Micronucleus formation causes perpetual unilateral chromosome inheritance in mouse embryos

Significance

Early mammalian embryos frequently constitute a mixture of euploid and aneuploid cells, termed embryo mosaicism. Although this is considered a major cause of fertility problems, the mechanistic explanation for the mitotic errors that give rise to the aneuploid cells within mosaic embryos remains mysterious. Here using long-term live imaging of chromosome segregation in mouse embryos we show that individual chromosomes are frequently encapsulated within small nucleus-like structures called micronuclei. We show that micronucleus-enclosed chromosomes lack proper kinetochores and are therefore unable to be correctly segregated, causing them to be randomly inherited by just one of the daughter cells during subsequent embryonic cell divisions. This unexpected pattern of chromosome inheritance provides a novel explanation for mosaicism in early embryos.

Abstract

Chromosome segregation defects in cancer cells lead to encapsulation of chromosomes in micronuclei (MN), small nucleus-like structures within which dangerous DNA rearrangements termed chromothripsis can occur. Here we uncover a strikingly different consequence of MN formation in preimplantation development. We find that chromosomes from within MN become damaged and fail to support a functional kinetochore. MN are therefore not segregated, but are instead inherited by one of the two daughter cells. We find that the same MN can be inherited several times without rejoining the principal nucleus and without altering the kinetics of cell divisions. MN motion is passive, resulting in an even distribution of MN across the first two cell lineages. We propose that perpetual unilateral MN inheritance constitutes an unexpected mode of chromosome missegregation, which could contribute to the high frequency of aneuploid cells in mammalian embryos, but simultaneously may serve to insulate the early embryonic genome from chromothripsis.

Accurate chromosome segregation is achieved by correct attachment of spindle microtubules to kinetochores, complex proteinaceous structures that assemble on centromeric DNA. Misattachment can cause so-called lagging chromosomes during anaphase (1), which are a hallmark of chromosomally unstable cells (2), and can result in micronuclei (MN)—small nucleus-like bodies that form if a chromosome remains separate from the main group of chromosomes at the time of nuclear envelope reformation. MN have long been used as a marker of genetic fidelity. For example, MN may be predictive of tumorigenicity and can be used to screen chemicals for genotoxicity (3–5). However, the cellular impact of MN formation is poorly understood. Recent studies found that chromosomes within MN become heavily damaged, causing DNA rearrangements (6–9). Paired with reincorporation of the MN chromosome into the main nucleus (principal nucleus; PN) during the next cell cycle (7, 9, 10), this provided an elegant explanation for chromosome-specific extensive rearrangements seen in many cancer cells, termed chromothripsis (11, 12). In the present study we show that the outcome of MN formation is markedly different in the preimplantation mammalian embryo.

Chromosomal mosaisicm is common in mammalian preimplantation embryos, up to 50% of human embryos produced in fertility clinics containing some aneuploid cells resulting from early mitotic errors (13–16), for which there is currently no clear cellular explanation. Simultaneously, early embryos frequently exhibit MN, the causes and consequences of which are unknown (17, 18). Here we pair long-term live 4D microscopy with high-resolution fixed embryo analysis to analyze the causes and consequences of naturally occurring MN in mouse embryos. Our experiments reveal an unexpected series of events in which chromosomes from within MN do not rejoin the principal nucleus, but are repeatedly inherited by only one daughter cell. We propose that this mechanism should generate a cascade of aneuploid cells, while protecting the genome from chromothripsis-like rearrangements.

Results

Micronucleus Formation in Mouse Embryos.

The preimplantation mouse embryo develops from a fertilized zygote to a 16- to 32-cell morula and then a 64- to 128-cell blastocyst without intervening cell growth over the course of ∼4 d, providing a tractable setting in which to examine cell divisions. We first analyzed MN occurrence using fixed-cell analysis of embryos cultured in vitro under standard conditions. MN were rare in early embryos, but from the 16-cell stage onward, embryos typically possessed 1–5 MN-containing cells (Fig. 1A and Fig. S1). Embryos that developed in vivo and were fixed for examination immediately after isolation at morula stage exhibited a similar number of MN as in vitro cultured embryos, revealing that in vitro culture does not affect MN abundance (Fig. 1A). MN could potentially arise via a variety of mechanisms (5, 19–21). To determine how MN form in embryos we performed medium-term (20-h duration) live confocal 4D time-lapse imaging of histone 2B:RFP-expressing (H2B:RFP) morulae, analyzing the dynamics of chromosome segregation in a total of 453 cell divisions. The vast majority (70%) of divisions occurred without obvious defect, the two sets of sister chromatids moving apart synchronously in anaphase. Fifteen percent of divisions exhibited mildly lagging chromosomes, which did not normally result in MN formation. However, in 10% of cell divisions, one or more severely lagging chromosomes were detected that resulted in MN formation in one of the daughter cells (Fig. 1B). Immunolabeling of embryos fixed in midanaphase revealed that lagging chromosomes almost always possessed CREST-labeled kinetochores at their leading edge (27/29 cases; Fig. 1C), indicating that anaphase-lagging chromosomes in embryos are usually intact chromatids as opposed to DNA fragments and thus that newly formed MN generally contain single whole chromatids. Consistent with this, MN diameter varied little between cells or developmental stage (1.74 +/− 0.06 μm in diameter in morula, 1.63 +/− 0.11 μm in blastocysts). Recent studies in cancer cells revealed that DNA in MN is subject to high levels of damage during the subsequent S-phase as a result of defective MN nuclear envelope function (6, 7). Analogous to this we found that MN in morulae exhibit faint or absent staining for the nuclear envelope structural component nuclear Lamin B1 (Fig. 1D) and also for LSD1, a marker of nuclear import (6), suggesting that MN nuclear envelope function is defective. Labeling with γH2AX antibodies revealed very high levels of damage in approximately half of all MN (Fig. 1D), consistent with DNA damage occurring in S-phase, as in cancer cells (6, 7, 22). In summary, lagging whole chromatids are a normal feature of preimplantation development in mouse, giving rise to MN within which DNA is subject to damage.

Fig. 1.

Cause and impact of MN formation in mouse embryos. (Ai) Typical example of a fixed morula illustrating the appearance of an MN (arrow). (Aii, Aiii) Fixed-cell analysis of MN number in in vitro cultured embryos during development, illustrating the emergence of MN at the 16- to 32-cell stage. (Aiv) Contemporaneous comparison of MN abundance in morulae that were either cultured from two-cell stage or flushed from the uterus at morula stage. Note no difference in MN abundance. Sixteen to sixty embryos per datapoint. (B) Live imaging of cell division using H2B:RFP. (Bi) Examples of chromosome segregation (white arrows), illustrating anaphases with and without lagging chromosomes that cause an MN (yellow arrows). (Bii) Analysis of 453 cell divisions from 58 embryos. (Ci) Example of embryo fixed for kinetochore examination in midanaphase. (Cii) Note that lagging chromosomes possessed clear CREST-labeled kinetochores in 27/29 cases. (D) Immunofluorescence analysis of MN structure and function using Lamin B1, LSD1, and γH2AX antibodies. Yellow circles highlight the MN. Quantitative fluorescence intensity analysis of PN vs. MN is presented for each. Minimum 20 MN-containing cells per group. Error bars represent SEM. T tests were used where appropriate, asterisks indicating P < 0.01.

Fig. S1.

MN abundance during early embryo development. Analysis of MN abundance in fixed embryos. Dataset the same as presented in Fig. 1A, presented so as to illustrate the number of MN-containing cells per embryo at different developmental stages. Total of 153 embryos examined.

Micronuclei Are Unilaterally Inherited at the Time of Cell Division.

In cancer cells, MN chromosomes are segregated along with chromosomes from the PN in the next mitosis, such that an MN often persists for only one cell cycle (7, 10). To explore the fate of MN during early development we examined the behavior of MN in live H2B:RFP-expressing morulae. In contrast to cancer cells, in mouse embryos we found that the MN was segregated normally along with the main chromosome mass in only 2 of 34 cases (6%) (Fig. S2). In the other 94% of MN-containing cell divisions, the MN was inherited by only one of the two daughter cells and persisted as an MN in that daughter cell (Fig. 2A). In most cases the MN remained visibly separate from the main chromosome mass throughout the whole of M-phase and was continuously observed to travel into one daughter (Fig. 2A). We saw no evidence of MN fragmentation that might lead to the reincorporation of a portion of MN DNA back into the PN, but cannot formally exclude very small DNA fragments below the imaging resolution limit. To determine why the chromosomal content of MN is not segregated normally, H2B:RFP-expressing embryos were observed with live 4D confocal microscopy, and individual embryos were fixed for examination by immunofluorescence when an MN-containing cell entered mitosis. Chromosomes from within MN exhibited a rounded hypercondensed appearance in ∼50% of cases (Fig. 2B), whereas the normal condensed chromatin structure typical of mitosis was evident in the other ∼50% (Fig. S3). Chromosomes from MN were usually spatially divorced from the spindle with no evidence of spindle microtubule interaction (Fig. 2B). Notably, regardless of the morphology of the DNA, chromosomes from MN lacked prominent CREST staining in all cases, indicating a failure to maintain a proper kinetochore, whereas prominent CREST-labeled kinetochores were evident on other chromosomes (Fig. 2B and Fig. S3). Because lagging chromosomes that form MN possess clear kinetochores (Fig. 1C), this indicates that the ability to support a functional kinetochore is lost while the chromatid is enclosed within the MN, concomitant with DNA damage. As a result, chromosomes from MN fail to be segregated in the subsequent mitosis and are inherited by only one of the daughter cells.

Fig. S2.

MN reincorporation. As described in the text, the vast majority of MN-containing cell divisions result in the persistence of the MN and its inheritance into one daughter cell. Displayed here is a rare example of the alternate outcome, where an MN reincorporates into the PN following the subsequent mitosis, highlighted with a white arrow.

Fig. 2.

MN are unilaterally inherited during embryogenesis. (Ai) Example of MN inheritance observed with H2B:RFP imaging. Note that the MN (red arrow) remains separate from the rest of the chromosomes (white arrow) during M-phase and is inherited by one daughter cell. (Aii) Analysis is of 34 MN-containing cell divisions. (B) H2B:RFP embryos were observed using live imaging and then fixed when an MN-containing blastomere was observed to enter mitosis. The embryo was then immunolabeled for microtubules and CREST. Note that the MN chromosome (yellow arrow) lacks clear CREST-labeled kinetochores and is separate from the spindle.

Fig. S3.

Absence of kinetochore staining on MN chromosomes in metaphase. A second example (supplemental to that in Fig. 2B), in which the MN chromosome assumes a normal mitotic condensed chromosome appearance , but nonetheless lacks clear evidence of a CREST-labeled kinetochore (arrow). Note clear CREST-labeled kinetochores on all other chromosomes.

Micronucleus Inheritance Does Not Prevent Preimplantation Development.

An advantage of the early mouse embryo as a model for analyzing cell divisions is that all cells remain constrained within a ∼100-µm sphere throughout preimplantation development, enabling observation of sequential cell divisions following MN formation. We therefore analyzed MN inheritance using datasets comprising complete Z-projections acquired every 225 s using spinning disk microscopy for 90 h, encompassing development from one-cell stage through to blastocyst stage (Fig. 3). This provides the spatial and temporal resolution to observe the genesis and fate of all MN, without affecting the health of the embryo, as determined by the ability to generate live pups after embryo transfer (23). We performed a comprehensive analysis of every cell division within each embryo, recording the incidence of lagging chromosomes and MN, as well as cell cycle durations, cell fate, and lineage relationships during preimplantation development. Consistent with the previous experiments, chromosomes from MN were not segregated in the subsequent cell cycle, but were unilaterally inherited, resulting in a single MN in one daughter cell (Fig. 3A). Importantly, these long-term datasets allowed us to continuously track the same MN over the course of several cell divisions. Taking advantage of this, we observed that second-generation MN were similarly inherited by one daughter cell, and MN were continually inherited in this manner up to four times within the observed period of preimplantation development (Fig. 3A and Fig. S4). Analysis of cell cycle timings across our datasets revealed that MN had no impact upon the duration of M-phase (Fig. 3B), consistent with the lack of coherent kinetochores on MN chromosomes. Interphase durations were also unaltered, indicative of a failure to activate DNA-damage responses (Fig. 3B). Moreover, cell cycle durations were similar in cells containing first- or subsequent-generation MN (Fig. 3B). For example, the 16- to 32-cell mitosis was 51.6 +/− 7.7 min in cells containing first-generation MN and 55.8 +/− 7.9 min in cells containing second-generation MN (P > 0.1). Thus, once formed, MN persist throughout preimplantation development, with their DNA contents being inherited at each subsequent mitosis without apparent impact upon the durations of subsequent cell divisions.

Fig. 3.

Long-term 4D imaging reveals repeated MN inheritance. (Ai) Live imaging of H2B:RFP-expressing embryos throughout preimplantation development and (Aii) lineage analysis of the same embryo. Blue box shows timing details of the divisions illustrated in the images in Ai. Note that in this example the MN, which is generated by a lagging chromosome, is then unilaterally inherited in the next two divisions. (B) Analysis of cell cycle durations in MN-containing cells compared with MN-free cells with no history of lagging chromosomes. Analysis of 12 examples of imaging complete continuous development from one-cell stage to blastocyst. Error bars represent SEM.

Fig. S4.

Repetitive MN inheritance. An additional example of lineage tracking following live imaging from one-cell stage through to blastocyst stage, similar to that shown in Fig. 3. Note that in this example, an MN is formed as a result of a lagging chromosome and is subsequently inherited four times.

Passive Inheritance Results in Randomly Distributed Micronuclei in Embryos.

Finally, we performed three series of experiments to understand the spatiotemporal dynamics of MN inheritance during cell division. First, we analyzed the 3D positioning of MN within live embryos by colabeling chromosomes and the plasmalemma using H2B:RFP and CAAX:GFP, respectively. In interphase, MN oscillated over small distances in three dimensions, but exhibited little or no net displacement over time (Fig. S5). No obvious inheritance bias was detected based on cell morphology, the MN being equally likely to be inherited by the larger or smaller daughter cell or, similarly, the daughter cell with the larger or smaller nucleus (Fig. S5). Second, as a further way of tracking the movement of MN, H2B:RFP-expressing embryos were coinjected with inert fluorescent beads to provide a passive maker of cytoplasmic dynamics during cell division. MN movement in anaphase closely mirrored that of nearby beads, suggesting that MN move passively as a result of normal cytoplasmic dynamics, rather than by specific cytoskeleton-directed mechanisms (Fig. 4A). Third, we analyzed the distribution of MN in ∼128-cell stage blastocyst-stage embryos and found no difference in the proportion of MN-containing cells in the inner cell mass (ICM; 3.6 +/− 0.8% of cells with MN) and the trophectoderm (TE; 3.2 +/− 0.5%; P > 0.1) (Fig. 4B). Together, these experiments indicate that MN movement in embryos is undirected, resulting in an even distribution of MN between the two major cell lineages at the end of preimplantation development.

Fig. S5.

Monitoring MN behavior in relation to cell shape. Two-cell stage embryos were microinjected with cRNA encoding H2B:RFP to label chromosomes, and CAAX:GFP to label plasmalemma, and live imaging performed using confocal microscopy, as elsewhere in the study. (A) Timecourse illustrating interphase motion of MN. Note that the MN exhibits little to no net movement during the period of imaging. (B) Example of mitosis in an embryo expressing CAAX:GFP and H2B:RFP (n = 11). The dotted lines highlight the two daughter cells. (C) Quantitative analysis of cell and nucleus size in relation to MN inheritance. Note that the MN is equally likely to be inherited by the larger or smaller cell.

Fig. 4.

Passive MN inheritance and even distribution of MN in blastocysts. (A) Analysis of anaphase MN motion relative to inert Dragon-green beads (n = 9). (Ai) Individual time-points are shown charting the division of an MN-containing cell within a morula. Note that MN motion closely matches that of nearby beads (white and green arrowheads, respectively), indicating that MN motion is not directed by specific mechanisms. (Aii) Tracking analysis from the same cell showing the movement of anaphase chromosomes (red), the MN (gray), and the same two beads highlighted in Ai (green). Bold circles mark positions at the end of the experiment. Note that the movement of the MN reflects that of nearby beads. (Aiii) Quantitative analysis of distances between the beads and MN. Distance between anaphase chromosomes shown on the same chart to illustrate timings. (Bi) Example MN location analysis using Oct4 antibodies to label the inner cell mass (ICM; green). Trophoectoderm (TE) nuclei appear blue, as they are unlabeled by Oct4 antibodies. Note the blastocyst shown is hatching and therefore takes a “figure-8” appearance. Arrow marks MN. (Bii) Analysis is of 34 embryos. Error bars represent SEM. (C) Cartoon depiction of perpetual unilateral inheritance. See further description in text.

Discussion

Our experiments show that, in mouse embryos, encapsulation within an MN leads to DNA damage and an apparently irreversible loss of the ability to assemble a normal kinetochore. As a result, MN chromosomes repeatedly fail to be segregated and are instead perpetually inherited (Fig. 4C). Although we have not formally counted chromosomes in embryo cells here, we suggest that repeated MN inheritance must result in whole chromosome aneuploidy (depicted in Fig. S6). If the lagging chromosome is initially correctly inherited such that first-generation MN-containing cells possess the correct number of chromosomes (albeit one within an MN), as is frequently the case in cancer cell lines (10, 24), then unilateral inheritance would render MN-free progeny of that cell hypoploid. Alternatively, if the MN first forms in the incorrect cell, then the MN-containing cell and its MN-containing progeny would be hyperploid, whereas the original MN-free cell and its progeny would be hypoploid (Fig. S6). Perpetual MN inheritance thus presents means of generating a cascade of aneuploid cells from a single initial lagging chromosome. Direct observation of the ploidy impacts of MN inheritance will require new methods for counting chromosomes in individual blastomeres in live embryos in situ. However, we suggest that this pattern of chromosome dynamics may provide at least a partial explanation for the high level of mitotic chromosome segregation errors and mosaicism detected in human embryos.

Fig. S6.

Cartoon explanation of hypothetical ploidy outcomes following unilateral MN inheritance. The cartoon depicts two possible scenarios, depending upon whether the MN is initially formed in the “correct” cell (blue numerals indicate chromosome numbers), or whether the MN is initially formed in the wrong cell (green numerals). For further explanation, see text.

Whether the apparent absence of kinetochores on DNA from MN in embryos involves specific mechanisms or is a serendipitous by-product of extreme chromosome damage remains to be explored. In addition to CREST-labeling being absent on chromosomes from within MN, we have also found that MN typically lack foci of the centromeric histone CENP-A, which are readily detectable in the principal nucleus in embryos (25) (Fig. S7). A simple explanation therefore is that damage to the MN chromosome may lead to loss of centromeric identity, precluding proper kinetochore assembly. Our data thus raise the broader question of whether lower levels of DNA damage, perhaps causing more subtle kinetochore assembly defects, may cause de novo segregation defects and thus be a more general driver of segregation errors in some contexts (26).

Fig. S7.

MN lack CENP-A foci; immunofluorescence analysis using CENP-A antibodies. (Ai) A typical confocal image of an MN lacking a pronounced CENP-A signal, whereas clear discrete foci are observed in the PN. (Aii) A rare example (∼10% of cases) where a clear, albeit reduced, CENP-A signal was observed in the MN. Arrows highlight MN. (B) Quantification of CENP-A foci intensity in PN and MN across all MN-containing cells. Data from a total of 55 MN from 63 embryos.

We conclude that perpetual unilateral MN inheritance is a previously unidentified mode of chromosome segregation error and joins chromothripsis as a direct consequence of MN formation. In addition, we speculate that MN inheritance may simultaneously serve an important genome protective role in early development. Although we did see MN rejoining the main group of chromosomes on a very small number of occasions (6%; Fig. S2) and so do not exclude the possibility that some low level of chromothripsis-like rearrangement could occur in early embryos (27), our data show clearly that in the vast majority of cases, MN chromosomes do not rejoin the PN during mitosis in preimplantation development. We also observed no evidence in any of our datasets of MN reabsorption into the PN in interphase. Indeed, our data raise the possibility that MN observed in embryonic stem cells (28) reflect segregation errors in preimplantation development. Thus, whereas MN in somatic cells rejoin the genome and risk cell transformation, MN in embryos generally do not. Instead, MN inheritance generates a small cascade of aneuploid cells, which likely undergo apoptosis later in development (29). MN inheritance may therefore provide a means of sacrificing a modest number of cells to prevent the incorporation of damaged DNA into the genome, thereby insulating the early embryo from chromothripsis. Whether some somatic tissues retain this strategy remains to be determined.

Materials and Methods

Embryo Culture and Microinjection.

All experiments were approved by the Centre de Recherche du Centre Hospitalier de l'Université de Montréal (CRCHUM) Comité Institutionnel de Protection des Animaux (CIPA). Embryos were harvested from superovulated BDF1 female mice mated with BDF1 males and cultured in KSOM media in 5% CO₂ at 37 °C. mRNA was manufactured using Ambion mMessage Machine according to the manufacturer’s instructions and microinjected into embryos using a picopump (World Precision Instruments) and micromanipulators (Narishige) mounted on a Leica DMI4000 inverted microscope, as previously described (30). Plasmids used were CAAX:GFP in pcDNA3.1 (gift from Guillaume Charras, University College London, London) and H2B:RFP in pRN4 (gift from Alex McDougall, Observatoire Océanologique de Villefranche-sur-Mer, Villefranche Sur Mer, France). Dragon-green beads were purchased from Bangs Laboratories and injected at a 1:5 dilution.

Immunofluorescence and Live Imaging.

Embryos were fixed using 4% paraformaldehyde (PFA; 40 min) and permeabilized by Triton-X (0.25%, 10 min) (31). Primary antibodies used as follows: CREST (gift from William Earnshaw, University of Edinburgh, Edinburgh, 1:200), α-tubulin (Sigma, 1:1,000), γH2AX (Trevigen, 1:800), Oct3/4 (SantaCruz, 1:300), LSD1 (Cell Signaling Technology, 1:400), and Lamin B1 (Abcam, 1:1000). Where CENP-A antibodies were used (Cell Signaling Technology, 1:200), embryos were fixed with 2% PFA. Alexa-labeled secondary antibodies were from Life Technologies. Live imaging was performed on a Leica SP8 confocal microscope fitted with a 20x 0.75NA objective and a HyD detector. Imaging was performed for 20 h at the morula stage, and embryos were included for analysis only following morphologically normal cavitation, which typically occurred in ≥50% of embryos in any given experiment. Complete 4D datasets of embryo development (as in Fig. 3) were obtained as previously described (23, 32). Briefly, 61 optical sections were obtained at 225-s intervals for 90 h using a CSU10 Yokagawa Nipkow disk system, mounted on an Olympus IX-71 inverted microscope.

Analysis and Statistics.

All data analysis was performed using ImageJ/Fiji. For tracking, positional coordinates were extracted from Z-stack datasets to calculate distances in 3D using the TrackMate Fiji plugin. Where shown, error bars represent SEM. T tests were used where appropriate, asterisks indicating P < 0.01.