A large number of mitotic regulators are degraded at the end of mitosis. These include Cdc20, Aurora B, Plk1, etc. and are most likely the substrates of APC^Cdh1 (Table 1) 16171819202122232425262728293031323334353637383940414243. Although many of these mitotic regulators did accumulate in the absence of Cdh1, they were eventually degraded, probably because of the stabilization of Cdc20 that compensates for the loss of Cdh1 1314. As a result, Cdh1-deficient cells could still proliferate. However, these cells did accumulate mitotic errors and display difficulties in completing cytokinesis, resulting in the formation of binucleated cells at a high frequency 1314.

The lethality associated with the loss of Cdh1 is largely a result of failed development of placenta, an essential organ for embryonic life in mammals 1314. In the placenta, there is a special cell type, giant cells which are polyploid. The polyploidy is acquired through endoreplication. In the absence of Cdh1, the placental giant cells failed to form, suggesting that APC^Cdh1 is required for DNA endoreplication. To undergo endoreplication, a cell must not enter mitosis after a round of DNA synthesis. One critical factor for mitotic entry is cyclin B1, a substrate of APC^Cdh1 33. Therefore, it is possible that the failed formation of giant cells in Cdh1 mutant embryos was a result of the inability of the would-be giant cells to block the accumulation of cyclin B1. Indeed, cyclin B1 was readily detectable in the mutant placenta in the area where the giant cells should reside 14.

Like an endoreplicating cell, cells with their DNA damaged do not enter mitosis either. A recent study revealed that APC^Cdh1-medicated degradation of Plk1 played an important role in preventing mitotic entry of DNA-damaged cells 44. Under normal conditions, Cdh1 is kept inactive from late G1 to early mitosis. When cells suffered DNA damage, Cdc14B is translocated out of nucleoli into nuclearplasm where the phosphatase dephophorylates and activates Cdh1 44. It is unknown if a similar mechanism is employed by the placental giant cells to activate Cdh1 in G2 to prevent mitotic entry.

In Drosophila, loss of fzr leads to an extra division cycle in the epidermis, a likely result of the accumulation of mitotic cyclins in G1 45. In HeLa cells, knocking down the expression of Cdh1 causes stabilization of Skp2 4647, a F-box containing protein responsible for bringing p27^Kip1 to the SCF complex for ubiquitination 4849. As a result, p27 is destabilized in the cells with impaired Cdh1 function and the G1 phase is shortened in these cells.

Our recent work identified Ets2 as a new substrate of APC^Cdh1 14. Ets2 is a member of the Ets family of transcription factors which share a unique DNA binding domain, the ETS domain 5051. The first Ets protein was identified as transduced mutant form of Ets1, v-Ets, in a retrovirus, E26 (

E

T

s

It was shown previously by Nasmyth's group that inactivation of APC/C by deleting its subunit APC2 in adult hepatocytes induced these otherwise quiescent cells to re-enter cell cycle 68. This is most likely a result of the loss of APC^Cdh1 activity and subsequent accumulation of Skp2 and Ets2.

To enter S-phase, APC^Cdh1 must be inactivated. Several different mechanisms are in place to contain APC^Cdh1 6970. First, the ubiquitination of APC-specific ubiqutin-conjugating enzyme (E2) UbcH10 by APC^Cdh1 itself provides a negative feedback mechanism that would eventually destroy APC^Cdh1 activity 7172. Second, as Cdk activity accumulates, Cdh1 is phosphorylated. The phosphorylation promotes Cdh1 dissociation from APC 1073. Third, phosphorylated Cdh1 is targeted by SCF liagse 74, further limiting the activity of APC^Cdh1. Finally, in late G1 phase, E2F activates the transcription of early mitotic inhibitor-1 (Emi1)/Rca1, which inhibits the activity of APC/C^Cdh1 as a pseudo-substrate 7576.

Opposite to the expected high proliferation rates in Cdh1-deficient cells, Cdh1^-/- mouse embryonic fibroblasts (MEFs) proliferate poorly and entered senescence after only a few passages 77. We found that the induction of p16^Ink4a in these cells was the cause for the reduced proliferation potential. The reason for the increased levels of p16 in Cdh1^-/- MEFs could simply be that p16 is a substrate of APC^Cdh1. Alternatively, it could be that a transcriptional activator of p16 is a substrate. It turned out that the latter was the case 14. We demonstrated that Cdh1-deficiency induced upregulation of p16 was a result of Ets2 stabilization. It was shown previously that Est2 could activate p16 expression 78. The expression of p16 is regulated by a number of factors in response to various stimuli. Under normal conditions, Ets2 levels are not enough to induce p16 expression and senescence because of Id1 and Bim1 (Fig. 2A). Id1 interacts with Ets2 and blocks its transcriptional activation of p16 7879, whereas Bmi1 directly represses p16 promoter 80. Our identification of the APC^Cdh1-Ets2-p16 axis shows the delicacy of the balance that determines the level of p16 expression. Loss of Cdh1 leads to increases in the levels of Ets2 to the point that it overcomes Id1's inhibition and results in senescence (Fig. 2B). This would imply that Ets2 activity required for p16 activation and that for other targets are different, which remains to be determined.

Overactivation of Ras signaling could also cause senescence in primary cells 818283. It is likely that Ets2 mediates, at least in part, this senescence effect of Ras signaling (Fig. 2C). It remains to be determined however if Ras signaling interacts with APC^Cdh1. The phosphorylation of Ets2 by Erk may interfere with the recognition of Ets2 by Cdh1, for example, leading to stabilization of Est2. Therefore, when Ras signaling is overactivated, APC^Cdh1 would only be able to provide limited balancing function against Ets2. By placing p16 under the transcriptional control of Ets2, evolution has setup a failsafe mechanism to prevent unwarranted proliferation.

Two places where appreciable levels of Cdh1 (and APC subunits) but not Cdc20 are expressed in adult mice are the brain and the liver 84. In the liver, APC^Cdh1 is likely required for preventing hepatocytes from reentering the cell cycle spontaneously (see above). What is the function of APC^Cdh1 in postmitotic neurons? The depletion of Cdh1 expression in primary neuron of the cerebellar cortex promotes significant elongations of axons, indicating a role of APC^dh1 in limiting axonal growth 85, possibly through regulating the abundance of SnoN and Id2 8687. APC^Cdh1 may also control pattering of axon growth in the mammalian brain 85.

We assessed neural functioning of mice that are heterozygous Cdh1 mutant. Electrophysiology studies of hippocampus revealed a deficit in late phase long-term potentiation (L-LTP), a process that underlies synaptic plasticity and is known to depend on both protein synthesis and degradation 88. In behavior tests, we found that Cdh1 heterozygous mice performed poorly in contextual fear conditioning 14, a hippocampus-dependent process 8990, but no difference was observed in cued fear condition which is less dependent on hippocampus 8990. Similar behavior findings were also reported by Malumbres' group 13. It is unclear at the moment what are the substrates of APC^Cdh1 that play a role in memory formation in mammals. However, studies from lower organisms may provide a clue. In C. elegans, the abundance of GLR-1 glutamate receptors in the ventral nerve cord is regulated by anaphase promoting complex, albeit unlikely to be in a direct manner 91. In Drosophila, Liprin-α, a multidomain scaffolding protein which is localized to synapses and regulates synaptic activities, seems to be a substrate of APC^Cdh1 92. It remains to be determined if mammalian homologues of Liprin-α and GLR-1 are regulated by APC^Cdh1 and if their misregulation can account for the learning and memory defects displayed by Cdh1 heterozygous mice.