APC/C- and Mad2-mediated degradation of Cdc20 during spindle checkpoint activation

Introduction

The Spindle Assembly Checkpoint (SAC) is elicited by improperly aligned chromosomes during prometaphase. A single unattached or tensionless kinetochore can activate the SAC and prevent the onset of anaphase.1 Mechanistically, this inhibition is achieved through the delayed degradation of securin and Cyclin B by APC/C^Cdc20.² APC/CCdc20 is a multi-subunit E3 ubiquitin ligase that targets many regulatory proteins for proteasome-dependent degradation during mitosis allowing first the metaphase-anaphase transition and then mitotic exit (the latter in collaboration with APC/C^Cdh1).³ As the activator of APC/C in mitosis, Cdc20 is the essential target of the SAC,^4,5 but how Cdc20 is inhibited by the SAC remains unclear. Our current understanding of the SAC suggests that Mad2, as part of the Mitotic Checkpoint Complex (MCC; containing Mad2, Mad3/BubR1, and Bub3), binds Cdc20, inhibiting its ability to activate APC/C, impeding the transition to anaphase.² The SAC also dampens APC/C activity through inhibitory phosphorylation of multiple Serine/Threonine residues on Cdc20 by Cdk1 and the spindle checkpoint protein Bub1,^6,7 although it is unclear how phosphorylation reduces the ability of Cdc20 to activate APC/C. In Saccharomyces cerevisiae, SAC activation leads to the targeted ubiquitination of Cdc20 by APC/C, with consequent degradation by the proteasome.⁸ Failure to reduce the level of Cdc20 during SAC activation causes cells to slip through the mitotic checkpoint, resulting in aneuploidy. Here, we investigated the regulation of Cdc20 stability during SAC activation in mammalian cells and explored the contribution of spindle checkpoint proteins to Cdc20 stability.

Results and Discussion

The level of Cdc20 protein is regulated by the cell cycle. Beginning with accumulation in S phase, the level of Cdc20 peaks in mitosis before decreasing in the late stages of mitosis after the degradation of Cyclin B and securin, reaching its nadir in G₁.³ Upon performing a nocodazole synchronization, shake-off, and release (Fig. 1A), we surprisingly found that the Cdc20 level of cells in prometaphase was not dramatically higher than that of G₁ cells and was much lower than that of the G₂ cells, which remain attached following nocodazole treatment and shakeoff (Fig. S1 and S2). To further investigate this observation, we lysed prometaphase and G₂ cells from both HeLa and U-2 OS cells, and the differences in the levels of Cdc20 and several other mitotic proteins, including Cyclin A, Cyclin B, Plk1 and Cdh1, were analyzed by immunoblotting (Fig. 1B). As expected, the levels of Cyclin B, Plk1 and Cdh1 did not decrease in nocodazole-arrested mitotic cells, and Cyclin A was degraded in a spindle checkpoint-independent manner, as previously described.⁹ Notably, Cdc20 levels were reproducibly lower in the nocodazole-arrested mitotic cells in comparison to the G₂ cells. We reasoned that this downregulation of Cdc20 in prometaphase could be a specific response to mitotic spindle checkpoint activation by nocodazole and that Cdc20 levels would not decrease in normal mitotic cells without spindle poison treatment. To test this hypothesis, we released U-2 OS cells from a thymidine arrest (S phase) and collected mitotic cells by shake-off 13 hours after release. As expected, these normal (i.e., not spindle-poisoned) mitotic cells had a similar amount of Cdc20 compared to the attached, G₂ cells (Fig. 1C). To exclude the possibility that Cdc20 downregulation was due to prolonged (16 hours) treatment with nocodazole, we released cells from a thymidine block into medium containing nocodazole for only eight hours. As shown in Figure 1D, Cdc20 was still degraded in the mitotic cells exposed to nocodazole for the shorter time period, suggesting that the downregulation of Cdc20 is not due to mitotic adaptation.

Next we tested whether inhibition of the proteasome could rescue the Cdc20 level in nocodazole-arrested mitotic cells. Adding MG132 to nocodazole-arrested cells restored Cdc20 to a level comparable to G₂ cells (Figure 2). The increase in Cyclin A demonstrated the effectiveness of MG132 in inhibiting the proteasome. To confirm that the degradation of Cdc20 in mitosis was specific to spindle checkpoint activation, we treated cells with taxol, another microtubule poison, or a taxol and MG132 combination, and results similar to nocodazole treatment were observed (Fig. S3). Therefore, we concluded that the degradation of Cdc20 is a specific response to SAC activation, and this degradation is proteasome dependent.

The ubiquitin ligase required for Cdc20 degradation during SAC activation was unknown. The only ubiquitin ligase known to induce Cdc20 degradation is APC/C^Cdh1, which ubiquitinates Cdc20 during the G₁ phase of the cell cycle. In mammalian cells, Cdh1 recognizes the KEN box motif at the N-terminus of Cdc20 to target it for degradation.¹⁰ Recently, another degron motif, the CRY box, was also reported to be partially responsible for the APC/CCdh1-mediated degradation of Cdc20 in mouse oocytes.¹¹ However, in mitosis, Cdh1 is largely inactive due to phosphorylation by mitotic kinases,³ making it unlikely that APC/C^Cdh1 is activated by SAC to induce degradation of Cdc20. Alternatively, the binding of spindle checkpoint proteins, such as Mad2 or BubR1, to Cdc20 in response to SAC activation could induce degradation by the APC/C core complex. Studies in S. cerevisiae suggest that the SAC regulates Cdc20 stability in mitosis, and the degradation of Cdc20 induced by SAC activation does not depend on Cdh1, instead requiring Mad2 and the core components of the APC/C complex.⁸ To test this possibility in mammalian cells, we performed siRNA-mediated knockdowns of Cdh1 and two core components of the APC/C, Cdc27 and APC2 (Fig. 3). In accordance with the findings in budding yeast, knockdown of Cdh1 did not rescue the Cdc20 protein level, while knockdown of either Cdc27 or APC2 increased the level of Cdc20. Thus, the core APC/C is required for the destruction of Cdc20 during SAC activation in mammalian cells, and it remains unknown whether a novel, unidentified adapter is required for this process.

Finally, we tried to identify the motifs required for Cdc20 degradation during SAC activation in mitosis. To this end, we constructed a panel of characterized^6,10-15 and novel Cdc20 mutants (Fig. 4A and B and Fig. S4), focusing on functional domains required for Cdh1-dependent degradation, Cdc27 binding (as a readout of APC/C binding), Cdk1 phosphorylation, Mad2 binding and BubR1 binding. We generated HeLa cells stably expressing the Cdc20 mutants at relatively low level by retroviral transduction and assessed the stability of these mutants in nocodazole-arrested mitosis in comparison to the remaining, attached G2 cells (Fig. 4C). In accordance with our siRNA results, the 166-499 and 1-470 mutants showed increased stability, likely because they failed to bind APC/C.¹³ Some mutants, such as 1-479, 1-489, IRAA and ΔC-Box, were still unstable in mitosis, despite reduced binding to APC/C (ref. ¹³ and data not shown). Notably, R132A, which retains robust binding to both APC/C and BubR1 but lacks Mad2 binding (Fig. 4B), is more stable than wild type Cdc20, though not as stable as the two APC binding mutants, suggesting that Mad2 is involved in the APC/C-dependent degradation of Cdc20 during SAC activation.

These results, combined with previous studies in budding yeast, recommend a model in which the proteasome-dependent degradation of Cdc20 is the initial response to checkpoint activation, preventing premature chromosome segregation. In such a model, upon SAC activation, immediate, Mad2- and APC/C-mediated degradation of Cdc20 is required to enact acute inhibition of APC/C^Cdc20 activity by reducing Cdc20 levels before a sustained checkpoint response can be implemented. Without this early step in checkpoint establishment, the existing levels of Cdc20 could produce the threshold APC/C activity for overriding early attempts to implement the checkpoint, causing some cells to progress through mitosis and become aneuploid. Notably, this scenario appears to occur in yeast, where mild overexpression of Cdc20 can abrogate the SAC.⁸

The observed Mad2 binding dependence for Cdc20 degradation is also supported by studies in yeast demonstrating that Mad2 binding mutants of Cdc20 can bypass the SAC.⁴ The potential roles of other spindle checkpoint proteins, such as BubR1, in this process remain unknown, but it is likely that they also contribute to Cdc20 degradation during SAC activation. While this manuscript was in preparation, it was reported that Cdc20 preferentially binds to BubR1/Bub3 in mitotic cells arrested with spindle poisons and that Mad2 is more responsible for initiating the formation of this Cdc20, BubR1 and Bub3 complex.¹⁶ Other proteins, such as a SAC-specific APC/C activator, may also be required for Cdc20 degradation in response to SAC activation. Finally, it remains unclear why a small population of Cdc20 is spared from Mad2- and APC/C-mediated degradation. Perhaps this Cdc20 is sequestered in another complex or it may remain stable due to deubiquitinating enzyme activity, as with other APC/C substrates.¹⁷ Clearly, further research into the coordination of spindle checkpoint protein activity for the induction of Cdc20 degradation and checkpoint establishment is required.

Materials and Methods