Capturing aberrant cell behaviors producing defects in human embryos via live imaging

Abstract

The generation of human embryos in vitro has revolutionized reproductive medicine, and also made it possible to study fundamental aspects of early human development. However, human preimplantation embryos often display an array of morphological defects associated with poor development and implantation. Here, we used live-embryo imaging and computational analysis to capture how these defects can be produced in real time. We record various forms of mitotic errors including lagging chromosomes producing micronuclei, multipolar spindles causing abnormal chromosome organization recapitulated by daughter cells, and uncontrolled scattering of condensed chromosomes. In addition, we capture abnormal cleavage furrow dynamics during cytokinesis producing binucleated and enucleated cells. Finally, we find cells with disrupted mitotic progression ultimately leading to blebbing and fragmentation. Thus, these results document specific aberrant cell behaviors producing morphological defects in real time, and indicate that errors during mitosis and cytokinesis represent a major cause of developmental failures in human embryos.

Live-imaging captures in real time mitotic errors and defects in the human embryo.

INTRODUCTION

Following the landmark demonstration of successful human in vitro fertilization (IVF) and the first embryo transfer, ~12 million babies have thus far been born from IVF globally (1, 2). Currently, over 3 million cycles are performed annually, resulting in 769,977 births, and the use of IVF continues to grow in assisted reproduction for multiple indications including infertility, fertility preservation, and preimplantation genetic testing (3). Moreover, surplus embryos donated for research have provided a unique gateway to study key aspects of early human development (4, 5). During preimplantation development, the cells of the embryo undergo the first mitotic divisions and must accurately segregate their genetic contents. They also adopt apical-basal polarity and tissue organization. In addition, they undergo the first lineage segregation allocating daughter cells to the outer trophectoderm that forms the placenta, and the inner cell mass (ICM) that forms the fetus and allantois. Finally, the blastocyst hatches from the zona pellucida to start implantation (6–9).

All these morphogenetic events require precise control of cell behaviors. Yet, human embryos frequently display a wide range of defects including aneuploidy, cell cycle arrest, and fragmentation, which can be detected across all, or some of the cells of the early embryo (10–18). Indeed, the gold standard of embryo grading still relies on morphological assessment including cell number, symmetry, and fragmentation (19). Moreover, embryos with morphological defects are known to have lower blastulation, implantation, and live birth rates (19–24). However, we lack a comparative documentation of the aberrant cell behaviors that produce each of these defects in real time. For instance, it is not fully understood if fragmented cells are preferentially produced during interphase or during specific stages of mitosis (16, 25), or whether binucleated cells derive primarily from failure during cytokinesis or cell fusion events (15).

Current approaches to visualize human embryos and clinical grading protocols are largely based on the use of standard light microscopy tools, which can acquire static or time-lapse images. This enables evaluation of overall embryo appearance, presence of fragmented cells, and signs of delayed development or cell cycle arrest (14, 26–30). A more detailed subcellular analysis can be achieved using stainings and immunofluorescence in fixed embryos, followed by confocal imaging in three dimensions (3D). This can detect, for example, the presence of micronuclei and multinucleated cells (31). Yet, this approach only allows the study of static snapshots precluding live-embryo tracking. Technologies such as preimplantation genetic testing for aneuploidy (PGT-A) offer a powerful tool to identify genetic defects, including the origin of aneuploidy (32), but require invasive removal of cells from the embryo (33, 34). Moreover, given the extent of mosaic aneuploidy reported in human embryos, techniques like PGT-A may not necessarily reflect the genetic status or defects of all cells across the embryo (35).

Live-embryo imaging enables the capture of cell dynamics in real time (36). Using standard light microscopes integrated with incubation systems has enabled the visualization of various cellular defects including cell fragmentation and multinucleation during the first mitotic divisions in human embryos (12, 37). However, standard light optics restrict this approach to two-dimensional (2D) planes, without axial (depth) resolution. Moreover, tracking individual cells in 2D becomes unfeasible when the embryo undergoes compaction and its cell-cell boundaries become difficult to resolve, even via differential interference contrast (DIC). Furthermore, since the first embryonic cleavages occur without subsequent cell growth (i.e., increase in daughter cell volume), the tracking accuracy decreases incrementally, as each division produces progressively smaller cells (38). Recent studies have exploited the use of confocal microscopy to reveal mitotic errors with higher spatial resolutions during the first cell divisions of the embryo (39, 40), but it remains a challenge to apply similar approaches to more advanced stages of preimplantation development, characterized by increased cell number and architectural complexity.

We have recently combined the use of confocal microscopy, fluorescent dyes, and computational segmentation to track cell behaviors with a high spatiotemporal resolution in human embryos (41). This allowed us to characterize some of the main morphogenetic events underlying human preimplantation development including embryo compaction, lineage segregation, and blastocyst formation. Here, we tested this live-imaging approach to capture and characterize aberrant cell behaviors that can produce defects recurrently found in human embryos.

RESULTS

Static characterization of cellular defects in fixed human embryos

To characterize defects in human embryos, we first performed static 3D analyses in 15 fixed embryos, which were obtained at various stages of preimplantation development (Fig. 1). We combined the use of the marker 4′,6-diamidino-2-phenylindole (DAPI), which labels DNA, and Phalloidin-Alexa488, which labels filamentous actin (F-actin). This allowed us to comprehensively assess defects in chromosome and nuclear integrity in parallel with cell morphology, as F-actin is enriched at the cell cortex and highlights the contour of the entire cell and embryo.

Fig. 1. Morphological defects in fixed human embryos.

Embryos fixed at the indicated stages, stained with DAPI and Phalloidin-Alexa-488. (A) DAPI allows detection of a micronucleus structure in an 8-cell stage embryo. Insets show magnified views (top panel shows 3D view, bottom panel shows a thinner 2D plane). (B) Cytoplasmic DNA structures in the trophectoderm layer may represent micronuclei derived from mitotic errors or nuclear DNA shedding, reported to occur in a small cell population during cavitation (41). Insets show cytoplasmic DNA structures from different angles (arrowheads). (C) In this blastocyst, a cell was captured during anaphase. A DNA-positive structure is detected in the central spindle, which likely represents a lagging chromosome (arrowhead). (D and E) Detection of cells containing abnormal nuclear number in an 8-cell embryo (D) and blastocyst (E). Segmentation of cell contours allows precise determination of multiple nuclei confined within the same cytoplasm. In the trinucleated cell, a micronucleus is also detected. (F) Examples of blastomeres displaying disrupted chromosome integrity. In two cells the nuclei exhibit a fragmented morphology, instead of the normal spherical organization found in interphase. It is unfeasible to establish whether these fragments were produced during interphase or mitosis. (G) 16-cell stage embryo exhibiting normally-appearing cells surrounded by cellular fragments, indicative of fragmentation. The cortex signal was used to reconstruct the fragments (inset, bottom). (H) 2D planes reveal abnormally-large cells, indicative of cell cycle arrest, in both trophectoderm and ICM. cytDNA, cytoplasmic DNA; ZP, zona pellucida; ICM, inner cell mass; TE, trophectoderm. Scale, 10 μm.

Consistent with previous work, our static analysis revealed multiple defects in human embryos including the presence of micronuclei and cytoplasmic DNA structures during interphase at both late cleavage and blastocyst stage (Fig. 1, A and B). Furthermore, we could capture a cell in anaphase where a lagging chromosome remained near the forming cytokinetic furrow (Fig. 1C). We also detected cells in interphase displaying two and three cell nucleus-like structures contained within the same cytoplasm (Fig. 1, D and E), which we refer to as multinucleated cells following previous criteria (42, 43). Additionally, we detected the presence of cells with disrupted chromosome distribution (Fig. 1F), cellular blebs and fragments devoid of DNA signal (Fig. 1G), and a subset of abnormally large cells displaying signs of cell cycle arrest (Fig. 1H).

Together, these static analyses enabled us to characterize cellular defects in human embryos with a high spatial resolution. However, they lack the temporal information needed to establish how each of these defects were produced. Based on our data and previous work, cytoplasmic DNA structures most likely originate from chromosome segregation errors during mitosis or nuclear shedding of interphase nuclei during blastocyst cavitation (40, 41). However, multinucleated cells could originate from incomplete cytokinesis or cell fusion events, and fragmented cells may arise from blebbing and fragmentation of initially-intact blastomeres (16, 44). To clarify the causal relationships of these plausible aberrant cell behaviors and the defects recurrently found in human embryos, we next applied a live-imaging approach.

Capturing aberrant cell behaviors causing disrupted chromosome organization and nuclear integrity in real time

We first focused on aberrant cell behaviors leading to defects in chromosome dynamics and nuclear integrity. For this study, we obtained 35 preimplantation embryos cryopreserved at the 8-cell stage. Following thawing, the embryos were co-stained with the fluorescent dyes SPY555-DNA and SPY650-FastAct and live-imaged. The SPY555-DNA dye labels chromosomes similarly to DAPI, and SPY650-FastAct labels F-actin similarly to Phalloidin in fixed embryos (41). Imaging live embryos requires rapid scanning with low laser exposure to minimize photodamage. Consequently, individual confocal planes typically show a lower signal-to-noise ratio, compared to experiments imaging fixed samples. However, computational segmentation of the chromatin and cell cortex in each 2D plane yields a reconstructed 3D view of the cell of interest with a superior signal-to-noise ratio, which can be analyzed from multiple angles and followed over time (Fig. 2). This approach allowed us to collect 4D imaging data (xyz plus time) of individual cells at various developmental stages. The number of cells that showed a successful division within the imaging window varied from zero to six per embryo, with experiments lasting up to 18 hours and with temporal resolution of 2 to 30 minutes between frames.

Fig. 2. Live-imaging disruptions in chromosome dynamics and nuclear integrity.

(A) SPY555-DNA and SPY650-FastAct allows high-resolution tracking of chromosome dynamics during a normally-occurring mitotic division in a human embryo. Left panels show the cell of interest at the start and end of imaging. The cell cortex is segmented to highlight its external contour and position within the embryo. The higher magnification views show the dynamics of the mitotic chromosomes, which undergo condensation, congression, and segregation into two daughter cells. (B) In this embryo, a DNA-containing structure fails to position at the metaphase plate with the rest of the condensed chromosomes, and is subsequently identified as a lagging chromosome during anaphase. The chromosome lagger forms a micronucleus-like structure which remains in the cytoplasm of one of the daughter cells for at least ~12 hours. (C) Additional live-imaging example of an aberrant mitotic behavior is captured at higher temporal resolution. Here, mitotic chromosomes condensed and congressed in an apparently-normal fashion. However, a lagging chromosome structure is produced during anaphase. (D) This cell displays mitotic entry characterized by chromosome condensation and congression towards the metaphase plate. However, the chromosomes fail to transition into anaphase and to segregate. The resulting cell does not undergo cytokinesis and establishes an abnormally-shaped chromosomes mass. (E) Segmentation analysis allows comparison of cell and nuclear volumes and sphericity. The cell on the left underwent a normal division [the same embryo from (A)] producing daughter cells and nuclei of similar size. The cell on the right, which failed to undergo mitosis in (D), produces a more irregularly shaped nuclear morphology. Scale, 20 μm.

We first collected movies of cells undergoing proper cell division (Fig. 2A and Movie S1). Here, chromosomes followed a stereotypical segregation process, and the cytokinetic furrow evenly separated the cytoplasm. Subsequently, the daughter cells started new cell cycles as evidenced by a spherical, interphasic chromosome organization (Fig. 2A and Movie S1). Conventional time-lapse microscopes employed in clinics provide limited subcellular information and impede detailed classification of mitotic phases. By contrast, our approach allowed us to reveal the main phases of mitosis via assessment of chromosome organization and cell shape. The mean duration of mitosis was 91 ± 30 minutes (n = 19 mitotic blastomeres, 8 cleavage embryos). This was obtained by measuring the time between prophase and cytokinesis (or from the last interphase time-point prior to mitotic entry and cytokinesis) (Fig. 2A and Movie S1, also see Methods). Computational segmentation also allows quantification of cell and nuclear volume, which revealed an expected halving following division (Fig. 2E).

Previous work has shown that the first mitotic division in human embryos displays frequent chromosome segregation errors (39, 40). Our tracking analysis at later developmental stages revealed additional cases of mitotic errors producing lagging chromosomes (Fig. 2, B and C; Movie S2). Interestingly, we could capture two different types of chromosome segregation errors. In one of them a putative chromosome failed to align at the metaphase plate, and remained peripheral while the other chromosomes aligned at the metaphase plate (Fig. 2B). In such cases, the transition from metaphase to anaphase occurred within 31.7 ± 12 minutes (n = 3 blastomeres from 3 embryos), which was comparable to the normally-appearing mitotic event (25.1 ± 7 minutes, 5 blastomeres from 3 embryos, P = 0.29; Fig. 2A). However, this structure remained separated from the two main chromosome masses, as they segregated away from each other during anaphase. Following cytokinesis, the DNA structure established a micronucleus, which was not immediately degraded, but persisted in the cytoplasm (Fig. 2B). This micronucleus was still detectable 12 hours from the onset of mitosis (Fig. 2B). In another case, all chromosomes appeared to have successfully condensed, congressed, and aligned at the cell equator by metaphase (Fig. 2C and Movie S2). However, a lagging structure appeared during anaphase, which remained near the equator as the sister chromatids segregated towards the two opposite cell poles (Fig. 2C and Movie S2).

We recently showed that a small fraction of nuclei undergoes a budding process, shedding DNA into the cytoplasm, specifically in trophectoderm cells during blastocyst cavitation (41). Consistent with this, we did not detect DNA shedding events at earlier cleavage stages (i.e., pre-blastocyst). This is in line with the notion that micronuclei produced prior to the blastocyst stage are derived primarily from mitotic errors, and not from disruption of the cell nucleus during interphase (41). In addition to segregation errors, we also detected a cell in which chromosomes initiated condensation, suggesting mitotic entry (Fig. 2D). However, after ~90 minutes of adopting a metaphase-like chromosome morphology, the cell showed no signs of cytokinetic furrow formation, and the chromosomes progressively became less condensed during the subsequent 90 minutes. This cell retained its overall shape and volume. However, it re-established an irregularly shaped nucleus-like structure, as demonstrated by measurements of nuclear sphericity (Fig. 2, D and E), compared with the more compacted and spherical nuclei produced during a normal division (Fig. 2A).

Capturing multipolar mitotic errors that can be recapitulated by daughter cells

In addition to the aberrant chromosome behaviors described above, previous work has identified major defects in spindle morphology. In these cases, some cells fail to establish a bipolar spindle, a defects thought to arise from aberrant centrosome regulation. Instead, an abnormal multipolar spindle is assembled, which can display three, instead of two spindle poles (42, 43). Recent imaging studies have captured this type of aberrant tripolar organization during the first mitotic division (39, 40), yet previous fixed-embryo analyses suggest they may also be prevalent during more advanced stages of preimplantation development (42).

Importantly, our live-imaging data could capture these multipolar mitoses during the transition from compacted morula to blastocyst stage (Fig. 3A and Movie S3). We first tracked the dynamics of the chromosome masses spatially arranged in this tripolar organization. As the multipolar mitosis progressed, each of the chromosome masses segregated outwards forming three separate nucleus-like structures (Fig. 3A to F). The duration of the metaphase-to-anaphase transition in normally-appearing and multipolar cells was relatively similar: bipolar division (n = 3 blastomeres); 24 ± 10 minutes and multipolar division (n = 4 blastomeres); 33 ± 15 minutes (P = 0.66, Fig. 3D). However, combining tracking with segmentation of the chromosomes and cortex showed that this multipolar segregation was followed by two sequential cytokinetic events (Fig. 3F). The first cytokinetic event segregated approximately one third of the cytoplasm and one of the chromosome masses, producing a mononucleated daughter cell (~20 minutes following metaphase) (Fig. 3, E and F). Subsequently, a second furrow formed at the center of the two remaining chromosome masses (~45 minutes from metaphase), as revealed by an analysis of the SPY650-FastAct labeling in virtual 2D planes (Fig. 3F). Unlike the first one, this furrow retracted and failed to complete cytokinesis, producing a binucleated daughter cell (Fig. 3, E and F).

Fig. 3. Capturing multipolar mitosis recapitulated by daughter cells.

(A) Live embryo imaged during the morula to blastocyst shows a cell with tripolar organization (cell #1). (B) Selected time-frames show cell #1 producing three sperate chromosome masses. Cell #2 undergoes a similar tripolar division. By contrast, cell #3 displays a normal bipolar organization. (C) Selected section from (A) shows additional cells (cell #4 and #5) at later time point. (D) These cells undergo a bipolar (cell #4) and tripolar (cell #5) mitosis. (E) Relative cell volumes shows that bipolar division produces similar sized cells, while tripolar divisions produced a smaller cell containing one chromosome mass and a larger cell with two masses. (F) Detailed analysis of multipolar division (from cell #1). Higher magnification views (second row) reveal a micronucleus. Third row shows chromosome masses and cortex revealing a binucleated cell (magenta, daughter #1) and a smaller mononucleated cell (green, daughter #2). Fourth row shows furrow dynamics. The furrow successfully cleaves the first daughter, but fails to cleave the remaining cell causing binucleation. (G) Additional time points show embryo during cavitation. The daughters from (F) are tracked and also shown to undergo aberrant mitoses. (H) Single section of blastocyst. Here, both daughter cells are detectable. (I) Final time points showing multipolar mitoses for cells tracked from (F) to (G). (J) Lineage tree identifying each cell based on division pattern and lineage position (18 hour window). (K) 3D views of the blastocyst corresponding to the lineage tree analyzed in (J). Scale, 20 μm.

Tracking the daughter cells revealed that they could subsequently enter mitosis (18 hours following the parental cell mitosis). Notably, both the binucleated trophectoderm cell (daughter #1) and the mononucleated ICM cell (daughter #2) went on to produce new multipolar mitotic events (Fig. 3G to I), suggesting the presence of supernumerary centrosomes (45, 46). Thus, such multipolar mitoses can be recurring, indicating that a defective mitotic apparatus can be inherited across cell generations. This was confirmed by the reconstruction of a lineage tree in which multipolar mitotic events perpetuated and could give rise to cells allocated to both the trophectoderm and ICM (Fig. 3, H to K).

Together, our live-imaging data captured distinct aberrant cell behaviors which caused compromised chromosome dynamics and nuclear integrity in human embryos. These include lagging chromosomes appearing during metaphase or anaphase producing micronuclei (Figs. 2, B and C), disrupted mitotic progression causing abnormal nuclear morphology (Fig. 2D), and multipolar spindle organization causing inheritance of abnormal nuclear structures, which can be recapitulated across a cell lineage (Fig. 3).

Capturing disrupted cytokinesis causing binucleated and enucleated cells

Previous work (15) and our initial analyses in fixed embryos (Fig. 1) also demonstrated the presence of cells with abnormal nuclear composition. As these defects may originate from aberrant cell behaviors during cytokinesis, we also focused on this phase of cell division. The SPY650-FastAct dye allows visualization of the contractile cytokinetic furrow assembled by actomyosin, which physically separates the two daughter cells by the end of division (47). Consistent with this, we could detect the formation of the furrow during anaphase and its ingression in telophase across 2D planes of a normally-dividing cell (Fig. 4, A). Moreover, quantification of relative cell volumes showed that the resulting daughter cells were similar in size, and displayed a cell nucleus localized close to the cell center of mass following cytokinesis (Fig. 4B).

Fig. 4. Errors in cytokinesis producing binucleated and enucleated cells.

(A) Live-imaging allows high-resolution analysis of cytokinesis. This example shows a normally-dividing cell. Left panels show the cells of intertest before and after division. The higher-magnification views (right panels) show selected time-frames of the dividing cell. Computational segmentation (middle row) and analysis of 2D planes (bottom rows) through the dividing cell reveal the ingression of the cleavage furrow, which eventually cleaves the two cells through the middle of the central spindle (i.e., between the two chromosome masses). (B) Spatial analysis of cleavage furrow and chromosome positioning within the dividing cell showing relative distance between the cytokinetic cleavage furrow (yellow thick line) and cell poles. (C and D) A similar live-imaging experiment reveals an abnormal cytokinetic event producing one binucleated (binucl.) and one enucleated (enucl.) daughter cell. In this case, the cleavage furrow ultimately ingresses off-center. Instead of cleaving the mitotic cell at the center of the two prospective nuclei, it cleaves the cell on one side of the spindle. (E and F) Additional example of an aberrant cytokinetic event in which the cleavage furrow ultimately ingresses off-center producing a binucleated and enucleated daughter cell. Scale, 20 μm.

By contrast, our live-imaging data also revealed cases where the cytokinetic furrow started to form with similar timing (30 ± 14 minutes after metaphase, n = 2 blastomeres from 2 embryos) to that of the normally-dividing cell (23 ± 4.8 minutes, n = 5 blastomeres from 3 embryos), but ingressed ectopically. In these cases, chromosome condensation and metaphase alignment initially appeared normal (Fig. 4C to F). However, the cytokinetic furrow formed away from the cell equator and subsequently cleaved the cytoplasm unequally, resulting in one daughter cell inheriting a larger cytoplasm with two nuclei, and the other inheriting a smaller cytoplasm devoid of chromosomes (Fig. 4C to F). These results indicate that, during cytokinesis, dysregulation of furrow positioning can lead to an uneven partitioning of cellular components producing binucleated and enucleated daughter cells.

Capturing fragmentation events originating during mitosis

The results described above demonstrate cases in which lagging chromosomes produced micronuclei (Fig. 2), multipolar mitoses caused inheritance and recapitulation of a malfunctional mitotic spindle apparatus (Fig. 3), and cytokinetic errors led to unequal inheritance of cytoplasm and chromosome masses (Fig. 4). Thus, we finally explored cell behaviors associated with fragmentation. While this phenomenon is frequent in human embryos (Fig. 1), it remains unclear whether fragmentation occurs during interphase or mitosis (12, 14, 25, 48). We first identified cells in which chromosomes became condensed and displayed a congression- and anaphase-like segregation. However, the chromosome masses subsequently shattered into multiple fragments and became dispersed throughout the cytoplasm during the following hours (Fig. 5, A and B), resembling those found in our analysis of fixed embryos (Fig. 1F) and in previous studies (15).

Fig. 5. Capturing cells undergoing chromosome dispersal throughout the cytoplasm and fragmentation.

(A and B) Live-imaging reveals cells in which condensed chromosomes undergo a mitotic congression and anaphase-like process. However, instead of establishing two distinct separate chromosome masses, the chromosomes become dispersed throughout the cytoplasm in a disorganized manner. (C) This embryo shows a similar behavior to (A) and (B). Chromosomes become condensed and undergo an initial congression-like process. However, they eventually become disorganized throughout the cytoplasm and display a fragmented or pyknotic-like morphology. The cell also undergoes fragmentation generating fragments either containing or devoid of DNA. (D and E) Additional examples of cells with aberrant chromosome dispersal throughout the cytoplasm followed by cell blebbing and fragmentation. Cellular fragments captured during the imaging window are segmented in magenta. Scales, 20 μm.

Furthermore, we identified another embryo in which a cell exhibited mitotic chromosome congression followed by a similar aberrant scattering of chromosome fragments throughout the cytoplasm (Fig. 5C). This cell eventually underwent cytoplasmic swelling followed by fragmentation (within 2 hours after mitotic entry) (Fig. 5C). Tracking the SPY555-DNA signals showed that some of the resulting cellular fragments inherited DNA, while others remained devoid of DNA (Fig. 5C). In a separate embryo, a similar aberrant cell behavior associated with fragmentation was detected (Fig. 5D). In this case, the chromosomes underwent abnormal dispersal throughout the cytoplasm, which was followed by cell blebbing and fragmentation (Fig. 5D). Finally, we could capture an embryo with numerous bleb-like fragments in the perivitelline space. Some of these fragments contained DNA and persisted for over 10 hours (Fig. 5E). Together, these results indicate that aberrant mitotic progression is associated with dispersal of chromosomes throughout the cytoplasm and can precede cell blebbing and fragmentation, producing both fragments containing DNA and devoid of DNA.

DISCUSSION

Morphological embryonic defects are well known markers of poor embryo quality, associated with lower implantation and live birth rates (19–23). Yet, prior to the application of live-imaging approaches it had remained unattainable to directly establish which specific abnormal cell behavior can give rise to each of these defects in real time, particularly during morula to blastocyst stages, when the embryo’s three-dimensional complexity increases significantly. The limited availability of human embryos for basic research also precludes extensive statistical analyses (4). Nevertheless, by combining live-imaging with computational segmentation, our study could capture in real time distinct aberrant cell behaviors producing defects in live embryos (Fig. 6).

Fig. 6. Summary of aberrant cell behaviors producing major defects in human embryos captured by live-imaging.

Schematic summary of the main aberrant cell behaviors causing defects in human embryos including: 1) Lagging chromosomes produced during the metaphase or anaphase can produce micronuclei; 2) Failure to transition to anaphase leading to disrupted nuclear shape and organization; 3) Multipolar mitoses can produce abnormally smaller cells and binucleated cells, which can further recapitulate the error in subsequent mitosis; 4) Aberrant regulation of the cytokinetic cleavage furrow causes binucleated and enucleated cells; 5) Aberrant dispersal of chromosomes throughout the cytoplasm can lead to arrested development or cell blebbing and fragmentation.

The data revealed that micronuclei and disrupted nuclear integrity resulted from lagging chromosomes and multipolar mitosis, while abnormally-nucleated cells (binucleated or enucleated) derived from cell divisions displaying failure in cytokinetic furrow dynamics. These results highlight cell division and cytokinesis as predominantly fragile steps of the cell cycle. Interestingly, despite such prominent errors in chromosome segregation or cytokinetic furrow positioning, these abnormal cells did not display extreme delays in the total duration of mitosis, compared to normally-appearing divisions. Current time-lapse microscopes used clinically can help to estimate the total duration of mitosis, but lack detailed sub-cellular information needed to assess chromosome organization throughout each mitotic phase. Thus, such microscopes may not readily capture some of these errors in chromosome organization or cytokinetic furrow positioning.

Our data also raise the question of why mitosis and cytokinesis may be more prone to errors in the early embryo, compared to somatic cells. Factors including oocyte age, accumulation of mutations, aneuploidies, freeze-thaw cycles, and culture conditions likely impose significant stress, which may render IVF embryos more fragile than those developing in vivo (49, 50). In addition, cell cycle checkpoints may be less efficient in preimplantation embryos than somatic cells (51). Furthermore, during preimplantation development the embryo undergoes an important transition from a meiotic to mitotic form of spindle organization and chromosome segregation. We also recently showed that mitotic fidelity during early development is initially aided by a network of actin cables regulating chromosome and spindle organization (52). Moreover, following embryo compaction cells start to adopt apical-basal polarity (8, 9, 53). Combined, all of these large-scale developmental changes may render the coordination between mechanisms regulating spindle assembly, chromosome organization, and cortex dynamics more error-prone in some cells of the embryo and along their progeny. Thus, the overall high rate of defects found in human embryos may result from a combined effect of stresses such as age, in vitro conditions, and physiological developmental transitions, all taking place prior to implantation.

It should also be noted that most of the aberrant cell behaviors recorded in our study occurred in a subset of cells of the embryo, while other cells could develop in an apparently normal fashion throughout the same imaging window. The extent of cells displaying distinct errors varied within each embryo, with 19 ± 13% binucleation, 20 ± 10% micronucleation, and 49 ± 25% blebbing. Establishing the longer-term impact of this variability in embryo development and implantation, or their specific impact to the cell lineages derived from trophectoderm and ICM precursors remains unfeasible due to the limited availability of human embryos, lack of human implantation assays, and restrictions on long-term embryo culture. Furthermore, while this study is based on the use of fluorescent dyes marking DNA and F-actin, implementation of dyes marking additional cell components may enable discovery of additional aberrant cell behaviors producing defects in IVF embryos.

In line with previous work in mouse embryos (54), we found that micronuclei were not immediately cleared, but remained within the cytoplasm. Similarly, binucleated cells did not undergo a stereotypical form of programmed cell death. Importantly, daughter cells resulting from a multipolar division could undergo a subsequent mitosis, recapitulating the anomalous multipolar organization of the previous mitosis. This suggests a conservation of an error-causing defect across the cell progeny. The recurrence of multipolar divisions suggests the presence of supernumerary centrosomes, yet this is paradoxical as supernumerary centrosomes can cause cell cycle arrest and centrosome disintegration (55). Plausibly, these control mechanisms may be less active or defective in preimplantation human embryos, or some cells of the embryo. Alternatively, multipolar spindles may result from premature centriole disengagement (56), caused by replication stress as reported in human embryos (57). We note that our analyses of fixed embryos also identified trinucleated cells. This condition may have arisen from an endoreplication of the cell with supernumerary centrosomes, yet thus far we have not captured this event in real time.

Together, these results indicate that human embryos have a higher inherent tolerance for defective cells, compared to somatic tissues and organs, in which defective cells are rapidly eliminated by cell death mechanisms to maintain homeostasis (58, 59). Retaining defective cells during early stages of human development may be beneficial to avoid disruption of embryo architecture and facilitate rapid expansion of the trophectoderm layer required for implantation, particularly as this extraembryonic lineage does not ultimately contribute to the fetus (60). Interestingly, this also implies an inherent capacity of human embryos to continue development despite accumulating defects, consistent with the regulative nature of mammalian development compared to the more deterministic development of other, non-mammalian species (61).

We also found a more severe case of aberrant chromosome dynamics in which cells entered mitosis, but the congressed chromosomes underwent fragmentation and dispersal throughout the cytoplasm. This could be followed by cellular blebbing and fragmentation. It will be interesting to investigate whether these dispersed chromosomes undergo some form of shattering, similar to the process of chromothripsis reported in cancer-associated cells (62–64). As spindle microtubules are key regulators of the cell cortex and cytokinetic furrow dynamics (65, 66), it is plausible that defects in spindle organization cause dysregulated cortex behavior triggering spatially ectopic ingression of the cleavage furrow. In addition, abnormal chromosome positioning near the cortex may induce localized blebbing or fragmentation, as DNA has a potent effect on cortex dynamics. This effect is observed during physiological processes including the DNA-induced cortex polarization found in the mouse oocyte (67), and after cytokinesis in the mouse embryo, when chromatin is transiently positioned near cortical poles triggering local F-actin clearance and cell blebbing (68). Unlike the mouse embryo, we show here and in previous work (41) that during cytokinesis in the human embryo, the newly-formed cell nuclei assemble near the cell center of mass, without reaching the cortical cell poles (Fig. 2A). However, in cases where chromosome dynamics and nuclear integrity are disrupted, the chromatin abnormally localized near the cortex may cause cortical dysregulation and blebbing (69). As the mechanisms driving cell fragmentation in the early embryo remain unclear, it will also be important to explore whether embryonic fragmentation shares some mechanistic basis with apoptotic blebbing, or represents a more uncontrolled process occurring only during preimplantation development. It also remains plausible that blebs and fragments may have some functional impact on development, as exposed recently in cultured cells (44). Given the scarcity of human embryos available for research, we do not exclude that some form of fragmentation may occur not only during or immediately following mitosis, but also interphase.

In conclusion, access to human embryos for research remains a critical factor limiting widescale analyses. However, our study provides direct examples of aberrant cell behaviors leading to morphological defects typically associated with poor implantation and live birth rates. Furthermore, direct live-imaging suggests that these mitotic and cytokinetic errors could represent the principal causes of preimplantation defects in human embryos.

MATERIALS AND METHODS

Human embryo samples

The experiments were performed at two locations including Boston IVF, Waltham MA, and Carlos Simon Foundation, Paterna, Spain. Surplus human embryos were donated for research following informed consent and guidelines by the New England institutional review board (WO 1–6450-1, USA) and National Commission of Human Assisted Reproduction (CNRHA, 2023–098, Spain) and the Ethics Committee of Hospital Clinico de Valencia (reference code 2023/098). Embryos were obtained at Boston IVF and the Centre for Reproductive Medicine (Next Fertility, Valencia, Spain). Samples were de-identified prior to thawing. Discarded embryos were determined as “Not Human Subjects Research.”

Live-embryo imaging

Vitrified embryos were thawed according to the manufacturer’s protocol using Vit Kit-Thaw (Fujifilm, Irvine Scientific, USA) and cultured for 1 hour in individual drops of 75 μl of Continuous Single Culture Complete (CSC) media with human serum albumin (HSA) (Fujifilm, Irvine Scientific, USA), covered with mineral oil in an incubator at 37°C, 7% CO₂ and 6% O₂ until further staining. Embryos were left to recover for 1 to 2 hours after thawing. Live embryos were stained for 1 hour with the live-cell probes SPY555-DNA (1:1,000) and SPY650-FastAct (1:2000), all from Spirochrome. Embryos were cultured in μ-Slide 8/18 Well Glass Bottom (Ibidi) at 37°C and 5% CO₂ in an incubator adapted for the microscope system (Leica) using a water Apochromat 40X 1.1 NA objectives and highly sensitive HyD detectors (Leica Stellaris 8 and Nikon A1RHD25). Confocal three-dimensional (3D) scans of the embryos were performed at 10 to 30 minute time intervals, except for Fig. 2C which was performed at 2 minute interval.

Staining of fixed embryos

For static analysis in fixed samples, embryos were fixed with 4% paraformaldehyde in DPBS-0.1% Triton X-100 for 20 min at 37°C, permeabilized in DPBS-0.5% Triton X-100 for 30 min, incubated in blocking solution (2% bovine serum albumin in DPBS-0.1% Triton X-100) for 1 h and incubated with Phalloidin-Alexa Fluor 488 (1:500, Invitrogen) and DAPI (Sigma) at 1:1,000.

Image analysis

2D and 3D visualizations of embryos were performed using Imaris 9.7 software (Bitplane AG). The manual surface rendering module was used for cell segmentation. Cell and nuclear volumes and sphericity were obtained from segmented data using the Imaris statistics module. The ortho and oblique slicer modules were used for analysis of virtual 2D planes (0.6 to 1.2 μm optical sections) and assessment of cytokinetic furrow ingression. Analysis of distances within single cells was performed in 2D virtual planes using the measurement point module. For this analysis, the longest distance was calculated between the chromosome masses and cell poles. For the estimations of the duration of mitosis, we quantified the time between prophase, or last frame of interphase to cytokinesis, which can be assessed by changes in chromatin compaction and cell shape.

Statistical analyses

Statistical analyses were performed in Excel and GraphPad Prism. Variables were analyzed using an unpaired, two-tailed Mann-Whitney U test.

Supplementary Materials

The PDF file includes:

Legends for movies S1 to S3

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S3