Report

Anaphase Inactivation of the Spindle Checkpoint

See allHide authors and affiliations

Science  04 Aug 2006:
Vol. 313, Issue 5787, pp. 680-684
DOI: 10.1126/science.1127205

Abstract

The spindle checkpoint delays cell cycle progression until microtubules attach each pair of sister chromosomes to opposite poles of the mitotic spindle. Following sister chromatid separation, however, the checkpoint ignores chromosomes whose kinetochores are attached to only one spindle pole, a state that activates the checkpoint prior to metaphase. We demonstrate that, in budding yeast, mutual inhibition between the anaphase-promoting complex (APC) and Mps1, an essential component of the checkpoint, leads to sustained inactivation of the spindle checkpoint. Mps1 protein abundance decreases in anaphase, and Mps1 is a target of the APC. Furthermore, expression of Mps1 in anaphase, or repression of the APC in anaphase, reactivates the spindle checkpoint. This APC-Mps1 feedback circuit allows cells to irreversibly inactivate the checkpoint during anaphase.

Chromosome segregation and cell cycle progression are coordinated. Before mitotic chromosome segregation, sister chromatids are held together by cohesin (1). To segregate properly, a pair of sister chromatids must be bioriented, with the two sister kinetochores attached to and pulled toward opposite poles of the spindle. Cohesin resists these pulling forces, which produces tension between the sister kinetochores (2). The spindle checkpoint detects chromosomes that are not bioriented, by monitoring both tension at kinetochores and the presence of kinetochores that are unattached to spindle microtubules (35). As mitosis proceeds and sister chromatid pairs biorient, the number of kinetochores lacking tension falls. The spindle checkpoint is sensitive enough to detect a single unattached kinetochore (6) or a single chromosome whose sisters are attached to the same pole (mono-orientation) (7, 8) and delays cell cycle progression until biorientation is complete.

Activation of the spindle checkpoint inhibits the onset of anaphase, arresting cells in mitosis until all chromosomes are bioriented (9). The target of the spindle checkpoint is Cdc20, an accessory subunit of the anaphase-promoting complex (APC) (10). Cdc20 induces the mitosis-specific form of the APC (APCCdc20). The ubiquitin ligase activity of APCCdc20 targets securin (Pds1) for destruction, relieving the inhibition of separase (Esp1), a protease that cleaves the cohesin subunit Scc1 (1). Thus, when all chromosomes are bioriented, the spindle checkpoint is satisfied and the cohesin linkage is eliminated, which allows sister chromatids to separate.

Anaphase kinetochores are no longer under tension; they lack a sister chromatid to pull against, and once they reach the spindle pole, any drag forces created by movement through a viscous medium disappear. However, the cell cycle does not arrest as a result of chromosome segregation, which suggests that the spindle checkpoint is inactivated as cells enter anaphase. Cdc15 is a part of the mitotic exit network in budding yeast (Saccharomyces cerevisiae); cdc15-2 strains proliferate at 23°C and arrest in late anaphase at 37°C (11). To confirm that the checkpoint cannot be activated during anaphase, we arrested cdc15-2 cells at 37°C and transferred them to medium containing the microtubule poison nocodazole, a known activator of the spindle checkpoint (4). The stability of epitope-tagged Pds1, an APCCdc20 target, was measured to monitor the checkpoint's activity (12). Anaphase-arrested cells treated with nocodazole did not stabilize Pds1, which showed a lack of spindle checkpoint activity (Fig. 1A). In contrast, Pds1 was stable in cells that had been arrested in metaphase with nocodazole at 37°C (Fig. 1A). Pds1 was also stabilized in anaphase-arrested cells in which Cdc20 expression had been inhibited, which confirmed that the anaphase instability of Pds1 was due to the activity of APCCdc20, the target of the spindle checkpoint (Fig. 1A).

Fig. 1.

Lack of tension at the kinetochore fails to activate the spindle checkpoint in anaphase, and APCCdc20 activity is necessary for Pds1 degradation in anaphase. (A) (Top, left) A cdc15-2 strain with hemagglutinin (HA) epitope–tagged Pds1 expressed by the GAL1 promoter (PGAL-PDS1-HA) was grown at 37°C to induce anaphase arrest, and then grown for a further hour at 37°C in medium containing galactose in the presence or absence of nocodazole (±Noc). In all the time courses, transcription of Pds1 from the GAL1 promoter was repressed after 1 hour by adding glucose, and samples were taken at the indicated times for Western blot analysis. Pds1 epitope-tagged with HA runs as two distinct bands, the more slowly migrating of which is phosphorylated (32). (Top, right) As a control, a CDC15 strain with PGAL-PDS1-HA was grown in medium containing nocodazole at 37°C to induce a metaphase arrest, and then grown for a further hour at 37°C in medium containing galactose and nocodazole. (Bottom) A cdc15-2 strain with PGAL-PDS1-HA and Cdc20 expressed under the control of the repressible MET3 promoter (PMET -CDC20) was grown at 37°C in medium lacking methionine to induce an anaphase arrest and then grown for a further hour at 37°C in complete synthetic medium (CSM) + galactose + methionine to selectively repress the expression of Cdc20. Equal loading was confirmed using actin (top panels) and Cdc28 (bottom). (B) A cdc15-2 strain with epitope-tagged Pds1 expressed under the endogenous promoter (Pds1-Myc) and PMET -CDC20 was grown at 37°C without methionine to induce an anaphase arrest and then transferred to methionine-containing medium to repress the expression of Cdc20; samples were taken at the indicated times for Western blot analysis with antibodies to the Myc epitope. A cdc15-2 Pds1-Myc strain without PMET -CDC20 (wild type) was grown under the same conditions as a control.

Spindle checkpoint signaling in anaphase matters only because Cdc20 is required after the metaphase-to-anaphase transition. Although inhibiting APCCdc20 stabilizes Pds1 and arrests cells in metaphase (10), stable Pds1 also keeps cells from leaving anaphase (13, 14), and repressing Cdc20 in anaphase leads to the accumulation of Pds1 (Fig. 1B). This shows that Cdc20 activity is required to destabilize Pds1 in anaphase cells, which corroborates the finding that APCCdc20 is needed for anaphase exit (15) and suggests that activating the spindle checkpoint in anaphase cells would prevent mitotic exit.

We asked how the spindle checkpoint is kept inactive during anaphase. Mps1 is a protein kinase whose activity is essential for spindle checkpoint signaling (16), and Mps1 overexpression arrests cells in metaphase by activating the checkpoint (17). Mps1 mRNA levels show little change during the cell cycle (18). In contrast, Mps1 protein abundance varied during the cell cycle (Fig. 2A), rising and falling with those of the mitotic cyclin Clb2, which accumulates as cells enter mitosis and is degraded as they leave it (15). This fall does not occur in cells released from G1 and then arrested at metaphase, which suggests that Mps1 is degraded after the metaphase-to-anaphase transition (Fig. 2A).

Fig. 2.

Mps1 is a cell cycle–regulated protein, and expression of Mps1 in anaphase reactivates the spindle checkpoint. (A) A strain with epitope-tagged Mps1 (Mps1-Myc) under the control of the endogenous MPS1 promoter was released from α-factor arrest and then arrested at either the subsequent metaphase (with nocodazole) or G1 (using α factor). Samples were taken at the indicated times for Western blot analysis with antibodies to the Myc epitope and Clb2 protein. Full-length Mps1-Myc is indicated with an arrow, and the lower band is a degradation product. The relative levels of the different proteins were quantified over the time course. (B, C, and D) MAD2 and mad2Δ strains with epitope-tagged Pds1 (Pds1-Myc) and MPS1 driven by the GAL1 promoter (PGAL-MPS1-myc) were released from a G1 (α factor) block to an anaphase (cdc15-2) arrest at 37°C in the absence of galactose. The anaphase-arrested cells were transferred to medium containing either raffinose to maintain Mps1 repression (–Gal-Mps1) or galactose to induce Mps1 expression (+ Gal-Mps1); at the indicated times after the transfer, samples were taken for Western blot analysis with antibodies to the Myc epitope (B) and antibody to Mad1 to monitor Mad1 phosphorylation (C). (D) After 120 min, the cells were transferred to the permissive temperature for cdc15-2 (23°C), and samples were taken at the indicated times to monitor rebudding as a measure of cell cycle arrest.

As cells enter anaphase, Mps1 levels fall, and the spindle checkpoint is turned off. This coincidence prompted us to ask if the checkpoint could be reactivated in anaphase by overexpressing Mps1 from the GAL1 promoter. Synchronized cdc15-2 cells were released from G1, arrested in anaphase at 37°C, and then treated with galactose to induce Mps1 expression. Spindle checkpoint reactivation, even in the absence of microtubule disruption, was demonstrated by an increase in Pds1 stability, Mad1 phosphorylation, and the inability of cells to rebud after they were returned to the permissive temperature (Fig. 2, B, C, and D). Pds1 stabilization and Mad1 phosphorylation are hallmarks of spindle checkpoint activation (19). Checkpoint reactivation depended on Mad2, an essential component of the spindle checkpoint (4). We ruled out a role for the Bub2-dependent spindle-positioning checkpoint, which keeps cells from leaving anaphase until one spindle pole body has entered the daughter cell (20) (fig. S1, A and B). The dependence on Mad2 indicates that Mps1 overexpression reactivates the spindle checkpoint and that all the components of this pathway downstream of Mps1 are functional in anaphase. The result also suggests that reducing the level of Mps1 helps to inactivate the checkpoint in anaphase.

The amount of Mps1 fell by about 50% as synchronized cells went from metaphase to anaphase (Fig. 3A, 140-min time point). If Cdc20 expression was then repressed while the cells were kept in anaphase, Mps1 accumulated, which indicated a role for APCCdc20 in reducing Mps1 stability during anaphase (Fig. 3A, 140-min to 220-min time points).

Fig. 3.

Mps1 protein abundance falls in anaphase, and Mps1 stability is regulated by the APC. (A) Cells containing epitope-tagged Mps1 (Mps1-Myc) and PMET -CDC20 were released from a G1 (α-factor) arrest to an anaphase (cdc15-2) arrest at 37°C in medium without methionine. After 140 min, 4′,6′-diamidino-2-phenylindole (DAPI) staining revealed that 97% of the cells had arrested in anaphase with separated chromosomes. At this time, the anaphase-arrested cells were transferred to medium containing methionine to repress expression of Cdc20. At the indicated times, samples were taken for Western blot analysis with antibodies to the Myc epitope, and a calibration curve was used to quantify Mps1 levels. (B) Strains containing Mps1N-LacZ fusion proteins expressed under the GAL1 promoter (PGAL-MPS1N-LacZ) were arrested in metaphase (nocodazole), anaphase (cdc15-2), or G1 (α factor), and then transferred to medium containing galactose for 2 hours to induce expression of the fusion protein. Glucose and cycloheximide were added to shut off Mps1 transcription and translation; samples were taken at the indicated times for Western blot analysis with antibody to β-galactosidase. Proteins contained the wild-type Mps1 N-terminal fragment, or fragments with mutation of putative D-boxes (RXXL to AXXA), at the indicated positions. Mutation of the 356D-box in full-length Mps1 also stabilized the protein (fig. S5).

The destruction box (D-box) is a recognition site for APCCdc20 and APCCdh1, the form of the APC active in G1 (21). There are three putative D-boxes (RXXL) in the noncatalytic N-terminal half of the Mps1 protein (fig. S2) (22). We fused the N-terminal portion of Mps1 (Mps1N, amino acids 1 to 450) to Escherichia coli β-galactosidase (LacZ), mutated the putative Mps1 D-boxes, and expressed the protein from the GAL1 promoter. By using this noncatalytic fragment, we avoided the complication that increasing Mps1 expression would activate the checkpoint and would reduce APCCdc20 activity. Mps1N-LacZ expression was induced for 2 hours before transcription was shut off by the addition of glucose. In metaphase-arrested cells, Mps1N-LacZ was largely stable for the duration of the experiment (Fig. 3B). In contrast, in cells arrested in anaphase or G1, this fusion protein was unstable (Fig. 3B). Mutation of the three D-box motifs (267RELL, 319RRAL, 356REVL) together was effective in stabilizing the protein construct in anaphase and G1 (Fig. 3B). Mutation of the 356REVL D-box alone substantially increased the stability of the fusion protein in anaphase and G1, which suggested that this motif is particularly important for regulating Mps1 stability. Deletion of CDH1 or repression of the APC subunit Cdc16 in G1 also stabilized full-length, wild-type Mps1 (fig. S3). These results suggest that Mps1 is a substrate for APCCdc20- and APCCdh1-dependent destruction in anaphase and G1, respectively.

The observations that a spindle checkpoint protein is a target of the APC and that APCCdc20 is inhibited by the checkpoint suggest that these two activities oppose each other in a double-negative feedback loop (Fig. 4A). In this model, the checkpoint inhibits the APC before chromosome biorientation, and this inhibition stabilizes Mps1 and potentiates the checkpoint. Once the chromosomes are aligned, the checkpoint's activity is reduced, the APC becomes more active, and Mps1 is degraded, which prevents subsequent reactivation of the checkpoint.

Fig. 4.

Mps1 and the APC mutually inhibit each other to create a double-negative feedback loop. (A) A model for Mps1 acting via the spindle checkpoint pathway to inhibit APCCdc20 activity, thereby increasing Mps1 stability. (B) Mps1 stability in anaphase-arrested (cdc15-2) cells with PGAL-MPS1-HA grown in galactose to induce expression of Mps1 for 1 hour. Glucose was added to the medium, and samples were taken at the indicated times for Western blot analysis with antibodies to HA epitope. Mps1 stability in anaphase was Mad2-dependent, and Mps1 phosphorylation (clearly visible when Mps1 is tagged with the HA epitope) results in a shift of the Mps1-HA band (indicated with an asterisk) that largely depends on the presence of Mad2. (C) Anaphase-arrested (cdc15-2) MAD2 and mad2Δ cells with Pds1-Myc and a kinase-repressible epitope-tagged Mps1 protein expressed under the GAL1 promoter (PGAL1-mps1-as1-myc) were transiently grown in galactose to induce Mps1 expression. After 2 hours, the cells were shifted to the permissive temperature (23°C) in medium containing glucose, with or without the 1NM-PP1 kinase inhibitor. Rebudding of cells was measured at the indicated times and the samples were used for Western blot analysis with antibody to the Myc epitope (see fig. S4). (D) Checkpoint activation without Mps1 overexpression, in anaphase-arrested (cdc15-2) MAD2 and mad2Δ cells with PMET -CDC20-HA grown in medium containing methionine to repress Cdc20 expression. After 45 min, the cells were shifted to the permissive temperature for anaphase release (23°C), and samples were taken at the indicated times to measure rebudding and used for Western blot analysis with antibodies to Mad1 and the HA epitope (see fig. S6). A strain containing Cdc20 expressed under the endogenous promoter was grown under the same conditions as a control.

To test whether this feedback loop depends on the presence of other checkpoint components, such as Mad2, we measured the stability of Mps1 after a period of high Mps1 expression from the GAL1 promoter in either MAD2 or mad2Δ cells arrested in anaphase. Mps1 was present in greater amounts at steady state and was more stable in MAD2 cells than in mad2Δ cells, which indicated that the feedback loop was activated in the MAD2 cells but not in the checkpoint mutant (Fig. 4B). The stabilized Mps1 in MAD2 cells was highly phosphorylated, consistent with the observation that the stability of Mps1 is influenced by its phosphorylation state (23, 24).

If Mps1 and APC activity form a feedback loop, transient expression of Mps1 should be able to switch the circuit into a state of sustained Mps1 stability and APC inactivity; this state should not be dependent on continued MPS1 transcription. Furthermore, because the checkpoint depends on the kinase activity of Mps1, it should be possible to switch the feedback circuit to its original state by inhibiting this activity. We tested these hypotheses by transiently expressing a repressible allele of Mps1 (mps1-as1) in anaphase. The kinase activity of this allele can be specifically inhibited with 4-amino-1-tert-butyl-3-(1′-naphthylmethyl)pyrazolo[3,4-d] pyrimidine (1NM-PP1), an inhibitor of genetically manipulated protein kinases (25). mps1-as1 was expressed in anaphase-arrested, cdc15-2 cells; the cells were shifted to 23°C; and Mps1 expression was shut off with glucose in the presence or absence of the inhibitor. We measured the ability of the checkpoint to maintain the “on” state by monitoring the time taken for cells to rebud and the stability of epitope-tagged Pds1 and Mps1 (Fig. 4C and fig. S4). In the absence of 1NM-PP1, prolonged checkpoint activation was observed in MAD2 cells, but not in checkpoint (mad2Δ) mutant cells. Treatment with 1NM-PP1 reduced the duration of checkpoint activation in MAD2 cells. Although continued kinase activity was essential, continued Mps1 expression was not required to prolong checkpoint signaling and delay rebudding. This suggests that transient Mps1 expression reduces APC activity below the threshold at which Mps1 is sufficiently stable to maintain the feedback circuit in the “checkpoint on” state.

If Mps1 stability regulates checkpoint activity in anaphase, increasing the stability of Mps1 should reactivate the spindle checkpoint, without overexpressing the protein. Repression of Cdc20 in anaphase stabilizes Mps1 (Fig. 3A), and we tested whether this repression was sufficient to activate the spindle checkpoint. Cdc20 expression was repressed in anaphase-arrested (cdc15-2) cells for 45 min at 37°C before transfer to 23°C to reactivate Cdc15. We compared rebudding time in checkpoint-proficient (MAD2) and checkpoint-deficient (mad2Δ) cells (Fig. 4D). The MAD2 cells stayed arrested for the duration of the experiment, whereas the checkpoint-deficient mutant cells and control cells in which Cdc20 expression had never been repressed rebudded (Fig. 4D). We found high levels of Mad1 phosphorylation and reduced Cdc20 abundance in the checkpoint-proficient cells, consistent with checkpoint activation (19, 26) (fig. S6). The simplest interpretation of these results is that a small amount of Cdc20 remains in checkpoint-deficient cells after a short period of Cdc20 repression, which allows exit from anaphase. In the checkpoint-proficient cells, repressing Cdc20 stabilizes checkpoint components, which allows the absence of tension at anaphase kinetochores to reactivate the checkpoint. Checkpoint activation further inhibits and destabilizes Cdc20 (26), which initiates a positive-feedback loop that makes it difficult for cells to exit mitosis, even after Cdc15 activity has been restored at the permissive temperature. We argue that the cell cycle engine in these cells has switched to a metaphase-like state. This result suggests that the anaphase stabilization of Mps1, and potentially other checkpoint components, permits the reactivation of the spindle checkpoint. We propose that it is the destabilization of these components that inactivates the checkpoint in anaphase in a normal cell cycle.

The regulation of other spindle checkpoint or kinetochore components likely plays a role in anaphase checkpoint inactivation. The budding yeast aurora kinase, Ipl1, activates the spindle checkpoint by creating unattached kinetochores, and Ipl1 moves from kinetochores to spindle microtubules shortly after the initiation of anaphase (5, 27, 28). Microtubule attachment to kinetochores in anaphase may be stabilized by the loss of Ipl1, helping to keep the checkpoint inactive. However, Ipl1 mutants respond to treatment with nocodazole, whereas anaphase-arrested cells do not (Fig. 1A), which suggests that additional factors, such as Mps1 degradation, have turned off the checkpoint in anaphase (29). The organization of other “chromosomal passenger proteins” also changes as cells enter anaphase (30), as do spindle microtubule dynamics (31), and these factors may also influence checkpoint behavior in anaphase. Finally, the checkpoint destabilizes Cdc20, as well as inhibits its activity, which reinforces the mutual antagonism between the checkpoint and APCCdc20.

We have presented evidence for a mechanism that inactivates the spindle checkpoint as yeast cells enter anaphase. When mitosis starts, the APC is off, the checkpoint is on, and checkpoint proteins are stable. As long as one chromosome has not aligned, the checkpoint inhibits the APC. When this chromosome biorients, a threshold is crossed, the APC becomes active, cells enter anaphase, and the destruction of Mps1 (and possibly other checkpoint proteins) permanently inactivates the checkpoint. The opposing activities of the checkpoint and the APC let cells switch rapidly between prometaphase, when they can sensitively monitor chromosome alignment, and anaphase, when they are irreversibly committed to entering the next cell cycle, despite the lack of tension at the kinetochores.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1127205/DC1

SOM Text

Figs. S1 to S6

Table S1

References and Notes

View Abstract

Navigate This Article