Maternal Smc3 protects the integrity of the zygotic genome through DNA replication and mitosis

Infertility impacts the psychological well-being, economic status, mental health, sexual and marital relationships, and quality of life of women and couples (Luk and Loke, 2015; Namdar et al., 2017), with 4.65 million couples having experienced infertility in the USA in 2013 alone (Thoma et al., 2013). Among numerous factors, oocyte quality contributes significantly to infertility. Female reproduction is profoundly affected by age (Hassold and Hunt, 2001; Hunt, 2017); fertility starts to decline during the mid-thirties and accelerates thereafter. Women are born with all the eggs for their lifetime and these decrease in quality and quantity with age. Although faulty meiosis is a major contributor to female infertility, early embryonic failure also contributes, making it desirable to have visual markers of in vitro fertilization that indicate successful outcomes. Many factors essential for developmental competence are loaded into oocytes, so-called ‘maternal-effect genes’.

Chromosome segregation during early development in mammals is highly error prone and aneuploidy abounds in oocytes and early embryos (Hassold and Hunt, 2001; Vázquez-Diez and FitzHarris, 2018; Radonova et al., 2019; Pauerova et al., 2020), underscoring the need to understand how the genome is transmitted from one generation to the next. Cohesin is a genome maintenance protein complex required for successful meiosis and mitosis. It is a ring-shaped multiprotein complex that has multiple crucial functions in sister chromatid cohesion, DNA damage repair, higher-order chromosome structure and gene expression, including epigenetic reprogramming. The cohesin ring consists of four modular subunits: SMC1α or β, SMC3, RAD21 or REC8 (Scc1), and stromalin 1 (STAG1) or stromalin 2 (STAG2). Cohesin holds sister chromatid pairs together from DNA replication until mitosis (Uhlmann and Nasmyth, 1998; Losada et al., 1998; Murayama et al., 2018; Nasmyth and Haering, 2005). The biorientation of sisters at metaphase requires cohesion, which can be established between sister chromatids during DNA replication and in G2 following a double-strand break (DSB) (Uhlmann and Nasmyth, 1998; Ström and Sjögren, 2005). Cohesin enforces DSB repair from the sister template. Other SMC protein complexes include condensin and SMC5/6. SMC5/6 plays roles in DNA repair, whereas condensin is crucial for condensation of chromosomes (Aragón, 2018; Thadani and Uhlmann, 2015). A complete picture of how the genome is transmitted from the germline into the next generation requires the field to understand how SMC complexes support the maintenance of the genome during this crucial transition.

The molecular events underlying the earliest stages of embryogenesis and how they are linked to the previous meiosis are foundational to development. Relatively few maternal-effect genes have been demonstrated to be essential to support blastocyst development in mouse oocytes, although more are suspected to exist (Kim and Lee, 2014; Condic, 2016). Knowledge of the working mechanisms of maternal-effect genes is essential to understand developmental competence. Most maternal-effect genes have been shown to support imprinting and zygotic genome activation. The RAD21 subunit of cohesin and the cohesin-associated DNA-binding protein CCCTC-binding factor (CTCF) have both been shown to act as maternal-effect genes, with crucial roles in reprogramming and gene expression in the embryo (Ladstätter and Tachibana-Konwalski, 2016; Fedoriw et al., 2004; Wan et al., 2008). Smc5 depletion in oocytes with the Zp3-cre driver leads to segregation-incompetent bivalents during meiosis I (Hwang et al., 2017). Depletion of cohesin subunits in oocytes may leave cohesin associated with chromosomes because sister chromatid cohesion is established during premeiotic S phase and cohesin does not turn over, although age-associated erosion can lead to aneuploidy (Burkhardt et al., 2016; Tachibana-Konwalski et al., 2010; Hodges et al., 2005). Furthermore, the dosage of cohesin can affect chromosome structure and behavior in mouse oocytes (Murdoch et al., 2013). Among the four modular subunits, SMC3 is the only subunit that exists in all cohesin complexes; thus, its ablation will affect all mitotic and meiotic complexes. Our previous study indicated that depletion of histone deacetylase 8 (Hdac8), a key recycling factor for SMC3, in mouse oocytes does not block embryogenesis (Singh et al., 2019). However, how Smc3 contributes to genome integrity and transmission in mouse oocytes and zygotes has not been examined.

We demonstrate herein that Smc3 in oocytes is required for female fertility and integrity of the zygotic genome. We report that depletion of SMC3 in mouse oocytes causes infertility as a result of failed embryogenesis. Unlike most maternal-effect genes characterized to date, Smc3 is required to maintain the integrity and segregation of zygotic chromosomes. Depletion of SMC3 in oocytes leads to a series of molecular events that include: (1) persistence of DNA damage following DNA replication; (2) cohesion defects and lagging chromosomes at the first cleavage; (3) sister chromatid missegregation with elongated spindles; (4) micronuclei; and (5) arrest at the two-cell stage. Remarkably, and in contrast to adult females, the dosage of maternal SMC3 in juvenile mutant females is sufficient to support developmental competence, despite compromised centromeric cohesion in the zygote. We propose that maternal SMC3 in the oocyte is required to support the integrity and transmission of the zygotic genome following DNA replication. Despite the high rates of aneuploidy observed during early embryogenesis, these results highlight the fundamental importance of euploidy at the two-cell stage. Furthermore, our study suggests that zygotes can bypass chromosome cohesion defects and chromosome missegregation at the first cleavage, but that aneuploidy may be a deal breaker for the progression of two-cell stage embryos.

To study the role of Smc3 in female germ cells, we used a conditional gene knockout strategy (cKO) based on two Cre drivers. Gdf9-iCre and Zp3-Cre each express Cre recombinase in oocytes at different stages of development (Lan et al., 2004). Whereas Zp3-Cre expression in oocytes by postnatal day (P) 3 is uncontested, the expression of Gdf9-iCre at 13.5 days post conception (dpc) has been debated (Ladstätter and Tachibana-Konwalski, 2016; Lan et al., 2004). To detect the timing of Gdf9-iCre expression, we examined female gonads with a lacZ reporter. We observed β-galactosidase staining in female gonads at 13.5 dpc (Fig. S1), suggesting that Gdf9-iCre deletes the floxed-Smc3 allele in oocytes during premeiotic DNA replication. Next, we validated the efficiency of the Cre recombinases in western blots, detecting SMC3 levels in Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre oocytes along with wild-type (Smc3^fl/fl) controls. The level of SMC3 significantly decreased by 54% and 77% in Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre germinal vesicle (GV)-stage oocytes, respectively, compared with controls (Fig. S2A-C), indicating that the cKO strategy was successful, but that SMC3 persists after gene deletion.

We next investigated whether deletion of Smc3 affects female fertility by carrying out a breeding trial. Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre adult female mice were both sterile (Fig. 1A, Table 1). Although there was variation among individuals, the heterozygous KO Smc3^+/fl Zp3-Cre female mice remained fertile. This indicates that a single copy of Smc3 is sufficient to maintain developmentally competent oocytes. The level of SMC3 in heterozygotes was comparable to cKO zygotes and 66% lower than in controls (Fig. S2C). However, protein levels measured by western blot may be an imperfect indicator of function. Overall, we conclude that maternal Smc3 is required in mouse oocytes for female fertility.

We next examined how deletion of Smc3 in oocytes blocks reproduction. Based on female infertility with both Gdf9-iCre and Zp3-Cre drivers and by scoring the number of follicles in ovaries, we investigated whether loss of Smc3 impacts ovarian reserve. We performed Hematoxylin and Eosin (H&E) staining on tissue sections of ovaries from adult female mice without hormone stimulation. Corpus lutea were observed in Smc3^fl/fl, Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre female mice, suggesting that these animals ovulate, and ruling out lack of ovulation as a reason for sterility (Fig. S3A). Furthermore, the total number of follicles was comparable in adult Smc3^fl/fl, Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre female mice (Fig. S3A,B), suggesting that differences in ovarian reserve do not account for infertility. Furthermore, ovaries from Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre female mice contained a higher ratio of secondary and antral follicles (Fig. S3C), suggesting that follicle maturation is not compromised but instead appears to be accelerated. Although we cannot explain why loss of Smc3 would accelerate follicle maturation, our results suggest that loss of Smc3 does not compromise fertility via depletion of the ovarian reserve. Next, we examined whether maternal Smc3 in oocytes was required for meiosis. Both Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre metaphase II oocytes had visibly normal meiotic spindles (Fig. 1B). We further asked whether loss of Smc3 caused aneuploidy in oocytes. To measure ploidy, we performed chromosome spreads in metaphase I and metaphase II oocytes. There were no obvious defects in chromosome structure in Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre oocytes (Fig. 1C; Fig. S4). Furthermore, we observed a normal number of chromosome pairs in Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre metaphase II oocytes (Fig. 1D). Despite the differences in timing of the cre drivers, the levels of SMC3 in the oocytes appeared to be sufficient for chromosome segregation during meiosis in both cases.

Next, we asked whether loss of Smc3 impacts embryogenesis. Zygotes derived from Smc3^fl/fl, Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre female mice crossed with wild-type males were isolated and cultured in vitro. Given that the phenotype of zygotes is dominated by proteins inherited from the oocyte, and transcription is negligible prior to the first mitotic division (Hamatani et al., 2004; Aoki et al., 1997), we designated the genotype of the zygote according to the maternal allele. Zygotes from Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre adult female mice arrested predominantly at the two-cell stage (Fig. 1E, Table 2), whereas Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre embryos failed to mature to morula and blastocyst. Given that the Smc3^fl/fl;Gdf9-iCre and Smc3^fl/fl;Zp3-Cre zygotes shared the same arrest phenotype, together they provide experimental replication. Moving forward, we focused on zygotes from the Smc3^fl/fl;Zp3-Cre driver because the Zp3-Cre driver is the standard driver used to study maternal-effect genes. Hereafter, for simplicity, we refer to Smc3^fl/fl;Zp3-Cre zygotes as Smc3^Δ/Δ zygotes, recognizing that gene deletion leads to protein depletion. Our findings strongly suggest that depletion of maternal SMC3 in oocytes causes infertility by impacting embryogenesis rather than oogenesis.

To examine in more detail how depletion of maternal SMC3 affects embryogenesis, we monitored DNA damage during different phases of the zygote cell cycle (Fig. 2A). Zp3-cre-mediated deletion of Rad21, a somatic-specific cohesin subunit, leads to unrepaired paternal DNA lesions derived from paternal reprogramming in G1-phase zygotes (Ladstätter and Tachibana-Konwalski, 2016), preventing examination of DNA damage and chromosome segregation at later stages of embryogenesis. Thus, DNA lesions were quantified based on γH2AX foci. In contrast to RAD21 depletion, DNA lesions were similar in Smc3^fl/fl and Smc3^Δ/Δ G1-phase zygotes (Fig. 2B,C). The SMC3 subunit can be recycled via an acetylation/deacetylation mechanism from one cell cycle to the next (Deardorff et al., 2012; Xiong et al., 2010; Borges et al., 2010; Beckouët et al., 2010). Our results suggest that the recycling of SMC3 is sufficient for the repair of DNA lesions associated with paternal genome reprogramming in Smc3^Δ/Δ G1-phase zygotes.

DNA replication generates DNA lesions, such as DSBs (Arnaudeau et al., 2001). Spontaneous DNA lesions were observed during S phase in both wild-type and mutant zygotes (Fig. 2D,E), as expected. However, whereas DNA lesions following S phase were detected at low levels in Smc3^fl/fl zygotes, indicating repair, these lesions persisted in Smc3^Δ/Δ G2-phase zygotes (Fig. 2F,G). Given that we did not induce exogenous DNA damage, our findings strongly suggest that DNA lesions were produced spontaneously during S phase. Therefore, we speculate that maternal Smc3 is required for the repair of spontaneous DNA lesions that arise during DNA replication, as postulated based on work in budding yeast (Ström and Sjögren, 2005).

To examine further the persistent DNA lesions in G2-phase zygotes, we probed for single-stranded DNA (ssDNA), which is generated during DNA replication. We found that RPA70, a marker of ssDNA, was significantly elevated in Smc3^Δ/Δ G2-phase zygotes (Fig. 3A,B). Although ssDNA can be generated from either active transcription or replication, transcription is negligible in mouse zygotes (Aoki et al., 1997; Jukam et al., 2017). Therefore, ssDNA is predicted to derive primarily from DNA replication. Our data suggest that maternal Smc3 is required to complete repair events associated with DNA replication in zygotes.

Homologous recombination is a common method used to restart stalled replication forks (Arnaudeau et al., 2001). We investigated whether RAD51 persists into G2 phase when SMC3 is depleted. We observed a significantly higher incidence of RAD51 foci in Smc3^Δ/Δ zygotes compared with Smc3^fl/fl (Fig. 3C,D), consistent with the γH2AX data, and demonstrating that accumulation of DNA lesions in Smc3^Δ/Δ zygotes coincides with the accumulation of homologous recombination proteins. Thus, our findings suggest that maternal SMC3 is necessary to complete the repair of spontaneous DNA damage from S phase in zygotes but is not essential for the recruitment of homologous recombination machinery.

The first detectable phenotype following loss of Smc3 in oocytes was persistent DNA damage in G2-phase zygotes. However, zygotes completed the first mitotic division. We next asked why Smc3^Δ/Δ embryos arrest at the two-cell stage. Although the two-cell stage is a common arrest point in mouse embryo development, the checkpoints involved are not well characterized (Zanoni et al., 2009). Given that zygotic genome activation (ZGA) occurs late in the two-cell stage in mice, faulty ZGA can lead to arrest (Schultz, 2002, 1993). In addition, because cohesin may contribute to ZGA in zebrafish (Meier et al., 2018), we examined whether loss of maternal Smc3 is required for ZGA in mice. We measured global RNA transcription in Smc3^Δ/Δ embryos at the two-cell stage using a 5-ethynyl uridine (5-EU) click-iT assay. In this assay, G2-phase zygotes and two-cell stage embryos were pulse labeled with 5-EU, which incorporates into nascent RNA and can be quantified via fluorescence (see Materials and Methods). The signal was comparable between Smc3^fl/fl and Smc3^Δ/Δ two-cell stage embryos (Fig. S5). Although we did not determine whether the zygotic transcriptome was affected, ZGA-associated nascent transcription occurred with normal timing and to grossly normal levels in Smc3^Δ/Δ embryos, suggesting that ZGA can initiate.

Micronuclei are a hallmark of genome instability (Soto et al., 2018). They often form when a chromosome or fragment fails to incorporate into the daughter nucleus during cell division. We asked whether depletion of maternal SMC3 results in micronuclei during the first zygotic division, detected by DNA staining in two-cell stage embryos. Remarkably, we found that Smc3^Δ/Δ embryos had significantly more micronuclei compared with Smc3^fl/fl embryos (Fig. 4A,B). This strongly suggests that loss of maternal Smc3 in the oocyte results in chromosome instability during the first mitotic division. To examine the formation of micronuclei in more detail, we used live cell imaging. DNA in zygotes was stained with the low cell-toxicity SIR-DNA dye and imaged every 3 min by spinning disk confocal microscopy. This allowed us to visualize the entire process of chromosome segregation during the first mitotic division in unperturbed zygotes [Movie 1 (wild type), Movie 2 (Smc3^Δ/Δ), Fig. 4C-E]. The morphology of sister chromatids in Smc3^Δ/Δ zygotes during metaphase appeared more stretched, consistent with reduced centromeric cohesion (Fig. 4C,D). Moreover, we observed lagging chromosomes during anaphase that formed a micronucleus (Fig. 4E). In all 20 movies of Smc3^Δ/Δ embryos, lagging chromosomes were observed, as compared to one in 37 movies of Smc3^fl/fl embryos. We speculate that weakened centromeric cohesion results in merotelic kinetochore attachments, with kinetochores attached to microtubules from both spindle poles, a configuration that evades detection by the spindle assembly checkpoint, culminating in lagging chromosomes that form micronuclei (Haarhuis et al., 2013; Thompson and Compton, 2011). Our results demonstrate that maternal Smc3 is required for accurate chromosome segregation during the first mitotic division. Importantly, chromosome missegregation and aneuploidy preceded ZGA and the two-cell arrest.

Cohesin is well known for its role in sister chromatid cohesion, which facilitates biorientation and chromosome segregation. Given that our live cell imaging revealed lagging chromosomes in Smc3^Δ/Δ zygotes, we next asked whether loss of maternal Smc3 affects sister chromatid cohesion. To achieve this, we induced metaphase arrest with the pharmacological anaphase-promoting complex/cyclosome (APC/C) inhibitor proTAME and then examined centromeres and spindles. We observed a complete loss of sister chromatid cohesion and separation in all Smc3^Δ/Δ zygotes (n=21), and maintenance in all Smc3^fl/fl zygotes (n=30) (Fig. 4F). Given that precocious separation of whole chromosomes was not apparent in live cell imaging of unperturbed mutant zygotes, we suggest that proTAME exacerbates the weakened centromeric cohesion observed in live cell imaging because of prolonged metaphase arrest, with intact microtubules generating cohesion fatigue and precocious sister separation in Smc3^Δ/Δ zygotes. We conclude that maternal SMC3 is required in zygotes to create robust sister chromatid cohesion and prevent lagging chromosomes, micronuclei formation and aneuploidy.

We investigated whether a DNA-damage checkpoint is activated in Smc3^Δ/Δ embryos at the two-cell stage. To examine the canonical ATR and ATM checkpoints at this stage, we labeled Chk1 and phospho-ATM (p-ATM), respectively (Fig. S6). We first validated the antibodies by introducing replication stress using hydroxyurea (HU) treatment in wild-type G2-phase zygotes. We found that Chk1 increased in G2-phase zygotes, whereas p-ATM remained low (Fig. S6A,B). Although we did not detect p-ATM signal in G2-phase zygotes with HU treatment, this antibody was validated in our lab in mouse placental tissue in a recent study (Singh et al., 2020). We next examined Chk1 and p-ATM levels in two-cell stage embryos. Neither Chk1 nor p-ATM increased at this stage in Smc3^Δ/Δ embryos, suggesting that the canonical ATM and ATR checkpoints were not activated (Fig. S6C,D). However, checkpoints in early embryogenesis are weak or may be non-canonical (Conn et al., 2004; Vázquez-Diez et al., 2019; Adiga et al., 2007; Kato and Tsunoda, 1992). Nonetheless, aneuploidy at the two-cell stage is not tolerated because embryos do not progress to further developmental stages (Pauerova et al., 2020), suggesting that aneuploidy (i.e. micronuclei) underpins the two-cell arrest in Smc3^Δ/Δ embryos.

If persistent DNA damage and aneuploidy underpin the two-cell arrest, then rescuing this single round of DNA repair and chromosome segregation may permit embryogenesis, once Smc3 can be expressed from the paternal genome during ZGA. To determine whether rescue could be achieved, we delivered exogeneous Smc3 mRNA into Smc3^Δ/Δ zygotes using microinjection (Fig. 5A,B) prior to S phase. Whereas all mock-injected Smc3^Δ/Δ zygotes arrested at either the zygote or two-cell stage (31 total), ∼22% of microinjected mutant zygotes (nine of 44 embryos) continued past the two-cell stage, six embryos successfully matured to a morula and two matured to a blastocyst (Fig. 5C, Table 3). Therefore, our data suggest that increased levels of SMC3 in zygotes enable the first mitosis and rescue embryogenesis. Microinjection is a challenging method; the imperfect rescue could arise from several factors, including: (1) heterogeneity in the timing of mating among different female mice, such that a few zygotes enter S phase prior to microinjection, although the time was selected based on staging from Fig. 2; or (2) insufficient translation of SMC3 in some zygotes. However, the rescue of embryogenesis in a subset of zygotes is consistent with the idea that sufficient SMC3 provided during the correct time window can enable developmental competence.

In addition to microinjection, we asked whether endogenous SMC3 protein could rescue the two-cell arrest. We hypothesized that SMC3 levels in oocytes may decline starting at P3 with induction of the Zp3-Cre driver and continue until sexual maturity, leading to a gradual loss of developmental competence. If true, juvenile females might have more SMC3, which would enable the first mitotic division and developmental competence. We asked whether oocytes from juvenile Smc3^Δ/Δ female mice were developmentally competent. Given that most mouse strains reach sexual maturity at 4-7 weeks of age, we could not conduct a breeding trial. Instead, 3- to 4-week-old juvenile Smc3^Δ/Δ female mice were treated with hormones for superovulation, followed by mating, zygote collection and culture in vitro. Remarkably, zygotes from juveniles matured to blastocysts (Fig. 6A, Table 4). Also, the level of SMC3 in juvenile Smc3^Δ/Δ GV-stage oocytes was comparable to Smc3^fl/fl levels by western blot analysis (Fig. S7A,B). Therefore, our data strongly suggest that the dosage of maternal SMC3 in oocytes declines between the juvenile and sexually mature stages in the mutant. Increased levels of SMC3 in the juvenile-derived oocytes enable successful chromosome segregation at the first mitosis, such that developmental competence is achieved.

We then asked whether levels of SMC3 in juveniles were sufficient for repair of DNA lesions in Smc3^Δ/Δ zygotes. We quantified γH2AX foci in juvenile-derived Smc3^fl/fl and Smc3^Δ/Δ zygotes in G2 phase (Fig. 6B,C). Surprisingly, we found that paternal DNA lesions increased in zygotes derived from Smc3^fl/fl juvenile females compared with mature females, suggesting either a general insufficiency of repair or an elevation in the incidence of lesions. Paternal DNA lesions were further increased in zygotes derived from juvenile Smc3^Δ/Δ females, whereas maternal DNA lesions showed no difference, suggesting a partial rescue. To examine whether DNA repair machinery was affected in zygotes derived from juvenile Smc3^Δ/Δ females, we quantified RAD51 foci as described previously (Fig. 6D,E). RAD51 foci were similar in zygotes derived from juvenile and mature Smc3^fl/fl females. However, RAD51 foci were elevated in zygotes derived from juvenile Smc3^Δ/Δ females compared with Smc3^fl/fl females, similar to the observations in zygotes derived from mature Smc3^Δ/Δ females. Our findings suggest that zygotes from wild-type juvenile females have persistent DNA lesions compared with zygotes from mature females, complicating the interpretation of the results. However, juvenile-derived mutant oocytes appear to contain sufficient SMC3 to partially prevent persistent DNA damage and enable the first mitotic division in Smc3^Δ/Δ zygotes. Importantly, once ZGA occurs, Smc3 provided by the paternal genome appears to be sufficient for developmental competence, once the first zygotic division is successfully completed.

Deletion of maternal Smc3 in oocytes from mature females resulted in the loss of cohesion and sister chromatid separation in zygotes arrested in metaphase (Fig. 4F). We asked whether cohesion was intact in zygotes from juvenile Smc3^Δ/Δ female mice based on the analysis of spindles and centromeres from metaphase-arrested juvenile-derived Smc3^Δ/Δ zygotes. Remarkably, chromosomes appeared stretched on the spindle, but precocious separation of entire sister chromatids did not occur (Fig. 7A). The length/width (L/W) ratio of the metaphase plate in juvenile-derived Smc3^Δ/Δ zygotes was comparable to that in Smc3^fl/fl zygotes (Fig. 7B), suggesting that, unlike in adult-derived mutant zygotes, chromosomes were not completely stretched apart (Fig. 4C,D). Notably, we observed a Y-shaped chromatid structure in spindles from juvenile-derived Smc3^Δ/Δ metaphase zygotes. We speculate that this Y structure is a consequence of losing centromeric cohesion on telocentric sister chromatids during proTAME arrest, consistent with the idea that prolonged arrest with intact microtubules may challenge centromeric cohesion. We also observed Y-shaped chromatids in chromosome spreads (Fig. S8), indicating that intersister cohesion in the short p-arm of a telocentric chromosome is insufficient to withstand prolonged mitotic arrest. The stretched morphology of sister chromatids on the mitotic spindle of juvenile-derived mutant zygotes in proTAME (Fig. 7A) was reminiscent of the stretched appearance of unperturbed adult-derived mutant zygotes in the live cell imaging experiment (Fig. 4C), suggesting that the stretched appearance represents poor centromeric cohesion.

Interkinetochore distances increase with age in oocytes, indicative of a loss of centromeric cohesion (Duncan et al., 2012), but the physiological relevance of this is not understood. To quantify centromeric cohesion, we measured the interkinetochore distance in zygotes. Intensity profiles of sister kinetochore signals were traced by double Gaussian fitting to determine interkinetochore distances in metaphase (Fig. 7C). We measured all distinguishable chromosome pairs with appropriate orientation in individual zygotes. In adult-derived Smc3^Δ/Δ metaphase zygotes, we were unable to quantify the interkinetochore distance because sister chromatid cohesion was completely lost and pairs were unidentifiable. In juvenile-derived Smc3^Δ/Δ metaphase zygotes, the interkinetochore distance was significantly increased relative to Smc3^fl/fl zygotes (Fig. 7C,D). This strongly suggests that juvenile-derived Smc3^Δ/Δ zygotes exhibit centromeric cohesion defects, but, strikingly, can sustain developmental competence with cohesion in chromosome arms. However, by western blot, we were unable to detect significantly decreased levels of SMC3 in juvenile-derived Smc3^Δ/Δ oocytes compared with adult-derived Smc3^fl/fl oocytes (Fig. S7A,B). This imperfect correlation between the level of SMC3 and cohesion could be a result of the insensitivity of western blotting, and that protein levels on a western blot are an imperfect indicator of SMC3 actively engaged in cohesion.

Loss of cohesion is associated with increased spindle length in yeast (Stephens et al., 2011), because there is less inward force constraining the elongation (Nannas et al., 2014), although this has not been examined in mouse zygotes. We measured the morphologies of sister chromatids and spindles in proTAME-treated zygotes from both adult and juvenile females (Fig. 7E,F). The L/W ratio and length of the spindle in juvenile-derived Smc3^Δ/Δ zygotes were significantly lower than in adult-derived Smc3^Δ/Δ metaphase zygotes, suggesting that the metaphase plate and spindle morphologies improved and cohesion was more substantial (Fig. 7E,F). Moreover, the spindle length and L/W ratio of the spindle in the mutant were significantly higher than in juvenile-derived wild-type zygotes, consistent with the loss of centromeric cohesion. In sum, the improved spindle length, chromosome morphology and DSB repair in juvenile-derived mutant zygotes are consistent with more cohesion than in the adult-derived mutant zygotes. Overall, we conclude that sufficient maternal SMC3 in Smc3^Δ/Δ juvenile-derived oocytes exists to support chromosome integrity, which enables developmental competence.

Here, we demonstrate that Smc3 is a maternal-effect gene and that sufficient levels of protein must be present in the oocyte for developmental competence. Smc3 joins the 60 or so maternal-effect genes that are known to block progression from the two-cell stage (Condic, 2016). SMC3 supports the fundamental processes of chromosome duplication and segregation, distinct from the cohesin subunit RAD21 and the cohesin-binding partner CTCF, which are documented to support gene expression and imprinting (Ladstätter and Tachibana-Konwalski, 2016; Fedoriw et al., 2004; Wan et al., 2008). We demonstrate that maternal SMC3 is required to repair spontaneous DNA lesions following DNA replication in zygotes and to support accurate chromosome segregation during the first mitotic division post fertilization to prevent aneuploidy. Despite the high rates of aneuploidy observed in early embryos in other studies, which suggest that aneuploidy is tolerated, our results suggest that the integrity and transmission of the zygotic genome is essential for successful embryogenesis and rely on proteins stored into the oocyte. Furthermore, we speculate that our findings are relevant to in vitro fertilization, in which preimplantation development is often monitored, because our results suggest that elongated spindles in the zygote and micronuclei at the two-cell stage serve as visual indicators of chromosomal instability correlating with poor outcomes.

Depletion of SMC3 in oocytes results in a unique phenotype relative to all other cohesin subunits that have been studied thus far. REC8- and SMC1β-null mutations are sterile because of meiotic failure, whereas Smc1β^fl/fl;Gdf9-iCre female mice are fertile, in contrast to the infertility observed in our breeding trials (Xu et al., 2005; Revenkova et al., 2004, 2010; Bannister et al., 2004). Elimination of meiotic cohesin genes via Gdf9-iCre or Zp3-Cre drivers may not affect meiotic competence because sufficient protein remains. Given that the zygote starts using mitotic cohesin subunits instead of meiotic subunits after fertilization, elimination of meiotic cohesin may not affect developmental competence. By contrast, elimination of RAD21 in oocytes via Zp3-Cre impacts developmental competence via paternal reprogramming and severely compromises entry into mitosis (Ladstätter and Tachibana-Konwalski, 2016). Unlike SMC3, which can be recycled, RAD21 is cleaved and destroyed after each cell division. We speculate that recycling of SMC3 allows cohesin-dependent repair of reprogramming-derived DNA lesions during G1. Alternatively, RAD21 may be uniquely involved in this repair. The first phenotype detected in Smc3^Δ/Δ zygotes is persistent DNA lesions in G2 phase. However, zygotes efficiently enter mitosis and the majority then arrest at the two-cell stage. We failed to detect canonical ATR and ATM checkpoints in the two-cell-blocked Smc3^Δ/Δ embryos. Determining the checkpoint mechanism will require further experimentation, but our findings suggest that persistence of DNA damage and loss of cohesion in the zygote do not prompt a checkpoint arrest.

Although many maternal-effect genes point to the vital role of reprogramming and embryonic genome activation, our study suggests a crucial role for chromosome maintenance and euploidy at the two-cell stage, despite many studies demonstrating frequent aneuploidy in early embryos. Studies of the repair of spontaneous DNA damage from DNA replication and sister chromatid cohesion in a single cell cycle have mainly been carried out in mammalian cultured cells and yeast. Zygotes are an elegant model because factors can be depleted in the oocyte prior to the first mitotic division; in addition, zygotes reveal vital functions required in early embryos. Fifty DSBs are estimated to occur during a normal S phase in a human cell (Vilenchik and Knudson, 2003). Our study is consistent with the proposal that cohesin is required to support their repair (Singh et al., 2020; Mondal et al., 2019; Ström and Sjögren, 2005), including replication fork restart (Frattini et al., 2017). Persistence of spontaneous DNA lesions in mutant zygotes is followed by sister chromatid missegregation. Lagging chromosomes and their subsequent incorporation into micronuclei could result from merotelic kinetochore attachment as well as from compromised DNA replication, resulting in tangled chromosomes that lag in segregation and resolve into micronuclei (Fig. 8), generating aneuploidies that are incompatible with developmental progression. Filia and Bcas2 are other maternal-effect genes required to maintain genomic integrity during cleavage-stage mouse embryogenesis, supporting the proposal that euploidy is a prerequisite for developmental competence (Zheng and Dean, 2009; Xu et al., 2015). A recent study found that aneuploidy from two-cell embryos does not propagate into further developmental stages (Pauerova et al., 2020), consistent with our proposal that the mutant embryos arrest at the two-cell stage because of mitosis-derived aneuploidy, although we cannot rule out faulty ZGA.

Centromeric cohesion is considered essential for accurate sister chromatid biorientation and segregation. Our study is consistent with previous studies that show that loss of centromeric cohesion affects the physical architecture of the spindle, resulting in an elongated spindle (Stephens et al., 2011; Deehan Kenney and Heald, 2006). In a surprising twist, the Y-shaped chromatid structure in chromosome spreads derived from juvenile Smc3^Δ/Δ zygotes, coupled with their ability to progress to the blastocyst stage, suggests that loss of centromeric cohesion can be compatible with successful mitosis and developmental competence. We speculate that the Y-shaped chromatid indicates that cohesion is completely lost surrounding the centromere and throughout the short arm of telocentric chromosomes. Cohesion in the q-arm of sister chromatids in the zygote is then sufficient to enable successful biorientation, mitotic division and subsequent embryogenesis. This observation is paradoxical in light of models based on cultured mammalian cells in which arm cohesion is removed by the prophase pathway and does not contribute to sister chromatid biorientation and segregation (Waizenegger et al., 2000; Sumara et al., 2002; Losada et al., 2002; Warren et al., 2000). Future efforts will be needed to ascertain how and where cohesion is needed for sister chromatid biorientation and segregation in early embryos.

Although the exact mechanism remains unclear, cohesin participates in homologous recombination (Litwin et al., 2018). Lack of cohesin or its loader leads to the accumulation of DNA lesions after γ-irradiation (Ferreira and Cooper, 2004; Sjögren and Nasmyth, 2001; Ström et al., 2007, 2004). However, our RAD51 analysis indicates that depletion of SMC3 does not disable the recruitment of homologous recombination proteins, consistent with a previous study in human cultured cells (Kong et al., 2014). Cohesin surrounds induced DSBs and holds sister chromatids together to facilitate their repair (Litwin et al., 2018; Watrin and Peters, 2006). We speculate that cohesion is established during DNA replication at sites of spontaneous damage and that these normally invisible repair events contribute to the establishment of sister chromatid cohesion during S phase. In this speculative model, first postulated in budding yeast (Ström and Sjögren, 2005), spontaneous DNA damage during DNA replication from stalled forks results in the recruitment of cohesin to support sister chromatid repair. This cohesion then persists and contributes to sister chromatid cohesion for chromosome biorientation during mitosis. However, more work will be required to test this working model.

We demonstrate that genetically identical zygotes can differ in developmental competence depending on the age of the mother. Although SMC3 is considered stable in aging oocytes (Tsutsumi et al., 2014), whether a natural age-associated decline in other cohesin subunits in the oocyte could impact developmental competence is an open question. Our study indicates that fertility is dramatically affected in mutant females over the span of ∼2 weeks, as they progress from a juvenile to an adult. Distinct from studies that demonstrate aging-related cohesion failures in meiosis and infertility (Tsutsumi et al., 2014; Lister et al., 2010; Chiang et al., 2010; Tachibana-Konwalski et al., 2010; Duncan et al., 2012; Hodges et al., 2005; Subramanian and Bickel, 2008; Murdoch et al., 2013), our study suggests that mutant juvenile females achieve developmental competence via higher dosage in oocytes, which benefits zygotes. However, the juvenile stage is also associated with increased DNA lesions and altered spindle morphology in zygotes, even those from wild-type females. Oocyte quality declines with age, but prepubertal females also have poor-quality oocytes (Karavani et al., 2019; Oktay et al., 2021). In the future, careful consideration should be accorded to maternal age when assessing maternal-effect genes because we would have concluded that Smc3 is not a maternal-effect gene if we had only analyzed zygotes from juveniles, a protocol recommended by some textbooks (Behringer et al., 2014; Luo et al., 2011). Future efforts will benefit from additional depletion strategies, analysis at multiple ages, visualizing cohesin on chromosomes to better appreciate the functional pool, and analysis of additional proteins involved in developmental competence, chromosome segregation and spindle function (Ma et al., 2013; De and Kline, 2013).

In conclusion, our study provides insights into the working mechanism of maternal Smc3 during early embryonic development. Oocyte-stored SMC3 is required to support the integrity of the zygotic genome during the very first round of DNA replication and sister chromatid segregation to pass successfully through the first and second mitotic divisions in the embryo. Our findings are broadly consistent with the recent proposal that meiosis-derived aneuploidy persists into embryogenesis, but that mitosis-derived aneuploidy triggers arrest (Wartosch et al., 2021). Furthermore, our results suggest that elongated spindles in zygotes and micronuclei in the two-cell embryo are visual markers of poor developmental outcomes, which could be useful for in vitro fertilization.

All mice experimental protocols were approved by the Institutional Animal Care and Use Committee of the Stowers Institute for Medical Research (SIMR; Kansas City, MO, USA) and were performed accordingly. Mice were housed in a barrier facility with constant temperature, humidity and light at SIMR.

Mice were genotyped as described previously (de Vries et al., 2000; Lewandoski et al., 1997; Viny et al., 2015; Lan et al., 2004). Smc3 heterozygous mice (Smc3^fl/+) were maintained in the C57BL/6J genetic background (Viny et al., 2015). Smc3^fl/fl;Zp3-Cre and Smc3^fl/fl;Gdf9-iCre male mice were generated by crossing Smc3^fl/+ females with Tg(Zp3-Cre) (The Jackson Laboratory, #003651) or Tg(Gdf9-iCre) males (The Jackson Laboratory, #011062) in the C57BL/6J genetic background. Experimental mice were generated from heterozygous- or homozygous-floxed females with homozygous-floxed males positive for Zp3-Cre or Gdf9-iCre.

Fertility tests were carried out by crossing experimental females (6-8 weeks old) with wild-type C57BL/6J males. Female mice were checked for plugs every morning and each litter was recorded. Females were mated continuously for at least 6 months. The number of pups on the first day after parturition was counted as the litter size.

Male Gdf9-iCre mice were crossed with female Gt(ROSA)26Sor^tm1(lacZ)Cos mice (R26R, The Jackson Laboratory, #003474) and embryos were collected in PBS at 13.5 dpc. Gonads were dissected from each embryo and fixed at 4°C in fixative [1% formaldehyde, 0.2% glutaraldehyde, 2 mM MgCl₂, 5 mM EGTA, in wash solution (0.02% NP40 in PBS)] for 10 min. Gonads were washed three times for 15 min each, in wash solution at room temperature. After washing, gonads were put in staining solution [5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆.3H₂O), 2 mM MgCl₂, 0.01% sodium deoxycholate, 0.02% NP40 and 1 mg/ml X-gal] overnight and, the next morning, embryos were washed, imaged and stored in fixative at 4°C. Images were recorded using a Zeiss Axiovert microscope.

Histological tissue sections of the whole ovaries were prepared as previously described (Singh et al., 2019). In brief, whole ovaries were fixed in Modified Davidsons fixative (Electron Microscopy Sciences; 64133-50) for 6 h at room temperature and then overnight at 4°C. Each ovary was dehydrated, embedded in paraffin blocks and then serially sectioned at a thickness of 5 μm. The slides were stained by H&E. Images were acquired by a VS120 Slide Scanner (Olympus) with a 40× objective. Primordial, primary, secondary and antral follicles were scored in every five sections throughout the entire ovary as previously described (Bristol-Gould et al., 2006; Duncan et al., 2017).

To obtain fully grown GV-stage oocytes, 6- to 8-week-old females were injected with 5 IU pregnant mares' serum gonadotropin (PMSG; Genway Biotech; GWB-2AE30A). Prophase I-arrested GV-stage oocytes were isolated by physical dissection of ovaries in M2 medium (Sigma-Aldrich; MR-015-D) with 5 μM milrinone (Sigma-Aldrich; M4659). Isolated GV-stage oocytes were released from prophase I by briefly washing in Lebovitz's L-15 medium (Thermo Fisher Scientific; 11415-064). Oocytes were incubated in 90 μl drops of M16 medium (Sigma-Aldrich; M7292) covered with mineral oil at 37°C and 5% CO₂. Metaphase II oocytes were obtained 14 h after incubation.

To obtain zygotes, timed mating was performed by consecutive intraperitoneal injection of 5 IU PMSG followed by injection of 5 IU human chorionic gonadotropin (hCG; Sigma-Aldrich; C1063) 46 h later. Superovulated female mice were crossed with C57BL/6J male mice after hCG injection. The referenced time point was defined by hCG injection for all related experiments (hours post-injection; hpi). Females were sacrificed 20 hpi and oviducts were isolated by surgical dissection and placed in M2 medium. Zygotes were released from cumulus cells by a brief incubation with 500 µg/µl hyaluronidase/M2 medium (Sigma-Aldrich; H4272). Fertilized zygotes were selected based on visible pronuclei. Unfertilized oocytes were discarded, and only fertilized zygotes were incubated in ∼90 µl drops of KSOM medium (Sigma-Aldrich; MR-101-D) at 37°C with 5% CO₂.

To increase the yield of zygotes, superovulation was performed by injection of 7.5 IU CARD HyperOva (Cosmo Bio; KYD-010-EX) followed by injection of 7.5 IU hCG 46 h later. Zygotes were then isolated and incubated in vitro. The Smc3 template was amplified from the Mouse SMC3 cDNA Clone in Cloning Vector (Genomics-online; ABIN4098669) by PCR using Q5 High-Fidelity DNA Polymerase (New England Biolabs; M0515), T3 and T7 primer pairs. mRNA was transcribed in vitro followed by 3′ tailing by ultra T7 mMessage mMachine kit (Invitrogen; AM1344) following the manufacturer's instructions. The purified mRNA was resuspended in either DEPC-treated water or TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4) and then diluted to 10 ng/μl in TE buffer prior to microinjection of between 1 and 2 picoliters (pl).

To identify the cell cycle phase accurately, zygotes were pulsed with 100 mM EdU for 30 min. G1-, S- and G2-phase zygotes were classified by both the timing of fixation (G1, 20.5-22.5 hpi; S, 24-26 hpi; G2, 26-28 hpi) and the presence of EdU incorporation in the pronucleus. Only S-phase zygotes incorporated EdU into the pronucleus. Any zygote collected from the G1- or G2-phase time points with EdU signal was disqualified as a G1- or G2-phase pronucleus. To verify global RNA transcription, zygotes were pulsed with 1 mM 5-EU for 2 h before fixation.

To examine checkpoint activation, two-cell stage embryos were collected at 53.5 hpi (∼10% of wild-type embryos reached the four-cell stage at this time point). To introduce replication stress and exogenous DNA damage, zygotes were cultured with 4 mM hydroxyurea at 24 hpi for 4 h before fixation.

Metaphase zygotes were obtained from incubation with 10 µM ProTAME (Boston Biochem; I-440) to inhibit the anaphase-promoting complex (APC) for 3-4 h. Oocytes and zygotes were fixed with 2% paraformaldehyde (PFA)/PBS for 1 h at room temperature. Fixed samples were washed twice with blocking solution [0.3% bovine serum albumin (BSA), 1× PBS, 0.01% Tween-20 (Sigma-Aldrich; P1379), 0.02% sodium azide (NaN₃)] for 5 min and then stored at 4°C. Samples were incubated in a permeabilization solution [0.3% BSA, 1× PBS, 0.1% Triton X-100 (Sigma-Aldrich; T9284), 0.02% NaN₃] for 20 min and then washed twice with blocking solution for 5 min. Samples pulsed with EdU or 5-EU were processed following the guidelines of the Click-iT EdU Alexa Fluor 488 or 647 imaging kit (Invitrogen) as needed. Samples were immediately washed with blocking solution three times for 5 min each. anti-α-Tubulin 488 Alexa Fluor conjugate (Sigma-Aldrich; 16-232; 1:200), rabbit anti-γH2A.X (Cell Signaling; 9718; 1:500), mouse anti -γH2A.X (Sigma-Aldrich; 05-636; 1:500), RPA70 (Life Technologies; PA5-21976; 1:100), RAD51 (Sigma-Aldrich; PC130; 1:250), centromere serum (ImmunoVision; HCT-0100; 1:100), Chk1 (Cell Signaling; 2360; 1:100) and p-ATM (Active Motif; 39530; 1:100) primary antibodies were diluted in blocking solution and incubated with the samples overnight at 4°C. The next day, primary antibodies were washed out with blocking solution five times each for 10 min and incubated in Alexa Fluor 647 anti-human (Thermo Fisher Scientific; A-21445; 1:500), Alexa Fluor 647 anti-mouse (Life Technologies; A31571; 1:500), Alexa Fluor 594 anti-rabbit (Thermo Fisher Scientific; A-11072; 1:500), Alexa Fluor 594 anti-mouse (Thermo Fisher Scientific; A-11020; 1:500) or Alexa Fluor 488 anti-mouse secondary antibodies (Thermo Fisher Scientific; A-28175; 1:500) diluted in blocking solution for 2 h at room temperature. Secondary antibodies were washed out with blocking solution five times each for 10 min. Samples were mounted on glass slides in VECTASHIELD Antifade Mounting Medium with DAPI (Vector Laboratories; H-1200-10). Image acquisition was performed on a Carl Zeiss LSM-710 confocal microscope with a 63× oil immersion objective and z-stacks were taken every 0.5 µm.

Slides of chromosome spreads were prepared as described before (Hodges and Hunt, 2002). In brief, zona pellucida was removed in acidic M2 medium. Oocytes in metaphase II or zygotes in metaphase were briefly washed with M2 medium and then deposited on a glass slide with ∼70 µL drops of spread solution [1% PFA, 0.15% Triton X-100, 3 mM dithiothreitol in distilled H₂O (pH 9.5)]. Slides were incubated in a humid chamber for 1 h and then dried overnight. Slides could be stored at −20°C before staining. Debris was washed from slides using 0.4% Kodak Photo-Flo 200 solution twice for 5 min, 1× PBS for 5 min and 1× PBS-0.1% Tween-20. Slides were incubated in the blocking solution for 1 h. Subsequent immunofluorescence was performed as mentioned earlier, using an Alexa Fluor 488 anti-human secondary antibody (Thermo Fisher Scientific; A-11013; 1:500).

Live zygotes were incubated with 100 nM SIR-DNA dye (Cytoskeleton; CY-SC007) diluted in ∼90 µl KSOM medium for 30 min. Stained embryos were then transferred to an Interchangeable Coverglass Dish (Bioptechs; 190310-35) with a 30 mm diameter, 1.5-thickness coverslip (Bioptechs; 30-1313-0319) in SIR-DNA/KSOM medium. Image acquisition was performed on a Nikon 3PO spinning disc microscope with a 40× water objective. Microscopy was controlled by Nikon's Confocal NIS-Elements Package. Time-lapse images were acquired with ∼13 z-sections of 3 µm every 3 min.

Fully grown GV-stage oocytes from each group were isolated as mentioned earlier, and briefly washed with L-15 medium. Then, 30 oocytes per tube were collected in 10 µl of L-15 medium and snap-frozen in liquid nitrogen. Frozen oocytes were stored at −80°C before testing. Proteins were extracted using Bio-Rad sample buffer (161-0747) with 2-mercaptoethanol (Sigma-Aldrich; M7522) at 95°C for 10 min and then centrifuged at 16,000 g for 2 min; the supernatant was then collected. Proteins were separated using Bio-Rad 4-15% gradient gels (456-1084) and blotted on a PVDF membrane (GE Healthcare; 10061-494). Blots were blocked with 0.3% ECL Prime Blocking Reagent (GE Healthcare; RPN418) in Tris-buffered saline supplemented with 0.01% Tween (TBSTw) for 1 h. SMC3 (Abcam; ab9263; 1:2000) and γ-tubulin (Abcam; ab11316; 1:1000) primary antibodies were diluted in 0.3% blocking buffer at 4°C overnight. Blots were washed with TBSTw three times for 10 min each and then incubated with anti-rabbit or anti-mouse HRP-conjugated secondary antibodies (GE Healthcare; NA931V and NA934V, respectively; 1:5000). Blots were washed with TBSTw three times each for 10 min and then developed with a ECL Prime Western Blotting Detection Reagent kit (GE Healthcare; RPN2232).

Image analysis was performed using open-source Fiji software (Schindelin et al., 2012). The intensities of the western blots were quantified using the gel analysis method outlined in ImageJ. Foci quantification was performed as described previously with minor modifications (Cai et al., 2009). Briefly, image stacks were background subtracted by a rolling radius of 1.32 µm or 3.86 µm, and foci counting was performed by analysis of particles >0.5 µm² or >0.01 µm² corresponding to the cell size covered by the entire z-stack range. Thresholds were kept at (20, 255) and (150, 255) within all experiments.

To quantify the L/W ratio of the metaphase plate and spindle, the z-projections of stack images were Gaussian blurred by a radius of 1.37 µm in the live cell imaging and by 1.5 µm in the immunostaining of metaphase zygotes. Thresholds were automated within all experiments to visualize the boundaries for length measurement.

The quantification of interkinetochore distance was performed as previously described (Cahoon et al., 2017). In brief, only image stacks with an appropriate orientation were selected for the quantification. All distinguishable centromere pairs were counted in each xy image. The line profile of CREST signal along the x-axis was generated from aligning the center of two peaks. The average profiles were fitted into a two-Gaussian model to determine the width of the two peaks.

The statistical parameters and tests used are described in the figure legends. Statistical analysis was performed using Microsoft Excel 365 and R studio. The parametric two-tailed, unpaired Student's t-test and one-way ANOVA test with the Tukey HSD test were performed for datasets, whereas the nonparametric unpaired Mann–Whitney U-test was used for datasets not passing the Iglewicz and Hoaglin's robust test for multiple outliers within a z score of 3.5. Fisher's exact test was performed to examine the contingency table datasets whereas the Chi-Square test was used to examine the difference between categorical variables.

We are thankful to Ross Levine's lab for sharing the Smc3 floxed mouse line. We thank Katja Wassmann and Elvira Nikalayevich [Institut de Biologie Paris Seine (IBPS), Sorbonne Universités, Centre National de la Recherche Scientifique UMR7622, Paris, France] for kind suggestions and tips for the SIR-DNA staining and live cell imaging of mouse zygotes. We are grateful to Heidi Monnin, Timothy Corbin and the Lab Animal Service Facility of the Stowers Institute for Medical Research (SIMR) for mouse husbandry and microinjection. We are grateful for helpful comments and suggestions from W.-T.Y.’s thesis committee members Robb Krumlauf (SIMR), Jay Unruh (SIMR), and John Marko (Northwestern University). We sincerely thank Tamara Potapova (Gerton lab) and Francesca Duncan (Northwestern University) for helpful discussions and support.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199800.