Eif4enif1 haploinsufficiency disrupts oocyte mitochondrial dynamics and leads to subfertility Free

Infertility affects an estimated 48 million couples and 186 million individuals worldwide (Mascarenhas et al., 2012). Premature ovarian insufficiency (POI), alongside generalized late marriage and childbirth, is one among many factors contributing to infertility. POI refers to the loss of ovarian function before 40 years of age. Individuals with POI often show elevated follicle-stimulating hormone (FSH) and decreased estrogen levels (Webber et al., 2016). The causes of POI are heterogeneous, and genetic abnormalities, including chromosome number abnormalities and single-gene mutations, account for up to 30% of all POI cases (Qin et al., 2015).

In recent decades, efforts have been made to explore POI caused by single-gene mutations. Mutations in numerous genes related to oocyte development and meiotic maturation have been associated with POI onset. However, the mechanisms by which many clinical gene mutations promote POI remain unclear.

Mutations in the EIF4ENIF1 (4E-T) gene, including missense and nonsense mutations, have been observed in families with dominantly inherited POI (França et al., 2020; Kasippillai et al., 2013; Zhao et al., 2019; Shang et al., 2022). EIF4ENIF1 is an RNA-binding protein (RBP). Human EIF4ENIF1 and its orthologs are mainly located in the cytoplasmic ribonucleoprotein complex P-body (Kamenska et al., 2014a; Minshall et al., 2007; Nakamura et al., 2004). It is an EIF4E-binding protein with reported translation repressing function (Kamenska et al., 2014a,2016). EIF4ENIF1 is highly expressed in developing oocytes (Villaescusa et al., 2006). P-body-like granules have been observed in oocytes in early-stage follicles and are reported to gradually disassemble with oocyte growth (Flemr et al., 2010). Fully grown mammalian oocytes have been reported to store mRNAs in the mitochondria-associated ribonucleoprotein domain (MARDO), which is involved in mRNA translation repression, storage and degradation (Cheng et al., 2022). EIF4ENIF1 is a MARDO protein component (Cheng et al., 2022). In an RNAi screen for essential genes in mouse oocyte meiosis, researchers found that Eif4enif1 knockdown in developing follicles causes meiotic arrest at the germ vesicle (GV) stage (Pfender et al., 2015). Collectively, clinical and experimental evidence indicates EIF4ENIF1 involvement in oocyte development and female fertility. However, Eif4enif1 deficiency in mouse models has been insufficiently studied to elucidate the mechanism by which it impairs fertility. No previous studies have investigated the translatome changes underlying EIF4ENIF1 deficiency in oocytes. The effect of EIF4ENIF1 deficiency on embryonic development also remains unknown.

We first analyzed the ovarian expression pattern of the EIF4ENIF1 protein in mice. EIF4ENIF1 was co-stained with the germ cell marker DAZL in mouse ovaries at various developmental stages (Fig. 1A). The results showed that EIF4ENIF1 is expressed in ovarian tissues throughout development. The EIF4ENIF1 protein expression level is similar in germ cells and somatic cells within fetal ovaries and relatively elevated in germ cells from the primordial follicle stages after birth. Notably, EIF4ENIF1 formed granule-like structures in early-stage oocytes, most prominently in 0 days post partum (dpp) ovaries, which underwent germ cyst breakdown. These granules contained several known P-body markers, including DDX6, LSM14B and EIF4E, which diminish gradually with oocyte growth (Fig. 1B,C).

We also analyzed EIF4ENIF1 expression patterns in human ovaries using published single-cell RNA-seq data (Gu et al., 2019; Li et al., 2017; Zhang et al., 2018). Human EIF4ENIF1 mRNA levels remained relatively low in female germ cells until the oogenesis stage (Fig. S1A). In all five follicle development stages of oogenesis, the EIF4ENIF1 mRNA levels of the oocytes were significantly higher than those in the corresponding granulosa cells (Fig. S1B).

Consistent ovarian EIF4ENIF1 expression patterns were observed in humans and mice. Moreover, EIF4ENIF1 peptide sequences were highly conserved among species. Thus, we proposed that EIF4ENIF1 might function during oocyte development via a mechanism conserved between humans and mice.

Five heterozygous mutations in EIF4ENIF1 have been described in individuals with clinically reported POI (Fig. S1C; Table 1). The first reported mutation in EIF4ENIF1 was a nonsense mutation that introduces a translation termination codon in place of Ser429. The resulting truncated protein lacks two nuclear export sequences (NES), its C-terminal internal disordered region (IDR) and the LSM14A interacting domain. Subsequently discovered mutations include missense and deletion mutations found in the IDR, N-terminal region and LSM14A-interacting domain. Various mutations correspond to different losses of EIF4ENIF1 protein functions, and pathogenicity arises in heterozygous backgrounds. We thus assumed that the clinically observed EIF4ENIF1 mutations might represent a variety of loss-of-function mutations, and that general pathogenicity could be due to EIF4ENIF1 haploinsufficiency.

We thus constructed an Eif4enif1 heterozygous knockout mouse model to mimic clinical conditions in POI (Fig. 2A; Fig. S1D). Eif4enif1^+/− mice of either sex showed no signs of developmental defects (Fig. S1E,F). We crossed Eif4enif1^+/− mice with wild-type (WT) or Eif4enif1^+/− mice and observed litter conditions among female mice from 4 to 35 weeks of age. Compared with the WT breeding pairs, female heterozygotes mated with WT males exhibited significantly reduced litter size, a higher litter frequency and mildly decreased total litter size (P=0.07) (Fig. 2B,C; Fig. S1G). No Eif4enif1 knockout homozygotes were observed among the offspring of heterozygous breeding pairs, indicating embryonic lethality.

Ovaries were collected from 3-week-old, 20-week-old and 9-month-old WT and Eif4enif1^+/− mice. Ovarian histology showed that Eif4enif1 haploinsufficiency did not result in significant changes in ovary size at any of the three stages, and follicle numbers in 3- and 20-week-old Eif4enif1^+/− ovaries were comparable with those in corresponding WT ovaries. However, ovaries from 9-month-old Eif4enif1^+/− mice showed reduced total follicle numbers and a more ‘solid’ section view (Fig. 2D-L). These results indicate that Eif4enif1 haploinsufficiency leads to middle-aged follicle loss and subfertility.

Mature oocytes remain transcriptionally silent until fertilization, with gene expression mainly regulated at the post-transcriptional level during maturation (De La Fuente et al., 2004). Translational control of gene expression in oocytes is important for proper physiological functioning. As EIF4ENIF1 has been reported to be involved in regulating translation, we hypothesized that Eif4enif1 haploinsufficiency could lead to oocyte dysfunction and translation abnormalities.

We first characterized the EIF4ENIF1 expression pattern upon meiosis resumption. Immunostaining of WT GV, MI and MII oocytes showed that, in GV oocytes, EIF4ENIF1 clustered into bright cloud-like structures spanning the cytoplasm (Fig. 3A). This structure gradually faded as meiosis progressed, surrounded the nuclear region in the MI phase, and was completely disassembled in the MII phase. Western blotting showed that mouse oocyte EIF4ENIF1 protein levels decrease with meiotic progression, being highest at the GV stage and lowest at the MII stage (Fig. 3B,C).

To further investigate how Eif4enif1 haploinsufficiency leads to impaired fertility in female Eif4enif1^+/− mice, we examined the in vitro maturation capacity of WT and Eif4enif1^+/− oocytes. Eif4enif1^+/− oocytes exhibited a significantly decreased germ vesicle breakdown (GVBD) ratio at 4 h of culture and reduced first polar body extrusion (PBE) at 17 h of culture, suggesting that Eif4enif1 deficiency retarded oocyte development in vitro (Fig. 3D-F).

The EIF4ENIF1 protein is involved in translational repression. To examine the change in translation with Eif4enif1 haploinsufficiency, we used T&T-seq, a sequencing method that uses affinity beads to capture actively translating ribosomes and construct libraries for total and ribosome-bound mRNAs (Hu et al., 2022), to evaluate the gene expression states of WT and Eif4enif1^+/− GV oocytes. WT and Eif4enif1^+/− oocytes displayed distinct transcriptomes and translatomes (Fig. 4A; Fig. S3A,B). There were more upregulated than downregulated genes in Eif4enif1^+/− oocytes at both mRNA and translation levels. Numbers of differentially expressed genes (DEGs) between Eif4enif1^+/− and WT groups were similar when measured at mRNA or translation level (Fig. 4B). At the transcriptional level, genes upregulated in Eif4enif1^+/− oocytes were enriched in pathways including chromatin binding, oxidoreduction-driven active transmembrane transporter activity and apoptotic signaling. At the translational level, Eif4enif1^+/− oocyte upregulated genes were enriched in oxidative phosphorylation, catabolic processes and the ERK1/2 cascade (Fig. 4C,D). These alterations suggest that Eif4enif1-haploinsufficient oocytes have abnormal gene expression profiles.

We next tried to find the major biological processes altered upon Eif4enif1 haploinsufficiency. As EIF4ENIF1 is reported to repress translation in somatic cells and there were more upregulated genes at both translation and transcription levels, we mainly focus on the pathways enriched by genes with increased expression in mutant oocytes. When examining mitochondria-related Gene Ontology (GO) biological processes we noted that, in Eif4enif1^+/− oocytes, all 13 mitochondria-encoded genes were upregulated at both translational and mRNA levels (Fig. 4E,F). This result drew our attention to changes in mitochondrial dynamics upon Eif4enif1 haploinsufficiency.

Mitochondrial DNA (mt-DNA) copy number assay and ATP content analysis showed that Eif4enif1^+/− GV oocytes had increased mt-DNA copy number and higher ATP content. ATP content and mt-DNA copy number increased by ∼10% and ∼60%, respectively (Fig. 5B,C). By contrast, mitochondrial membrane potential (MMP) was not significantly altered in Eif4enif1^+/− oocytes (Fig. 5D). MitoTracker Red and JC-1 staining revealed an abnormal mitochondrial distribution pattern in Eif4enif1^+/− oocytes. Mitochondria were distributed relatively evenly in the WT ooplasm, with marginal enrichment in the perinuclear region, whereas in heterozygous oocytes, mitochondria strongly assembled ring-like structures surrounding the nuclear region, with little signal in the outer cytoplasmic region (Fig. 5A).

Mitochondrial dynamics are mediated by mitochondrial fission and fusion (Giacomello et al., 2020). We found that the increased mt-DNA level, elevated ATP level and the perinuclear mitochondrial distribution of Eif4enif1^+/− oocytes resembled the previously reported phenomenon of mitochondrial hyperfusion (Table S1) (Yan et al., 2019; Giacomello et al., 2020; Wakai et al., 2014; Adhikari et al., 2022).

To verify this hypothesis, Eif4enif1^+/− and WT GV oocytes were subjected to transmission electron microscopy to examine mitochondrial appearance. A fraction of Eif4enif1^+/− oocyte mitochondria displayed increased size and a ‘fused’ appearance (Fig. 5F). The average mitochondrial size significantly (P<0.0001) increased in Eif4enif1^+/− oocytes (Fig. 5E). Based on these results, we concluded that Eif4enif1^+/− oocytes undergo mitochondrial hyperfusion.

We also stained the essential fission and fusion regulatory proteins DRP1 (also known as DNM1L) and MFN1 in GV oocytes. These two proteins exhibited similar cellular distributions as the mitochondrial inner membrane protein CYC (CYCS) and the MARDO protein EIF4ENIF1 (Fig. S2A,B). This suggests the possible proximity of these essential mitochondrial dynamics regulatory proteins to MARDO and that disruption of MARDO proteins might interfere with the regulatory functions of these mitochondrial dynamics proteins.

A recent report revealed that mature mammalian oocytes store mRNAs in the MARDO, where polyA-containing mRNAs and RBPs gather around the mitochondria in the ooplasm (Cheng et al., 2022). EIF4ENIF1 is a MARDO RBP component (Cheng et al., 2022).

We stained reported MARDO components in WT and Eif4enif1^+/− surrounded nucleolus (SN) oocytes and found that EIF4ENIF1 co-localizes with CYC (mitochondrial inner membrane protein) in MARDO. In Eif4enif1^+/− oocytes, EIF4ENIF1 displays the same alteration in distribution as mitochondria (represented by CYC), coalescing into ring-like structures in the perinuclear region (Fig. 6A). Intensity profiling along the radial direction of the oocyte showed that the fluorescent signal peaks of EIF4ENIF1 and CYC, which corresponded with dot-like MARDO structures, existed in both perinuclear and peripheral cytoplasmic regions in WT oocytes. In the peripheral region of the Eif4enif1^+/− oocytes, the signal curve became relatively smooth, indicating that few MARDO structures are present (Fig. 6B,C).

The same phenomenon was observed in the staining of polyA-containing mRNAs and other reported MARDO RBP, such as YBX2, LSM14B and DDX6 (Fig. 6D). These MARDO components exhibit a ring-like pattern upon Eif4enif1 haploinsufficiency. Moreover, the mitochondrial dynamics proteins DRP1 and MFN1 showed similar changes in distribution (Fig. S2A,B). These observations suggest that known MARDO components can still be recruited; however, the MARDO pattern was altered and accumulated in the perinuclear region of oocytes upon Eif4enif1 haploinsufficiency.

To examine the effect of EIF4ENIF1 haploinsufficiency on embryonic development, we performed western blot and immunostaining to observe its expression pattern in preimplantation embryos. Western blotting of early embryos showed that EIF4ENIF1 protein levels decreased after cleavage and remained low until the blastocyst stage (Fig. S2E). EIF4ENIF1 appeared to be dispersed in fertilized eggs and formed a granular structure in the peripheral cytoplasmic region. These granules were large and predominant in four- and eight-cell embryos and became inapparent in blastocysts (Fig. 7A).

We collected one-cell stage embryos from WT and Eif4enif1^+/− mice after mating with WT males and observed their developmental conditions 48, 72 and 120 h after human chorionic gonadotropin (hCG) injection during in vitro culture. The number and fertilization rate of embryos from Eif4enif1^+/− mice were comparable with those of WT mice. However, the ratios of fertilized eggs developing into two-cell, four-cell and blastocyst embryos were significantly lower in Eif4enif1^+/− groups (Fig. 7B,C), indicating that Eif4enif1 deficiency could lead to embryo development failure.

We conducted RNA-seq experiments to examine the gene expression profiles of WT and Eif4enif1^+/− one- and two-cell embryos and Eif4enif1^+/− embryos arrested at the one-cell stage (Fig. 7D,E; Fig. S3C-F). Principal component analysis (PCA) showed that, in the direction of PC1, the arrested one-cell embryos fell between normal one-cell and two-cell embryos, indicating that these embryos might be in an intermediate stage in the transition from one-cell to two-cell embryos.

GO analysis showed that, compared with WT one-cell embryos, Eif4enif1^+/− zygotes arrested at the one-cell stage had downregulated pathways related to DNA repair, mitotic cell cycle processes and spliceosomal complexes (Fig. 7E), indicating that these defective embryos might be undergoing cell cycle retardation.

Little is known about the molecular changes occurring in Eif4enif1-deficient oocytes. In contrast to the generally reported role of specifically repressing translation, we demonstrated that in GV oocytes, Eif4enif1 haploinsufficiency significantly altered both the transcriptome and translatome (Fig. 8). There are several possible explanations. The first possibility is due to the cumulative effects of Eif4enif1 haploinsufficiency throughout oocyte development. It is likely that Eif4enif1 haploinsufficiency starts to affect the translation of downstream genes from early oocyte developmental stages, which may include transcription regulators. These downstream genes could further alter the oocyte transcriptome at later stages. A second possible explanation is that EIF4ENIF1 affects mRNA stability instead of directly regulating transcription. Cheng et al. (2022) reported that MARDO functioned in oocyte mRNA storage and decay. It is possible that the composition of MARDO-stored mRNAs was altered upon distribution abnormality brought by EIF4ENIF1 deficiency, resulting in different mRNA storage and decay patterns. Another possible mechanism is through metabolic changes. Our results show that Eif4enif1 haploinsufficiency disrupts oocyte mitochondrial dynamics and impacts mitochondria function such as ATP production. It is well established that transcription is an energy-dependent process. ATP is involved in the functioning of the RNA polymerase II enzyme (RNA Pol II) (Conaway and Conaway, 1988; Sawadogo and Roeder, 1984). Therefore, Eif4enif1 haploinsufficiency might affect RNA Pol II function by changing oocyte cellular ATP levels, disrupting the transcriptome.

The crosstalk between eIF4Es (4Es) and 4E-binding proteins (4E-BPs) is essential in translation regulation. In eukaryotic cells, the universally expressing eIF4E participates in forming the eIF4F complex and helps with the recruitment of ribosome small units to the 5′ cap of mRNAs (Kamenska et al., 2014b). EIF4ENIF1 (4E-T) is capable of binding both eIF4E1 and eIF4E2 (Kamenska et al., 2016), the latter of which is nearly absent in late oocytes and early embryos (Guo et al., 2023). Instead, a germ cell-specific eIF4E1 isoform, eIF4E1b, showed confined expression within the late oocyte to the early embryo time window (Guo et al., 2023; Minshall et al., 2007; Yang et al., 2023). In Xenopus oocytes, eIF4E1b, rather than 1a, is the only 4E isoform that was observed to interact with the CPEB complex which contains EIF4ENIF1 but not 4E-BP1 (Minshall et al., 2007). In mouse GV oocytes, the cellular location of the eIF4E1b protein is similar to EIF4ENIF1 (Guo et al., 2023); however, we failed to observe co-localization of eIF4E with EIF4ENIF1 in oocytes at the GV stage using an antibody that should mainly target eIF4E1a, unlike in primordial oocytes or 293FT cells (Fig. S2C,D). It appears that eIF4E1b/EIF4ENIF1 is the specific 4E/4E-BP pair that takes part in the CPEB-involved translation control in late oocytes and early embryos. Given that EIF4ENIF1 is involved in oocyte mitochondria dynamics, it is worth exploring whether eIF4E1b and CPEB also take part in this process as well as their association with the MARDO.

We observed more upregulated than downregulated genes at both transcriptional and translational levels in Eif4enif1 haploinsufficiency. This was in contrast to the situation in ZAR1 (another MARDO RBP component) knockout oocytes (Rong et al., 2019; Cheng et al., 2022), which also exhibits in vitro maturation (IVM) defects (Rong et al., 2019) and MARDO disruption (Cheng et al., 2022). In addition, mitochondrial hyperfusion and perinuclear MARDO accumulation were observed in Eif4enif1 haploinsufficient oocytes, whereas they appeared to be completely dispersed in Zar1 knockout oocytes (Cheng et al., 2022). It has been previously reported that cells adopt mitochondrial fusion as a complementary strategy when facing stress and energy demands (Youle and van der Bliek, 2012; Meyer et al., 2017) and as stress level continues to increase, mitochondrial morphology tends to become more networked first, then more fragmented (Meyer et al., 2017). Thus, we conjectured that complete knockout and haploinsufficiency of MARDO genes might correspond to ‘harsh’ and ‘mild’ stress levels, respectively, inducing seemingly opposite cellular responses. It is noteworthy that, to date, clinically characterized EIF4ENIF1 mutation-carrying individuals with POI have all been heterozygous. Therefore, the condition of these individuals could be more similar to that in the heterozygous knockout model.

Cheng et al. recently reported a close relationship between mitochondria and GV-enriched RBPs, including EIF4ENIF1, in mammalian oocytes (Cheng et al., 2022). Our results, which showed the proximity of MARDO to essential mitochondrial dynamics regulatory proteins, further suggest a possible role for MARDO in regulating mitochondrial dynamics. However, how Eif4enif1 haploinsufficiency influences the mitochondrial dynamics-MARDO-gene expression network remains unclear. One possibility is that EIF4ENIF1 prevents DRP1 from inhibiting phosphorylation, either by blocking the kinase binding site, competing for kinase activity (as EIF4ENIF1 is an abundant, large protein with multiple phosphorylation sites), or via other indirect mechanisms. Previous studies have reported that PKA-mediated phosphorylation of Ser637 in human DRP1 inhibits its activity and leads to mitochondrial fission defects (Chang and Blackstone, 2007) and that PKA colocalizes with mitochondria in fully grown mouse oocytes (Webb et al., 2008). EIF4ENIF1 haploinsufficiency may increase PKA-mediated Ser637 phosphorylation of DRP1, causing subsequent mitochondrial hyperfusion without dispersing the MARDO structure.

Oocyte development is accompanied by dynamic changes in RNP complexes. P-body-like granules have been observed in early stage mouse oocytes and blastocysts (Flemr et al., 2010). In fully grown oocytes, MARDO shares several RBP components with P-bodies. However, it remains unclear where and how RBPs function during the MII to two-cell phase, where the maternal-to-zygotic transition occurs through the degradation of maternal transcripts and transcription of zygotic RNAs (Li et al., 2013). Further research is necessary to elucidate the role of RBPs during this transition to fill in the blanks in this picture.

Wild-type C57BL/6 mice used in this study were raised in Tsinghua animal facilities or purchased from the Tsinghua University Laboratory Animal Resources Center or Beijing Charles River Laboratories. Mice were housed in specific-pathogen-free animal facilities and raised on a 12-h light/dark cycle with free access to sterile pellet food and water at 22-26°C. The animal protocols used in this study were approved by the Institutional Animal Care and Use Committee (IACUC) of Tsinghua University. Comparisons in this study were made between littermates.

Primary antibodies used were: rabbit anti-EIF4ENIF1 (Abnova, PAB29372), rabbit anti-EIF4ENIF1 (Abcam, Ab95030; for western blot), mouse anti-DAZL (Bio-Rad, MCA2336), mouse anti-DDX6 (Santa Cruz Biotechnology, sc-376433), rabbit anti-eIF4E (Abcam, ab33766), mouse anti-cytochrome C (Santa Cruz Biotechnology, sc-13561), mouse anti-α-tubulin (Easybio, BE0031), rabbit anti-YBX2 (Abcam, AP154829), rabbit anti-LSM14B (Thermo Fisher Scientific, PA5-66371), rabbit anti-DRP1 (Abcam, AP184247) and mouse anti-MFN1 (Proteintech, 66776-1-Ig).

For immunofluorescence, primary antibodies were diluted at 1:100, except for mouse anti-DAZL (1:25) and mouse anti-cytochrome C (1:25). For western blot, primary antibodies were diluted at 1:1000 for rabbit anti-EIF4ENIF1(Abcam) and 1:2000 for mouse anti-α-tubulin.

Secondary antibodies used were: Alexa Fluor 647 donkey anti-mouse IgG (Invitrogen, A-31571), Alexa Fluor 488 donkey anti-rabbit IgG (Invitrogen, A-21206), horseradish peroxidase (HRP) goat anti-mouse IgG (Easybio, BE0102) and HRP goat anti-rabbit IgG (Easybio, BE0101). Secondary antibodies were diluted at 1:1000 for immunofluorescence and 1:2000 for western blot analysis.

Eif4enif1 heterozygous knockout mice were generated at the Xiamen University Laboratory Animal Center on a C57BL/6 background. A pair of sgRNAs were designed to remove the third and fourth exon of the Eif4enif1 gene from the mouse genome, causing a frameshift mutation that led to a nonsense mutation after the 32nd amino acid in the corresponding polypeptide. The remaining product was 3% of the original polypeptide and was considered nonfunctional. Successful editing was confirmed using genomic PCR of the joined fragments flanking the third and fourth exon sites and Sanger sequencing. F0 mice with the desired edit were mated with WT C57BL/6 mice and offspring were selected for further experiments. Frozen sperms of this strain are stored at Tsinghua Laboratory Animal Research Center.

Female mice were continuously mated with fertile males from 4 to 35 weeks of age. Litter duration and size were recorded at the beginning of each week. The total number of pups was calculated by summation of all litter sizes.

Ovarian tissues were dissected and fixed in 4% paraformaldehyde (PFA) for 48 h. After dehydration in serial ethanol and dimethylbenzene, samples were embedded in paraffin with the maximum projection face perpendicular to the embedded surface and continuously sectioned at 5 μm thickness. One of every five successive sections was selected and three consecutive sections were placed on glass slides. Sectioned samples were subjected to Hematoxylin-Eosin (HE) staining and examined using a histological scanner (3DHISTECH Pannoramic Scan). The section with the maximum ovary tissue diameter was determined using the CaseViewer software (version 2.4, 3DHISTECH), and all ovary sections on five consecutive slides surrounding the one containing the maximum section were selected for follicle counting. Follicles were classified according to well-established criteria as primordial, primary, secondary, antral and atretic, and the average number of follicles was calculated for comparison.

Paraffin-embedded tissue sections were dewaxed, rehydrated in dimethylbenzene and graded in ethanol solutions. Antigen retrieval was performed using a citrate buffer (3 g/l sodium citrate tribasic dihydrate, 0.366 g/l citrate, pH 6.0) at 95°C for 10 min. Samples were permeabilized with 1% Triton X-100 in PBS-T (0.1% Tween-20 in PBS) for 10 min and blocked in BDT buffer {0.3% Triton X-100 in blocking solution [3% (w/v) bovine serum albumin (Sigma-Aldrich, A3083) in PBS-T]} for 60 min at room temperature (25°C). Samples were then incubated with primary antibodies diluted in BDT buffer at 4°C overnight. After incubation, samples were washed thrice with PBS-T and incubated with secondary antibodies diluted in BDT buffer for 60 min at room temperature. Subsequently, sections were washed thrice with PBS-T and stained with 4′,6-diamidino-2-phenylindole (DAPI) for 15 min at room temperature. Finally, samples were soaked in an anti-fade reagent (Thermo Fisher Scientific, P10144) to prevent quenching and covered with a coverslip. Slices were sealed with nail polish and dried overnight before confocal microscopy.

Female mice (3-4 weeks of age) were sacrificed. Dissected ovaries were chopped with blades in M2 medium (Sigma-Aldrich, M7167) containing 10 μM cilostamide (Cayman Chemical, 14455), which prevented spontaneous GVBD. GV oocytes were collected via mouth pipetting.

Oocytes were collected in PBS and lysed in 2× RIPA buffer (Bestbio, BB-32013). Samples were incubated at 95°C for 10 min with the addition of sodium dodecyl sulfate polyacrylamide gel electrophoresis (s.d.S-PAGE) sample buffer (GenStar, E53-01). Samples were then separated by 10% s.d.S-PAGE and transferred onto 0.22 μm polyvinylidene fluoride (PVDF) membranes (Millipore, ISEQ00010). Membranes were blocked in 5% nonfat milk in TBS-T (0.1% Tween-20 in Tris-Buffered Saline) and incubated with primary antibodies at 4°C overnight. After incubation, membranes were washed thrice with TBS-T and incubated with secondary antibodies. After washing again with TBS-T three times, bands were visualized using Immobilon Western reagents (Millipore, WBKLS0500) in a ChemiDoc XRS+ imaging system (Bio-Rad).

GV oocytes collected as described above in cilostamide-containing M2 medium were transferred and washed in M16 medium [prepared in house: 10% fetal bovine serum (Gemini, 900-108), 0.16% penicillin-streptomycin (Gibco, 5070063), 0.228% sodium pyruvate (Gibco, 11360070), 0.05U/ml FSH (Solarbio, F8470) in α-MEM (Gibco, 12561049)] to remove cilostamide, then placed in a cell incubator. The GVBD rate at 4 h and PBE rate at 17 h of culture were observed and recorded under a Ti-e microscope (Nikon).

Oocytes and early embryos were washed in acidic Tyrode's solution to remove the zona pellucida and fixed in 1% PFA overnight at 4°C or 4% PFA for 10-15 min at room temperature. Oocytes were then washed three times for 10 min each in PBS-T and incubated in penetrating solution (0.25% Triton X-100 in PBS-T) for 25 min at room temperature. After penetration, oocytes were washed three times, transferred into a blocking solution [3% (w/v) bovine serum albumin (Sigma-Aldrich, A3083) in PBS-T] and blocked for 1 h at room temperature. Oocytes and early embryos were then incubated with primary antibodies diluted at respective titrations in a blocking solution overnight at 4°C. After washing three times in PBS-T, oocytes and early embryos were incubated with secondary antibodies and DAPI in PBS-T for 1 h and 30 min at room temperature, respectively. After incubation, oocytes and early embryos were washed three times with PBS-T and mounted on a glass-bottomed dish or glass slide for confocal microscopy.

Oocytes were fixed in 1% PFA overnight at 4°C and permeabilized in 70% ethanol for 20 min at room temperature. Oocytes were then rehydrated in washing buffer [2× SSC buffer (Sangon Biotech, B548109) with 10% formamide (Sigma-Aldrich, F9037)] and incubated with 0.5 μM oligo dT30-cy5 probe (RiboBio) at 37°C overnight. Following incubation, oocytes were washed three times in washing buffer for 20 min at 37°C, incubated with DAPI in 2× SSC buffer for 1 h at room temperature and again washed once with 2× SSC buffer. Oocytes were then transferred to PBS-T for further imaging.

ATP content was analyzed according to the manufacturer's instructions using an Enhanced ATP Assay Kit (Beyotime, S0027). Briefly, oocytes were lysed in an ATP lysis buffer. Samples and standard ATP solutions were added to the diluted ATP detection solution in a flat white 96-well plate, and sample luminescence was measured using a luminometer. ATP concentration was calculated using a standard curve.

Between 13.00 and 15.00, female mice of 4-5 weeks of age were intraperitoneally injected with 5 IU of pregnant mare serum gonadotropin (PMSG) (NSHF; 50 IU/ml dissolved in sterile filtered saline) followed by hCG (NSHF; 50 IU/ml dissolved in sterile filtered saline) 46-48 h later. Female mice were mated with fertile WT male B6 mice immediately after hCG injection and examined for vaginal plug formation the following morning. Mice with vaginal plugs were euthanized and their oviducts were dissected. Ampullar portions of the oviducts were torn in the M2 medium to release the granulosa-surrounded embryos. Secondary polar bodies were characteristic of the fertilized eggs. Embryos were briefly digested in hyaluronidase (Yuanye, S10060; 1 mg/ml, dissolved in M2 medium) to remove surrounding granulosa cells, transferred to KSOM medium (Sigma-Aldrich, MR-106) and placed in a cell incubator for further development.

The culture medium was added to a 35 mm Petri dish and covered with mineral oil (Sigma-Aldrich, M8410). Oocytes and embryos were cultured at 37°C in a humidified atmosphere containing 5% CO₂ in a cell incubator. The oil-covered medium was preequilibrated under culture conditions for at least 30 min before use. Cell developmental stages were observed using a Nikon Ti Eclipse microscope.

Library construction for RNA-seq was performed using the SMART-seq2 strategy. Reverse transcription was performed according to the manufacturer's instructions using a Single Cell Full Length mRNA Amplification Kit (Vazyme, N712). The library was constructed using a TruePrep DNA Library Prep Kit V2 for Illumina (Vazyme, TD502).

T&T-seq was performed according to a protocol previously established by Hu et al. (2022). Briefly, 30 oocytes from each group were lysed in sample buffer (Vazyme, N712) and divided into two subgroups at a 20:80 ratio. The major subgroup was used for translatome sequencing and the minor subgroup for transcriptome sequencing: 20% of the lysate was used directly for reverse transcription and library construction, as described above, and 80% of the lysate was incubated with pre-activated ribosome magnetic beads (Immagina Biotechnology, RL001) to capture actively translating ribosomes. The beads were then washed and the captured ribosomes were digested with proteinase K to release bound mRNAs. The released mRNAs were purified using clean VAHT RNA beads (Vazyme, N412-01-AA) for reverse transcription and library construction, as described above.

The processing of T&T-seq and RNA-seq data is roughly the same as that in the original study (Hu et al., 2022). Briefly, data generated from sequencing experiments were trimmed of adaptors, removed of duplicates, mapped to the mm10 genome and counted of mapped reads using Trim Galore (v0.6.4), Hisat2 (v2.1.0), Samtools (v1.6) and Featurecounts (v1.6.5). Count data were converted into TPM values and heatmaps were constructed using TBtools (v2.008) (Chen et al., 2020). DEGs were analyzed by DESeq2 (v1.26.0).

GV oocytes were incubated with 200 nM Mito-Tracker Red CMXRos (Beyotime, C1035) dissolved in M2 medium containing 10 μM cilostamide for 30 min at 37°C. After incubation, oocytes were washed three times in fresh cilostamide-containing M2 medium, fixed overnight in 1% PFA at 4°C and stained with DAPI for 30 min at room temperature. Oocytes were then washed thrice in PBS-T, mounted on glass slides and covered with coverslips for confocal microscopy.

Denuded GV oocytes were collected in an M2 medium containing 10 μM cilostamide, lysed in DNA lysis buffer (50 mM Tris-HCl, 1 mM EDTA, 0.5% Tween-20, 2 mg/ml proteinase K) for 120 min at 55°C and heated for 10 min at 95°C to inactivate proteinase K.

The mt-DNA copy number was directly assessed by quantitative PCR using the oocyte lysates prepared as described above. The mt-DNA copy number was calculated as ⁠. A shared codon sequence of histone 3, having 12 copies in a single set of chromosomes, was selected as an internal control.

The mt-DNA copy numbers were normalized to the average of the WT groups in each experiment to obtain comparable results. Normalized mt-DNA copy numbers from three repetitive experiments were pooled for unpaired two-tailed Student's t-tests.

The primers used were: Cox2-F, ATAACCGAGTCGTTCTGCCAAT; Cox2-R, TTTCAGAGCATTGGCCATAGAA; H3-F, CTACCAGAAGTCGACCGAGC; H3-R, GTCCTTGGGCATGATGGTGA.

GV oocytes were incubated with 200 nM JC-1 (Beyotime, C2006) and dissolved in M2 medium containing 10 μM cilostamide, centrifuged before use, for 30 min at 37°C in the dark. After incubation, oocytes were washed three times in fresh cilostamide-containing M2 medium and transferred to another drop of cilostamide-containing M2 medium for live-cell confocal microscopy. The intensities of red (polymer) and green (monomer) JC-1 fluorescence were quantified using Nikon NIS-Elements (v5.41).

GV oocytes were fixed in a mixture of paraformaldehyde (2%) and glutaraldehyde (2.5%) for 1 h at room temperature and washed four times with PB buffer (0.1 M). The tissue underwent post-fixation with osmium tetroxide (1%) and tetrapotassium hexacyanoferrate trihydrate (1.5%) for 2 h at 23°C, followed by ethanol dehydration in graded solutions (50, 70, 80, 90, 100, 100 and 100%) for 10 min each followed by 1, 2-epoxypropane twice for 10 min each and infiltrated with a mixture of 1, 2-epoxypropane and Epon 812 resin for 8 h (SPI Supplies). Subsequently, tissues were incubated in pure Epon 812 twice and polymerized in an oven at 60°C. Polymerized resin blocks were sectioned using a Leica EM UC7 ultramicrotome. Ultrathin sections (80 nm) were mounted and dried on coated copper grids. Grid-mounted sections were stained with 2% uranyl acetate (25 min) and lead citrate (5 min). Imaging was performed using an FEI Tecnai Spirit transmission electron microscope.

At least ten fields were analyzed for each single oocyte. Mitochondrial identification was achieved by training a segmentation AI using v and manual binary editing. The mitochondrial area was calculated using the NIS-Elements software.

Statistical details, including the statistical test type used, replicate numbers and statistical significances, are specified in the figure legends. All statistical analyses were performed using the Prism software (v8.4.0). For Student's t-test, equal variance and normal distribution are assumed. Error bars indicate standard deviation (s.d.).

We thank the staff at the Tsinghua Nikon Imaging Center for their help with confocal and transmission electron microscopy image analyses. We thank the staff at the Tsinghua Laboratory Animal Research Center for their help with our animal-related experiments. We thank the staff at the Tsinghua Cell Biology Facility for their help with electron microscopy sample preparation. We thank L. Xie for helpful discussions and suggestions.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202151.reviewer-comments.pdf