The anaphase-promoting complex (APC) mediates the ubiquitination and degradation of key M-phase regulators, including cyclins and the anaphase inhibitor securin. Intriguingly, securin can also inhibit the degradation of cyclin B. This competition between substrates permits the accumulation of enough cyclin to drive entry into M phase.
The APC is a large ubiquitin ligase that promotes the degradation of several cell-cycle regulators during G1 and M phase. Securin is a key APC substrate and the one whose degradation gave rise to the name APC: securin binds to and inhibits a protease termed separase, which, when released following securin degradation, cleaves proteins that hold sister chromatids together, thereby initiating anaphase. On page 445 of this issue, Marangos and Carroll1 show that securin also functions as an APC inhibitor, facilitating the accumulation of another APC substrate, cyclin B, to promote mitosis. Thus, securin joins a growing family of mechanistically related APC inhibitors that range from pure inhibitors to those that are also APC substrates.
As with other ubiquitin ligases (or 'E3s'), the APC brings substrates into close proximity of a ubiquitin-conjugating enzyme (an 'E2') so that the E2 can transfer ubiquitin to the substrate. Although it contains at least thirteen distinct proteins, the APC still needs help to bind most substrates. The so-called APC activators, Cdc20 and Cdh1, fulfill this role in somatic cells by binding both APC and the substrates. APC substrates contain degradation signals, the most common of which are the Destruction Box and the KEN Box. Cdc20 and Cdh1 recruit substrates to the APC by binding directly to these degradation motifs. An emerging theme in the regulation of the APC is the competitive inhibition of substrate binding to Cdc20 or Cdh1.
The spindle-assembly checkpoint inhibits APCCdc20 even when only one chromosome is not properly attached to the mitotic spindle. Many proteins are involved in this checkpoint, and nearly as many models have been proposed for how it functions biochemically. Recent findings2,3 focus on the role of Mad3/BubR1 (termed Mad3 in yeast and BubR1 in all other species), which, along with Mad2, have long been known to bind to Cdc20. Mad3/BubR1 proteins contain two conserved KEN Boxes and Mad3 from Saccharomyces cerevisiae also contains an important Destruction Box. Despite the presence of these degradation motifs, yeast Mad3 is not an APCCdc20 substrate and is stable during mitosis. Instead, these motifs dock Mad3 to Cdc20, thereby preventing substrate binding and stabilizing securin. Because of its mitotic stability, Mad3 has been called a pseudosubstrate inhibitor of the APC, a fancy term for a competitive inhibitor that is not itself a substrate. In most species, Mad3/BubR1 binding to Cdc20 depends on the simultaneous binding of Mad2 to Cdc20. Mad2, in turn, is conformationally activated at chromosome kinetochores that are not attached to the mitotic spindle. Thus, a plausible mechanism for extinguishing APCCdc20 inhibition by Mad3/BubR1 is through the attachment of all kinetochores to the spindle to prevent Mad2 activation, coupled with the spontaneous conformational inactivation of Mad2.
The vertebrate protein Emi1 was the first characterized pseudosubstrate inhibitor of the APC, in this case of APCCdh1 (ref. 4). In contrast to Mad3/BubR1, which functions in the spindle assembly checkpoint, Emi1 controls the normal timing and coordination of the cell cycle. Emi1 has a conserved Destruction Box through which it binds to Cdh1 (and probably also to the APC directly). This binding inhibits APCCdh1, particularly during S phase, thereby facilitating the accumulation of mitotic cyclins5. Emi1 also contains a zinc-binding region (ZBR) that is necessary for APC inhibition. Interestingly, mutation of the ZBR converts Emi1 from a pseudosubstrate inhibitor into an APC substrate. This observation hints at the close connection between pseudosubstrates and actual substrates, and raises the question of why other pseudosubstrate inhibitors, such as Mad3/BubR1, are not themselves ubiquitinated by the APC. Emi1 is not an APC substrate but instead, is inactivated by degradation following ubiquitination by the SCF, the other major class of cell-cycle ubiquitin ligases.
Although there is no orthologue of Emi1 in S. cerevisiae, the recently described Acm1p protein seems to be its functional equivalent6,7. Similarly to Emi1, Acm1p inhibits APCCdh1 during S phase and into mitosis, and can inhibit APCCdh1 activity in vitro. Acm1p contains candidate Destruction Box and KEN Box motifs and is rapidly degraded late in mitosis; however, it is not an APC substrate8 and its degradation motifs are required for its binding to Cdh1p and for its ability to inhibit the APC, both in vitro and in vivo (Denis Ostapenko, J.L.B., Ruiwen Wang, and M. J. S., unpublished data). As with Emi1, Acm1p function is terminated by its rapid degradation by the proteasome. However, unlike Emi1, Acm1p is not ubiquitinated by an SCF complex and it is stabilized during S phase and M phase by cyclin-dependent kinase-mediated phosphorylation.
In contrast to Mad3/BubR1, Emi1 and Acm1p, the fission yeast Mes1 protein is both a competitive APC inhibitor and an APC substrate9. Mes1 functions during meiosis in Schizosaccharomyces pombe to prevent the complete degradation of a B-type cyclin by the APC between the first and second meiotic M phases. As with the pseudosubstrate inhibitors, Mes1 uses a Destruction Box and a KEN Box to compete with substrates for binding to the APC activator. Interestingly, release of Mes1 required its APC-mediated ubiquitination. Thus, Mes1 is both an APC inhibitor and an APC substrate. It is still unclear whether Mes1 is an APC substrate — perhaps a weak one — and at the same time, an APC inhibitor, or whether these two phases are temporally distinct. Perhaps Mes1 is an inhibitor of one APC activator and a substrate of a second one that is expressed at a slightly later time in meiosis.
This range of APC inhibitors brings us back to the work of Marangos and Carroll on securin1. They studied prophase-arrested mouse oocytes, which have a low constitutive APC activity. This activity creates a balance between the synthesis and degradation of cyclin B, which is necessary for the activation of the Cdc2/CDK1 cyclin-dependent protein kinase and entry into M phase. They found that the modest increase in securin level produced by injection of securin mRNA increased the rate at which oocytes entered M phase. This effect was seen only with wild-type securin and not when the Destruction Box and KEN Box of securin were mutated or when mRNA for an unrelated APC substrate was injected, and resulted in a small increase in cyclin B levels. Importantly, ablation of securin mRNA reduced the rate of cyclin B accumulation and significantly reduced the rate at which cells entered M phase, an effect that required APCCdh1 function. Thus, securin seems to be a bona fide inhibitor of the APC-mediated degradation of cyclin B. Similarly to the APC inhibitors discussed previously, securin seems to inhibit the APC directly, as the effect required that securin contain its degradation motifs, which are necessary for Cdh1 binding. As with Mes1, extinguishing securin inhibition of the APC is accomplished through its APC-mediated degradation.
What is remarkable here is that securin is simultaneously an APC inhibitor and an APC substrate. The small effect of a modest change in securin level on cyclin accumulation indicates that the system is very sensitive to absolute cyclin B concentrations and suggests that the APC, or its activators, are operating near capacity, so that the level of one APC substrate affects the degradation of another substrate. This type of competition may be a consequence of a process described as substrate ordering10, in which a processive APC substrate such as securin, which undergoes multiple rounds of ubiquitination without substrate dissociation from the APC, will be poly-ubiquitinated and degraded more rapidly than a substrate that is ubiquitinated in a distributive manner. If the capacity of the APC is limiting, then substrate ordering will also produce a situation in which a processive substrate inhibits the degradation of a distributive substrate.
To continue this work, future studies should include biochemical recapitulation and extension of the findings in oocytes. It will also be important to determine the generality of the APC inhibitor function of securin. Is this effect limited to meiosis or is it a universal feature of M-phase regulation? Intriguingly, evidence suggesting that this may be a general feature of M-phase regulation comes from studies in S. cerevisiae showing that securin/Pds1p inhibits destruction of mitotic cyclin11,12, although further work is needed to determine whether this effect is mediated by direct inhibition of the APC. An important unanswered question is why M phase has evolved this way. The ability of securin to inhibit the APC and permit cyclin B accumulation may be a means of coupling these processes so that the cell does not enter M phase before attaining a sufficient level of securin, to prevent premature initiation of anaphase. Finally, why is competition, either with pseudosubstrates or ordinary substrates, such a common mechanism for inhibiting the APC? Perhaps it is a natural step to create an APC inhibitor from an APC substrate. Alternatively, regulation at the level of substrate binding to an APC activator may allow the cell to finely tune the inhibition to only certain APC substrates at specific times. For example, some APC substrates (such as Nek2A and cyclin A) that seem able to bind to the APC directly, could still be degraded when Cdc20 or Cdh1 are bound to an inhibitor protein. This possibility may explain how cyclin A is actively degraded during the spindle-assembly checkpoint13. Such regulatory subtleties would not be possible if an inhibitor targeted the APC or a ubiquitin-conjugating enzyme directly.