Spindle checkpoint can secure additional cheating time for selfish expanded centromeres

Summary

Expanded centromeric satellite repeats can violate Mendel’s Law of Segregation by preferentially segregating to the egg. In mice, these selfish centromeres enrich microtubule-destabilizers at pericentromeres to detach from the spindle and flip towards the egg side of the meiotic spindle, thereby achieving preferential segregation. However, despite consistent enrichment of destabilizers upon centromere expansion, such enrichment alone is insufficient to drive preferential retention of expanded centromeres, suggesting a missing component in understanding their non-Mendelian segregation. Here we propose that prolonged spindle checkpoint activation is crucial for expanded centromeres to cheat the segregation process by providing sufficient time for them to flip towards the egg side. By experimentally manipulating kinetochore size in a species-specific manner, we found that assembling larger kinetochores triggers robust spindle checkpoint activation, leading to anaphase delay and preferential retention of expanded centromeres in the egg. Comparisons across multiple hybrid mouse models revealed that centromeric satellite asymmetry does not consistently lead to kinetochore asymmetry and anaphase delay, explaining why satellite asymmetry does not always result in preferential retention of larger centromeres. Altogether, this work highlights the significance of checkpoint activation in exploiting the inherent asymmetry in female meiosis and the distinct responses of kinetochore proteins and microtubule-destabilizers to centromere expansion.

Graphical Abstract

eTOC Blurb:

Expanded centromere DNA repeats violate Mendel’s Law of Segregation by preferentially segregating to the egg in animals and plants. Using oocytes collected from hybrid mice, Walton et al. show that such selfish centromeres exploit the spindle checkpoint system to increase their own transmission rates.

Introduction

Selfish genetic elements can manipulate meiosis to preferentially transmit themselves to the next generation^1–5. This genetic cheating during meiosis, meiotic drive, has significant impacts on genetics and reproduction as the cheating impacts transmission ratios and manipulates the gametogenesis process, often leading to fertility issues⁶. Female meiosis provides a unique opportunity for genetic elements to cheat because of its inherent asymmetry: only chromosomes that segregate to the egg transmit to the next generation⁷. Centromeres direct chromosome segregation and therefore have an ideal opportunity to alter the segregation pattern. Indeed, expanded centromeric satellites bias their transmission in mice and plants (centromere drive)^8–12. Expanded mouse centromeric satellite repeats preferentially remain in the egg by exploiting asymmetry within the oocyte spindle^13,14 (Figure 1A). The mouse oocyte spindle initially assembles at the oocyte center and subsequently migrates towards the cortex in metaphase I. Spindle asymmetry establishes upon spindle migration with the cortical side (destined for the polar body) enriched with tyrosinated microtubules (Tyr MTs) and the interior side (destined for the egg) enriched with detyrosinated (dTyr) MTs. Selfish centromeres are initially oriented randomly on the metaphase I spindle, but the spindle asymmetry provides a spatial cue for selfish centromeres to flip towards the egg side (Figure 1A). Selfish centromeres achieve this flipping by enriching MT-destabilizing activity (i.e., MCAK, a member of the kinesin-13 family and the Chromosomal Passenger Complex (CPC))^15–18, which allows the centromere to detach from the spindle, providing an opportunity for the chromosome to re-orient. When the selfish centromere is attached to the “wrong” side (i.e., the cortical side), the synergistic effect of unstable Tyr MTs and the high MT-destabilizing activity at the selfish centromere preferentially destabilizes this attachment, leading to detachment and flipping^14,19–21. Upon facing the more stable dTyr MTs on the egg side, they are less likely to flip back, providing directionality to the flipping. Biased destabilization of spindle attachment followed by directional flipping is considered to be the source of preferential retention of selfish expanded centromeres in the egg^1,8,11,12.

Figure 1. Asymmetry in microtubule-destabilizer is required but not sufficient for centromere drive.

(A) Centromere drive in mouse oocytes occurs through biased chromosome flipping. The mouse oocyte meiosis I spindle initially assembles at the oocyte center^67–69. Chromosomes undergo flipping on the meiotic spindle at this stage without any spatial cues^19,70. Upon spindle migration towards the cortex, spindle asymmetry in tyrosination is established, which provides a spatial cue for selfish centromeres (larger circles) to preferentially detach from the cortical side of the spindle (green lines) and flip towards the interior side (yellow lines) to segregate to the egg. (B) Hybrids crossed between a M. m. domesticus strain (C57BL/6J) and a M. spretus strain (SPR2) or a M. m. musculus strain (PWD/PhJ) were utilized as experimental systems to investigate centromere drive. SPR2 and PWD/PhJ have significantly larger centromeres compared to C57BL/6J^24,38,71,72 (also see Figure 1D). (C) dom x spr oocytes were matured to meiosis II, treated with Monastrol to individualize chromosomes, and fixed and stained for TOP2A (a marker for dom centromeres⁴⁰) and ACA to label all centromeres. Graph shows the percentage of spr chromosomes among total chromosomes in each meiosis II egg to examine preferential segregation; n = 79 cells; each dot represents one egg cell; One sample t test was used to analyze the deviation from 50% (D) dom x mus oocytes expressing dCas9-EGFP with gRNA targeting centromeric Minor satellite were fixed at metaphase I and stained for HEC1 (kinetochore) and MCAK (MT-destabilizer). Graph shows centromere signal ratios, calculated as the mus centromere divided by the dom centromere signal for each bivalent; each dot represents one meiotic bivalent; n = 79 and 81 bivalents for HEC1 and MCAK, respectively; Wilcoxon signed rank test was used to analyze deviations from 1; “d” and “m” indicate dom and mus centromeres, respectively. (E) dom x mus oocytes expressing dCas9-EGFP with gRNA targeting centromeric Minor satellite were matured to metaphase II and fixed and stained for HEC1 and MCAK. Chromosomes with higher dCas9-EGFP and MCAK levels at sister centromeres were identified as mus chromosomes to count the number of mus and dom chromosomes. Graph shows the proportion of mus chromosomes among total chromosomes in each meiosis II egg; n = 38 cells; each dot represents one egg cell; One sample t test was used to analyze the deviation from 50%. The images are maximum projections showing all the chromosomes (left) and optical sections to show individual chromosomes (right); red lines, median. (F) Comparison of three mouse hybrid systems to study centromere drive. Centromeric satellite asymmetry consistently leads to MT-destabilizer asymmetry but not kinetochore asymmetry and centromere drive. A hybrid system with kinetochore asymmetry exhibits anaphase delay, implying a molecular link between these two features.

Results

High microtubule-destabilizing activity is not sufficient to retain expanded centromeres.

Although MT-destabilizing activity is required to confer selfishness to centromeres, it may not be sufficient to preferentially retain larger centromeres in the egg. For example, in oocytes from hybrid mice between Mus musculus domesticus and Mus spretus (hereafter dom x spr), the larger spretus centromeres enrich higher MT-destabilizing activity but do not show biased orientation towards the egg side before anaphase I onset unless anaphase onset was experimentally delayed¹⁹ (Figure 1B). Rather, there is a slight trend for spr centromeres to face towards the cortical side¹⁹. To confirm this observation, we analyzed the segregation pattern by analyzing meiosis II eggs (Figure 1C). If eggs are euploid, we can count the number of spr chromosomes in each egg to test if the distribution is deviated from the expected distribution assuming random segregation. However, total chromosome numbers vary among dom x spr eggs due to mis-segregation in meiosis I^22,23. Therefore, we quantified the proportion of spretus chromosomes among the total in each egg to examine if it deviates from 50% (Figure 1C). Consistent with the previous chromosome orientation analysis, we found that spr chromosomes were not preferentially retained in the egg (45.48%) (Figure 1C).

To further confirm that destabilizer asymmetry is not sufficient for centromere drive, we crossed the same Mus musculus domesticus strain with a Mus musculus musculus strain (hereafter dom x mus) where the centromeric satellite is significantly expanded on musculus chromosomes compared to domesticus^24,25 (Figures 1B and 1D). Similar to dom x spr oocytes, dom x mus oocytes showed significant asymmetry in MT-destabilizers across the meiotic bivalent without any preferential retention of larger centromeres (i.e., mus centromeres) in the egg (48.17%) (Figures 1D and 1E). These results indicate a missing component to drive non-Mendelian segregation of larger centromeres to the egg in these systems.

Possible models for how enlarged kinetochores contribute to centromere drive.

By comparing three mouse hybrid models with centromeric satellite asymmetry, we noticed that the larger centromere consistently enriches a higher MT-destabilizer level but does not always assemble a larger kinetochore compared to the paired centromere: dom x spr and dom x mus oocytes don’t show kinetochore asymmetry, whereas oocytes from the hybrid crossing the same domesticus strain with another domesticus strain with smaller centromeres (hereafter dom x dom (s)) do^13,19 (Figure 1F). Importantly, dom x dom (s) oocytes preferentially retain larger centromeres in the egg¹³, raising a possibility that assembling larger kinetochores is a crucial requirement for their biased transmission.

If larger kinetochores contribute to centromere drive, what could be the underlying mechanism? Here we propose two possible models on how kinetochore size impacts centromere drive. First, it is intuitive to speculate that the larger kinetochore on the selfish centromere impacts its MT attachment, modulating chromosome dynamics on the spindle to alter the segregation pattern^26–28. In this case, we expect to see differences in kinetochore-MT attachments and chromosome distribution on the spindle depending on the kinetochore size of the larger centromere. Second, the larger kinetochore may result in anaphase delay, providing additional time for selfish centromeres to flip towards the egg side. Anaphase delay has been observed in dom x dom (s) oocytes where kinetochore asymmetry and centromere drive are detected^12,19. However, it has not been experimentally tested (1) if anaphase delay is directly caused by the increased kinetochore size at the larger centromere and (2) what the molecular basis underlying the anaphase delay is. These two possible models are not mutually exclusive since modulating MT attachments can cause anaphase delay^29–33.

Enlarged kinetochores do not substantially impact spindle interactions.

To test these hypotheses, we employed a bottom-up approach to experimentally increase the kinetochore size of larger spr centromeres in dom x spr hybrid oocytes where destabilizers are asymmetric but kinetochores are symmetric (Figure 2A). We targeted a centromere protein, CENP-T, which recruits outer kinetochore proteins^34–37 (e.g., HEC1), using a TALE construct that specifically targets centromeric Minor satellites on spr chromosomes³⁸. The CENP-T targeting strategy increased the kinetochore size on spr centromeres without impacting MT-destabilizer asymmetry (Figure 2A). Throughout the study, we removed oocytes that are over-expressing CENP-T-TALE (top 10% of the oocytes expressing CENP-T-TALE) based on its fluorescent signals to avoid analyzing overly large kinetochores that compromise their basic kinetochore functions (see below).

Figure 2. Kinetochore enlargement does not significantly affect spindle interactions.

(A) Strategy to enlarge kinetochores by targeting CENP-T to spr centromeres in dom x spr oocytes (top left). dom x spr metaphase I oocytes expressing CENP-T-TALE were used to prepare chromosome spreads and stained for HEC1 and MCAK. Graph shows centromere signal ratios, calculated as the spr centromere divided by the dom centromere signal for each bivalent; each dot represents one meiotic bivalent; MCAK, n = 60 and 44 bivalents for control and CENP-T-TALE, respectively; HEC1, n = 99 and 84 bivalents for control and CENP-T-TALE, respectively; unpaired Mann-Whitney test was used to analyze statistical significance; “d” and “s” indicate dom and spr centromeres, respectively. (B) dom x spr oocytes were analyzed for cold-stable microtubules at metaphase I. ACA (anti-centromere antibodies) signals were used as a kinetochore marker. Enlarged images are optical slices showing individual kinetochore-MT attachments (i.e., end-on attachment, lateral attachment, or unattached). Graph shows the percentage of kinetochores in each attachment category after pooling two independent experiments (n = 91, 86, and 26 chromosomes were analyzed for unexpressed control, +CENP-T-TALE, and +CENP-T-TALE overexpressed, respectively); Chi-square test for independence was used to examine statistical significance. (C) dom x dom oocytes and dom x spr oocytes expressing either control TALE or CENP-T-TALE were imaged live during metaphase I in the presence of SPY650-DNA to visualize chromosomes. Graph shows the metaphase plate width at 8 hr 10 min after the nuclear envelope breakdown (NEBD); n = 24, 26, and 31 cells for dom x dom, dom x spr +TALE, and dom x spr +CENP-T-TALE, respectively; each dot represents an oocyte cell; unpaired Mann-Whitney test was used to analyze statistical significance; white circles, oocyte cortex; white dashed lines, spindle area; left-right arrows, metaphase plate width; the images are maximum projections; red lines, median. See also Figure S1.

Using this enlarged kinetochore system, we first tested if kinetochore size directly impacts spindle interactions to drive preferential segregation of larger centromeres to the egg. To analyze MT interactions, we visualized kinetochore-MT attachments in oocytes with enlarged kinetochores and found that increased kinetochore size does not significantly impact spindle interactions (Figure 2B). Kinetochores can be unattached, attached laterally, or attached end-on to spindle microtubules. While end-on attachment leads to proper chromosome segregation, unattached and laterally attached kinetochores can result in mis-segregation³⁹. We previously found that spr kinetochores have a higher end-on attachment rate compared to dom kinetochores⁴⁰, and this trend was not significantly impacted by the increased kinetochore size (Figure 2B, unexpressed control vs +CENP-T-TALE). In contrast, CENP-T-TALE overexpression led to kinetochore deformation and significant mis-attachment (predominantly lateral attachment), compromising basic kinetochore functions (Figure 2B, +CENP-T-TALE vs +CENP-T-TALE overexpressed). These results justify removing CENP-T-TALE overexpressing oocytes from the analyses and demonstrate that the rest of the CENP-T-TALE expressing oocytes have similar attachment profile with control unexpressed oocytes.

As a functional readout of spindle interactions, we performed time-lapse imaging of metaphase I oocytes. The time-lapse imaging confirmed the previous finding that dom x spr hybrid oocytes in general show basal off-centeredness of chromosomes compared to pure species dom x dom oocytes, which have symmetric centromeres^19,23 (Figure 2C). It was previously shown that centromere asymmetry leads to chromosome off-centeredness on the spindle and that larger off-centeredness is associated with stronger centromere drive^19,28,41. When we compared control TALE and CENP-T-TALE expressing oocytes, we found that kinetochore enlargement (+CENP-T-TALE) did not further increase the off-centeredness on top of the basal off-centeredness in hybrid oocytes (+TALE) (Figure 2C). These results suggest that kinetochore asymmetry does not substantially impact spindle interactions and chromosome dynamics in this hybrid oocyte system. Importantly, the time-lapse imaging also confirmed that control TALE and CENP-T-TALE expressing dom x spr oocytes show similar chromosome flipping rates (Figures 1A and S1), which are crucial to orient larger centromeres towards the egg side of the spindle.

Enlarged kinetochores drive anaphase delay.

Next, we tested the second hypothesis regarding kinetochore size and anaphase delay by measuring the duration of meiosis I division (Figure 3A). dom x spr oocytes typically enter anaphase I and extrude the first polar body 8 – 10 hours after nuclear envelope breakdown, similar to pure dom x dom oocytes²³ (Figure 3A unexpressed control and Figure S2A unexpressed control). Meiosis I duration was not affected by the control TALE expression, whereas the expression of CENP-T-TALE significantly delayed anaphase onset (Figure 3A, TALE and CENP-T-TALE). We found that the anaphase delay was due to the activation of the spindle checkpoint, a surveillance system that monitors kinetochore-MT attachments, since the delay was completely canceled by treating the oocytes with a chemical inhibitor, Reversine, which inhibits a spindle checkpoint kinase, MPS1^42,43 (Figure 3A, CENP-T-TALE Rev. and Figure S2A, Rev.). As an additional control experiment, we expressed CENP-T-TALE in pure spr x spr oocytes, which would enlarge both kinetochores on the bivalent, and confirmed that enlarged kinetochores delay anaphase onset without substantially affecting chromosome alignment (Figure S2C and S2D).

Figure 3. Enlarged kinetochores drive spindle checkpoint activation and preferential segregation of larger centromeres.

(A) dom x spr oocytes expressing either control TALE or CENP-T-TALE were matured for 20 hr to examine the timing of the polar body extrusion (a readout for anaphase I onset). Reversine was added to oocyte culture upon NEBD to cancel the spindle checkpoint. Timepoints with the orange rectangle indicate the timing of the polar body extrusion. Graph shows the proportion of oocytes that have extruded the polar body; n = 70, 18, 35, 21, and 33 oocyte cells for control, TALE, CENP-T-TALE, control Rev., and CENP-T-TALE Rev., respectively. (B) dom x spr prometaphase I oocytes expressing CENP-T-TALE were used to prepare chromosome spreads and stained for MAD1. Graph shows centromeric MAD1 signal intensities; each dot represents one centromere; n = 42, 42, 38, and 38 centromeres for control spr, control dom, +CENP-T-TALE spr, and +CENP-T-TALE dom, respectively; Kruskal-Wallis test was used to analyze statistical significance. (C, D) dom x spr oocytes expressing control TALE or CENP-T-TALE were matured to meiosis II with or without Reversine, treated with Monastrol to individualize chromosomes, and fixed and stained for TOP2A (a marker for dom centromeres⁴⁰). Graph in D shows the percentage of spr chromosomes in each meiosis II egg to examine preferential segregation; n = 48, 34, and 22 cells for TALE, +CENP-T-TALE, and +CENP-T-TALE + Reversine, respectively; each dot represents one egg cell; unpaired two-sided t test was used to analyze statistical significance. Table in D shows the quantification of the deviation from 50% using One sample t test; CENP-T-TALE expressing oocytes preferentially retained spr chromosomes, while the other two groups preferentially retained dom chromosomes. The images are maximum projections. (E) Model of centromere drive in mouse oocytes, highlighting how prolonged spindle checkpoint activation facilitates preferential retention of selfish larger centromeres. See also Figures S2, S3.

Spindle checkpoint signals are emitted from the kinetochore. Unattached or mis-attached kinetochores enrich spindle checkpoint proteins such as MPS1 and the MAD1/MAD2 complex to inhibit the anaphase-promoting complex/cyclosome (APC/C), leading to metaphase arrest^29,44–47. We tested if the spindle checkpoint activation by the CENP-T-TALE expression results from the enlarged kinetochore recruiting more spindle checkpoint proteins. Consistent with this idea, enlarged kinetochores on spr centromeres enriched substantially higher MAD1 levels compared to dom centromeres on the same bivalent and compared to centromeres in dom x spr oocytes without the CENP-T-TALE expression (Figure 3B). We also analyzed MAD2 levels in hybrid oocytes that naturally show centromere drive (i.e., dom x dom (s) oocytes) and found that larger kinetochores enrich higher levels of MAD2 compared to smaller kinetochores (Figure S2B). Collectively, these results suggest that assembling larger kinetochores induces spindle checkpoint-mediated anaphase delay by enriching checkpoint proteins.

Assembling larger kinetochores and enriching destabilizers lead to centromere drive.

Anaphase delay would provide more time for chromosome flipping, leading to non-Mendelian segregation. Therefore, we tested if the prolonged spindle checkpoint activation by kinetochore enlargement facilitates the retention of larger centromeres in the egg. We matured dom x spr meiosis I oocytes with enlarged spr kinetochores to meiosis II to examine the result of meiosis I segregation (Figures 3C, 3D, and S2E). We found that the CENP-T-TALE expression leads to larger spr centromeres more often segregating to the egg (55.28%) in contrast to control TALE expressing oocytes where 44.85% of spr centromeres remain in the egg. To directly test the contribution of the spindle checkpoint on this segregation pattern, we treated CENP-T-TALE expressing oocytes with Reversine. Canceling the checkpoint by Reversine also reversed the segregation pattern (Figures 3C and 3D), consistent with our model. If a significant number of chromosomes mis-segregate in meiosis I, the proportion of spr centromeres in meiosis II eggs may not reflect chromosome orientation at metaphase I. Therefore, we re-analyzed the dataset after removing eggs with significant aneuploidy (Figure S2F), which confirmed that the segregation pattern does not substantially change from the original analysis in Figure 3D. Together with the previous study¹⁹, these results suggest that the combination of destabilizing activity (chromosome flipping) and increased kinetochore size (spindle checkpoint activation) confers selfishness to expanded centromeric satellites in mice (Figure 3E).

Discussion

This study provides the first cell biological insights into the contribution of checkpoints on non-Mendelian transmission. Asymmetry in centromeric satellites is thought to drive non-Mendelian chromosome segregation in oocytes, but observations in previous and current studies suggest that satellite asymmetry does not consistently result in preferential segregation^9,10,19. Here we provide cell biological evidence that larger centromeres do not consistently assemble larger kinetochores, which are critical to drive prolonged spindle checkpoint activation to retain expanded centromeres (Figure 3E). The original centromere drive theory proposed that centromeric satellite expansion leads to more MT-attachment sites, for example by assembling a larger kinetochore, to preferentially segregate to the egg¹⁰. Our data suggest that kinetochores in our hybrid mouse system have an essential role in centromere drive by activating a strong spindle checkpoint in addition to serving as MT-attachment sites (Figure 3E). Investigating how selfish centromeres cheat in oocytes leads to a deeper understanding of the centromere itself. Our observations imply that kinetochore proteins and MT-destabilizers exhibit distinct responses to centromere evolution.

Little variations in kinetochore size upon centromere expansion.

The kinetochore location and size have been used as functional readouts of centromere evolution^28,48–51. However, kinetochore proteins usually do not fully occupy centromeric satellites^50,52, and our study shows that kinetochore size does not consistently respond to centromere expansion, at least in mouse oocytes. This observation is consistent with the idea that kinetochore proteins have undergone adaptive evolution not to follow centromere evolution to minimize functional differences between homologous centromeres and suppress centromere drive^10,53. Interestingly, the kinetochore size is asymmetric when one of the homologous centromeres is significantly small (i.e., dom (s) or the CHPO strain). At these extremely small mouse centromeres, kinetochore proteins fully occupy the centromeric satellite¹³, implying that centromere shrinkage can regulate the kinetochore size by limiting the spreading of kinetochore proteins outside the centromeric satellite. On the other hand, for centromeres that are larger than a certain size, kinetochore proteins may have evolved a mechanism not to scale their abundance with the satellite size, making kinetochores insensitive to centromere expansion and preventing centromere drive. Alternatively, kinetochore size may not scale with satellite size due to a fitness cost associated with overly large kinetochores and/or a lack of mechanisms to recruit additional kinetochore proteins.

Microtubule-destabilizer level as a functional readout of centromere evolution.

MT-destabilizers localize at the pericentromere, which is a chromatin region juxtaposed to the centromere^54–57. Our results show that MT-destabilizer levels scale with centromere size more constantly than kinetochore proteins (Figure 1), implying that centromeric satellites can regulate MT-destabilizer levels, consistent with our recent study using Peromyscus mice⁵⁸. Therefore, our findings suggest pericentromeric MT-destabilizer levels as a new functional readout for centromere expansion.

It remains largely unknown how centromeric Minor satellites (together with the pericentromeric Major satellites) regulate MT-destabilizer levels (Figure S3). Shugoshin serves as a scaffold for MT-destabilizers and is recruited to the pericentromere by directly binding the histone H2A-pT121 mark^55–57. This histone phosphorylation is catalyzed by BUB1 kinase, which localizes at the kinetochore⁵⁹. When Minor satellite asymmetry leads to kinetochore asymmetry as in dom x dom (s) oocytes, BUB1 localization is asymmetric, leading to asymmetries in H2A-pT121 and MT-destabilizers¹⁹. When Minor satellite asymmetry does not lead to kinetochore asymmetry, centromere geometry can be different between homologous centromeres, influencing the accessibility of kinetochore-localized BUB1 kinase to pericentromeric chromatin, thereby modulating H2A phosphorylation¹⁹. In dom x spr oocytes, dom centromeres are relatively more decondensed because of the presence of more Major satellites at those centromeres, which enrich less condensin II and cause chromosome de-condensation, spatially separating BUB1 from pericentromeric chromatin^19,40. However, in dom x mus oocytes, domesticus and musculus centromeres have similar Major satellite copy numbers²⁴, suggesting a distinct mechanism of larger Minor satellites to recruit more MT-destabilizers. Much as they say, ‘all roads lead to Rome’, increasing MT-destabilizing activity appears to be a general strategy for mouse centromeres to cheat, but expanded centromeres from different species have evolved distinct strategies to enrich the same activity.

Selfish centromeres exploit a surveillance system to cheat in female meiosis.

In addition to enriching high levels of MT-destabilizing activity, centromeres require additional time after spindle migration to perform directional flipping (Figure 3E). We provide evidence that larger kinetochores can induce prolonged activation of the spindle checkpoint to secure this additional “cheating” time. Our enlarged kinetochore approach enriched higher levels of MAD1 on the larger kinetochore with little impact on chromosome dynamics (Figures 2B, 2C, and 3B). Consistent with this observation, kinetochore MAD1 levels scale with kinetochore size in mitosis⁶⁰. As a future direction, it would be interesting to enlarge kinetochores by targeting other kinetochore proteins such as CENP-C, which has both common and distinct contributions to kinetochore assembly from CENP-T^34,35,61.

Vertebrate oocytes typically have a weak spindle checkpoint, which is counterintuitive because of the risk of producing aneuploid eggs^44,45,62. We speculate that this weak checkpoint could be a consequence of adaptive evolution, limiting the cheating time for selfish centromeres¹⁹. Continuous cycles of centromere drive and suppression would lead to the evolution of diverse cheating mechanisms. Indeed, another mouse hybrid model for centromere drive exhibits directional flipping before spindle migration, implying that selfish centromeres in this particular system do not depend on anaphase delay to flip towards the egg side⁶³. Investigating various hybrid animals and plants with centromere asymmetry would reveal if there is any specific biological process that is prone to exploitation by selfish centromeres^64–66.

In conclusion, this study revealed that selfish elements exploit checkpoint systems to increase their own transmission rates and provides a new conceptual framework to investigate how the genome can adaptively evolve to suppress genetic cheating to maintain Mendelian transmission.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Takashi Akera (takashi.akera@nih.gov).

Materials availability

Resources and reagents are available upon reasonable request.

Data and code availability

• Microscopy data have been deposited at FigShare and are publicly available as of the date of publication at 10.25444/nhlbi.29316011.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR Methods

Experimental model and study participant details

Mouse strains

Mouse strains were purchased from Envigo (NSA, stock# 033 corresponds to CF1, Mus musculus domesticus), Jackson Laboratory (C57BL/6J, stock# 000664, Mus musculus domesticus; ZALENDE/EiJ, stock #001392 corresponds to CHPO, Mus musculus domesticus; PWD/PhJ, stock# 004660, Mus musculus musculus), and RIKEN BioResource Research Center (SPR2, stock# RBRC00208, Mus spretus). Mus musculus domesticus C57BL/6J females were crossed to Mus spretus SPR2 males to generate the F1 hybrid (dom x spr), because the other direction is technically challenging to produce offspring. Reciprocal crosses were used when crossing Mus musculus domesticus C57BL/6J to Mus musculus musculus PWD/PhJ; however, the majority of F1 females used resulted from crosses where the sire was PWD/PhJ, as this direction is more efficient to produce F1 hybrid mice. Mice were housed in an animal facility at room temperature, 30–70% humidity, and with a ventilated rack system. Mice were exposed to a 12 hr light/dark cycle year-round. Individuals used in experiments could be sisters or unrelated and ranged in age from 6 weeks to 5 months. Mice were euthanized with CO₂ followed by cervical dislocation prior to dissection of the ovaries. All animal experiments were approved by the Animal Care and Use Committee (National Institutes of Health Animal Study Proposal#: H-0327) and were consistent with the National Institutes of Health guidelines.

Method details

Mouse oocyte collection and culture

All manipulations of mouse oocytes/eggs were performed with a mouth-operated plastic pipette with either a 75 or 100 μm diameter pipette tip (Cooper Surgical, Inc., Cat# MXL3–75 or MXL3–100, respectively). For oocyte collection, germinal vesicle (GV)-intact oocytes were collected from 6 week to 5 month-old female mice in M2 media (Sigma-Aldrich, Cat# M7167) supplemented with 5 μM milrinone (Sigma, Cat# 475840) to prevent meiotic resumption⁷³. Oocytes were then transferred to M16 media (Millipore, Cat# M7292), covered with paraffin oil (Nacalai, Cat# NC1506764), and incubated at 37°C in a humidified atmosphere of 5% CO₂ in air. To induce meiotic resumption, milrinone was washed out, and oocytes that did not undergo nuclear envelope breakdown (NEBD) within 1.5 h after the milrinone washout were removed from the culture. Oocytes were matured for variable times according to each assay. For analysis of prometaphase I and metaphase I stages, oocytes were matured for 3 and 7 hr, respectively. For analysis of metaphase II eggs, oocytes were matured for 16 hr. To cancel the spindle checkpoint, 0.5 μM Reversine (Sigma-Aldrich, Cat# R39-4-1MG) was added to oocyte culture upon NEBD. For the in situ chromosome counting assay, oocytes were matured for 13 hr after NEBD and subsequently treated with 100 μM Monastrol (Millipore, Cat# 475879) in the organ culture dish (Falcon, Cat# 353037) for 2 hr 15 min prior to the fixation⁷³. To label chromosomes during live-imaging, SPY650-DNA (1:1000 dilution, Spirochrome, Cat# SC501) was added to oocyte culture upon NEBD.

Oocyte microinjection

GV-intact mouse oocytes were microinjected with ~5 pl of cRNAs or proteins in M2 containing 5 μM milrinone, using a micromanipulator TransferMan 4r and FemtoJet 4i (Eppendorf). Following the microinjection, oocytes were maintained at prophase I in M16 supplemented with 5 μM milrinone overnight to allow protein expression. cRNAs used for microinjections were TALE (TALE construct that recognizes M. spretus Minor satellite repeats fused to mRuby2 at the C terminus³⁸; addgene, Cat# 47879) at 600 ng/ul, CENP-T-TALE (TALE construct that recognizes M. spretus Minor satellite repeats fused to mRuby2 and M. musculus CENP-T at the C terminus) at 650 ng/ul. cRNAs were synthesized using the T7 mMessage mMachine Kit (Ambion, Cat# AM1340) and purified using the MEGAclear Kit (ThermoFisher, Cat# AM1908). The dCas9-EGFP-gRNA complex that targets the Minor satellite sequence was assembled in vitro by mixing 5 uM of dCas9-EGFP protein (Novateinbio, Cat# PR-137213G) with 5 uM of the gRNA (target sequence: 5’-ACACTGAAAAACACATTCGT-3’) synthesized using GeneArt Precision gRNA Synthesis kit (ThermoFisher, Cat# A29377) in the reaction buffer (2 mM HEPES, 10 mM NaCl, 5 mM MgCl2, 10 μM EDTA, pH 6.5) and incubating at room temperature for 10 min⁷⁴.

Immunostaining of whole-mount oocytes and chromosome spreads

Mouse oocytes/eggs were fixed in freshly prepared 2% paraformaldehyde (Electron Microscopy Sciences, Cat# 15710) in 1x PBS (Quality Biological, Cat# 119-069-101CS) with 0.1% Triton X-100 (Millipore, Cat# TX1568–1) for 20 min at room temperature (RT), permeabilized in 1x PBS with 0.1% Triton X-100 for 15 min at RT, placed in the blocking solution (0.3% BSA (Fisher bioreagents, Cat# BP1600–100) and 0.01% Tween-20 (ThermoFisher, Cat# J20605-AP) in 1x PBS) overnight at 4°C, incubated 2 hr with primary antibodies at RT, washed three times for 10 min with the blocking solution, incubated 1 hr with secondary antibodies at RT, washed three times for 10 min in the blocking solution, and mounted on microscope slides with the Antifade Mounting Medium with DAPI (Vector Laboratories, Cat# H-1200).

For chromosome spreads, zona pellucida was removed from oocytes with Acidic Tyrode’s Solution (Millipore, Cat# MR-004-D), and the oocytes were fixed with 1% paraformaldehyde, 0.15% Triton X-100, and 3 mM DTT (Sigma, Cat# 43815) at prometaphase I (3 hr from NEBD) or at metaphase I (7 hr from NEBD).

For the cold stable microtubule assay, oocytes were cultured for 6 hr after NEBD and placed into ice cold M2 for 3 min before fixation with freshly prepared 2% paraformaldehyde in 1x PBS with 0.1% Triton X-100 for 20 min at RT, permeabilized in 1x PBS with 0.1% Triton X-100 for 15 min at RT, and placed in the blocking solution (0.3% BSA and 0.01% Tween-20 in 1x PBS) overnight at 4°C. Fixed oocytes were stained for α-tubulin and ACA (kinetochore). TALE signals were used to distinguish domesticus and spretus centromeres.

The following primary antibodies were used: rabbit anti-Topoisomerase II (1:100, Abcam, Cat# ab109524), mouse anti-human HEC1 antibody (1:100, Santa Cruz, Cat# sc-515550), mouse anti-mouse MAD1 antibody (1:50, Santa Cruz, Cat# sc-137025), rabbit anti-human MCAK (1:1000, a gift from Duane Compton), CREST human autoantibody against centromere (ACA, 1:100, Immunovision, Cat# HCT-0100), and anti- α-tubulin (DM1A, 1:500, Sigma, Cat# CP06–100UG). Secondary antibodies were Alexa Fluor 488–conjugated donkey anti-rabbit (1:500, Invitrogen, Cat# A21206) or donkey anti-mouse (1:500, Invitrogen, Cat# A21202), Alexa Fluor 568–conjugated goat anti-rabbit (1:500, Invitrogen, Cat# A10042), Alexa Fluor 647–conjugated donkey anti-mouse (1:500, Invitrogen, Cat# A32787), or Alexa Fluor 647–conjugated goat anti-human (1:500, Invitrogen, Cat# A21445).

Microscopy and Image analysis

Fixed mouse oocytes/eggs were imaged with a microscope (Eclipse Ti; Nikon) equipped with 100x / 1.40 NA oil-immersion objective lens, CSU-W1 spinning disk confocal scanner (Yokogawa), ORCA Fusion Digital CMOS camera (Hamamatsu Photonics), and 405, 488, 561 and 640 nm laser lines controlled by the NIS-Elements imaging software (Nikon). Confocal images were acquired as Z-stacks at 0.3 μm intervals. For live imaging, oocytes were placed into 3 μl drops of M16 covered with paraffin oil in a glass-bottom tissue culture dish (fluoroDish, Cat# FD35–100) in a stage top incubator (Tokai Hit) to maintain 37°C and 5% CO₂ in air. Time-lapse images were collected with a microscope (Eclipse Ti2-E; Nikon) equipped with the 20x / 0.75 NA objective, CSU-W1 spinning disk confocal scanner (Yokogawa), ORCA Fusion Digital CMOS camera (Hamamatsu Photonics), and 405, 488, 561 and 640 nm laser lines controlled by the NIS-Elements imaging software (Nikon). Confocal images were collected as Z-stacks at 1 μm intervals to visualize chromosome dynamics (Figures 2C, S1A, S2C) and polar body extrusion events (Figures 3A, S2A, S2D). Images are displayed as maximum intensity Z-projections.

Fiji/ImageJ (NIH) was used to analyze all the images^75,76. In general, optical slices containing chromosomes were added to produce a sum intensity Z-projection for pixel intensity quantifications. To quantify centromeric signal intensities (HEC1, MCAK, and MAD1), ellipses were drawn to encompass centromere regions of each chromosome based on TALE signals and the DAPI staining, and the signal intensity was integrated over each ellipse after subtracting background signals, obtained near the centromere. For the in situ chromosome counting assay to test preferential chromosome segregation, the number of chromosomes were counted in metaphase II eggs using the DAPI, TALE, and TOP2A signals⁷³. In dom x spr eggs, strong TOP2A signals at pericentromeres were used to identify domesticus chromosomes. In dom x mus eggs, stronger MCAK and dCas9-EGFP signals recognizing Minor satellites were used to identify musculus chromosomes.

Quantification and statistical analysis

Data points were pooled from two to five independent experiments. Data analysis was performed using Microsoft Excel and GraphPad Prism 10. Scattered plots were created with GraphPad Prism 10. Unpaired Mann-Whitney test (Figures 2A, 2C, and 3D), one sample t test for deviations from 50% (Figures 1C, 1E, and 3D), Chi-square test for independence (Figure 2B), Wilcoxon signed rank test for deviations from 1 (Figure 1D), and Kruskal-Wallis test (Figure 3B) were used for statistical analyses, and the actual P values are shown in each figure panel.

Key Resources Table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
rabbit anti-Topoisomerase II	Abcam	RRID:AB_10859793
mouse anti-human HEC1	Santa Cruz	sc-515550
mouse anti-mouse MAD1	Santa Cruz	sc-137025
rabbit anti-human MCAK	gift from Duane Compton	N/A
anti-α-tubulin	Millipore	RRID: AB_2617116
CREST human autoantibody against centromere	Immunovision	RRID:AB_2744669
Alexa Fluor 488 donkey anti-rabbit IgG	Invitrogen	RRID:AB_141708
Alexa Fluor 488 donkey anti-mouse IgG	Invitrogen	RRID:AB_141607
Alexa Fluor 568 goat anti-rabbit IgG	Invitrogen	RRID:AB_2534017
Alexa Fluor 647 donkey anti-mouse IgG	Invitrogen	RRID: AB_2762830
Alexa Fluor 647–conjugated goat anti-human	Invitrogen	RRID:AB_2535862
Bacterial and virus strains
Biological samples
Chemicals, peptides, and recombinant proteins
M2 media	Sigma-Aldrich	M7167
M16 media	Millipore	M7292
Paraffin oil	Nacalai	NC1506764
Reversine	Sigma-Aldrich	R39-4-1MG
Milrinone	Sigma Millipore	M4659
T7 mMessage mMachine kit	Ambion	AM1340
MEGAclear Transcription Clean-up kit	Thermo Fisher Scientific	AM1908
dCas9-EGFP protein	Novateinbio	PR-137213G
GeneArt Precision gRNA Synthesis kit	Thermo Fisher Scientific	A29377
Acidic Tyrode’s Solution	EmbryoMax	MR-004-D
Antifade Mounting Medium with DAPI	Vector Laboratories	H-1200
Paraformaldehyde	Electron Microscopy Sciences	15710
Monastrol	Millipore	475879
SPY650-DNA	Spirochrome	SC501
Critical commercial assays
Deposited data
Experimental models: Cell lines
Experimental models: Organisms/strains
Mouse: NSA (CF1)	Envigo	033
Mouse: C57BL/6J	Jackson Laboratory	000664
Mouse: ZALENDE/EiJ	Jackson Laboratory	001392
Mouse: PWD/PhJ	Jackson Laboratory	004660
Mouse: SPR2	RIKEN BioResource Research Center	RBRC00208
Oligonucleotides
gRNA (target sequence: 5’-ACACTGAAAAACACATTCGT-3’)	Clark et al.73	N/A
Recombinant DNA
TALE-mRuby2-CENP-T	This paper	N/A
TALE-mRuby2	Addgene	47879
Software and algorithms
Fiji/ImageJ	Schindelin et al.74; Schneider et al.75	https://fiji.sc/
GraphPad Prism 10	GraphPad	http://www.graphpad.com/
Other
Plastic pipette tip	Cooper Surgical Inc.	MXL3-75, MXL3-100

Highlights:

• Enlarged kinetochores recruit more spindle checkpoint proteins

• Extended spindle checkpoint activation promotes transmission of selfish centromeres

• First mechanistic link between selfish genetic elements and checkpoints