A leading hypothesis to explain maternal age-related aneuploidy is a deterioration of sister chromatid cohesion. Experiments in yeast show that sister chromatids can only be effectively held together when the mitotic cohesin protein SCC1 is expressed before S phase, indicating that cohesion is established in S phase [56
]. Chromosome cohesion is lost at anaphase onset when the protease separase cleaves SCC1. A mutant form of SCC1, which cannot be cleaved by separase, blocked chromosome segregation only when expressed before S phase [57
]. In mammalian cells expressing GFP-tagged cohesin proteins, results of fluorescence recovery after photobleaching experiments showed that chromatin-bound cohesins are stable from S phase until anaphase [58
]. Together these results show that mitotic cohesion is established in S phase, and once loaded, cohesin proteins remain stably bound to chromosomes until cleavage by separase at anaphase. Meiotic cohesion is similarly established in S phase in yeast [59
]. Because cohesion must remain functional for up to 50 yr until meiosis resumes in human oocytes, cohesion is a good candidate for a process that might fail with maternal age and lead to increased aneuploidy.
Cohesion along chromosome arms keeps bivalents intact in MI (A), and centromere cohesion holds sister chromatids together in MII (D). A defect in cohesion distal to crossover sites may result in a shift of chiasmata placement (chiasmata slippage; B) or even premature bivalent separation in MI (C), whereas reduced centromere cohesion may result in premature separation of sister chromatids in MII. The relationship between premature chromosome separation, different positions of chiasmata, and maternal age was first documented in mice in 1968 [61
]. The distal movement of chiasmata is now recognized as chiasmata slippage, suggesting that loss of cohesion occurs with age. In mice and Drosophila
deficient in the meiotic cohesin protein SMC1B, chiasmata slippage and premature chromosome separation in oocytes were also observed. In both cases, the loss of cohesion phenotype worsened with maternal age [62
], consistent with the idea that cohesion defects may contribute to age-related aneuploidy.
Transgenic mice have been engineered to test two critical parts of the cohesion hypothesis: 1) whether new cohesion can be established after S phase, and 2) the stability of cohesins with age. To address the first question, cleavage sites specific for TEV protease were inserted into the endogenous locus of REC8, the meiotic counterpart of SCC1 [64
]. The engineered sites allow TEV protease to mimic separase by cleaving REC8 and releasing cohesion. Microinjection of TEV protease into these oocytes thus triggers chromosome separation in the absence of endogenous wild-type REC8. To determine when cohesion can be established, an additional transgene encoding wild-type REC8 was conditionally activated. Wild-type REC8 expression in the fetus, before the onset of meiosis, prevented TEV-mediated chromosome separation (). In contrast, when wild-type REC8 expression was driven after
birth, exposure to TEV protease led to prematurely separated chromosomes. These results show that cohesion can only be established in S phase before birth, and once loaded, cohesin proteins do not exchange after birth. The second part of the hypothesis—stability of cohesin proteins—was tested in mice by conditionally deleting the Smc1b
]. When Smc1b
inactivation occurred shortly after birth, the initial amount of SMC1B was sufficient to hold bivalents together and maintain chiasmata placement for at least 8 mo in aging mice [65
]. Together, these results suggest that cohesins load onto chromosomes and establish cohesion only during fetal development, and the cohesin complexes, once loaded, remain functional until meiosis resumes.
FIG. 3. Schematic of experiments to test when functional chromosome cohesion is established. Transgenic mice with cleavage sites specific for TEV protease on REC8 were engineered . A) Microinjecting TEV protease into oocytes where bivalents were held together (more ...)
Results from genetically perturbed models are consistent with the cohesion hypothesis, but do not address what occurs during the natural aging process. Staining for either SMC1B or REC8 in naturally aged oocytes showed an age-dependent decrease of chromosome-associated cohesins [55
]. For example, chromosome-associated REC8 levels gradually decreased with age, by at least 90% by 12 mo of age [67
]. Interestingly, aneuploidy incidence remained low at 12 mo old, but increased dramatically by 15 mo. These results suggest that cohesins are originally loaded in excess, and aneuploidy occurs only when REC8 levels fall below a threshold. In addition, total REC8 protein was similar between young and old oocytes, further supporting the idea that once REC8 is lost from chromosomes, it cannot be effectively reloaded. Although a shift of chiasmata toward the distal end was also observed in old oocytes when compared to their young counterparts, complete separation of bivalents at MI was rare, suggesting that the observed aneuploidies are not due to complete loss of cohesins from chromosome arms [55
Centromere cohesion, on the other hand, has an important function at MI: to physically link sister kinetochores to promote attachment to one pole, or mono-orientation, so reductional segregation of homologous chromosomes can take place [59
]. Loss of cohesion at the core centromere in fission yeast led to increased separation of sister kinetochores and thus promoted erroneous biorientation in MI (B) [68
]. Similarly, in plants centromere cohesion is required for mono-orientation, and increased separation of sister kinetochores resulted in lagging chromosomes and chromosome segregation errors at anaphase [69
]. In mouse, an increased separation of sister kinetochores at MI in old oocytes suggests that centromere cohesion is indeed weakened with maternal age. In addition, live imaging of individual young and old oocytes through anaphase I showed a higher incidence of lagging chromosomes in old oocytes, consistent with erroneous MI attachments. Chromosome counts in the resulting eggs at metaphase II often contained aneuploidies of single chromatids, indicating a premature loss of centromere cohesion [55
Although the results in natural aging models are consistent with the cohesion hypothesis, whether chromosome cohesion is actually reduced in naturally aged oocytes was not directly tested by any experimental perturbation. To directly target cohesion, separase was prematurely activated by depleting securin and preventing separase phosphorylation by CDK1 at S1121 and T1342 [71
]. Chromosome cohesion in old oocytes was more susceptible to premature separase activation compared to chromosome cohesion in young oocytes, as shown by premature chromosome separation, demonstrating that cohesion is indeed reduced with natural aging [74
]. Together, all of these results in naturally aged mice indicate that a deterioration of cohesion, especially at the centromere, occurs with increasing age and leads to increased aneuploidy.
In humans, there are also clues that cohesion becomes defective with age. When aneuploidy types were characterized in oocytes from women of different ages by cytogenetic approaches, aneuploidies due to single chromatids generally exceeded whole chromosome aneuploidies [75
]. This result argues that defective cohesion must make a major contribution to age-related aneuploidy. A recent study of human oocytes found Smc1b
mRNA levels to be similar between young and old oocytes, but it is unclear whether SMC1B protein is made in the mature oocyte [80
]. Studies in mouse oocytes suggest that cohesin proteins can load onto chromosomes only during S phase, but what occurs in human oocytes remains unknown.