2484 genes were significantly upregulated and 2309 205 downregulated ( Supplementary Figure 5a), and as expected, ferroptosis drivers 206 tended to be downregulated in DOR , and suppressors were associated with NOR 207 patients ( Figure 3 a and Supplementary Figure 5b, c). The above data suggested 208 ferroptosis is up-regulated in DOR and may be a contributing factor, and human DOR 209 at least superficially resembles the erastin-treated young mice. 210 Cystine is imported into the cell via the SLC3A2/SLC7A11 (System X c-), a 211 glutamate/cystine transporter. Cystine is then reduced to cysteine by TXN to modulate 212 GPX4 activity 48,49. Erastin promotes ferroptosis by inhibiting SLC3A2/SLC7A11 213 transportation, suggesting that this process may be disrupted in DOR and aged human 214 GCs. We confirmed by RT-qPCR that the ferroptosis suppressor genes (TXN, GPX4, 215 and SLC7A11) were down -regulated ( Figure 3 b), whilst the ferroptosis drivers 216 ELOVL5, NOX4, and the ferroptosis activator NNMT 50 were up-regulated (Figure 3c) 217 in GCs from young, middle, and aged human GCs. This pattern was matched in RNA-218 seq data from NOR and DOR GCs ( Figure 3d). Interestingly, SLC7A11, GPX4, and 219 TXN were downregulated at the protein level in old GCs from humans, whilst NOX4 220 was up-regulated in old GCs ( Figure 3e and f). Immunofluorescence imaging agreed 221 as SLC7A11, TXN, GPX4 were downregulated in the old GCs , whilst ELOVL5 and 222 NOX4 were up-regulated (Figure 3g). There was a similar pattern in the ovaries of aged 223 mice, as IHC confirmed that TXN and SLC7A11 were downregulated in 52 -week-old 224 mice, whilst NOX4 was up -regulated (Figure 3h). GPX4 expression was somewhat 225 mixed in this mouse model experimental system (Figure 3h). IHC of human GCs, 226 however, showed that GPX4, TXN, and SLC7A11 were all reduced in GCs from old 227 patients, whilst NOX4 was up -regulated ( Figure 3 g). These data suggest that 228 dysfunction of TXN, GPX4, SLC7A11, and NOX4 may induce or modulate ferroptosis 229 in aged human GCs. 230 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 9 231 TXN delays H2O2-induced senescence in KGN GC-like cells by inhibiting 232 ferroptosis 233 To address whether deficiency of the TXN-based antioxidant system triggers ferroptosis 234 in aged GCs, we knocked down TXN using shRNAs (shTXN#1, sh TXN#2) or 235 overexpressed (TXN OE#1, TXN OE#2) in KGN cells (Figure 4a). Overexpression of 236 TXN significantly increased cell viability and rescued cell proliferation inhibition in 237 H2O2-treated KGN cells ( Supplementary Figure 6a). Notably, and indicative of 238 ferroptosis, Fe2+ accumulation (Figure 4b) and peroxidized lipids (Figure 4 c and 239 Supplementary Figure 6b) were significantly increased in TXN knockdown cells 240 compared with the sh LUC controls. In contrast, TXN overexpression resulted in the 241 opposite effect: Fe2+ accumulation and lipid peroxidation were low (Figure 4c and 242 Supplementary Figure 6b). Notably, in cells stressed with H2O2, overexpression of 243 TXN reduced lipid peroxidation and Fe2+ levels (Figure 4b, c, and Supplementary 244 Figure 6b). Notably, mitochondrial depolarization was reduced in TXN OE cells and 245 increased in TXN knockdown cells ( Supplementary Figure 6c), suggesting altered 246 mitochondrial function . In addition, SA -β-gal production was reduced in TXN OE 247 H2O2-treated cells but was increased in the TXN knockdown cells (Figure 4d). In 248 support for a role for TXN in ferroptosis and senescence, the senescence markers 249 (p21CDKN1A, p16CDKN2A, and LMNB1) and ferroptosis -related markers (GPX4, 250 SCL7A11, and NOX4) were increased in TXN knockdown cells and matched the effect 251 of H2O2 (Figure 4e). Overall, TXN levels inversely correlated with higher levels of 252 Fe2+ (Figure 4b), peroxidized lipids (Figure 4c), SA-β-gal (Figure 4d), and ferroptosis 253 marker gene expression (Figure 4e). 254 We next used the TXN activators (TXNIP-IN-1 and NADPH) and an inhibitor (PX-255 12) to explore the role of TXN in ferroptosis. Cell viability was significantly increased 256 by TXNIP-IN-1 when combined with NADPH, and this pattern extended to cells treated 257 with H2O2 (Supplementary Figure 6d). GSH was decreased in PX-12-treated cells and 258 increased when TXN was activated (TXNIP -IN-1 or NADPH ) (Supplementary 259 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 10 Figure 6e). Conversely, MDA levels were significantly increased after PX-12 treatment 260 and decreased in cells treated with TXNIP-IN-1 and/or NADPH ( Supplementary 261 Figure 6f). Furthermore, mitochondrial membrane potential, accumulation of Fe 2+, 262 lipid peroxidation, and ROS levels were all decreased in PX-12-treated cells (Figure 4f 263 and Supplementary Figure 6 g-j). Notably, these ferroptosis -related factors were 264 significantly increased by PX-12 treatment and reduced by TXNIP-IN-1 or NADPH 265 (Figure 4f and Supplementary Figure 6g-j). These changes were reflected in changes 266 in protein levels, as the ferroptosis suppressor proteins SLC7A11 and GPX4 were 267 significantly up-regulated by PX-12 treatment (Figure 4g). Conversely, treatment with 268 the TXN activators TXNIP -IN-1 and NADPH led to upregulation of ACSL 4 (Figure 269 4g). Finally, SA-β-gal was increased in H 2O2 and PX-12-treated cells and could be 270 ameliorated by TXN -IN-1 and NADPH ( Figure 4 h). These results suggested that 271 activation of TXN significantly inhibited ferroptosis in H2O2-treated KGN cells. 272 273 TXN deficiency blocks mitophagy in aged ovarian GCs 274 To study the mechanism underlying the ferroptosis induced by TXN deficiency in 275 ovarian aging, we performed a 4D -DIA proteomics to determine the differentially 276 abundant proteins in human ovarian GCs from young and old individuals and identified 277 8127 proteins, of which 71 and 436 were differentially abundant in old versus young 278 patients (Figure 5a and Supplementary Table 2). KEGG pathway enrichment analysis 279 showed that, among the top 10 enriched pathways for the young vs old, mitophagy and 280 autophagy-related pathways were significantly enriched in young (Figure 5b and 281 Supplementary Figure 7 a and b), including mitophagy -related proteins , such as 282 BCL2L1, OPTN, and BNIP3L ( Figure 5 c and Supplementary Figure 7 b). 283 Interestingly, BNIP3L RNA was modestly downregulated (Supplementary Figure 7c) 284 and was significantly downregulated in the mass spec data , western blot, and 285 immunofluorescence of old and young GCs (Figure 5c-e). 286 LC3/GABARAP proteins are important for autophagy 51 (MAP1LC3B) was not 287 detected in the protein mass spec data, however the autophagy LC3B/ATG8 -related 288 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 11 protein GABARAPL1 was significantly down-regulated at the RNA, and protein level 289 in old GCs (Figure 5d, f and Supplementary Figure 7d and e), suggesting impaired 290 autophagy and/or mitophagy. To reveal autophagic flux, KGN cells were transfected 291 with a dual reporter plasmid containing GFP and RFP fused to LC3B (GFP-RFP-LC3B) 292 52. Knockdown of TXN in KGN cells led to increased RFP and autophagic turnover, 293 conversely, overexpression of TXN led to increased autophagic flux (Figure 5g and h), 294 suggesting TXN is driving increased autophagy . To observe mitophagy, we examined 295 the formation of mitophagosomes represented by c o-localizing mitochondria and 296 lysosomes 45. Indeed, in young GCs, there was a substantial increase in mitophagy and 297 co-localization with lysosomes, and this pattern was absent in GCs from old patients 298 (Figure 5 i). Similarly, when TXN was transfected into young and old patient GCs , 299 young GCs responded with increased LC3II, and presumably autophagy, whilst old 300 cells were weak to respond (Figure 5d). These results provide evidence for TXN-driven 301 impaired autophagy and mitophagy in old GCs, due to disrupted TXN regulation. 302 We explored mitophagy in aged mice ovarian tissues. T ransmission electron 303 microscopy (TEM) of mitochondrial membranes suggested dysfunction in 304 mitochondrial membranes as they appeared fuzzy or broken (Supplementary Figure 305 7f). The total number, mitochondrial size , and healthy/unhealthy ratio were all 306 significantly decreased in mouse GCs in 52-week-old mice compared to young 8-week 307 mice (Figure 5j). Inhibition of ferroptosis via intraperitoneal injection of the ferroptosis 308 inhibitor Fer-1 improved mitochondrial morphology of GCs ( Supplementary Figure 309 7g). Notably, mitochondrial dysfunction, swelling , and membrane potential collapse 310 were all present in Erastin and POF model mice and were rescued with the addition of 311 Fer-1 ( Figure 5k and Supplementary Figure 7 h). The a ging-related marker 312 p21CDKN1A was decreased in Fer-1-treated old and POF model mice, and LMNB1 was 313 increased (Supplementary Figure 7h). Finally, the ROS markers TXN, NOX4, and 314 ferroptosis markers SLC7A11, GPX4, and ACSL4 were all increased (Supplementary 315 Figure 7h), and NAD dehydrogenase -related proteins were also higher in old GCs 316 (Supplementary Figure 7 i). These data suggest a link between ROS, mitochondrial 317 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 12 dysfunction, and ferroptosis. 318 319 TXN regulates the expression of BNIP3L to trigger mitophagy in GCs 320 The mitochondrial outer membrane protein BNIP3L serves as a mitophagy receptor by 321 recognizing autophagosomes through ATG8 53. It was up -regulated in old GCs in the 322 human GC mass spec ( Figure 6a), suggesting the involvement of BNIP3L -mediated 323 mitophagy. We wondered if BNIP3L expression was downstream of TXN. When TXN 324 was knocked down or overexpressed in KGN cells BNIP3L protein and RNA matched 325 the TXN changes (Figure 6b, c and Supplementary Figure 8a and b), additionally, 326 there was a shift in LC3 from LC3I (cytoplasmic) to the LC3II (autophagosome, 327 membrane-bound) form that signifies increased autophagy, and LC3II/LC3I ratio was 328 reduced in KGN cells treated with H2O2 and could be rescued when TXN was 329 overexpressed ( Figure 6 b and c). This suggests that mitophagy enhanced by TXN 330 might be regulated in a BNIP3L-dependent manner. 331 TXN regulates the activity of transcription factors (TFs), for example, p53 54, NFkB 332 55, and AP-1 56,57. Evidence suggests TXN may directly interact with TFs bound to DNA 333 54,57,58, possibly through APE1/REF1 55,56. Hence, we wondered if TXN could bind to 334 the genome, essentially acting as an epigenetic factor to modulate TF activity, and if so, 335 could this identify downstream regulatory targets of TXN. To this end, we performed 336 CUT&Tag for TXN in KGN cells 59. Peak discovery identified 17,963 TXN-bound loci, 337 which were primarily clustered around transcription start sites (TSSs) (Figure 6d and 338 e). Motif discovery identified AP-1 motifs at the TXN -bound loci (Figure 6f), which 339 suggests indirect binding of TXN to the genome. Interestingly, gene ontology analysis 340 of TXN -bound genes showed pathways related to mitophagy, proteolysis, and 341 autophagy (Figure 6g and Supplementary Figure 8c), and this was exemplified by 342 TXN binding to the promoters of BNIP3L, GPX4, TXN, and SLC7A11 (Figure 6h and 343 Supplementary Figure 8d). 344 We confirmed the binding of TXN to the BNIP3L promoter using c hromatin 345 immunoprecipitation (ChIP-qPCR) (Supplementary Figure 8e and f). To test whether 346 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 13 TXN binding regulates BNIP3L, we generated a luciferase vector containing the 347 BNIP3L promoter (Supplementary Figure 8e). The luciferase activity in the TXN-348 overexpressing cells was increased, whereas when TXN was knocked down, luciferase 349 activity was reduced in both KGN and 293T cells ( Figure 6i). These data suggested 350 that TXN binds to DNA to regulate mitophagy and autophagy -related genes through 351 AP-1 and binds to the promoter and activates BINP3L. 352 To confirm that the mitophagy protein BNIP3L was involved in ferroptosis, we 353 transfected a vector containing an shRNA targeting BNIP3L and a control scrambled 354 shRNA into TXN OE KGN cells. Cell viability was significantly impaired when 355 BNIP3L was knocked down in TXN OE cells (Supplementary Figure 8g). The 356 ferroptosis-related metabolite GSH was significantly lower in the BNIP3L knockdown 357 KGN cells , whilst MDA was significantly higher (Supplementary Figure 6h). 358 Furthermore, Fe2+, lipid ROS, and mitochondrial depolarization were all significantly 359 increased in BNIP3L knockdown cells and could be partially rescued in the TXN OE 360 cells. (Figure 6j, k, and Supplementary Figures 8 i and j). Similarly, in KGN cells 361 treated with H2O2, the ferroptosis markers (lipid ROS and MMP levels) were increased 362 when BNIP3L was overexpressed (Supplementary Figure 8k and l). Finally, western 363 blot showed that th e ferroptosis suppressor s SLC7A11 and GPX4 were significantly 364 down-regulated, while the ferroptosis driver ACSL4 was up -regulated in BINP3L 365 knockdown cells ( Figure 6l). These results establish a mechanistic link suggesting 366 TXN regulates BNIP3L to inhibit ferroptosis in aged ovarian GCs. 367 Up-regulation of ferroptosis in GCs may have a knock -on effect on the oocytes 368 they surround and support. Interestingly , we observed an accumulation of Fe 2+ in the 369 cytoplasm of oocytes from old women that was absent in young women (Figure 7a and 370 b). This suggests ferroptosis in ovarian GCs may affect oocyte development (or vice 371 versa) to cause iron overload. This also matches the up-regulation of ferroptosis seen in 372 POI 30. Notably, these changes cannot be detected from RNA -seq of oocytes, as 373 ferroptosis and mitophagy-related genes are unchanged between young and old oocytes 374 (Figure 7c) 60. Overall, ferroptosis in GCs leads to impaired oocyte quality, which leads 375 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 14 to reduced fertility in older women. 376 377

378 Ferroptosis is caused by a redox imbalance between the production of oxidants and 379 antioxidants, which is driven by the abnormal expression and activity of multiple redox-380 active enzymes that produce or detoxify free radicals and lipid oxidation products , 381 whereas selective mitochondrial autophagy may inhibit ferroptotic death. Here, we 382 show that reduced TXN leads to consequent reduced expression of BNIP3L, and 383 reduced mitophagy, which ultimately causes a feedback loop that leads to increased 384 ferroptosis (Figure 7d). Here, we found that TXN binds to the promoter of and activates 385 the mitophagy-related protein BNIP3L, which was consistently down-regulated in aged 386 GCs, blocking the occurrence of non -canonical mitophagy and resulting in the 387 accumulation of damaged mitochondria and release of ROS in human GCs, in vitro cell 388 models, and mouse models of aging. 389 Ferroptosis could be a potential therapeutic target for ART and infertility treatment. 390 Notably, supplementation of NADH can rescue age-related declines in mouse fertility 391 61,62, possibly by inhibiting ferroptosis . Similarly, supplementation with antioxidants 392 such as coenzyme Q10 could improve the quality of oocytes by reducing apoptosis in 393 a mouse model 63,64, and human ART patients under 40 years old 65, although it remains 394 controversial in clinical practice 66. In addition, ErZhiTianGui decoction (a mixture of 395 ten herbs) was reported to regulat e mitochondrial homeostasis, reduce ROS 396 accumulation, and inhibit ACSL4-mediated lipid peroxidation in aged mice 67. Finally, 397 direct inhibition of ferroptosis in mice by injection of deferoxamine improved fertility 398 68. Generally, antioxidant therapy by supplementation is thought to improve 399 mitochondrial function, thereby improving oocyte quality, embryonic quality, and age-400 related reproductive outcomes. These data suggest ferroptosis could be both a marker 401 of ovarian aging and a target for improved infertility 68. However, inhibition of 402 ferroptosis in a mouse model of PCOS did not affect GC viability 69. Ultimately, the 403 efficacy of direct intervention against ferroptosis in human patients remains unclear. 404 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 15 In c onclusion, ovarian aging is closely associated with declining fertility and 405 reproductive dysfunction, which has long been a challenging issue in reproductive 406 medicine. We have determined that ferroptosis is associated with ovarian aging, and 407 inhibiting ferroptosis could ameliorate some of the features of aging in a mouse model. 408 We also discovered that TXN is down-regulated in aging GCs. TXN inhibits ferroptosis 409 in aging GCs by binding (indirectly) to DNA and activating autophagy/mitophagy 410 genes and BNIP3L-dependent mitophagy, therefore ameliorating ovarian aging. These 411

show how a TXN/BNIP3L-driven mitophagy pathway regulates ferroptosis in 412 ovarian aging, thereby offering a basis for targeted prevention and delay of age-related 413 ovarian dysfunction. 414 415 Acknowledgments 416 This work was supported by grants from the National Natural Science Foundation of 417 China (32270597 to A.P.H). The authors acknowledge the assistance of SUSTech Core 418 Research Facilities. 419 420 Author contributions 421 W.T.Y . designed the study. W.T.Y . and R.X. prepared figures, performed the 422 experiments, analyzed the data, and wrote the first draft of the manuscript. Yi Liu 423 performed experiments, analyzed the data, supervised the study, reviewed , and edited 424 the manuscript. ZQ.H., Z.L. performed the bioinformatic analysis. M.G. analyzed the 425 immunofluorescence. A.P.H. revised the manuscript, acquired funding, and assisted in 426 the bioinformatic analysis. GQ.T. acquired funding and designed and supervised the 427 study. All authors assisted in revising the manuscript , performing experiments, 428 analyzing the data, and/or preparing figures. 429 430 Competing interests 431 The authors declare no competing interests. 432 433 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 16 Ethical approval 434 The Institutional Review Board of the First Affiliated Hospital of Xi’an Jiaotong 435 University approved this study (LLSBPJ -2024-250). All participants provided 436 informed written consent. For animal studies, ethical approval was obtained from the 437 Biomedical Ethical Committee of the Health Science Center of Xi ’an Jiaotong 438 University (XJTUAE2024-716). The use of animals in this research adhere s to ethical 439 guidelines, including minimizing pain and distress, providing proper housing and care, 440 and using the minimum number of animals necessary to obtain valid scientific results. 441 442 Data Accessions 443 The mass spectrometry data generated in this study were submitted to iProX 70 with the 444 accession number: IPX0011098000. CUT&Tag sequence data was deposited in the 445 Gene Expression Omnibus with accession number: GSE297159. This study includes a 446 reanalysis of GSE255690 42, GSE232306 4, and GSE158802 60. 447 448

449 1. Jin, C., Wang, X., Yang, J., Kim, S., Hudgins, A.D., Gamliel, A., Pei, M., Contreras, D., Devos, 450 M., Guo, Q., et al. (2024). Molecular and genetic insights into human ovarian aging from single-451 nuclei multi-omics analyses. Nat Aging. 10.1038/s43587-024-00762-5. 452 2. Broekmans, F.J., Soules, M.R., and Fauser, B.C. (2009). Ovarian aging: mechanisms and 453 clinical consequences. Endocr Rev 30, 465-493. 10.1210/er.2009-0006. 454 3. Isola, J.V .V ., Ocanas, S.R., Hubbart, C.R., Ko, S., Mondal, S.A., Hense, J.D., Carter, H.N.C., 455 Schneider, A., Kovats, S., Alberola-Ila, J., et al. (2024). A single-cell atlas of the aging mouse 456 ovary. Nat Aging 4, 145-162. 10.1038/s43587-023-00552-5. 457 4. Jia, Z.C., Li, Y .Q., Zhou, B.W., Xia, Q.C., Wang, P.X., Wang, X.X., Sun, Z.G., and Guo, Y . 458 (2023). Transcriptomic profiling of human granulosa cells between women with advanced 459 maternal age with different ovarian reserve. J Assist Reprod Genet 40, 2427 -2437. 460 10.1007/s10815-023-02915-8. 461 5. Zhou, C., Guo, Q., Lin, J., Wang, M., Zeng, Z., Li, Y ., Li, X., Xiang, Y ., Liang, Q., Liu, J., et al. 462 (2024). Single-Cell Atlas of Human Ovaries Reveals The Role Of The Pyroptotic Macrophage 463 in Ovarian Aging. Adv Sci (Weinh) 11, e2305175. 10.1002/advs.202305175. 464 6. Winkler, I., Tolkachov, A., Lammers, F., Lacour, P., Daugelaite, K., Schneider, N., Koch, M.L., 465 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 17 Panten, J., Grunschlager, F., Poth, T., et al. (2024). The cycling and aging mouse female 466 reproductive tract at single-cell resolution. Cell 187, 981-998 e925. 10.1016/j.cell.2024.01.021. 467 7. Wang, X., Wang, L., and Xiang, W. (2023). Mechanisms of ovarian aging in women: a review. 468 J Ovarian Res 16, 67. 10.1186/s13048-023-01151-z. 469 8. Kobayashi, H., and Imanaka, S. (2024). Mitochondrial DNA Damage and Its Repair 470 Mechanisms in Aging Oocytes. Int J Mol Sci 25. 10.3390/ijms252313144. 471 9. Yang, Y., Shi, L., Fu, X., Ma, G., Yang, Z., Li, Y., Zhou, Y., Yuan, L., Xia, Y., Zhong, X., et al. 472 (2022). Metabolic and epigenetic dysfunctions underlie the arrest of in vitro fertilized human 473 embryos in a senescent-like state. PLoS Biol 20, e3001682. 10.1371/journal.pbio.3001682. 474 10. Wang, T., Peng, J., Fan, J., Tang, N., Hua, R., Zhou, X., Wang, Z., Wang, L., Bai, Y., Quan, X., 475 et al. (2024). Single -cell multi-omics profiling of human preimplantation embryos identifies 476 cytoskeletal defects during embryonic arrest. Nat Cell Biol 26, 263-277. 10.1038/s41556-023-477 01328-0. 478 11. Alberico, H.C., and Woods, D.C. (2021). Role of Granulosa Cells in the Aging Ovarian 479 Landscape: A Focus on Mitochondrial and Metabolic Function. Front Physiol 12, 800739. 480 10.3389/fphys.2021.800739. 481 12. Blengini, C.S., and Schindler, K. (2023). Follicular communication breakdown in aging ovaries. 482 Nat Aging 3, 636-637. 10.1038/s43587-023-00435-9. 483 13. Carabatsos, M.J., Sellitto, C., Goodenough, D.A., and Albertini, D.F. (2000). Oocyte-granulosa 484 cell heterologous gap junctions are required for the coordination of nuclear and cytoplasmic 485 meiotic competence. Dev Biol 226, 167-179. 10.1006/dbio.2000.9863. 486 14. Wang, H., Huang, Z., Shen, X., Lee, Y., Song, X., Shu, C., Wu, L.H., Pakkiri, L.S., Lim, P.L., 487 Zhang, X., et al. (2024). Rejuvenation of aged oocyte through exposure to young follicular 488 microenvironment. Nat Aging 4, 1194-1210. 10.1038/s43587-024-00697-x. 489 15. Liu, C., Zuo, W., Yan, G., Wang, S., Sun, S., Li, S., Tang, X., Li, Y ., Cai, C., Wang, H., et al. 490 (2023). Granulosa cell mevalonate pathway abnormalities contribute to oocyte meiotic defects 491 and aneuploidy. Nat Aging 3, 670-687. 10.1038/s43587-023-00419-9. 492 16. Yang, Q., Chen, W., Cong, L., Wang, M., Li, H., Wang, H., Luo, X., Zhu, J., Zeng, X., Zhu, Z., 493 et al. (2024). NADase CD38 is a key determinant of ovarian aging. Nat Aging 4, 110-128. 494 10.1038/s43587-023-00532-9. 495 17. Zhang, D., Zhang, X., Zeng, M., Yuan, J., Liu, M., Yin, Y ., Wu, X., Keefe, D.L., and Liu, L. 496 (2015). Increased DNA damage and repair deficiency in granulosa cells are associated with 497 ovarian aging in rhesus monkey. J Assist Reprod Genet 32, 1069-1078. 10.1007/s10815-015-498 0483-5. 499 18. Wang, S., Ren, J., Jing, Y., Qu, J., and Liu, G.H. (2024). Perspectives on biomarkers of 500 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 18 reproductive aging for fertility and beyond. Nat Aging 4, 1697 -1710. 10.1038/s43587-024-501 00770-5. 502 19. Dixon, S.J., Lemberg, K.M., Lamprecht, M.R., Skouta, R., Zaitsev, E.M., Gleason, C.E., Patel, 503 D.N., Bauer, A.J., Cantley, A.M., Yang, W.S., et al. (2012). Ferroptosis: an iron-dependent form 504 of nonapoptotic cell death. Cell 149, 1060-1072. 10.1016/j.cell.2012.03.042. 505 20. Jiang, X., Stockwell, B.R., and Conrad, M. (2021). Ferroptosis: mechanisms, biology and role 506 in disease. Nat Rev Mol Cell Biol 22, 266-282. 10.1038/s41580-020-00324-8. 507 21. Yu, Y., Yan, Y., Niu, F., Wang, Y., Chen, X., Su, G., Liu, Y., Zhao, X., Qian, L., Liu, P., and 508 Xiong, Y . (2021). Ferroptosis: a cell death connecting oxidative stress, inflammation and 509 cardiovascular diseases. Cell Death Discov 7, 193. 10.1038/s41420-021-00579-w. 510 22. Tang, D., Chen, X., Kang, R., and Kroemer, G. (2021). Ferroptosis: molecular mechanisms and 511 health implications. Cell Res 31, 107-125. 10.1038/s41422-020-00441-1. 512 23. Dixon, S.J., and Stockwell, B.R. (2019). The Hallmarks of Ferroptosis. Annual Review of 513 Cancer Biology 3, 35-54. https://doi.org/10.1146/annurev-cancerbio-030518-055844. 514 24. Mazhar, M., Din, A.U., Ali, H., Yang, G., Ren, W., Wang, L., Fan, X., and Yang, S. (2021). 515 Implication of ferroptosis in aging. Cell Death Discov 7, 149. 10.1038/s41420-021-00553-6. 516 25. Zhang, J., Su, T., Fan, Y ., Cheng, C., Xu, L., and LiTian (2024). Spotlight on iron overload and 517 ferroptosis: Research progress in female infertility. Life Sci 340, 122370. 518 10.1016/j.lfs.2023.122370. 519 26. Qin, S., Chi, X., Zhu, Z., Chen, C., Zhang, T., He, M., Gao, M., Zhao, T., Zhang, J., Zhang, L., 520 et al. (2025). Oocytes maintain low ROS levels to support the dormancy of primordial follicles. 521 Aging Cell 24, e14338. 10.1111/acel.14338. 522 27. Lai, Q., Xiang, W., Li, Q., Zhang, H., Li, Y ., Zhu, G., Xiong, C., and Jin, L. (2018). Oxidative 523 stress in granulosa cells contributes to poor oocyte quality and IVF-ET outcomes in women with 524 polycystic ovary syndrome. Front Med 12, 518-524. 10.1007/s11684-017-0575-y. 525 28. Zhang, L., Wang, F., Li, D., Yan, Y ., and Wang, H. (2021). Transferrin receptor -mediated 526 reactive oxygen species promotes ferroptosis of KGN cells via regulating NADPH oxidase 527 1/PTEN induced kinase 1/acyl -CoA synthetase long chain family member 4 signaling. 528 Bioengineered 12, 4983-4994. 10.1080/21655979.2021.1956403. 529 29. Tan, W., Dai, F., Yang, D., Deng, Z., Gu, R., Zhao, X., and Cheng, Y. (2022). MiR -93-5p 530 promotes granulosa cell apoptosis and ferroptosis by the NF-kB signaling pathway in polycystic 531 ovary syndrome. Front Immunol 13, 967151. 10.3389/fimmu.2022.967151. 532 30. Wang, F., Liu, Y., Ni, F., Jin, J., Wu, Y., Huang, Y., Ye, X., Shen, X., Ying, Y., Chen, J., et al. 533 (2022). BNC1 deficiency-triggered ferroptosis through the NF2-YAP pathway induces primary 534 ovarian insufficiency. Nat Commun 13, 5871. 10.1038/s41467-022-33323-8. 535 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 19 31. Wang, C.K., Chen, T.J., Tan, G.Y.T., Chang, F.P., Sridharan, S., Yu, C.A., Chang, Y.H., Chen, 536 Y. J . , C h e n g , L . T. , a n d H w a n g-Verslues, W.W. (2023). MEX3A Mediates p53 Degradation to 537 Suppress Ferroptosis and Facilitate Ovarian Cancer Tumorigenesis. Cancer Res 83, 251-263. 538 10.1158/0008-5472.CAN-22-1159. 539 32. Basuli, D., Tesfay, L., Deng, Z., Paul, B., Yamamoto, Y ., Ning, G., Xian, W., McKeon, F., Lynch, 540 M., Crum, C.P., et al. (2017). Iron addiction: a novel therapeutic target in ovarian cancer. 541 Oncogene 36, 4089-4099. 10.1038/onc.2017.11. 542 33. Picca, A., Faitg, J., Auwerx, J., Ferrucci, L., and D'Amico, D. (2023). Mitophagy in human 543 health, ageing and disease. Nat Metab 5, 2047-2061. 10.1038/s42255-023-00930-8. 544 34. Carvalho, K.F., Machado, T.S., Garcia, B.M., Zangirolamo, A.F., Macabelli, C.H., Sugiyama, 545 F.H.C., Grejo, M.P., Augusto Neto, J.D., Tostes, K., Ribeiro, F.K.S., et al. (2020). Mitofusin 1 546 is required for oocyte growth and communication with follicular somatic cells. FASEB J 34, 547 7644-7660. 10.1096/fj.201901761R. 548 35. de Bruin, J.P., Dorland, M., Spek, E.R., Posthuma, G., van Haaften, M., Looman, C.W., and te 549 Velde, E.R. (2004). Age -related changes in the ultrastructure of the resting follicle pool in 550 human ovaries. Biol Reprod 70, 419-424. 10.1095/biolreprod.103.015784. 551 36. Zhang, Y ., Bai, J., Cui, Z., Li, Y ., Gao, Q., Miao, Y ., and Xiong, B. (2023). Polyamine metabolite 552 spermidine rejuvenates oocyte quality by enhancing mitophagy during female reproductive 553 aging. Nat Aging 3, 1372-1386. 10.1038/s43587-023-00498-8. 554 37. Kagan, V .E., Mao, G., Qu, F., Angeli, J.P., Doll, S., Croix, C.S., Dar, H.H., Liu, B., Tyurin, V .A., 555 Ritov, V .B., et al. (2017). Oxidized arachidonic and adrenic PEs navigate cells to ferroptosis. 556 Nat Chem Biol 13, 81-90. 10.1038/nchembio.2238. 557 38. Jingyun, Z., Zhaoyan, N., Xianglong, K., Liqian, Na, Z., Lvcuiting, Lina, G., and Haifeng, W. 558 (2023). Study on the relationship between SlRTl and oxidative stress in aged patients 559 undergoing in vitro fertilization and embryo transfer cycles. J Gynecol Obstet Hum Reprod 52, 560 102516. 10.1016/j.jogoh.2022.102516. 561 39. Bai, L., Yan, F., Deng, R., Gu, R., Zhang, X., and Bai, J. (2021). Thioredoxin -1 Rescues 562 MPP(+)/MPTP-Induced Ferroptosis by Increasing Glutathione Peroxidase 4. Mol Neurobiol 58, 563 3187-3197. 10.1007/s12035-021-02320-1. 564 40. Yang, W.S., SriRamaratnam, R., Welsch, M.E., Shimada, K., Skouta, R., Viswanathan, V.S., 565 Cheah, J.H., Clemons, P.A., Shamji, A.F., Clish, C.B., et al. (2014). Regulation of ferroptotic 566 cancer cell death by GPX4. Cell 156, 317-331. 10.1016/j.cell.2013.12.010. 567 41. Hernandez Mora, J.R., Buhigas, C., Clark, S., Del Gallego Bonilla, R., Daskeviciute, D., 568 Monteagudo-Sanchez, A., Poo-Llanillo, M.E., Medrano, J.V ., Simon, C., Meseguer, M., et al. 569 (2023). Single-cell multi-omic analysis profiles defective genome activation and epigenetic 570 reprogramming associated with human pre -implantation embryo arrest. Cell Rep 42, 112100. 571 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 20 10.1016/j.celrep.2023.112100. 572 42. Wu, M., Tang, W., Chen, Y., Xue, L., Dai, J., Li, Y., Zhu, X., Wu, C., Xiong, J., Zhang, J., et al. 573 (2024). Spatiotemporal transcriptomic changes of human ovarian aging and the regulatory role 574 of FOXP1. Nat Aging 4, 527-545. 10.1038/s43587-024-00607-1. 575 43. Zhou, N., Yuan, X., Du, Q., Zhang, Z., Shi, X., Bao, J., Ning, Y ., and Peng, L. (2023). FerrDb 576 V2: update of the manually curated database of ferroptosis regulators and ferroptosis -disease 577 associations. Nucleic Acids Res 51, D571-D582. 10.1093/nar/gkac935. 578 44. Liang, D., Feng, Y ., Zandkarimi, F., Wang, H., Zhang, Z., Kim, J., Cai, Y ., Gu, W., Stockwell, 579 B.R., and Jiang, X. (2023). Ferroptosis surveillance independent of GPX4 and differentially 580 regulated by sex hormones. Cell 186, 2748-2764 e2722. 10.1016/j.cell.2023.05.003. 581 45. Nguyen, C.T.N., Kim, S.M., and Kang, Y .P. (2022). Mass spectrometry-based approaches to 582 explore metabolism regulating ferroptosis. BMB Rep 55, 413 -416. 583 10.5483/BMBRep.2022.55.9.109. 584 46. Friedmann Angeli, J.P., Schneider, M., Proneth, B., Tyurina, Y .Y ., Tyurin, V .A., Hammond, V .J., 585 Herbach, N., Aichler, M., Walch, A., Eggenhofer, E., et al. (2014). Inactivation of the ferroptosis 586 regulator Gpx4 triggers acute renal failure in mice. Nat Cell Biol 16, 1180 -1191. 587 10.1038/ncb3064. 588 47. Lee, E.H., Han, S.E., Park, M.J., Kim, H.J., Kim, H.G., Kim, C.W., Joo, B.S., and Lee, K.S. 589 (2018). Establishment of Effective Mouse Model of Premature Ovarian Failure Considering 590 Treatment Duration of Anticancer Drugs and Natural Recovery Time. J Menopausal Med 24, 591 196-203. 10.6118/jmm.2018.24.3.196. 592 48. Zhao, Y ., Li, Y ., Zhang, R., Wang, F., Wang, T., and Jiao, Y . (2020). The Role of Erastin in 593 Ferroptosis and Its Prospects in Cancer Therapy. Onco Targets Ther 13, 5429 -5441. 594 10.2147/OTT.S254995. 595 49. Yan, R., Lin, B., Jin, W., Tang, L., Hu, S., and Cai, R. (2023). NRF2, a Superstar of Ferroptosis. 596 Antioxidants (Basel) 12. 10.3390/antiox12091739. 597 50. Zhang, X., Du, L., Qiao, Y ., Zhang, X., Zheng, W., Wu, Q., Chen, Y ., Zhu, G., Liu, Y ., Bian, Z., 598 et al. (2019). Ferroptosis is governed by differential regulation of transcription in liver cancer. 599 Redox Biol 24, 101211. 10.1016/j.redox.2019.101211. 600 51. Martens, S. (2016). No ATG8s, no problem? How LC3/GABARAP proteins contribute to 601 autophagy. J Cell Biol 215, 761-763. 10.1083/jcb.201611116. 602 52. Kimura, S., Noda, T., and Yoshimori, T. (2007). Dissection of the autophagosome maturation 603 process by a novel reporter protein, tandem fluorescent -tagged LC3. Autophagy 3, 452-460. 604 10.4161/auto.4451. 605 53. Novak, I., Kirkin, V ., McEwan, D.G., Zhang, J., Wild, P., Rozenknop, A., Rogov, V ., Lohr, F., 606 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 21 Popovic, D., Occhipinti, A., et al. (2010). Nix is a selective autophagy receptor for 607 mitochondrial clearance. EMBO Rep 11, 45-51. 10.1038/embor.2009.256. 608 54. Jabbar, S., Mathews, P., Wang, X., Sundaramoorthy, P., Chu, E., Piryani, S.O., Ding, S., Shen, 609 X., Doan, P.L., and Kang, Y . (2022). Thioredoxin-1 regulates self-renewal and differentiation of 610 murine hematopoietic stem cells through p53 tumor suppressor. Exp Hematol Oncol 11, 83. 611 10.1186/s40164-022-00329-3. 612 55. Ando, K., Hirao, S., Kabe, Y ., Ogura, Y ., Sato, I., Yamaguchi, Y ., Wada, T., and Handa, H. (2008). 613 A new APE1/Ref -1-dependent pathway leading to reduction of NF -kappaB and AP -1, and 614 activation of their DNA -binding activity. Nucleic Acids Res 36, 4327 -4336. 615 10.1093/nar/gkn416. 616 56. Karimpour, S., Lou, J., Lin, L.L., Rene, L.M., Lagunas, L., Ma, X., Karra, S., Bradbury, C.M., 617 Markovina, S., Goswami, P.C., et al. (2002). Thioredoxin reductase regulates AP-1 activity as 618 well as thioredoxin nuclear localization via active cysteines in response to ionizing radiation. 619 Oncogene 21, 6317-6327. 10.1038/sj.onc.1205749. 620 57. Wei, S.J., Botero, A., Hirota, K., Bradbury, C.M., Markovina, S., Laszlo, A., Spitz, D.R., 621 Goswami, P.C., Yodoi, J., and Gius, D. (2000). Thioredoxin nuclear translocation and interaction 622 with redox factor-1 activates the activator protein-1 transcription factor in response to ionizing 623 radiation. Cancer Res 60, 6688-6695. 624 58. Das, K.C., and Muniyappa, H. (2010). c -Jun-NH2 terminal kinase (JNK) -mediates AP -1 625 activation by thioredoxin: phosphorylation of cJun, JunB, and Fra -1. Mol Cell Biochem 337, 626 53-63. 10.1007/s11010-009-0285-0. 627 59. Kaya-Okur, H.S., Wu, S.J., Codomo, C.A., Pledger, E.S., Bryson, T.D., Henikoff, J.G., Ahmad, 628 K., and Henikoff, S. (2019). CUT&Tag for efficient epigenomic profiling of small samples and 629 single cells. Nat Commun 10, 1930. 10.1038/s41467-019-09982-5. 630 60. Llonch, S., Barragan, M., Nieto, P., Mallol, A., Elosua-Bayes, M., Lorden, P., Ruiz, S., Zambelli, 631 F., Heyn, H., Vassena, R., and Payer, B. (2021). Single human oocyte transcriptome analysis 632 reveals distinct maturation stage-dependent pathways impacted by age. Aging Cell 20, e13360. 633 10.1111/acel.13360. 634 61. Yang, Q., Cong, L., Wang, Y., Luo, X., Li, H., Wang, H., Zhu, J., Dai, S., Jin, H., Yao, G., et al. 635 (2020). Increasing ovarian NAD(+) levels improve mitochondrial functions and reverse ovarian 636 aging. Free Radic Biol Med 156, 1-10. 10.1016/j.freeradbiomed.2020.05.003. 637 62. Bertoldo, M.J., Listijono, D.R., Ho, W.J., Riepsamen, A.H., Goss, D.M., Richani, D., Jin, X.L., 638 Mahbub, S., Campbell, J.M., Habibalahi, A., et al. (2020). NAD(+) Repletion Rescues Female 639 Fertility during Reproductive Aging. Cell Rep 30, 1670 -1681 e1677. 640 10.1016/j.celrep.2020.01.058. 641 63. Zhang, M., ShiYang, X., Zhang, Y ., Miao, Y ., Chen, Y ., Cui, Z., and Xiong, B. (2019). Coenzyme 642 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 22 Q10 ameliorates the quality of postovulatory aged oocytes by suppressing DNA damage and 643 apoptosis. Free Radic Biol Med 143, 84-94. 10.1016/j.freeradbiomed.2019.08.002. 644 64. Ben-Meir, A., Burstein, E., Borrego-Alvarez, A., Chong, J., Wong, E., Yavorska, T., Naranian, 645 T., Chi, M., Wang, Y., Bentov, Y., et al. (2015). Coenzyme Q10 restores oocyte mitochondrial 646 function and fertility during reproductive aging. Aging Cell 14, 887-895. 10.1111/acel.12368. 647 65. Florou, P., Anagnostis, P., Theocharis, P., Chourdakis, M., and Goulis, D.G. (2020). Does 648 coenzyme Q(10) supplementation improve fertility outcomes in women undergoing assisted 649 reproductive technology procedures? A systematic review and meta -analysis of randomized -650 controlled trials. J Assist Reprod Genet 37, 2377-2387. 10.1007/s10815-020-01906-3. 651 66. Rodriguez-Varela, C., and Labarta, E. (2021). Does Coenzyme Q10 Supplementation Improve 652 Human Oocyte Quality? Int J Mol Sci 22. 10.3390/ijms22179541. 653 67. Zhicheng, J., Yongqian, L., Peixuan, W., Kai, Y ., Mengyu, S., Wen, C., Qihui, L., and Ying, G. 654 (2024). ErZhiTianGui Decoction alleviates age -related ovarian aging by regulating 655 mitochondrial homeostasis and inhibiting ferroptosis. J Ovarian Res 17, 12. 10.1186/s13048-656 023-01341-9. 657 68. Zeng, W., Wang, F., Cui, Z., Zhang, Y ., Li, Y ., Li, N., Mao, Z., Zhang, H., Liu, Y ., Miao, Y ., et 658 al. (2025). Inhibition of ferroptosis counteracts the advanced maternal age -induced oocyte 659 deterioration. Cell Death Differ. 10.1038/s41418-025-01456-0. 660 69. Cai, Z., He, S., Liu, R., Zhou, L., and Zhao, L. (2022). Plumbagin rescues the granulosa cell's 661 pyroptosis by reducing WTAP -mediated N6 -methylation in polycystic ovary syndrome. J 662 Ovarian Res 15, 126. 10.1186/s13048-022-01058-1. 663 70. Ma, J., Chen, T., Wu, S., Yang, C., Bai, M., Shu, K., Li, K., Zhang, G., Jin, Z., He, F., et al. 664 (2019). iProX: an integrated proteome resource. Nucleic Acids Res 47, D1211 -D1217. 665 10.1093/nar/gky869. 666 667 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint GSH/GSSG ratio Old Y oung p=0.026 Ratio GSH GSSG Old Y oung Old Y oung p=0.36 p=0.039 Arachidonic acid Intensity Adrenic acid Old Y oung Old Y oung p=0.047p=0.049 0.2 0.60.2 0.6 −log10(p-value)Down (Young) Up (Old) Microbial metabolism in diverse environments Biosynthesis of amino acids Carbon metabolism Tryptophan metabolism Biosynthesis of various antibiotics Alanine, aspartate and glutamate metabolism Tropane, piperidine and pyridine alkaloid biosynthesis Glycerophospholipid metabolism Ferroptosis Valine, leucine and isoleucine biosynthesis Caffeine metabolism Butanoate metabolism Ratio 2.0 1.8 1.6 Count 2 4 6 8 10 12 CoA PE-AA-O-OHPE-AAAA-CoA PE-AA-OH No change Down in old Up in old Not in data AA/AdA ACSL4 Ferroptosis 0 1 2 3 4 5 −4 40 Log2(fold-change) −log10 (p-value) Old Up (Old) Down(Young) No change q Mitobright LT FerroOrange Old GC Middle GC Young GC Old GC Middle GC Young GC JC-1aggregatesJC-1monoemrs Merge Merge h i g 15 10 5 0 Normalized expression Sum of ferroptosis suppressors 15.0 12.5 10.0 7. 5 5.0 2.5 0.0 Sum of ferroptosis 0.0 0.5 1.0 Mean expression in group drivers 20 60 100 Fraction of cells PGRMC1 ATF4 SAT1 RPL8 KDM6B ATF3 ZFAS1 CIRBP GABARAPL1 LPIN1 MEG3 NR5A2 COPZ1 PPP1R13L TMSB4X PARP1 CDH1 NOS2 BEX1 HMOX1 SLC16A1 Young Old Driver SuppressorGCsd e f Young Old kDa b p21CDKN1 LMNB1 A Old GC Middle GC Young GC 72 34 16 55 27 43 ca SA-β-gal Middle GOld GC C Young GC 50μm 50μm 50μm GAPDH in group (%) Figure 1 j p MergeDAPI Oxidized C11-BODIPYC11-BODIPY Young GC Middle GC Old GC k m l n o Cystine Cysteine GCL γGC GSS GSH GSSG LPCAT3 ALOX15 LysoPE GPX4 Young GSH Adrenic acid Young Old LMNB1 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 23 Figure 1. Aging induces senescence and ferroptosis in human ovarian granulosa 668 cells 669 a Representative images of senescence -associated β-galactosidase (SA-β-gal) staining in GCs 670 (Granulosa cells) from old (n=6, >36 years), m iddle-aged (n=6, 30-36), or young (n=6, <30) 671 women. 672 b Western blot of LMNB1 (Lamin B1), p21CDKN1A, and GAPDH in GCs from old, middle-aged, 673 and young women. The experiment was performed six times with similar results. 674 c Quantitation of the protein levels of LMNB1 and p21CDNK1A based on western blots relative to 675 GAPDH. The p -value is from a one-way analysis of variance (ANOV A) and Tukey's HSD 676 (Honestly Significant Difference) test (n=6). 677 d tSNE (t-stochastic neighbor estimation) embedding of scRNA-seq data from young and old GCs. 678 Cells are colored by the young and old classification, as in the original study. Data is from a 679 reanalysis of Ref. 42. 680 e Bubble plot showing the mean expression in the scRNA-seq data for a selection of ferroptosis-681 driver and suppressor genes. 682 f tSNE plots of scRNA-seq data (as in panel d) colored by the sum of expression of ferroptosis-683 drivers and suppressors, as defined in Ref. 43. 684 g Violin plot of the concentration of the end-product of lipid peroxidation , malondialdehyde 685 (MDA), in GCs from old (n=30), middle (n=28), and young (n=22) women. The error bars and 686 statistical analysis were measured by one-way ANOV A and Tukey's HSD test. 687 h The levels of glutathione (GSH) were measured by colorimetric assay for GCs from old (n=30), 688 middle (n=28), and young (n=22) women. The error bars and statistical analysis were measured 689 by one-way ANOV A and Tukey's HSD test. 690 i Microscope images comparing ferrous ion (Fe2+; FerroOrange) levels and active mitochondria 691 (Mitobright LT) in old, middle, and young patients. Experiments were performed three times 692 with similar results. 693 j Microscope images of mitochondrial membrane potential (MMP) as detected via JC-1 694 monomers (low MMP; green) and aggregates (high MMP; red) in GCs from old, middle-aged, 695 and young women. Experiments were performed three times with similar results. 696 k Vo l c a n o p l o t o f m e t a b o l i t e s i n y o u n g a n d o l d G C s . S i g n i f i c a n c e w a s d e t e r m i n e d a s an absolute 697 fold-change > 2 and a Bonferroni-Hochberg adjusted p-value of < 0.05. 698 l Ontology analysis of the significantly up- and down-regulated metabolite-related terms. 699 m Schematic map for Ferroptosis based on KEGG Ferroptosis pathway (hsa04216). Metabolites 700 are in circles; proteins are in square boxes. 701 n Violin plots showing quantitation of arachidonic acid, adrenic acid, GSH, and GSSG for young 702 and old GCs. 703 o Violin plot showing the ratio of GSG/GSSG metabolites from the mass spec data. 704 p Microscope images of lipid peroxide levels in GCs as measured by C11 -BODIPY 581/591 705 fluorescent probes for the peroxidized (green) and total (red) lipids. The experiment was 706 performed three times with similar results. 707 q Quantification of the ratio of lipid peroxidation levels by FACS in GCs from old, middle, and 708 young GCs measured using C11-BODIPY . The error bars indicate the mean ± standard deviation 709 (SD). Statistics are from a two-sided Kruskal-Wallis test. 710 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint a d c f g h Figure 2 i Young mice SPSS Fer-1 b Old mice SPSS Fer-1 Erastin SPSS POF mice Fer- e 1 SPSS Erastin Fer-1 Old mice Young mice POF mice Old mice Young mice POF mice SPSS Fer-1 SPSS Erastin SPSS Fer-1 100μm 100μm 100μm 100μm 100μm 100μm .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 24 Figure 2. Increased ferroptosis in GCs of aged and premature ovarian failure 711 model mice 712 a Schematic of the mouse models used in this study. Young (52 weeks) mice 713 were treated with Fer -1 (Ferroptosis inhibitor; 1 mg/kg) or Erastin ( 20 mg/kg; Ferroptosis 714 inducer) via intraperitoneal injection once a day for 21 days. To induce POF (premature ovarian 715 failure), 12-week-old mice were administered a single dose of busulfan and cyclophosphamide. 716 Six mice were used for each group. 717 b Hematoxylin and eosin-stained ovary sections from 12- or 52-week-old mice treated with SPSS 718 (vehicle), Fer-1, Erastin, or BuCy. Scale bar = 625 μm. 719 c Ovary weight of 12- and 52-week-old mice treated with Erastin, or in POF model mice (n=6). 720 Statistics are from an unpaired two-tailed Student’s t-test. 721 d Bar chart of the number of follicles per ovary for the indicated mice and treatments for primary, 722 secondary, antral, and atretic follicles (n=6 per group). Statistics are from an unpaired two-tailed 723 Student’s t-test. 724 e Example images of litter size from 12 -week and 52 -week mice treated with the indicated 725 chemicals. 52-week-old SPSS (vehicle) and 12 -week-old BuCy-treated mice did not produce 726 offspring. 727 f Bar chart of average litter sizes of Erastin, BuCy, 12-week-old (young) POF model mice treated 728 with F er-1 or SPSS (vehicle), and 52 -week-old mice treated with Fer-1 (n=6 per group). 729 Statistics are from an unpaired two-tailed Student’s t-test. 730 g Bar chart of the levels of estradiol and progesterone for mice with the indicated age and 731 treatments (n=6 per group). Statistics are from an unpaired two-tailed Student’s t-test. 732 h Images of 2 -cell stage embryos collected from 12 and 52 week-old mice treated with SPSS 733 (vehicle), Fer-1, Erastin, or POF model mice treated with Fer-1. 734 i Bar chart showing the 2-cell cleavage rate (n=6) for embryos from the indicated age mice and 735 treatments. Statistics are from a two-sided Kruskal-Wallis test. 736 737 738 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint b Figure 3 h Young (8W) Old (52W) TXN SLC7A11 NOX4GPX4 e SLC7A11 GPX4 Old Middle Young 16 NOX4 GAPDH 72 TXN 55 72 27 16 10 55 34 43 kDa c f TXN/ g Young GC Middle GC Old GC DAPI SLC7A11/DAPI GPX4/DAPI NOX4/DAPI ELOVL5/DAPI 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm a Depleted ovarian reserve Normal ovarian reserve Ferroptosis drivers NES=1.41 q-value=3.7e-2 0.2 0.0 Ferroptosis suppressors NES=−1.47-0.2 q-value=3.1e-2 Enrichment Score Rank Gene rank d DOR NOR NNMT GPX4 TXN ELOVL5 SLC7A11 NOX4 -3 3 0 Z-score Driver Suppressor ELOVL5 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 25 Figure 3. The TXN antioxidant system is impaired in GCs of aged human and 739 mouse ovaries 740 a GSEA for drivers and suppressors for old women (n=6) versus young women (n=6) . Data is 741 from a reanalysis of GSE232306 4. Ferroptosis driver or suppressor genes, as defined in FerrDB 742 43. 743 b RT-qPCR for the ferroptosis suppressors TXN, GPX4, and SLC7A11. GCs were obtained from 744 old women (n=10, >36 years), middle -aged women (n=18, 30~36), and young women (n=6, 745 <30). The error bars indicate the mean ±SD. Statistics are from an independent samples two-746 sided Student’s t-test. 747 c RT-qPCR for the ferroptosis drivers ELOVL5, NOX4, and the ferroptosis activator NNMT in 748 GCs. The error bars indicate the mean ±SD. Statistics are from an independent samples two-749 sided Student’s t-test. 750 d Heatmap of selected differentially regulated ferroptosis drivers and suppressors in DOR 751 (depleted ovarian reservoir) and NOR (normal ovarian reservoir) human samples. Data is from 752 a reanalysis of GSE232306 4. Ferroptosis driver or suppressor genes, as defined in FerrDB 43, 753 and the ferroptosis activator NNMT. 754 e Western blot of SLC7A11, GPX4, TXN, NOX4, and GAPDH in GCs from old, middle, and 755 young women. 756 f Bar chart of quantitated levels of SLC7A11, GPX4, TXN, and NOX4 in GCs from old, medium, 757 and young women, relative to GAPDH and relative to the old GC samples. The experiment was 758 performed 6 times with similar results. 759 g Immunofluorescence of TXN, SLC7A11, GPX4, NOX4 and ELOVL5, in GCs from old, middle, 760 and young women (n=3 per group). 761 h Immunohistochemistry of TXN, SLC7A11, GPX4, and NOX4 in sections of mouse ovaries at 762 52 weeks old and 8 weeks old (n=3 per group). 763 764 765 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint c Figure 4 17 34 kDa 43 27 GAPD a H TXN shLUC#1 shTXN# 1 shTXN#2shLUC#2 17 34 Empty OE#1TXN OE#1 TXN OE#2 GAPDH TXN kDa 43 27 shLU b C shTXN Empty OE TXN OE H2O2-Empty OE H2O2-TXN OE Merge Mitobright LT FerroOrange 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm 20μm shLUd C shTXN Empty OE TXN OE H2O2 PBS 50μm 50μm 50μm 50μm 50μm 50μm 50μm 50μm Empty OE#2 e 0 200 400 600 SLC7A11 H2O2 (μM) GPX4 NOX4 NRF2 TXN YAP BCL2 55 55 55 55 10 27 55 0 0.5 1 2 GAPDH 34 shTXN (μg) TXNRD1 72 kDa 72 72 72 17 55 72 72 34 43 h DMSO PX-12 TXNIP-IN-1 50μm 50μm 50μm 50μm 50μm50μm DMSO TXNIP-IN-1 TXNIP-IN-1+NADPH H2O2 PBS f DMS g O PX-12 TXNIP-IN-1 H2O2 NADPH + - - - - - + - - - - - + - - - - - + - - - - + + - + - + - - + - + + 10 TXN 10 GPX4 72 LMNB1 55 SLC7A11 GAPDH 34 kDa 17 55 27 72 43 ACSL4 72 90 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 26 Figure 4. TXN delays the aging phenotype in ovarian GCs by inhibiting 766 ferroptosis 767 a Western blot of TXN protein levels in KGN cells transfected with shRNAs against TXN 768 (shTXN#1 and shTXN#2) or overexpressed (TXN OE#1 and TXN OE#2), and their appropriate 769 controls. 770 b Comparison of the ferrous ion (Fe2+) level in TXN knockdown and overexpressing KGN cells 771 by the Fe2+ indicator FerroOrange fluorescent probe (red), and MitoBright LT Green. 772 c Lipid peroxidation levels were measured with the C11 -BODIPY 581/591 dye by FACS 773 (green/red for oxidized/reduced lipids) in KGN cells transfected with an shRNA targeting TXN 774 or LUC as a control, or a plasmid containing a TXN overexpression construct. Cells were also 775 treated with H2O2 or PBS. The error bars indicate the mean±SD, and the p-value is from a two-776 sided Kruskal-Wallis test. 777 d SA-β-gal staining in TXN-silenced and overexpressed KGN cells treated with H 2O2 or PBS. 778 Scale bar = 50 μm. 779 e Western blots for the indicated proteins in KGN cells treated with increasing concentrations of 780 H2O2 or transfected with increasing amounts of an shRNA targeting TXN. 781 f Quantitation of oxidized lipids using C11 -BODIPY dye (left plot) and total ROS (reactive 782 oxygen species) (right plot) in KGN cells treated with the indicated factors : PX -12 (TXN 783 inhibitor), TXNIP -IN-1 (TXN activator), NADPH, or H 2O2. The error bars indicate the 784 mean±SD, and the p -value is from a two -sided Kruskal -Wallis test. The experiment was 785 performed in biological quadruplicate. 786 g Western blot of the indicated proteins in KGN cells treated with DMSO as a control, H2O2, PX-787 12 (TXN inhibitor), TXNIP-IN-1 (TXN activator), or NADPH. 788 h SA-β-gal staining in KGN cells treated with DMSO, PX-12, TXN-IP-1, with PBS as a control, 789 or H2O2. Scale bar = 50 μm. 790 791 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 5μm 5μm 5μm 5μm 5μm Mitophagy Red MeLyso Tracker Green rge Young GC Old GC 5μm %M i t o c h o n d r i a Healthy Mitochondria Damged Mitochondria %M i t o c h o n d r i a 100 80 60 40 20 0 8W52W j k p=5.4e-5*** 1.2 0.9 0.6 0.3 Mitochondria Size (%)0.0 8W52W p=0.0046** Number of Mitochondria per field 50 40 30 20 10 0 8W52 W Figure 5 g i Young GC Old GC BNIP3L DAPI Merge 10μm 10μm 10μm 10μm 10μm 10μm e Old Young 4 3 2 1 0 −log10 (p- value) −4 0 4 Log2(fold-change) BNIP3L HRAS SQSTM1 TOMM40L OPTN a c f b q = 1.2e-3 NES = -1.69 0.0 -0.1 -0.2 -0.3 -0.4 Enrichment Score Mitophagy Pathway Rank YoungOld BCL2L1 RAB7A HRAS CTTN BNIP3L SQSTM1 CALCOCO2 OPTN −2 0 2 Log2(fold-change) Old Y oung Youn gG C - Em ptd y kDaOld GC- Emp ty 17 10 TXN Youn gG C - TXN OldGC- TX N BNIP3L LC3I LC3II GAPDH 43 34 17 10 43 34 shLUC shTXN empty OE TXN OE GFPRFPMerge Autophagy turnover eff h iciency p=0.027* p=0.003** 80 60 40 20 0 Old mice+ Fer- shLUC shTXN empty OETXN OE  RFP GFP / RFP GFP (%) 1 PO Fm ice +Fe r-1 POF mice Yo un gmice+ Erast in 120 100 80 60 40 20 0 Healthy Mitochondria Damaged Mitochondria 5μm5μm5μm5μm 5μm5μm 5μm5μm 5μm5μm 5μm5μm p=0.17 n.s. p=0.026* 40000 20000 Old MiddleYoung GABARAPL1 Intensity .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 27 Figure 5. TXN deficiency blocks mitophagy in aged ovarian GCs 792 a Vo l c a n o p l o t o f 4D-DIA proteomics for human ovarian GCs from young women (36 years old, n=6). A protein was considered differentially abundant if 794 its p-value was 1.5. 795 b GSEA for genes in the mitophagy pathway in old and young individuals. Proteins were ranked 796 by their fold-change from old to young. 797 c Heatmap of selected mitophagy-related proteins. 798 d Western blot for TXN, BNIP3L, LC31/II and GAPDH as a control for Young and Old GCs, and 799 GCs transfected with a plasmid overexpressing TXN. 800 e Immunofluorescence of BNIP3L in GCs from old and young women. 801 f Dot plot of the mass spec intensity of BNIP3L in young, middle, and aged GCs. The error bars 802 indicate the mean±SD, and the p-value is from a two-sided Mann-Whitney U test 803 g Images of autophagic activity based on a GFP-RFP-LC3B reporter. Green puncta indicate 804 nascent auto phagosomes, yellow indicates autophagosomes, and red puncta mark 805 autolysosomes. The experiment used KGN cells transfected with an shRNA targeting LUC or 806 TXN, or an Empty or TXN overexpressing vector. 807 h Quantification of autophagic flux across different conditions (as in panel g), represented by the 808 number of colocalized RFP/GFP voxels per cell. Statistical significance was assessed using a 809 one-sided unpaired Student’s t-test. 810 i Fluorescence imaging of mitophagy (Mitophagy Red; becomes red when pH drops in 811 mitochondria) and LysoTracker (lysosomes) in GCs from old and young women. 812 j The number of mitochondria (left plot), size (middle plot) and percentage of healthy/damage 813 mitochondria (right plot) as estimated from transmission electron microscopy (TEM) in mice 814 ovary GCs from young mice (8W) and old mice (52W) (See also Supplementary Figure 7f 815 and g). The error bars indicate the mean ±SD (n=6 for each group). Statistics are from an 816 unpaired two-sided Student’s t-test. 817 k Dot plot of t he number of mitochondria (left plot), and bar chart of the percentage of 818 healthy/damage mitochondria (right plot) , as estimated from TEM , in mice ovary GCs from 819 young mice (8W) treated with Erastin, POF model mice treated with F er-1 or SPSS and old 820 mice (52W) treated with Fer-1. The error bars indicate the mean ±SD (n=3 for each group). 821 Statistics are from an unpaired two-sided Student’s t-test. 822 823 824 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint Figure 6 i shTXN shLUC TXN Empty TXN b BNIP3L LC3I LC3II GAPDH 43 34 Empty Empty H2O2H2O2 TXN TXN 43 17 kDa c 10 34 17 1043 34 10 kDa 17 43 34 17 10 j FerroOrange Mitobright LT Merge shBNIP3L Empty OE shBNIP3L TXN OE shScramble TXN OE Empty OE shScramble shScramble shBNIP3L TXN OE TXN BNIP3L SLC7A11 + - l - - + - + - + - + + 55 72 34 34 34GPX4 10 kDa 17 43 72 27 5543 k 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm TXN BNIP3L 10μm LC3I LC3II GAPDH TSS-10 0d e kbp 01 0 kbp TXN KGN cells 10 50-10 0 kbpkbp 5 4 3 2 1 -100 -50 kbp 50 100 kbp8% 16% 24% 100 200-200 -100 kbp kbpGene desert f CCA G CG TG A AT T TAC GC A AT AGT TC TCGG CT GTA ACCAG G T TT GGCAGCC CACTAA C GGCGGAC ATTAG AT T CGCG CTTT ZFX 1e-67 C TTAA GTTAGGT CGGA CACAAAGC -1k TXN 1k C TAC C TC TCG GAT AGG AG AGTTTCTA GG CCGG CAAT CATA 17963 peaks 100 200 300 Tag density AT CA AGTGT C ETS-family 1e-52 YY-family 1e-49 TATA/ZBTB 1e-48 C GC CG G TA TAA AT AGT TC GG TGG TC C A AC C g Ubiquitin mediated proteolysis(hsa04120) Protein processing in endoplasmic reticulum(hsa04141) RNA transport(hsa03013) Cell cycle(hsa04110) Mitophagy - animal(hsa04137) Human T-cell leukemia virus 1 infection(hsa05166) p53 signaling pathway(hsa04115) Platinum drug resistance(hsa01524) 0.0 2.5 -log10(q-value) 10kbp 51 1kbp 109 SBNO2 51 chr8:26368499-26413609 10kbp 51TXN (KGN cells) BNIP3L BNIP3LSDAD1P1 SDAD1P1 BNIP3L SDAD1P1 BNIP3L chr19:1093025-1112305 1kbp TXN (KGN cells) 109 POLR2E GPX4 GPX4 SBNO2 10kbp 51 chr9:110240063-110274307 TXN (KGN cells) TXN TXN h GAPDH ACSL4 Autophagy - animal(hsa04140) Peroxisome(hsa04146) p=4.8e4*** Old Middle a Young BNIP3L p=0.093 n.s. p=0.026* 20000 Intensity 0 Intensiy Intensity Intensity Intensity 1e-258 AP-1 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 28 Figure 6. TXN regulates the expression of BNIP3L and subsequent ferroptosis in 825 GCs 826 a Protein level of GABARAPL1 in GCs from old, middle-aged, and young human GCs 827 from mass spec data. Significance is from a two-sided Mann-Whitney U test. 828 b Western blot for TXN, BNIP3L , and LC3I/II in TXN-knockdown and overexpressing KGN 829 cells, and the relative expression levels relative to GAPDH (bar chart below). 830 c Western blot for TXN, BNIP3L, and LC3I/II in TXN overexpress ing KGN cells treated with 831 H2O2 and analysis of relative protein levels relative to GAPDH (bar chart below). 832 d Heatmap and pileup of TXN binding to DNA in KGN cells. 833 e Genome distribution of the TXN CUT&Tag peaks. Peaks were annotated to the nearest TSS and 834 allocated to bins either 5’ (negative numbers) or 3’ relative to the TSS. A random background is 835 shown in grey for comparison. 836 f TF motif discovery at TXN-bound loci. The TF family and q-value are indicated on the left. 837 g Gene ontology for KEGG pathways for TXN-bound genes. 838 h Genome views around the promoters of BNIP3L, GPX4, and TXN. 839 i Dual luciferase reporter assay for the BNIP3L promoter region cloned in front of luciferase in 840 TXN-knockdown and overexpressing KGN and 293T cells. Significance is from a two-way 841 ANOV A and Tukey test. 842 j Fe2+ and lipid ROS levels in BNIP3L knockdown and TXN overexpressing KGN cells. 843 k Quantitation of oxidized lipids using C11-BODIPY dye in KGN cells transfected with shRNAs 844 against BNIP3L or a scrambled control, or a TXN, or an Empty overexpression vector. The error 845 bars indicate the mean values±SDs, and the p-value is from an unpaired two-tailed Student’s 846 t-test. 847 l Western blot for TXN, BNIP3L, SLC7A11, GPX4, and ACSL4 in KGN cells transfected with 848 an shRNA targeting BNIP3L or a scrambled control, or with a TXN overexpression vector. 849 850 851 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint p=0.41 SLC7A11 p=0.65 10 5 0 MII OldYoung GV OldYoung MII OldYoung GV OldYoung ELOVL5 p=0.54 10 5 0 p=0.74p=0.33 TXN p=0.29 10 5 0 MII OldYoung GV OldYoung MII OldYoung GV OldYoung BNIP3L p=0.064 10 5 0 p=0.22 Young Middle Old BF FerroOrange MItobright LT Merge Figure a 7 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm 10μm b Log2(Normalized tag count) c d .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint 29 Figure 7. Advanced maternal age oocytes have increased iron accumulation and 852 decreased mitochondria. 853 a Co-staining of Fe2+ (red) and mitochondria (green) in mature MII oocytes (IVM culture from 854 GV oocytes, 3 independent experiments with a total of >10 oocytes) from old, middle, and 855 young women. 856 b Quantitation of FerroOrange fluorescence in MII oocytes from old, middle and young patients. 857 Statistics are from an ANOV A test with Tukey’s HSD. 858 c Violin plots of RNA-seq of aged, middle-aged, and young MII oocytes for selected ferroptosis 859 and mitophagy genes. Note that NOX4, GPX4, and NNMT are not expressed/detectable. Data is 860 from a reanalysis of GSE158802 60. Significance is from a two-sided Welch’s t-test. 861 d Model of ferroptosis in age-related GC and oocyte degradation. 862 863 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted July 14, 2025. ; https://doi.org/10.1101/2025.07.12.664497doi: bioRxiv preprint