Who guards the guardian? Mechanisms that restrain APC/C during the cell cycle.

Abstract

The cell cycle is principally controlled by Cyclin Dependent Kinases (CDKs), whose oscillating activities are determined by binding to Cyclin coactivators. Cyclins exhibit dynamic changes in abundance as cells pass through the cell cycle. The sequential, timed accumulation and degradation of Cyclins, as well as many other proteins, imposes order on the cell cycle and contributes to genome maintenance. The destruction of many cell cycle regulated proteins, including Cyclins A and B, is controlled by a large, multi-subunit E3 ubiquitin ligase termed the Anaphase Promoting Complex/Cyclosome (APC/C). APC/C activity is tightly regulated during the cell cycle. Its activation state increases dramatically in mid-mitosis and it remains active until the end of G1-phase. Following its mandatory inactivation at the G1/S boundary, APC/C activity remains low until the subsequent mitosis. Due to its role in guarding against the inappropriate or untimely accumulation of Cyclins, the APC/C is a core component of the cell cycle oscillator. In addition to the regulation of Cyclins, APC/C controls the degradation of many other substrates. Therefore, it is vital that the activity of APC/C itself be tightly guarded. The APC/C is most well studied for its role and regulation during mitosis. However, the APC/C also plays a similarly important and conserved role in the maintenance of G1 phase. Here we review the diverse mechanisms counteracting APC/C activity throughout the cell cycle and the importance of their coordinated actions on cell growth, proliferation, and disease.

Overview

It is well established that cell cycle progression is driven by Cyclin-CDK kinase complexes. Cyclin expression is tightly controlled during cell cycle, with upregulation determined by transcriptional changes and downregulation triggered by proteolytic destruction. The destruction of multiple Cyclins, as well as dozens of other cell cycle regulated proteins, is controlled by the APC/C, a megacomplex E3 ligase and core component of the cell cycle oscillator. In this review, we will introduce the APC/C by providing a brief, historical overview related to the discovery of Cyclin proteolysis and how this precipitated the discovery of the enzyme controlling their degradation. This review differentiates itself from other excellent reviews on APC/C in that we focus on the growing body of evidence highlighting the role and regulation of APC/C outside of mitosis. We discuss the contribution of APC/C to G0/G1 phase and its potential role in tumor suppression. We highlight recent studies that suggest APC/C inactivation commits cells to cell cycle progression, that it could be inactivated or antagonized in cancer to promote proliferation, and the consequences of its activation state on cell cycle control, proliferative decision-making, and G1/S checkpoint function. Throughout the review, we often refer to APC/C as “active” or “inactive”, a clear over-simplification given the dynamic and complex enzymology of APC/C. Nevertheless, this captures the notion that at specific points in the cell cycle (e.g. late mitosis) its activity is very high, and at others (e.g. in S-phase) its activity is quite low or undetectable by conventional approaches.

Cyclin degradation and the discovery of APC/C

Cyclin-CDK complexes serve as signaling hubs that direct progression through the cell cycle. The activity and identity of Cyclin-CDK rose to prominence with the convergence of genetic screens in yeast and biochemical studies in human cells and amphibian oocytes. These latter studies uncovered diffusible factors, which could promote mitosis in cultured cells [1], and maturation in oocytes [2]. The molecular identity of the maturation promoting factor (MPF) was pursued by several groups. Classic studies showed that MPF activity cycled as eggs moved through cell division and was precipitously lost at the end of mitosis [3]. Then, leveraging a recently developed frog egg extract system, the Kirschner lab demonstrated that Cyclin synthesis, as well as Cyclin destruction, played an important role in the oscillating activity of MPF [4]. Shortly thereafter, the identity of MPF became clear, along with the realization that it was composed of two components [5–7]. One was Cyclin, which had been discovered years earlier as a protein whose abundance oscillates during early development in the embryos of sea urchins and clams [8,9]. The second was a kinase, later named CDK, that corresponded to the conserved Cdc2/Cdc28 kinase, which had been shown to be a key driver of cell cycle progression that was conserved across enormous evolutionary timescales [10,11].

The embryonic Cyclins, Cyclin A and Cyclin B, first described by Hunt and colleagues, control MPF activity during early development [8,9]. While transcriptional changes could account for Cyclin accumulation and MPF activation, how MPF was inactivated at the end of mitosis remained elusive. The answer came with the discovery that Cyclin degradation is controlled by the ubiquitin pathway [12]. It is notable that at the time that Cyclin was first observed, the ubiquitin system had just recently been discovered; however, Alexander Varshavsky predicted the role of ubiquitination in controlling Cyclin abundance years before it was shown experimentally [13].

In 1995, companion papers from the Kirschner and Hershko labs reported the identification of a large, multi-subunit complex capable of promoting the ubiquitination of Cyclin B, which they independently named the Anaphase Promoting Complex and Cyclosome (APC/C), respectively [14,15]. Notably, many APC/C subunits had been identified decades earlier by Hartwell and colleagues as cell division cycle (cdc) mutants in budding yeast [16].

Since the discovery of APC/C, many labs have scrutinized when and how it is regulated and the mechanisms underlying the ubiquitination of its substrates. Most studies have focused on the essential role of the APC/C in mitotic exit. Importantly, APC/C is the key downstream effector of the mitotic spindle assembly checkpoint (SAC), which prevents exit from mitosis until all sister chromatid pairs achieve bipolar attachment to mitotic spindle microtubules. Interested readers are directed to reviews that describe the details of SAC signaling [17,18]. More recently, the role of mitotic post-translational modifications of APC/C in controlling its dynamic activity has furthered our understanding of the biochemical regulation of cell division, and represents an important advance in our understanding of APC/C biology. Interested readers are pointed to papers detailing the role and mechanisms of phosphorylation [15,19–25], SUMOylation [26,27], and acetylation [28].

In addition, while primarily studied in mitosis and G1 phase of actively proliferating cells, APC/C is re-activated after DNA damage. Specifically, DNA damage caused by doxorubicin during S/G2 leads to APC/C re-activation to avoid entry into mitosis while allowing cells to intiate repair [29]. Furthermore, in the nervous system, APC/C restrains axonal growth to maintain neurons in a quiescent state [30,31].

Over the last five years, advances in cryo-electron microscopy, coupled with the ability to purify the APC/C from baculovirus-infected insect cells, provided atomic level understanding of the APC/C structure and enzymology. Curious readers should refer to other, excellent reviews discussing APC/C structure, enzymology, spatiotemporal regulation, and substrate recognition [17,32–35]. Nevertheless, a brief introduction to APC/C structure and architecture follows in order to better understand its role and regulation in the cell cycle.

APC/C structure and function

The core of the APC/C is composed of at least fourteen different proteins, some in multiple copies, that assemble into a tremendous 1.2 MDa macro-molecular ubiquitinating machine. To our knowledge, APC/C represents the largest E3 ligase ever described. It is hitherto unknown why the APC/C is so large. To put its size in perspective, the APC/C is larger than the catalytic core of a ribosome. Many E3 ubiquitin ligases rely on substrate receptor subunits to designate targets for ubiquitination. Likewise, during somatic cell cycles, the APC/C uses either of two substrates receptors, Cdc20 and Cdh1/Fzr1 (hereafter referred to as Cdh1). The activity of APC/C depends on its association with these coactivators, whose primary role is to recruit substrates to the APC/C for ubiquitination.

The Cdc20-bound form of APC/C functions during the metaphase to anaphase transition where it catalyzes the degradation of Cyclin B and Securin to promote mitotic progression. Then in late mitosis, the Cdh1-bound form of APC/C is activated. Important for the discussion below, APC/C^Cdh1 remains active throughout G1 as well as during quiescence (G0). The APC/C also relies on the sequential activity of a pair of E2 ubiquitin conjugating enzymes, UBCH10/UBE2C and UBE2S. First, UBE2C decorates substrates with a combination of very short ubiquitin chains and/or ubiquitin monomers. Then, UBE2S extends these chains, generating degradation signals for substrates [36,37]. Notably, UBE2S assembles non-canonical ubiquitin chains, linked through K11 in ubiquitin [38–41]. In addition, more recent evidence points to an ability of APC/C to form branched or heterotypic chains, and these unique chain topologies could provide strong signals for organizing the temporal ordering of substrate degradation at mitotic exit [42–44].

APC/C in G1 control

Since the APC/C literally “promotes anaphase,” it is essential for the exit from mitosis. However, the APC/C also plays an important and less well-understood role in G1. Similar to its role in mitosis, the function of APC/C during G1 is evolutionarily conserved and vital to homeostatic cell cycle dynamics.

Prior to the biochemical identification of APC/C, studies from Amon and Nasmyth demonstrated that proteolysis of the budding yeast B-type Cyclin Clb2 begins in mitosis but then continues into the ensuing G1. Therefore, the enzyme which catalyzes Cyclin destruction is active not just during mitosis but also in G1-phase [45]. Cdh1 was subsequently identified as the APC/C substrate receptor that controls Clb2 degradation in G1 [46–48]. Early experiments, foreshadowing the importance of APC/C^Cdh1 in G1 maintenance, showed that the loss of Cdh1 could not be tolerated in budding yeast that had also lost the CDK inhibitor Sic1, which itself restrains S-phase entry [46,47]. This genetic relationship suggested that Cdh1 works in collaboration with other G1 restriction factors to restrain G1/S. Further studies in budding yeast highlighted the continued activity of APC/C^Cdh1 up to the point of S-phase entry [49]. In the fission yeast S.pombe, Cdh1 promotes Cyclin B degradation in G1 phase [50]. Similarly, Cdh1 prevents the unscheduled accumulation of Cyclin B during G1 in Drosophila [51]. Likewise, Cyclin B is unstable when introduced into cell extracts produced from both quiescent and G1 phase cells but remains stable in extracts produced from cells in S-phase [52]. Thus, the activity of APC/C, which is initiated in mitosis to promote anaphase, continues throughout G1, and this functionality represents an evolutionarily conserved feature of the cell cycle.

These studies in diverse experimental systems pointed to a potential role for APC/C in G1-phase. The G1/S border represents a major barrier to proliferation in eukaryotes. During this time in the cell cycle, cells integrate diverse extracellular and intracellular signals to decide whether to enter the cell cycle. In support of a role for APC/C^Cdh1 in G0/G1 maintenance and restraining DNA replication, Cdh1 depletion in budding yeast renders cells unable to arrest in G1-phase in response to the hormone alpha-factor, which induces a cell cycle arrest in wild-type cells [47]. This important result pointed to a role for APC/C^Cdh1 in preventing unscheduled proliferation. Consistently, Cdh1 loss in Drosophila leads to an extra cell cycle in the epidermis during development whereas its overexpression blocked cell division [51]. Similarly, in chicken DT40 cells the loss of Cdh1 promoted the premature accumulation of Cyclins A and B in late G1-phase and rendered cells unable to arrest in G1 in response to pharmacological targeting of the mTOR pathway [53]. Further confirmation that the premature accumulation of APC/C substrates led to an accelerated rate of G1 progression came from several groups using mouse embryonic fibroblasts that were deleted for Cdh1 [54,55]. These results have been further validated in diverse human cell systems using Cdh1 overexpression [56] and depletion [54,57–60]. which altogether showed a critical role for Cdh1 in restraining progression through G1 and S-phase entry. Finally, evidence that the role of Cdh1 in G1/S is directly related to the function of APC/C came from conditional knockout studies in mouse livers. The loss of APC/C activity, triggered by APC2 inactivation, caused cell cycle entry even in the absence of proliferative signals [61]. Taken together, these studies demonstrated that the APC/C is active in G0 and G1 phase cells, where it acts as a critical restriction factor to prevent unscheduled proliferation. This places APC/C^Cdh1 among a small group of key regulators known to restrain the entry of cells into S-phase, including the retinoblastoma tumor suppressor (RB) and CDK inhibitory proteins [62,63]. In the next section, we describe emerging data highlighting the overlapping role of APC/C and RB in controlling the G1/S transition.

APC/C and RB coordinate G1/S progression

The G1/S boundary represents a major barrier to proliferation and oncogenesis. The G1/S border is controlled in part by CDK4/6 which binds to any of the three D-type Cyclins and phosphorylates the tumor suppressor protein RB [64]. The hyper-phosphorylation of RB by Cyclin D-CDK4/6 and also by Cyclin E-CDK2 triggers its dissociation from the E2F transcription factor, initiating a feedback loop that promotes S-phase entry (analogous proteins exist in yeast and their regulation of cell cycle entry is similar [65]). Due to its role in restraining G1/S, RB is a potent tumor suppressor in human cancers. RB is lost through mutation in a subset of malignancies, most notably in retinoblastoma and small cell lung cancer [66,67]. In addition, RB is functionally inactivated across diverse human cancers as evidenced by the aberrant expression of the E2F transcriptional program and resistance to CDK4/6 inhibition in specific, aggressive subtypes of cancers [68,69].

Several studies have demonstrated a collaboration between APC/C^Cdh1 and RB in restraining cell cycle entry. The first evidence came from a genetic screen in C.elegans, which identified Cdh1 as a synthetic genetic interactor with the sole worm RB orthologue, Lin-35 [70]. Importantly, mutations in Cdh1 and Lin-35 led to profound hyperproliferative defects throughout the worm [70]. Similarly, genetic studies in flies showed that forced E2F expression was insufficient to drive cell cycle entry, whereas concomitant activation of E2F together with loss of APC/C activity triggers proliferation [71]. In human cells, there is evidence that their coordination could be direct: APC/C was identified in a proteomic screen for physical RB interactors. It was shown that Cdh1 is required for RB induced cell cycle arrest by triggering the degradation of the ubiquitin ligase that targets the CDK inhibitory (CKI) protein p27 [72]. The molecular mechanisms of p27 regulation and activity are discussed later. Furthermore, Cdh1 is required for cell cycle arrest induced by CDK4/6 inhibition in both worms and human cells [73]. Taken together, these studies point to an important role for APC/C function in restraining cell cycle entry in collaboration with RB. Consistent with the important tumor suppressor role for RB, Cdh1 haploinsufficiency produces epithelial tumors in mice, suggesting a potential role in tumor suppression [55]. However, Cdh1 is not recurrently mutated or transcriptionally silenced in human cancers. While this could be due to an essential function, consistent with bi-allelic loss being lethal in mice [55,74], it raises the question as to whether Cdh1 is truly a tumor suppressor in human cancers.

Given its vital role in regulating the G1/S transition, it is essential to inactivate APC/C^Cdh1 to execute timely S-phase initiation. Several mechanisms of APC/C inactivation exist. The relationship and role of these pathways in potentially restraining APC/C in both normal and cancer cells is discussed below.

Mechanisms modulating APC/C inactivation

During progression through the cell cycle, APC/C is controlled by several overlapping and often interconnected mechanisms. The most well-studied mechanism is the control of APC/C activity by the mitotic spindle checkpoint, which keeps APC/C activity in check until metaphase, when all chromosome achieve bipolar microtubule attachments [17]. However, numerous other pathways control APC/C activity in late G1, S, and G2 phases. Interestingly, the mechanisms governing APC/C, and the CDKs that it controls, are largely analogous. For example, as is detailed below, both APC/C and CDKs are controlled by coactivators that are essential for their activity, and whose expression and degradation are tightly controlled throughout the cell cycle. Figure 1 summarizes some of the mechanisms described below.

Figure 1.

Schematic overview of the pathways that restrain APC/C activity post-translationally. The relative timing of their activation is generally shown.

The APC/C is most highly active from mid-mitosis and then throughout G1 phase. A schematic representation of the pathways that control APC/C and the degradation of its substrates during cell cycle progression is shown, with emphasis on those that operate outside of mitosis. Prior to metaphase, the activity of APC/C is prevented by the spindle assembly checkpoint, or SAC. The Cdc20-bound form is activated at metaphase (APC/C^Cdc20). In late mitosis, the Cdh1-bound form of APC/C becomes active (APC/C^Cdh1). The APC/C uses two E2 ubiquitin conjugating enzymes, UBE2C and UBE2S, which form K11-linked chains on substrates. The degradation of APC/C substrates is antagonized by the deubiquitinase Cezanne, which reverses K11-linked ubiquitin chains formed on APC/C substrates. The APC/C is then inactivated at G1/S through myriad mechanisms. The SCF^{Cyclin F} E3 ubiquitin ligase triggers Cdh1 degradation. The APC/C inhibitory protein EMI1 accumulates and binds APC/C. CDK2 phosphorylates Cdh1, preventing its binding to APC/C and promoting its cytoplasmic localization. The APC/C E2s are also degraded. A second ubiquitin ligase, SCF^βTRCP controls the destruction of Cdh1 and Emi1 later, and the degradation of both via this mechanisms is dependent on PLK1, which becomes active just prior to mitotic entry.

Gene expression:

The expression of Cdc20 and Cdh1 is controlled at the mRNA level during the cell cycle. In systematic studies examining cell cycle transcriptional dynamics, both transcripts emerged in multiple studies displaying oscillatory expression. In addition, the APC/C utilizes two E2 ubiquitin conjugating enzymes, UBCH10/UBE2C and UBE2S, which are also controlled at the mRNA level throughout the cell cycle. In fact, in the five most well-validated studies examining cell cycle transcriptional dynamics, UBE2S and UBCH10/UBE2C are among a small group of approximately 100 genes that show cell cycle regulation in all studies [75–79]. Likewise, both Cdh1 and Cdc20 expression oscillated in four out of five of these studies. Much like the role of Cyclins expression in regulating CDK activity, the cyclical expression patterns of coactivator and E2 mRNAs suggest an important role for transcription in organizing the activity of APC/C during cell cycle.

Cdh1 phosphorylation:

Cdh1 is a phosphoprotein with more than ten apparent CDK consensus motifs as well as other less well-studied phosphorylation sites [80]. Phosphorylation of Cdh1 on these CDK sites plays an important role in preventing APC/C activation. In budding yeast, elegant gene replacement strategies have demonstrated the essentiality of Cdh1 phosphorylation at CDK consensus sites [81]. Mechanistically, phosphorylation of Cdh1 regulates its function in at least two ways. First, Cdh1 phosphorylation regulates its binding to APC/C. A non-phosphorylatable version of Cdh1, harboring alanine mutations at 11 CDK phospho-consensus motifs, is constitutively bound to the APC/C and triggers the destruction of Clb2 [48]. These sites can be dephosphorylated by the cell cycle regulated phosphatase Cdc14 [82]. Consistent with the regulation of Cdh1 by CDK phosphorylation, the inactivation of APC/C requires the accumulation of S-phase Cyclins, which are not under APC/C control in yeast. Accordingly, overexpression of Sic1, which inhibits both CDK activity and Cdh1 phosphorylation, reactivates the APC/C by promoting the association of Cdh1 with the APC/C complex [49]. Cdh1 is also heavily phosphorylated during cell cycle progression in humans on its myriad of CDK consensus motifs and at other sites as well [83]. Consistent with the aforementioned observations in yeast, a non-phosphorylatable mutant version of human Cdh1 constitutively binds and activates APC/C, impairing S-phase entry [56,84,85].

In addition to controlling binding to the APC/C, Cdh1 phosphorylation also regulates its localization. It is important to note that the core APC/C complex is largely thought to localize to the nucleus. In yeast, Cdh1 localization is cell cycle regulated and depends on CDK, with phosphorylation enhancing its nuclear export and contributing to APC/C inactivation [86]. More recent studies showed that selected phosphorylation sites can control APC/C binding, whereas others are responsible for regulating Cdh1 localization [87]. Likewise, in humans, a version of Cdh1 harboring alanine substitutions at CDK consensus motifs was exclusively nuclear, whereas a phospho-mimetic mutant version was cytoplasmic [88].

In human cells, the role of different Cyclin-CDK complexes in phosphorylating Cdh1 continues to evolve, raising the important question as to the identity of the kinase(s) responsible for Cdh1 phosphorylation. Originally, Lukas et al. showed that Cdh1 co-precipitated with Cyclin A, but not Cyclin E, and that Cyclin A-Cdk2 could phosphorylate Cdh1 in an in vitro kinase assay [85], consistent with earlier findings in cell extracts [52]. This evidence is corroborated by a recent study showing that Cdh1 is an excellent Cyclin A-CDK2 substrate [89]. Perhaps most significantly, depletion of Cyclin A, but not Cyclin E, strongly increased the association of Cdh1 with the APC/C core complex [85]. Since the substrate selectivity of Cyclin-CDK complexes most often relies on the Cyclin, these data indicate that in humans Cyclin A-Cdk2 or Cyclin A-Cdk1 phosphorylate Cdh1 to prevent its association with the APC/C complex. This suggests a positive feedforward loop wherein Cyclin A promotes its own stability by preventing activation of the ligase that catalyzes its degradation. Recent biochemical data showed that Cyclin B could also phosphorylate Cdh1 in vitro, adding an additional layer of regulation that likely accounts for the inability of Cdh1 to bind APC/C until late mitosis, only after Cyclin A and Cyclin B have been largely destroyed [90].

In contrast to these observations, more recent studies implicated additional Cyclin-CDK complexes in inhibiting APC/C. First, it was shown that the worm ortholog of Cdh1 could be phosphorylated by both human and worm Cyclin D-CDK4 complexes on several CDK consensus sites in the Cdh1 amino-terminus [73]. These data were corroborated by a second study, which also showed that human Cdh1 was a substrate of Cyclin D-CDK4/6 [91]. In a third study, using single cell live imaging of APC/C biosensors, the authors showed that depletion of Cyclin E, but not Cyclin A, altered APC/C activation, although it is not known if this is due to direct phosphorylation of APC/C or Cdh1 by Cyclin E/CDK2 [60]. At this time, it remains unknown whether Cdh1 is a promiscuous substrate that can be phosphorylated by these myriad kinases or if there is greater selectivity in vivo than what is currently appreciated. Resolving these differences and defining precisely when Cdh1 becomes phosphorylated, by which kinase, on which sites, and the importance of these modifications represents important future questions. Notably, in yeast, different phosphorylation sites on Cdh1 differentially contribute to changes in APC/C binding and localization [87].

Finally, it was recently shown that Cdh1 could be phosphorylated by a non-Cyclin-CDK entity, ERK, the downstream kinase in the MAPK cascade, and that phosphorylation also inhibits APC/C^Cdh1 activity [91]. It is notable that CDK and ERK are both proline directed kinases that phosphorylate serine and threonine residues that have a proline in the +1 position. Further, while most cell cycle focused research groups studying APC/C would envision these phosphorylation sites as being controlled by CDK, there is ample space to imagine the coordinate control by the MAPK pathway, which also plays an important role in controlling proliferation. Recent studies in yeast have highlighted cooperation between CDK and MAPK pathways in cell cycle associated signaling [92].

Emi1 binding:

Early mitotic inhibitor 1 (Emi1) is a member of the F-box family, a set of substrate receptors proteins for the Skp1-Cul1-F-box (SCF) family of E3 ubiquitin ligases. In the SCF, F-box proteins, of which there are 69 in humans, designate substrates for degradation. [93]. However, while Emi1 can assemble into an SCF complex, it is unknown if it functions as a substrate receptor. Instead, Emi1 was identified and is known as an important inhibitor of APC/C [94,95]. In humans, Emi1 prevents the APC/C dependent degradation of Cyclin A and Cyclin B [96,97]. Furthermore, depletion of Emi1 causes cell cycle arrest in late S-phase and G2 and re-replication that is due to the reactivation of APC/C [96,97]. The Emi1 ortholog in Drosophila, Regulator of Cyclin A (RCA1), is so-named because it is also required to prevent Cyclin degradation in G2 phase [98]. Moreover, rca1 mutant fly imaginal cells have enlarged nuclei, consistent with re-replication due to endocycling [98]. Emi1 is also reported to contribute to APC/C inactivation at the G1/S border. Emi1 overexpression can drive cell cycle entry and proliferation in cells overexpressing Cdh1 or RB [95]. In addition, the rate of APC/C inactivation is slower in Emi1 depleted cells [60]. However, few studies have reported a strong G1 arrest in Emi1 depleted cells. Thus, it is unknown to what extent Emi1 contributes to APC/C shut-off versus maintaining APC/C in an inactive state.

At the onset of mitosis, Emi1 becomes a substrate of another SCF ligase using the substrate receptor protein βTRCP1, and its destruction in early mitosis is dependent on the phosphorylation of Emi1 by PLK1 [99–102]. The degradation of Emi1 in early mitosis is vital to the subsequent activation of APC/C at the metaphase to anaphase transition.

Cdh1 and E2 degradation:

Cdh1 protein levels oscillate during the cell cycle in both yeast and humans, lending credence to the notion that dynamics in Cdh1 abundance contribute to APC/C activity [48,84]. This phenomenon is analogous to the mechanism by which Cyclins control CDK. In human cells, Cdh1 levels are high in G1, decrease significantly in S-phase, and then reappear in G2/M [59,84,89,103,104]. Multiple E3 ligases have been implicated in Cdh1 destruction. First, APC/C was suggested to control its own degradation through auto-catalytic degradation [103]. Later, the SCF family of E3 ubiquitin ligases were implicated in Cdh1 degradation [105]. Two independent studies have recently implicated two different F-box substrate receptors in Cdh1 destruction.

First, the Wei laboratory showed that Cdh1 degradation could be triggered by SCF, in complex with its F-box protein, βTRCP [89]. While βTRCP often binds to the degron sequence DSGxx(x)S, it recognizes a non-canonical D-box in Cdh1 (DDGxxxS). This recognition is controlled by sequential phosphorylation of the βTRCP degron in Cdh1, first by Cyclin A-CDK2 and then by PLK1. More recently, our laboratory showed that Cdh1 is also a substrate of Cyclin F, the founding member of the SCF family of substrate receptors and a non-canonical Cyclin that neither binds nor activates a CDK [59,106,107]. Notably, Cyclin F is among the most highly cell cycle regulated of all F-box proteins and is the only F-box containing protein whose mRNA emerged in all cell cycle transcriptional studies that have been performed to date [108]. In a remarkable twist of fate, Cyclin F is itself a substrate of APC/C-mediated degradation in late mitosis and early G1-phase, suggesting a tightly coordinated, reciprocal relationship between Cyclin F and Cdh1 in regulating S-phase entry. In addition, this latter study noted that the previously identified degron in Cdh1, which was thought to mediate its degradation by APC/C [103], overlaps with the Cyclin F binding sequence [59]. Thus, the extent to which APC/C and SCF^{Cyclin F} coordinate the destruction of Cdh1 at the G1/S border remains an open question and an important area of investigation. Furthermore, since PLK1 is only active in the hours preceding mitotic entry, it remains unknown how these convergent, SCF-related, proteolytic pathways coordinately control the abundance of Cdh1 throughout the cell cycle.

Finally, the two APC/C E2s, UBE2C and UBE2S, are both substrates of the APC/C [37,109]. The auto-catalytic degradation of its E2s suggests that the APC/C potentially functions as an “autonomous oscillator” that inactivates itself by destroying its own E2s after all of its substrates have been consumed [109]. However, this interpretation is complicated by a conflicting study that analyzed UBE2C expression and role in cell cycle, and showed that it accumulates in late G1/early S-phase before Cyclin A [110]. Determining the mechanisms underlying degradation of the APC/C E2s and how their degradation contributes to S-phase entry will be important to fully understand the nature of APC/C inhibition.

Deubiquitinating enzymes:

Ubiquitin is conjugated to substrates through an isopeptide bond between its carboxyl-terminal amino acid (Glycine) and the epsilon amino group in substrate lysine residues. Similarly, the formation of polyubiquitin chains occurs by the addition of ubiquitin onto other ubiquitin molecules, either through any of the seven lysines in ubiquitin or through its amino terminus. These variations lead to different ubiquitin linkages with unique physical topologies that promote diverse cellular outcomes. This is commonly referred to as the ubiquitin code [111]. In the same way that phosphatases regulate kinase-signaling cascades by controlling the dephosphorylation of substrates, so-called deubiquitinating enzymes (DUBs) regulate ubiquitin signaling cascades to control the ubiquitination and degradation of substrates. There are approximately 100 DUBs in humans, which are implicated in various aspects of cellular physiology and disease [112]. Because of its tremendous importance in cell physiology and disease, some studies suggest that the APC/C could be controlled by deubiquitination and several DUBs have been linked to APC/C and the degradation of its substrates.

In budding yeast, UBP15 deubiquitinates the Cyclin Clb5 and promoting S-phase entry [113]. In humans, USP37 was identified as an APC/C^Cdh1-associated deubiquitinating enzyme ([114], corroborated by our unpublished data) that promotes cell cycle progression via Cyclin A deubiquitination, and USP37 is also an APC/C substrate [114]. Thus, USP37 and Cyclin F both negatively regulate APC/C and are also APC/C substrates, highlighting complex feedback mechanisms involved in APC/C control [114]. In early mitosis, APC/C activation is prevented by the SAC, which sequesters Cdc20 away from APC/C in the mitotic checkpoint complex (MCC) [17]. The inactivation of the SAC, and thus the full activation of APC/C, is driven by auto-ubiquitination of Cdc20 [115]. The DUB USP44 emerged from an RNAi-based screen for regulators of the SAC [116]. USP44 reversed Cdc20 ubiquitination, suggesting that USP44 is a SAC component. However, USP44 is unique among spindle checkpoint proteins in that it is non-essential, since USP44 knockout mice are viable and have a seemingly intact SAC [117].

Finally, we recently identified OTUD7B/Cezanne as a key DUB that antagonizes APC/C substrate ubiquitination. The specificity of Cezanne for specific ubiquitin chains has remained controversial, with the Komander group reporting specificity of Cezanne for K11-linked ubiquitin chains and others suggesting a role in K63 and K48-linked deubiquitination [118–121]. Our results support an extraordinary specificity of Cezanne for K11-linkages and demonstrate that it is itself cell cycle regulated, and importantly, antagonizes APC/C substrate ubiquitination in mitosis [122]. Cezanne depletion induces defects in chromosome segregation and the formation of micronuclei, indicative of a role in cell division [122]. In the future, it will be important to understand how Cezanne is regulated during cell cycle progression, how it binds substrates, and the mechanism by which it restrains APC/C-mediated substrate ubiquitination.

Together, these studies support the notion that the ubiquitinating activity of the APC/C is regulated and counteracted by DUBs. Since the APC/C is widely considered the master E3 ligase involved in cell division, it is highly possible that additional DUBs and new layers of regulation for the ones already described are yet to be discovered.

Convergent CDK-APC/C networks and commitment to S-phase

In response to suboptimal proliferative conditions, cells arrest at a so-called “restriction point” prior to the start of S-phase. These observations in human cultured cells date back several decades [123–125]. The restriction point is controlled, in part, by the CDK-RB-E2F pathway. Despite the importance of CDK-RB in G1/S control, cells continue to proliferate in vivo and in vitro following pathway ablation (e.g. loss of CDK4 and CDK6) and, remarkably, have a functional G1/S checkpoint [126,127]. Therefore, additional pathways coordinate G1/S control and the commitment to cell cycle progression at the G1/S boundary.

A recent study from Cappell et al. shed light on the role of APC/C in the commitment to S-phase. They demonstrated that APC/C^Cdh1 inactivation at the G1/S boundary is rapid, exhibits characteristics of a bistable switch, and this switch to an “APC/C off” state is triggered by Cyclin E/CDK2 [60]. Once inhibited, APC/C is irreversibly maintained in an inactive state by the inhibitory protein Emi1 [60]. Together, these data suggest that Cyclin E initiates the inactivation of APC/C that is then locked down by Emi1 Using single cell fluorescent reporters combined with immunostaining, they also showed that RB phosphorylation precedes APC/C inactivation by several hours, and APC/C inactivation, not RB phosphorylation, temporally coincided with S-phase entry. The phosphorylation of RB has long been considered the point of no return in G1, after which cells are committed to starting S-phase. Significantly, these recent studies showed that cells could still arrest in G1 following mitogen withdrawal or stress after RB had become phosphorylated, so long as APC/C had not yet been inactivated. These data led to the proposition that APC/C inactivation at G1/S represents a “commitment point” for cell cycle entry [60].

The above study places Cyclin E activation upstream of APC/C. In contrast, single cell analysis combined with molecular modeling had previously suggested that the key role of APC/C^Cdh1 in controlling S-phase entry was due to its role upstream of CDK2 activation [58]. The activation of CDK2 is restrained by the CKI protein p27, whose degradation is controlled by an SCF ubiquitin ligase in combination with the adaptor protein SKP2. Importantly, SKP2 is a substrate for the APC/C [128,129]. Therefore, APC/C inactivation causes an increase in the abundance of SKP2. This allows SKP2 to assemble into SCF^SKP2 ligase complexes and trigger p27 degradation. The degradation of CKI allows for full Cyclin E/CDK2 activation and entry into S-phase. Thus, in cells lacking Cdh1, the amount of Cyclin E which is needed to drive S-phase entry is lowered, explaining why cells without Cdh1 progress rapidly into S-phase [58].

Notwithstanding the differences in mechanistic explanations between these studies, together, they highlight the central role of the APC/C in organizing regulatory networks involved in G1/S control. Modeling these systems and networks, incorporating additional regulators discussed above, and testing hypothesis derived from this modeling is an important area of future study. Doing so will further elucidate the intricate role of APC/C in cell cycle control and define the molecular features controlling the G1/S boundary, a major barrier to transformation which is perturbed almost universally in cancer.

Relevance of APC/C inactivation in cancer

APC/C^Cdh1 is among a small group of key regulators implicated in restraining the start of DNA replication. The other key members of this group, RB and CKIs, are definitively linked to cancer. As discussed above, RB is lost or functionally inactivated in many malignancies. In addition, the CKI p16^INK4A, which inhibits CDK4/6, is silenced and mutated in cancer, and the CKI p21 is a downstream transcriptional target of the tumor suppressor p53. However, it is largely unknown if Cdh1 is similarly inactivated in cancers.

The role of Cdh1 in restraining S-phase is conserved throughout evolution and is evident in budding and fission yeast, flies, worms, chickens, mice, and humans. Moreover, single allelic loss of Cdh1 leads to the formation of epithelial tumors [55], pointing to a role in tumor suppression. APC/C^Cdh1 assembly has been reported to be influenced by MAPK signaling, and Cdh1 loss cooperates with additional oncogenic lesions to promote transformation in melanocytes [91]. However, Cdh1 is not recurrently mutated, deleted by copy-number loss, or transcriptionally silenced in human malignancies, suggesting that cancer cells might use alternative mechanisms to block APC/C^Cdh1 function. Nevertheless, reductions in Cdh1 abundance have been noted in aggressive breast and colorectal cancers [130,131]. Since Cdh1 levels are reduced in S-phase, these changes could be the indirect consequence of the start of S-phase. Alternatively, transient reductions in Cdh1 levels could weaken the barrier to cell cycle entry, promoting S-phase entry, similar to phosphorylation of RB, which drives and coincides with the start of S-phase. Unlike RB, it is unknown if Cdh1 is repressed in cancer to lower the barrier to cell cycle entry. Moreover, and in contrast to RB, Cdh1 is essential for animal development and survival, with knockout mice embryos dying around day E9.5 [132].

The results above indicate that repression of Cdh1 dosage or activity could weaken the G1/S boundary and promote tumorigenesis. Importantly, changes in gene dosage play an important role in human cancers, highlighted by the fact that copy-number alterations represent the most pervasive recurrent changes across all human cancers. Moreover, in any given cancer, copy-number changes are recurrent and not stochastic. For example, the landscape of copy-number changes observed in specific subtypes of breast cancer are distinct (e.g. amplification of 1q and 8q), not random, and point to a functional importance for these regions in disease [68,133,134]. The enrichment of proliferative drivers in copy-number amplified regions (e.g. Myc on chromosome 8q), and loss of genes that impair growth in deletions, further supports this notion [135].

How then might APC/C^Cdh1 be inactivated in human cancers? Studies to date highlight three potential mechanisms. The first comes from the analysis of the human papillomavirus (HPV) E7 oncoprotein. The E7 oncoprotein is well known as a suppressor of RB. A recent study showed that the presence of E7 leads to abnormally high-level expression of Emi1, the APC/C inhibitory factor [136]. Elevated Emi1 in turn restrains the degradation of APC/C substrates and leads to mitotic abnormalities.

The next two studies come from our own laboratory. We previously showed that Cyclin F is targeted for degradation by APC/C in late mitosis and early G1, and subsequently, Cyclin F targets Cdh1 for degradation at the end of G1 and in S-phase. Moreover, we demonstrated that Cyclin F is phosphorylated and activated by the oncogenic kinase AKT/PKB [137]. AKT is a key downstream effector in the phosphoinositide 3-kinase (PI3K) signaling cascade [138]. Importantly, we showed that interfering with AKT–mediated phosphorylation of Cyclin F altered the stability of both Cyclin F and Cdh1. Significantly, activating mutations in PI3K represent the most recurrent, activating mutations in all cancers [139]. Thus, activation of PI3K-AKT could trigger the degradation of Cdh1 in human cancers, thereby lowering the barrier to cell cycle entry. These data also imply that APC/C is an information switchboard that integrates extracellular signaling through the PI3K-AKT cascade with the core of the cell cycle machinery. This is similar to that of CDK4/6, which integrates information on mitogen availability via the transcriptional induction of Cyclin D through the Ras-MAPK pathway [140].

The final mechanism relates to the aforementioned DUB Cezanne. As discussed above, changes in gene expression and gene copy number are hallmarks found in virtually all cancers. In many cases, including breast and ovarian cancers, amplifications on chromosome 1q represent a hallmark of disease. However, little is known about how 1q amplification drives oncogenesis. Interestingly, Cezanne is located at 1q22, the center of a genomic amplicon in breast cancer [141]. In fact, Cezanne is one of three most amplified and over-expressed DUBs in all of breast cancers. Cezanne mRNA and gene copy number are increased in a remarkable 32% of breast cancers. By comparison, the Myc-oncogene is amplified or over-expressed in approximately 21% of breast tumors. Cezanne over-expression could have myriad effects on the cell cycle. By counter-balancing the APC/C [122], Cezanne could alter kinetics of mitosis that impact chromosome stability, as well as the G1/S commitment point, promoting aberrant cell cycle progression, a hallmark of cancer. Thus, Cezanne amplification/overexpression could contribute to tumorigenesis through various and yet to be characterized pathways, on top of its newly described role in restraining APC/C-mediated ubiquitination.

Future questions

There still remain many important questions to understand the role and regulation of APC/C in cell cycle entry at the G1/S boundary and how it coordinates with CDK-RB to facilitate proliferation. First, as discussed above, myriad mechanisms can contribute to APC/C inactivation, including Cdh1 phosphorylation and ubiquitination, E2 degradation, and Emi1 binding. However, it is unknown how these pathways interconnect and the relative utilization of each pathway in a specific cell line or system is unknown. It is also unknown if specific pathways are deployed to functionally inactivate distinct sub-populations of APC/C that in turn allow for the stabilization of specific substrates but not others. This is a particularly interesting idea given the localization of APC/C to distinct structures in the cell [33,142–144]. In considering the role and potential importance of the diverse mechanisms which function to control APC/C, it is useful to consider the paradigm of Cyclin-CDK complexes, which are perceived as playing essential roles as core components of the cell cycle oscillator yet are individually dispensable for cell and animal growth and viability. Similarly, we imagine that each of the pathways described above cooperatively function to regulate APC/C but are likely individually dispensable, or play context dependent roles, and could be compensated for by other, redundant mechanisms.

Another open question relates to how APC/C substrates accumulate as cells enter S-phase. When looking across various studies, it becomes clear that not all substrates accumulate with similar kinetics as cells enter S-phase. This variability is likely controlled by substrate transcription, translation, and degradation by other E3s. In addition, emerging evidence has pointed to substrate level regulation of APC/C-mediated degradation, the paradigm being that of Cdc6, which becomes phosphorylated and protected from APC/C [145]. The question of substrate accumulation kinetics is the mirror of the differential degradation kinetics observed at mitotic exit, where some proteins are degraded early and others much later. Although this notion is well documented in the case of mitotic substrates, where different binding affinities lead to different processivity rates, the mechanisms and enzymes underlying this regulation and the consequences on S-phase entry and progression remain entirely unstudied.

Finally, like RB, APC/C represents a strong G1/S restriction factor and mounting evidence points towards its role in tumor suppression. In the same way that CDK4/6 inhibition “re-awakens” RB for therapeutic benefit, it will be interesting in the future to determine if APC/C can be similarly “re-awakened”. This could potentially be achieved by inactivating PI3K-AKT, since we have shown that this pathway activates Cyclin F to trigger Cdh1 degradation. Alternatively, one could imagine inhibiting the ligases that trigger Cdh1 degradation, despite the fact that such pharmacological tools are not yet available. Nevertheless, future studies are needed to understand how APC/C function changes and contributes to the process of tumorigenesis, and if as we predict, APC/C activity is reduced in cancer to lower the barrier to cell cycle entry. Addressing these questions will provide vital insight regarding mechanisms that guard against the inappropriate or untimely inactivation of APC/C, a key guardian of cell cycle, proliferation and genome integrity, and the consequences of this regulation in pathological settings.

Highlights.

• The APC/C ubiquitin ligase is active in G1 phase, and guards against premature entry into S-phase.

• Due to its vital role in controlling myriad substrates, multiple, coordinated mechanisms guard against inappropriate APC/C activation

• A combination of kinases, deubiquitinating enzymes, transcription factors and other E3 ligase restrain APC/C activity during S, G2 and early mitosis

• Alterations in the mechanisms that control APC/C are likely to be important in cancers where the barrier to proliferation has been weakened.