| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Center for the Study of Reproduction and Women’s Health and Departments of Molecular Genetics (P.E.C.), Obstetrics, Gynecology and Women’s Health (P.E.C., S.E.P., J.W.P.), and Developmental and Molecular Biology (J.W.P.), Albert Einstein College of Medicine, Bronx, New York 10461
Correspondence: Address all correspondence and requests for reprints to: J. W. Pollard, Developmental and Molecular Biology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, New York 10461. E-mail: pollard{at}aecom.yu.edu
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Formation of the...III. Recombination Events during...IV. Cell Cycle Regulation...V. Studies of Meiotic...VI. PerspectivesReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Formation of the...III. Recombination Events during...IV. Cell Cycle Regulation...V. Studies of Meiotic...VI. PerspectivesReferences |
|---|
The defining event of meiosis is prophase I, during which homologous (maternal and paternal) chromosome pairs locate one another and become physically connected in a process termed "synapsis." Two features of prophase I ensure that homologous chromosomes remain tethered together until the appropriate time during metaphase I. The first is a tripartite protein structure called the "synaptonemal complex" (SC), which consists of two lateral elements (aligning each homolog) and a central element that holds sister chromatids together and maintains the homologs in a tethered state. This structure appears early in prophase I and persists until the maturation of recombination events (crossovers) into chiasmata, the second feature that ensures persistent pairing before metaphase I. The formation of crossovers and the appearance of the SC are very much intertwined events and are largely conserved among meiotic species. Interestingly, however, Drosophila melanogaster males, Schizosaccharomyces pombe, Tetrahymena thermophila, and Aspergillus nidulans, all of which undergo meiosis, do not assemble SCs and do not exhibit reciprocal recombination (reviewed in Refs. 2, 3, 4).
The regulation of SC formation and recombination in mammals has been the focus of much attention recently, and the advent of genetic and biological tools for mammalian systems has resulted in increased interest in these processes. Comparative studies have revealed increasing levels of complexity with regard to the meiotic regulation in mammals, as a result of the larger genome size and/or because of the extrinsic regulation imposed by the physiological and endocrinological environment of the gonads. This is well illustrated by the observation that meiotic error rates rise with increasing genome and organismal complexity. Thus, meiotic errors that give rise to unequal separation of homologous chromosomes at the first meiotic division (nondisjunction) occur at a rate of 1:10,000 in budding yeast, at 1–2% in mice and men, and as great as 20% in women (reviewed in Ref. 5). Understanding how meiosis is regulated in mammals, and extrapolating between mouse and human, will allow for a greater understanding of the mechanisms by which maternal meiotic errors arise in women. In addition, our understanding of these processes is significantly enhanced by examining the meiotic phenotypes of mutant mouse lines harboring targeted gene mutations and by examining the biology of patients with genetic syndromes that result in sterility or reduced fertility. Both of these approaches have yielded a large number of genes that are now known to have roles in meiosis, and the current review is aimed at providing a comprehensive examination of these genes and their function in meiotic recombination.
|
II. Formation of the SC and Its Role in Meiosis |
|---|
|
TopAbstractI. IntroductionII. Formation of the...III. Recombination Events during...IV. Cell Cycle Regulation...V. Studies of Meiotic...VI. PerspectivesReferences |
|---|
). Extension of the central element results in a zippering effect such that, as cells enter pachynema (adjective: pachytene), the longest stage of prophase I, the homologs are synapsed along their entire length. Chromosomes remain in this state for several days in mice, and up to weeks in human oocytes, and they continue to compact throughout this period. Toward the end of the pachytene stage, the central element begins to disassemble. The homologous chromosomes repel each other, progressing into the next stage of prophase I, diplonema. They remain connected at chiasmata, the physical manifestation of reciprocal recombination (crossover) events, through to the final stage of prophase I, diakinesis, and it is these hybrid DNA structures that are responsible for ensuring that homologs remain paired through to metaphase. By diakinesis, much of the SC structure is lost, and only remnants of the lateral elements remain.
|
), the two structural elements that align along each pair of sister chromatids and which begin to assemble in early leptonema of prophase I (6, 7, 8, 9, 10, 11). SYCP3 is present in meiotic nuclei in early leptonema and persists until diplonema when the SC begins to break down. The importance of the axial elements is underscored by the fact that Sycp3–/– male mice are sterile as a result of a mass apoptosis of the spermatocyte population before pachynema (12). Not surprisingly, axial elements fail to form during leptonema in Sycp3–/– spermatocytes, resulting in a failure to assemble meiotic nodule (MN) components and an absence of SYCP1 recruitment at zygonema. However, transverse filaments, reminiscent of those that make up the central element of the SC, accumulate throughout the chromatin of spermatocytes from Sycp3–/– mice, suggesting that SYCP1 accumulation is not wholly dependent on axial element formation (13). Interestingly, Sycp3–/– females are fertile and generate viable offspring, albeit with drastically reduced litter sizes coupled with increased oocyte aneuploidy (14). Thus, although SYCP3 is a central feature of the axial element in both male and female mice, and is critical for appropriate synapsis in both sexes, SYCP3 is not replaced by another protein in Sycp3 nullizygous females; therefore, female meiosis appears not to be entirely dependent on proper SC formation.
During zygonema, the SYCP3 filament becomes continuous along each chromosome, and a third structure, the central element, begins to accumulate. The mammalian central element consists of SYCP1 (Fig. 1C) and is responsible for synapsis and bivalent formation through zygonema and into pachynema. Other components of the central element have recently been identified in mice, including FK506 binding protein-6 (FKBP6), which localizes to synapsed regions of chromosomes and physically associates with SYCP1 (15). Targeted inactivation of Fkbp6 in mice results in aspermic males and the absence of normal pachytene spermatocytes. Instead, loss of FKBP6 results in abnormal pairing and misalignments between homologous chromosomes, nonhomologous partner switches, and autosynapsis of X-chromosome cores in meiotic spermatocytes. Interestingly, however, fertility and meiosis are normal in Fkbp6 mutant females, indicating that FKBP6 is not essential for meiotic progression in oocytes (15). Two other proteins, SYCE1 and CESC1, were identified by microarray profiling and appear to be novel components of the mammalian central element (16).
Cohesins are also intertwined with the SC structure, and meiosis-specific cohesin complexes are present in mammalian germ cells. Mitotic cohesin contains four subunits; two SMC (structural maintenance of chromosome) proteins, SMC1
and SMC3, and two non-SMC subunits, RAD21 and SCC3. Collectively, this complex is essential for ensuing sister chromatid cohesion through mitosis until metaphase when cohesion is lost between the chromosome arms, followed by loss of centromere cohesion (reviewed in Ref. 17). In meiotic cells, SMC1 is replaced by SMC1ß, whereas RAD21 and SCC3 are replaced by REC8 and STAG3, respectively. Recent evidence has indicated that an intact cohesin complex is essential for progression through prophase I of meiosis. Smc1ß–/– male and female mice are sterile as a result of meiotic failure during pachynema and early metaphase, respectively. Spermatocytes from these mice exhibit impaired axial element formation, shortened axial element lengths, chromosome synapsis errors, and a failure to process recombination events, whereas oocytes progress through prophase I and accumulate a few crossovers that enable them to progress through metaphase I and beyond, but with increased rates of nondisjunction and premature sister chromatid separation (18). Similarly, two recently isolated mutations in Rec8 result in sterility in mice as a result of similar synapsis defects and recombination failure (19, 20). Interestingly, the absence of REC8 in one of these mouse lines results in SC formation between sister chromatids (20), indicating a role for the cohesin complex in directing appropriate SC assembly. Thus the SC, although essential for meiotic progression, is not sufficient to ensure sister chromatid cohesion without specific participation of the meiosis-specific cohesin complex.
An important feature of the developing SC is the meiotic (or recombination) nodule (MN), an electron-dense structure first described by Adelaide Carpenter in 1975 (21). The original term for the MN was the recombination nodule, but this is somewhat misleading because not all MN/recombination nodule structures will become true recombination events. Molecular and cytogenetic analysis has revealed that the components of the SC vary through prophase I, as does the frequency and placement of MNs (reviewed in Ref. 2). In mice, the number of MN structures is highest in leptonema and early zygonema and declines steadily thereafter to reach a stable plateau at midpachynema (reviewed in Ref. 23). The number of MNs at this time correlates well with the number of eventual crossovers observed at metaphase. Indeed, studies with the phosphatase inhibitor, okadaic acid, which accelerates prophase I progression through pachynema and induces precocious chiasmata formation, have demonstrated that two key MN components, the MutL homologs MLH1 and MLH3, are also found at nascent chiasmata (24). That the stabilization of MNs and ultimate crossover formation are intricately linked, and that both are dependent on a functional SC, is demonstrated by mouse gene knockout studies showing MN loss in mice lacking components of the SC. For example, the frequency of MNs is severely reduced in spermatocytes from Smc1ß–/– mice, suggesting a role for the cohesin complex in mediating events at these sites (18). An important question arises, however, as to the function of the excess MNs observed in early prophase I. Because these sites presumably represent recombination initiation sites (see below), their function in promoting meiotic progression is unclear because only a fixed number of MNs persist through to late pachynema. The events that give rise to this fixed number of late MNs remains poorly understood, particularly in mammals, but will be discussed in the context of recombination events in the following sections.
The timing and length of prophase I vary widely between species and even between sexes. In male mice, for example, prophase I lasts over 10 d, with pachynema constituting 5–7 of those days (Fig. 2). The first cohort of spermatogonia enters prophase I at around d 13 postpartum and progresses through leptonema and zygonema, reaching pachynema by around d 17 (25). They remain in pachynema until d 22–23 postpartum, and the first cohort of metaphase I cells is observed at this time. Mature spermatozoa are then observed from d 27 postpartum onward, with cohorts of new cells progressing through prophase I in waves from that time onward.
|
). By e17, cells enter pachynema and progress through pachynema, reaching early diplonema by birth (26). Instead of progressing through to metaphase I, however, the cells enter a stage of meiotic arrest known as dictyate arrest in which all meiotic activity ceases and the cells undergo a protracted stage of diplonema. The nascent chiasmata are established and maintained through early postnatal development, and the meiotic events do not reinitiate until after puberty, when cohorts of cells are stimulated to resume meiosis by the action of the gonadotropins, principally FSH. These oocytes progress through metaphase I and into meiosis II, arresting once again soon after ovulation and resuming the second meiotic division upon fertilization. Thus, some oocytes will reinitiate meiosis soon after sexual maturity is achieved, whereas others may remain in dictyate arrest for much longer periods of time. A similar temporal framework exists for human female meiosis. Oogonia begin to populate the primitive ovary as early as 7–9 wk gestation, with mitotic activity being evident by 10–12 wk (27). Meiosis is initiated soon thereafter, with prophase I cells being observed between 12 and 20 wk gestation (27), and maximal pachytene cell numbers being observed by 26 wk gestation (27, 28, 29, 30). By 20 wk gestation, two thirds of the total germ cells are intrameiotic, and one third are still undergoing mitosis. By 7 months gestation, germ cell mitosis ceases, and the last of the germ cells enter meiosis (29). Others are either still in the process of meiosis, undergoing atresia, or are arrested at the dictyate stage (31). By term, many of the oocytes are at dictyate arrest, although a high proportion are still progressing through meiosis, highlighting the less synchronous manner of human oocyte meiosis (31, 32, 33, 34, 35).
|
III. Recombination Events during Mammalian Meiosis |
|---|
|
TopAbstractI. IntroductionII. Formation of the...III. Recombination Events during...IV. Cell Cycle Regulation...V. Studies of Meiotic...VI. PerspectivesReferences |
|---|
Homologous recombination is initiated in almost all meiotic species by the formation of a double-strand break (DSB) in one sister chromatid of one homolog. The SPO11 topoisomerase is responsible for the formation of DSBs in a conserved manner throughout all meiotic species, including Saccharomyces cerevisiae, S. pombe, Coprinus cinereus, Caenorhabditis elegans, and mammals (reviewed in Ref. 36). Not surprisingly, therefore, Spo11–/– mutant mice are infertile as a result of a failure to induce recombination during leptonema of prophase I (37, 38). In male mice, the failure to induce DSBs is paired with a failure of chromosome condensation and synapsis in leptonema and zygonema, coupled with an absence of recombination markers (37, 38). Cisplatin treatment, which induces DSB formation, restores recombination events in spermatocytes from Spo11–/– males, resulting in resumption of prophase I and at least partial spermatogenesis in these animals (38). Interestingly, oocytes from Spo11–/– females reach the diplotene/dictyate stage of prophase in nearly normal numbers, but most are destroyed soon after birth as the meiotic defects become apparent to the checkpoint machinery (37, 39).
The events that proceed from SPO11-induced DSB induction are not yet fully elucidated in mammals but have been extensively characterized in other species such as yeast, flies, and worms. Our current understanding of these events is that DSB resolution can occur through several pathways (summarized in Fig. 3), but that recombination events are directed via DSB repair (DSBR) that involves double Holliday junction formation or via synthesis-dependent strand annealing (SDSA) (40, 41). In both pathways, DSB ends are degraded, providing 3'-overhangs that are then capable of invading an opposing homolog. This event results in the displacement of one homolog, producing stretches of heteroduplex DNA, and the formation of a D-loop structure that then becomes stabilized by subsequent branch migration via DNA synthesis. In the DSBR pathway, the second end of the DSB becomes captured, resulting in a mature double Holliday junction structure that may be resolved as a gene conversion event with or without crossing over (to produce a recombinant molecule), whereas in SDSA the invading strand gets displaced and reanneals with the other end of the DSB (after DNA synthesis). Repair of mismatches then results in a gene conversion tract without crossover formation (42). Other mechanisms for crossover formation have been identified in yeast but have yet to be verified in mammals (43, 44).
|
The recombination process in mammals is less clearly defined than that outlined above for yeast. However, as with yeast meiosis, the ultimate crossover number is extremely conserved in mammals, arising from a starting number of DSB events that outnumber the final chiasma count by more than 10-fold (46). Analysis of orthologs of yeast genes involved in the recombination process has revealed that many conserved gene products are involved in these selection events in mammals. These gene products include the RecA homologs, RAD51 and DMC1, and members of a number of DNA repair families including the breast cancer susceptibility genes, Brca1 and Brca2, the RecQ helicases, Blm (BS mutated) and Wrn (Werner syndrome-associated), the MMR family, and the Atm/Atr [ataxia telangiectasia (A-T)-mutated and A-T-related] genes (see below). Given the importance of these genes in DNA repair and modification in general, it is not surprising that targeted mutations within this group result in embryonic lethality, cancer predisposition, and/or premature aging of mice. Of those mutations that do allow for viable pups, it is interesting to note that the homozygous mutant animals also exhibit severely reduced fertility (or complete sterility) along with their cancer predisposition/genomic instability phenotypes, highlighting the similarities in function of these genes in DNA repair and recombination events. As discussed below, these mutant animals thus allow us to explore the function of their respective genes in the meiotic process and to expand our knowledge of recombination events in mammalian systems. The ultimate goal of such studies is to understand how recombination is regulated and what role the DNA repair mechanisms play in the decision between crossover and noncrossover (NCO) fates.
|
IV. Cell Cycle Regulation and Checkpoints in Prophase I |
|---|
|
TopAbstractI. IntroductionII. Formation of the...III. Recombination Events during...IV. Cell Cycle Regulation...V. Studies of Meiotic...VI. PerspectivesReferences |
|---|
and Ref. 35).
|
The actions of cyclins and their associated kinases are inhibited by CDK inhibitors that fall into two families, the INK4 and Cip/Kip families (57). Cyclin D/CDK4 and CDK6 complexes are subjected to negative regulation by both the INK4 (a–d) and the Cip/Kip (p21Cip/Waf1, p27Kip1, and p57Kip2), families of inhibitors (Cckn1 and 2), whereas cyclin E/CDK2 complexes are negatively controlled only by the Cip/Kip family (58). Although this sequence of events has been largely validated in cell culture experiments, recent genetic analyses of cell cycle events in mice lacking these proteins have not been straightforward to interpret, and the results suggest considerable compensation within and between gene families (59). Indeed ablation of the supposedly essential cyclin E1 (Ccne1), or its partner CDK2 (Cdk2), did not result in embryonic lethality, and mice were fully viable (60, 61). Even ablation of both Ccne1 and e2 did not block embryogenesis until late in development (60). These data indicate that cell cycles can progress normally even in the absence of any CDK2 activity. However, deletion of cyclin A2 (Ccna2), a widely expressed cyclin, results in early embryonic development immediately after implantation, probably as a result of loss of cyclinA2/CDK2 whose activity is required for S phase transverse (62).
Disruption of the genes encoding the cell cycle-regulatory proteins has revealed important roles in meiosis. Mice with null mutations in Cdk2 show severe abnormalities in both male and female meiosis with complete loss of germ cells in both sexes (61). However, the timing of loss is a little different between sexes. CDK2 is required for full chromosome synapsis in males. Thus, Cdk2–/– males trigger a pachytene checkpoint and exhibit massive apoptosis of spermatocytes before diplonema even though mitotic divisions of spermatogonia seem unaffected. In contrast, Cdk2–/– females do not display defects at pachynema, but significant defects were found in oocytes as they reached dictyate with the aberrant distribution of SYC3 along chromosomes and a random dispersal of centromeres. Immediately after dictyate arrest oocyte loss occurs, and all oocytes are lost by postnatal d 14 (61).
Cyclin E2 (Ccne2)-deficient mice also display meiotic abnormalities in males but, in this case, females are fully viable and fertile. The block in spermatogenesis in cyclin E null males appears incomplete with the report of aberrant meiotic figures and low sperm count but is made more severe in mice that are also heterozygous for a mutation in the gene encoding cyclin E1 (60). Mice completely lacking all E-type cyclins die during embryogenesis, and in these mice meiotic analysis in females was not performed (60). Male meiotic cells uniquely express cyclin A1 (Ccna1), and targeted ablation of this gene blocks spermatogenesis and, as expected from the expression data, not oogenesis (63, 64, 65). This block is not in premeiotic S phase, but instead manifests itself at the end of prophase before MI and is associated with desynapsis abnormalities followed by the propulsion of cells into apoptosis (65).
Mice lacking all D-type cyclins (66), as well as those mice lacking their partners CDK4 (Cdk4) and CDK6 (Cdk6), also die during embryogenesis (67), whereas those lacking a single cyclin D are viable (68). Cyclin D2 (Ccnd2) null male mice are fertile but have reduced testicular size and lower sperm number, whereas mice harboring mutations in the other cyclin D subtypes (D1 and D3) do not have any apparent abnormalities in spermatogenesis or oogenesis (66). Cyclin D2, however, is required for FSH-regulated granulosa cell proliferation, and the females are infertile due to a failure in follicle maturation (69). Null mutant mice for Cdk4, the cyclin D partner, have hypoplastic testis with abnormal spermatogenesis consisting of reduced numbers of spermatogonia and spermatocytes but exhibiting normal oogeneis until CDK4 is required together with cyclin D2 for FSH-induced follicular development (69, 70). These data suggest that at least one cyclin D is required in partnership with CDK4 for normal spermatogenesis but that this is not involved in the generation of female germ cells. It is unclear exactly where the defect in spermatogenesis lies, although it does seem to involve meiotic phases.
It is unfortunate that mutations in Rb1 result in embryonic lethality (71) because the absence of Rb during meiosis cannot be analyzed without the creation of a conditional Rb1 allele. However, the recent rescue of this early embryonic lethality by tetraploid aggregation techniques (72) paves the way for analysis of female meiosis during later embryonic development. In the case of the E2F transcription factors that are downstream of Rb, there is remarkable redundancy, and all three E2F family members need to be removed before an impairment in cell cycle progression is observed (73). This has not yet been performed in mice. Thus, it is still unclear what the role for these nuclear transcription factors is in the regulation of meiosis.
The activity of the CDKs is regulated by the CDK inhibitors (74). P27kip1 is a ubiquitous regulator of the cyclins, and mice homozygous for a null mutation in the Cdkn1b gene show testicular hypertrophy but without apparent effect upon ovarian maturation; however, there are endocrine disruptions that make female mice infertile (75, 76). Homozygous null mutations for both Cdkn2c and Cdkn2d in mice result in male infertility due to a failure of spermatogenesis but without any effect on female fertility. However, the disruptions in spermatogenesis appear additive, with defects in the timing of mitotic exit and in progression through meiosis. This apparently triggers an apoptotic pathway(s) that results in a gradual depletion of germ cells with few spermatids and spermatozoa being detected in the seminiferous tubules of these mice (77). These data suggest that these inhibitors act to extinguish the activity of a D-type cyclin that, together with the activity of p27kip1 during normal physiology, allows appropriate exit from mitosis without reactivation of DNA replication and normal progression through meiosis. However, the data also suggest that some of these cell cycle regulators may continue to have functions through the meiotic cell cycles maintaining proper timing in association with correct chromosome segregation (Table 1).
These studies of mice carrying mutations in cell cycle-regulatory genes are only preliminary and as yet do not explain the control over timing of the transitions from mitosis to meiosis in mammals, or their mode of action in regulating the chromosome dynamics after chromosome replication or the progression in M1. The dramatic loss of germ cells in Cdk2 null mutant mice evident in prophase I without apparent effects during mitosis strongly suggests that, in addition to known targets of proteins involved in DNA replication, elements of the SC are either direct or indirect targets for this kinase. Furthermore, these studies indicate that without CDK2 action, SC proteins fail to assemble on the homologous chromosome pair, resulting in desynapsis and nonhomologous pairing. Consistent with this is the unusual association of CDK2 with chromosomes during anaphase (78). Furthermore, lacking in the analysis of these mouse mutants is an understanding of why female meiosis is unaffected by mutations in some of these cell cycle-regulatory molecules despite apparently similar progression through mitotic and meiotic cycles in both sexes. Only the absence of CDK2 appears to profoundly affect both male and female meiosis, whereas the other mutants have a meiotic phenotype that is restricted to effects on spermatogenesis (Table 1).
In addition to the regulation of DNA synthesis by cyclin-dependent mechanisms, DNA replication also requires the organization of the prereplicative complex (pre-RC) at the origins of replication. This "licenses" the DNA to be copied, and the assembly of this complex is tightly controlled to ensure one, and only one, round of DNA replication per cell cycle (79, 80). Origin licensing begins with the recognition of the origin by the origin replication complex, a heterohexameric complex (81). The association of CDT1 and CDC6 with this complex facilitates the loading of a hexameric complex consisting of the minichromosome maintenance (MCM) proteins 2–7 in stoichiometric amounts (82). This, in turn, triggers a conformational change that allows firing of the pre-RC by the action of CDK and CDC7, resulting in recruitment of DNA polymerase, proliferating cell nuclear antigen, and CDC4 (82). The MCM complex melts the chromatin, allowing replication to be initiated by the primase activity of DNAse polymerase a (82). This licensing complex is then removed to block reinitiation of DNA synthesis. This removal includes down-regulation of origin replication complex 6, CDT1, and MCM proteins as well as geminin expression that binds to CDT1, thereby blocking the loading of MCM proteins onto the chromatin (81, 82).
Proteins of the pre-RC complex are required for premeiotic DNA replication in yeast (83, 84). Given the importance of these proteins for DNA synthesis, it has not been possible to analyze the meiotic phenotypes of mice carrying inactivating mutations in the pre-RC complex genes. Nevertheless, it would be expected that these genes would be down-regulated during early prophase I to prevent unscheduled DNA synthesis. Interestingly, this is not the case, and instead the down-regulation of CDC6 and CDT1 does not occur until the early spermatid stage, whereas MCM2 expression is not lost until late spermatid stage (85). Similarly in females, MCM2 and CDT1 are found in oocytes beyond dictyate arrest. Moreover, regulation may be different between males and females because geminin is expressed in primary spermatocytes but not in primary oocytes, whereas CDC6 expression is down-regulated in oocytes but is maintained in early spermatocytes (85). This suggests two different mechanisms for the suppression of replication firing in males and females. The continued expression of many components of the pre-RC complex in prophase I of both sexes suggests additional roles in meiosis such as in recombination or DNA repair. However, identification of such roles will necessitate the existence of conditional or hypomorphic alleles.
The final phase of prophase I is the entry into the first meiotic division (MI). Like the mitotic cycle, meiosis is regulated by maturation-promoting factor (MPF), the complex that includes cyclin B and CDK1 (49, 86). Although the regulation of MI is beyond the scope of this review, it is clear that MPF is essential during mammalian meiosis because, despite the lack of conditional alleles to ablate these proteins specifically in germ cells and the very early embryonic lethality of CDK1 null mutants (53), genetic ablation of the CDK1-interacting protein, CKS2, results in normal progression through prophase I including normal chromosome synapsis, but then followed by a failure to undergo the first meiotic division in both male and female mice (87). Similarly, the inhibitor of CDK1, -2, and -5, roscovitine, inhibited MI in cocultures of rat spermatocytes and Sertoli cells in a germ cell-specific manner (88). However, there are no mutations that directly allow the analysis of MPF in mammalian systems. Nevertheless, the extensive biochemistry that has been performed in Xenopus and the genetic analysis in yeast (49, 86) should pave the way for a similar analysis in mammals using conditional alleles of the relevant genes.
B. Checkpoints in prophase I
The absence of cell cycle-regulatory proteins, particularly CDK2 and cyclin A1, triggers apoptotic pathways in developing germ cells. In somatic cells a major checkpoint is triggered through detection of DSBs in a process that involves ATM and AT/Rad3-related (ATR) gene products (89). These kinases propagate their signal through p53–p21 and CHK1 and 2 (CDC25A and CDC25C-mediated pathways), resulting in inhibitory phosphorylation of CDK2 and its subsequent degradation and consequent cell cycle arrest (89). This is coupled to DSBR that requires another substrate, the Nijmengen breakage syndrome (NBS1) protein (see Section V) that may also be involved in the checkpoint itself. However, if repair is not mediated, the cells are triggered to die via apoptosis. The proteins p53 and p21 are required for this response because their targeted ablation partially rescues ATM-deficient somatic and meiotic cells (90, 91, 92). During meiosis, as will be discussed in the next section, the absence of ATM or NBS1 triggers apoptosis in both male and female germ cells with few cells surviving beyond pachytene. These data indicate the importance of the ATM DNA damage-sensing pathway during meiosis, perhaps resulting in the blockage of passage toward MI, while the DSBs introduced as a necessary prelude to homologous recombination are repaired. Thus in the absence of ATM, cells are sent into apoptosis at what has been determined as the epithelial stage IV in the testis, approximately equivalent to midpachynema (92), to avoid the inevitable aneuploidy that would follow improper chromosome synapsis. Whether the mechanism is the same as found in somatic cells remains undetermined, although absence of NBS1 and CDK2 (downstream of CHK2-CDC25), both ATM direct or indirect targets, result in meiotic abnormalities during pachytene and apoptosis as discussed in Section IV.A. However, it has not been established, for example, whether CDK2 accumulates in the absence of ATM, resulting in the failure of cell cycle arrest, as would be predicted from experiments in somatic cells. Furthermore, this checkpoint, unlike those in somatic cells, is p53 independent (92). This suggests that the pathway through CHK1 and -2 proteins is emphasized in meiosis, consistent with the association of these proteins with meiotic chromosomes (93).
Analysis of other meiotic mutants in mammals suggests that there are at least two checkpoints during prophase I (94, 95). The first senses unpaired chromosome and/or unprotected double-strand breaks. Its presence is graphically demonstrated by the complete loss of germ cells through apoptosis in males and females in mutant mice lacking Msh4 and Msh5 (96, 97) as discussed in detail below. This checkpoint is p53 independent (Ref. 94 and our unpublished results). A similar checkpoint appears to be initiated after loss of DMC1, a RecA homolog (98) and may also be congruent with that used by ATM. However, the depletion of meiocytes in Atm–/– mice appears more rapid and synchronized, whereas in females lacking MutS homologs (MSH)4 or -5, checkpoint activation is associated with disruption of ovarian structures that is not observed in ATM-deficient mice (96, 97). This may indicate the presence of more than one checkpoint or of other signaling pathways that modify a single process. Genetic epistasis experiments may help to sort out these alternatives.
A second meiotic checkpoint is revealed by the loss of MutL homolog 1 (Mlh1), and this appears to sense univalent chromosomes without appropriate chiasmata as the meiocytes exit from pachytene (99, 100). In this case, apoptosis is executed either at or before MI (101). This checkpoint is also p53 independent (our unpublished data). Mlh1 null females show no oocyte loss, and ovulation proceeds in an apparently normal manner. However, these oocytes fail to fertilize properly and rapidly lose viability, with few progressing past MI. Thus, although the chiasmata loss is the cause of the meiocyte loss, the timing of cell death is quite different. This may reflect on the different processes with the female oocytes entering dictyate arrest before the checkpoint is activated. The oocytes instead appear to be eradicated as a result of improper chromosome segregation and subsequent developmental failure, rather than through apoptosis that is initiated by the actions of a stage-specific checkpoint per se. A similar sex difference is observed in mice nullizygous for the MLH1 partner, Pms2, where males are infertile due to a prophase I defect, whereas females are fertile (102). The loss of such a stringent checkpoint post-pachytene in females may be a partial explanation for the increased numbers of meiotic abnormalities in this sex.
|
V. Studies of Meiotic Mutants Involved in DNA Recombination in Mice and Man |
|---|
|
TopAbstractI. IntroductionII. Formation of the...III. Recombination Events during...IV. Cell Cycle Regulation...V. Studies of Meiotic...VI. PerspectivesReferences |
|---|
A. The RecA homologs, RAD51 and DMC1
Bacterial RecA protein is involved in homologous recombination and repair of DNA damage caused by ionizing radiation and/or chemical mutagens (103, 104). The eukaryotic homolog, RAD51, was identified by Shinohara and colleagues (105, 106, 107, 108) and was shown to catalyze strand transfer between homologous DNA strands in yeast and humans. Human RAD51 maps to chromosome 15q15.1 and promotes homologous pairing and strand exchange (109). A second RecA homolog, DMC1, appears to be restricted to meiosis (110) and colocalizes with RAD51 at early MNs (111) in a variety of organisms, including Escherichia coli (112), mice (109, 111, 113, 114), and humans (109, 115). Interestingly, a number of RAD51-like paralogs have been identified, of which there are five in mammalian cells (RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3; reviewed in Ref. 116).
RAD51 attaches to the 3'-end of a DNA DSB site and mediates the transfer of the resulting nucleoprotein filament to a complementary homologous strand on the intact chromatid. This invasion displaces the noncomplementary intact strand, an essential step in D-loop formation (reviewed in Refs. 116 and 117). Yeast rad51 mutants are defective in meiotic DSBR, displaying an accumulation of recombination intermediates coupled with increased chromosomal nondisjunction and decreased spore viability (108, 118). The requirement for other factors in the Rad51-driven meiotic events is demonstrated by the observation that (replication protein A) RPA, Brca2 Rad52, and Rad55 all augment the recombinational activity of Rad51 in yeast (119, 120, 121, 122, 123) and in human cells (124).
In human spermatocytes and oocytes, RAD51 forms nuclear foci from early zygonema to late pachynema (109, 115), colocalizing with the lateral element SC protein, SYCP3, most strongly on unsynapsed segments. In contrast to the situation in mice, however, where RAD51 is rapidly lost from the chromosome core after synapsis (109, 111, 113), RAD51 immunoreactivity persists on synapsed human chromosomes until late pachynema, although the number of RAD51 foci declines rapidly through pachynema to reach a level that is comparable to the frequency of mature MNs that are thought to give rise to the final population of chiasmata (115). Thus, in both mouse and human meiocytes, RAD51 focus numbers start out at a frequency of approximately 10-fold higher than the final chiasma tally (>250 in the mouse and >350 in humans) and decline thereafter to undetectable in the mouse and less than 100 in humans.
The meiosis-specific RecA homolog, DMC1, is highly conserved among meiotic species, ranging from yeast to lilies to mice (98, 110, 125, 127). Yeast dmc1 mutants are meiotically inactive, yet have no demonstrable mitotic defects (110). Yeast Dmc1 and Rad51 proteins colocalize during meiotic prophase I and are localized to the SC before synapsis. Interestingly, although DMC1 fails to localize to these sites in rad51 mutants, Rad 51 is found in dmc1 mutants, indicating a temporal loading order for the RecA homologs (110, 112). Similar to Rad51, Dmc1 protein forms a nucleoprotein filament on single-stranded DNA in the presence of ATP (128, 129, 130, 131, 132).
Mice that are homozygous for a null mutation in Dmc1 are sterile, exhibiting meiotic disruption early in prophase I before pachynema. Spermatocytes from Dmc1–/– males show apparently normal progression through leptonema and early zygonema, with no complete pachytene-stage cells being observed (98, 127). No attempts at synapsis are apparent, and chromosome cores remain in their uncompacted state, indicative of early prophase I. Interestingly, RAD51 continues to be present on the chromosome cores in numbers indicative of leptotene to zygotene stages, but no diminution of RAD51 focus frequency (that would be indicative of a progression through to pachynema) is observed (98).
In Dmc1–/– females, the loss of DMC1 causes a similar meiotic disruption, but the consequences to the overall biology of the gonad are more severe. By e17, the number of oocytes within the fetal ovary is significantly diminished in Dmc1–/– females compared with wild type (98) and, by d 4 of life, fewer than 10% of oocytes remain (39). The consequence of this global loss of oocytes from the Dmc1–/– ovary is the subsequent degeneration of ovarian structures through life, with adult females displaying a "streak-like" ovarian morphology reminiscent of ovaries from Turner syndrome patients.
The loss of ovarian structures after mutation of a key early recombinogenic gene is now a common theme in murine meiosis. Several of the genes that will be discussed in this review, when mutated, give rise to a similar female phenotype. Interestingly, the earliest of the recombinogenic genes, Spo11, is quite different. As described earlier, SPO11 is thought to be the enzyme responsible for generating DSBs in mammalian meiocytes and, indeed, its activity is essential for DSB formation in mice (37, 38, 133). However, unlike the genes, the products of which process these DSB events, mutation of Spo11 in females results in at least some progression through prophase I, with oocytes progressing through to the diplotene/dictyate stages before succumbing, after birth, to apopototic elimination (37). Interestingly, however, some oocytes from Spo11–/– neonates progress further than diplotene, with 15–20% of oocytes being capable of initiating primordial follicle formation indicative of entry into dictyate arrest (39). By contrast, few or no oocytes from Dmc1–/– females remain to this point; instead they die by diplotene. These observations suggest that once DSBs are formed, they must be dealt with to ensure the health and survival of the oocyte. In line with this theory, oocytes from Spo11–/–Dmc1–/– double-mutant females are a phenocopy of oocytes from Spo11–/– single-mutant animals (39). Taken together, these results demonstrate that the failure to induce DSBs in Spo11 mutant animals results in meiotic disruption that is independent of a DNA damage checkpoint and that is therefore distinguishable from the DNA damage checkpoint to which oocytes from Dmc1 null females succumb.
By contrast with the situation in females, male meiocytes are wholly dependent on DSB formation and recombination to progress through prophase I. Spermatocytes from Spo11–/– and from Dmc1–/– males show similar meiotic phenotypes, arresting their meiotic program at or before pachytene (37, 38, 98, 127); in contrast, meiotic progression in Spo11 nullizygous males is partially restored by introducing cisplatin-induced DSBs in spermatocytes, whereas Dmc1–/– males are unaffected by such treatment (37). Taken together, these results indicate that RAD51 and DMC1 cooperate in the downstream processing of SPO11-induced DSB events, and that the checkpoint machinery monitoring their actions may differ in stringency between male and female mammals.
B. The MMR machinery in mammalian meiosis
The MMR pathway, as its name implies, was first identified in its capacity as the primary mechanism through which DNA mismatches are detected and corrected. The system acts primarily to correct mismatches arising as a consequence of errors during replication in S phase, but also participates in other DNA stability functions associated with stabilization of repeat structures, such as trinucleotide repeats (134, 135, 136), cisplatin-induced cross-links, and damage from oxidation/alkylation agents (reviewed in Ref. 137). This pathway also prevents recombination between divergent sequences, known as homeologous recombination, which could result in potentially lethal chromosomal imbalance. Components of the MMR pathway are highly conserved throughout evolution and are closely linked with components of the cell cycle and checkpoint machinery in higher eukaryotes (reviewed in Refs. 138 and 139). That these events are essential for cell viability and integrity is demonstrated by the observation that defects in the mammalian MMR system result in cancer predisposition (reviewed in Ref. 140). In humans, germline mutations in certain MMR genes are responsible for hereditary polyposis colorectal cancer, characterized by early-onset colorectal cancer, along with cancers of the endometrium, stomach, urinary tract, and ovary (reviewed in Ref. 141). Mutations in the MMR gene family are also responsible for a high number of spontaneous colorectal cancer cases.
In bacteria, where the MMR system was first defined, mismatch recognition is effected by the MutS protein, which binds as a homodimer to the site of the mismatched bases. An ATPase-containing protein, MutL, which binds to the MutS homodimer is then recruited, and this MutS-MutL complex then activates the MutH endonuclease (reviewed in Ref. 138). MutH is a methylation-sensitive endonuclease that binds to a methyl group on either side of the mismatch. Taking advantage of the fact that, in prokaryotes, newly synthesized DNA is transiently unmethylated, MutH preferentially nicks the daughter strand, thereby directing the repair machinery away from the template strand and instead marking the erroneous strand for exonucleolytic resection and resynthesis (142). Depending on the side of the MutH activity, one of four exonucleases is then recruited, possessing either 5'-3' exonuclease (exonuclease VII or RecJ) or a 3'-5' exonuclease (exonucleases I or X) activity, to degrade the mismatch-containing nicked strand (reviewed in Ref. 138). These events are assisted by the MutU (UvrD) helicase that unwinds the DNA from the nick through to the mismatch.
In eukaryotes, the MMR system is complicated by the presence of several homologs of MutS and several of MutL. No MutH homolog exists in eukaryotes, possibly because of the absence of methylation status-directed strand discrimination. Instead, strand discrimination is thought to come about as a result of interactions of the MMR family with the replication fork machinery, such as proliferating cell nuclear antigen (143, 144). Seven MSHs have been identified, five of which exist in mammals, MSH2–MSH6. These act as heterodimers, with MSH2 being the principal MutS homolog involved in repair events, in conjunction with either MSH3 or MSH6. Together, the MSH2–MSH6 (known as MutS) and the MSH2–MSH3 (known as MutSß) bind to and repair all common mismatches, but each complex has its own substrate specificity. Mammalian MutS can bind to, and repair, DNA-containing base-base mismatches and insertion-deletion loops, whereas MutSß can only bind to insertion-deletion loops. In addition, evidence from yeast suggests that MutSß is capable of binding to unpaired loops as large as 5.6 kb that could arise as a result of heteroduplex formation during meiotic recombination (145).
Like the MutS homologs, MutL also has several homologs in eukaryotes, and in mammals these are named MLH1, MLH2, MLH3, and PMS2, the latter being named for its postmeiotic segregation (PMS) phenotype in yeast (where it is confusingly called PMS1). MLH1 is the principal MutL homolog acting in MMR events, and it primarily functions in this regard with PMS2 (MutL), although interactions with MLH2 (to form MutLß) and MLH3 (to form MutL
) may provide repair mechanisms for specialized classes of mutational intermediates, at least in yeast (146, 147). In mammals, the most likely function for MutL heterodimers other than MutL is in meiosis, where MutL is essential for recombination and crossing over, as it is in yeast (see below). MLH3 appears to be the primary binding partner for MLH1 in these meiotic processes, both in yeast and in mammals (42, 148, 149, 150, 151), whereas the role for MLH3 in MMR and colorectal cancer is contentious at the present time (149, 152, 153, 154, 155, 156, 157, 158).
In addition to its role in DNA repair and cell cycle control, the MMR pathway is also essential for meiosis in a variety of species including S. cerevisiae, Arabidopsis thaliana, C. elegans, and Mus musculus (23, 42, 159, 160, 161, 162). For these purposes, however, other MutS homologs are recruited, and the MSH4–MSH5 heterodimer (MutS) appears to function solely in meiotic recombination; these proteins lack the N-terminal MutS/MSH domain that is necessary for mismatch binding (163). In mice, mutation of Msh4 or Msh5 results in meiotic disruption early in prophase I and sterility in both males and females (96, 97, 164). At the cytogenetic level, chromosomes replicate in premeiotic S phase, apparently normally, and then enter meiosis, accumulating SYCP3 protein and forming complete axial elements. The axial elements appear somewhat elongated, however, and stain less intensely for SYCP3 (P. E. Cohen, unpublished observations). Most significantly, chromosome synapsis is severely reduced in both Msh4–/– and Msh5–/– animals, with SYCP1 rarely appearing between the axial elements (P. E. Cohen, unpublished observations), and with a high degree of mispairing, self-synapsis, and nonhomologous pairing being observed (Fig. 4). As a result, spermatocytes succumb to apoptosis early in prophase I, at or before zygonema, such that the testis of adult Msh4–/– and Msh5–/– males is largely devoid of spermatogenic cells except for one or two layers of spermatogonia and early spermatocytes.
|
Given the stoichiometry of the MMR complexes in repair functions, it was not surprising to find that MSH4-MSH5 interacts with both MLH1 and MLH3 in yeast and in mice (165, 166). Mutations of either of these MutL homologs in mice also result in meiotic failure but at a distinctly later time point to that seen for Msh4 and Msh5 mutations (99, 100). Spermatocytes from Mlh1 and Mlh3 null animals progress through to pachynema apparently normally, with complete synapsis being observed (99, 148). Then at diplonema, when the central element of the SC disassembles and the chromosomes begin to move apart, instead of remaining together at chiasmata sites, the chromosomes in Mlh1–/– and Mlh3–/– males fall apart prematurely. This premature desynapsis is thought to occur as a result of a failure to stabilize these crossovers, and, as a consequence of this destabilization, fewer than 10% of chiasmata are observed at metaphase in Mlh1–/– males, whereas none are seen in Mlh3–/– males (99, 100, 148, 170).
More recently, studies of MLH3 and MLH1 function in meiotic cells revealed some interesting dynamics with respect to the accumulation of these proteins in SCs during prophase I. MLH3 is recruited first to the chromosome core in late zygonema or early pachynema, with MLH1 being recruited in midpachynema in mice (170). This observation is further supported by the persistence of a few MLH3-positive foci on chromosome cores from Mlh1–/– males but the absence of MLH1 on chromosomes from Mlh3–/– males (148, 170). Furthermore, this persistence of MLH3 in Mlh1–/– is coupled with the retention of 10% chiasma numbers at metaphase (171). Thus MLH3 may function independently of MLH1 to select MN sites for further processing into crossover events. The phenotype of Mlh1 and Mlh3 null females reflects the later function of MLH1 and MLH3 in mammalian meiosis as compared with MSH4 and MSH5. Thus, chromosomes progress through to pachynema and into early diplonema (99, 149, 172). However, at birth, when the oocytes enter dictyate arrest, meiotic failure has not yet become apparent, and so the ovary assumes its normal dictyate developmental pattern, even though the oocytes are destined to fail in the resumption of the meiotic program. Thus, adult Mlh1–/– or Mlh3–/– females have normal ovaries possessing oocytes at all stages of follicular development. These oocytes are capable of resuming meiosis upon ovulation, but at metaphase, it becomes apparent that few, if any, intact chiasmata remain. As a consequence, the chromosomes fail to establish stable bipolar spindle orientation, the spindle poles are destabilized and continue to elongate, and most oocytes fail to achieve an anaphase configuration (172). In vitro fertilization studies demonstrate that oocytes from Mlh1–/– and Mlh3–/– females progress poorly through early fertilization events (99, 148). Although first polar bodies are, in fact, extruded, it appears unlikely that these polar bodies contain the appropriate chromosome count because chromosomes appear to be arrayed randomly over the first meiotic spindle (Ref. 172 , and N. Kolas and P. E. Cohen, unpublished observations).
In line with the phenotypes of the meiotic MMR genes, their gene products are intrinsic to the chromosome cores of prophase I meiocytes. MSH4 (and presumably MSH5) is localized to meiotic chromosome cores of mouse spermatocytes and oocytes as early as leptonema (97), at a frequency that is thought to reflect the status of DSB-processing events (Fig. 5). That MSH4 is localized to MNs at this time is shown by colocalization with RAD51 and DMC1 (173) and, in human oocytes at least, MSH5 (115). However, MSH4 localization on chromosome cores persists beyond these other early MN markers, through until at least late pachynema and, from early to midpachynema onward, interacts with components of the late MNs, including MLH1 and MLH3 (165, 166). Thus, MSH4 and MSH5 constitute core components of the MNs that are resident through most of the DSB-processing events, and the frequency of which diminishes gradually through prophase I, reflecting the sequential selection of recombination intermediates along a reciprocal recombination pathway (Figs. 5 and 6). Biochemical data indicate that the MSH4/5 heterodimer associates specifically with Holliday junction substrates and that this binding stimulates ATP hydrolysis (174). This, in turn, instigates a sliding clamp configuration for MSH4/5 that moves away from the Holliday junction and encircles the heteroduplex arms (174). It is postulated that MLH1–MLH3 stabilizes these clamp structures, allowing more to accumulate along the heteroduplex stretch, and thus ensuring the progression of this site through to later crossover stages (23, 174, 175). This suggestion is confounded, however, by studies in yeast, which demonstrate the existence of three distinct crossover pathways, only one of which involves MSH4–MSH5 (161, 176).
|
|
C. The RecQ helicases, Bloom syndrome (BS) mutated (BLM) and Werner syndrome-associated (WRN)
BS is an autosomal-recessive disorder characterized by growth deficiency, dermatological sun-sensitive pigmentary lesions, immunodeficiency, predisposition to malignancies, and chromosomal instability, including increased frequency of sister chromatid exchanges (178). The BS gene product, BLM (also known as RecQ-like 3, RECQL3), is a member of the RecQ family of 3'-5' single-stranded DNA helicases, molecules that recognize and bind to and expand single-stranded gaps in duplex DNA (179), and include the human Werner’s syndrome-associated (WRN) and Rothmund-Thomson syndrome (RECQL4) gene products, as well as S. cerevisiae SGS1. Men with BS are sterile due to azospermia; women, although sometimes fertile, have a shortened reproductive life span, displaying normal puberty, but a high incidence of premature ovarian failure (178, 180). Human male heterozygote carriers of BLM mutations display sperm with higher frequencies of structural chromosome abnormalities (181). Interestingly, the rare occurrences of pregnancy among women with BS are associated with higher rates of preterm labor (182, 183).
The human BLM gene spans 250 kb on chromosome 15q, constitutes a 4437-bp cDNA, and encodes a 1417-amino acid protein (179). Transfection of BLM coding sequence into cells from BS patients restores BLM activity and partially complements the phenotype of S. cerevisiae sgs1 top3 mutant strains (184). Similarly, murine Blm spans 4780 bp of chromosome, contains an open reading frame of approximately 1420 bp, and shares 81% identity with human BLM cDNA, whereas the proteins are 76% similar (185). Blm is expressed at high levels in the testis where the protein localizes to germ cells in prophase I.
BLM protein contains a C-terminal nuclear localization sequence, whereas the RecQ helicase domain is located in the central portion of the protein (179). BLM unwinds DNA in 3'- to 5'-direction, but can also recognize and disrupt other DNA structures such as Holliday junctions, triple helices, and complex duplex structures (186, 187, 188, 189). Other RecQ-specific domains exist in the C-terminal portion of the protein, including a zinc-binding RecQ-Ct domain that appears to direct interactions with the DNA and to modulate the ATPase and helicase functions of BLM (190). The C terminus also regulates nucleolar localization and interactions with other proteins that function in DNA metabolism, including topoisomerase III (TOPOIII), RAD51, and MLH1, whereas other regions direct binding of BRCA1, ATM, p53, FEN1, NBS1/MRE11, MSH2, and MSH6 (191, 192, 193, 194, 195, 196, 197, 198, 199, 200), to name a few. The interaction with ATM is functionally important because ATM may at least partially phosphorylate the helicase (201).
All of these interactions point to a role for BLM in homologous recombination and the maintenance of DNA integrity. Accordingly, BS cells exhibit profound genomic instability, characterized by excessive chromosomal breakage, hyperrecombination, and increased mutation frequency, and are hypersensitive to genotoxic agents (202, 203, 204). A defining feature of BS cells is an elevated frequency of sister chromatid exchanges that arise from crossing over of chromatid arms during homologous recombination resulting from damaged replication forks. Recombination in mitot
