Inhibiting bridge integrator 2 phosphorylation leads to improved oocyte quality, ovarian health and fertility in aging and after chemotherapy in mice

Female ovaries degenerate about 20 years earlier than testes leading to reduced primordial follicle reserve and a reduction in oocyte quality. Here we found that bridge integrator 2 (BIN2) is enriched in mouse ovaries and oocytes and that global knockout of this protein improves both female fertility and oocyte quality. Quantitative ovarian proteomics and phosphoproteomics showed that Bin2 knockout led to a decrease in phosphorylated ribosomal protein S6 (p-RPS6), a component of the mammalian target of rapamycin pathway and greatly increased nicotinamide nucleotide transhydrogenase (NNT), the free-radical detoxifier. Mechanistically, we find that phosphorylation of BIN2 at Thr423 and Ser424 leads to its translocation from the membrane to the cytoplasm, subsequent phosphorylation of RPS6 and inhibition of Nnt translation. We synthesized a BIN2-penetrating peptide (BPP) designed to inhibit BIN2 phosphorylation and found that a 3-week BPP treatment improved primordial follicle reserve and oocyte quality in aging and after chemotherapy-induced premature ovarian failure without discernible side effects. The authors found that deleting an ovary-rich gene called Bin2 improved fertility and oocyte quality in mice. Inhibiting BIN2 with a cell-penetrating peptide improved ovarian function in aging and in chemotherapy-treated mice.

R eproductive health is important for the overall health of the individual self as well as the next generation. Due to their distinct physiology, women's ovaries begin to degenerate at least 20 years earlier than men's testes. Moreover, the ovary is more vulnerable than the testis to various adverse factors 1,2 . In addition, chemotherapy against various tumors always severely reduces ovarian primordial follicle reserve (PFR) and lowers oocyte quality 3,4 . An ideal medication would improve ovarian PFR and oocyte quality but not interfere with normal physiology; however, such a medication has yet to be realized. Till now, clinical studies on female fertility improvement have been scarce due to ethical and safety issues; however, multiple translational studies using mice or rats have shown promising improvement of ovarian lifespan, providing important references for future clinical studies and applications.
For maintaining ovarian PFR, the complicated 'dialog' between oocytes, granulosa cells, theca cells and other cells is crucial for ovarian function. Many factors (ligands) and receptors play important roles in the dialog process [5][6][7][8] and one of the critical downstream signal mediators is the mammalian target of rapamycin (mTOR), particularly mTOR complex 1 (mTORC1) [9][10][11][12] . Based on these processes, for translational medicine, inhibiting the activators of the mTOR pathway can theoretically inhibit primordial follicle (PMF) activation, thereby prolonging follicle reserve during the natural aging process. For example, inhibition of cyclooxygenase-1 can prolong the life of postnatal follicles 13 . Cyclooxygenase-1 upregulation is interleukin (IL)-4-dependent and regulated by Fes-Akt (serine-threonine protein kinase)-mTOR 13 . The sirtuin 1 (Sirt1) activator SRT1720 could also improve follicle reserve and prolong ovarian life by inhibiting the mTOR pathway 14 . As a direct inhibitor of mTOR, rapamycin has been shown to preserve the PMF pool and thereby increases ovarian lifespan in young and aging female mice 15 . Caloric restriction can also protect follicle reserve by inhibiting the mTOR pathway 16 . However, despite the positive results above, several concerns remain. First, kinases within the mTOR pathway are mostly universally active and important; inhibition of these kinases could substantially damage overall body health. For example, rapamycin is also an immunosuppressant and might lead to adverse metabolic reactions 17 . Second, patients typically have apparent adverse reactions (such as diarrhea and nausea) to these chemicals 18 . Third, inhibition of mTOR activators is a doubleedged sword, as excessive inhibition could reduce follicle survival and lower oocyte quality because the activators are critical for these aspects [5][6][7][8][9][10][11][12] . In addition, other strategies, such as activating SIRT1/ FOXO1 (forkhead box O1) signaling through human adiposederived stem cell-secreted hepatocyte growth factor and fibroblast growth factor 2 (ref. 19 ) or upregulating Erk (mitogen-activated protein kinase 1) phosphorylation through protein tyrosine kinase 2β knockout (KO) 20 , also could not specifically target ovaries and might induce tumorigenesis 21,22 .
For improving oocyte quality, although the physiological level of reactive oxygen species (ROS) is essential for follicular rupture and oocyte maturation, excessive ROS under disadvantageous Inhibiting bridge integrator 2 phosphorylation leads to improved oocyte quality, ovarian health and fertility in aging and after chemotherapy in mice Female ovaries degenerate about 20 years earlier than testes leading to reduced primordial follicle reserve and a reduction in oocyte quality. Here we found that bridge integrator 2 (BIN2) is enriched in mouse ovaries and oocytes and that global knockout of this protein improves both female fertility and oocyte quality. Quantitative ovarian proteomics and phosphoproteomics showed that Bin2 knockout led to a decrease in phosphorylated ribosomal protein S6 (p-RPS6), a component of the mammalian target of rapamycin pathway and greatly increased nicotinamide nucleotide transhydrogenase (NNT), the free-radical detoxifier. Mechanistically, we find that phosphorylation of BIN2 at Thr423 and Ser424 leads to its translocation from the membrane to the cytoplasm, subsequent phosphorylation of RPS6 and inhibition of Nnt translation. We synthesized a BIN2-penetrating peptide (BPP) designed to inhibit BIN2 phosphorylation and found that a 3-week BPP treatment improved primordial follicle reserve and oocyte quality in aging and after chemotherapy-induced premature ovarian failure without discernible side effects.
circumstances, such as psychological stress, aging or chemotherapy, is detrimental 23,24 . Therefore, many translational studies supported that oocyte quality could be improved by reducing ROS. For example, nicotinamide mononucleotide supplementation eliminated ROS to suppress apoptosis in aged oocytes 25,26 . Coenzyme Q10 supplementation decreased ROS levels and prevented abnormal mitochondrial distribution in oocytes of obese mice 27 . Brown adipose tissue transplantation could also improve the oocyte quality of aging mice 28 . However, neither the upper (and most other) antioxidants nor brown adipose tissue transplantation was able to increase PFR and nonenzymatic antioxidants are less efficient than enzymatic ROS detoxifiers such as superoxide dismutase and glutathione peroxidase 23,25,26 . It has been reported that supplementation with melatonin 29 , quercetin 30 or resveratrol 31 could improve both PFR and oocyte quality by reducing ROS and apoptosis; however, melatonin can inhibit the hypothalamic-pituitary-gonadal axis or directly act on the ovary to reduce the content of androgens, estrogens and progesterone 32 . Quercetin is reported to be mutagenic 33 . Resveratrol has to be taken daily for 12 months in mice to show a significant increase in PFR 31 .
In the current study, we found that the target BIN2 might be a critical regulator of both PFR and oocyte quality in mice. Because BIN2 is fairly dominant in ovaries and oocytes, the target inhibition of BIN2 to prolong female fertility and improve oocyte quality did not affect mouse normal physiology.

Results
BIN2 is essential for primordial follicle activation. Many important cancer-related genes are essential for ovarian function and female fertility. We herein have termed these cancer/ovary genes, in concert with reported cancer/testis genes. BIN2 (gene ID 668218) has also been referred to as breast cancer-associated protein 1. It has been reported that in diverse human cancers, including breast cancer and ovarian cancer, BIN2 has over 40 new emerging posttranslational modification sites, among which, most are phosphorylation sites (https://www.phosphosite.org/proteinAction?id=9451& showAllSites=true), suggesting that BIN2 might be over-activated through hyper-phosphorylation to promote tumorigenesis. We primarily showed that BIN2 overexpression in human epithelial ovarian cancer (EOC) cells significantly upregulated phosphorylated ribosomal protein S6 (p-RPS6) of the mTOR pathway while not altering p-Akt (Fig. 1a,b), whereas Bin2 knockdown in human EOC cells significantly downregulated p-RPS6 while also not altering p-Akt (Fig. 1c,d). p-RPS6 over-activation is a major cause of premature activation of the PMF pool 34 . Therefore, we are interested in determining whether BIN2 is involved in female fertility, which is as yet unknown.
We first found that the BIN2 protein was the most abundant in ovaries, compared to other major tissues (Fig. 1e, red arrow) and that the BIN2 protein level sharply increased as initial follicle recruitment occurred (Fig. 1f). Bin2 mRNA level in oocytes was much higher than the other two Bin family members, Bin1 and Bin3 (Fig.  1g), and remained constant during meiosis (Fig. 1h). Moreover, the BIN2 protein was more highly expressed in oocytes than in granulosa cells (Fig. 1i-k). These findings indicated that BIN2 might have important functions within ovaries and primarily within oocytes.
To uncover its function, we used the cas9 technique to delete the whole exon 9 (83 bases) of the Bin2 genome to obtain Bin2 global KO mice (Fig. 1l). Western blots showed that BIN2 protein completely disappeared in the KO ovaries (Fig. 1m,n). Of note, Bin2 KO did not affect ovary weight, body weight or ovary/body weight ratio, (Supplementary Fig. 1a-d), oocyte maturation or zygote development ( Supplementary Fig. 1e). However, the 60-week-long fertility assay showed that the cumulative pup number per female (Fig. 1o), the average pup number per female from 2 to 10 months old (Fig.  1p) and the average litter size from 8 to 12 months old (Fig. 1q) all significantly increased in the Bin2-KO group, although within some months the average litter size did not show a significant increase (only a trend to increase; Supplementary Fig. 1F). Next, follicle counting showed that from 6 months of age, both the total follicle number and the PMF number were significantly higher in Bin2-KO ovaries ( Fig. 1r-u). In addition, Bin2 KO did not affect major immunity or inflammation indexes, including C-reactive protein (CRP), CD4 (T-cell surface glycoprotein), IL-6 ( Supplementary Fig. 2a,b) and ovaries ( Supplementary Fig. 2c,d) when compared to the control. These results indicated that Bin2 depletion could significantly improve PMF survival and increase fertility without affecting the overall health of female mice.
Phosphorylation translocated BIN2 to activate RPS6. From the above results, BIN2 might be a new target for the safe augmentation of PMF reserve and ovarian lifespan. Because BIN2 is highly conserved between diverse mammalian species, including mice and humans, investigating the mechanism by which BIN2 functions in female mice might ultimately contribute to improvement of ovarian performance in human females. We used TMT-labeled quantitative phosphoproteomics to systematically identify differentially phosphorylated proteins (≥1.2-fold or ≤0.833-fold) between control and Bin2-KO 3-week-old ovaries and carefully observed pathways known to be involved in PMF survival and activation and oocyte and quantification showed that BIN2 overexpression in human eOC cells A2780 significantly upregulated p-RPS6 (red arrow) of the mTOR pathway, while not altering p-AKT; n = 3 biologically independent replicates. c,d, Western blot and quantification showed that Bin2 knockdown by siRNA in A2780 cells significantly downregulated p-RPS6 (red arrow) of the mTOR pathway, while not altering p-AKT; n = 3 biologically independent replicates. e, Western blot showed that BIN2 protein was the most abundant in ovarian tissue (indicated by arrow), compared to other major tissues. Li, liver; Br, brain; Ov, ovary; Sp, spleen; Ki, kidney; Lu, lung. f, BIN2 protein level gradually increased from PND-1 to PND-21. PND, postnatal day. g, RT-PCR in oocytes showed that the Bin2 mRNA level was much higher than the other two Bin family members, Bin1 and Bin3. h, Western blot showed that BIN2 level remained constant during meiosis. The experiments in e-h,m,n were all repeated three times independently with similar results. i-k, Western blot and immunofluorescence showed that BIN2 protein was more highly expressed in oocytes than in granulosa cells (GCs); n = 3 biologically independent replicates. in k, DNA in blue, BIN2 in green. l,m, Cas9 technique was used to delete the entire exon 9 (83 bases, indicated by red rectangle) of Bin2 genome to obtain Bin2 global KO mice, western blot showed that BIN2 protein completely disappeared in Bin2-KO ovaries. n, Genotypying of Bin2-KO mice through PCR and DNA gel electrophoresis. Blue rectangle indicates WT mouse and red rectangle indicates Bin2-KO mouse. experiments in m,n were repeated three times independently with similar results. o-q, A 60-week-long fertility assay showed that in the Bin2-KO group (n = 7 WT mice, n = 5 KO mice), the cumulative litter number per mouse significantly increased (o); the average litter numbers per female from 2 to 10 months significantly increased (p); and the litter size (pups per litter) from 8 to 12 months also significantly increased (q). r-u, Follicle counting showed that at 6 months old (r,s, n = 6 WT mice, n = 6 KO mice) or 10 months old (t,u, n = 6 WT mice, n = 6 KO mice), total follicle numbers and PMF numbers significantly increased in Bin2-KO ovaries. Moreover, the numbers of PFs, secondary follicles (SFs) and antral follicles (AFs) at 10 months old all significantly increased (u). Scale bars, 50 µm (k) and 100 µm (r,t). β-Actin, GAPDH or Gapdh was used as a loading control. For bar graphs in b,d,j,p,q,s,u, data are presented as mean ± s.e.m. Statistical significance was determined using two-sided unpaired t-test; exact P values are labeled above the graphs.
quality maintenance, such as the mTOR pathway (Fig. 2a). We found that only RPS6 of the mTOR pathway had decreased phosphorylation (Fig. 2b,c, arrow), which was verified to be significant by western blot and immunofluorescence ( Fig. 2d-f). In addition, TMT-labeled quantitative proteomics (Fig. 2g) and RNA sequencing ( Fig. 2h) showed that major components of the mTOR pathway did not change at transcriptional or translational levels. Hence, BIN2 might interact with and regulate RPS6 at a post-translational (phosphorylation) level.
However, BIN2 was enriched on the membrane, whereas RPS6 was within the cytoplasm, rendering it unreasonable to conclude that BIN2 interacts with RPS6. Notably, BIN2 has many predicted phos-phorylation sites; hence, we hypothesized that BIN2 might be activated through phosphorylation and translocated into the cytoplasm. We identified several phosphorylation sites (Supplementary Table 1); among these, Thr423 and Ser424 were in the C-terminal domain of unknown function (DUF) (Fig. 3a). We also found that these two residues and the adjacent residues are highly conserved between humans, rats, mice, rhesus monkeys and cynomolgus macaques (Fig. 3b) and we therefore hypothesized that these two residues might be the key residues for activation and translocation of BIN2. We generated a phosphorylation antibody using a core phospho BIN2 peptide (pBP) 'QSKRAASIQR(pThr) (pSer)A' and verified its specificity through BIN2 knockdown with specific siRNAs (Fig. 3c) and a comparative dot-blot with pBP and the corresponding non-phospho BIN2 peptide (BP) 'QSKRAASIQRTSA' (Fig. 3d). We also constructed plasmids expressing BIN2 wild-type (WT) or the BIN2-T423A and S424A mutation (BIN2-AA) and transfected either into A2780 cells. Blotting with p-BIN2 antibody detected a clean single band in BIN2-WTtransfected cells but nothing was detected in BIN2-AA-transfected cells; these results further support the specificity of the p-BIN2 antibody (Fig. 3e). Immunostaining showed that BIN2 was enriched on the oocyte membrane, whereas p-BIN2 was enriched within the nucleus at the germinal vesicle (GV) stage and within spindles during meiosis, which was a quite similar pattern to that observed for p-RPS6 (Fig. 3f,g and Supplementary Fig. 3). Next, to further verify the relationship between BIN2 and RPS6, we injected the BP into the GV oocytes and speculated that this BIN2 peptide could compete with the endogenous BIN2, thereby decreasing phosphorylation of endogenous BIN2, which will subsequently decrease RPS6 phosphorylation (Fig. 3h). Blotting and immunofluorescence both showed that BP injection significantly decreased p-BIN2 levels, while the p-RPS6 level also significantly decreased ( Fig. 3i-n). In all, these results indicated that p-BIN2 level was positively correlated with p-RPS6 level.
In vitro characterization of p-BIN2 activity on RPS6. Next, we attempted to directly verify the relationship between BIN2 and RPS6 and the key roles of Thr423 and Ser424 on BIN2 activation. We used the Bac-to-bac system to express BIN2-WT and BIN2-AA proteins (Fig. 4a). We also expressed and purified RPS6 from Escherichia coli. We found that the primary advantage of bacteria-derived RPS6 was that it was fully inactive (nonphosphorylated). First, BIN2-WT can phosphorylate RPS6 more effectively than an equal amount of BIN2-AA (Fig. 4b). Next, an in vitro dose-dependent phosphorylation assay showed that BIN2-WT could phosphorylate RPS6 much more strongly than BIN2-AA (Fig. 4c,d). Immunostaining on baculovirus-infected sf9 cells showed that BIN2-WT localized both on the membrane and within the cytoplasm, whereas BIN2-AA was almost exclusively localized on the membrane; accordingly, p-RPS6 intensity in BIN2-WT baculovirus-infected sf9 cells was significantly higher than in BIN2-AA baculovirus-infected sf9 cells (Fig. 4e,f).
We next used the Bac-to-bac system to express and purify two different motifs of BIN2-BAR (1-240 AA) and DUF (241-505 AA) and examined the difference between them (Fig. 4g). Firstly, we found that BIN2-DUF, which includes T423 and S424, could phosphorylate RPS6 more effectively than an equal amount of BIN2-BAR (Fig. 4h). Secondly, an in vitro dose-dependent phosphorylation assay showed that BIN2-DUF could phosphorylate RPS6 much more strongly than BIN2-BAR ( Fig. 4i,j). These results further supported the conclusion that BIN2 phosphorylation at Thr423 and Ser424 is the primary mechanism for activating RPS6. . β-Actin was used as a loading control. Scale bar, 50 µm. For bar graphs in c,e,g,h, data are presented as mean ± s.e.m. Statistical significance in e was determined using two-sided unpaired t-test; exact P values are labeled above the graphs.
Thus, we hypothesized that Bin2 KO could improve oocyte quality through NNT. Live-dye staining showed that the ROS level significantly decreased (Fig. 5i,j), whereas the JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-imidacarbocyanine) level significantly increased (Fig. 5k,l) within Bin2-KO oocytes. These observations indicated that Bin2 KO could even improve oocyte quality. Next, we attempted to address whether p-BIN2 or p-RPS6 binds the translation-regulating regions of Nnt to inhibit its translation. We found that Nnt 5′ untranslated region (UTR) are highly conserved between multiple species, including humans, rats, mice, rhesus monkeys and cynomolgus macaques ( Supplementary Fig. 4). We subdivided Nnt 5′ UTR into seven shorter UTRs (SUTRs) or four longer UTRs (LUTRs) (Fig. 6a). We used BIN2 and RPS6 purified from sf9 cells because these proteins were verified to have phosphor-  Table 1) through BIN2 antibody immunoprecipitation (IP) and LC-MS; among these, Thr423 and Ser424 were in the C-terminal DUF domain. b, Thr423 and Ser424 and the adjacent residues of BIN2 are highly conserved between humans, rats, mice, rhesus monkeys and cynomolgus macaques. c, Western blot with the custom-made p-BIN2 antibody (Zoonbio Biotechnology) showed that BIN2 knockdown with specific siRNAs in A2780 cells dramatically reduced p-BIN2 levels, supporting the specificity of the p-BIN2 antibody. d, Dot-blot with the custom-made p-BIN2 antibody on different amounts of pBP QSKRAASIQR(pThr)(pSer)A or BP QSKRAASIQRTSA showed that p-BIN2 could detect pBP at a level of 100 pg to 10 ng, whereas it could not detect BP at any level, supporting the specificity of the p-BIN2 antibody. e, Western blot with the custom-made p-BIN2 antibody on BIN2-WT or BIN2-AA (T423A and S424A non-phosphorylable mutant)-expressing A2780 cells showed that the p-BIN2 antibody detected an expected-size band in BIN2-WT-expressing cells, whereas it did not detect any signal in BIN2-AA-expressing cells, supporting the specificity of the p-BIN2 antibody. f, Immunofluorescence showed that p-BIN2 was enriched within the nucleus of GV oocytes, similar to p-RPS6. DNA is indicated in blue, p-BIN2 or p-RPS6 in green and tubulin in red. g, Immunofluorescence showed that during meiosis, BIN2 was enriched on the oocyte membrane, whereas p-BIN2 was enriched within spindles. h, We hypothesized that BP could compete with endogenous BIN2 to be phosphorylated, therefore BP injection could reduce p-BIN2, consequently conduced to p-rps6 reduction. i,j, Western blot and quantification showed that BP injection into oocytes significantly reduced p-BIN2 level and p-RPS6 level was also significantly reduced, supporting the hypothesis in h; n = 3 biologically independent replicates. k-n, Immunofluorescence and quantification showed that BP injection into oocytes significantly reduced p-BIN2 level (k,l, n = 12 WT oocytes and 13 Bin2-KO oocytes) and p-rps6 level was also significantly reduced (m,n, n = 15 KO oocytes and 14 Bin2-KO oocytes), supporting the hypothesis in h. DNA is indicated in blue and p-BIN2 or p-RPS6 in green. Scale bar, 20 µm. β-Actin was used as a loading control. The experiments in c-g were repeated three times independently with similar results. For bar graphs in j,l.n, data are presented as mean ± s.e.m. Statistical significance was determined using two-sided unpaired t-test; exact P values are labeled above the graphs.
ylated portions (Fig. 6b,c). Protein-RNA UV crosslinking and a gelshift assay showed that p-BIN2 did not bind any SUTRs or LUTRs (Fig. 6d,e), whereas p-RPS6 preferred to bind SUTR 3 + 4, from 42 to 95 bp of Nnt 5′ UTR, instead of other UTR or coding sequence (CDS) regions ( Fig. 6f,g, arrow). Next, we constructed expression vectors wherein EGFP-myc was conjugated to different regions of Nnt 5′ UTR at the upstream location and co-transfected it with the RPS6-WT vector into A2780 cells and found that only RPS6-WT co-transfected with SUTR 3 + 4-fused EGFP-myc significantly decreased EGFP-myc protein level (Fig. 6h,i), providing further evidence of the binding preference of RPS6 on SUTR 3 + 4 of Nnt.
On the other hand, we overexpressed RPS6-WT or RPS6-S235A and S236A (RPS6-AA) and found that only RPS6-WT decreased endogenous Nnt translation (Fig. 6j,k) and decreased the protein level of SUTR 3 + 4-fused EGFP-myc protein level (Fig. 6l-n). These results suggested that p-BIN2 negatively regulates Nnt translation through p-RPS6, whereas p-RPS6 negatively regulates Nnt translation by directly binding to the Nnt 5′ UTR.
Inhibiting p-BIN2 prolongs ovarian lifespan in aging mice. As stated above, female ovarian function and fertility generally begin to decrease at 35 years of age. The existing strategies for prolonging ovarian lifespans in mice have demonstrated several side effects. Because Bin2 KO increased fertility without discernible side effects, we proposed that BIN2 might be a target for safely prolonging ovarian function and female fertility. We planned to use 8-month-old aging mice as models to mimic 35-year-old women 37 (Fig. 7a) and we tested whether inhibiting BIN2 phosphorylation through the non-phosphopeptide 'QSKRAASIQRTSA' could largely mimic the effects of Bin2 KO in mice.
To facilitate permeation of the non-phosphopeptide 'QSKRAASIQRTSA' in vivo, we fused PEP1, the known cellpenetrating peptide 38 , to it and renamed it BIN2-penetrating peptide (BPP) (Fig. 7b). We first verified that FITC-labeled BPP can efficiently enter ovaries (Fig. 7c) or in vitro-cultured oocytes and A2780 cells ( Supplementary Fig. 5a,b) and BPP had typical dose-dependent ( Supplementary Fig. 5c) and time-dependent ( Supplementary Fig. 5d) pharmacokinetics pattern. Second, we found that BPP injection significantly decreased both p-BIN2 and p-RPS6 levels within ovaries (Fig. 7d,e). Next, after a 3-weeklong BPP injection cycle on 8-month-old aging mice, a fertility assay from 36 to 63 weeks of age demonstrated that the number  ). b, In vitro phosphorylation and western blot showed that BIN2-WT can phosphorylate RPS6 more than an equal amount of BIN2-AA. c,d, In vitro, dose-dependent phosphorylation assay and western blot showed that BIN2-WT could phosphorylate RPS6 much more strongly than BIN2-AA. e,f, Immunofluorescence and quantification on baculovirus-infected sf9 cells showed that BIN2-WT localized both on the membrane and within the cytoplasm, whereas BIN2-AA almost exclusively localized on the membrane. Accordingly, cytoplasmic p-RPS6 intensity in BIN2-WT baculovirus-infected infected sf9 cells was significantly higher than in BIN2-AA baculovirus-infected sf9 cells. DNA is indicated in blue, BIN2 in green, RPS6 in magenta and p-RPS6 in red. n = 22 SF9 cells for BIN2-WT group, n = 16 SF9 cells for BIN2-AA group. g, Two different motifs of BIN2, BAR (1-240 AA) and DUF (241-505 AA) were expressed and purified as above; SDS-PAGe and Coomassie staining showed that both purified proteins had good purity and expected size (indicated by arrow). h, In vitro phosphorylation and western blot showed that BIN2-DUF can phosphorylate RPS6 more than an equal amount of BIN2-BAR. i,j, In vitro dose-dependent phosphorylation assay and western blot showed that BIN2-DUF could phosphorylate RPS6 much more strongly than BIN2-BAR. Scale bar, 10 µm. experiments in a-d,g-j were repeated three times independently with similar results. Data in f are presented as mean ± s.e.m. Statistical significance was determined using two-sided unpaired t-test; exact P values are labeled above the graphs.
of cumulative pups per female and litters per female significantly increased (Fig. 7f-h and Supplementary Fig. 6a), although the monthly cumulative pups per female (at 10, 11, 12, 13 and 14 months old) were not always significant ( Supplementary Fig. 6a). Follicle counting showed that total follicles, PMFs and primary follicles (PFs) (Fig. 7i,j and Supplementary Fig. 6b,c) all increased. Accordingly, p-RPS6 also significantly decreased (Fig. 7k,l), whereas the NNT level within oocytes significantly increased (Fig. 7m-p). In addition, major organ weight ( Supplementary Fig.  7a,b) and major blood biochemical indexes ( Supplementary Fig.  7c) did not change after BIN2 inhibition. These results indicated that ovary function in aging mice was improved by BIN2 inhibition without discernible side effects. Finally, we examined how well BPP could revive ovarian function in the aging mice at the transcriptional level. RNA sequencing showed that at the fold threshold of |log 2 (2 months/9 months)| ≥ 2, there were 1,858 differentially expressed genes (DEGs) between 2and 9-month-old ovaries; at the fold threshold of |log 2 (9 months BPP/9 months)| ≥ 2, there were 1,138 DEGs between 9-month-BPP ovaries and 9-month ovaries; and between these two clusters of DEGs, 622 DEGs overlapped, suggesting that BPP injection could significantly recover the 9-month ovarian profile, bringing it close to the 2-month ovarian profile. (Fig. 7q,r). KEGG analysis showed that multiple follicle survival, oocyte quality and oocyte steroidogenesis-related pathways were among the top ten enriched signaling pathways (Fig. 7s).
Inhibiting p-BIN2 protects ovaries from CPA-induced premature ovarian failure. Chemotherapy can cause premature ovarian failure (POF). One of the major mechanisms of this failure is mTOR pathway over-activation. Therefore, mTOR inhibitors could rescue ovarian function and fertility 13,14,17 . However, mTOR-inhibiting chemicals also have multiple side effects 17,18 . We wanted to test whether BPP could also rescue chemotherapy-induced POF safely and effectively. We used 2-month-old mice to mimic pubescent women (approximately 16 years of age; Fig. 6a).
Western blots showed that a harsh, 3-week-long cyclophosphamide (CTX) treatment cycle significantly elevated p-RPS6, whereas BPP treatment reduced p-RPS6 levels close to the control level (Fig.  8a,b). Moreover, CTX treatment significantly decreased oocyte NNT levels, but BPP injection recovered NNT levels close to the control (Fig. 8c-f). Consequently, a fertility assay from 4 to 9 months old showed that CTX treatment almost induced complete infertility, whereas BPP significantly recovered cumulative pups per female and litters per female (Fig. 8g-i), although the monthly numbers of pups per female (at 5, 6, 7, 8 and 9 months old) were not always significant ( Supplementary Fig. 8). Follicle counting at 3 months old illustrated that CTX treatment sharply decreased the numbers of follicles at each stage, whereas BPP significantly upregulated follicle numbers, although at levels still less than those of the control (Fig. 8j-l). Moreover, we showed that CTX increased double-strand breaks (DSBs) as labeled by γH2AX staining; in contrast, BPP treatment significantly decreased DSBs (Fig. 8m,n). These results Among the differentially expressed gross proteins, NNT was the most upregulated (Bin2-KO/WT, >sixfold, indicated by a red arrow); n = 2 biologically independent replicates. c,d, Western blot and quantification verified that the NNT protein level in the Bin2-KO group increased more than sixfold compared to the control; n = 3 biologically independent replicates. e,f, RT-PCR and quantification showed that Bin2 KO did not affect the mRNA level of NNT; n = 3 biologically independent replicates. g,h, Western blot and quantification showed that the NNT protein level in the RPS6-knockdown group significantly increased; n = 3 biologically independent replicates. i,j, Live-dye staining and quantification showed that the ROS level significantly decreased in Bin2-KO oocytes; n = 28 WT oocytes and 22 Bin2-KO oocytes. k,l, JC-1 staining and quantification showed that the mitochondria membrane potential significantly increased in Bin2-KO oocytes. Aggregate shown in red and monomer in green. Mitochondria membrane potential is indicated as aggregate/monomer ratio (red/green); n = 49 WT oocytes and 42 Bin2-KO oocytes. Scale bar, 20 µm. β-Actin or Gapdh was used as a loading control. For bar graphs in b,d,f,h,j,l, data are presented as mean ± s.e.m. Statistical significance was determined using two-sided unpaired t-test; exact P values are labeled above the graphs.
indicated that BPP could effectively preserve the ovarian PMF pool and simultaneously maintain oocyte quality during chemotherapy.

Discussion
BIN2 belongs to the BAR family proteins, which are mostly involved in the phagocytosis of synaptic vesicles, regulation of microfilaments, cell division and migration in different cells 39,40 . The nonconserved C-terminal region might explain the functional diversity of different BAR proteins 39,40 . An in vitro study has shown that human Bin2 knockdown led to decreased leukocyte migration, pseudopod density and pseudopod kinetics 41 . Recently, BIN2 was shown to play crucial roles in platelet activation in thrombosis and thrombo-inflammation 42 ; however, there have been no reports on its function in reproduction. According to the National Center for Biotechnology Information, the C-terminal region of human BIN2 contains infection cell polypeptide 4, a phosphoprotein, suggesting that BIN2 could be activated or deactivated through phosphorylation or dephosphorylation, respectively. We have independently characterized two key phosphorylation sites (Thr423 and Ser424) within the C-terminal region (temporarily named DUF; Fig. 4). Notably, although these regions are highly variable among humans, cynomolgus monkeys, macaque monkeys, mice and rats, the peptide sequences around Thr423 and Ser424 are highly conserved ( Supplementary Fig. 9), suggesting that the conserved activation/   Fig. 6 | p-RPS6 binds to Nnt 5′ UTR to inhibit Nnt translation. a, Nnt 5′ UTR region was subdivided into seven SUTRs or four LUTRs, then single-strand RNA against these regions was transcribed in vitro. b,c, SDS-PAGe, Coomassie staining (left) and western blot (right) showed that RPS6 (b) and BIN2 (c) proteins had good purity and phosphorylated portions. d,e, Protein-RNA UV crosslinking and gel-shift assay showed that p-BIN2 did not bind any SUTR or LUTR. f,g, Protein-RNA UV crosslinking, gel-shift and quantification showed that p-RPS6 prefers to bind SUTR 3 + 4, from 42 to 95 bp of Nnt (red arrow), instead of other SUTR or CDS region. The shift ratio was obtained by the integral intensity of shifted upper SUTR bands divided by the integral intensity of corresponding SUTR only (as a loading control). h,i, eGFP-myc conjugated to different regions of Nnt SUTR (1 + 2, 3 + 4, 5 + 6 and 7) or CDS upstream was constructed into expressing vectors pcDNA3.1( + ), each vector was co-transfected with Rps6-WT-TagRFP-flag vector. Western blot and quantification showed that Rps6-WT overexpression significantly reduced eGFP-myc level only when eGFP-myc was fused to Nnt SUTR 3 + 4; n = 3 biologically independent replicates. j,k, Western blot and quantification showed that Rps6-WT overexpression significantly decreased NNT level, whereas RPS6-AA (RPS6-S235A and S236A, non-phosphorylable form) overexpression did not; n = 4 biologically independent replicates. l-n, Immunofluorescence (n), western blot (l) and quantification (m) showed that Rps6-WT co-transfected with SUTR 3 + 4-fused eGFP-myc significantly decreased eGFP-myc protein level, whereas Rps6-AA co-transfection did not; n = 3 biologically independent replicates. Scale bar, 50 µm. β-Actin was used as a loading control for blots. The experiments in b-e,n were repeated three times independently with similar results. For bar graphs in i,k,m, data are presented as mean ± s.e.m. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by the Tukey post hoc test; exact P values are labeled above the graphs.   Fig. 7 | Inhibiting p-BIN2 prolongs ovarian lifespan of aging mice. a, Analogical time axis between mouse and human; 8-month-old aging mice were used to mimic 35-year-old women. b, Cell-penetrating peptide PeP1 was fused to BP and renamed as BPP to facilitate permeation. c, FITC-labeled BPP can efficiently enter ovaries 3 d after injection. d,e, Western blot and quantification showed that BPP injection significantly decreased both p-BIN2 and p-RPS6 within ovaries; n = 3 biologically independent replicates. f-h, A fertility assay from 36 to 63 weeks old showed that a 3-week BPP injection cycle into 8-month-old aging mice significantly increased the number of cumulative pups per female (f-g) and litters per female (h); n = 7 mice for control group and n = 8 mice for BPP group. i,j, Follicle counting showed that both total follicles and PMFs significantly increased in the BPP-treated group; n = 3 biologically independent replicates. k,l, Western blot and quantification showed that p-RPS6 significantly decreased in BPP-treated ovaries; n = 6 biologically independent replicates. m-p, Immunohistochemistry (n = 10 oocytes for control group, n = 11 oocytes for BPP group (m,n)) and western blot (n = 5 biologically independent replicates (o,p)) showed that the NNT level within oocytes significantly increased in the BPP-treated group. q,r, RNA sequencing showed that at the fold threshold of |log 2 (2 months/9 months)| ≥2, there were 1,858 DeGs between 2-month-old ovaries and 9-month-old ovaries; at the fold threshold of |log 2 (9-month-BPP/9 months)| ≥2, there were 1,138 DeGs between 9-month-BPP ovaries and 9-month ovaries; and between these two clusters of DeGs, 622 DeGs overlapped (Supplementary Dataset 5). s, KeGG analysis of functionally annotated DeGs showed that multiple follicle survival, oocyte quality or oocyte steriodogenesis-related pathways are among the top ten enriched signaling pathways (red rectangle). Scale bars, 200 µm (c) and 100 µm (i,m). β-Actin or GAPDH was used as a loading control. experiments in c were repeated three times independently with similar results. For bar graphs, data in e,g,h,j,l,n,p are presented as mean ± s.e.m. Statistical significance was determined using two-sided unpaired t-test; exact P values are labeled above the graphs.
inactivation mechanism might rely on these two residues. On the other hand, human BIN2 is also referred to as breast cancer-associated protein 1 and has 40 reinforced post-translational modification sites in diverse human tumors, suggesting that its over-activation may promote cell proliferation. In the present study, we found that Bin2 KO specifically downregulated p-RPS6 of the mTOR pathway. RPS6 hyper-phosphorylation is closely related to follicular overactivation 34 , suggesting that targeted downregulating p-RPS6 by Western blot showed that a harsh, 3-week-long CTX daily treatment significantly elevated both p-BIN2 and p-RPS6 levels, whereas BPP treatment reduced their level close to control; n = 4 biologically independent replicates for p-BIN2 detection; n = 5 biologically independent replicates for p-RPS6 detection. c-f, Quantification of immunohistochemistry (n = 27 oocytes for control group, n = 39 oocytes for CTX group, n = 31 oocytes for CTX + BPP group (c,d)) and western blot (n = 3 biologically independent replicates (e,f)) showed that CTX treatment significantly decreased the oocyte NNT level, whereas BPP injection recovered the NNT level close to control. g-i, A fertility assay from 4 to 9 months old showed that CTX treatment caused almost completed infertility, whereas BPP significantly recovered cumulative pups per female and litters per female (although less than control); n = 8 mice for control group, n = 6 mice for CTX group, n = 7 mice for CTX + BPP group. j-l, Follicle counting at 3 months old showed that CTX treatment sharply decreased follicle numbers at each stage, whereas BPP significantly increased follicle numbers at each stage; n = 5 biologically independent replicates. m,n, Immunofluorescence and quantification showed that CTX treatment increased DSBs as labeled by γH2AX staining, whereas BPP significantly decreased DSBs. DNA is indicated in blue and γH2AX in green; n = 18 oocytes for control group, n = 17 oocytes for CTX group, n = 17 oocytes for CTX + BPP group. Scale bars, 100 µm (c,j) and 20 µm (m). β-Actin was used as a loading control. For bar graphs in b,d,f,h, i,k,l,n, data are presented as mean ± s.e.m. Statistical significance was determined by one-way ANOVA followed by the Tukey post hoc test; exact P values are labeled above the graphs.
Bin2 KO or inhibition could inhibit PMF activation and, therefore, increase PFR. These conclusions are supported by multiple results in our present study. RPS6 is a component of the ribosome 40S subunit. The peculiar feature that differentiates RPS6 from other ribosomal proteins is that it can bind chromatin as well as localize within the matrix and cytoplasm 43 . Conceivably, it can regulate various important cellular processes. In fact, RPS6 is known to be regulated by multiple kinases and phosphatases [44][45][46] . In the present study, we showed that BIN2 might be a new kinase that directly phosphorylates RPS6 (Fig. 4). Thus, Bin2 KO or inhibition is predicted to moderately downregulate p-RPS6. On the other hand, RPS6 itself was also essential for PMF survival; excessive RPS6 suppression or Rps6 knockout caused loss of PMFs 46 ; hence, the extent of RPS6 reduction must be moderate to ensure both the survival and increase of PMFs. In our study, Bin2 KO or inhibition seemed to cause an optimum p-RPS6 reduction, which is the critical basis of the dual function of BIN2.
The main function of NNT is to regenerate NADP into NADPH and maintain the mitochondrial redox balance 35 ; NNT deficiency causes vascular ROS production and exacerbates atherosclerotic plaque development 47 . On the other hand, NNT levels decrease as aging and mitochondrial NAD levels also decrease in somatic cells, whereas NNT overexpression could restore NAD levels and promote reprogramming of aged somatic cells 48 . In our study, Bin2 KO and inhibition both substantially increased NNT and this change could protect oocyte quality even in cases in which p-RPS6 is significantly reduced. Moreover, we discovered that p-BIN2 regulates Nnt translation through p-RPS6 and thus it is likely that p-RPS6 directly binds to the 42-95 bp Nnt 5′ UTR to inhibit Nnt translation. Therefore, we provided a clear model for how BIN2 knockdown or inhibition can improve oocyte quality through p-RPS6 and NNT.
In summary, we showed that Bin2 KO or inhibition could have a dual effect (Supplementary Fig. 10a). At the post-translational level, p-BIN2 specifically activated p-RPS6 of mTOR; thus, p-BIN2 inhibition improved the PFR by specifically and moderately downregulating p-RPS6. At the translational level, p-RPS6 binds to 42-95 bp of Nnt 5′ UTR mRNA to inhibit Nnt translation; thus, p-BIN2 inhibition upregulated NNT protein levels to improve oocyte quality. Because BIN2 is fairly predominant in ovaries and oocytes, the effect of p-BIN2 inhibition might be limited primarily to these areas. Therefore, compared to current strategies, BIN2 could be a specific and safe target for improving PFR and oocyte quality ( Supplementary Fig. 10a,b). However, our conclusion is limited to the mouse model and further studies are needed to test the effects of Bin2 inhibition in other mammalian models (such as pigs and monkeys), which might provide stronger references for future clinical studies and application on improving and prolonging female fertility in humans.

Methods
Bin2 knockout mice. Animal experimental procedures in our study were all approved by the Animal Ethics Committee of Nanjing Medical University (approval no. IACUC-1809011) and all mice were housed under standard specificpathogen-free conditions of the Animal Core Facility (ACF).
The global Bin2-knockout C57/B6 mice were made in the ACF of the State Key Lab of Stem Cell and Reproductive Biology of the Institute of Zoology, Chinese Academy of Sciences and then transferred to the ACF of the State Key Lab of Reproductive Medicine of Nanjing Medical University and re-cleansed. CRISPR/ Cas9 technology was used. We planned to delete the whole exon 9, which is shared by three isoforms of Bin2. For this, we designed two 20-base, gene-complementary oligonucleotides of single-guide (sg)RNA within the introns upstream and downstream of exon 9, respectively (Supplementary Table 2). Each of these four oligonucleotides were inserted into pUC57-T7-genomic RNA. The sgRNA was produced by MEGAshortscript kit (Thermo Fisher Scientific) using linearized pUC57-T7-Bin2 gRNA as a template and purified by MEGAclear kit (Thermo Fisher Scientific). The cas9 mRNA was first produced by mMessage mMachine T7 kit (Thermo Fisher) using linearized pST1374-N-NLS-flag-linker-Cas9 as a template. Then the mRNA was poly A-tailed (to increase mRNA stability) using a poly A-tailing kit (Thermo Fisher) and purified by RNeasy Micro kit (QIAGEN). sgRNA and cas9 mRNA were sent to the ACF of the State Key Lab of Stem Cell and Reproductive Biology; pronuclear microinjection, embryo transfer and mouse parturition was accomplished by professional staff in the ACF. Genotyping of Bin2-KO mice (Fig. 1i) was performed by PCR and DNA sequencing, where the forward primer was 5′-GATTCTAAGCTGCTTCCACATTAC-3′; the reverse primer was 5′-TGGCTGGTCTCAGATTTATGAC-3′; and the middle reverse primer was 5′-ACAAAGACTTTGTTGGAATGTTG-3′. The genotyping PCR program was 94 °C for 5 min, 35 cycles of melting at 94 °C for 30 s, annealing at 57 °C for 15 s and extension at 72 °C for 30 s, with additional extension at 72 °C for 5 min at the end. PCR with forward and reverse primers will result in a band size difference of about 650 bp (Fig. 1n).
For all fertility assays, WT mating male mice were rotated on a monthly basis between cages according to a random allocation table (Supplementary Dataset 1).
Model mice. The relative age between mice and humans was determined according to a previous report 37 (Fig. 7a). Forty 8-month-old C57/B6 retiring female mice (equal to 34.5-year-old human female) were bought from ACF and randomly divided into two groups: control group, treated with 0.5 mg kg −1 control peptide, PEP1 (KETWWETWWTEWSQPKKKRKV); and BPP group, treated with 0.5 mg kg −1 BPP (KETWWETWWTEWSQPKKKRKVQSKRAASIQRTSA) once daily for 3 weeks. BPP and PEP1 were synthesized by Shanghai Bootech BioScience & Technology Company. The treated mice were subjected to diverse assays (Fig. 6) at 9 months old (equivalent to a 36.5-year-old human female).
Sixty 2-month-old ICR female mice (equivalent to a 16-year-old human female) were bought from ACF and randomly divided into three groups: control group, treated with 0.5 mg kg −1 PEP1; CTX-only group, treated with 75 mg kg −1 CTX (Cayman) once daily for 3 weeks; and CTX + BPP group, treated with 75 mg kg −1 CTX and 0.5 mg kg −1 BPP once daily for 3 weeks. The treated mice were subjected to diverse assays (Fig. 7) at 3 months old (equivalent to a 20-year-old human female).
All procedures involving mice samples were approved by the Ethical Committee of Nanjing Medical University (approval no. IACUC-1809011).
Plasmid construction. All to-be-expressed gene fragments were amplified by high-fidelity DNA polymerase (Vazyme) from complementary DNA reversetranscribed by SSRT VI (Thermo Fisher), then the fragments were digested by high-fidelity restriction enzymes (NEB), purified and inserted into related empty plasmids by Quick ligase (NEB). All primer sequences for the constructs are described in Supplementary Table 5. In all construct, BIN2, RPS6 or other fragments were fused to strep II or flag for rapid detection with strep II or flag antibody. All insert and vector information along with the corresponding figure panels employing the related constructs is given in Supplementary Table 7.
Oocyte collection and in vitro culture. Fully grown GV oocytes were collected from 3-week-old female mice. Oocytes were released by puncturing follicles with a sterile syringe needle in MEM+ medium (0.01 mM EDTA, 0.23 mM sodiumpyruvate, 0.2 mM penicillin/streptomycin and 3 mg ml −1 BSA in MEM). After washing away the cumulus cells from the cumulus-oocyte complexes, oocytes were cultured in 100 µl mini-drops of MEM+ containing 20% fetal bovine serum (FBS) (Thermo Fisher) covered with mineral oil at 37.0 °C in an incubator with 5% O 2 and 5% CO 2 in a humidified atmosphere.
Cell culture, plasmid transfection and siRNA knockdown. Human EOC cells, A2780, were from ECACC (Procell Life Science & Technology, cat. no. 93112519). Cells were cultured in DMEM with 10% FBS. For plasmid transfections, cells were grown at 50-70% confluence and plasmids were transfected with Lipofectamine 3000 Transfection Reagent (Thermo Fisher) according to the manufacturer's protocol. pcDNA3.1(+) was used as a control according to the distinct treatment in various experiments.
For siRNA knockdown, each double-strand short-interfering (si)RNA was produced by the T7 RiboMAX express RNAi system (Promega) according to the manufacturer's instructions. Two complementary single-stranded RNAs were first separately transcribed from corresponding double-stranded DNA templates (Supplementary Table 4), then annealed to form the final double-stranded siRNA. siRNA was then purified by conventional phenol/chloroform/isopropanol precipitation, aliquoted and stored at −80 °C after a quality check on an agarose gel. A ready-to-use siRNA mixture was set by mixing siRNAs against four distinct target regions at an equal molar ratio to a final concentration of 5 µmol l −1 . For siRNA transfection into A2780 cells, we again used Lipofectamine 3000 Transfection Reagent as above. The DNA templates for control siRNA is a sequence that doesn't target to any mouse mRNAs (Supplementary Table 4).
Immunofluorescence staining of oocytes and image taking. Oocytes were permeabilized with 0.5% Triton X-100/PHEM (60 mM PIPES, 25 mM HEPES, pH 6.9, 10 mM EGTA and 8 mM MgSO 4 ) for 5 min and then were fixed in 3.7% FPA in PHEM for 20 min at room temperature. After being washed with PBS/0.05% PVP (polyvinylpyrrolidone) three times at 10 min each, oocytes were blocked in blocking buffer (100 mM glycine and 1% BSA in PBS) for one hour at room temperature. Primary antibodies were then diluted in blocking buffer and oocytes were incubated in it overnight at 4.0 °C. Fluorescent secondary antibodies (Jackson ImmunoResearch Laboratories) were used at 7.5 μg ml −1 . DNA was stained with 0.3 μg ml −1 Hoechst 33258. Next, buffers were slowly and gently removed from around the oocytes, which immobilized the oocytes on the slide; then a drop (about 5-10 μl) of anti-fade solution (0.25% n-propyl gallate and 90% glycerol in PBS) was mounted onto the oocytes and oocytes were covered by a cover glass. To avoid the deformation of the oocytes, a double-stick tap was pre-placed between the slides and coverslips. Specimens were imaged with IQ2 on an Andor Revolution spinning-disk confocal system (Andor Technology) mounted on an inverted Ti-E microscope (Nikon) with a ×60, 1.4 NA objective and captured with a cold charge-coupled device camera (Andor). Most images are displayed as maximum intensity projections of the captured z stacks.
Ovarian hematoxylin and eosin staining and follicle counting. Ovaries were obtained, washed and fixed in 10% buffered formalin or 4% PFA overnight, embedded in paraffin, continuously sectioned at 5-µm thickness, then stained with hematoxylin and eosin. Follicle stages were classified according to Pederson's standard. Only follicles with visible nuclei were counted. In brief, an oocyte surrounded by a single layer of flattened or cubical granulosa cell was defined as a primordial or primary follicle; an oocyte surrounded by more than one layer of cuboidal granulosa cells without visible antrum was defined as a secondary follicle; a follicle possessing a clear antral space containing follicular fluid was defined as an antral follicle. For the follicle number, we counted every other two slices (as the same follicle will appear at adjacent different slices) and the final follicle number of a certain stage was a cumulative number of all counts of the corresponding stage.
Identifying p-BIN2 sites and generating p-BIN2 antibody. Generally, the phosphorylated part of a protein is about 0.01-0.1%, making it almost impossible to use oocytes for identification of phosphorylation sites. Therefore, we used NIH3T3 cells instead, then generated a phospho-specific antibody and verified the antibody in oocytes. We used five immunoprecipitation reactions, each of which employed 2.5 × 106 NIH3T3 cells, 30 μl protein A/G beads and 2 μg BIN2 antibody. Then the immuno-complexed beads were eluted by 0.2 M glycine (pH 2.7) and the phosphorylated portion of the immuno-complex was enriched by Pierce TiO 2 Phosphopeptide Enrichment and Clean-up kit (Thermo Scientific) and sent to the Testing and Analysis Center, Nanjing Medical University for liquid chromatography-mass spectrometry (LC-MS). We chose Thr423 and Ser424 (phosphorylation possibility 99.8%) and the whole process of antibody production and purification was accomplished by Zoonbio Biotechnology. A short phosphopeptide QSKRAASIQR(pThr)(pSer)AC ('C' at the C terminus is an extra residue for conjugation) was synthesized and injected into rabbits for serum production. The phospho-specific antibody was purified from the serum through a column filled with phosphopeptide-conjugated resin and then absorbed through a column filled with non-phosphopeptide (QSKRAASIQRTSA)-conjugated resin to remove any residual non-phospho-specific antibody.
Expression and purification of BIN2 and RPS6 in SF9 cells. EGFP-strep-II-tagged BIN2-WT, BIN2-AA (T423A, S424A), BIN2-BAR (BAR domain of BIN2), BIN2-DUF (DUF domain of BIN2) and TagRFP-flag-tagged RPS6 proteins were cloned and expressed through a Bac-to-bac system (Thermo Fisher). In brief, corresponding sequences (Supplementary Table 3) were cloned into pFastBacHTA (Supplementary Table 4) and then transformed into DH10Bac E. coli cells. The bacmid was isolated from DH10Bac E. coli (Vazyme) with QIAfilter Plasmid Purification kit (QIAGEN) and then transfected into Sf9 cells (Genetimes ExCell Technology, cat. no. ATCC CRL-3357) with Cellfectin II Transfection reagent (Thermo Fisher) in the static upper plane of an orbital shaker (Shanghai Zhichu Instrument) at 27 °C. The virus-containing supernatant was used to infect fresh Sf9 cells for 72 h and the procedure was repeated twice to amplify the baculovirus. Then, 20 μl of the third cycle supernatant was added to Sf9 cells in 250 ml SFM900-II medium (Thermo Fisher) with 5% FBS to express proteins in the lower bed of an orbital shaker at 27 °C and 200 r.p.m. Next, infected cells were resuspended in a lysis buffer (containing 50 mM Tris, 10% sucrose, 50 µM ATP, 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mM dithiothreitol (DTT), 1% NP40, 10 mM imidazole, 1× protease inhibitor and phosphatase inhibitor, pH 7.0 by HCl) and lysed with a high-pressure cell disrupter (Union Biotech), centrifuged and the resulting supernatant was incubated with 1 ml Ni-NTA Superflow resin (QIAGEN) at 4 °C for 1 h. The resin was then transferred into a 5-ml chromatography column (Biocomma), washed with four column volumes of wash buffer (40 mM imidazole in resuspension buffer without PMSF) and eluted with 500 mM imidazole in resuspension buffer without PMSF. The eluted protein was concentrated by a size-exclusion spin column and exchanged into BRB80 (80 mM HEPES, 1 mM MgCl 2 and 1 mM EGTA, pH 6.8 by KOH) with 10% glycerol, 50 μM ATP and 5 mM DTT. The protein was aliquoted and kept at −80 ˚C for future use.
Expression and purification of RPS6 proteins in E. coli. For a nonphosphorylated form of RPS6, the recombined RPS6-tagRFP-flag sequence was cloned into pRSETB (primers are listed in Supplementary Table 5) and transformed into BL21-DE competent E. coli (Vazyme). Transformed E. coli was grown in 4 × 1 l LB + medium in an orbital shaker (Shanghai Zhichu Instrument Co.) at 37 °C and 220 r.p.m. until OD 600 reached 1.0. Next, 0.2 mM isopropyl-b-d-thiogalactoside was added into the medium to induce the recombined RPS6 expression overnight at 16 °C, 220 r.p.m. The next morning, E. coli was spun down, washed once with cold PBS, lysed and purified as described above.
In vitro phosphorylation assays. BRB80 with 10% glycerol, 1 mM ATP and 5 mM DTT was used for the in vitro phosphorylation assays. Appropriate amounts of BIN2 proteins and RPS6 were incubated at room temperature for 20 min. Finally, the reaction was subjected to western blotting.

TMT-labeling quantitative proteomics and phosphoproteomics.
Approximately 200 3-week-old ovaries per repeat (about 200 mg), two repeats for WT and Bin2-KO, respectively, were sent to Hangzhou Jingjie. In brief, ovaries were cracked and the supernatant was digested by trypsin into peptides. Then the peptides from individual samples were isobaric-mass tagged by TMT 6 -126, TMT 6 -127, TMT 6 -128 and TMT 6 -129, respectively, according to the manufacturer's protocol for TMT kit/ iTRAQ kit (Thermo Fisher). Next, TMT-labeled tryptic peptides were fractionated by high pH reverse-phase HPLC using a Thermo Betasil C18 column, then peptide fractions were subjected to a nanospray ionization source followed by tandem mass spectrometry (MS/MS) in Q ExactiveTM Plus (Thermo) coupled online to the UPLC for identification of peptides (for proteomics) or phosphorylation sites (for phosphoproteomics). The resulting MS/MS data were processed using the Maxquant search engine (v.1.5.2.8).
Assay of mitochondrial transmembrane potential. Oocytes were incubated at 37 °C for 20 min with the fluorescent potentiometric indicator JC-1 (cat. no. 40706ES60, Yeasen) diluted 1:200, then washed twice with PBS and added to droplets (50 µl) of culture medium. Images of green fluorescence (JC-1 as monomers at low membrane potentials) and red fluorescence (JC-1 as aggregates at higher membrane potentials) were captured using confocal microscopy as described above. Mitochondrial depolarization is indicated by a decrease in the red/green fluorescence intensity ratio.
Detection of ROS generation. The ROS assay kit (cat. no. S0033, Beyotime) was used to detect ROS generation in oocytes. In brief, oocytes were incubated with a dichlorofluorescein diacetate probe for 20 min at 37 °C in the dark, washed and mounted on slides for confocal imaging.
RNA sequencing and analysis. RNA samples were collected from mouse ovaries. RNA isolation, library construction and RNA sequencing (RNA-seq) were carried out by the Beijing Genomics Institute following standard protocols. The library products were sequenced using a BGISEQ-500. Standard bioinformatics analysis was performed by the Beijing Genomics Institute. For gene expression analysis, the significance of the DEGs was defined by the bioinformatics service of the Beijing Genomics Institute according to the combination of the absolute value of |log 2 (treated/control)| ≥2 and q value <0.001. All original sequence datasets have been submitted to the database of NCBI Sequence Read Archive under accession number: For RNA-seq in Fig. 2h, the accession numbers for WT1-3 and KO1-3 are SAMN22783185-SAMN22783190 (six continuous numbers); for RNA-seq in Fig. 7q-s, the accession numbers for 2M-1-3, 9M-1-3 and 9M+BPP-1-3 are SAMN22786754-SAMN22786762 (nine continuous numbers).

RNA-protein crosslinking and gel-shift assay.
To determine whether p-BIN2 or p-RPS6 binds the 5′ UTR and to meticulously analyze which subregion of the 5ʹ UTR is preferentially bound, we used two different scales of subdivision (Fig.  6a). For one scale, we subdivided the whole 5′ UTR of Nnt into seven shorter subregions (SUTRs), 20-30 bp in length; for another scale, we subdivided the whole 5′ UTR of Nnt into four longer subregions (LUTRs), 40-50 bp in length.
Single-strand RNAs (primers for DNA templates are listed in Supplementary  Table 6) were produced and purified using the T7 RiboMAX Express RNAi System (Promega) as above, then aliquoted and stored at −80 °C. The difference with these is that the RNA is single-stranded from one transcription reaction.
p-BIN2 or p-RPS6 were incubated for 20 min with each of SUTRs or LUTRs. After incubation, the mixtures were placed on a parafilm-coated plate on top of an ice plate, then UV-irradiated three times using CL-1000 UV Crosslinker (Upland) at 254 nm, 5 mJ cm −2 energy. Next, each crosslinked complex was separated on a 0.6% agarose gel side by side. The corresponding RNA alone was loaded as a control.
Statistical analysis. All experiments were repeated at least three times. Data are presented as mean ± s.e.m. Comparisons between two groups were made by Student's t-test. Differences among more than two groups were compared using one-way ANOVA. P < 0.05 was considered statistically significant. Statistical analyses were conducted with GraphPad Prism.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request. Supplementary Datasets 1-5 have been deposited into Zenodo (https://doi.org/10.5281/zenodo.4006605). Raw data and extracted text files for quantitative proteomics and phosphoproteomics have been deposited into PRIDE, under accession nos. PXD028776 and PXD028777, respectively.