Peculiarities of the Mammalian Oocyte Cell Cycle

Abstract

Oocyte meiosis, the process of egg cell formation, requires a highly regulated cell cycle with many unique features compared to somatic cell division. On the journey to create a healthy embryo, this special cell carries a heavy responsibility and must navigate a remarkable number of complex challenges. Most oocytes will never complete this journey, less than 0.1% are ever ovulated, and fewer are viable. However, the few that do complete, manage by the execution of a series of extraordinary adaptations through two rounds of cell division. In this review we discuss some of these challenges and the adaptations that have evolved to mitigate them. This is not intended to be a comprehensive review of the cell cycle in oocytes meiosis, but to highlight some of the differences between oocyte meiosis and a typical mitosis. We discuss features that make this cell unique and the cell cycle regulatory mechanisms that support them. A salute to the few that make it and those that are sacrificed along the way.

Introduction

The mammalian oocyte is a highly specialized cell responsible for carrying female genetic material and for providing almost all the initial resources for embryonic development after fertilization. Accordingly, the production of a competent oocyte is a unique and tightly regulated process. Unfortunately, this is also a highly error prone process. In humans, up to two thirds of infertility cases include a problem of female origin (1,2), and though the underlying molecular mechanisms of many infertilities remain idiopathic, errors in oocyte meiosis are singled out as a main perpetrator (3). Cell cycle regulation in oocyte meiosis is particularly complicated, but by understanding the peculiarities of this process we are better able to understand infertility, a devastating disease and now a global health problem.

The World Health Organization recently reported that approximately 17.5% of the adult population experience infertility (4). Moreover, the rate of infertility is increasing, exacerbated by increasing maternal age (5–8). Many turn to Assisted Reproductive Therapies (ARTs) to bear children, and In-Vitro Fertilization (IVF) is normal in human reproduction. Indeed, predictions are that in many parts of the world up to 10% of children will soon be conceived through IVF (9). Yet IVF is not a magic wand. Like natural pregnancy, IVF can be limited by the quality of the fertilizing sperm cell, but more often by the quality of the oocyte. An oocyte must be produced that is both capable of being fertilized and that contains all the resources necessary to support the development of an early embryo. Indeed, while sperm cells are essentially differentiated to form a motile nucleus, oocytes develop a remarkably complex cytoplasm. They are unique in their ability to generate and store large amounts of maternal mRNAs, proteins, organelles and metabolic substrates. Their contents must both initiate and maintain the development of the early embryo, and this is by no means simple. Even in fertile couples, the success of planned natural conception is approximately 20-30% per ovulatory cycle (10,11). Moreover, in assisted reproduction, IVF success rates remain dependent upon the production of viable correct-stage oocytes, the number of which varies across age and fertility pathology (12,13).

Infertility and oocyte quality are not only human concerns but a threat to livestock industries. For example, poor bovine fertility is perhaps the most economically important animal health issue globally. In the UK alone, subfertility in dairy cattle incurs annual losses of approximately £500M and this problem is escalating. First service conception rates are currently less than 40% compared to 50% in the 1980s (14,15). A direct effect of subfertility is that many more animals are required to generate replacement animals. This drives down welfare (fewer resources per animal) and increases emissions (an estimated 11-19% of green-house emissions are produced by farm animals, largely by cattle) (16). IVF has emerged as a critical tool for bovine livestock maintenance and in the propagation of endangered species. However, while IVF is practical in these species (not least because it allows the transfer of genetic material across long distances), procedures are yet to match natural pregnancy success rates.

A clear problem in larger mammals is that many oocytes are not capable of developing into healthy offspring. Many different errors can produce an outcome that means either the oocyte does not reach a stage where it can be fertilized, or cannot support embryogenesis if it is fertilized. The molecular origins of these errors are often difficult to determine, but perhaps the best studied errors are aneuploid divisions - missegregation events that produce oocytes with incorrect numbers of chromosomes. Indeed, in humans, aneuploidy in the oocyte is the leading genetic cause of infertility, miscarriage, and congenital disorders in babies that survive to term (17). Even in younger fertile women, human oocytes exhibit a high baseline rate of aneuploidy of approximately 20%, increasing significantly with maternal age to approximately 80% by age 40 (18–20). Similarly, aneuploidy is estimated to be present in up to 30% of young bovine oocytes (21). By comparison, the incidence of aneuploidy in mouse oocytes is a mere 2–4% (22–24). Indeed, the disparity in the competencies of the same cell type in different organisms is itself peculiar.

In addition to aneuploid failures, it is important to understand that many oocytes do not reach a stage where they could become aneuploid, instead they arrest their cell cycle before an attempted division. There are many reasons why oocytes might stall and arrest in this way, but often due to a gross error that has no chance of resolution. This might be considered a failure (the oocyte will not produce offspring) but can also be considered biological success. An error has occurred, but the damaged cell is prevented from causing harm or forming an embryo that would ultimately be lost. In contrast, if an aneuploid oocyte is produced, we can consider this a two-step failure. First the oocyte had the propensity to missegregate chromosomes, and second, errors leading to missegregation were either not detected or there was no response to them. In this situation, not only has the cell cycle failed to support a healthy division, but it has also not prevented it. Consequently, aneuploidies are inherited by the embryo if the oocyte is fertilized. This is starkly contrasted in healthy mitosis, where irrevocably misaligned or missegregated cells are typically removed by apoptosis, so genome instability does not cause damage (see further discussion below).

Through all error types, an understanding of oocyte cell cycle control is paramount to understanding the origins of error. This in turn is critical to understanding the causes of female infertility and aneuploid miscarriages. Yet, despite the importance of this cell type, cell cycle regulation has been understudied in mammalian oocytes compared to mitotic cells. Thankfully, many recent advances in oocyte research are rapidly expanding our understanding, revealing unique regulatory mechanisms and remarkable differences compared to the mitotic cell cycle. In addition, beyond meiosis, the study of oocytes presents a broader opportunity; most cell cycle proteins in oocyte meiosis are the same as those in mitosis, but this does not mean they behave or are regulated in the same way. Oocytes thereby provide a different context system in which to interrogate alternative behaviors of canonical cell cycle proteins as they navigate the challenges of oocyte meiosis. For example, this is relevant to our understanding of tumorigenesis, where aneuploid divisions drive genomic instability. Furthermore, increasing numbers of studies now report the inappropriate and oncogenic reappearance of ‘meiotic proteins’ in cancer cells (25).

In this review we provide context to the oocyte cell cycle by first summarizing mammalian oogenesis and oocyte maturation, alongside some of the broad brushstrokes of cell cycle regulation. We will then highlight challenges that are unique to oocyte meiosis, and their cell cycle solutions that often appear compromised. Specifically, we will discuss cell cycle adaptations compared to mitosis, and features of oocyte meiosis progression that might expose the oocyte to non-recoverable division errors.

Oogenesis and prophase arrest

Oocyte production (or oogenesis; see Figure 1 for summary) begins in early fetal development with mitosis and the production primordial germ cells (PGCs). PGCs are one of the first lineages to be established in development, and are the precursors of both female and male gametes (26). In mammals, they originate outside of the embryo proper, before migrating to the genital ridge that in a male fetus will develop into testis, and in a female fetus the ovaries (27,28). From this stage, PGCs in the female (that later give rise to oogonia) colonize the developing ovary, dividing several times again by rapid mitosis before entry into meiosis (29) - a long single-iteration cell cycle that might last decades.

Figure 1. The cycle of mammalian female meiosis using human as an example.

Oocyte development begins with germ cell proliferation during fetal development, where many primordial germ cells are produced by mitotic divisions to form oogonia. Oogonia enter meiosis I (MI) and become primary oocytes that remain arrested in prophase I until they are periodically selected for ovulation from puberty. Following this, though each reproductive cycle, hormonal signals stimulate a cohort of primary oocytes to resume meiosis. Typically, only one completes MI, producing an oocyte arrested in meiosis II (MII) and a small polar body that eventually degenerates. Ovulation releases the MII oocyte from the ovary, which will only complete MII if fertilized. The completion of MII is marked by the extrusion of a second waste polar body, and female and male pro-nuclei form and fuse to create a single cell embryo or zygote. Mitotic divisions resume for embryonic development. The whole process then begins again in the ovaries of the fetus if female.

How meiosis initiation is triggered is not yet fully resolved (for review see (30)), but several important factors are known. For example, S-phase is considered to mark the mitosis-meiosis transition, coinciding with the activation of meiotic genes by STRA8 and MEIOSIN (31–34). Concurrently, external WNT signaling supports female GCs to become competent in the meiotic cell cycle (35,36). Up stream of these events, Retinoic Acid (RA) is a key signaling factor, and has been considered the trigger of meiosis by instructing GCs to express meiosis genes (37,38). However, recent genetic studies suggest RA is not the sole determinant of initiation (39,40).

Following the mitosis-meiosis transition, cells enter prophase I and are now referred to as oocytes. Entry into meiosis is not entirely synchronous (41), and is variable across species, but always within fetal life (42). For example, from approximately embryonic day 12.5 to 16.5 in mouse (43), and largely between the 9^th and 22^nd week of gestation in humans (44,45).

Over the following few days oocytes progress through substages of prophase I: leptotene, zygotene, and pachytene. This is a busy time. First in leptotene, chromosomes begin to condense, and double-strand breaks (DSBs) are deliberately introduced by the enzyme Spo11 (46). In zygotene, homologous chromosomes begin pairing, and a synaptonemal complex (SC) scaffold assembles between homologs. SC formation stabilizes homolog pairing and facilitates the repair of double-strand breaks (DSBs) through homologous recombination. Notably, this repair forms crossovers at selected DSB sites for the reciprocal exchange of genetic material - a major contributing element to the delivery of genetic diversity by meiosis (47–49). SC formation and homologous recombination complete in pachytene, and crossovers become visible as chiasmata as the SC disassembles in diplotene. Chiasmata then serve as the primary physical linkages that ensure homologous chromosomes remain connected until they are segregated in anaphase of MI (for review see (50)). Without these tethers homologs could segregate randomly, leading to aneuploidy. Finally, oocytes arrest with fully condensed chromosomes in diakinesis (or dictyate), an extended diplotene configuration, until just prior to ovulation (see Figure 2 for a schematic representation of chromosome pairing and meiosis divisions) (51).

Figure 2. Homologous chromosome pairing and chromosome division steps through meiosis I and meiosis II.

A. Meiosis I begins with homologous chromosome pairing during prophase I. Maternal and paternal chromosomes have been replicated, and sisters are bound together with cohesion rings. Each chromosome must align with its homolog to form a tetrad. B. SPO11 induces double strand breaks. DNA is then resected to produce single-strand overhangs, which invade the homologous duplex DNA and form Holiday junctions. C. The synaptonemal complex is built between homologous chromatids, forming a zipper like structure between maternal and paternal origin chromatids. The Holiday junctions are resolved as crossovers, leaving the homologs connected at chiasmata. NB for clarity only one crossover event is shown in the figure, but there will be multiple per chromatid. D. The resulting structure is a 4-chromatid tetrad, with cohesin rings on the arms now holding homologous chromosomes together. Tetrads then align at the metaphase plate. E. The first meiotic division is reductional. In anaphase I, homologous chromosomes are pulled to opposite poles, but sister chromatids remain attached. Cytokinesis follows, producing two haploid cells, each with duplicated chromosomes. F. The second meiotic division is equational. Like mitosis, chromosomes align individually at the metaphase plate and sister chromatids segregate to opposite poles in anaphase II. In male meiosis the outcome is four genetically distinct haploid cells. In female meiosis, the production of two polar bodies creates just one functional oocyte.

Given that oocytes are arrested from just a few days after entry into meiosis, this means two key points. First female mammals have all the oocytes they will ever ovulate present at birth^*1. Second, oocytes remain in prophase until at least the minimum age of sexual maturity, and at most until the end of female reproductive life. In humans, this range typically spans more than 4 decades and this is complex biology. Longevity without division is rare, most cells either divide or die. Maintaining this state without losing developmental potential is a major evolutionary challenge that requires tight cell cycle regulation. Critically this involves the suppression of CDK1 activity that will ultimately drive the oocyte out of prophase arrest at the correct time. To achieve this, high levels of cyclic adenosine monophosphate (cAMP) activate protein kinase A (PKA), and this is sustained by G-protein-coupled receptor 3 (GPR3) (52–54). PKA both activates and stabilizes WEE1/MYT1 inhibitory kinases, while inactivating CDC25B phosphatase to prevent it’s CDK1 activating role (for review see (55)). Importantly, to further ensure CDK1 activity inhibition, WEE1B (an oocyte-specific member of WEE1/MYT kinase family) and CDC25B and are compartmentalized in the nucleus and cytoplasm respectively (56). This is a critical kinase-phosphatase molecular balancing act; any signal disruption prior to the appropriate cue can lead to premature meiotic resumption or oocyte death. To support this, the translation of specific cell cycle proteins is repressed (including the CDK1 activator cyclin B1 (57)).

Importantly, oocytes are not isolated in the ovary, but rely on surrounding granulosa cells for signals that also promote arrest. Transzonal projections (TZPs), thin cytoplasmic extensions from cumulus granulosa cells, penetrate the zona pellucida of the oocyte to form gap junctions. These structures are essential for the exchange of signaling molecules like cAMP and cGMP (cyclic guanosine monophosphate, another second messenger molecule), to inhibit the phosphodiesterase enzyme (PDE3A) that would otherwise degrade cAMP (58,59).

Furthermore, TZPs are not only essential for communication to the oocyte, but for mechanical anchoring, and nutrient, ion and metabolite delivery. Moreover, bidirectional communication is important to coordinate granulosa cell proliferation, differentiation and follicular development (58). Indeed, from the moment the primary oocyte becomes enveloped in a primary follicle their fates are sealed together. This ensures synchronized development of the oocyte and its follicular environment but adds another layer of complexity that must be maintained for many months to decades depending on species.

The ovary’s job of producing enough oocytes with the level of fitness required for ovulation and fertilization competence is thought to be supported by the vast numbers it starts with. Not surprisingly, this number differs dependent on reproductive strategies and life span. For example, approximately 12,000 primary oocytes are formed by E15.5 in mouse, and approximately 7 million oogonia (precursors to primary oocytes) are formed by 20 weeks of gestation in humans (60–62). These are peak numbers before extensive waves of atresia begin reducing the pool available for folliculogenesis and ovulation. In human females, the number of primary oocytes is reduced to 1–2 million by birth, and to 300,000–500,000 by puberty (60,61). Indeed, it is a rarity that an oocyte makes it to the end of meiosis, and typically fewer than 500 will ever be ovulated. Notably, this same pattern of germ cell death exists across all mammalian species studied, with a strong wave of atresia during the formation of primordial follicles and then continuous loss until depletion (63). However, the function of mass atresia is only partially understood. Some have viewed it as a random process, exacerbated by nutritional deficits and other environmental factors. But others describe the elimination of germ cells with chromosomal abnormalities or defective mitochondria; a quality control process to promote the retention of only genetically intact and developmentally competent oocytes (64). This selective process would help prevent the transmission of mutations.

Another possibility proposed in mouse is that some cells die to serve a nursing function for surviving oocytes. This is supported by the biology of germ cell formation, growing in cysts of many cells together that maintain contacts. Only a fraction of the cells within a completed cyst (typically only one) survive to form mature oocytes (65). It is also proposed that the first waves of germ cell death helps synchronize follicle formation and ensures proper enclosure of the remaining oocytes by granulosa cells (66). Moreover, maintaining large numbers of metabolically active follicles would be energetically costly. Therefore, as well as removing potentially damaged cells, atresia likely conserves energy and maintains hormone equilibrium. In support of this theory, prophase arrested oocytes can engage effective repair mechanisms (specifically DNA damage repair by homologous recombination (67)). However, the preference following damage is instead apoptosis (mediated by TAp63) (68,69). Notably, as oocytes enter the growth stage, TAp63 expression reduces, favoring a switch to repair mechanisms (69,70).

Oocyte growth towards the end of prophase arrest

By the time a mammal reaches puberty, oocyte atresia has slowed but is still the fate of most cells. Many more oocytes begin preparation for ovulation than are ever released. For example, in humans, approximately 20 primordial follicles are recruited per ovulatory wave (each containing one primary oocyte) but typically only one is ovulated.

Primary follicles exit the ovarian reserve around four months before ovulation and grow through secondary, small antral, large antral, and preovulatory follicular stages. Similarly, the oocyte grows with the follicle over this same period. In humans for example, a typical oocyte increases from approximately 35 μm in diameter in a primary follicle to over 110 μm in diameter by the end of the growth phase (71,72), an increase in volume approximately 35-fold. Concurrently metabolic activity increases, and the surrounding pre-granulosa cells proliferate and differentiate. Typically, only one oocyte is ovulated, while the rest undergo atresia. This pattern is typical of mammalian ovulation at younger age. In older age, baseline reserves have diminished, fewer follicles are recruited per wave, and many oocytes will be of lower quality. The species can no longer afford the more stringent selection criteria of a younger animal (for review see (60)).

Through the growth stage, oocytes must accumulate sufficient mRNAs to drive progression through both meiotic divisions and into early embryonic development. Accordingly, as they prepare for meiosis resumption, oocytes become increasingly vulnerable to CDK1 activity that could override prophase arrest prematurely. To prevent this, an additional arresting activity becomes increasingly important, the destruction of cyclin B1. Cyclin B1 is the major regulatory component of CDK1 activity (cyclin B1:CDK1 activity), and thereby cyclin B1 levels must be kept below a threshold level to inhibit premature activation. This is achieved by the E3 ligase Anaphase Promoting Complex/Cyclosome (APC/C) and an APC/C activator, CDH1 (73). Critically, CDH1 must remain dephosphorylated to activate the APC/C, necessitating phosphatase activity. In this role, evidence in mouse demonstrates the importance of the phosphatase CDC14B (previously implicated in supporting APC/C^CDH1 activity in somatic cells (74)). Indeed, CDC14B-depleted oocytes prematurely escape prophase arrest due to elevated CDK1 activity, a phenotype that can be partially restored by CDH1 depletion. In contrast, CDC14B over expression delays meiotic resumption by reducing CDK1 activity and cyclin B1 levels (75).

As oocytes increase in size, they also undergo extensive nuclear and cytoplasmic reorganization. These changes first support the production and storage of proteins and mRNAs, and following this, transcriptional silencing. There are several key aspects of this transformation, including the accumulation of MARDOs (Mitochondria-Associated Ribonucleoprotein Domain) that regulate mRNA storage (76), and cytoplasmic lattices that provide a structural platform for maternal protein storage (77). Moreover, to support the additional production needs of growth, a transient compartment recently identified in mouse oocytes concentrates mitochondria, RNAs, ribonucleoproteins (RNPs), and endoplasmic reticulum (ER) around a single MTOC. This concentration boosts translation and protein synthesis, and this is important to avoid the cytoplasm becoming increasingly dilute in rapidly growing oocytes (78).

Towards the end of the oocyte growth stage, ribosome biogenesis shuts down and transcription is silenced resulting in the disassembly of the fine ultrastructure of an active nucleolus (79,80). Concomitantly, the nucleus of the oocyte undergoes a dramatic rearrangement in chromatin from a cloudy dispersion throughout the nucleoplasm (Non-Surround-Nucleolus; NSN), to a densely packed sphere around the nucleolus (Surround-Nucleolus; SN). The difference between SN and NSN nucleolar chromatin state was first evidenced in mouse in 1995 (81), and then in human in 2002 (82) and is highly indicative of competence. Specifically, while oocytes with either chromatin arrangement can resume meiosis, they do not have comparable developmental potential (83–86); embryos derived from NSN oocytes often halt development at the two-cell stage, while embryos from SN oocytes are capable of developing to the blastocyst stage and full term. Full size, SN, GV stage oocytes are thereby the goal for meiosis resumption (see Figure 3 for schematic representation of oocyte growth and meiosis resumption) (87).

Figure 3. Murine oocyte nuclear and cytoplasmic reorganisation from growth initiation through to spindle assembly.

A) A small primary oocyte at the start of the growth phase, chromatin (purple) is mostly decondensed in the nucleus with (usually) a single large nucleolus. B) As the oocyte starts to rapidly grow, a transient structure supports high levels of translation by concentrating mitochondria, RNAs, RNPs, and ER around an MTOC (green). Cytoplasmic lattice structures (white) store large amounts of protein for future use, including epigenetic reprogramming factors for the embryo. C) A fully developed oocyte ready to be ovulated. Chromatin has largely condensed around the nucleolus to form a structure called the karyosphere, although some remains in contact with the nuclear membrane. The cell is transcriptionally silent, and MTOCs can be found around the nucleus and in the cytoplasm. Protein aggregates that have built up and would otherwise pose a threat to the health of the oocyte are accumulated in endolysomal vesicular assemblies (orange). mRNAs are stored in a membraneless mitochondria-associated ribonucleoprotein domain around the nucleus. D) The nuclear envelope breaks down, the nucleolus decondenses and shrinks, and chromatin condenses further into individualised chromosomes. Some MTOCs fragment and migrate to form what will become the spindle poles. E) Spindle formation is supported by the concentration of microtubule regulatory factors at the poles by a TACC3-based phase-separated compartment (the liquid-like spindle domain).

Meiosis resumption and oocyte maturation

In mitosis, prometaphase immediately follows prophase without interruption. In the oocyte however, this cell cycle transition requires a trigger, and importantly, a trigger that is linked to the female reproductive cycle. This is provided, by a surge in luteinizing hormone, which reduces cGMP levels in the surrounding granulosa cells of select oocytes. This allows PDE3A to degrade cAMP, resulting in a drop in PKA activity that permits CDC25B activation and thereby cyclin B1:CDK1 activation (55,88). Importantly, the mechanism of CDC25B activation relies on its relocation, shuttled into the nucleus prior to Germinal Vesicle Breakdown (GVBD), while WEE1B is exported to the cytoplasm (56). Indeed, these relocation events occur at a time that roughly coincides cyclin B1:CDK1 translocation to the nucleus (15–20 min before GVBD in mouse oocytes (89)). Alongside this, MIS12 (a component of the kinetochore complex) is thought to suppress CDC14B activity, inhibiting APC/C^CDH1 directed cyclin B1 degradation (90). Together, these mechanisms co-ordinate meiosis resumption and GVBD, triggering a cascade of events that drive the oocyte through meiosis I (MI) to eventually arrest again in metaphase of meiosis II (MII) until fertilization.

Briefly, GVBD in oocytes (akin to nuclear envelope breakdown in mitosis) initiates prometaphase I and chromosome alignment. Like mitosis, this involves the formation of a bipolar microtubule spindle that aligns chromosomes on a metaphase plate in preparation for segregation in anaphase I. However, in this first division, homologous chromosomes are aligned (rather than sister chromatids in mitosis), with each homolog attached by kinetochores to microtubules emanating from opposite spindle poles (Figure 2E).

Like mitosis, chromosome alignment in meiosis is critically supported by the activity of CDK1, mediated by increasing cyclin B1 binding and CDK1 phosphorylation. Importantly, CDK1 is only active when bound to a cyclin partner that bestows the capacity to phosphorylate and regulate upwards of 100 substrates (91,92).

Once chromosomes are aligned in MI, the loss of CDK1-bound cyclin B1 is equally important to drive anaphase. This is achieved by the destruction of cyclin B1, triggering the poleward movement of microtubules and the segregation of chromosomes to complete anaphase I (93–95). Cyclin B1 destruction is again mediated by the APC/C, but now directed by a second activator, CDC20.

Critically, the APC/C is not just important for cyclin B1 regulation, but a key component of networks regulating many cell cycle proteins by allowing their accumulation and timed destruction at transition points (96). For example, another important target is securin, an inhibitory pseudosubstrate of the separase enzyme that cleaves the cohesins holding chromosomes together (97–101). It is crucial that separase is inhibited while chromosomes are aligning, but equally important that removal of separase inhibition is timely so that chromosomes only cleave in anaphase. Indeed, cyclin B1 and securin destruction targeting are coupled, ensuring separase activation and the poleward movement of chromosomes are synchronous (102,103).

Accordingly, it is essential that the APC/C only initiates substrate destruction at the correct time relevant to each cell cycle stage. Here the spindle assembly checkpoint (SAC) plays a key role in regulating APC/C activity by generating a ‘“wait anaphase’ signal at unattached kinetochores (Figure 4). The SAC exerts this effect by catalyzing the formation of a multiprotein complex comprising of MAD2, BUBR1 and BUB3 proteins, and the APC/C activator CDC20. From its discovery, this complex is known as the mitotic checkpoint complex (MCC), but this same complex forms in oocyte meiosis and directly sequesters both APC/C-bound and free CDC20 (104). In principle, while MCC bound, the APC/C^CDC20 is unable to form the co-receptor that accepts a destruction box motif (D-box) present in both cyclin B1 and securin (105,106). This inhibits anaphase robustly in healthy mitosis (a single unattached kinetochore generates sufficient MCCs to prevent all chromatid pairs from segregating) (107). Subsequently, as kinetochores attach to microtubules under tension, MCCs are no longer produced. This signals the completion of chromosome alignment and the inhibitory hold on the APC/C^CDC20 is appropriately lifted (106). However, though the SAC is active and essential in mammalian oocyte meiosis, it is not as effective, and this will be discussed below.

Figure 4. A simplified summary of SAC activity and APC/C targeting in mitosis and meiosis.

The SAC is a crucial surveillance mechanism assisting accurate chromosome segregation in mitosis and mammalian oocyte meiosis. It becomes active in prometaphase as chromosomes begin to attach to spindle microtubules via their kinetochores. SAC activity promotes the high levels of cyclin B1:CDK1 activity necessary to align chromosomes, and keeps the enzyme separase inactive. In mitosis, proper kinetochore-microtubule tension and bipolar attachment are required before progression to metaphase. If kinetochores are not correctly attached, the SAC halts cell cycle progression by inhibiting the APC/C. This action is directed by mitotic checkpoint complexes (MCCs), which form at unattached kinetochores and inhibit APC/C activity by binding to its activator CDC20. This prevents the degradation of cyclin B1 (the regulatory subunit of cyclin B1:CDK1 activity) and securin (an inhibitor of separase), and the cell remains in metaphase. Once all chromosomes are properly aligned and attached, MCCs disassemble, allowing the APC/C to ubiquitinate securin and cyclin B1 driving anaphase. This system robustly inhibits aneuploidy in healthy mitosis. However, while the same mechanism operates in oocyte meiosis, where chromosomes misalign, aneuploidy is often not inhibited.

On release from SAC inhibition anaphase is triggered. In MI, anaphase is known as the reductional division; one of each homologous chromosome pair is retained in the oocyte while the other is extruded in a first polar body (PB1; an extreme example of an asymmetric division). Notably, one of either the paternal or maternal origin homologs is lost, although extensive crossover-based recombination ensures a contribution from both parents regardless.

Unlike mitosis, there is no interphase, S-phase or reformation of the nucleus. Instead, almost immediately the oocyte enters MII; cyclin B1:CDK1 activity and securin levels increase for a second time, driving chromosome alignment and separase inhibition (108). High levels of cyclin B1:CDK1 activity stabilize the MII spindle with sister chromatids aligned for segregation (109). However, despite alignment, the cell cycle arrests, and anaphase is prevented by an APC/C inhibitor, EMI2. MII arrest is unique to vertebrate oocytes and termed Cytostatic Factor (CSF) arrest. CSF ensures that an oocyte does not proceed until fertilization, thereby synchronizing the completion of meiosis with the entry of sperm. Sperm entry activates the cell cycle once again with a series of intracellular calcium oscillations, activating CaMKII to direct the degradation of EMI2. This cascade results in cyclin B1:CDK1 inactivation and the oocyte completes anaphase II with the extrusion of a second polar body (PB2) (110–112). This second segregation is referred to as an equational division and is more analogous to mitosis. Subsequently, a female pronucleus forms, able to merge with the male pronucleus and initiate embryonic development.

Above are some of the broad brushstrokes of important mechanisms that regulate MI and MII progression. The following sections will discuss in more detail several of the most challenging differences between mitosis and mammalian oocyte meiosis. How the cell cycle copes with these peculiarities, and how they evidence vulnerabilities to division errors.

Assembling a microtubule spindle in a large cell volume

Compared to a somatic cell of 10 μm diameter, a human oocyte (115 μm diameter) is approximately 1000-fold greater in volume. This size is crucial for supporting the developing embryo after fertilization. Several rounds of division take place in the same volume without significant external contribution. Thereby, through MI and MII the oocyte expels as little cytoplasm as possible with its waste genetic material.

Assembling a microtubule spindle and aligning homologous chromosomes in this volume is complex. First, mammalian oocytes do not carry centrosomes, an integral spindle assembly tool in mitotic cells. In mitosis it is widely accepted that a ‘search and capture’ mechanism operates using the rapid growth and shrinkage of microtubules emanating from centrosomes to ‘search’ the cytoplasm, probing for kinetochores on chromosomes (113,114). Once a microtubule encounters a kinetochore, it forms a stable kinetochore-microtubule attachment in a ‘capture’ event, supported by motor proteins (e.g., dynein, kinesin) to aid chromosome positioning. In the relatively small volume of a mitotic cell this process is very efficient, and prometaphase (the chromosome aligning period) typically takes 10-15 minutes (114).

In contrast in oocytes, the role of centrosomes is replaced by mechanisms that vary even between mammals, with many independent Microtubule Organizing Centers (MTOCs) taking this role in mice (115), and the chromosomes themselves nucleating microtubules in a Ran GTPase-dependent manner in humans (116). Importantly, in all oocytes studied to date, spindle assembly is thought to require the concentration of microtubule regulating factors at key sites. In human oocytes, these factors are concentrated before GVBD in a TACC3-dependant manner by the huoMTOC (human oocyte MTOC). This structure later fragments upon GVBD and localizes to kinetochores (the sites of spindle microtubule nucleation in these cells) (117). In contrast, in the case of species not known to nucleate spindle microtubules at kinetochores (such as murine, porcine and ovine), an analogous role is performed by a TACC3-defined phase-separated compartment called the Liquid-like Spindle Domain (LISD). The LISD concentrates microtubule nucleation and regulation factors at spindle poles (118).

Over many hours of prometaphase I (6-7 hours in mouse and approximately 16 hours in humans (116)), nucleated microtubules establish a microtubule spindle with two independent and opposite poles. However, a significant number of initial kinetochore microtubule attachments are incorrect. For example, in mouse oocytes, quantification suggests that close to 90% of all chromosomes undergo at least one, and up to 6 rounds of ‘error correction’ (kinetochore-microtubule depolymerization and reattachment) before achieving correct biorientation (119). Similarly, Schuh et al (2015) reported that 80% of observed human oocytes transiently loose bipolarity at least once during spindle assembly, suggesting these alternative assembly methods strongly correlate with bipolar spindle instability, and in turn chromosome segregation errors (120). Indeed, human oocytes spindle assembly is likely further challenged by a deficiency in at least one key molecular motor that stabilizes spindles in other mammalian oocytes (121).

Furthermore, depolymerization of kinetochore-microtubule attachments and subsequent reattachment, was observed to take up to 50-minutes even in mouse oocytes (119). Meaning that a single round of error correction may last longer than the full duration of prometaphase in most mitotic cells. Like mitosis, error correction is dependent on aurora kinase, the catalytic subunit of the Chromosome Passenger Complex (CPC; a multi-function regulator of chromosome alignment, kinetochore-microtubule attachments, cohesion, and the spindle assembly checkpoint (122)). However, while aurora B acts as the essential subunit in mitosis, a germline specific aurora C homolog takes on this role in mouse oocyte meiosis (123,124). Importantly however, not all kinetochore-microtubule attachment errors are detectable in oocytes, and this will be discussed later.

In addition to a microtubule spindle, mammalian oocyte chromosome segregation takes place within an actin skeleton (125). Meiotic spindle F-actin stabilizes kinetochore-bound microtubules to promote accurate chromosome segregation. While this is not strictly a cell cycle peculiarity, this feature supports cell cycle progression and is in stark contrast to mitotic chromosome segregation. Removal of F-actin in mouse oocytes results in the premature segregation and scattering of sister chromatids, particularly in aged oocytes (126). Indeed, a recent study identifies a novel pool of F-actin that localizes directly to the meiotic spindle in mouse oocytes. This challenges the traditional view that microtubules alone regulate spindle assembly and organization. Indeed, F-actin is observed to form a cage-like structure around polar microtubule organizing centers (pMTOCs), which if disrupted leads to unfocused spindle poles and chromosome missegregation (127).

Spindle migration for polar body extrusion

Both MI and MII divisions produce a large cell and a much smaller polar body (for example PBs in human oocytes are approximately 1% of the original oocyte volume (128)). Critically, this asymmetry maintains cytoplasmic volume, and preserves maternal factors that later become essential through the oocyte-embryo transition (129–131). This is complicated by the central location of spindle formation (dictated by prior GV positioning). Consequently, a period of migration is necessary to ‘off-center’ the spindle.

In mitotic cells, astral microtubules play a large role in positioning the spindle (132). However, oocytes lack traditional astral microtubules that might facilitate positioning, and instead several unique mechanisms support spindle migration. These include a dynamic meshwork of F-actin able to transmit forces over long distances to position the spindle (133), and a chromatin-driven Ran-GTP gradient that acts as a spatial cue to polarize the cell (134).

Best evidence in mouse oocytes demonstrates that straight (formin-2, spire 1/2) and branched (ARP2/3) actin nucleators play a role in controlling spindle positioning in mouse oocytes (135–138) (for review see (139)), and an actin mediated ‘cytoplasmic streaming’ activity is essential to both maintain the direction of MI spindle migration, and to promote spindle anchoring (140,141). Importantly, as the spindle polarizes, a thick F-actin cap forms at the cortex towards the leading edge of the approaching spindle, marking the future site of PB1 extrusion. Evidence in Xenopus suggests this is driven by the accumulation of CDC42–GTP (a well-known polarity effector) that recruits the ARP2/3 complex to the cortex to nucleate branched F-actin. In these cells, F-actin then acts to generate the pushing force required for membrane protrusion and PB formation (142). Given the conservation of CDC42-ARP2/3 signaling across species, this same mechanism is likely to operate in mammals. Importantly, CDC42 is recruited to the same site in preparation for PB1 extrusion in mouse, and disruption of CDC42 activity prevents asymmetric division (143). Finally, cortical softening is crucial for proper spindle positioning and PB1 extrusion (144,145), Notably this is in direct contrast to somatic cells known to stiffen as they round up in mitosis.

In addition to F-actin mediated spindle migration, recent evidence in mouse demonstrates that a subset of MTOCs contribute to spindle positioning, acting ‘like’ astral microtubules that anchor spindles in mitotic cells (146). Previously, oocyte MTOCs have been observed to nucleate astral-like microtubules, but alone they are too short to extend to the cortex from a central spindle location (115,147,148). Instead, a subset of cytoplasmic MTOCs (mcMTOCs) solve this problem in MI by bridging the gap (149). Specifically, short astral-like microtubules reach mcMTOCS, which then provide connection to the cortex. Critically, mcMTOCs localize asymmetrically, antipodal to the site of F-actin enrichment and subsequent PB1 extrusion. The authors propose a model in which the role of F-actin driving oocyte spindle migration, is balanced by forces exerted from mcMTOCs, supporting timely migration and accurate positioning (149).

Following PB1 extrusion, the oocyte almost immediately enters MII, and the second spindle assembles parallel to the cell membrane, near the PB1 extrusion site. Notably, positioning of the acentrosomal MII spindle is also dependent on CDC42 and the ARP2/3 complex (150,151). Between MI and MII, the spindle apparatus remains anchored to the cortex. Certainly in mouse, MII spindle assembly is almost immediate, likely due to the rapid increase in cyclin B1:CDK1 activity levels (152). At this point the oocyte is released from the follicle into the oviduct to await fertilization. Subsequently, MII arrested oocytes remain actin-rich with a continuous flow of actin nucleation to maintain the cortical positioning of the spindle (153).

Less is known about the microtubule dynamics that reassemble an MII bipolar spindle from the monopolar assembly retained on MI exit. However, unusually high levels of the cyclin-dependent kinase regulator cyclin A2 are known to be important for spindle stability (154), and the spindle-associated proteins MISS and DOC1R are specifically essential for MII spindle architecture (155,156). Interestingly, the MII spindle is distinct from MI in its requirement for Ran-GTP in anaphase. MI chromosome segregation is relatively normal in Ran-inhibited oocytes, but MII spindles are defective and highly prone to segregation errors (157).

Aligning chromosomes in oocyte meiosis with a compromised SAC

Particularly in MI, the complexity and duration of prometaphase leaves the oocyte highly vulnerable to errors in division. Like in mitotic cells, chromosome segregation errors are limited in mammalian oocytes by the activity of the SAC monitoring the attachment of chromosomes to microtubules (104,158). However, unless damage or misalignment is severe, it functions less as a checkpoint and more as a gentle brake (159–162).

To date, all mitotic SAC components studied in oocytes localize as expected at unattached kinetochores in early MI (22,160,163–170). Accordingly, in several of these examples, the loss of SAC components accelerates MI and/or permits chromosome missegregations. In addition, exposure to even low dose spindle poisons triggers the recruitment of SAC components to kinetochores (171). Initially, this stabilizes cyclin B1 and securin protein levels, delaying PB1 extrusion (152,172). However, these stimulated arrest points are often only transient, and oocytes eventually missegregate their chromosomes and exit MI. The SAC is unable to respond to a small (or even moderate) number of misaligned (and therefore presumably mis-attached) chromosomes (158). Indeed, in mutant oocytes that fail to establish stable bipolar chromosome-microtubule attachments, anaphase still takes place (173).

One suggested reason for the lack of a robust SAC response is the volume of the oocyte. The function of the SAC relies on the formation and rapid diffusion of MCCs which are short-lived in the cytoplasm. Accordingly, the strength of the SAC is then determined by the rate of MCC formation (controlled by the number of unattached kinetochores) and the volume over which MCCs must diffuse (174,175). Strong evidence for this is presented in C. elegans embryogenesis where SAC strength gradually increases over multiple divisions as cell size decreases (175). Effective checkpoint signaling is thereby likely problematic in large cells with just one or two unaligned chromosomes.

Significantly, Kyogoku and Kitajima 2017 (176) manipulated the volume of GV mouse oocytes to either double or halve their original size, revealing that decreased cytoplasmic volume produced spindles with better-focused poles and enhanced SAC stringency. In contrast, the reverse was true in oocytes with increased volume. Interestingly, the spindle volumes in these manipulated oocytes scaled almost perfectly with cell volume. The authors proposed a two-pronged impact of oocyte volume on meiotic integrity in mouse oocytes. First, that the ability of MTOCs to form a stable bipolar spindle is compromised by a larger volume (unstable bipolar spindles and severe misalignment errors fail to inhibit anaphase in human oocytes (116)). Second, that an increased cytoplasmic volume dilutes the pre-formed nuclear factors (such as MCCs) necessary for anaphase inhibition following GVBD (177). This is compelling, yet in some cases there is little correlation between oocyte size and meiosis fidelity between organisms, other factors are clearly important. Whether the effect of size will be more apparent in human oocytes (approximately three to four times greater in volume than mouse) and whether the same principle causes the SAC to strengthen during early embryonic development remains to be determined. Notably, a high frequency of errors is also permissible through early divisions in human embryos (178).

Generating sufficient MCCs over volume is only one aspect of checkpoint function challenged by oocyte meiosis. A second problem is the duration of prometaphase I. Even in mouse oocytes, the duration of prometaphase I is longer than a robustly activated checkpoint can hold many mitotic cells (Figure 5). Mitotic cells in persistent SAC arrest eventually either ‘slip’ out of mitosis (caused by slow degradation of cyclin B1 (179)), or trigger apoptosis before exit. In the case of slippage, the outcome is often still post-mitotic cell death in healthy organisms (180–182), removing potentially damaging cells. In the case of death in mitosis, it is understood that pro-apoptotic signals gradually increase while the cell is SAC active. Eventually, a threshold of apoptotic signaling is reached sufficient to initiate cell death, again removing potentially damaging cells. Therefore, whether or not a cell dies in mitosis depends upon whether the apoptotic threshold is breached before exit from mitosis - the ‘competing-networks’ model (183,184). Yet neither of these scenarios take place in the oocyte. On the one hand this might appear to promote the delivery of a compromised genome to offspring. However, a buildup of apoptotic signals is incompatible with the drawn-out process of chromosome alignment in mammals. A prolonged prometaphase I is a feature of every healthy mammalian oocyte.

Figure 5. Relative timeline of cyclin B1 accumulation and destruction between mitosis and MI in mouse oocytes.

Lines reflect a typical healthy mitosis (green line), mitosis that stalls due to persistent checkpoint activation (red line), and a typical healthy mouse oocyte MI (blue line). If arrest in mitosis persists, the SAC cannot hold the cell indefinitely (red line). This situation indicates an insurmountable problem. The desirable outcome is that these cells do not divide, and either ‘slip’ for later demise or undergo apoptosis. To ensure this, pro-apoptotic signals accumulate over the arrest time and eventually trigger cell death if the cell does not achieve anaphase in time. Thereby, mitotic cells with persistent errors (that might lead to aneuploidy and cancer) are eliminated. In this model, two opposing molecular networks operate simultaneously: one promotes mitotic progression, and the other initiates cell death. However, in mammalian oocyte meiosis, apoptotic signaling is suppressed. It is not difficult to see why this might be necessary. All oocytes spend longer in prometaphase than would be tolerated in mitosis.

These example timings are based on our own work in mouse oocytes from GVBD to PB1 extrusion (blue line), and MEF (Mouse Embryonic Fibroblast) cells for same species comparison (258). Green line is the average time a MEF takes from nuclear envelope breakdown to anaphase/cytokinesis, red line is the average time a nocodazole arrested MEF takes to exit mitosis (judged by returning from a rounded mitotic shape to interphase flattening).

A third feature to challenge the SAC in oocytes is their potential for inappropriate microtubule-kinetochore attachments, and this will be discussed below.

Erroneous kinetochore-microtubule attachments do not reliably inhibit anaphase in MI oocytes

Considering the above, it is perhaps not surprising that erroneous kinetochore-microtubule attachments do not always inhibit anaphase in MI oocytes (7,23,185,186). Experiments in mouse oocytes support this, and suggest that kinetochore attachment rather than alignment dictates SAC satisfaction (23). For example, on ablation of the protein NuMA (Nuclear Mitotic Apparatus), oocyte homologs formed attachments, but with alignment defects and a lack of tension. In these oocytes the SAC was silenced and anaphase forged ahead, but with severe aneuploidies (187).

Indeed, it is widely accepted that the structure of MI chromosome pairs (bivalents) presents a substantial complication for error sensing (158). Pairs of sister chromatids must remain coupled while homologs separate. Two functionally separate sister kinetochores must connect to microtubules emanating from the same pole (contrary to either mitosis or MII), while the kinetochores of the opposing homologous pair must connect to the opposite pole. This situation all but excludes the interkinetochore tension which is thought to aid SAC silencing in mitosis.

Briefly, the favored model of tension sensing in mitosis describes that while chromosomes are aligning, aurora B kinase at the inner centromere, phosphorylates kinetochore substrates to prevent the stabilization of microtubule attachments. Once bi-orientated attachments are formed (i.e., kinetochores are attached to opposite spindle poles), this creates tension across kinetochores, spatially separating aurora B is from its kinetochore substrates. The result is reduced phosphorylation, which stabilizes attachments. In the case of most incorrect attachments (e.g., syntelic or merotelic), aurora B remains close to and phosphorylates kinetochore substrates, destabilizing these attachments to allow a correction event (188).

In contrast, MI oocytes cannot operate this same spatial separation mechanism due to the tetrad structure of meiotic chromosomes. The kinetochores of sister chromatids remain paired during MI anaphase, meaning that even after bivalent stretching, aurora B/C remains close to the kinetochore-microtubule interface. This proximity causes persistent phosphorylation, which destabilizes even correct attachments. This situation necessitates an alternative mechanism to eventually dephosphorylate kinetochore substrates and stabilize attachments. PP2A-B56 phosphatase takes on this role in mouse oocytes, recruited gradually, dependent on BUBR1 phosphorylation by CDK1 (182). However, this mechanism has limited feedback; the gradual increase in CDK1 activity is described as a timer, meaning that PP2A recruitment is coupled to meiosis progression regardless of tension (190). This lack of coordination between bivalent stretching and kinetochore phosphoregulation inhibits the distinction between correct and incorrect attachments (189).

Similarly, in MII, merotelic attachments (the bi-directional connection of a single kinetochore to both poles) still satisfy the SAC (chromosomes are attached and under tension), despite not being in the correct orientation for accurate segregation (191).

Given above, it not surprising that aneuploid cell divisions are more prevalent in mammalian oocytes than in other cell types. Compounding this, it is likely more difficult for the SAC to distinguish correct from incorrect alignment in ageing, where the spatial separation of sister kinetochores increases over time, contributing to the age-associated increase in segregation errors (192,193).

Mechanisms assisting the SAC in preventing chromosome segregation errors

Given the duration of prometaphase I, mammalian oocytes are vulnerable to premature APC/C activity for several hours before chromosomes align in MI. However, despite their prolonged prometaphase mouse oocytes are rarely aneuploid. In these cells two mechanisms are thought to help extend prometaphase, compensating for a less effective SAC in oocytes. The first mechanism is a reversal of the order of APC/C co-activator binding compared to mitotic cells. CDC20 is the APC/C co-activator that will ultimately drive MI exit by targeting cyclin B1 for destruction. However, in prometaphase I oocytes, APC/C^CDH1 is active first to target CDC20 itself. APC/C^CDH1 activity thereby acts as a brake, preventing early accumulation of APC/C^CDC20, slowing down the first half of prometaphase while the spindle is forming (194). Eventually however, it becomes necessary to switch to APC/C^CDC20 activity, and a gradual exchange is made later in prometaphase I in preparation for MI exit. Now with increasing APC/C^CDC20 activity, the oocyte becomes vulnerable to premature cyclin B1 degradation. Indeed, as the SAC’s inhibitory hold over the APC/C reduces, cyclin B1 and securin degradation initiate in oocytes before all chromosomes are correctly attached. In mitosis this would have catastrophic consequences, but in mouse oocytes, where an additional mechanism slows down the second half of prometaphase I, this rarely leads to aneuploidy.

In contrast to mitosis, cyclin B1 is expressed in excess of CDK1 in mouse oocytes (152,195). This alone indicates distinct regulation of cyclin B1:CDK1 activity in meiosis compared to mitosis. Evidence further suggests that non-CDK1-bound and CDK1-bound cyclin B1 have different temporal degradation in mouse oocytes, whereby non-CDK1-bound cyclin B1 is degraded ahead of the bound form. This means some degradation can initiate before chromosomes have fully aligned, but without negative consequence. By existing in excess, cyclin B1 does not become limiting until later in prometaphase I and cyclin B1:CDK1 activity is prolonged (152). This extends the opportunity for chromosomes to complete alignment and error correction, compensating for a weakened SAC. Following this, the reduced pool of CDK1-bound cyclin B1 is degraded rapidly. By this mechanism, cyclin B1 can be degraded in prometaphase, but by a mechanism that inhibits rather than promotes division errors.

In addition to this, a similar strategy exists to remove securin, which is also present in excess of its binding partner separase (172). Notably, securin is not essential in MI oocytes (196) (other separase inhibitors exist and this will be discussed below), but nevertheless it must still be removed to release separase (102). To facilitate this, we suggest that non-separase-bound securin degradation initiates ahead of metaphase and ahead of separase-bound securin. This prevents a situation where securin levels might exceed the amount an oocyte can remove in metaphase alone (172). It is plausible that the removal of the non-bound forms of cyclin B1 and securin in prometaphase ensure the APC/C is not overloaded in the metaphase-anaphase transition. CDK1-bound cyclin B1 and separase-bound securin are thereby destroyed simultaneously, and cyclin B1:CDK1 inactivation and separase release remain coupled.

Whether free-then-bound destruction ordering mechanisms also operate in other mammalian oocytes remains to be determined. This will be particularly interesting in the oocytes of species that suffer high rates of aneuploidy. It is also possible that other cyclins are subject to similar targeting pattens. Of particular interest is cyclin A2 where degradation also occurs in two waves in oocytes. Most cyclin A2 is degraded in prometaphase I, but a small fraction is localized to kinetochores and escapes targeting, later degraded in anaphase I (197).

A further critical protein with a late destruction timing in oocyte MI is cyclin B3 (198,199). In mammalian oocytes, cyclin B1 is widely accepted as the principal B-type cyclin driving prometaphase, with some CDK1 activity promoting roles that can be compensated for by cyclin B2 (200)^*2. In contrast however, the redundant functions of cyclin B1 and B2 are not shared by cyclin B3, which only functions in the gametes of female vertebrates. Critically, cyclin B3 knockout mice are viable, and males are fertile (201), but in contrast females are infertile (199). The oocytes of these mice achieve MI alignment, but cyclin B1 and securin are stabilized rather than degraded and oocytes fail to exit metaphase. This suggests that high levels of cyclin B3 are necessary to activate the APC/C and promote anaphase - the opposite of high levels of cyclin B1:CDK1 activity that prevent anaphase (198). In mouse oocyte MI, cyclin B3 is then degraded in anaphase I (198,199), after cyclin B1, but also in an APC/C-directed manner. Cyclin B3 mRNA then decreases substantially in the MI-MII transition, and protein does not reaccumulate in MII (202). However, if expressed ectopically in MII mouse oocytes, cyclin B3 again has an APC/C activating role, overriding CSF arrest to force MII exit independent of fertilization (198).

Curiously, regarding the late timing of cyclin B3 degradation, the region of cyclin B1 that promotes its earlier prometaphase destruction, is not conserved in B3. This contrasts with cyclin B2 and cyclin A, in which this region is well conserved. It would be interesting to swap this region between cyclin B1 and B3 and observe destruction dynamics.

Coping with two rounds of cell division without an intervening DNA synthesis phase

A further stark difference between meiosis and mitosis is the need to suppress interphase and DNA synthesis (S-phase) between MI and MII. This is essential to produce a haploid MII oocyte ready for fertilization.

To facilitate this, unlike mitotic exit, cyclin B1 is not completely destroyed between MI and MII, and some CDK1 activity remains (203) (152). Alongside this DNA replication licensing factors are suppressed, including the key factor CDC6. CDC6 suppression is perhaps best understood in Xenopus oocytes (204), with strong supporting evidence in mouse. In Xenopus and mouse CDC6 is translated from GVBD but only able to accumulate in MII (previously suppressed by high CDK1 activity in MI (205,206)). Through the MI-MII transition CDC6 is then under antagonistic regulation between B-type cyclins (that stabilize CDC6) and the MOS–MAPK pathway (that negatively controls CDC6 accumulation) (207). Subsequently, evidence in MII Xenopus oocytes demonstrates that CDC6 escapes negative regulation and accumulates, stabilized by newly synthesized cyclin B1. The molecular mechanisms here are not fully understood, but they are crucial on two accounts, to suppress S-phase in the MI-MII transition, and to prepare the oocyte for S-phase post fertilization.

Following MI exit, EMI2 is essential to immediately inhibit APC/C^CDC20 activity and thereby also inhibit S-phase (208). Cyclin B1:CDK1 activity and separase inhibition are quickly restored and oocytes CSF arrest in MII until fertilization. CSF arrest is a unique feature of vertebrate oocyte fertilization, and this presents a further cell cycle challenge; EMI2 levels must be kept low during MI to allow proper progression through the first meiotic division. Several mechanisms are thought to contribute to this, including evidence in Xenopus oocytes of MI specific translation suppression (209,210). In addition, cyclin B1:CDK1 and cyclin B2:CDK1 destabilize EMI2 in MI, but more importantly cyclin B3:CDK1 preferentially phosphorylates EMI2 at a highly conserved site resulting in EMI2 degradation in MI (198). EMI2 is thereby only upregulated after MI, following B-type cyclin destruction (208,211,212).

EMI2 accumulates in MII and is stabilized by c-MOS. Indeed, c-MOS was first thought to constitute CSF since knockout mouse oocytes failed to maintain MII arrest (213). However, it is now understood to function by enhancing APC/C inhibition, promoting not only the stability, but the activity of EMI2 through a MOS-MAPK signaling regulator (p90^Rsk) that directly phosphorylates EMI2. Experiments in Xenopus oocytes demonstrate that p90^Rsk recruits PP2A-B56 to remove inhibitory phosphorylations (deposited by CDK1 and another important cell cycle kinase, PLK1) to prevent EMI2 degradation (214– 217).

Upon fertilization EMI2 is marked for degradation by PLK1 (110,218), introducing yet another challenge to meiotic cell cycle regulation. PLK1 has many crucial regulatory roles throughout prometaphase and is fully active in metaphase (219), yet EMI2 degradation must only take place after fertilization. This ordering is achieved by the need for a ‘priming’ event delivered at fertilization by the signaling molecule CaMKII (220,221). CaMKII directly phosphorylates EMI2, driving PLK1 to further phosphorylation by EMI2 (110,218,222). This generates a degron that is recognized by an SCF family E3 ligase, leading to EMI2 ubiquitination and subsequent degradation (223). Therefore APC/C inhibition is only removed once both CaMKII and PLK1 are active. Notably, at fertilization, the activity of CaMKII also mediates the destruction of MOS.

The need for two rounds of cell division in the absence of DNA synthesis introduces a further unique cell cycle challenge. In MI, separase must enable the segregation of homologous chromosome pairs but leave sister chromatids attached until appropriate segregation in MII. This requires the stepwise removal of cohesin, first from the arms of chromosomes in MI, and then at centromeric regions in MII. Briefly, in MI the meiosis-specific cohesin subunit REC8 is protected at centromeric regions of chromosomes by shugoshin 2 (SGO2) (224). SGO2 localizes to centromeres (mediated by BUB1 and MPS1 kinase activity, proteins more usually associated with SAC activity (225)) and recruits PP2A phosphatase to dephosphorylate REC8. Dephosphorylated REC8 at centromeres is resistant to cleavage by separase, ensuring that only arm cohesin is removed in MI. This protection allows homologous chromosomes to segregate while sister chromatids remain paired. Subsequently, SGO2 is targeted for degradation (226) and largely removed from centromeres on MI exit (227,228). REC8 is then readily phosphorylated and susceptible to separase-mediated cleavage through the second division (for review see (229)).

Notably, this mechanism of centromeric cohesin protection has limits. Cohesin and SGO2 are both depleted in ageing, and weakened centromere cohesion contributes significantly to age-related aneuploidies (230,231). This situation is characterized in MII by increased inter-sister kinetochore distances, chromosome misalignment, and Premature Sister Chromatid Segregation (PSCS). Indeed, cohesin is particularly susceptible to ageing since strong evidence suggests it is only loaded in fetal life, and there is no capacity to reload once lost (232).

The rapid sequential progression from MI to MII further challenges the management of separase inhibition. Securin is degraded by the APC/C on MI exit, but levels do not recover to the same extent in MII-stage oocytes (233,234). In young mouse oocytes this is not problematic. In contrast, in the oocytes of older animals, Nabti et al., report an increase in APC/C-mediated securin destruction, resulting in a further reduction of MII securin (235). This is likely due to a reduction in SAC activity since oocytes from older mice have reduced levels of SAC components (236,237). Nabti and colleagues suggest that reduced securin in aged MII oocytes leads to increased separase activation, which in turn promotes segregation errors. Indeed, increased inter-sister kinetochore distances and PSCS phenotypes, were recovered by either introducing more securin or enhancing SAC activity in aged oocytes (235).

Importantly, securin is not the only inhibitor of separase activity in mouse oocytes as cyclin B1:CDK1 activity also takes on this role. Indeed, securin is dispensable in MI, and in young oocytes both securin levels and cyclin B1:CDK1 activity must be diminished to observe premature release of separase activity (226,238,239).

How the existence of dual mechanisms of separase inhibition impacts aged oocytes is not clear. But presumably, if SAC components are depleted in aged oocytes and APC/C activity is enhanced, this would also negatively impact cyclin B1 regulation in aged oocytes. In this situation the addition of either securin or cyclin B1 could recover depleted separase inhibition (particularly given that at the MI-MII transition, cyclin B1 is suggested to play the major inhibitory role (239)).

The existence of dual mechanisms of separase inhibition speaks to the importance of tightly regulated control. Indeed, further mechanisms of inhibition have been documented in mitotic cells. Most significantly, a mechanism of separase inhibition has been described in the mitotic cell cycle involving SGO2 and MAD2 (240). However, this same inhibition is unlikely to contribute significantly to separase regulation in mammalian oocytes (226,239).

On MI exit it is important that significant levels of cyclin B1 and securin are destroyed before separase is released (226). This could suggest that a late/final function of diminishing cyclin B1:CDK1 activity is separase inhibition. Reaching a low threshold of cyclin B1:CDK1 activity before separase release could ensure that chromosomes are only cleaved at the point of the poleward movement of microtubules. Indeed, it is feasible that changing threshold levels of cyclin B1:CDK1 might help control the ordering of key events more generally through the MI-MII transition; cyclin B1:CDK1 activity thresholds are already known to order events on entry into mitosis, and to gradually stabilize kinetochore-microtubule attachments in prometaphase I mouse oocytes (190,241).

Tight regulation of phosphatase activity is equally important to tight regulation of kinase activity. For example, a sharp switch in the metaphase-anaphase transition is not only reliant on the cessation of phosphorylation, but on the reversal of many CDK1-, PLK1- and aurora-imposed phosphorylations. However, while waves of protein dephosphorylation are known to be critical through each stage of meiosis, the details of phosphatase activity are understudied in mammalian meiosis. Most research in oocytes has focused on the serine/threonine phosphoprotein phosphatase (PPP) family. This is not surprising given their importance in mitosis, where the subfamily of PP2A-like protein phosphatases account for the majority of cellular serine/threonine dephosphorylation events (242,243). In mitotic cells, PP2A acts in coordination with protein phosphatase 1 (PP1), where each phosphatase has unique temporospatial regulation to oppose CDK1 activity at critical stages (244). This is likely to be similar in oocytes. However, a challenge in studying phosphatase activities stems from the lack of specificity of inhibitors targeting both phosphatase activities. For example, when using PP1/PP2A inhibitors such as Okadaic Acid and Calyculin A, the defects observed are so substantial that it is difficult to identify individual mechanisms perturbed by their loss. Some studies have sought to separate PP2A and PP1 function, and evidence supports heavy reliance on each at multiple stages. Briefly, evidence in mouse suggests PPI activity is critical to the acquisition of meiotic competence and GVBD; to proper chromosome condensation and bivalent formation; and during the MI-MII transition to allow MI exit (245–247). Similarly, in an oocyte-specific PP2A deletion model, oocytes escaped prophase arrest, had elongated spindles, and precocious separation of sister chromatids (248). Conversely, a PP2A-like family member PP6 was reported to have a more specific role dispensable for oocyte maturation, but depletion of which perturbed MII spindle formation (and this was accompanied by increased aurora A activity) (249).

Coping with DNA damage over decades of prophase arrest and a prolonged prometaphase

Prophase arrest is one of the most striking features of the mammalian oocyte cell cycle and one of its greatest challenges. Oocytes are particularly vulnerable to oxidative stress and DNA damage over this timeline, and this contributes significantly to age-related fertility decline. But of course, not all oocytes suffer insurmountable damage, and several surveillance and repair mechanisms contribute to their protection when damage is only mild.

First the ATM and ATR kinase pathways that detect DNA DSBs are active and can trigger downstream effectors such as CHK1 and CHK2 (250,251). These effectors can halt cell cycle progression or trigger apoptosis in prophase if damage is irreparable. This ensures that oocytes with overly compromised genomes are eliminated. Second, RAD51 and DMC1, proteins involved in resolving DSBs formed through homologous recombination, can also be employed to repair DSBs that arise from endogenous stress (252). Third, mitochondrial quality control and antioxidant defenses act to mitigate oxidative stress (a major source of DNA damage), though this defense diminishes in ageing (253). In addition, oocytes exhibit low transcriptional activity, which inherently reduces the risk of transcription-associated DNA damage.

Subsequently, canonical responses to DNA damage are compromised in full grown oocytes, and this is thought to be linked to the altered condensed-chromatin state that restricts DNA damage response (DDR) protein access (254). DNA damage is still detected (evidenced by the persistence of γH2AX foci at multiple stages in oocytes (255)), but the outcome changes; only severe DNA damage prevents meiotic resumption in full-grown oocytes (256). By comparison, in somatic cells, even mild DNA damage is sufficient to induce a cell cycle arrest (by activating the G2/M checkpoint), providing sufficient time for damage repair (257). Furthermore, if DNA damage is not repaired in somatic cells, the p53-dependent apoptotic pathway is activated (258). In stark contrast, p53- and p63-mediated apoptoses are not functional in full grown oocytes (69). Moreover, Sun et al., (254) report that while γH2AX foci do persist following damage, oocytes do not initiate autophagy.

Curiously, the reduced state of autophagy might contribute to the weakened DDR mechanism in damaged oocytes, since experimental induction of autophagy can improve developmental competence (259–261) and decreased DNA damage (262). This implies the possible participation of autophagy in mitigating DNA damage. Nevertheless, without chemical stimulation, mammalian oocytes are perhaps the only non-transformed cell to fail to produce a robust G2 phase DNA damage checkpoint (256). Instead, full-grown oocytes leverage SAC activity to inhibit meiotic progression following DNA damage (263). This is not a robust strategy; first the SAC isn’t active until oocytes are in prometaphase of MI (meaning damaged oocytes already reach a stage where they can be ovulated and this is wasteful), and second, the SAC is known to be weak in oocytes compared to somatic cells (176,185,187,236). Indeed, while chemical induction of moderate-to-severe levels of DNA damage were shown to activate the SAC in mouse and human oocytes, many still resumed meiosis and matured to MII (251,263–265). For example, following damage to stimulate a three-fold increase in γH2AX levels in mouse oocytes, 40% of oocytes completed anaphase, of which 80% were aneuploid (compared to 3% aneuploidy in controls) (254). Similarly, human oocytes harboring DNA damage can progress through MI and reach MII (265).

It is also worth noting that γH2AX foci are consistently reported to be higher in aged mouse and human oocytes (252). Yet, while abundant γH2AX foci might imply the presence of a robust DDR, this does not appear to be the case. Following DNA damage induction, aged mouse oocytes took twice as long for γH2AX foci levels to resolve to baseline, and this is likely due to a reduction in efficacy of other DDR response elements. Indeed, oocytes from older mice were observed to have reduced levels of the DDR signaling protein MDC1, and reduced mobility of DNA break sites. Moreover, the DDR has also been shown to be suppressed in aged porcine oocytes (266).

Conclusions/Summary

The mammalian oocyte cell cycle is unique, characterized by multiple prolonged arrests, novel mechanisms, asymmetry, and a different chromosome division goal. Many cell cycle proteins underpinning oocyte meiosis are the same as those in mitosis, but they are often governed by separable regulatory processes, acting over different time frames. Moreover, meiosis-specific proteins have been identified, and many are essential. Together these differences support remarkable adaptations that allow the oocyte to cope with responsibilities that cannot be compromised - for example supporting an extreme cell volume, the readiness for fertilization, and the ability to sustain the earliest stages of life. Understanding these features not only expands our knowledge of reproductive biology, but beyond meiosis, oocyte cell divisions present an opportunity to study proteins out of their most well understood mitotic context.

In this review we aimed to focus on mammalian oocyte meiosis references. However, in many cases, processes are still best understood in non-mammalian species such as Xenopus and yeast. This provides a challenge and an opportunity. On the one hand there are several examples of mechanisms that differ significantly between species (even between mammals), making drawing translational conclusions from model organisms more difficult. But on the other hand, comparison of these differences provides valuable insights into key mechanisms that underpin the production of all female gametes, and indeed more generalizable principles of all cell biology.

One factor that seems specific only to the oocytes of some larger mammals, is their propensity for cell division errors. Many oocyte-specific cell cycle features expose this cell to periods of maturation where it is vulnerable to damage or inappropriate progression before readiness. Why has the oocyte evolved this way? Some frequent events appear to be a risky compromise, for example division in the presence of unaligned chromosomes. But here we can consider the number of resources required to support each ovulatory cycle. A lower quality MII oocyte that might support life is perhaps still worth presenting if the other option is no oocyte.

It also seems possible that in some species, it might be beneficial that not every ovulated oocyte results in a birth. Perhaps human evolution has not selected against a high frequency of errors. For an individual, pregnancy loss is clearly devastating. However, human females have a possible 25-year reproductive window and each carried birth, and infant, is a heavy burden. Reduced birth rates have arguably only become problematic following recent trends towards delayed maternal age. Similarly, intensive livestock breeding has forced a situation where we demand productivity in animals above their natural birth rate, which might otherwise be sufficient to sustain their population.

Studying mammalian oocytes is technically difficult, they are produced in low numbers, deeply embedded in ovarian tissue, and sensitive to manipulation. However, the challenges of this material are diminishing. Knowledge of both generalizable mechanisms of cell cycle regulation, and species-specific differences are expanding. Furthermore, as the capabilities of spatial proteomics, phospho-proteomics and sequencing sensitivities also expand, their application to oocytes further accelerates our understanding. This is exciting discovery science that ultimately informs our ability to treat infertility, and perhaps in the future, to routinely recover some erroneous oocytes for success in ARTs. Importantly, as our environment and lifestyles change, knowledge of oocyte cell cycle regulation will also help us to better understand the mechanistic impacts of influences external to the oocyte, for example heat stress, microplastics, mycotoxin contamination and obesity, many of which are suggested to negatively impact oocyte quality.