Mouse female meiotic spindles assemble from acentriolar microtubule-organizing centers (aMTOCs) that fragment into discrete foci. These are further sorted and clustered to form spindle poles, thus providing balanced forces for faithful chromosome segregation. To assess the impact of aMTOC biogenesis on spindle assembly, we genetically induced their precocious fragmentation in mouse oocytes using conditional overexpression of Plk4, a master microtubule-organizing center regulator. Excessive microtubule nucleation from these fragmented aMTOCs accelerated spindle assembly dynamics. Prematurely formed spindles promoted the breakage of three different fragilized bivalents, generated by the presence of recombined Lox P sites. Reducing the density of microtubules significantly diminished the extent of chromosome breakage. Thus, improper spindle forces can lead to widely described yet unexplained chromosomal structural anomalies with disruptive consequences on the ability of the gamete to transmit an uncorrupted genome.
The microtubule (MT) spindle enables equal segregation of chromosomes between daughter cells during mitosis and meiosis. Checkpoints monitor the attachment of each chromosome to the spindle apparatus to ensure that this process is faithful (Etemad and Kops, 2016; Touati and Wassmann, 2016). Even though numerical errors induced by defects in spindle assembly have been extensively studied in mitosis and meiosis, not much is known about the mechanisms or the origins of structural errors arising during M-phase (Brunet et al., 2003; Nagaoka et al., 2011; Kolano et al., 2012; Sacristan and Kops, 2015). However, a link between abnormal mitosis and DNA damage has been established (Ganem and Pellman, 2012). For example, prolonged mitosis is often related to DNA damage (Rieder and Palazzo, 1992; Rieder and Maiato, 2004). Merotelic attachments, in which one kinetochore of a chromosome is attached to both spindle poles, are common in tumor cells with extra centrosomes (Ganem et al., 2009). These attachment errors cause chromosome mis-segregation at mitotic exit (Cimini et al., 2001) and subsequent formation of micronuclei and chromothripsis in daughter cells (Cimini et al., 2001; Crasta et al., 2012; Zhang et al., 2015).
In the absence of canonical centrosomes, which are made from a pair of centrioles and pericentriolar material (PCM), meiotic spindles form with an important contribution from chromosome-mediated MT nucleation pathways. Thus MT assembly is enhanced in the vicinity of meiotic chromosomes (Heald et al., 1996; Dumont et al., 2007; Dumont and Desai, 2012; Bennabi et al., 2016). In mouse oocytes, PCM aggregates remain despite centriole loss, a feature that describes them as acentriolar MT-organizing centers (aMTOCs). It is unclear how aMTOCs assemble in the absence of centrioles, but they largely contribute to meiotic spindle assembly (Maro et al., 1985; Dumont et al., 2007; Schuh and Ellenberg, 2007). In fully grown prophase I–arrested oocytes, aMTOCs constitute typically one to three large foci (Luksza et al., 2013). At meiotic entry, these foci stretch around the nuclear envelope and fragment in a dynein- and MT-dependent manner (Luksza et al., 2013; Clift and Schuh, 2015). The importance of such timely reorganization of aMTOCs has not been addressed.
Plk4 is the master regulator of centriole duplication (Bettencourt-Dias et al., 2005; Habedanck et al., 2005) and is also essential for meiotic and mitotic acentriolar spindle assembly in oocytes and early mouse embryos by stimulating MT growth (Coelho et al., 2013; Bury et al., 2017). In the present study, we took advantage of the role of Plk4 to perturb aMTOC organization in oocytes (Luksza et al., 2013). Transgenic overexpression of Plk4 (Marthiens et al., 2013) in oocytes during their growth phase in the ovary led to precocious aMTOC fragmentation before nuclear envelope breakdown (NEBD). It also led to earlier spindle bipolarization associated with greater MT nucleation. After NEBD, oocytes completed meiosis I. However, one bivalent broke into two pieces, consistent with breaks occurring near the recombined Lox P sites present in the genome of three different mouse lines. Crucially, a decrease in MT density significantly diminished the efficiency of bivalent breakage. Altogether, our work suggests that increasing MT density and probably forces exerted on fragilized chromosomes can lead to bivalent breakage during meiosis I, as shown here for three different chromosomes. This in turn leads to mis-segregation of chromosome fragments. Chromosome rearrangements and fragment losses are relatively frequent in human embryos and lead to severe pathologies. This is the case for intellectual disability syndromes such as Wolf–Hirschhorn caused by a 4p deletion (Wieczorek et al., 2000), the 9p deletion (Okten et al., 2009), or the 6q terminal deletion (Striano et al., 2006). We hypothesize that our work may identify one mechanism by which structural chromosome anomalies can arise during meiosis.
Plk4 overexpression in growing oocytes causes premature aMTOC fragmentation
As previously published (Bury et al., 2017), we confirmed that Plk4 is present in mouse oocytes. For this, we labeled endogenous Plk4 by immunofluorescence in control oocytes blocked in prophase I of meiosis (Fig. 1 A, bottom left). By costaining with the PCM component pericentrin (Fig. 1 A, top left), we showed that Plk4 localizes to aMTOC, arguing that it is not a limiting factor for centriole assembly in oocytes.
To test the importance of aMTOC organization on spindle assembly (Bennabi et al., 2016), we took advantage of the role of Plk4 (Bury et al., 2017) to try and perturb them in oocytes using a novel approach. Conditional Plk4 overexpression in oocytes was achieved by crossing a line expressing the Cre-recombinase under the ZP3 promoter with a line containing a random insertion of a chicken β-actin promoter (CAG) fused to a floxed stop-codon and to mCherry-Plk4 (Fig. S3 A, referred to as Plk4 OE; Marthiens et al., 2013). The ZP3 promoter is active after birth, after homologous chromosomes have recombined and formed chiasmata, and only during oocyte growth (Lewandoski et al., 1997). In this way, we obtained excision of the stop-codon and expression of the m-Cherry-Plk4 transgenic construct exclusively in oocytes during their growth period at puberty.
In both Plk4 OE prophase I–arrested oocytes (from Plk4flox/wt; Cre+ female mice) and control oocytes (coming either from Plk4flox/wt; Cre− or Plk4wt/wt; Cre+ female mice), Plk4 was detected on aMTOCs (Fig. 1 A, bottom). However, Plk4 OE oocytes displayed modified aMTOC characteristics with numerous small foci in the vicinity of the nucleus (Fig. 1 A, right). In Plk4 OE oocytes, the amount of Plk4 at aMTOCs was more than six times higher than in controls (Fig. 1 B), while the amount of pericentrin at aMTOCs remained unchanged (Fig. 1 C). Because of the relative size of the oocyte and the aMTOCs, the local aMTOC signal corresponds to only 0.1% of the signal in the entire oocyte. It is thus not so surprising that no detectable increase in the total amount of either pericentrin or Plk4 was observed in whole Plk4 OE oocytes compared with controls (Fig. S1, A and B). These data argue that exogenous Plk4 preferentially accumulates on aMTOCs and is present there in large excess.
To describe the changes in the organization of aMTOCs, we performed 3D quantitative analysis of competent, fully grown Plk4 OE oocytes in prophase I (Fig. 1, D and E). The number and the cumulated surfaces of all aMTOCs were both significantly higher in Plk4 OE compared with controls. These observations suggested that aMTOCs were more fragmented in Plk4 OE oocytes.
aMTOC fragmentation occurs at the end of the prophase I block
We used 3D structured illumination microscopy (3D-SIM) to look more closely at the detailed morphology of PCM in prophase I–arrested Plk4 OE oocytes, as has been done in centriolar cells (Fu and Glover, 2012; Lawo et al., 2012; Mennella et al., 2012; Sonnen et al., 2012). In controls, oocytes display one major aMTOC, harboring a globular shape with the pericentrin matrix displaying multiple cavities filled with Plk4 (Luksza et al., 2013). In contrast, fully-grown Plk4 OE oocytes, competent to resume meiosis I, contained multiple aMTOCs that assembled into thread-like structures when running parallel to the nuclear envelope (Fig. S1 C, right). The pericentrin internal cavities were no longer detected in these long thread-like structures, indicating a dramatic change in PCM organization.
The superresolution microscopy data and the absence of increase in pericentrin labeling, together with the 3D quantifications of aMTOC organization, argue for a premature fragmentation of the PCM in Plk4 OE oocytes (Fig. S1 D). This fragmentation occurs at the end of oocyte growth, in prophase I, as incompetent Plk4 OE oocytes present only one globular pericentrin matrix, filled with Plk4, as in controls (Fig. S1, C [left] and D; Luksza et al., 2013). Thus, even if induced early on during the growth phase, Plk4 overexpression results in major changes in the organization of aMTOCs only at the end of the prophase I block, once the oocyte reaches competence to resume meiosis I.
Overexpression of Plk4 during the growth phase of the oocyte is needed for aMTOC fragmentation
The fragmentation of aMTOCs occurred at the end of the oocyte growth phase; thus, we tested whether a similar phenotype could be reproduced by directly injecting the cRNA encoding for mCherry-Plk4 into fully grown oocytes in prophase I (Fig. 2 A). When the mCherry-Plk4 cRNA is injected, much higher levels of Plk4 are reached than with Plk4 OE, both globally and associated with aMTOCs (Fig. 2, B–D). In the case of the cRNA, the total number of aMTOCs per oocyte was higher than in controls but was comparable to the number observed in the transgenic line overexpressing Plk4 (Fig. 2, B and E). However, instead of several aMTOCs presenting widths ranging from 1 to 5 µm2 as in Plk4 OE oocytes (Fig. 2 F), in each injected oocyte we detected one large aMTOC and multiple tiny foci of mCherry-Plk4 (smaller than 1 µm2; compare left vs. right panels in Fig. 2 B and see size quantification in Fig. 2 F). Although we do not know how these very small aMTOCs are generated, the large aMTOC is the dominant site for MT nucleation in these oocytes (Fig. 2 G, left). This is strikingly different from what can be observed in the Plk4 OE transgenic prophase I oocytes, where all fragments of aMTOCs show significant MT nucleation (Fig. 2 G, right). Altogether, our results suggest that Plk4 overexpression during the growth phase of oocytes (via transgenic Plk4 OE), but not when limited to the competent/fully-grown stage (via cRNA injection), leads to premature fragmentation of aMTOCs into smaller functional units able to promote significant MT nucleation.
Fragmented aMTOCs have increased MT-nucleating capacity
To test the impact of precocious aMTOC fragmentation, oocytes were then allowed to resume meiosis, and local MT density was observed around the condensing chromosomes at the time of NEBD as done previously (Breuer et al., 2010). The intensity of EB3-GFP fluorescence in the vicinity of chromosomes was significantly higher (1.4×) in Plk4 OE oocytes, indicating that they contained more MT than controls (Fig. 3, A and B). Importantly, the increase in EB3-GFP intensity at NEBD correlated with the number of aMTOCs observed in prophase I (Fig. 3 C). However, such an increase in MT density at NEBD was not observed after overexpression of mCherry-Plk4 via cRNA injection into fully grown oocytes (Fig. 2 H), despite higher levels of overexpression reached by this approach (Fig. 2, C and D). This result suggests that the extent of aMTOCs fragmentation in prophase I impacts on the MT density at meiosis resumption.
To estimate the consequences of an increased MT density at NEBD, we followed the formation of meiotic spindles in live EB3-GFP–injected oocytes (Fig. 3 D and Video 1). In control oocytes, the entire meiosis I lasts ∼9 h from NEBD until polar body extrusion (PBE). At early stages of meiosis resumption after NEBD, the newly assembled mass of MT self-organizes into a bipolar array ∼3–4 h after NEBD (Dumont et al., 2007; Schuh and Ellenberg, 2007). As expected from previous work (Dumont et al., 2007; Kolano et al., 2012), increased density of MT observed in Plk4 OE oocytes resulted in earlier spindle elongation and bipolar axis formation (on average 45 min earlier in Plk4 OE oocytes compared with controls; Fig. 3, D and E). This result was confirmed on fixed samples observed 4 h after NEBD and analyzed by immunofluorescence (Fig. S2, A and B). More oocytes (61.4% vs. 26.1%) contained a robust bipolar spindle in Plk4 OE oocytes compared with controls at 4 h after NEBD (Fig. S2 B). Thus, in Plk4 OE oocytes, meiotic spindle bipolarity is established earlier than in controls.
Plk4 OE oocytes are endowed with chromosome breakage
We then analyzed the consequences of premature increased density of MT early on during meiosis I. Unexpectedly, Plk4 OE oocytes frequently contained 21 individual chromosome fragments, instead of the normal 20 bivalent complement of mouse. Remarkably, we observed the appearance in live cells of a round chromosome, smaller than all others, that oscillated back and forth across the metaphase plate, suggesting it was not correctly bi-oriented (Fig. 4 A, black arrows; and Video 2). Indeed, this small DNA fragment had only one pair of kinetochores positive for Mad2-YFP (Fig. 4 A, right inset). We wanted to better characterize this breakage event, which has never been described previously in the first meiotic division without prior treatment with a DNA-damaging agent (Collins et al., 2015; Marangos et al., 2015; Lane et al., 2017). Therefore, oocytes were fixed at NEBD + 6h00, centromeres were labeled with CREST, and 3D reconstruction was used to count the number of bivalents in oocytes derived from two independent insertions of the same Plk4 transgene (transgenic lines A and B; see Materials and methods, Fig. S3 C, and Videos 3, 4, and 5). In transgenic line A, we observed the breakage of one bivalent as depicted in Fig. 4 A: one large and one small fragment, both of which harboring a pair of centromeres labeled with CREST (Fig. 4 B, upper; and Video 4, compare with Video 3). The pattern of breakage was very reproducible within this line (Fig. 4 C showing various examples artificially highlighted in blue). In line B, we also observed the breakage of a single bivalent, but different from line A, generating a fragment of DNA lacking CREST signal (Fig. 4, D and E; and Video 5) and much smaller than any fragment detected in oocytes from line A (Fig. 4 F). Without any centromere and thus devoid of the ability to assemble a kinetochore, this smaller fragment was not retained in the meiotic spindle (Fig. 4 D, left, and other examples of the small fragment artificially highlighted in green in Fig. 4 E). The frequency at which we could observe chromosome breakage in the two lines differed: it was more frequent in line A (27.0%) than in line B (10.2%; Fig. 4 G and summarized in Fig. S3 C). Differences in the rate of chromosome breakage between the two strains were not a result of different overexpression levels of Plk4, as no difference in mCherry-Plk4 expression was found at aMTOCs between the two lines (Fig. S4, A and B).
Because the breakage occurred only on one bivalent, in a specific manner for both transgenic lines, we hypothesized that it took place at the insertion site of the mCherry-Plk4 transgene—this being the only major difference between the two lines. The site of insertion was sequenced for both lines. In line A, it was found to be close to the centromeric region of chromosome 9, theoretically generating a small 45-Mb DNA fragment if the cut site occurs in the region of the transgene. In line B, it was close to the telomeric region of chromosome 12, which would generate a very small 15-Mb acentric DNA fragment if the cut occurs in the region of the transgene (Figs. 4 H and S3 C). These sites of insertion are fully consistent with the fragments observed in both lines: two fragments containing centromeres for line A (Fig. 4 B), and a smaller acentric fragment for line B (Fig. 4 D).
To support the sequencing data, we performed FISH experiments on oocytes from line A. The chromosome fragments corresponded to the genome location of transgene insertion, as 100% of the observed broken chromosomes were labeled with a FISH probe for chromosome 9 (Fig. 4 I). Chromosome breaks were never observed in oocytes coming from Plk4flox/wt; Cre− mice or Plk4wt/wt; Cre+ female mice (our negative controls that do not overexpress Plk4) or in the mCherry-Plk4 cRNA-injected oocytes (Fig. S2 C). Furthermore, we know that recombination and repair of the mCherry-Plk4 transgene did happen in all Plk4 OE oocytes from both lines, since 100% of them were positive for mCherry-Plk4. Therefore, it can be excluded that the break took place early on during oocyte growth. Our data strongly support the view that the chromosome breakage occurs at or near the site of the recombined transgenic insertion at late stages of oocyte growth or even later during the division (Fig. 4 J).
Chromosome breakage can occur at any Cre-recombined Lox P site provided aMTOCs are fragmented
We then tested whether the creation of another fragile site, different from the transgenic Plk4 insertion, could also be associated with breakage of one bivalent. For this, we crossed Plk4 line Aflox/wt; Cre+ mice with MyoXflox/wt; Cre− mice. The MyoXflox/wt; Cre− line has been constructed by insertion of two Lox P sites between exon 26 and exon 30 of the myosin X endogenous locus in chromosome 15 (Fig. S3 B). These mice are fertile and display no obvious phenotype in their oocytes (not depicted and Fig. 5, lower panels). Oocytes coming from the Plk4 line Aflox/wt; Cre+; MyoXflox/wt females were fixed at NEBD + 6h00 (as done in Fig. 4, B and D), and their chromosomes were counted. 12.5% of these oocytes displayed a broken bivalent resembling the one observed in Plk4 line A (Fig. 5 A, upper panels, blue arrows), while 15.6% presented a new type of break: a large piece of DNA, with no centromere (Fig. 5 A, lower panels, orange arrow; and Video 6). We did not observe oocytes presenting the two types of break simultaneously. Their frequency might be too low, and we did not look at enough oocytes coming from this double cross (Fig. 5 B). The volume of the new acentric fragment was significantly larger than the smallest DNA piece observed in both line A and line B (Fig. 5, C and D). We interpret this new type of break as a breakage at the MyoX locus on chromosome 15 (Fig. 5, E and F), since it is consistent with the larger size of the acentric fragment as well as with the potential chiasma location on this chromosome (Morgan et al., 2017). Acentric fragments of chromosomes are not retained in the spindle (see Fig. 4 D). Remarkably, we observed 2/3 of oocytes with the large acentric fragment close to the metaphase plate (Fig. 5 D, left), while 1/3 was away from the spindle (Fig. 5 D, right). This observation could indicate that the fragile bivalent is cut into two pieces while meiosis I progresses. Importantly, no chromosome break was observed in control oocytes from Plk4 line Aflox/wt; Cre−; MyoXflox/wt or Plk4 line Awt/wt; Cre+; MyoXflox/wt females (Fig. 5 A, lower panels). This indicates that the break occurs only when two conditions are met: Plk4 overexpression associated with aMTOC fragmentation and recombined Lox P sites. These data argue that sites breaking under tension can be created in the genome, outside the Plk4 insertion.
Extent of bivalent breakage correlates with the extent of aMTOCs spreading and MT density
Plk4 OE oocytes have fragmented aMTOCS, nucleate more MTs and thus present an accelerated spindle formation. All this could cause changes in forces applied on chromosomes during meiosis I and favor chromosome breakage. We went on to characterize further the relationship between spindle dynamics and the generation of broken bivalents.
First, we compared line A and line B. As mentioned previously, more chromosome fragments are observed in line A than in line B (Fig. S3 C). Consistently, we observed that the extent of aMTOC fragmentation in prophase I was higher in line A than in line B (Fig. 6 A), despite them having the same level of overexpression at the end of oocyte growth (Fig. S4, A and B). Kinetics of synthesis and/or recruitment of Plk4 might be different between lines during the growth phase of oocytes, potentially impacting aMTOC biogenesis.
We then counted oocytes presenting a broken chromosome at different time points in meiosis I (in Plk4 OE from line A). At 4h00 after NEBD, we observed only 5.4% of oocytes with a broken bivalent, compared with 27.0% at 6h00 after NEBD (Fig. 6 B). At this stage, oocytes already have individualized chromosomes easier to count than at 6h00 after NEBD, when they are aligned on the metaphase plate (Fig. S2 A and Video 7). The percentage of oocytes that presented a broken chromosome increased significantly as meiosis I progresses, as the spindle assembles and k-fibers are formed (Dumont et al., 2007; Schuh and Ellenberg, 2007; Breuer et al., 2010; Bennabi et al., 2016).
Because MT density is increased in Plk4 OE, maybe owing to the increase in the surface of nucleation from fragmented aMTOCs (Figs. 1 E and 2 B), we tested whether lowering MT density would impact the extent of bivalent breakage. To achieve this, Plk4 OE oocytes from line A were incubated with very low doses of nocodazole (ranging from 50 to 100 nM). As previously shown, such doses impact on spindle MT density (Fig. 6, C and D). The treatment with 50 nM nocodazole had a tendency to reduce the percentage of breakage events normally observed 6 h after NEBD (from 23.3% in controls down to 14.0% with 50 nM nocodazole, P = 0.172) while higher doses (100 nM) significantly reduced breakage events (from 23.3% in controls down to 6.1% with 100 nM nocodazole, P = 0.0068; Fig. 6 E and Video 8). This suggested that bivalent breakage was a consequence of increased MT density. Altogether, our results argue for a positive correlation between aMTOC fragmentation, earlier spindle bipolarization due to increased density of MTs, and DNA breakage.
Plk4 OE oocytes progress less efficiently throughout meiosis I
We then tested if the presence of a broken bivalent would affect cell cycle progression of Plk4 OE oocytes. We live-imaged Plk4 OE oocytes throughout meiosis I. These oocytes were delayed by almost 2 h in PBE timing compared with controls (Plk4 OE, 10h50 ± 2h00; controls, 9h05 ± 1h36; Fig. 7 A). The delay was most probably due to spindle assembly checkpoint (SAC) activation, since it could be rescued using the Mps1 inhibitor, reversine, as previously described (Hached et al., 2011; Kolano et al., 2012). In the presence of reversine, Plk4 OE underwent PBE with kinetics and efficiency comparable to controls (Fig. 7 A). However, the maintenance of an active SAC was not due to the presence of a broken bivalent. Indeed, by following the chromosomes using histone-GFP in live cells, we could detect the presence of a small broken fragment, yet anaphase I occurred on time at 9h30 (in seven oocytes with a fragmented chromosome, four matured on time; Fig. 7 B). This observation is fully consistent with previous work showing that drug-induced breakage of one bivalent does not induce SAC activation (Collins et al., 2015). Because we do not observe any delay in anaphase I timing in mCherry-Plk4 cRNA-injected oocytes (not depicted), which display higher Plk4 levels (Fig. 2, C and D), we conclude that it is not the presence of an excess active kinase but rather global perturbation of spindle assembly that delays meiosis I progression. As a consequence of this failure to arrest, some mature eggs will be produced with structural anomalies, as observed at anaphase I (Fig. 7 C).
Fragmentation of aMTOCs at NEBD is normally concomitant with a burst in local MT assembly around chromosomes (Dumont et al., 2007; Schuh and Ellenberg, 2007). Here we found that precocious fragmentation of aMTOCs in conditions of Plk4 overexpression was associated with increased MT polymerization at NEBD. We showed that fragmentation of aMTOCs significantly increases their surface area, and so propose that augmentation of their capacity to polymerize MT is related to this surface area increase, providing larger sites for γ-TuRC recruitment. Previous work has shown that MT density controls the timing of first meiotic spindle bipolarization (Dumont et al., 2007; Breuer et al., 2010; Kolano et al., 2012). Consistent with this, Plk4 OE oocytes displayed accelerated rates of spindle bipolarization. In mouse oocytes, the aMTOC morphology appears adapted to mediate the nucleation of an appropriate amount of MT at a given time, and we show here that alterations from this timely regulation have deleterious consequences on meiosis outcome.
We observe that spindles assembled with higher MT density at NEBD correlated with the presence of three different broken chromosomes (chromosome 9 and probably also 12 and 15) in different regions, near centromeres or close to telomeres (Fig. S3 C). The capacity of spindle MT to exert sufficient tension on chromosomes resulting in their breakage is still a matter of debate (Ganem and Pellman, 2012). However, here we provide evidence that this can happen in meiosis in regions where a Cre-mediated recombination event may have generated a region of fragility. The mechanism at the origin of this fragility remains elusive, and we cannot exclude that Plk4 overexpression during oocyte growth may have favored a fragility that is revealed only after meiosis resumption. However, it is rather unlikely that the break occurred before meiosis resumption, since Plk4 OE oocytes at all stages of their growth from both transgenic lines displayed overexpression of the mCherry-Plk4 transgene (data not shown). These data argue that Cre-mediated recombination and repair occurred in all oocytes, at the beginning of their growth phase, when the ZP3-Cre turns on (Lewandoski et al., 1997) to allow Plk4 transgene expression. Furthermore, the percentage of oocytes presenting one broken bivalent increased significantly (5×) from 4h00 to 6h00 after NEBD (Fig. 6 B). Our data thus suggest that most breaks are revealed while the meiosis I spindle is being assembled (Fig. S5). Meiosis I spindle assembly in both mouse and human oocytes is progressive, starting with the setup of bipolarity, then K-fiber assembly around NEBD + 5h00, and ending with spindle poles assembly (Dumont et al., 2007; Schuh and Ellenberg, 2007; Breuer et al., 2010; Kitajima et al., 2011; Holubcová et al., 2015; Bennabi et al., 2016). Even though forces during the different critical steps of meiosis I spindle assembly have never been measured directly, we can speculate that these forces increase as more bundles are made and K-fibers become progressively solidly anchored into the two spindle poles. This could potentially explain the increase in breakage events observed in our model.
In mitosis, the consequences of Plk4 overexpression have been extensively studied in various model systems (Habedanck et al., 2005; Rodrigues-Martins et al., 2007; Marthiens et al., 2013), but with no reported evidence for Plk4-induced chromosome breaks. Furthermore, whole-genome sequencing of Plk4 OE cells from several mouse tissues have not identified acentric chromosomes (Serçin et al., 2016; Levine et al., 2017). One main difference between previous studies and ours is that in mitotic spindles, forces are mostly exerted on kinetochores, whereas in meiosis, spindle forces exerted on kinetochores are transmitted to chiasmata and to chromosome arms. This suggests that the specific architecture of meiosis I chromosomes, bivalents held by chiasmata, might be more susceptible to break under potentially increased spindle forces, in condition of fragility (Figs. 4 J and 5 F). Yet, another important difference between meiosis and mitosis, is the extensive duration of meiosis I in mouse (8 to 9 h) and human (up to 17 h) oocytes (Holubcová et al., 2015; Bennabi et al., 2016) compared with mitosis (2 h). Indeed, it is known that weak forces applied repetitively can efficiently break DNA in vitro (Liang et al., 2013). The extended length of meiosis I of mammalian oocytes might favor the occurrence of such breaks.
Importantly, the presence of these breaks, in particular those producing acentric chromatin fragments not retained in the spindle (in Plk4 OE from line B and in Plk4 line A × MyoX), will not be detected by the SAC, as it functions through detecting MT occupancy at the kinetochore. Hence kinetochore-free chromatin is not recognized by any component of this checkpoint pathway. The other type of break we describe in line A will not be sufficient to elicit a robust SAC response (our results and Collins et al., 2015). Both types will lead to a class of chromosome abnormalities in the resulting egg that have not been observed during meiosis I before without prior treatment with DNA-damaging agents. Yet such types of structural anomalies, where terminal small pieces of chromosomes are missing, are quite frequent, from 1/20,000 to 1/50,000 for Wolf–Hirschhorn syndrome (Ho et al., 2016). It would be interesting to investigate the status of Plk4 activity in Wolf–Hirschhorn syndrome, as affected children are precisely characterized by mutations in the Plk4 gene (Martin et al., 2014). Even though we describe here a very artificial system, our work might shed new light on the etiology of human diseases associated with terminal deletions.
Materials and methods
Mouse strains and genotyping
Zp3-Cre [C57BL/6-Tg(Zp3-cre)93Knw/J] breeding pairs were obtained from the Jackson Laboratory. mCherry-Plk4flox/wt mice were generated by random insertion of a pCAG-loxCATlox-mCherryPlk4SV40pA construct in the genome of C57BL/6N mice (Marthiens et al., 2013). Two transgenic lines were amplified from two different founder lines with two different insertion sites and different transgene copy number and are referred to, for simplicity, as line A and B. Line A corresponds to the F3 line in Marthiens et al. (2013). Unless specified, most experiments were performed on line A, which presented a higher extent of aMTOC fragmentation and DNA breakage. For detection of Cre-mediated multiple stop codon excision, genotyping was performed using the following primers: for the Cre, 5′-GCGGTCTGGCAGTAAAAACTATC-3′ and 5′-GTGAAACAGCATTGCTGTCACTT-3′; and for the mCherry, 5′-CGCCACCATGGTGAGCAAGGG-3′ and 5′-CTCGTCCATGCCGCCGGTGG-3′. Control oocytes came from either Plk4flox/wt; Cre− or Plk4wt/wt; Cre+ female mice, while Plk4 OE oocytes came from Plk4flox/wt; Cre+ female mice. In Fig. 5, we also used oocytes coming from a cross of the Plk4 line Aflox/wt; Cre+ with the MyoXflox/wt; Cre− mice. The MyoXflox/wt; Cre− have been constructed by Genoway by insertion of two Lox P sites between exons 26 and 30 of the Myosin X endogenous locus (Fig. S3 B).
Transgene integration sites identification
Target locus amplification and next-generation sequencing, made by Cergentis, were used to determine the transgenic single-stranded oligodeoxynucleotide integration sites as previously described (de Vree et al., 2014).
Oocyte collection and culture
Oocytes were collected from 8- to 12-wk-old female mice into M2 + BSA medium supplemented with 1 µM milrinone (Reis et al., 2006) to ensure a block in prophase I. Resumption of meiosis was triggered by culturing oocytes in milrinone-free medium. All drugs were stored in DMSO at −20°C and diluted in M2 + BSA.
Plasmid construction and in vitro transcription of cRNA
We used previously described constructs, pRN3-H2B-RFP, pRN3-EB3-GFP (Breuer et al., 2010), and pMDL-Mad2-YFP (Lane et al., 2012). Histone2B-GFP (a gift from G. Halet, Institut de Génétique et Développement de Rennes, Rennes, France) was subcloned into the pRN3 vector. The mCherry-Plk4 construct used to produce the transgenic lines (Marthiens et al., 2013) was subcloned into the pRN3 vector. All cRNAs were synthesized using the T3 mMessage mMachine kit (Ambion) and resuspended in RNase-free water as previously described (Verlhac et al., 2000).
Injection of in vitro–transcribed cRNAs into the cytoplasm of prophase I–arrested oocytes was performed using an Eppendorf Femtojet microinjector as described (Verlhac et al., 2000) and the oocytes were further kept for 1–3 h in milrinone to allow expression of fusion proteins. Oocytes were then released from their prophase I arrest by transferring and washing into milrinone-free M2 medium.
Both nocodazole and reversine were added at NEBD. Oocytes were collected in M2 medium + milrinone and then washed in milrinone-free M2 medium containing either 50 or 100 nM nocodazole or 100 nM reversine as in Hached et al. (2011) and Kolano et al. (2012). Oocytes were kept in nocodazole until fixation at NEBD + 6h00. Control oocytes were treated with DMSO at the same dilution as the drugs.
Spinning disk videos were acquired using a Plan APO 40×/1.25-NA objective on a Leica DMI6000B microscope, enclosed in a thermostatic chamber set at 37°C (Life Imaging Services) and equipped with a CoolSnap HQ2/CCD-camera (Princeton Instruments) or EMCCD camera (Evolve) coupled to a Sutter filter wheel (Roper Scientific) and a Yokogawa CSU-X1-M1 confocal scanner. MetaMorph software (Universal Imaging) was used to collect the data.
For MT observations, oocytes were fixed using glutaraldehyde as previously described (Terret et al., 2003). For Plk4 endogenous labeling, oocytes were fixed using ice-cold (−20°C) methanol for 10 min. For CREST staining, oocytes were fixed at 30°C for 30 min in PHEM (Pipes, Hepes, EGTA, and MgCl2) buffer containing 2% formaldehyde and 0.05% Triton X-100, and were then permeabilized for 10 min in PBS containing 0.05% Triton X-100. Oocytes were extensively washed with PBS buffer between solutions. Oocytes were incubated at 4°C overnight in PBS supplemented with Tween-20 before primary antibody incubation (CREST, Abcam, 1:50). We used a rat monoclonal antibody against tyrosinated α-tubulin (YL1/2, Abcam) at 1:200 and mouse monoclonal anti-pericentrin (BD) at 1:400. For Plk4 labeling, a rabbit anti-Plk4 antibody was used at 1:500 (gift of A. Holland, John Hopkins School of Medicine, Baltimore, MD). After primary antibody incubation and subsequent washes, oocytes were labeled with the corresponding secondary antibodies (Jackson ImmunoResearch Laboratories). Coverslips were mounted in Prolong Gold with DAPI (Life Technologies). SIM samples were treated with lower concentrations of antibodies to avoid signal saturation and mounted in Prolong Gold without DAPI.
Image acquisition of fixed oocytes was performed on the SP5/AOBS confocal microscope equipped with a Plan APO 63×/1.4-NA objective or on the Spinning disc used for live imaging, and deconvolution was applied using Huygens software (SVI).
SIM acquisitions were performed on workstations of the Nikon Imaging Centre of the Curie Institute. Acquisitions were performed in 3D SIM mode with an n-SIM Nikon microscope before image reconstruction using NIS-Elements software based on Schermelleh et al. (2008). The system is equipped with an APO TIRF100 1.49-NA oil immersion and an EMCCD DU-897 Andor camera.
Oocytes were cultured in M2 medium and fixed once in 1× PBS, 1% PFA, and 1 mg/ml polyvinylpyrrolidone (PVP), for 1 min at RT. Cells were permeabilized at RT for 1 min in 1× PBS, 0.4% Triton X-100, and 0.5% PFA. Cells were further fixed for 10 min at RT in 1× PBS, 4% PFA, 1 mg/ml PVP, and 0.05% Triton X-100. Cells were then permeabilized in 1× PBS, 0.5% Triton X-100, 1 mg/ml PVP, and 0.02% RNase for 1 h at 37°C. Samples were preincubated in 1 µg/ml Cot-1 DNA, 10% dextran sulfate, 2× SSC, 0.5 mM EDTA, 0.05% Triton X-100, 1 mg/ml PVP, 0.5 mg/ml BSA, and 4 M urea, pH 7.3, overnight at 37°C, then denatured at 83°C for 10 min and prehybridized for 1 h at 37°C. Here urea has been used instead of formamide (Sinigaglia et al., 2018). After overnight hybridization with a chromosome paint against chromosome 9 (Metasystem, see supplier’s protocol) at 37°C, oocytes were washed at 45° twice for 15 min in 0.2× SSC, 0.05% Triton X-100, and 1 mg/ml PVP and mounted in Prolong Gold with DAPI (Life Technologies). Denaturation and hybridization steps were performed in glass-bottom Petri dishes. Other steps were performed in 4% agarose and 9 mg/ml NaCl-coated Petri dishes.
3D analysis of aMTOCs and chromosome fragment morphology was performed using Imaris (Bitplane). Thresholding of objects was done in a semiautomatic way. First, objects were thresholded automatically via Imaris, then we checked the accuracy of the threshold manually. Comparison of mCherry-Plk4 overexpression levels in cRNA-injected oocytes versus transgenic oocytes was performed in Metamorph by measuring the absolute intensity of fluorescence on summed Z-projections (70 Z, spaced every 0.5 µm, arbitrary units). Background values were measured within a region of interest outside the cell and were subtracted before quantification. Plk4 and pericentrin levels in Plk4OE oocytes over endogenous protein in control oocytes were quantified on summed Z-projections (70 Z, spaced every 0.5 µm, arbitrary units). SIM acquisitions were treated in Fiji after reconstruction and realignment then rendered in Imaris 3D viewer after calibration. The total and local MT fluorescence signal intensities were measured in oocytes expressing EB3-GFP on summed Z-stack projections (six Z planes spaced every 4 µm). Timing of bipolarization was assessed in oocytes expressing EB3-GFP imaged at 10-min intervals.
Statistical analyses were performed using Graphpad Prism 5.0 software. ns, nonsignificant (P > 0.05); *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.
Online supplemental material
Fig. S1 shows quantification of the pericentrin and Plk4 signals in the whole oocyte and SIM acquisition of aMTOCs in Plk4 OE oocytes. Fig. S2 shows that Plk4OE oocytes present accelerated rates of spindle bipolarization as observed in fixed oocytes. Fig. S3 illustrates the different genetic insertions used in the study. Fig. S4 shows that the amount of mCherry-Plk4 on aMTOCs is comparable in line A and B in prophase I oocytes. Fig. S5 shows a scheme recapitulating our model. Video 1 shows that spindle bipolarization is accelerated in oocytes overexpressing Plk4. Video 2 shows a chromosome fragment moving back and forth across the metaphase plate in Plk4OE oocyte. Video 3 shows 20 bivalents observed in control oocytes at NEBD +6h00. Video 4 shows 21 pieces of DNA counted in Plk4 OE line A oocytes at NEBD +6h00. Video 5 shows 21 pieces of DNA counted in Plk4 OE line B oocytes at NEBD +6h00. Video 6 shows that a new form of DNA break can be observed in Plk4 line Aflox/wt; Cre+; MyoXflox/wt oocytes at NEBD +6h00. Video 7 shows that bivalents can be easily counted in Plk4 OE oocytes observed at NEBD +4h00. Video 8 shows that bivalents can be easily counted in Plk4 OE oocytes treated with 50 nM nocodazole and observed at NEBD +6h00.
We thank Ludovic Dumont and Anchi Chang, masters students, for their help at the beginning of the project; Carole Pennetier and Manon Chartier at the Curie Institute for genotyping; Adel Al Jord from our laboratory for critical reading of the manuscript; Lucie Sengmanivong from the Nikon Imaging Centre at Institut Curie-Centre National de la Recherche Scientifique (Paris, France); Guillaume Halet for the gift of the HistoneH2B-GFP plasmid; Andrew Holland for the gift of the anti-Plk4 antibody; and Katia Ancelin and Edith Heard at the Curie Institute for the DNA-FISH protocol.
M. Łuksza was supported by a Fondation ARC pour la Recherche sur le Cancer fellowship. This work was supported by a grant from the Ligue Contre le Cancer (EL/2012/LNCC/MHV), the Agence Nationale de la Recherche (ANR-DIVACEN to M.-H. Verlhac and R. Basto, 14-CE11), a Fondation pour la Recherche Médicale Label (DEQ20150331758 to M.-H. Verlhac), and an Institut National du Cancer grant (PLBIO 2016-270-TRAN). This work has received support from the Fondation Bettencourt Schueller, support under the program “Investissements d’Avenir” launched by the French government and implemented by the Agence Nationale de la Recherche (ANR-10-LABX-54 MEMO LIFE, ANR-11-IDEX-0001-02 PSL* Research University). The work was also supported by Biotechnology and Biological Sciences Research Council funding to K.T. Jones (Mechanisms of DNA damage and repair in mature oocytes, BBSRC, BB/L006006/1) and American Institute for Cancer Research (now Worldwide Cancer Research; 13-0170) to V. Marthiens and R. Basto.
The authors declare no competing financial interests.
Author contributions: M. Manil-Ségalen and M.-H. Verlhac conceived and supervised the project. M. Manil-Ségalen performed most experiments. M. Łuksza and J. Kanaan participated in some experiments. M. Manil-Ségalen and M.-H. Verlhac analyzed most experiments. S.I.R. Lane and K.T. Jones performed the mad2-YFP experiment. V. Marthiens, M.-E. Terret, and R. Basto provided scientific input, mouse strains, and reagents. M. Manil-Ségalen and M.-H. Verlhac wrote the manuscript, which was seen and corrected by all authors.
S.I.R. Lane's present address is Institute for Life Sciences and Faculty of Chemistry, University of Southampton, Southampton, UK.