Global SUMOylation in mouse oocytes maintains oocyte identity and regulates chromatin remodeling and transcriptional silencing at the end of folliculogenesis Free

In mammals, oocyte development begins embryonically but is not completed until the oocyte is fertilized following ovulation in adults. Prior to birth, mouse oocytes begin meiosis and proceed through the early stages of prophase I to arrest in the diplotene stage (Bullejos and Koopman, 2004; Schindler, 2011). Oocytes remain arrested in prophase I while stored as primordial follicles and following recruitment into the growing pool of follicles, where they develop to fully grown oocytes capable of resuming meiosis, i.e. meiotic competence. During folliculogenesis, prophase I oocytes are transcriptionally active, producing and storing transcripts needed for meiotic maturation and early embryogenesis; once fully grown, oocyte transcription is silenced and the chromatin undergoes structural changes within the nucleus (also called the ‘germinal vesicle’ or GV) (De La Fuente and Eppig, 2001; Albertini, 2002). Following the luteinizing hormone (LH) surge, oocytes resume meiosis, the nuclear envelope breaks down (GV breakdown or GVBD), and the process of spindle assembly begins and ovulation occurs. Oocytes proceed through pro-metaphase to reach metaphase I (MI) and must satisfy the spindle assembly checkpoint (SAC) before homologous chromosomes are separated and the first polar body is extruded, marking the end of meiosis I. A second meiotic arrest then occurs at metaphase II (MII), where the oocytes, now referred to as eggs, remain arrested until fertilization (Tripathi et al., 2010). Oocytes are transcriptionally silent throughout meiotic maturation and transcription does not resume until after fertilization when zygotic genome activation (ZGA) occurs during the maternal-to-zygotic transition (Lee et al., 2014; Aoki, 2022).

SUMOylation is an essential covalent post-translational modification wherein a small ubiquitin-like modifier protein (SUMO1-4) is attached to a lysine of the target protein at a specific sequence. SUMOylation uses a stepwise enzymatic cascade (E1 ‘activating’, E2 ‘conjugating’ and E3 ‘ligase’ enzymes), similar to ubiquitination. There is only one known E2 enzyme in the SUMOylation pathway, UBE2I (also known as UBC9) and mice lacking Ube2i die early post-implantation with defects in chromosome condensation and segregation and a range of nuclear structural defects (Nacerddine et al., 2005). SUMOylation impacts diverse functions, but predominantly targets nuclear proteins and regulates DNA replication, DNA damage repair and chromatin (Neyret-Kahn et al., 2013; Cossec et al., 2018; Garvin, 2019; Theurillat et al., 2020; Abbas, 2021; Vertegaal, 2022). SUMOylation also maintains mouse embryonic stem cell identity and pluripotency (Nacerddine et al., 2005; Cossec et al., 2018; Vertegaal, 2022). Model systems, particularly yeast, have shown SUMOylation as a key regulator of meiosis, chromatin configuration, and transcriptional activity, but the role of SUMOylation during mammalian oocyte development is largely uncharacterized, particularly during the critical growth phase within ovarian follicles when prophase I-arrested oocytes attain meiotic and developmental competence (Buccione et al., 1990; Zhu et al., 2010; Yuan et al., 2014; Nottke et al., 2017; Feitosa et al., 2018; Feitosa and Morris, 2018; Rodriguez et al., 2019; Bhagwat et al., 2021).

We previously generated a mouse model with conditional deletion of Ube2i in oocytes beginning at the primordial follicle stage (i.e. prior to the start of folliculogenesis) (Rodriguez et al., 2019). Female mice were sterile with premature depletion of both resting and growing ovarian follicles in early adulthood, reminiscent of primary ovarian insufficiency. Loss of both resting (primordial) and growing follicles suggested that oocyte SUMOylation was required for the stability of both populations (Rodriguez et al., 2019), but it was unknown whether loss of primordial follicles was directly due to oocyte defects or an indirect consequence caused by loss of regulatory factors from growing follicles (Durlinger et al., 1999, 2002; Mizunuma et al., 1999; Fortune, 2002). To test this, we generated a Ube2i oocyte knockout using Zp3-cre+, an oocyte-specific cre recombinase line that deletes loxP-flanked (‘floxed’) alleles only in growing follicles (Lan et al., 2004). Here, we show that Ube2i Zp3-cre+ female mice are sterile but have a stable ovarian reserve. Phenotype analysis of Ube2i Zp3cre+ mice uncovered additional roles for UBE2I in mouse oocytes that include establishing a fully condensed chromatin configuration, regulating global transcriptional silencing and preventing premature activation of genes typically associated with ZGA. Thus, our approach uncovered SUMOylation as a major regulatory process for key events in mouse oocyte development, which must be established by the end of folliculogenesis.

To determine the role of UBE2I in growing oocytes, male mice carrying cre recombinase under control of regulatory sequences of the mouse zona pellucida 3 (Zp3) gene (Zp3-cre) were crossed to Ube2i^loxP/loxP female mice (Cox et al., 2021) to delete Ube2i in oocytes within growing primary follicles (Fig. 1A). This conditional allele of Ube2i deletes exons 3 and 4, whereas our prior conditional knockout model deletes Ube2i exons 2 and 3 (Rodriguez et al., 2019); both Ube2i conditional alleles generate null alleles following recombination (Rodriguez et al., 2019; Cox et al., 2021). We validated that the conditional deletion of Ube2i^loxP/loxP (Δ3-4) using Gdf9-icre+ generates a similar phenotype to Ube2i^loxP/loxP Gdf9-icre+ (Δ2-3) (Rodriguez et al., 2019) by testing fertility over 6 months and assessing ovarian histology. Both the new Ube2i^loxP/loxP (Δ3-4) Gdf9-icre+ and the published Ube2i^loxP/loxP Gdf9-icre+ (Δ2-3) were female sterile, with full depletion of ovarian follicles by 6 months of age (Fig. S1). The studies herein used the Ube2i^loxP/loxP (Δ3-4) allele for conditional deletion.

Ube2i^loxP/loxP Zp3-cre+ oocytes (termed ‘Ube2i Zp3-cre+’) were validated for loss of Ube2i using quantitative reverse transcription PCR (RT-qPCR) and indirect immunofluorescence. By RT-qPCR, fully grown, prophase I-arrested GV oocytes isolated from Ube2i Zp3-cre+ antral follicles showed a 99.4% reduction in Ube2i transcripts compared with transcript levels in control (Ube2i^loxP/loxP) littermates (P<0.0001) (Fig. 1B). Indirect immunofluorescence of UBE2I demonstrated loss of the UBE2I protein specifically in oocytes of growing follicles but not in oocytes in primordial follicles in Ube2i Zp3-cre+ mice (Fig. S2). Specificity of oocyte deletion was verified by RT-qPCR of Ube2i in granulosa cells and the presence of UBE2I protein in cumulus cells between genotypes (Figs S2 and S3). To determine whether loss of Ube2i in oocytes of growing follicles affected fertility, 6-week-old control or Ube2i Zp3-cre+ female littermates were pair-housed and continually mated with sexually mature wild-type males for 6 months. Although control females produced an average (±s.e.m.) of 6.4±0.5 pups per litter, Ube2i Zp3-cre+ females failed to produce any litters, demonstrating that oocyte-specific deletion of Ube2i using Zp3-cre+ causes female sterility (Fig. 1C, P=0.0003).

As both Ube2i Gdf9-icre+ models (Ube2i Δ3-4 and the published Ube2i Δ2-3) show complete depletion of all ovarian follicles by 6 months of age (Fig. S1) (Rodriguez et al., 2019), we assessed whether loss of Ube2i in oocytes of growing follicles using Zp3-cre+ also affected stability of the ovarian reserve. Ovaries from 6-month-old control and Ube2i Zp3-cre+ females were collected, and histology and ovarian follicle morphometrics determined. Unlike 6-month-old Ube2i Gdf9-icre+ ovaries (Rodriguez et al., 2019) (Fig. S1), 6-month-old Ube2i Zp3-cre+ ovaries were a similar size as that of control ovaries and contained similar follicle stages, including corpora lutea, indicating that ovulation had occurred (Fig. S4). Although ovarian histology was grossly similar between genotypes, some defects were noted in cumulus cells surrounding mutant oocytes. Although it was compact, the cumulus cell layer around oocytes of antral follicles showed fewer layers in Ube2i Zp3-cre+ ovaries (Fig. S4E,F). A linear regression analysis of cumulus cell width and antral follicle diameter showed that Ube2i Gdf9-icre+ cumulus cells maintained the positive correlation with follicle diameter similar to that of controls, but had an altered growth pattern (Fig. S5). By morphometric analysis, Ube2i Zp3-cre+ ovaries contained similar numbers of all follicle stages compared with those for control littermates, except for a small but significant decrease in the number of primary follicles (Fig. S4G). In line with maintenance of the ovarian reserve, Ube2i Zp3-cre+ female mice had similar steady state levels of estradiol and follicle-stimulating hormone (FSH) as those of their littermate controls (Fig. 1D,E). Taken together, these results indicate that although loss of Ube2i in growing oocytes caused sterility, it minimally disrupted folliculogenesis or ovarian hormonal feedback.

The presence of corpora lutea in histological sections of ovaries of Ube2i Zp3-cre+ mice (Fig. S4B,D) suggest that these mice ovulate despite being sterile. Therefore, we assessed the ability of mice to ovulate using stimulation by exogenous gonadotropin injection (‘superovulation’). Ube2i Zp3-cre+ females ovulated significantly fewer oocytes/eggs (P=0.0138) compared with control females (Fig. 1F). As expected, within the ampulla of the oviduct, eggs from control females were enclosed in an expanded cumulus-oocyte-complex. However, those ovulated from Ube2i Zp3-cre+ females had fewer layers of expanded cumulus cells (Fig. S6). In addition, control mice ovulated MII-stage eggs with visible polar bodies as expected; however, ovulated oocytes from Ube2i Zp3-cre+ females lacked a visible polar body (Fig. 1G-I). To verify the meiotic stage of the ovulated oocytes, we analyzed for the presence of a spindle via α-tubulin immunostaining and DNA labeling. Control ovulated eggs contained a well-organized MII spindle with chromosome alignment, whereas Ube2i Zp3-cre+ oocytes had disorganized spindles, misaligned chromosomes and no polar body, thus appearing to arrest in MI (Fig. 1J,K).

We then determined the ability of fully grown, prophase I-arrested GV oocytes to undergo meiotic resumption and progression in vitro. Fully grown, prophase I-arrested GV oocytes were harvested from control and Ube2i Zp3-cre+ ovaries following hormone stimulation into medium containing 2.5 µM milrinone, which maintains meiotic arrest, and any remaining cumulus cells were manually removed. Oocytes were collected, washed to remove the milrinone, then cultured for 8.5 h, fixed and analyzed for GVBD and meiotic progression. In this time frame, the majority of control oocytes progressed from GV to MI, but the progression of Ube2i Zp3-cre+oocytes was significantly altered (P=0.0077) (Fig. 1L). Similar percentages of oocytes died between genotypes (control, 5.3%; Ube2i Zp3-cre+, 8.8%) or reached pro-metaphase (control, 7.3%; Ube2i Zp3-cre+, 6.1%). However, the percentage of Ube2i Zp3-cre+ oocytes that remained at the GV stage (27.2%) was nearly double that of control oocytes (14.6%), and control oocytes were also more successful at reaching MI (72.8%) than Ube2i Zp3-cre+ oocytes (57.9%) (Fig. 1L). These data indicate that Ube2i Zp3-cre+ oocytes have reduced meiotic competence compared with that of control oocytes.

In vitro-matured MI oocytes were further analyzed for spindle architecture and chromosome alignment by confocal microscopy using anti-α-tubulin antibodies for microtubules and Hoechst 33342 (hereafter, ‘Hoechst’) to detect DNA. For classification, oocytes with multiple chromosomes lagging or not aligned on the metaphase plate were classified as having ‘poor’ alignment, whereas oocytes with all or nearly all chromosomes aligned at the metaphase plate were classified as having ‘good’ alignment. The majority of control MI oocytes had good chromosome alignment (81.2%) (Fig. 2A-C,G). In contrast, most of Ube2i Zp3-cre+ MI oocytes had poorly aligned chromosomes (60.6%; P<0.0001) (Fig. 2D-G). In addition to misaligned chromosomes, Ube2i Zp3-cre+ oocytes often displayed abnormal spindles, including those that appear stretched, shortened or disorganized (Fig. 2D-F). To quantify differences in spindle length and width, MI spindles were measured from pole to pole and at the widest point of the mid-spindle. Ube2i Zp3-cre+ spindles were significantly shorter (P=0.0444) and narrower (P=0.0285) than control MI spindles (Fig. 2H,I). Chromosome spreads from in vitro-matured MI oocytes showed that Ube2i Zp3-cre+ oocytes successfully formed bivalents; however, they appeared less condensed (Fig. 2J,K).

Because of the misaligned chromosomes, we further analyzed whether microtubule attachment to kinetochores was disrupted in Ube2i Zp3-cre+ oocytes. We used a cold-stable assay as low temperatures depolymerize microtubules not attached to kinetochores, whereas microtubules attached to kinetochores are relatively stable (Ma and Schultz, 2013). Fully grown, prophase I-arrested GV oocytes (stripped of cumulus cells) were cultured to mid-MI (7 h), then placed in ice-cold minimum essential medium (MEM) prior to fixation and analyzed for microtubules (α-tubulin) and kinetochores (CREST). Owing to the abnormal chromosome structure and poor alignment in Ube2i Zp3-cre+ oocytes at MI (Fig. 2), it was not possible to determine the precise number of kinetochore-microtubule attachments, so only the presence or absence of cold-stable microtubules was scored. Control oocytes were significantly more likely to have cold-stable microtubules (97.2%) than Ube2i Zp3-cre+ oocytes (79.2%, P=0.0333) (Fig. 3A-C). Because of the impaired kinetochore-microtubule interaction in Ube2i Zp3-cre+ oocytes, we further analyzed for changes in acetylation of histone H4 at lysine 16 (H4K16ac), as this histone modification maintains kinetochore function in somatic cells and possibly also in oocytes (Choy et al., 2011; Ma and Schultz, 2013). Fully grown control and Ube2i Zp3-cre+ oocytes were analyzed at GV (Fig. 3D-F), pro-metaphase (Fig. 3G-I) and MI (Fig. 3J-L) stages by immunofluorescence and the signals quantified by confocal microscopy. The relative amount of H4K16ac was unchanged between genotypes at any stage, indicating that Ube2i Zp3-cre+ oocytes have reduced kinetochore function than control oocytes, which is not due to differences in H4K16ac levels at the GV, pro-metaphase or MI stages.

Prior to meiotic resumption, fully grown, prophase I-arrested GV oocytes undergo chromatin remodeling, which changes the configuration of the chromatin around the nucleolus. This change involves the transition from a less compact, non-surrounded nucleolus (‘NSN’) configuration, to a more compact, surrounded nucleolus (‘SN’) configuration (Mattson and Albertini, 1990; Debey et al., 1993; Ma et al., 2013). In isolated fully grown GV oocytes, the majority (87%) of control oocytes were in the SN configuration as expected (Fig. 4A,B,E). Compared with control oocytes, significantly fewer Ube2i Zp3-cre+ oocytes were in the SN configuration (18%) (Fig. 4C-E, P<0.0001). No difference in oocyte diameter was found between control and Ube2i Zp3-cre+ samples (Fig. 4F); thus, the difference cannot be attributed to Ube2i Zp3-cre+ oocytes not reaching a fully grown oocyte size. The transition from NSN to SN is also associated with transcriptional silencing (Ma et al., 2013), and thus we analyzed transcriptional activity of control and Ube2i Zp3-cre+ oocytes using 5-ethynyl-uridine (5EU) labeling. As expected, fully grown control GV oocytes in the SN configuration showed no fluorescent 5EU signal, (Fig. 5A), whereas control NSN oocytes had a positive fluorescent signal (Fig. 5B). Both transcriptionally active and inactive Ube2i Zp3-cre+ oocytes were identified (Fig. 5C,D), but there was a significantly higher proportion of Ube2i Zp3-cre+ oocytes that were transcriptionally active (77%, P<0.0001) compared with similar sized control oocytes (28%) (Fig. 5E,F), suggesting that Ube2i Zp3-cre+ oocytes fail to suppress transcription.

Histone modifications such as methylation are associated with transcriptional activity in GV oocytes, and trimethylation of histone H3 at lysine 3 (H3K4me3) and lysine 9 (H3K9me3) in mouse oocytes are markers of transcriptional repression (Smith et al., 2022). Because of the active transcription observed in Ube2i Zp3-cre+ oocytes (Fig. 5), we analyzed overall H3K4me3 and H3K9me3 levels in fully grown, prophase I-arrested GV oocytes by immunofluorescence (Fig. 6). Compared with control GV oocytes, Ube2i Zp3-cre+ GV oocytes had significantly reduced levels of both H3K4me3 (34%, P<0.0001, Fig. 6A-C) and H3K9me3 (27%, P<0.0001, Fig. 6D-F), supporting the lack of transcriptional repression in the mutant oocytes.

Because Ube2i Zp3-cre+ fully grown, prophase I-arrested oocytes fail to silence transcription, we performed RNA sequencing of individual fully grown, prophase I-arrested oocytes from control and Ube2i Zp3-cre+ mice (n=17 oocytes per genotype) to determine changes in the transcriptome. Principal component analysis (PCA) identified two groups with 70% variance in principal component (PC) 1 (PC1), which readily separated the samples by genotype; 15% variance was explained by PC2, which showed a wider sample spread for Ube2i Zp3-cre+ oocytes than for control oocytes (Fig. 7A). Using a cutoff of |log₂(fold change or FC)|≥1 and adjusted P-value (P_adj)<0.05, we identified 7603 differentially expressed genes (DEGs) in Ube2i Zp3-cre+ oocytes compared with control oocytes (Fig. 7B; Table S1). Of these DEGs, more upregulated genes (6308) were found compared with the number of downregulated genes (1295). To investigate biological functions associated with the DEGs, Gene Ontology (GO) analysis was performed for upregulated and downregulated DEGs (Sherman et al., 2022). Separate analysis of either upregulated and downregulated genes returned the same top GO term: ‘regulation of transcription from RNA polymerase II promoter’. Differentially upregulated genes also returned the GO terms ‘signal transduction’, ‘multicellular organism development’, ‘cell differentiation’ and ‘cell adhesion’, whereas downregulated genes returned the GO terms ‘positive regulation of transcription from RNA polymerase II promoter’, ‘multicellular organism development’, ‘cell differentiation’ and ‘DNA-templated regulation of transcription’ (Fig. 7C).

Key oocyte-expressed genes were differentially regulated in Ube2i Zp3-cre+ fully grown, prophase I-arrested oocytes (Fig. 8A). Genes downregulated in Ube2i Zp3-cre+ oocytes included Bmp15, Gdf9, H1oo (also known as H1f8), Hormad1, Mater (or Mlrp5), Msy2 (Ybx2), Npm2, Oosp1, Rfpl4, Tgfb2, Zar1, Zp1, Zp2 and Zp3, whereas upregulated genes include Blimp1 (Prdm1), Pou5f1 (Oct4) and Trim28. Of note, both Bmp15 and Gdf9 are also significantly downregulated in Ube2i Gdf9-icre+ ovaries (Rodriguez et al., 2019) and may contribute to a reduced proliferation of the cumulus cell layer in antral follicles of Ube2i Zp3-cre+ ovaries and the reduction in cumulus expansion (Figs S4-S6). The oocyte-specific transcription factor Nobox was significantly upregulated in Ube2i Zp3-cre+ oocytes (Fig. 8B). Nobox was the only oocyte-specific transcription factor found to be differentially regulated. As both H3K4me3 and H3K9me3 were downregulated in Ube2i Zp3-cre+ oocytes (Fig. 6), we examined the DEGs for changes to genes encoding histone lysine methyltransferases and demethylases. The methyltransferase genes that were upregulated in Ube2i Zp3-cre+ oocytes included Kmt2e, Setd1b, Setdb1 and Setdb2, whereas Setd1a, Smyd2 and Suv39h2 were downregulated (Fig. S7A). The demethylase genes that were upregulated included Kdm4b, Kdm4c, Kdm4dl, Kdm5a and Kdm5b, whereas Kdm2b and Kdm7a were downregulated (Fig. S7B).

Major chromatin remodeling is not required for global transcriptional silencing in GV oocytes (De La Fuente et al., 2004). However, to verify that transcriptome changes were not simply due to persistence of the NSN configuration of mutant GV oocytes, the DEG list from Ube2i Zp3-cre+ oocytes was compared with gene expression profiles previously reported to be associated with SN and NSN oocytes (Ma et al., 2013). The genes downregulated in wild-type SN oocytes compared with their expression in wild-type NSN oocytes represent only 2% (152/7603) of the DEGs in the Ube2i Zp3-cre+ oocytes. Similarly, the genes upregulated in wild-type SN oocytes compared with their expression in wild-type NSN oocytes comprise a mere 3% (226/7603) of the Ube2i Zp3-cre+ DEGs (Fig. S8). The overlapping genes are provided in Table S2. Thus, the predominance of NSN oocytes in Ube2i Zp3-cre+ oocytes is unlikely to fully explain the large transcriptomic changes in the mutant Ube2i Zp3-cre+ oocytes.

To further analyze the differences in gene expression, Ingenuity Pathway Analysis (IPA) (https://digitalinsights.qiagen.com/IPA) of the DEGs was used to identify the top upstream regulators. This analysis revealed that several of the top upstream regulators were associated with ZGA, including Dppa2, Dppa4, Dux, Sox2 and Zscan4 genes (Fig. 8C) (Kramer et al., 2014). In the RNA-sequencing (RNA-seq) dataset, Dppa4 expression was significantly decreased in Ube2i Zp3-cre+ oocytes (FC=−3.8; P<0.0001), whereas the expression levels of Duxf3 (FC=+203.2; P<0.0001) and Zscan4c (FC +14.3; P<0.0001,) as well as those of Zscan4a, Zscan4b, Zscan4d, Zscan4e and Zscan4af were significantly increased; Dppa2 (FC=0.96; P=0.7) and Sox2 (FC=2.4; P=0.08) expression was unchanged between genotypes. Expression of the ZGA-associated genes, Duxf3 and Zscan4, was validated in independent samples of fully grown, prophase I-arrested control and Ube2i Zp3-cre+ GV oocytes by RT-qPCR, confirming that these genes were upregulated in Ube2i Zp3-cre+ oocytes compared with control oocytes (Fig. S9). Other ZGA-associated genes such as Eif1a, encoding an embryonic transcription factor, and Tmem92, encoding a downstream target of ZSCAN4, had significantly higher expression as seen in the list of DEGs from the RNA-seq dataset as well as with RT-qPCR validation (Fig. S8).

In mice, ZGA occurs in two waves: the minor ZGA beginning in the S phase of one-cell embryos, and the major ZGA beginning from mid to late two-cell stage (Aoki, 2022). Comprehensive transcriptome analysis of MII, one-cell-stage and two-cell-stage embryos by RNA-seq previously identified five classes of genes according to changes in expression during progression of MII to two-cell embryos (Abe et al., 2018). We used informatic analysis to overlap our DEGs with the published major ZGA genes (clusters 1 and 2), minor and major ZGA genes (cluster 4), and minor ZGA genes (cluster 5) [gene sets derived from the DNA Data Bank of Japan (DDBJ) Sequence Read Archive accession number DRA006557; Abe et al., 2018]. Overlapping gene sets with a |log₂FC|≥1 and P_adj<0.05 are provided in Table S3. For genes associated with the minor ZGA (cluster 5), 20.0% of our DEGs were upregulated, whereas 5.6% were downregulated. For genes associated with the major ZGA (clusters 1 and 2), 13.2% of our DEGs were upregulated and 3.1% were downregulated. Finally, for the cluster associated with both minor and major ZGA (cluster 4), 23.7% of our DEGs were upregulated and 3.5% were downregulated. Thus, there was unexpected upregulation of over 17% of the ZGA-associated genes when Ube2i was conditionally deleted from oocytes in growing follicles. To further examine the overlap between our DEG list sorted by minor ZGA, minor and major ZGA, and major ZGA clusters, we performed IPA on the genes that were upregulated or downregulated in the Ube2i Zp3-cre+ oocytes (treatment versus control, |log₂FC|≥1 and P_adj<0.05), and upstream transcription factors were identified and clustered with QIAGEN IPA and further annotated with PANTHER GO enrichment analysis (http://geneontology.org/). (Fig. 8D). Of these transcription factors, MYC had the higher number of affected downstream transcriptional target genes, followed by β-catenin (CTNNB1), huntingtin (HTT), TP63 (TRP63) and sex-determining region box 2 (SOX2). (Fig. 8D).

During ZGA, there is a transient upregulation of retrotransposons linked to totipotency (Low et al., 2021). The murine endogenous retrovirus-like gene MuERV-L (Erv4) is one of the earliest transcribed genes in mouse one-cell-stage embryos and a key target of the transcription factor DUX during ZGA (Kigami et al., 2003; De Iaco et al., 2017; Hendrickson et al., 2017). MuERV-L expression was analyzed by RT-qPCR and found to be significantly upregulated in Ube2i Zp3-cre+ oocytes (4014% compared with its expression in control oocytes, P<0.0001, Fig. S9) (Kigami et al., 2003). We additionally determined whether expression of other types of transposable elements (TEs) was altered using |FC|≥2 and false discovery rate (FDR)-adjusted P-value≤0.05. Eleven of 594 TEs were downregulated, whereas 583 TEs were upregulated. All major groups of retrotransposons, including long interspersed nuclear elements (LINEs), short interspersed nuclear elements (SINEs) and long terminal repeats (LTRs), were significantly upregulated (Fig. 9A) in fully grown, Ube2i Zp3cre+ GV oocytes compared with controls. Thus, the increase in the expression of genes associated with ZGA along with activation of TEs, including key retrotransposons, indicate that there are major disruptions in transcriptional regulation in the GV oocytes that arise by the end of folliculogenesis.

Based on the DEGs found to be associated with the minor and major ZGA, as well as the increase in TE activation that have previously been correlated to the two-cell stage (Evsikov et al., 2004), we compared our RNA-seq data from Ube2i Zp3cre+ and control oocytes with the previously published RNA-seq data from wild-type GV oocytes, early two-cell-stage and late two-cell stage (Zhang et al., 2016). Similar to the PCA for control versus Ube2i Zp3cre+ oocytes (Fig. 7A), these samples again clustered with 54% variance in PC1 and 26% variance in PC2 (Fig. 9B). The control samples readily clustered with the published GV dataset (Fig. 9B). In contrast, the Ube2i Zp3cre+ oocytes clustered independently in a range along PC1 between the GV stage and the early two-cell stage to the late two-cell stage, although, along PC2, they were clearly not GV, early two-cell stage or late two-cell stage (Fig. 9B). In total, the increase of genes associated with ZGA along with activation of TEs and the lack of clustering with GV oocytes indicate that by the end of folliculogenesis, there are major disruptions in transcriptional regulation in the fully grown Ube2i Zp3cre+ oocytes and a potential loss of cell identity.

The scope of regulation provided by SUMOylation in mammalian oocyte biology is not well known, particularly during oocyte growth within ovarian follicles. During folliculogenesis, critical hallmarks of oocyte development are established, including acquisition of meiotic and developmental competence. Towards the end of folliculogenesis, there is a major change in chromatin organization (NSN to SN), with the SN configuration associated with global suppression of transcription and increased developmental competence (Mattson and Albertini, 1990; Zuccotti et al., 1998; Bouniol-Baly et al., 1999; De La Fuente and Eppig, 2001; Miyara et al., 2003). Although SUMOylation is known to be a key process for transcriptional repression in general (Ouyang and Gill, 2009; Theurillat et al., 2020), a role for SUMOylation for global transcriptional repression in mammalian oocytes has not been previously described. Using a novel mouse model developed to delete Ube2i selectively in growing oocytes, we demonstrate that UBE2I is necessary for the processes of chromatin remodeling, transcriptional regulation and differentiation into fully grown GV oocytes, in addition to its more well-known roles in meiotic progression (Fig. S10).

The oocyte-specific knockout of Ube2i using Gdf9-icre demonstrated that Ube2i was required for stability of the ovarian reserve (Rodriguez et al., 2019). It was unclear if accelerated depletion of the ovarian reserve was due to direct effects of Ube2i in oocytes of primordial follicles or indirectly due to loss of growing follicles, which produce factors, including anti-Müllerian hormone, that regulate primordial follicle dynamics (Durlinger et al., 1999; Pangas, 2012). To test this, we used the oocyte-deleter strain Zp3-cre, which is temporally distinct from Gdf9-icre (Lan et al., 2004). Even though deletion at the primordial stage (Ube2i Gdf9-icre+) or from primary follicles onward (Ube2i Zp3-cre+) caused female sterility and oocyte arrest at MI, unlike Ube2i Gdf9-icre+ ovaries, Ube2i Zp3-cre+ ovaries showed a stable primordial follicle pool with limited effects on overall follicle populations. Thus, Ube2i has essential roles in oocytes of both primordial follicles and growing follicles. A detailed comparison of phenotypes is shown in Table S4.

Although the number of follicles at most stages of folliculogenesis were unchanged even in 6-month-old Ube2i Gdf9-icre+ mice, there was a 30% decline in primary follicles and alterations in growth of the cumulus cell layer surrounding oocytes, which resulted in a smaller area of cumulus cell expansion following ovulation in Ube2i Gdf9-icre+ mice compared with that in controls. This secondary effect on cumulus cells may reflect reductions in the oocyte-expressed genes Bmp15 and Gdf9. BMP15 and GDF9 function in a complex signaling loop that bidirectionally regulates oocyte and cumulus cell development, which includes maintaining meiotic arrest (Su et al., 2004; Peng et al., 2013; Wigglesworth et al., 2013). Even though the cumulus cell layer was reduced in Ube2i Gdf9-icre+ oocytes, it appeared to be sufficient to maintain bidirectional communication, as oocytes from Ube2i Gdf9-icre+ grew to a similar size as control oocytes and maintained meiotic arrest; additionally, Ube2i Gdf9-icre+ ovaries lacked hallmarks of severe cumulus cell defects that lead to follicle and oocyte loss (Pangas et al., 2006; Ploutarchou et al., 2015). More detailed experiments are required to understand the reduction in primary follicles and changes to cumulus cell growth that can impact oocyte quality.

In wild-type ovaries, by the end of folliculogenesis, a significant proportion of fully grown oocytes are in the SN configuration, with the remainder in NSN (Zuccotti et al., 1995). Although NSN oocytes resume meiosis normally, progress to MII and can be fertilized, they subsequently arrest at the two-cell stage (Zuccotti et al., 1998). The majority of Ube2i Zp3-cre+ GV oocytes fail to transition from the NSN to SN chromatin configuration; however, unlike wild-type NSN oocytes or other mutant mouse lines that retain high levels of NSN oocytes in antral follicles (Burns et al., 2003), Ube2i Zp3-cre+ GV have a twofold reduction in meiotic resumption. This suggests that the defects in Ube2i Zp3-cre+ are not solely due to the presence of the NSN stage. Of the oocyte-expressed genes downregulated in Ube2i Zp3-cre+, several have knockout mouse models that show an increased percentage of GV oocytes in the NSN configuration. These includes germline knockouts of the maternal effect genes Npm2 and Mater, which both show a later arrest (two-cell-stage embryo) compared with that of Ube2i Zp3-cre+ model (Tong et al., 2000; Burns et al., 2003; Monti et al., 2013). The maternal effect gene Zar1, for which the knockout mouse model arrests at the one-cell-stage embryo (Wu et al., 2003), is also downregulated in Ube2i Zp3-cre+ oocytes. During development, Zar1 and Mater transcripts are degraded from GV to MII (Su et al., 2007), with 90% of all maternal-effect gene transcripts in oocytes degraded by ZGA (Schellander et al., 2007; Walser and Lipshitz, 2011). The decreased levels of maternal-effect gene transcripts, in spite of the maintenance of active transcription in Ube2i Zp3-cre+, may additionally suggest that RNA stability or storage of subsets of transcripts could be affected, although this remains to be determined. One hypothesis is this could be due in part to a reduction in the levels of the RNA-binding protein MSY2, which is significantly downregulated approximately twofold, as Msy2^−/− GV oocytes have a 25% reduction in poly(A)-containing RNA (Medvedev et al., 2011). Other RNA-binding proteins (RBPs) are targets of SUMOylation or contain SUMO-interacting motifs (Richard et al., 2017; Ruta et al., 2021); however, the RBP SUMO targets in oocyte biology, which might include MSY2, remain to be characterized.

Although the NSN-to-SN chromatin configuration switch in fully grown oocytes is associated with transcriptional silencing, these processes can be uncoupled (Bouniol-Baly et al., 1999; De La Fuente et al., 2004; Ma et al., 2013) and mechanisms underlying transcriptional silencing are not well understood (Schultz et al., 2018). Transcriptional silencing in wild-type GV oocytes is associated with the accumulation of H3K4me3 and H3K9me3 (Kageyama et al., 2007; Chousal et al., 2018). Ube2i Zp3-cre+ GV oocytes show loss of these repressive marks and a failure to silence transcription, potentially in part because they remain in the NSN configuration, but this could also suggest that SUMOylation plays a role in the establishment of trimethylated H3K4 and H3K9 during oocyte development. Decreased levels of H3K4me3 and H3K9me3 are also associated with expression of ZGA-associated genes such as the Zscan4 gene family (Smith et al., 2022). Active removal of H3K4me3 by the lysine demethylases KDM5A and KDM5B (both of which are significantly upregulated in Ube2i Zp3cre+ oocytes) is required for ZGA, and reduction of these marks results in premature activation of the ZGA in GV oocytes (Dahl et al., 2016). Our informatic analysis of Ube2i Zp3-cre+ GV oocyte DEG revealed the category of ZGA-associated genes as top upstream regulators, including MYC, a proposed master regulatory gene in the two-cell-stage embryo (Zeng and Schultz, 2005). The upregulation of genes typically associated with ZGA suggests that the fully grown Ube2i Zp3cre+ oocytes had not acquired the appropriate differentiation state by the end of folliculogenesis or had an altered cell fate. SUMOylation is known to be involved in ZGA (Higuchi et al., 2019), but the upregulation of genes normally associated with ZGA is pathological in GV oocytes. Genes associated with ZGA in Ube2i Zp3-cre+ GV oocytes (Dux, Zscan4 and Eif1a-like genes) showed a similar upregulation when Ube2i was knocked down in embryonic stem cells, which converted embryonic stem cells to a more pluripotent, two-cell-embryo-like state (Cossec et al., 2018). The increase in genes associated with ZGA, along with activation of retrotransposons, suggests that the epigenetic program of oocytes either was not established during follicle development or, if established, was not maintained. Thus, our data support a hypothesis that SUMOylation in fully grown oocytes is required for the oocyte transcriptional program and SUMOylation may safeguard oocyte ‘fate’ or ‘identity’ until it is necessary to activate the zygotic genome, similar to its role in other cellular contexts (Cossec et al., 2018).

In summary, the phenotype of Ube2i Zp3-cre+ GV oocytes indicates new roles for UBE2I in oocytes during folliculogenesis, which are necessary for fertility. The phenotype includes loss of transcriptional silencing prior to meiotic resumption, improper chromatin remodeling at the end of folliculogenesis, decreased production or regulation of maternal-effect gene transcripts and loss of suppression of transcripts associated with totipotency (i.e. the two-cell stage). In terms of genetic regulation, the increase in gene expression associated with ZGA and upregulation of retrotransposons suggests that the epigenetic program of Ube2i Zp3-cre+oocytes is likely altered. Understanding these genetic changes in combination with the improper chromatin state will provide insight into how the oocyte maintains its own genetic program while preparing for early events in embryogenesis (Hanna et al., 2018).

Experimental procedures utilizing mice were performed in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and Institutional Animal Care and Use Committee-approved animal protocols at Baylor College of Medicine (AN-4762). The Ube2i^loxP/loxP mice used in this study were previously generated and provided by Sean Hartig (Baylor College of Medicine, Houston, TX, USA) (Cox et al., 2021) and backcrossed to the F1 hybrid C57BL/6J/129S7/SvEvBrd background of our previous study (Rodriguez et al., 2019). Recombination by Cre generates a null allele through deletion of exons 3 and 4 as described (Cox et al., 2021). Ube2i^loxP/loxP females were mated to Zp3-cre [strain Tg(Zp3-cre)^93Knw/J; The Jackson Laboratory, strain 003651] males to generate Ube2i^loxP/+ Zp3-cre+ males, which were then crossed to Ube2i^loxP/loxP females to generate Ube2i^loxP/loxP Zp3-cre+ females, called Ube2i Zp3-cre+ in text. All cre lines were maintained on the male line. Generation and oocyte-restricted expression of Zp3-cre mice as well as genotyping has been previously characterized (de Vries et al., 2000; Lan et al., 2004). Genomic DNA from ear punches was used for genotyping. Cre-negative female littermates (Ube2i^loxP/loxP) were used as controls in all experiments to minimize differences in genetic backgrounds. Ube2i^loxP/loxP were also crossed to Gdf9-icre, and the sterility and ovarian follicle loss phenotype validated against the prior study (Rodriguez et al., 2019). Genotyping primers are listed in Table S5.

To test fecundity, sexually mature (6-week-old) Ube2i Zp3-cre+ females or control littermates were pair-housed with sexually mature (8-week-old) wild-type males of known fertility and continuously mated for 6 months (n=3 mating pairs per genotype). Females were monitored daily for new litters and the number of pups, date of birth and sex was recorded, similar to prior publications (Pangas et al., 2008; Rodriguez et al., 2019). Any resulting litters were weaned at 3 weeks of age.

Mice were anesthetized using isoflurane (Abbott Laboratories) inhalation and euthanized by cervical dislocation. Ovaries were collected and fixed in 10% neutral buffered formalin (Electron Microscopy Sciences) overnight, followed by standard paraffin processing and embedding at the Human Tissue Acquisition and Pathology Core at Baylor College of Medicine. Thirteen control ovaries (from 13 mice) and 12 Ube2i Zp3-cre+ ovaries (from 12 mice) were serially sectioned at 5 µm using an Ergostar HM200 microtome (Microm) and stained with periodic acid (Sigma-Aldrich) and Schiff's reagent (Sigma-Aldrich) (PAS) for analysis. Brightfield images were obtained using Axioscope 2 plus and Zeiss Zen Software v2.3 and Zen Blue. For morphometric quantification, ovaries were sectioned at 5 µm and the follicles in every fifth section were manually counted (Tilly, 2003). To avoid overcounting, only follicles with a visible oocyte nucleus were counted. For follicles smaller than the antral stage, a correction factor was applied by multiplying by 5 to estimate the number of total follicles (Hirshfield and Midgley, 1978; Tilly, 2003). Corpora lutea and antral follicles were counted in total by following each individually through serial images. Follicle classification was based on previous studies into primordial (type 2), primary (types 3a and 3b), secondary (type 4), preantral (types 5a and 5b) and antral (types 6-8) (Pedersen and Peters, 1968).

Blood was collected from isoflurane-anesthetized group-housed, 6-month-old mice by cardiac puncture, and the serum was separated by centrifugation in microtainer collection tubes (Becton, Dickinson, and Company) and frozen at −20°C until assayed. Mouse FSH and estradiol levels were analyzed by ELISA at the University of Virginia Ligand Core Facility [Specialized Cooperative Centers Program in Reproductive Research, National Institute of Child Health and Human Development (NICHD)/National Institutes of Health (NIH) U54-HD28934]. The assay method information is available online (https://med.virginia.edu/research-in-reproduction/ligand-assay-analysis-core/assay-methods/). Data were log transformed before statistical analysis by Welch's two-tailed unpaired t-test (FSH) or two-tailed unpaired t-test (FSH).

For analysis of cumulus-oocyte complexes or collection of oocytes following superovulation, female mice aged 5-6 weeks were injected with 5 IU pregnant mare's serum gonadotropin (PMSG; ProSpecBio), and 46 h later injected with 5 IU human chorionic gonadotropin (hCG; Pregnyl; Merck Pharmaceuticals). Mice were euthanized 16-18 h after injection with hCG and cumulus-oocyte complexes were harvested from the oviduct ampulla and imaged live, or incubated with 0.3 mg/ml hyaluronidase (Sigma-Aldrich) to detach cumulus cells, then the oocytes were fixed in 3.7% paraformaldehyde (PFA; Electron Microscopy Sciences) in 0.1% BSA (Sigma Aldrich) in PBS (BSA-PBS) for 15-20 min, then rinsed through five drops of BSA-PBS to prevent overfixation. Oocytes were stored in BSA-PBS at 4°C overnight, or directly used for immunostaining.

For collection of fully-grown, prophase I-arrested GV oocytes, 5- to 6-week-old female mice were injected with 5 IU PMSG and ovaries were excised after 44-46 h into M2 medium (Millipore Corporation) supplemented with 2.5 µM milrinone (Sigma Aldrich) to maintain prophase I meiotic arrest. Follicles were ruptured via puncturing with a 26G needle to release oocytes into the medium. Oocytes were denuded of cumulus cells via continuous pipetting (De La Fuente et al., 2004) and separated from cumulus cells by passing the medium through a 40 µm mesh strainer (oocytes were retained, whereas cumulus cells passed through the mesh strainer). Any remaining cumulus cells were mechanically stripped by gently pipetting the oocyte through a clinical IVF ‘stripper pipettor’ (Cooper Surgical) using a plastic pipette tip slightly smaller than the diameter of the oocyte. After denuding, oocytes were rinsed through ten drops of EmbryoMax CZB medium with Phenol Red (Millipore Corporation) supplemented with 2 mM L-glutamine (Thermo Fisher Scientific) to wash out milrinone and cultured in a fresh drop of medium covered with mineral oil (Sigma-Aldrich) at 37°C and 5% CO₂. Oocytes that had not undergone GVBD were removed from the culture after 2.5 h, and oocytes were cultured for 5 h to pro-metaphase I or 7.5-8 h to MI or for monitoring meiotic progression. After culturing to the desired stage, oocytes were fixed in 3.7% PFA in BSA-PBS for 15-20 min, then rinsed through five drops of BSA-PBS to prevent overfixation. Oocytes were stored in BSA-PBS at 4°C overnight, or directly used for immunostaining.

Chromosomes spreads were performed similar to prior studies (El Yakoubi et al., 2017; Nguyen et al., 2018). Oocytes were collected and cultured to pro-metaphase (5 h) or MI (8.5 h), followed by the removal of the zona pellucida (ZP) by transferring the oocytes through up to ten drops of acid Tyrode's solution (Sigma-Aldrich) under a brightfield microscope until the ZP was visibly removed. After removal of the ZP, oocytes were rinsed through ten drops of M2 medium and returned to the incubator to recover for 30 min. Then, 50 µl of chromosome-spreading solution [0.64% PFA, 3 mM dithiothreitol (Thermo Fisher Scientific), 0.16% Triton X-100 (Thermo Fisher Scientific)] was applied to the wells of PTFE printed slides (VWR), and three to five oocytes were transferred to the well. Lysing of the oocytes was confirmed by visual monitoring using a brightfield microscope. The slides were allowed to dry at room temperature and stored at −20°C until use.

Immunofluorescence was performed essentially as described previously (Rodriguez et al., 2019). After fixation, oocytes were permeabilized in 0.1% Triton X-100 in BSA-PBS for 20 min, then rinsed through five drops of BSA-PBS. Blocking was performed in 1.0% BSA and 0.01% Tween-20 for 45 min to 1 h at room temperature, or overnight at 4°C. Cells were incubated in primary antibody diluted in blocking buffer for 2 h, or overnight at 4°C, followed by three 10-min washes in BSA-PBS. After washing, oocytes were incubated in the secondary antibodies diluted in blocking buffer for 1 h, followed by three 10-min washes in BSA-PBS. DNA was stained using 20 µM Hoechst (Thermo Fisher Scientific) in BSA-PBS for 15 min, and briefly rinsed in BSA-PBS. Oocytes were mounted on Superfrost Plus glass slides (VWR) under no. 1.5 thickness coverglass (VWR) using ProLong Diamond Antifade Mount (Thermo Fisher Scientific) and allowed to dry overnight at room temperature. Slides were stored at 4°C until imaging. Information on antibodies used in this study can be found in Table S6.

Chromosome spreads were allowed to warm to room temperature, then rinsed three times in PBS. Blocking was performed with 3.0% BSA in PBS for 45 min. Primary antibodies were diluted in blocking buffer and incubated on the chromosome spreads for 2 h, followed by three 15-min washes with PBS. Secondary antibodies were diluted in blocking buffer and incubated on the spreads for 1 h, followed by PBS washes. DNA was then stained using 20 µM Hoechst for 15 min, followed by a PBS rinse. Negative controls were obtained by omitting the primary antibody, instead incubating in blocking buffer alone, before incubating in the secondary antibody diluted in blocking buffer. ProLong Diamond was applied to the samples, and a no. 1.5 coverslip was applied and allowed to dry at room temperature. Slides were stored at 4°C until imaged.

Immunofluorescence images were obtained at the Optical Imaging and Vital Microscopy Core at Baylor College of Medicine on a Zeiss LSM 780 or 880 confocal microscope using the 40× oil objective. Z-stacks were captured using optimal step sizes from the bottom of the region of interest to the top of the region. To acquire images that were used for fluorescence quantification, the parameters were held consistent across all samples. Z-stacks were exported from Zeiss Zen 3.2 Blue edition and the signal intensity was measured using NIH ImageJ software. Negative controls were used to determine background fluorescence signal intensities, which were subsequently subtracted from the fluorescence intensities of the experimental oocytes.

Chromosome alignment was analyzed using oocytes cultured for 8.5 h and fixed as described above. Oocytes were then stained for α-tubulin and DNA, then z-stacks of the chromosomes were collected for analysis. MI chromosome alignment was performed in Zeiss Zen 3.2 Blue edition. Oocytes with all chromosomes within 5 µm of the metaphase plate were classified as ‘good’, whereas oocytes with one or more chromosomes more than 5 µm away from the metaphase plate were classified as ‘poor’. To analyze SN and NSN chromosome configuration of GV oocytes, oocytes were collected, denuded and fixed, and DNA was stained with Hoechst. Oocytes with a solid ring of DNA around the nucleolus were classified as SN, whereas oocytes lacking a ring of DNA around the nucleolus or containing only a partial ring were classified as NSN. To ensure that immature oocytes were not used, oocyte diameter was measured across the midpoint of the oocyte, not including the ZP.

GV oocytes were collected as above and cultured in the presence of 2 mM 5-ethynyl uridine (5EU) (Click-IT RNA Alexa Fluor 488 Imaging Kit, Invitrogen) in EmbryoMax CZB medium supplemented with L-glutamine according to the manufacturer's recommended procedure. In brief, oocytes were cultured in 5EU for 45 min, then immediately fixed in 3.7% PFA for 20 min and rinsed through five drops of BSA-PBS. The oocytes were permeabilized with 0.1% Triton X-100 in BSA-PBS for 15 min, followed by BSA-PBS rinse. The Click-IT labeling reaction was performed by incubating the oocytes in the provided reaction cocktail according to the manufacturer's protocol. Oocytes were then counterstained with 20 µM Hoechst for 15 min, rinsed in BSA-PBS, mounted in ProLong Diamond and allowed to dry overnight. Z-stacks of each oocyte were obtained for analysis. Oocytes containing greater fluorescence signals than the signals in negative-control oocytes were considered to be transcriptionally active. To ensure that immature oocytes were not used, oocyte diameter was measured across the midpoint of the oocyte, not including the ZP.

GV oocytes were collected and denuded, then cultured for 7 h in EmbryoMax CZB medium supplemented with L-glutamine. Oocytes were then placed in ice-cold M2 medium (Invitrogen) for 10 min, fixed in 3.7% PFA for 20 min, then used in immunostaining (as above) for α-tubulin and kinetochores (anti-CREST).

Denuded, fully grown GV oocytes were collected from PMSG-stimulated control and Ube2i Zp3cre+ mice as independent pools as described above. Total RNA was generated using the PicoPure Isolation Kit (Applied Biosystems) and treated with DNase (QIAGEN) prior to the generation of cDNA. The average amount of RNA obtained per oocyte was 0.081 ng. 50 ng of RNA was used to generate cDNA using the High-Capacity RNA-to-cDNA reverse transcription kit (Invitrogen). cDNA was diluted 1:2 in water and 2 μl used for RT-qPCR. RT-qPCR was performed using the PowerUp SYBR Green Master Mix (Applied Biosystems) on an Applied Biosystems StepOnePlus real-time PCR system, with samples performed in duplicate. A melt-curve analysis was performed to verify that the primers amplified a single peak. Amplification of gene expression from RT-qPCR data was analyzed using the ΔΔCt method (Livak and Schmittgen, 2001) and normalized to ubiquitin C (Ubc) expression for each sample. In all graphs, the data are shown relative to the mean of the control value. The primers used for RT-qPCR are listed in Table S5.

The RNA-seq library was generated according to SMART-seq v4 Ultra low input RNA kit (Takara). In brief, the largest fully grown GV oocytes (n=17 oocytes per genotype) were collected from five control and five Ube2i Zp3-cre+ PMSG-primed females and were lysed in lysis buffer (Takara Bio), 3-SMART-seq CDS primer II  (Takara Bio) and V4 oligonucleotide  (Takara Bio) were added for first-stranded cDNA synthesis and stored at −80°C until cDNA was generated. cDNA was amplified using PCR Primer II A (Takara Bio) and subsequently purified using Ampure XP beads (Beckman). The Illumina library was prepared using Nextera XT DNA library preparation kit (Illumina) and sequenced using an Illumina Novaseq. Raw counts obtained by mRNA sequencing were Rfiltered (genes with counts≥34), normalized and differentially analyzed (|FC|≥ 2 and FDR-adjusted P-value≤0.05) using DEq2 v1.36.0 (Love et al., 2014). Reads were aligned to the mouse genome version mm39. PCA plot and volcano plots were generated using ggplot2 v3.3.5 (Wickham, 2016) and ggrepel v0.9.1 (https://ggrepel.slowkow.com). Ingenuity Pathway Analysis v.90348151 (Qiagen; https://digitalinsights.qiagen.com/IPA) was used to identify likely upstream transcriptional regulators that were linked to the DEG datasets based on changes in expression. We used Benjamini-Hochberg adjusted P-value<0.01 as the significance threshold. For analysis of TEs, TEs were quantified and annotated using TEtranscripts v2.2.3 (Jin et al., 2015). TE counts were filtered, normalized and processed using DESeq2 v1.36.0 (Love et al., 2014) in a similar way as RNA-seq analysis. Bar plots of the proportion of TE reads were generated using GraphPad Prism 9. Only TE classes that contributed a minimum of 1% of the total reads were presented. For analysis of ZGA genes, genes belonging to different ZGA categories (major, minor, and major and minor) were annotated using datasets identified in Abe et al. (2018). Bar plots were generated using ggplot2 v3.3.5 (Wickham, 2016). Upstream regulator analysis of significant differentially expressed ZGA genes was carried out with Ingenuity Pathway Analysis and clusters further annotated using PANTHER GO enrichment analysis (http://geneontology.org/).

All experiments were performed with a minimum of three independent replicates, with sample sizes indicated in the text or figure legends. Statistical analysis was performed using GraphPad Prism 9. Fertility data, immunofluorescence quantification and RT-qPCR analyses are displayed as the mean and standard error of the mean (s.e.m.). Unpaired Student's t-test was used to determine the difference between two groups. ELISA data were log-transformed prior to statistical analysis using unpaired Student's t-test or Welch's t-test as indicated. All t-tests were two-tailed. Chromosome alignment, chromosome configuration and transcriptional activity were analyzed using Fisher's exact test on raw data. P<0.05 was considered to be statistically significant.

The authors would like to thank the directors and staff at the following advanced technology core facilities at Baylor College of Medicine: Human Tissue Acquisition and Pathology Core, Optical Imaging and Vital Microscopy Core, and Dr Rui Chen and the Single Cell Genomics Core. Core facilities at Baylor College of Medicine are supported by a CPRIT Core Facility Support Award (CPRIT-RP180672) and the NIH (P30 CA125123). We additionally thank Dr Diana Monsivais and Bethany Patton (Baylor College of Medicine) for helpful discussions and comments, and Dr Cecilia Blengini (Rutgers University) for assistance with chromosome spreads. We thank additional members of the Pangas, Monsivais and Hartig laboratories for helpful discussions.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201535.reviewer-comments.pdf