Mad2 is dispensable for accurate chromosome segregation but becomes essential when oocytes are subjected to environmental stress Free

Clinically, many women face infertility or birth defects, such as embryo implantation failure, miscarriage, or congenital conditions such as Down syndrome (trisomy 21) and Edwards syndrome (trisomy 18), and the incidence increases markedly with maternal age. Such conditions are primarily due to the aneuploid gamete (Nagaoka et al., 2012), typically caused by chromosome mis-segregation during cell division. For mammalian oocyte maturation, this process is notoriously prone to error, especially during the first meiotic division (Hassold and Hunt, 2001). The spindle assembly checkpoint (SAC), as a highly conserved surveillance system, plays an indispensable role in faithful chromosome segregation of both mitotic and meiotic eukaryotic cells (Jones and Lane, 2013). Studies on the molecular mechanism underlying the SAC first started in somatic cells and revealed that the core components of the mitotic SAC include Mad1 (Mad1l1), Mad2 (Mad2l1), Bub1, Bub3, BubR1 (Bub1b), Mps1 (Ttk) and aurora B (Saurin et al., 2011). These proteins collaborate with numerous other factors to form a complex network of checkpoint signaling (Garcia et al., 2021; Musacchio, 2015). The most central function of the SAC is to control cell cycle progression, i.e. to check whether all kinetochores are properly attached to microtubules and block the metaphase-to-anaphase transition in case unattached kinetochores persist. SAC proteins are recruited to unattached kinetochores and generate a diffusible ‘NO’ signal resulting in the inhibition of an E3 ubiquitin ligase, anaphase-promoting complex/cyclosome (APC/C) (Alfieri et al., 2016), and a temporary or permanent arrest in metaphase. Mads, Bubs, BubR1 and Mps1 are all involved in the control of checkpoint arrest (Meraldi et al., 2004; Liu et al., 2003); however, among these proteins, mitotic progression was accelerated only in the absence of Mps1, BubR1 and Mad2, and mitotic timing is independent of their kinetochore localization (Meraldi et al., 2004; Tighe et al., 2008; Maciejowski et al., 2010). In addition, many SAC factors also play direct or indirect roles in chromosome alignment, such as Mps1 (Sarangapani et al., 2021), aurora B (Jelluma et al., 2008), BubR1 (Suijkerbuijk et al., 2012), Bub1 (Klebig et al., 2009) and Mad1 (Akera and Watanabe, 2016).

Mad2 is one of the key components of the mitotic checkpoint complex (MCC), the effector of the SAC. It is well-established that Mad2 is the most downstream component in the SAC. The Mad1–Mad2 complex is recruited to unattached kinetochores (Chen et al., 1999), and then catalyzes the formation of a complex between Mad2 and Cdc20 (Sironi et al., 2001). Indeed, many studies have confirmed that Mad2 is required for ordered cell cycle progression and genome stability in mitosis. In mouse studies, Mad2 knockout caused embryonic lethality, which was triggered by apoptosis due to an uncontrolled and accelerated cell cycle, and chromosome mis-segregation (Dobles et al., 2000; Burds et al., 2005). Similar phenotypes were observed in pre-implantation embryos in which Mad2 was knocked down (Wei et al., 2011) and in heterozygous adult mice (Michel et al., 2001). Chromosome mis-segregation without Mad2 may be explained by a finding in human cells that MAD2 stabilizes kinetochore–microtubule attachments. This function is independent of the checkpoint or its kinetochore localization (Kabeche and Compton, 2012).

Meiosis, a specialized cell division during germ cell development, exhibits unique features distinct from mitosis, such as recombination and segregation of homologous chromosomes during meiosis I. Numerous studies by us and others have demonstrated functional SAC signaling in mouse oocytes. During meiosis I, SAC ensures the timely and correct segregation of homologous chromosomes by monitoring whether all kinetochores are correctly attached to the bipolar spindle. Specifically, BubR1 (Touati et al., 2015), Bub1 (McGuinness et al., 2009) and Mps1 (Hached et al., 2011) not only have a controlling role, but also influence timing in meiosis I. Furthermore, oocytes showed an increased incidence of misaligned chromosomes and aneuploidy in the absence of BubR1 (Touati et al., 2015), Bub1 (McGuinness et al., 2009), Mps1 (Hached et al., 2011), Bub3 (Li et al., 2009) and Mad1 (Zhang et al., 2005). These proteins re-establish or stabilize attachments, or stabilize spindle microtubules, thus promoting alignment. From these elegant studies, we know that SAC proteins also perform functions beyond the checkpoint, promoting chromosome alignment and accurate segregation in mouse oocyte meiosis.

Meiotic Mad2 plays a crucial role in regulating oocyte progression. Previous studies by us and others showed that Mad2 localizes to unattached kinetochores and, driven by dynein, is removed from kinetochores once attachments are well-established in meiotic oocytes (Wassmann et al., 2003; Zhang et al., 2004, 2007). Depletion of Mad2 in oocytes through in vitro approaches, such as RNA interference (Wang et al., 2007), morpholino (Homer et al., 2005a,b), and mutant injection (Wassmann et al., 2003), all resulted in premature activation of APC/C, accelerated first meiotic progression, and aneuploidy. In addition, Mad2^+/− female mice exhibited impaired fertility, and their oocytes showed phenotypes similar to those described above (Niault et al., 2007). Owing to the lack of robust, genetics-based research, little is known about the relationship between Mad2 and other SAC proteins in meiotic oocytes. In this study, we depleted Mad2 in mouse oocytes using a conditional-knockout approach and present evidence that, although Mad2 controls the timing of meiosis I progression, it is dispensable for female fertility. However, Mad2 plays a crucial role in maintaining chromosome euploidy in response to environmental stimuli. Our study extends our understanding of the function of Mad2 in meiotic oocytes.

We first determined the expression of Mad2 protein in mouse oocytes at four stages, including germinal vesicle (GV), germinal vesicle breakdown (GVBD), metaphase I (MI) and metaphase II (MII), corresponding to culture for 0, 4, 8 and 12 h, respectively. The results showed that Mad2 was expressed during all stages of meiotic maturation, and the level was relatively constant from GV to MI but increased significantly at MII (Fig. 1A). To explore the function of Mad2, we conditionally depleted Mad2 in the mouse oocyte using previously described Mad2^Flox/Flox mice (Foijer et al., 2013) in which ‘LoxP’ sequences flank the Mad2 coding sequence (exons 2 to 5). We crossed Mad2^Flox/Flox mice with Gdf9-Cre mice to generate Mad2^Flox/Flox; Gdf9-Cre+ female mice (hereafter referred to as Mad2^−/− mice) (Fig. 1B). We determined genotypes of Mad2^Flox/Flox (as control mice) and Mad2^−/− mice by PCR (Fig. 1C). Western blotting analysis confirmed the absence of Mad2 in mutant oocytes (Fig. 1D).

To address the roles of Mad2 in oocyte meiotic progression, GV-stage Mad2^−/− oocytes were harvested and induced to resume meiosis (Fig. 2A). We found that GVBD rates between the two groups were not significantly different at 0.5, 1, 1.5 and 2 h after release from prophase I arrest by removal of 3-isobutyl-1-methylxanthine (IBMX) (Fig. S1). Therefore, Mad2 does not play a role in GVBD.

However, we found that polar body extrusion (PBE) was initiated at 6 h after release in Mad2^−/− oocytes, compared with 10 h in controls, indicating that PBE occurred dramatically earlier in Mad2-depleted oocytes (Fig. 2B). Furthermore, to confirm that the phenotype was due to the absence of Mad2, we performed rescue experiments. As expected, premature PBE of Mad2^−/− could be rescued by injecting full-length Mad2 mRNA (Fig. 2B). In addition, PBE rates within 14 h did not differ between the two groups (Fig. 2C). It is well-established that the core function of the SAC is to induce metaphase arrest when some kinetochores remain unattached. We treated oocytes with a low dose of nocodazole and found that control oocytes could arrest during meiosis I, whereas most Mad2^−/− oocytes could not (Fig. 2D). This indicates that SAC control is defective in Mad2^−/− oocytes.

We then wondered whether oocytes matured in the ovary also complete meiosis I prematurely. Because it is difficult to visualize the oocyte maturation in vivo, we harvested ovaries and pricked the follicles to collect oocytes 9 h after human chorionic gonadotropin (HCG) injection (Fig. 2E) and found that the proportion of PBE in Mad2^−/− oocytes was significantly higher than that in controls (Fig. 2F,G). Because it is difficult to guarantee that all oocytes in the ovary can be efficiently collected and analyzed, we performed a histological analysis of ovaries 9 h after HCG injection (Fig. 2E). Oocytes with a polar body were visible in most follicles of Mad2^−/− females; however, they were rarely found in control follicles (Fig. 2H,I). These findings suggest that PBE is also accelerated in Mad2^−/− oocytes in vivo.

Next, to understand in more detail the dynamics of meiotic progression in Mad2-deficient oocytes, we microinjected histone H2B mRNA into oocytes to label chromosomes. Live-cell imaging showed that premature chromosome segregation occurred at a dramatic rate at 5.5 h after GVBD on average in Mad2^−/− oocytes (Fig. 2J,K). Taken together, we demonstrated that the absence of Mad2 in oocytes not only impairs SAC control but also results in accelerated progression of meiosis I.

In mouse oocytes, separase (encoded by Espl1) is a cysteine protease that dissolves arm cohesion by cleaving kleisin Rec8 (Kudo et al., 2006). We detected the activity of separase by microinjecting mRNA for a separase sensor and following sensor activity by live-cell imaging. This reporter consists of a Rad21 fragment that is cleaved by activated separase, which causes an EGFP signal to disappear, followed by the appearance of red fluorescence (Fig. 3A). In Mad2^−/− oocytes, the red signal emerged early, whereas in control oocytes the signal did not appear until 9 h after GVBD (Fig. 3B), indicating premature activation of separase in the absence of Mad2.

During the metaphase-to-anaphase transition in meiosis I, securin (Pttg1) is depleted by activated APC/C, which results in separase becoming activated (Herbert et al., 2003). We hypothesized that the untimely anaphase I in Mad2^−/− oocytes could be attributed to the premature activation of APC/C. We analyzed the degradation of two APC/C substrates, cyclin B1 and securin, by time-lapse imaging. As expected, in Mad2^−/− oocytes, the destruction of both cyclin B1 (Fig. 3C,E) and securin (Fig. 3D,F) occurred much earlier compared with controls, with minimum intensities at an average of around 6 h after GVBD. These results suggest that in Mad2-deficient oocytes, APC/C is activated prematurely, resulting in premature degradation of cyclin B1 and securin, which in turn results in early activation of separase.

Given that the above phenotypes are similar to those observed after the loss of other SAC proteins, which are indispensable for fertility (Touati et al., 2015; McGuinness et al., 2009; Hached et al., 2011), we wondered about the necessity of Mad2 for female fertility. A six-month breeding assay was performed. To our surprise, the females lacking Mad2 in their oocytes were fertile (Fig. 4A), and the litter sizes were comparable to controls (Fig. 4B). Moreover, we performed superovulation (Fig. 4C) and found that the numbers of ovulated oocytes between the two groups were also comparable (Fig. 4D). In addition, the offspring from the two groups were similar in size, shape and mental state. None of the offspring suffered death or developmental delay (Fig. 4E).

Previous studies established that the sterility of female mice with oocytes lacking SAC proteins was caused by aneuploid eggs (Touati et al., 2015; McGuinness et al., 2009; Hached et al., 2011). To investigate whether chromosome segregation errors occur in Mad2-deficient oocytes, we collected ovulated MII oocytes for chromosome spreading. Most Mad2^−/− oocytes exhibited normal chromosome numbers (N=20). The euploidy rates were not significantly different between the two groups (Fig. 4F-H; see Table S1A for specific data). In summary, normal ploidy is consistent with normal fertility, indicating that Mad2 in oocytes is not indispensable for chromosome separation control and female fertility in mice.

However, when we cultured Mad2^−/− oocytes in vitro (Fig. 2A) and then spread chromosomes during MII, it was found that the euploidy rate (51.67%) was significantly lower than that of the control (84.70%) (Fig. 5A,B; see Table S1A for specific data). Specifically, the distribution of the number of chromosomes in each egg was wider than in the control (Fig. 5C). In addition, 27.61% of Mad2^−/− oocytes contained pre-separated sister chromatids, which only accounted for 1.45% of control oocytes, a difference that was statistically significant (Fig. 5D,E). These results do not seem to match the data in the previous section.

To explore the contradiction, we further studied in vitro-matured (IVM) oocytes. First, oocytes were fixed at 4 h after GVBD (the period before chromosome segregation in Mad2^−/− oocytes) or were induced to exogenously express Venus-β-tubulin to analyze the spindle, an important factor for chromosome segregation. The spindle morphology and size of Mad2^−/− oocytes was normal (Fig. S2). Additionally, previous studies have revealed that the activity of mitogen-activated protein kinase (MAPK) (Erk1/2; Mapk3/Mapk1) is gradually elevated after GVBD (Sun et al., 1999a,b) and is involved in microtubule assembly in mouse oocytes (Verlhac et al., 1993, 1994). We collected Mad2^−/− oocytes from three stages (2 h, 3.5 h and 5 h after release) before PBE and control oocytes from five stages (GV, 3.5 h, 5 h, 6.5 h and 8 h after release) before metaphase I. Western blotting showed that the phosphorylation level of Erk1/2 in Mad2^−/− oocytes was gradually increased after GVBD, synchronized with the level in controls (Fig. S3). Thus, we conclude that chromosome mis-segregation in Mad2^−/− oocytes in vitro is not related to the spindle organization.

Then, to visualize the dynamic activity of chromosomes, exogenous mCherry-H2B and Venus-β-tubulin were expressed in oocytes to track chromosomes and spindles through time-lapse imaging. Chromosomes were aligned before segregation in most of the control oocytes. However, only some Mad2^−/− oocytes aligned their chromosomes (see ‘a’) before segregation, whereas the remaining did not. In the latter oocytes, two obvious phenotypes were found: most chromosomes were aligned but individual ones were not (see ‘b’), or almost all chromosomes failed to align (see ‘c’) (Fig. 5F,G). For quantitative assessment, we measured the width of the chromosome plate in each oocyte (Fig. 5H). For the Mad2^−/− oocytes with misaligned chromosomes, the width was significantly larger than in the control. But for the Mad2^−/− oocytes with aligned chromosomes, the width was comparable to the control (Fig. 5I). Next, we investigated whether the difference was due to the premature PBE of Mad2^−/− oocytes (i.e. not enough time for alignment). Thus, we measured the width frame by frame. We found that the width of the control group became gradually reduced and tended to be stable. However, the width of the Mad2^−/− group did not decrease with a similar slope; it was accompanied by fluctuation (Fig. 5J). This confirms that chromosome alignment is indeed less efficient in IVM Mad2^−/− oocytes. Furthermore, nearly 40% of the Mad2^−/− oocytes displayed lagging chromosomes during segregation (see ‘a’), or individual chromosomes were separated from the spindle (see ‘b’) (Fig. 5K,L). However, what are the reasons for such impaired chromosome alignment and segregation of Mad2^−/−oocytes in vitro? Is it due to the absence of cumulus cells? We therefore cultured the cumulus–oocyte complex (COC) of Mad2^−/− mice and performed chromosome spreading. The results showed that the presence of cumulus cells was not sufficiently effective to compensate for segregation errors of Mad2^−/− oocytes, suggesting that cumulus cells are not the dominant factor causing differences in vivo and in vitro (Fig. S4). Additionally, no misalignment was observed during MII, and cytostatic factor-induced metaphase arrest was sustained in Mad2^−/− oocytes (Fig. S5). In conclusion, these results illustrate that a portion of IVM Mad2^−/− oocytes have abnormal chromosome segregation during meiosis I, resulting in aneuploidy.

In mouse oocytes, several studies showed that Mad2 loses the kinetochore localization after the deletion of BubR1 or Mps1 (Touati et al., 2015; Hached et al., 2011). We next explored whether Mad2 depletion would adversely affect the recruitment of other SAC proteins, or whether the checkpoint would be silenced. We performed chromosome spreading on Mad2^−/− and control oocytes at prometaphase I (3 h after GVBD) and stained BubR1, Mad1 and Bub3, respectively. In Mad2^−/− oocytes, each of the three proteins localized at the kinetochores and had fluorescence intensity comparable to the control (Fig. 6A), suggesting normal recruitment of these proteins.

Mps1 is another essential SAC kinase in meiosis (Hached et al., 2011). Reversine has an inhibitory effect on Mps1, thereby inhibiting SAC activity (Santaguida et al., 2010). After GVBD, we added reversine to the M2 medium to treat oocytes until the MII stage (Fig. 6B). We found that reversine did not further affect the meiosis I progression of Mad2^−/− oocytes. Except for the non-treated control group, the progression of the other three groups was relatively similar (Fig. 6C). Furthermore, through chromosome spreading of MII oocytes, we found that reversine exacerbated the aneuploidy of Mad2^−/− oocytes (Fig. 6D,E). This illustrates that, in Mad2^−/− oocytes, Mps1 plays a role in promoting chromosome alignment and proper segregation during meiosis I. Based on these results, we hypothesized that SAC proteins are able to be recruited and still perform some functions when Mad2 is absent.

Based on the different phenotypes in vivo and in vitro, we investigated whether environmental pressure has an effect on oocyte maturation. Mouse and human oocytes are known to be very sensitive to low temperatures, which affects events such as spindle assembly (Pickering and Johnson, 1987; Wang et al., 2001). For further exploration, we transferred the oocytes to a preset 30°C incubator after GVBD (Fig. 7A). Compared with the curves at 37°C (Fig. 2B), the curves of both groups at 30°C were elongated horizontally and depressed vertically, suggesting a slower meiosis I progression and lower cumulative percentage of oocytes that had undergone PBE (PBE%) at each moment. However, PBE of Mad2^−/− oocytes started 6 h after release, whereas it was significantly delayed to about 12 h in the control group (Fig. 7B), and the stable cumulative PBE% of Mad2^−/− oocytes was significantly higher than in the control (Fig. 7C). These data suggest their different ability to complete meiosis I when under the same environmental stress, which may also provide evidence of weak checkpoint control in Mad2^−/− oocytes.

Chromosome spreading showed that the euploidy rates of Mad2^−/− and control MII oocytes were significantly different (Fig. 7D,E, left). However, although the control had a high euploidy rate (69.13%), it actually showed a low PBE% (33.63%); although the Mad2^−/− group had a very low euploidy rate (1.28%), it showed a high PBE% (69.34%). Therefore, we can eliminate the impact of the PBE% in order to accurately reflect how many of the MI oocytes after GVBD can generate euploid MII oocytes. That is, in the control group about 33.63%×69.13%=23.25% of the MI oocytes developed into euploid MII eggs. In contrast, in the Mad2^−/− group, this rate was only about 69.34%×1.28%=0.89%. The difference between the two groups is still highly significant (P<0.01) (Fig. 7E, right) (see Table S1A for detailed data). Moreover, the distribution of the chromosome numbers in each Mad2^−/− MII oocytes was much more diffuse than in the control (Fig. 7F). In parallel, the percentage of oocytes with pre-separated sister chromatids was dramatically increased in Mad2^−/− compared with control oocytes (Fig. 7G,H).

Summarizing the data in previous sections, euploidy was impaired in both control and Mad2^−/− oocytes when the environment became harsher. However, there were differences in the slope of the dashed lines between the two groups (Fig. 8A,D). For the euploidy rate, two-way ANOVA indicated that there was an interaction effect between the genotype and environment. A further simple effect test showed that the euploidy rates of the Mad2^−/− oocytes differed significantly, not only between in vivo and IVM groups, but also between IVM 37°C and IVM 30°C groups, whereas the control did not (Fig. 8B; see specific analysis in Table S1B). As the environment became worse, the distribution of the number of chromosomes in Mad2^−/− oocytes became increasingly diffuse compared with controls (i.e. a greater proportion of Mad2^−/− oocytes contained more or fewer chromosomes, compared with 20) (Fig. 8C). In addition, for pre-separated chromatids, a similar trend was observed in Mad2^−/− oocytes (Fig. 8E,F; see specific data and analysis in Table S2A,B). In summary, we demonstrate that, with environmental pressure, the ability of MI Mad2^−/− oocytes to divide into euploid MII oocytes becomes weaker than in controls.

Chromosome segregation in mammalian oocytes is error prone (Pellestor et al., 2005). The APC/C activity must be accurately controlled to avoid mis-segregation. This heavy task is performed by SAC. Mad2, a core member of SAC, has been predominantly studied in mouse oocytes, but research on genetic levels is still lacking. Here, we for the first time investigated the role of Mad2 in oocytes by conditional knockout and showed that lack of Mad2 had no effect on the meiotic resumption, but accelerated meiosis I progression. However, mutant female mice generated euploid eggs and were fertile, but their oocytes showed chromosome mis-segregation and aneuploidy when matured in vitro, which was exacerbated when the environment was disadvantageous.

It is known that SAC plays an important role in the control and prolongation of meiosis I. First, in our study, Mad2-depleted oocytes in which spindle microtubules were destroyed failed to arrest (Fig. 2D). Mps1 inhibitor treatment had no effect on meiotic progression in Mad2^−/−oocytes (Fig. 6C). Second, Mad2^−/− oocytes extruded the first polar body about 4 h earlier than in the controls, which was almost identical to that observations in Bub1 and Mps1 knockout oocytes (McGuinness et al., 2009; Hached et al., 2011). Meiosis I progression of Mad2^−/− oocytes in vivo also showed an accelerated tendency. The absence of Mad2 caused premature degradation of securin and cyclin B1 as well as activation of separase, followed by early metaphase-to-anaphase transition, suggesting premature activation of APC/C. Taken together, these data are consistent with previous knockdown studies of Mad2 and knockout studies of other SAC proteins involved in meiosis. We thus further demonstrate the SAC function of Mad2 as a controller and timer of the first meiotic progression (Fig. 9A).

Defects in SAC are often accompanied by aneuploidy in both mitosis and meiosis. However, female mice with Mad2 knockout in oocytes unexpectedly had normal fertility, whereas IVM Mad2^−/− oocytes showed chromosome mis-segregation, which was further exacerbated at a lower temperature. Specifically, mis-segregation showed two features: unequal distribution of the homologous chromosomes (measured by euploidy rate) and precocious separation of sister chromatids (measured by the chromatid rate).

Unequal distribution of homologous chromosomes has been observed in Mps1, Bub1 and BubR1 knockout oocytes, leading to aneuploidy and sterility (Touati et al., 2015; McGuinness et al., 2009; Hached et al., 2011). However, the distribution was correct in Mad2^−/− oocytes matured in vivo but increased aneuploidy was observed in IVM Mad2^−/− oocytes, indicating sensitivity of the SAC machinery to environmental changes. Aneuploidy increased further when the kinase Mps1 was inhibited (Fig. 6D), suggesting that other SAC factors still play roles in the absence of Mad2, but with unstable efficiency and vulnerable SAC machinery.

Premature sister-chromatid separation is common in aging animals. In humans and mice, this is due to weakened sister-centromere cohesion (hereafter called ‘cohesion’) (Chiang et al., 2010). There are two possible causes for premature sister chromatid separations in Mad2^−/− eggs. (1) A pair of sister chromatids separate at meiosis II as a result of weakened cohesion and separated kinetochores, with no change in the total number of chromosomes. This occurred in Bub1 knockout or Mps1-inhibited oocytes (McGuinness et al., 2009; El Yakoubi et al., 2017). (2) In contrast to previous studies, predivision of univalents in meiosis I (3:1 segregation) was observed in Mad2^−/− oocytes. For example, some oocytes contained an odd number of chromosomes, indicating that the sister is not in the egg. Predivision is the main consequence of merotelic attachment during meiosis I (Chiang et al., 2010). In addition, we also observed lagging chromosomes in IVM Mad2^−/− oocytes (Fig. 5K). Previous studies showed that merotelic attachment is dissolved by aurora B/C through several rounds of corrections, and inhibition of aurora causes lagging chromosomes (Kitajima et al., 2011), but merotelic attachment in somatic cells cannot be detected by the SAC (Cimini et al., 2001). It has also been shown that partial lagging in the oocytes of aged mice directly causes aneuploidy (Mihajlović et al., 2021). So, we speculate that the weaker cohesion and inefficient correction of merotelic attachment together led to 3:1 segregation in Mad2^−/− oocytes. Bub1 and Mps1 are required for the correct localization of Sgo2 and therefore the protection of cohesion (McGuinness et al., 2009; El Yakoubi et al., 2017), and biochemical evidence shows that human MAD2 is an interactor of SGO2 (Orth et al., 2011). Thus, with environmental stress, sister chromatids separate at both meioses in Mad2^−/− oocytes, possibly as a result of weakened cohesion and misattachments.

Taken together, chromosome segregation in Mad2-deficient oocytes is sensitive to external changes and is error prone. When developing in vivo, under optimum conditions, despite the shortened prometaphase I, cell cycle-related biochemical events proceed in an orderly way; however, notably, other SAC proteins may be unstable but still functional in chromosome segregation, so fertility is normal in mutant females. It is known that the quality and cellular events of IVM oocytes are inherently disturbed (Krisher, 2013). When matured in vitro, rising environmental stresses increase the challenge of chromosome alignment beyond the threshold that the unstable SACs can cope with, even at 37°C. Furthermore, perhaps the unstable SACs themselves, which lack Mad2, are susceptible to stresses and become less efficient. Thus, mis-segregation occurs because of the combined effect of these factors as well as the weakened cohesion and misattachments. We speculate that the aneuploidy observed in previous Mad2 in vitro studies may be attributed to the co-reaction of environmental stress and Mad2 deletion, but not simply to the deletion of Mad2 itself. We compare other SAC proteins and Mad2 to a street cleaner and his coat. His duty is to sweep the leaves together. When the weather is sunny, it is no challenge for him to sweep the leaves, even without his coat. But on windy and cold days, without the coat he has no energy to sweep the more disordered leaves together (Fig. S6). We, therefore, suggest that Mad2 may directly or indirectly stabilize SAC proteins and protect cohesion in meiotic oocytes. This differs from the role of Mad2 in mitosis as well as other SAC proteins in meiosis (Fig. 9B).

The spindle assembly checkpoint is based on the signaling cascade of SAC proteins. According to the mainstream mitotic model, Mad1 is recruited to kinetochores through hierarchical recruitment of upstream aurora B, Mps1, Bub1 and Bub3. Then, Mad1 recruits Mad2, which participates in MCC assembly (Vigneron et al., 2004; Dou et al., 2019; Heinrich et al., 2012). However, the pathway in meiosis is not well established. Deletion of Mps1 or BubR1 causes loss of kinetochore localization of Mad2 (Hached et al., 2011; Touati et al., 2015), whereas deletion of Mad2 did not affect the localization of BubR1, Bub3 or Mad1 in our study (Fig. 6A). Mad2 may be the most downstream of these proteins, joining MCC as the final step (Fig. 9A), which is similar to its role in mitosis.

First, it is well known that chromosome segregation in mammalian oocyte meiosis is error prone, but the newfound identity of Mad2 in our study suggests, from the opposite perspective, that oocytes have protective mechanisms against changes in follicular microenvironments or external environments. This thereby reduces the risk of chromosome mis-segregation with subsequent serious consequences. Second, the molecular mechanisms that ensure equal segregation of homologous chromosomes and the timely segregation of sister chromatids in different organisms may be diverse and deserve further study. Future studies are required to elucidate in detail the mechanisms by which Mad2 stabilizes SAC proteins and chromosome alignment, as well as stabilizing cohesion during meiosis I. Third, it is perceived that SAC proteins are highly interconnected. A single protein may have multiple roles, and multiple proteins may also perform the same function. This suggests that SAC proteins make up a complex network and each protein cannot be studied in isolation. Similarly, events such as the checkpoint pathway, chromosome alignment, spindle assembly and kinetochore-microtubule attachment cannot be separated.

In this study, we demonstrate that, in mouse oocytes, Mad2 controls meiosis I progression and prolongs prometaphase, which is consistent with its role in mitosis. However, depletion of Mad2 does not affect in vivo-matured oocyte euploidy and fertility. Mad2 contributes to maintaining oocyte euploidy when exposed to environmental challenges. This study reveals a non-essential role of Mad2 in oocyte meiosis compared with the other SAC proteins or compared with mitosis.

The species of animal used in the experiments was Mus musculus. Mad2^Flox/Flox mice were obtained from Allan Bradley's laboratory (Foijer et al., 2013). To obtain oocytes with conditional knockout of Mad2, Gdf9-Cre mice, which express Cre recombinase via the Gdf9 promoter, were crossed with Mad2^Flox/Flox mice. The resulting Mad2^Flox/+; Gdf9-Cre+ male mice were then crossed with Mad2^Flox/Flox female mice to obtain Mad2^Flox/Flox; Gdf9-Cre+ female mice, which were used as the experimental animals, and littermates Mad2^Flox/Flox female mice were used as the control animals. The strain of mice was C57BL/6J with an ICR background. To measure the deletion of Mad2, the primers used for genotyping were: Mad2 forward 5′-AGGCTGAGCCGGGCCTTAGGAC-3′, Mad2 reverse 5′-GTAACCGTGTAATAACGTTTAAGTCTC-3′ (annealing temperature 66.4°C); Cre forward 5′-TCTGATGAAGTCAGGAAGAACC-3′, Cre reverse 5′-GAGATGTCCTTCACTCTGATTC-3′ (annealing temperature 60°C).

All animal handling protocols were approved by the Institutional Animal Care Committee of the Institute of Zoology (IOZ), University of Chinese Academy of Sciences (UCAS). Mice were housed in an environmentally controlled experimental animal center (IOZ, UCAS) with constant temperature and humidity, adequate food and water, and alternating cycles of 12 h of light and 12 h of darkness.

Three pairs of full-sibling control females and experimental females were used for breeding assessment. Eight-week-old female mice were initially caged with wild-type male mice (ICR, 12 weeks old) for 6 months. The offspring were counted on the day they were delivered. The appearance of the offspring was observed for 2 months.

Oocytes were collected from ∼6- to 10-week-old female mice. For the collection of ovulated MII oocytes, mice were injected intraperitoneally with 10 U of pregnant mare's serum gonadotrophin (PMSG). After 46-48 h, intraperitoneal injection of HCG was performed. After another 14 h, the mice were euthanized by cervical dislocation, and then the bilateral Fallopian tubes were carefully removed with scissors and briefly placed in pre-warmed drops of M2 medium (M7167, Sigma-Aldrich). Afterwards, the oviducts were transferred into another drop of M2 medium, containing hyaluronidase at a concentration of 1 mg/ml. Then, the ampulla of the oviduct was torn using a syringe, allowing the COC to flow out. After the cumulus cells were dispersed, oocytes were harvested with a flame-pulled glass mouth pipette for subsequent experiments.

For the collection of pre-ovulatory oocytes, ovaries were collected 9 h after HCG, then large follicles were pricked with a syringe in drops containing hyaluronidase. Oocytes were then allowed to flow into the droplets as much as possible and the cumulus cells were fully digested. Later, oocytes were washed and transferred to drops of the M2 medium, and after GV-stage oocytes were removed and the remaining cells were counted as soon as possible. Other steps were the same as described above.

For COC collection, mice were first injected with 10 U of PMSG. After 46-48 h, mice were killed in the same manner mentioned above to collect bilateral ovaries. Follicles were pierced sufficiently with a syringe to allow COCs to flow out. COCs were gently collected by mouth pipetting and then cultured in droplets of the M2 medium covered with mineral oil (M8410, Sigma-Aldrich) at 37°C, 5% CO₂. After 15 h, the cumulus cells were removed by blowing with a mouth pipette, and all oocytes except those in the GV stage were collected for subsequent experiments.

For GV-stage oocyte collection, mice were first injected intraperitoneally with 10 U of PMSG. After 40-42 h, the mice were killed to collect ovaries, followed by rapid mincing of the ovaries with a razor blade. The tissue was immediately mixed into M2 medium, containing 200 μM IBMX (M2-IBMX) to maintain the prophase I arrest of oocytes. Second, the full-grown oocytes were collected by mouth pipetting and washed several times in drops of medium to adequately remove cumulus cells. For cell incubations, the GV oocytes were either placed in drops of M2-IBMX medium for prophase arrest or washed nine times with drops of IBMX-free medium and then cultured for continuing maturation. When incubating oocytes, droplets of the M2 medium were covered with mineral oil. Unless otherwise stated, the culture conditions were: 37°C, 5% CO₂. Oocytes undergoing GVBD within 1.5 h of release were used for subsequent experiments. When cultured at a lower temperature, oocytes were first incubated at 37°C, 5% CO₂ until GVBD occurred, and then transferred to another incubator with the conditions set to 30°C, 5% CO₂. Except for the temperature, other conditions and procedures remained unchanged. For nocodazole treatment, the final concentration was 0.2 μM, and it was added into the M2 medium 3 h after GVBD. For reversine treatment, it was dissolved in DMSO and stored at −20°C and added into the M2 medium at a final concentration of 0.5 μM after GVBD.

At specific periods as described above, full-sibling control and Mad2-depleted female mice were sacrificed, and their ovaries were immediately fixed in 4% paraformaldehyde (PFA) at 4°C for 24 h. Tissues were dehydrated in graded ethanol series and embedded in paraffin (39601095, Leica). Then, each ovary was serially sectioned at a thickness of 5 μm with a microtome (Leica), and all sections were collected and sequentially attached to adhesive slides (188105, Citotest). After drying at 45°C overnight in an oven, sections were deparaffinized with xylene, hydrated through graded alcohol series, and stained with Hematoxylin and Eosin. Images were collected with a comprehensive digital pathology scanner (Aperio VESA8, Leica) at 10× magnification. For statistical analysis, all sections of the same follicle were observed consecutively. Oocytes in which polar bodies were found were considered as having undergone PBE and those in which no polar body was found were considered as meiosis I stage.

To quantify the level of endogenous protein, oocytes at different stages were first washed with PBS to remove the proteins in the medium, then counted and collected into the Eppendorf tubes, and, finally, placed into liquid nitrogen. After that, 6 µl of 2×SDS loading buffer (P0015L, Beyotime) was added to the tube, and the samples were lysed in boiling water for 5 min, then put on crushed ice for temporary storage, or stored at −20°C for subsequent analysis. For western blotting, sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE) was prepared in advance (the concentrations of stacking gel and resolving gel were ∼12-15%, and 5%, respectively). The above samples were concentrated at 80 V for about 30 min and then separated at 120 V for ∼90-120 min. Later, proteins were transferred from the gel to the polyvinylidene difluoride (PVDF) membrane (ISEQ00010, Millipore) by the semi-dry transfer method (Bio-Rad) for 28-30 min. After four washes with Tris-buffered saline with 0.1% Tween-20 (T8220, Solarbio) (TBST), the membranes were blocked with 5% Albumin Bovine V (bovine serum albumin, BSA) in TBST on a rocking platform for 1 h at room temperature (RT), or overnight at 4°C. The membranes were then incubated with primary antibodies overnight at 4°C. After four washes with TBST, the membranes were incubated with secondary antibodies for 1 h at 37°C, or for 2 h at RT. Finally, after washing the membrane four times, 8 min each, the luminescent solution (34095, Thermo Fisher Scientific) was added dropwise to the membranes, which were then exposed with an enhanced chemiluminescence imaging system (Bio-Rad, ChemiDoc XRS+).

The following primary antibodies were used to detect proteins: mouse anti-β-actin (1:1500; TA-09, ZSGB-Bio; RRID: AB_2636897), mouse anti-Mad2 (1:100; sc-47747, Santa Cruz Biotechnology; RRID: AB_627902), mouse anti-Erk1/2 (1:1000; 4696, Cell Signaling Technology; RRID: AB_390780), rabbit anti-phospho-Erk1/2 (1:1000; 4370, Cell Signaling Technology; RRID: AB_2315112). The following secondary antibodies were used: horseradish peroxidase-conjugated goat anti-rabbit IgG (1:2000; BE0101, Easybio) or horseradish peroxidase-conjugated goat anti-mouse IgG (1:2000; BE0102, Easybio; RRID: AB_2923205).

At each specific stage, oocytes were incubated in Tyrode's solution (T1788, Sigma-Aldrich) to remove the zona pellucida adequately, then immediately transferred back to the M2 medium for three washes. On one side of the adhesive slides (188105, Citotest), 1 cm×0.5 cm frames were drawn with a hydrophobic pen, and around 10 µl of the spread solution (1% PFA in distilled H₂O containing 0.15% Triton X-100 and 3 mM dithiothreitol, pH 9.2) were added to the frame areas. Oocytes were gently arranged one by one in the solution with mouth pipetting. Slides were kept as still as possible until the droplets were thoroughly dry. Fixed oocytes were blocked with 1% BSA in PBS for 1 h at RT and then incubated with specific primary antibodies overnight at 4°C. After three washes (10 min each time) with washing buffer (0.1% Tween-20 and 0.01% Triton X-100 in PBS), the corresponding secondary antibodies were added for 1 h at RT. Chromosomes were then stained with 1:10,000 diluted 4′-6-diamidino-2-phenylindole solution (DAPI, 10 μg/ml in PBS) for 10 min at RT after three washes (10 min each time). Finally, after dipping a small amount of DABCO (2% triethylenediamine and 10% 0.2 M Tris-HCl pH 7.4 in glycerol) into the frame with a pipette tip, the slides were covered with coverslips. Starting with blocking, each step of the incubation was performed in a wet box, to prevent the liquid from drying out.

To localize centromeres, a primary human anti-centromere antibody (ACA) antibody (1:50; 15-234, Antibodies Incorporated; RRID: AB_2687472) with a corresponding secondary antibody conjugated to Alexa Fluor Cy5 (1:200; 709-175-149, Jackson ImmunoResearch; RRID: AB_2340539) were used. To detect SAC proteins, the following antibodies were also used: rabbit anti-Mad1 antibody (1:600; GTX105079, GeneTex; RRID: AB_11173437) and rabbit anti-Bub3 antibody (1:500; ab133699, Abcam; RRID: AB_11155962) with a corresponding secondary antibody conjugated to Alexa Fluor 594 (1:1000; A11012, Thermo Fisher Scientific; RRID: AB_2534079), as well as primary mouse anti-BubR1 antibody (1:300; ab54894, Abcam; RRID: AB_940666) with a corresponding secondary antibody conjugated to Alexa Fluor 488 (1:1000; A11029, Thermo Fisher Scientific; RRID: AB_2534088).

For spindle and ACA staining, oocytes were first washed with the culture medium and then transferred to 0.5% Triton X-100 in PHEM buffer (containing PIPES, HEPES, EGTA and MgSO₄) for 5 min at RT to permeabilize. Next, oocytes were washed three times (5 min each time) in washing buffer (PBS containing 0.05% polyvinyl pyrrolidone, PVP) and fixed in a solution of 3.7% PFA in PHEM containing 0.00112% KOH for 20 min at RT. After three more washes (10 min each time) with the above washing buffer, oocytes were blocked with blocking solution (1% BSA and 0.75% glycine in PHEM) to block nonspecific binding sites for at least 1 h at RT. On the first night, oocytes were incubated with primary anti-ACA antibody diluted in blocking solution (described above) overnight at 4°C. On the following night, oocytes were co-incubated with the anti-ACA antibody (as described above) together with FITC-conjugated anti-α-tubulin antibody (1:1000; 322588, Thermo Fisher Scientific; RRID: AB_2532182) overnight at 4°C. On the third day, after four washes (15 min each time) with PBST buffer (0.05% Tween-20 in PBS), oocytes were incubated with the corresponding secondary antibody conjugated to Alexa Fluor Cy5 for 1 h at RT. Finally, after three washes with PBST (10 min each time), chromosomes were stained with DAPI for 15 min at RT. The four corners of the coverslips were evenly coated with lanolin, and the force was neither light nor heavy when covering.

Oocytes were analyzed by ultra high-resolution confocal microscopy (LSM880 Fast Airyscan, Carl Zeiss AG). For chromosome spreading, oocytes were imaged with a 40× water-immersion objective. In the z-axis direction, imaging was performed only at the section with the strongest punctate signal (e.g. ACA, Bub3, etc.). When imaging the SAC proteins, a consistent laser intensity was used for both control and Mad2^−/− samples, and the brightness was subsequently adjusted using the same parameters as well. For imaging of spindles, a 63× oil objective was employed, and imaging was performed at 1.0-μm intervals in the z-axis direction. ZEN software (version 2.3, Carl Zeiss AG) was used for image stack generation, processing, export and measurement. The fluorescence intensity of BubR1, Bub3 and Mad1 was quantified using ImageJ (version 1.53a, National Institutes of Health).

Euploidy assessment was based on counting kinetochores with ImageJ (version 1.53a, National Institutes of Health). In this study, n represents the number of separated sister chromatids and N represents the total number of chromosomes, i.e. the number of paired chromosomes ×1+the number of separated chromatids ×0.5. Oocytes were considered to be euploid only if N=20 and there were no separated chromatids (n=0).

The mouse Mad2 gene (NM_019499.5) was synthesized by the Beijing Genomics Institution and then recombined into the pCS2(+) vector. Deletion of the Myc sequence in Mad2-PCS2(+) resulted in a plasmid for Mad2 without any tag. Similarly, the human PTTG1 gene (NM_001282382), which encodes securin, was also synthesized by the Beijing Genomics Institution and recombined into the pcDNA3.1(+)-EGFP vector. The separase sensor plasmid has been previously described (Li et al., 2019).

Plasmids were digested with the corresponding restriction endonucleases, i.e. linearized, and the linear template DNAs were then purified in preparation for in vitro transcription. Corresponding to different promoters, cRNAs were transcribed with SP6 or T7 mMESSAGE mMACHINE Transcription Kit (Invitrogen). Afterwards, poly-A tails were synthesized with the Poly (A) Polymerase Tailing Kit (Lucigen). Finally, cRNAs were purified with RNeasy Mini Kits (QIAGEN) and dissolved with RNase-free water, then dispensed into 1 ml aliquots and placed at −80°C for long-term storage.

Prior to microinjection, in vitro-transcribed cRNAs were diluted with RNase-free water to appropriate concentrations: 100 ng/µl for Mad2-Myc, 1000 ng/µl for Mad2 (with no tag), 20 ng/µl for H2B-mCherry, 200 ng/µl for cyclin B1-Venus, 500 ng/µl for securin-EGFP, 100 ng/µl for β-tubulin-Venus and 400 ng/µl for separase sensor.

GV-stage oocytes were collected for microinjection and placed in a flattened drop of M2-IBMX (∼20 μl) onto the lid of a culture dish, covered with mineral oil. An inverted Nikon microscope and electrical micromanipulators were employed, as well as a FemtoJet Microinjector pump (Eppendorf). Holding pipettes and injection pipettes were made with a pipette puller and a microforge. To minimize the period outside of optimal cultural conditions, the duration of microinjection did not exceed 20 min. Oocytes were allowed to recover for at least 10 min, before being transferred to droplets for culture.

Oocytes injected with Mad2-Myc were cultured in M2-IBMX for 6 h for cRNA expression before being released for the subsequent rescue experiments shown in Fig. 2B. Mad2 (with no tag) cRNAs were overexpressed for 24 h for the western blotting shown in Fig. 1D. As for several other cRNAs, the duration was shortened to 3 h.

At 1.5 h after release, GVBD oocytes were collected and transferred to oil-covered microdrops (∼8-14 μl) of M2 medium in a glass-bottom Petri dish (Φ20 mm; NEST) for live-cell imaging. The number of oocytes placed depended on the volume of each drop, i.e. one oocyte per microliter. Confocal imaging was performed with a spinning-disk, inverted confocal microscope (Andor Dragonfly 200), equipped with an sCMOS camera, and a 20×/ NA0.75 objective was used. The microscope was controlled using Imaris software (version 9.2.1, Bitplane). Each droplet corresponds to one imaging field. Images were taken every 20 min or 30 min. Seven z-sections of whole-mount oocytes were acquired. The distance between z-sections was 7.14 μm.

The raw images were post-processed with Imaris software. All images were stacked and assembled first before analysis and quantification. To quantify the fluorescence intensity of cyclin B1-Venus and securin-EGFP, the mean fluorescence intensity of each oocyte at each time point was analyzed and calculated by the software. Normalization was then performed based on the intensity at the first time point. In addition, measurement of the length of the spindle (Fig. S2C) and the width of the chromosome distribution (Fig. 5H) was also performed with this software. Specifically, the distance between the most distantly distributed chromosomes was manually measured, with the long axis of the spindle as a reference. To minimize errors, measurements were not made for oocytes for which the long axis of the spindle was parallel to the z-axis.

GraphPad Prism 9 was used to compare the means of two independent samples. First, each sample was tested for normal distribution, and then an F-test was used to test for the homogeneity of variance between the two samples. If the above two results were not significantly different, a two-tailed, unpaired Student's t-test was performed. If the variances were not homogenous, a t-test with Welch's correction was performed. If the samples did not accord with a normal distribution, a non-parametric, two-tailed Mann-Whitney test was performed. For two-factor data, a two-way ANOVA was first carried out using R (version 4.2.2). If the interaction effect was significant, a simple effect test was then performed using SPSS (version 26.0). The number of independent replicates and the sample sizes are shown in the figures or figure legends. Data are all shown as mean±s.e.m., and P<0.05 was considered statistically significant.

We thank Dr Floris Foijer and Petra L. Bakker for generously providing suggestions in our experiments; Dr Floris Foijer for editing the manuscript and improving the language. We thank Shiwen Li for assistance with live cell imaging; Shiwen Li and Yue Wang for assistance with confocal imaging; Hua Qin for assistance with tissue imaging using a digital pathology scanner; and Xili Zhu for assistance with imaging acquisition and processing. We thank Dr Jian Li for providing plasmids. We thank all members of the Sun Lab for their help and discussions.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201398.reviewer-comments.pdf.