Maternal DCAF2 is crucial for maintenance of genome stability during the first cell cycle in mice Free

In mammals, female germ cells exit mitosis after their last round of DNA replication, enter meiosis on embryonic days 11.5–12.5, and are arrested at the diplotene stage of prophase I for an extended period (Smallwood and Kelsey, 2012). Within 3 days of birth, these oocytes are enclosed by ovarian somatic cells for formation of primordial follicles (Zhang and Liu, 2015). DNA replication is halted in these follicle-enclosed oocytes for months in experimental rodents or up to 50 years in humans.

Shortly after fertilization and pronucleus formation, the first round of DNA replication of the embryo is initiated with several distinct features: (1) the DNA replication machinery is assembled by maternal factors stored in the oocyte; (2) DNA replication is initiated and progresses separately in haploid male and female nuclei, respectively; (3) the DNA replication in the zygote is accompanied by drastic genome remodeling, including massive DNA demethylation, histone modifications and histone replacement (Akiyama et al., 2011; Murai et al., 2014; Swiech et al., 2007; Xu et al., 2015). It has been proposed that genome remodeling after fertilization, particularly in the male nucleus, induces DNA damage including both single- and double-strand breaks (Hajkova et al., 2010; Nakatani et al., 2015). The damaged DNA is repaired in the first round of replication (Ma et al., 2016; Wang et al., 2013). It is unclear whether repair of the genome remodeling-induced DNA damage is a prerequisite for subsequent cleavages, and if so, which maternal factors are necessary for repairing the DNA damage and for maintaining zygotic genome integrity during oocyte-to-embryo transition.

Our studies have shown that cullin ring-finger ubiquitin ligase 4 (CRL4) has multiple functions in the maintenance of oocyte survival, meiotic cell cycle progression, and genome reprogramming after fertilization (Yu et al., 2015a,b, 2013). We and others have also reported that the CRL4 components cullin 4A and DDB1 are essential for spermatogenesis in mice and Drosophila (Kopanja et al., 2011; Yu et al., 2016b). CRL4 plays versatile biological roles by employing >90 WD40 repeat-containing adaptor proteins, known as DDB1–cullin-4-associated factors (DCAFs) (Lee and Zhou, 2007). These DCAFs recruit a wide spectrum of substrates to CRL4 for poly- or monoubiquitylation (Bennett et al., 2010). DCAF2, also known as DTL or CDT2, is one of the best-studied substrate receptors of CRL4 (Abbas and Dutta, 2011). For ubiquitin-mediated degradation, DCAF2 targets a number of substrates, such as CDT1, p21 (also known as CDKN1A) and CHK1 (also known as CHEK1), involved in regulation of the cell cycle and survival (Havens and Walter, 2011).

DCAF2 overexpression has been reported in breast (Ueki et al., 2008) and ovarian (Pan et al., 2013) carcinomas and rhabdomyosarcomas (Missiaglia et al., 2009), and contributes to chemotherapy resistance of ovarian cancer cells (Pan et al., 2013). By contrast, the in vivo function of DCAF2 in development has been poorly studied. Targeted disruption of the Dcaf2 gene in mice causes early embryonic mortality (Liu et al., 2007). The zygotic Dcaf2 knockout embryos (Dcaf2^–/–) terminate development at the four- to eight-cell stage (Liu et al., 2007). This situation has prevented research into the role of DCAF2 in later developmental processes including gametogenesis. Furthermore, the maternally contributed DCAF2 might compensate for the loss of zygotic DCAF2 in the first two embryonic cell cycles.

To specifically determine whether DCAF2 is one of the key CRL4 substrate adaptors in oogenesis, we generated a novel mouse strain that carries a Dcaf2 allele flanked by loxP sequences, and deleted Dcaf2 in oocytes using cell type-specifically expressed Cre DNA recombinase. The results showed that the maternal DCAF2 protein is crucial for prevention of DNA re-replication in the first and unique mitotic cell cycle of the zygote. We also report that DCAF2 plays a previously unrecognized role in the regulation of DNA demethylation during oocyte-to-embryo transition in mice.

We first examined the expression and distribution of DCAF2 in oo­cytes. By conducting quantitative real-time PCR, we found that Dcaf2 mRNA is expressed much more strongly in fully grown oocytes and zygotes than in representative somatic tissues (Fig. 1A). Dcaf2 expression further increased in two- to eight-cell embryos after major zygotic genome activation, and then decreased in morulae and blastocysts (Fig. 1B). Unfortunately, the available commercial anti-DCAF2 antibodies failed to detect endogenous DCAF2 in mouse oocytes and zygotes. Therefore, we expressed an mCherry-tagged DCAF2 in oocytes and zygotes by mRNA microinjection followed by fluorescence staining to study its localization. The DCAF2 protein was found to be evenly distributed in germinal vesicle (GV)-stage oocytes, but showed larger amounts in pronuclei than in the zygotic cytoplasm after fertilization (Fig. 1C). The nuclear localization of DCAF2 became more prominent in two- to four-cell embryos (Fig. 1C). Collectively, these results suggested that DCAF2 is highly expressed in mouse oocytes and zygotes, and might be intimately involved in the first mitotic cell cycle progression.

To determine the in vivo function of maternal Dcaf2, we generated a mouse strain that carries a Dcaf2 allele flanked by loxP sequences (Dcaf2^fl/fl) (Fig. 1D) and deleted maternal Dcaf2 using oocyte-specifically expressed Gdf9-Cre DNA recombinase (Dcaf2^fl/fl;Gdf9-Cre). The Dcaf2 deletion was confirmed by quantitative RT-PCR in oocytes isolated from the ovaries of Dcaf2^fl/fl and Dcaf2^fl/fl;Gdf9-Cre females (Fig. 1E). The females were completely infertile when mated with wild-type (WT) males. To identify the reason for the infertility of Dcaf2^fl/fl;Gdf9-Cre females, we collected embryos from the oviducts of control and mutant females 48 h and 56 h after human chorionic gonadotropin (hCG) injection. While embryos from control females developed to two-cell (hCG 48 h) and four-cell (hCG 56 h) stages, embryos from Dcaf2^fl/fl;Gdf9-Cre females (later termed Dcaf2^♀−/♂+) were all arrested at the zygote stage and contained normal male and female pronuclei (Fig. 1F). These results indicated that maternal Dcaf2 is required for completion of the first mitotic division of the zygote.

Because DCAF2 deletion in somatic cells causes DNA damage and apoptosis (Kim et al., 2008; Pan et al., 2013), we investigated the potential involvement of DCAF2 in the maintenance of oocyte survival. Ovaries of adult (4 to 6 months old) Dcaf2^fl/fl;Gdf9-Cre mice are normal histologically (Fig. S1A). Immunohistochemical staining for the oocyte marker mouse vasa homolog (MVH) revealed that oocytes were present in both primordial follicles and in growing follicles (Fig. S1B). After superovulation, Dcaf2^fl/fl;Gdf9-Cre females ovulated metaphase II (MII) oocytes in numbers similar to those of the control mice (Fig. S1C,D). These ovulated Dcaf2-null oocytes contained normal spindles (Fig. S1D). When the GV stage-arrested fully grown Dcaf2-null oocytes were isolated and cultured in vitro, they had normal germinal vesicle breakdown (GVBD) and polar body-1 (PB1) emission rates, just as the control oocytes did (Fig. S1E). These results suggested that Dcaf2 is dispensable for oocyte survival from the primordial-follicle stage to ovulation; the loss of DCAF2 did not affect oocyte meiotic maturation.

In a rescue experiment, when the Dcaf2^{♀­­−/♂+} zygotes were microinjected with exogenous Dcaf2 mRNAs (mCherry-Dcaf2) and cultured in vitro, they could develop into blastocysts with significantly higher efficiency than the control zygotes (Dcaf2^{♀­­−/♂+} zygotes microinjected with mRNAs encoding for mCherry) (Fig. 1G,H). This result further indicated that maternal DCAF2 is specifically needed in zygotes but is dispensable for oocyte development and survival.

We then investigated the reason for the developmental arrest of Dcaf2^♀−/♂+ zygotes. DNA damage labeled with S139-phosphorylated histone H2AX (pH2AX-S139, also known as γH2AX) was detected in WT zygotes at the pronucleus (PN)-4 stage (Fig. 2A) but was repaired in two-cell embryos (Fig. 2B). By contrast, the Dcaf2^♀−/♂+ zygotes contained stronger pH2AX signals than the WT zygotes did at PN5 (Fig. 2A), and the signals increased further when the control embryos developed to two-cell stage (Fig. 2B). The pH2AX foci in male/female pronuclei were counted, and the numbers are shown in Fig. 2C. Occasionally, a few Dcaf2^{♀­­−/♂+} zygotes divided once but were arrested at the two-cell stage. These embryos contained abundant pH2AX signals in nuclei, while the control two-cell embryos were negative for pH2AX (Fig. 2B). In addition, RPA2, a marker of single-strand DNA breaks, was detected in Dcaf2^{♀­­−/♂+} zygotes but not in WT zygotes (Fig. 2D). When we overexpressed exogenous DCAF2 in Dcaf2^{♀­­−/♂+} zygotes, the formation of pH2AX foci was significantly repressed, indicating that the accumulation of DNA damage is a direct effect of maternal DCAF2 depletion (Fig. S2).

It has been reported that mammalian DCAF2 prevents DNA re-replication in cultured cancer cell lines (Pan et al., 2013). Therefore, we assessed the possible role of maternal DCAF2 in the regulation of the first round of DNA replication in zygotes. BrdU incorporation data indicated that DNA replication in WT zygotes was initiated at the PN3 stage and finished at the early PN5 stage (6 h and 12 h after in vitro fertilization, respectively; Fig. 3A). Nevertheless, DNA replication was initiated early in Dcaf2^{♀­­−/♂+} zygotes at PN2 stage and did not stop at the late PN5 stage. Even when the control zygotes developed to the two-cell stage and started the second round of DNA replication, BrdU incorporation was still detectable in developmentally arrested Dcaf2^{♀­­−/♂+} zygotes (Fig. 3A,B).

The DNA replication licensing factor CDT1 has been reported to be a target of CRL4^DCAF2 in the maintenance of genome integrity of proliferating somatic cells (Shibata et al., 2011). CDT1 is degraded by CRL4^DCAF2 after S phase entry to ensure that DNA replicates only once per cell cycle (Kim et al., 2008). We tested whether CDT1 might be involved in maternal Dcaf2 knockout-induced zygotic genome instability. Unfortunately, the available commercial anti-CDT1 antibodies failed to detect endogenous CDT1 in the mouse zygotes. Therefore, we expressed GFP-CDT1 in oocytes and zygotes by mRNA microinjection, to study its localization and cellular effects. Overexpressed CDT1 was localized in GVs of fully grown oocytes but did not affect meiotic maturation (Fig. S3). We also expressed GFP-CDT1 in WT and Dcaf2^{♀­­−/♂+} zygotes by mRNA microinjection. The exogenous GFP-CDT1 was strongly expressed in zygotes of both phenotypes at 7 h after microinjection, and located in both male and female pronuclei (Fig. 4A). GFP-CDT1 proteins were unstable and were remarkably degraded in WT zygotes at 21–29 h after microinjection. By contrast, the exogenous GFP-CDT1 remained stable in Dcaf2^{♀­­−/♂+} zygotes at these time points (Fig. 4A). This result provided direct evidence that loss of DCAF2 led to the persistence of CDT1 in zygotes.

Significantly, overexpression of CDT1 in zygotes induced massive DNA damage, as revealed by pH2AX and RPA2 immunofluorescence assays (Fig. 4B,C), and caused developmental arrest at one- to two-cell stages (Fig. 4D,E). Thus, CDT1 overexpression in oocytes and zygotes mimicked the phenotypes of maternal Dcaf2 deletion. This observation suggested that DCAF2 maintains zygotic genome stability by downregulating CDT1 in the first mitotic cell cycle.

Because DNA replication, DNA damage and DNA demethylation are interacting events in zygotes (Hajkova et al., 2010), we evaluated the potential effects of Dcaf2 deletion on epigenetic modifications in oocytes and zygotes. The oocytes of Dcaf2^fl/fl;Gdf9-Cre mice had similar numbers of DNA 5-methylcytosine (5mC)-to-5-hydroxymethylcytosine (5hmC) modifications (Fig. S4A). In WT zygotes, 5hmC levels were higher in male pronuclei than in female pronuclei, while 5mC levels showed the opposite trend. Nonetheless, female pronuclear 5hmC signals were remarkably increased in Dcaf2^{♀­­−/♂+} zygotes (Fig. 5A).

Our results suggested that formation of both DNA double-strand breaks and 5hmC in WT zygotes was more active in the male pronucleus than in the female pronucleus, and that both signals were upregulated in the female pronucleus of Dcaf2^{♀­­−/♂+} zygotes, suggesting that DNA damage and DNA demethylation may correlate. To confirm this notion, we examined pH2AX amounts in zygotes derived from Ddb1^fl/fl;Gdf9-Cre mice, in which the DNA demethylase methylcytosine dioxygenase (TET3) is inactive and 5hmC fails to accumulate in the male pronucleus (Yu et al., 2013). No pH2AX foci were detected in these zygotes at PN3–5 stages (Fig. 5B). This result indicated that drastic DNA modifications induce DNA damage in the zygotic genome, and maternally expressed DCAF2 is required for repair of this DNA damage to ensure normal development (Fig. 5C).

On the other hand, zygotic histone exchange and histone methylation were not affected by the maternal Dcaf2 knockout. After fertilization, de novo synthesized histone H3.3 could load onto chromatin of both male and female pronuclei (Fig. S4B), and histone methylation markers H3K4me3 and H3K9me3 showed a normal distribution pattern (more abundant in the female than male pronucleus) in Dcaf2-deleted zygotes (Fig. S4C).

The data above showed that both DNA over-replication and genome reprogramming contributed to the accumulation of DNA damage in Dcaf2^{♀­­–/♂+} zygotes. To determine whether suppression of DNA replication can protect Dcaf2^{♀­­–/♂+} zygotes from DNA damage, we treated zygotes with aphidicolin (5 μM), a specific inhibitor of DNA polymerase α. Aphidicolin efficiently blocked DNA replication because bromodeoxyuridine (BrdU) incorporation was abrogated in the aphidicolin-treated zygotes (Fig. 6A). Nonetheless, aphidicolin treatment did not reduce DNA damage signals in Dcaf2^{♀­­–/♂+} zygotes. Instead, DNA damage labeled with pH2AX increased in both WT and Dcaf2^{♀­­–/♂+} zygotes after aphidicolin treatment (Fig. 6B). In addition, immunofluorescence data revealed that RPA2, a marker of DNA single-strand breaks, was undetectable in normal WT zygotes but was significantly upregulated in aphidicolin-treated zygotes (Fig. 6C). Therefore, inhibiting DNA polymerase activity caused DNA replication stress and massive DNA damage instead of preventing it. In comparison with WT zygotes, Dcaf2^{♀­­–/♂+} zygotes are more susceptible to aphidicolin-induced DNA damage (Fig. 6B); this finding implied that they are more fragile and have an insufficient DNA repair ability in response to environmental stressors.

Moreover, the DNA 5hmC amount also increased in zygotes incubated with aphidicolin (Fig. 5D). Under DNA replication stress, the asymmetrical 5hmC distribution between male and female pronuclei disappeared, and the embryos were arrested at the one-cell stage (Fig. 6D). This phenotype is similar to the one we observed in Dcaf2^{♀­­–/♂+} zygotes. The ratio of female to male pronuclear 5hmC intensities was ∼0.4 in control zygotes but increased to ∼1.0 in aphidicolin-treated zygotes (Fig. 6E). The DNA demethylase TET3 that mediates 5hmC formation in zygotes was preferentially localized to male pronuclei of control zygotes (Gu et al., 2011), but was evenly distributed in male and female pronuclei of aphidicolin-treated zygotes, as shown by immunofluorescence using a TET3 antibody kindly provided by Dr Guoliang Xu (State Key Laboratory of Molecular Biology, Shanghai, China) (Fig. 6F). Therefore, DCAF2 deletion-induced or DNA polymerase inhibition-induced DNA replication stress impaired the asymmetrical TET3 localization in pronuclei and led to increased 5hmC generation in the female pronucleus.

Although Dcaf2 is not required for oocyte survival and meiotic maturation in a normal physiological environment, Dcaf2-null oocytes might be vulnerable to DNA-damaging stressors. To test this hypothesis, we used the anticancer drug etoposide, also known as VP-16, to induce DNA double-strand breaks in WT and Dcaf2-null GV oocytes. The oocytes were treated with 5 μM VP-16 for 3 h and then cultured in a VP-16-free medium (containing 2 μM milrinone to inhibit spontaneous GVBD) for self-repair overnight (Fig. 7A). VP-16 induced massive DNA damage, which was repaired fully in WT oocytes but not in Dcaf2-null oocytes after overnight culture (Fig. 7B,C). These results indicated that DCAF2 was recruited to repair DNA damage in GV oocytes, but its specific biochemical function in this process has yet to be determined.

The gene encoding DCAF2 was first discovered in yeast in 2003, and was later shown to be evolutionarily conserved and to perform key functions in the maintenance of genome stability (Jascur et al., 2011; Kim et al., 2008; Yoshida et al., 2003). On the other hand, its participation in postnatal development and hemostasis has never been studied in vivo, owing to the early mortality of Dcaf2 knockout embryos at the ∼eight-cell stage (Liu et al., 2007). After generating a novel Dcaf2-floxed mouse strain, we studied the function of Dcaf2 in germ cells for the first time.

Although DCAF2 appears to be necessary for survival and proliferation of cultured somatic cells, it is dispensable for oocyte growth and meiotic maturation (Coleman et al., 2015). This phenomenon is likely to be caused by the dormancy of DNA replication in oocytes. DCAF2 is mainly a nuclear protein in somatic cells but was found to be evenly distributed in GV oocytes, further suggesting that it is functionally inactive at this developmental stage (Pan et al., 2013). This novel finding suggests that DCAF2 is required only for maintenance of genome integrity during the S phase of the cell cycle. Currently, we do not know whether oocytes are the only nonproliferating cell type tolerant of Dcaf2 deletion. It will be interesting to determine whether DCAF2 is also dispensable in other terminally differentiated cell types arrested in the G0 phase, such as neurons, hepatocytes and adipocytes, by means of a tissue-specific knockout using the Dcaf2-floxed mice reported in this study.

Our study also identified DCAF2 as a previously unrecognized maternal factor that safeguards zygotic genome stability. Accompanying the initiation of the first round of zygotic DNA replication, maternal DCAF2 molecules translocate from the cytoplasm into pronuclei, indicating their functional activation in the S phase. Maternal Dcaf2-deleted embryos were arrested at one- to two-cell stages owing to prolonged DNA replication and accumulation of massive DNA damage. Biochemically, DCAF2 could play two roles in the maintenance of zygotic genome integrity. First, drastic genome reprogramming in zygotes, particularly in the male pronucleus, causes DNA double-strand breaks; DCAF2 may be directly involved in the timely repair of these DNA lesions to ensure embryonic cell cycle progression, as has been observed in cultured somatic cells (Blaikley et al., 2014; Han et al., 2015). Second, DCAF2 functions as a substrate adaptor mediating CRL4-dependent ubiquitylation and degradation of the DNA replication licensing factor CDT1, thereby preventing the DNA damage caused by re-replication (Higa et al., 2006; Jin et al., 2006). CDT1 overexpression in WT zygotes mimicked the phenotype of Dcaf2^{♀­­–/♂+} zygotes, indicating that maintaining appropriate CDT1 amounts is indeed crucial for zygotic cell cycle progression.

As we already pointed out in the Results section, most commercially available antibodies against DCAF2 and CDT1 were generated and selected to target human antigens, and therefore all failed to detect the corresponding antigens in mouse oocytes and zygotes. Owing to this technical obstacle, we could not describe fluctuations in the amount and localization changes of endogenous DCAF2 and CDT1 before and after fertilization. As a partial compensation, our results using exogenously expressed DCAF2 and CDT1 demonstrated that both proteins are stringently regulated during maternal-zygotic transition and are necessary for completion of the first mitotic cell cycle.

In addition to DNA replication and damage repair, we also studied the effects of maternal Dcaf2 deletion on DNA and histone methylation during maternal-zygotic transition. The results indicated that although histone loading and methylation were not obviously affected in Dcaf2^{♀­­–/♂+} zygotes, the asymmetrical pattern of DNA demethylation in pronuclei was disturbed (Nakamura et al., 2012). It has been demonstrated that DNA demethylase TET3 preferentially enters the male pronucleus and drives prompt DNA demethylation in the zygote (Gu et al., 2011; Yu et al., 2013). H3K9me2 on the maternal genome was reported to inhibit access of TET3 to the female PN (Nakamura et al., 2012). It would be interesting to see whether TET3 can be located in the female PN while asymmetric H3K9me2 distribution is maintained in maternal Dcaf2 knockout zygotes. In addition, DNA repair pathways may be involved in the zygotic DNA demethylation process. There is no direct evidence that maternal DCAF2 directly regulates DNA demethylation. Rather, our results indicate that replication stress or large-scale DNA damage (caused by Dcaf2 deletion or DNA polymerase inhibition) prevents the establishment of the asymmetrical DNA demethylation pattern in the zygote.

Dcaf2^fl/fl mice were generated by the Model Animal Resource Information Platform, Model Animal Research Center of Nanjing University. Embryonic stem cells used to generate this mouse strain were purchased from the European Conditional Mouse Mutagenesis Program (ES cell clone EPD0842_C05). Dcaf2^fl/fl;Gdf9-Cre mice were produced by crossing mice bearing the Dcaf2^fl allele with previously reported Gdf9-Cre transgenic mice (Yu et al., 2013, 2016b). Mice were genotyped by PCR using genomic tail DNA. All primers used for genotyping and other applications are listed in Table S1. Animal care and experimental procedures were in accordance with the Animal Research Committee guidelines of Zhejiang University.

Human Dcaf2 cDNA from an Invitrogen cDNA library was subcloned into mCherry-tagged expression plasmids. Mouse Cdt1 cDNA PCR-amplified from the mouse cDNA library was subcloned into GFP-tagged expression plasmids. Linearized plasmids were in vitro transcribed using a T7 and SP6 mMESSAGEmMACHINE Kit (Invitrogen, AM1344 and AM1340, respectively) following the manufacturer's instructions. mRNAs were recovered by lithium chloride precipitation and resuspended in nuclease-free water.

For microinjection, GV oocytes were harvested in M2 medium with 2 μM milrinone to inhibit spontaneous GVBD, and zygotes were cultured in KSOM containing 20 mM HEPES. All injections were performed using an Eppendorf transferman NK2 micromanipulator. Each oocyte or zygote was injected with 5–10 pl RNA samples. After injection, oocytes or zygotes were washed and cultured at 37°C with 5% CO₂.

Female mice (21–23 days old) were intraperitoneally injected with 5 IU pregnant mare serum gonadotropin (PMSG) and with 5 IU human chorionic gonadotropin (hCG) 44 h later. After an additional 16 h, oocyte/cumulus masses were surgically removed from the oviducts, and the numbers of oocytes were counted after digestion with 0.3% hyaluronidase (Sigma-Aldrich). Oocyte images were acquired using a Nikon SMZ1500 stereoscope. To obtain early embryos, superovulated females were mated with 10- to 12-week-old WT males. Successful mating was confirmed by the presence of vaginal plugs. Embryos were harvested from oviducts at the indicated times post-hCG injection. In some experiments, in vitro fertilization was performed as previously described to synchronize zygotic development.

Oocytes at the GV stage were harvested in M2 medium (M7167; Sigma-Aldrich) and cultured in mini-drops of M16 medium (M7292; Sigma-Aldrich) covered with mineral oil (M5310; Sigma-Aldrich) at 37°C in a 5% CO₂ atmosphere. Zygotes were harvested from fertilized females after the hCG injection, and cultured in KSOM medium (Millipore).

Oocytes were fixed in 4% paraformaldehyde in PBS for 30 min at room temperature and permeabilized for 15 min in 0.3% Triton X-100 in PBS. Antibody staining was performed using standard protocols described previously (Zhang et al., 2015). The antibodies used are listed in Table S2. Imaging of embryos following immunofluorescence was performed on a Zeiss LSM710 confocal microscope.

Ovaries were collected and fixed in formalin overnight, processed, and embedded in paraffin using standard protocols (Zhang et al., 2016). Ovaries were serially sectioned at 5 μm and stained with Hematoxylin and Eosin (H&E). Immunohistochemistry was performed using standard protocols (Sha et al., 2017). The antibodies used are listed in Table S2.

To detect DNA replication, zygotes were labeled at the indicated time points after fertilization with 1 mmol/l BrdU in KSOM for 2 h. The embryos were fixed in 2.5% paraformaldehyde in PBS containing 0.5 N NaOH for 15 min at room temperature. BrdU incorporation into DNA was detected with a mouse monoclonal antibody against BrdU (Table S2), followed by Alexa Fluor 488-conjugated goat anti-mouse IgG. The embryos were mounted in 80% glycerol and observed with a laser-scanning confocal microscope. Zygotes showing small foci of BrdU signals in pronuclei were defined as ‘zygotes with weak BrdU signals’, and those showing even BrdU signals in the whole pronuclei (excluding the nucleolus-like body) were defined as ‘zygotes with strong BrdU signals’.

First, 100 oocytes or embryos were lysed in protein loading buffer and heated at 95°C for 5 min. Then, SDS-PAGE and immunoblotting were performed using a Mini-PROTEAN Tetra Cell System (Bio-Rad) following standard procedures (Yu et al., 2016a). The antibodies used are listed in Table S2.

Total RNA was extracted from oocytes or embryos using a RNeasy Mini kit (Qiagen) according to the manufacturer's instructions, followed by reverse transcription (RT) using a Superscript RT kit (Bio-Rad). Quantitative RT-PCR was performed using a Power SYBR Green PCR Master Mix (Applied Biosystems, Life Technologies) with an ABI 7500 Real-Time PCR system (Applied Biosystems) and the primers listed in Table S1. The relative transcript level of Dcaf2 in the control sample (left-hand bar of each graph) was set as 1 in Fig. 1A,B,E. The relative transcript levels of other samples were compared to the control, and the fold changes shown. For each experiment, quantitative PCR reactions were performed in triplicate.