Two closely connected mechanisms safeguard the fidelity of chromosome segregation in eukaryotic cells. The mitotic checkpoint monitors the attachment of kinetochores to microtubules and delays anaphase onset until all sister kinetochores have become attached to opposite poles. In addition, an error correction mechanism destabilizes erroneous attachments that do not lead to tension at sister kinetochores. Aurora B kinase, the catalytic subunit of the CPC (chromosomal passenger complex), acts as a sensor and effector in both pathways. In this review we focus on a poorly understood but important aspect of mitotic control: what prevents the mitotic checkpoint from springing into action when sister centromeres are split and tension is suddenly lost at anaphase onset? Recent work has shown that disjunction of sister chromatids, in principle, engages the mitotic checkpoint, and probably also the error correction mechanism, with potentially catastrophic consequences for cell division. Eukaryotic cells have solved this ‘anaphase problem’ by disabling the mitotic checkpoint at the metaphase-to-anaphase transition. Checkpoint inactivation is in part due to the reversal of Cdk1 (cyclin-dependent kinase 1) phosphorylation of the CPC component INCENP (inner centromere protein; Sli15 in budding yeast), which causes the relocation of the CPC from centromeres to the spindle midzone. These findings highlight principles of mitotic checkpoint control: when bipolar chromosome attachment is reached in mitosis, the checkpoint is satisfied, but still active and responsive to loss of tension. Mitotic checkpoint inactivation at anaphase onset is required to prevent checkpoint re-engagement when sister chromatids split.
- Aurora B kinase
- chromosomal passenger complex (CPC)
- mitotic checkpoint
- spindle assembly checkpoint
The accuracy of chromosome segregation in eukaryotes underpins cellular fitness and survival, and prevents genomic imbalances that could give rise to pathological conditions. From their synthesis during DNA replication until their segregation during anaphase, a protein complex called cohesin holds sister chromatids together . After mitotic entry the MTs (microtubules) of the mitotic spindle capture sister chromatids by binding to KTs (kinetochores) , proteinaceous structures that assemble on centromeric DNA. Once all chromosomes have been aligned at the metaphase plate, the APC/C (anaphase-promoting complex/cyclosome), an E3 ubiquitin ligase that is bound to its co-factor Cdc20 (cell division cycle 20), marks securin and cyclin B for proteolytic degradation [3,4]. Destruction of securin liberates a protease called separase that cleaves the cohesin complex and triggers chromatid segregation to opposite poles . Proteolysis of cyclin B inactivates Cdk1 (cyclin-dependent kinase 1), the enzyme that drives cells into mitosis, to bring about cell separation (cytokinesis) and mitotic exit. Mitotic-exit phosphatases in turn then become active to remove modifications from Cdk1 substrates [6,7]. In conjunction, these processes give rise to two physically distinct daughter cells that have inherited a complete copy of the genome.
Wait and fix: error correction and the mitotic checkpoint
Accurate chromosome segregation can only be accomplished if KTs on sister chromatids are attached to MTs emanating from opposite spindle poles. In this configuration, which is referred to as bipolar attachment or biorientation, sister centromeres are under tension and cohesin resists the splitting forces of the mitotic spindle [2,8].
Although KT capture by the mitotic spindle is stochastic in nature, cohesin cleavage by separase occurs globally on all chromosomes and is irreversible. As a consequence, eukaryotic cells have evolved two conserved and closely linked mechanisms that ensure that all chromosomes have reached biorientation before splitting sister chromatids.
The first mechanism is known as the mitotic checkpoint (or ‘wait-anaphase signal’). This pathway acts as a timing and synchronization device that delays anaphase onset until all chromosomes have become bi-oriented . Mitotic checkpoint signalling is usually triggered by the recruitment of a number of mitotic checkpoint proteins including Bub1, BubR1, Mps1, Mad1 and Mad2 to unattached KTs and, in some cases, to KTs that are attached to MTs, but have failed to reach bipolar tension. Subsequently, a complex known as the MCC (mitotic checkpoint complex), containing Mad2, Cdc20, BubR1 and Bub3, inhibits the activity of the APC/CCdc20 . Thereby, the engaged checkpoint blocks the degradation of securin and cyclin B and prevents anaphase onset. A single unattached KT is known to be sufficient for mounting this checkpoint response . Once all chromosomes have undergone biorientation and aligned at the metaphase plate, the resulting tension stabilizes KT–MT attachments and thereby silences the mitotic checkpoint signalling to permit anaphase onset. Checkpoint satisfaction is achieved by dynein-mediated removal of mitotic checkpoint proteins , and possibly involves KT deformation [11,12].
The second mechanism is refered to as error correction. Erroneous KT–MT attachments that do not generate tension are eliminated during the search-and-capture process that leads to biorientation, while bipolar attachments are stabilized [13–16]. Erroneous attachments include configurations in which, e.g., both sister KTs are attached to the same spindle pole. Sister chromatid cohesion at centromeres allows cells to differentiate between erroneous and correct attachments, as only the latter result in tension between sister KTs [8,17]. Detachment of erroneous KT–MT connections generates unattached KTs that can subsequently undergo another round of capture by the mitotic spindle and also trigger the ‘anaphase-wait’ checkpoint mechanism [18,19].
The best candidate for the tension sensor that controls the error correction mechanism and possibly the mitotic checkpoint is a conserved multiprotein complex, known as the CPC (chromosomal passenger complex) [20,21]. The CPC localizes to the inner centromere until anaphase onset. The catalytic core of the CPC is Aurora B kinase (called Ipl1 in budding yeast), which interacts with a scaffold protein called INCENP (inner centromere protein; Sli15 in budding yeast). Aurora B promotes error correction and the mitotic checkpoint response by destabilizing KT–MT attachments through phosphorylating outer KT proteins and MT motors [18,19,22–28]. Recent experiments have raised the possibility that biorientation might remove Aurora B substrates from the kinase's reach by inducing structural changes in KT geometry [2,11,12,29,30]. Modified Aurora B targets can then be dephosphorylated by protein phosphatase 1 to stabilize attachments [31–33]. This could provide a molecular mechanism for how tension is detected and for how the mitotic checkpoint is satisfied and silenced.
Although error correction and the mitotic checkpoint co-operate to ensure correct chromosome segregation, the relationship between the two remains to be fully understood. Aurora B kinase might contribute to the generation of the mitotic checkpoint signal both by the creation of unattached KTs, as well as by directly promoting the recruitment of checkpoint components to tensionless or unattached KTs (discussed in ).
The ‘anaphase problem’
The above model does not address an important question of mitotic checkpoint control. What happens at anaphase onset when cohesin is cleaved? Why do the error correction pathway and the mitotic checkpoint not spring into action when sister centromeres are split and tension is suddenly lost? In this review, we will focus on recent advances in different model organisms that shine light on this poorly understood, but important, aspect of mitotic checkpoint control [35–38].
In principle, the tension status at KTs of segregating sister chromatids at anaphase resembles that of split sister chromatids caused by defects in sister chromatid cohesion. This condition would normally lead to the destabilization of attachments and engage the checkpoint . However, resurgence of the mitotic checkpoint and error correction pathway at anaphase onset would have unwanted consequences for chromosome segregation and cell division: (i) inhibition of the APC/CCdc20 would block or delay mitotic exit [35,37]; and (ii) KT–MT connections could be broken while chromatids are being pulled to opposite poles .
The significance of the ‘anaphase problem’ of the mitotic checkpoint is highlighted by several observations. Two recent studies, using syncytial Drosophila embryos  and budding yeast  as model systems, have tested the consequences of artificially cleaving an engineered version of cohesin using TEV (tobacco etch virus) protease in metaphase-arrested cells (Figure 1). Employing this system, chromatid disjunction could be separated from other processes that are normally associated with anaphase, such as activation of the APC/C, activation of separase and reversal of Cdk1 phosphorylation. In both flies and yeast, TEV-protease-induced splitting of sister chromatids in metaphase-arrested cells quickly led to the engagement of the mitotic checkpoint and the recruitment of checkpoint proteins to KTs [35,36] (Figure 1A). The erratic chromosome movements in Drosophila embryos that followed TEV-protease-mediated chromatid disjunction also suggest that KT–MT attachments became destabilized possibly reflecting the activation of the error correction pathway  (Figure 1A).
These experiments demonstrate that the very event that triggers anaphase in eukaryotic organisms, the cleavage of cohesin, engages the mitotic checkpoint and may activate the error correction pathway. This response is presumably initiated by the loss of tension and a change in KT geometry when sister chromatids are split. In vivo analysis of budding yeast KT architecture revealed that anaphase KTs indeed adopt a relaxed and possibly tension-less configuration . However, the error correction and mitotic checkpoint pathways are not engaged during anaphase in wild-type cells. Even disruption of MT attachment by nocodazole treatment in budding yeast cells arrested in anaphase failed to trigger mitotic checkpoint signalling . Thus mechanisms must exist in presumably all eukaryotic cells that disable both pathways at the metaphase-to-anaphase transition to render cells refractory to the loss of tension.
Reversal of Cdk1 phosphorylation disables the mitotic checkpoint at anaphase onset
What inactivates the mitotic checkpoint and error correction pathway? In budding yeast the phosphatase Cdc14 seems to lie at the heart of this process. Cdc14 is known to reverse Cdk1-dependent phosphorylation during mitotic exit . At the metaphase-to-anaphase transition, in addition to catalysing cleavage of cohesin, separase also induces the activation of Cdc14 [41,42] (Figure 2). Ectopic Cdc14 expression suppressed the mitotic checkpoint response after TEV-protease-induced sister chromatid disjunction  (Figures 1B and 2). Conversely, loss of Cdc14 function resulted in engagement of the mitotic checkpoint and APC/C inhibition after anaphase onset . These results suggest that in budding yeast, Cdc14 activation by separase at anaphase onset inactivates the mitotic checkpoint when sister centromeres are split and tension is lost (Figure 2). In Drosophila embryos Cdk1 inhibition by injection of p27 was similarly able to suppress mitotic checkpoint engagement, as well as the erratic chromosome movements upon artificial disjunction of sister chromatids  (Figure 1B).
Taken together, these experiments suggest that removing Cdk1 target phosphorylation disables the mitotic checkpoint and probably the error correction pathway when sister chromatids are split in anaphase. This provides an elegant solution for the ‘anaphase problem’: at anaphase onset, Cdk1 inactivation, and, at least in budding yeast, Cdc14 activation are coupled through APC/C-induced destruction of cyclin B and securin. The corollary of this solution is that one or more processes that are elicited by Cdk1 substrate dephosphorylation are responsible for inactivating the checkpoint response when cells exit mitosis. This interpretation is consistent with earlier studies in Xenopus, Drosophila and yeast that implicate Cdk1 activity in the regulation of the mitotic checkpoint and KT–MT interactions [43–46].
Moving the tension sensor out of the way: relocation of Aurora B and the CPC
At anaphase onset the CPC and its catalytic subunit Aurora B leave the inner centromere and translocate to the spindle midzone where they regulate the process of cytokinesis [20,21]. As first suggested by experiments in flies , this striking change in location is an excellent candidate for being responsible for disabling the mitotic checkpoint: (i) Aurora B's departure from centromeres is a conserved event in eukaryotic cells that is triggered by the loss of Cdk1 phosphorylation on INCENP at the metaphase-to-anaphase transition [44,47–49] (Figures 2 and 3A); (ii) since Aurora B acts as a tension sensor, if not removed from centromeres, Aurora B could respond to the loss of tension when sister chromatids are split at anaphase onset; (iii) fluorescence resonance energy transfer sensor experiments in mammalian cells have shown that the translocation of the CPC to the spindle midzone is associated with a loss of Aurora B phosphorylation along chromosomes ; and (iv) consistent with the notion that the relocation of the CPC inactivates the mitotic checkpoint and error correction pathway, CPC components are retained at centromeres in Drosophila when checkpoint proteins are recruited and chromosomes erratically move during anaphase triggered by TEV protease cleavage of cohesin or in the presence of non-degradable cyclin B [36,45].
Experiments in budding yeast have now shown that the engagement of the mitotic checkpoint upon artificial disjunction of sister chromatids at metaphase indeed depends on persistence of the CPC at centromeres  (Figure 2). A non-phosphorylateable mutant of Sli15INCENP, which mimics the dephosphorylated state independently of Cdc14 action , abrogated the engagement of the mitotic checkpoint after untimely separation of sister chromatids . This suggests that the dephosphorylation of Sli15INCENP and relocation of the CPC are important mechanisms that disable the mitotic checkpoint and error correction pathway at anaphase (Figure 2).
Is the relocation of the CPC the sole mechanism responsible for the inactivation of the mitotic checkpoint and error correction pathway at anaphase onset? This has now been tested in human cells . Analogous to the situation in budding yeast, the relocation of Aurora B to the spindle midzone in mammalian cells is regulated by the dephosphorylation of INCENP on the Cdk1 target site Thr59 . In addition, CPC relocation in human cells requires the kinesin Mklp2 (mitotic-kinesin-like protein 2)  and a cullin-like ubiquitin ligase . Depletion of Mklp2 or expression of INCENPT59E, which mimics constitutive phosphorylation by Cdk1, caused the retention of the CPC at centromeres throughout anaphase [38,47,48] (Figure 3B). These experimental conditions permitted the analysis of the consequences of preventing CPC relocation while maintaining other processes normally associated with anaphase, such as Cdk1 inactivation. Retention of the CPC at centromeres led to the recruitment of the mitotic checkpoint proteins, Bub1, BubR1 and Mps1, to KTs in response to sister chromatid disjunction at anaphase  (Figure 3). Inhibitor treatment experiments demonstrated that this recruitment required the kinase activity of Aurora B. However, CPC retention at centromeres during anaphase was insufficient to generate other hallmarks of error correction and mitotic checkpoint engagement, such as destabilization of KT–MT attachments, KT recruitment of Mad1 and Mad2, or inhibition of the APC/C  (Figure 3A). Mad2 did not even accumulate at entirely unattached KTs after nocodazole treatment in anaphase. Thus although CPC relocation appears to be an efficient mechanism to inactivate error correction and the mitotic checkpoint, additional pathways clearly exist, at least in human cells, that disable parts of the response when sister chromatids separate at anaphase.
Beyond the CPC
What could be the mechanisms that co-operate with the removal of the CPC from centromeres to inactivate the mitotic checkpoint? Several observations suggest that many processes that disable the checkpoint at anaphase onset are linked to the inactivation of Cdk1 [35,36,43], analogous to the relocation of the CPC. The additional mechanisms appear to safeguard the stability of KT–MT attachments and to prevent Mad2 recruitment to unattached KTs once cells have passed the metaphase-to-anaphase transition . The inactivation of proteins involved in the mitotic checkpoint and error correction pathway by Cdk1 site dephosphorylation could be responsible for this. Excellent candidates targets for this regulation are the activity of the checkpoint kinase Mps1, which is required for both Mad2 recruitment to KTs and error correction [37,52,53], the RZZ (Rod–Zw10–Zwilch) complex  and Mad1, both of which act as KT-targeting proteins for Mad2 . Several other proteins involved in checkpoint signalling, such as Bub1 and Cdc20, are also regulated by phosphorylation [46,55], as are KT proteins and kinesins that regulate the dynamic behaviour of MTs .
The activity and regulation of the APC/C could also contribute to preventing mitotic checkpoint engagement. In budding yeast, Mps1 is targeted for proteolysis by the APC/CCdc20 at anaphase . Furthermore, during mitotic exit the APC/C exchanges its co-factor Cdc20 for Cdh1. APC/CCdh1, in contrast with APC/CCdc20, is not sensitive to mitotic checkpoint inhibition . This switch, which is also controlled by reversal of Cdk1 phosphorylation [40,56], could therefore prevent mitotic checkpoint signalling from inhibiting APC/C activity later during anaphase .
The three states of the mitotic checkpoint
The mitotic checkpoint and error correction pathway can be described by three different states (Figures 2 and 3A): (i) during prometaphase, unattached or erroneously attached KTs trigger error correction and engage the mitotic checkpoint, which inhibits the APC/C; (ii) once stable bipolar attachment of all chromosomes is achieved at metaphase, the mitotic checkpoint is satisfied and error correction ceases, but both pathways are still active and continue to monitor the status of attachment; and (iii) after the metaphase-to-anaphase transition, the mitotic checkpoint and error correction pathway are disabled and no longer monitor the status of attachments. Thus loss of tension, which would have led to checkpoint engagement moments earlier, no longer elicits a response when sister centromeres are split in anaphase.
Eukaryotic cells have solved the ‘anaphase problem’ of the mitotic checkpoint by tightly coupling the inactivation of this pathway to the disjunction of sister chromatids [35,36,45]. This is achieved through the action of APC/CCdc20, which induces the simultaneous degradation of cyclin B and securin to inactivate Cdk1 and liberate separase respectively. In budding yeast, separase co-ordinates cohesin cleavage with activation of the Cdc14 phosphatase, which leads to CPC relocation. Less is know in higher eukaryotes about the nature and regulation of the INCENP phosphatase although human CDC14A has been suggested as a candidate . The departure of Aurora B from centromeres, caused by INCENP dephosphorylation, is a fundamental and possibly universally conserved mechanism that inactivates the mitotic checkpoint in anaphase [35,38]. However, it is not the only one. Elucidating the nature of the additional mechanisms will be an important task for future research. Conjointly, these redundant mechanisms might be essential for successful chromosome segregation. Furthermore, investigating them will undoubtedly provide insights into how the mitotic checkpoint and error correction pathway operate prior to anaphase when chromosomes have to be aligned and bi-oriented. Finally, mechanisms that incapacitate the mitotic checkpoint at anaphase may also play a role during mitotic slippage, a phenomenon that describes the escape of cells from mitosis despite the continued presence of conditions that normally engage the mitotic checkpoint. Mitotic slippage is of significant clinical interest because it dictates the outcome of the exposure of cancer cells to antimitotic drugs .
Work in the laboratory of F.U. and M.P. is supported by Cancer Research UK. M.D.V.-N. is supported by a Ramón Areces Foundation fellowship.
Chromosome Segregation and Aneuploidy: An Independent Meeting held at the Royal College of Surgeons, Edinburgh, U.K., 19–23 June 2010. Organized and Edited by Bill Earnshaw (Wellcome Trust Centre for Cell Biology, Edinburgh, U.K.), Kevin Hardwick (Wellcome Trust Centre for Cell Biology, Edinburgh, U.K.) and Margarete Heck (Queen's Medical Research Institute, Edinburgh, U.K.).
Abbreviations: APC/C, anaphase-promoting complex/cyclosome; Cdc, cell division cycle; Cdk1, cyclin-dependent kinase 1; CPC, chromosomal passenger complex; INCENP, inner centromere protein; KT, kinetochore; Mklp2, mitotic-kinesin-like protein 2; MT, microtubule; TEV, tobacco etch virus
- © The Authors Journal compilation © 2010 Biochemical Society