Cell cycle checkpoint regulators reach a zillion

Abstract

Entry into mitosis is regulated by a checkpoint at the boundary between the G₂ and M phases of the cell cycle (G₂/M). In many organisms, this checkpoint surveys DNA damage and cell size and is controlled by both the activation of mitotic cyclin-dependent kinases (Cdks) and the inhibition of an opposing phosphatase, protein phosphatase 2A (PP2A). Misregulation of mitotic entry can often lead to oncogenesis or cell death. Recent research has focused on discovering the signaling pathways that feed into the core checkpoint control mechanisms dependent on Cdk and PP2A. Herein, we review the conserved mechanisms of the G₂/M transition, including recently discovered upstream signaling pathways that link cell growth and DNA replication to cell cycle progression. Critical consideration of the human, frog and yeast models of mitotic entry frame unresolved and emerging questions in this field, providing a prediction of signaling molecules and pathways yet to be discovered.

Introduction

Cell division in eukaryotes is a highly conserved process that punctuates the life of an actively dividing cell. In the 1950s, Howard and Pelc, studying the periods of cell division and non-division in Vicia faba (fava bean), established the concept of a cell cycle divided into four sequential phases of unequal length: G₁ (gap phase 1), S (DNA synthesis), G₂ (gap phase 2) and M (mitosis).¹ Later studies with other cell types defined a quiescent phase, called G₀, outside the four-step life cycle of actively dividing cells.^2-5 Except under very specific experimental conditions,^6,7 the cell is driven “forward” from one phase of the cell cycle to the next (G₁ to S to G₂ to M),⁸ with a circuit from G₁ to G₁ constituting one cell cycle. The studies by Howard and Pelc, as well as subsequent studies in the field, recognized G₁ as the primary cellular growth phase, S as the phase in which genome duplication and the initial steps of mitotic spindle formation occur, G₂ as an additional growth phase and M as the phase in which both mitosis and cytokinesis occur. Correct timing of the transitions between cell cycle phases is critical for proper cell division. For example, if the genome does not fully replicate or is physically broken prior to chromosomal segregation, the resulting cells would not contain equivalent copies of the genome. To prevent precocious progression of the cell cycle and its ensuing detrimental outcomes, such as aberrant cell proliferation or death, checkpoints operate throughout the cell cycle, most often at the border between cell cycle phases.^9,10 A checkpoint is a point in the cell cycle at which cell cycle progression arrests until the previous stage of growth or division has been completed with fidelity, or until certain requirements for cell division are met. For example, in both yeast and mammalian cells, before progression into S phase, there is a nutrient-sensing cell growth checkpoint (for a review, see ref. 11). The checkpoint that is the focus of this review lies at the G₂/M border and regulates entry into M phase. Other checkpoints include those that monitor spindle position, chromosomal separation and mitotic exit.^12,13 The actual pre-conditions (e.g., cell size, nutrient availability, DNA integrity) that allow a cell to move through a checkpoint and into the next phase of the cell cycle without arrest (i.e., without activating the checkpoint) varies from checkpoint to checkpoint, from organism to organism and from somatic to embryonic cells. However, the core molecular mechanisms of checkpoint control remain highly conserved.

From a molecular viewpoint, oscillation of the activity of cyclin-dependent kinases (Cdks), when they are in complex with adaptor molecules known as cyclins, is the minimal engine of the cell cycle that temporally orders the phases of the cell cycle.⁷ Without Cdk activity, the cell cycle does not progress.¹⁴ In addition to kinase activation, cyclins confer substrate specificity (for a review see ref. 15). The activity of these Cdk/cyclin complexes, which phosphorylate serine/threonine residues, is regulated by inhibitory proteins and by post-translational modifications (e.g., phosphorylation).¹⁵ Prior to the biochemical purification and cloning of Cdks and their associated cyclins in the late 1980s and early 1990s, several types of Cdk-cyclin complexes were first identified physiologically as factors required for entry into specific phases of the cell cycle (e.g., S phase-promoting complex and M phase-promoting complex, also known as maturation-promoting factor or MPF).^8,16,17 Because specific classes of cyclins are expressed only during certain phases of the cell cycle, specific Cdk-cyclin complexes form in each phase of the cell cycle and prepare the cell for the next cell cycle phase through the phosphorylation of specific substrates.¹⁸ Thus, the cyclical expression of individual cyclins, in conjunction with the activation, degradation or inhibition of Cdk/cyclin complex regulators, creates a self-organized, hysteretic, temporal pattern.^19-21 Under certain experimental conditions, these mechanisms of regulation are dispensable for cell cycle progression.⁷ Quality control within the cell cycle is enforced by the aforementioned checkpoints.

One of the major checkpoints in cell division lies at the G₂/M boundary and controls entry into mitosis. At its core, this cell cycle checkpoint is a phospho-regulated switch. Phosphorylation of specific Cdk residues within M phase-specific Cdk/cyclin complexes can either inhibit or promote the activity of these complexes toward their substrates, thereby preventing or activating entry into mitosis.¹⁸ As an additional block to mitosis, mitotic Cdk substrates are kept dephosphorylated by the activity of protein phosphatase 2A (PP2A), which contributes to ordering cell cycle phase transitions.^22-25 Thus, a balance between the active levels of a mitosis-activating kinase and a mitosis-inhibiting phosphatase is thought to ensure that mitotic events do not occur precociously. Fine-tuning the balance between kinase and phosphatase activities is important for maintaining the fidelity of cell division. Cells that fail to accurately balance these opposing effects face catastrophe.^26,27 In certain cancer cells, the G₂/M checkpoint is especially important.²⁸ When cancer cells have a defective G₁ checkpoint, they rely strongly on the G₂/M checkpoint to prevent cell death during cellular replication.²⁹ As such, drugs that force cancer cells to ignore the G₂/M checkpoint are an exciting new therapeutic treatment for selectively inducing cell death in cancer cells.

G₂/M Checkpoint in S. pombe and Vertebrates

In the fission yeast Schizosaccharomyces pombe, cell size is coordinated with cell cycle progression (as opposed to Xenopus oocytes, which do not have a cell size checkpoint).³⁰ Regulation occurs principally at two control points, one in G₁ and another in G₂ just prior to mitosis, with deviations in cell size at these points indicative of misregulated cell cycle progression. A striking spectrum of cell size phenotypes, coupled with a forward genetic approach, aided the identification of many genes that regulate progression from G₂ to M, including cdc2⁺, wee1⁺ and cdc25⁺.^30-33 The products of these S. pombe genes and their homologs in S. cerevisiae and vertebrates constitute the conserved core mechanism of the G₂/M checkpoint, consisting of a Cdk (Cdc2), which is phospho-regulated by an inhibitory kinase (Wee1) and an activating phosphatase (Cdc25). Both Wee1 and Cdc25, with their opposing kinase and phosphatase activities, respectively, are dose-dependent regulators of the phosphorylation state of Cdc2 (Cdk).^31,33-36 In turn, Cdc2 (Cdk) and PP2A have opposing kinase and phosphatase activities, respectively, that regulate the phosphorylation state of Cdc2 (Cdk)/cyclin complex substrates. As an additional level of regulatory feedback, PP2A also dephosphorylates Cdc25.³⁷

In addition to monitoring DNA integrity,^38,39 the Wee1-dependent G₂/M checkpoint in S. pombe couples cell size to mitotic entry.^30,40,41 While this coupling has been known since the 1970s, the mechanism has only been elucidated recently. Fission yeast are rod shaped and divide following growth at the ends of the rod, with cellular scission occurring in the middle of the rod at the midzone. Prior to mitosis, a gradient of Pom1 kinase forms from each end to the midzone.^42,43 Pom1 regulates an inhibitory kinase of Wee1, known as Cdr1/Nim1 (Hsl1p in S. cerevisiae). As the cell grows in size, the ends extend away from the midzone, decreasing the concentration of Pom1 at the midzone, which relieves the inhibition of Nim1. The result is the degradation of Wee1, following phosphorylation by Nim1, and subsequent entry into mitosis.^42,43 The extent to which a Pom1-like mechanism is conserved has yet to be determined. However, there is some evidence that suggests conservation across species. A Pom1 homolog, called Yak1p, exists in S. cerevisiae and functions as an inhibitor of the cAMP-protein kinase A (PKA) stress response pathway.⁴⁴ If Yak1p regulates Hsl1p, it may serve to link the G₂/M checkpoint to cell stress responses in S. cerevisiae.

Cycling Xenopus egg extracts and Xenopus oocytes, which have been used to study cell cycle progression biochemically, have been integral experimental models for the elucidation of the G₂/M checkpoint.^45,46 Utilization of Xenopus cycling egg extracts showed that in addition to the destruction of Wee1,⁴⁷ a protein that maintains the Cdc2/Cdk in an inhibited state by phosphorylation of T15,⁴³ bypass of the G₂/M checkpoint and entry into mitosis also requires the removal of this inhibitory phosphate.³³ The Xenopus model was therefore the first to show that the Cdc25 phosphatase dephosphorylates T15.^34,49 In cycling Xenopus extracts, Cdc25 itself becomes hyper-phosphorylated with the approach of mitosis.³⁷ Cdc25 hyper-phosphorylation leads to increased enzymatic activity in vitro, with the hyper-phosphorylated form of Cdc25 in Xenopus and S. pombe acting as a mitotic activator,^37,50 consistent with the observation that deletion of cdc25⁺ in S. pombe⁵¹ causes a delay at G₂/M that results in elongated cells.³¹

G₂/M Checkpoint in S. cerevisiae

As in S. pombe and vertebrates, Cdk/cyclin complexes in the budding yeast S. cerevisiae promote progression through the cell cycle. In S. cerevisiae, a single Cdk, Cdc28p, which is orthologous to S. pombe Cdc2, is essential for cell cycle progression. Although S. cerevisiae possesses other Cdks, such as Pho85p, only the deletion of CDC28 is lethal.¹⁵ Clns and Clbs are the two main classes of cyclins in S. cerevisiae, with Clns 1–3 serving in G₁ and Clbs 1–6 serving in S, G₂ or M phase (for a review, see ref. 18). Cdc28p/Clb2p is the key Cdk/cyclin complex that regulates G₂ to M progression. Between the end of S phase and the beginning of mitosis, a checkpoint regulated by the Swe1p (Saccharomyces wee1⁺ homolog) kinase can inhibit Clb2/Cdc28p and thus mitotic entry.⁵²

Study of the core mechanism of G₂/M checkpoint regulation in S. cerevisiae has focused mainly on the phospho-regulation of the Cdc28p/Clb2p complex. When phosphorylated on residue T19 (functionally equivalent to T15 in S. pombe), Cdc28p has significantly reduced kinase activity, resulting in a delay in entry into mitosis.⁵³ The Swe1p kinase is responsible for this inhibitory phosphorylation.⁵² Both Swe1p and Mih1p (mitotic inducer homolog), which is a homolog of S. pombe Cdc25,⁵¹ shuttle in and out of the nucleus, regulating both nuclear and cytoplasmic pools of Cdc28p.⁵⁴ At G₂/M, Swe1p translocates from the nucleus to the bud neck, where it localizes to septin rings, dependent upon the activity of the Nim1p-like kinase Hsl1p and the prior localization of Hsl1p and Hsl7p (an arginine methyltransferase) to septin rings.^55-58 After Swe1p localizes to the mother-bud neck, it is sequentially phosphorylated by the Polo kinase Cdc5p⁵⁹ and the p21-activated kinase Cla4p.^60,61 This hyper-phosphorylated form of Swe1p becomes ubiquitinated and is thereby targeted for degradation by the Skp, Cullin, F-box-containing (SCF) ubiquitin ligase complex.⁶² Without Swe1p, no inhibitory phosphorylation of Cdc28p occurs. Efficient Swe1p degradation relies on hyper-phosphorylation and is necessary for entry into mitosis. However, phosphorylation of Swe1p is not just inhibitory. In a feedback loop, Cdc28p/Clb2p phosphorylates Swe1p,⁵⁹ further activating Swe1p kinase activity,⁶³ while at the same time priming Swe1p for degradation by the Cdc5p- and Cla4p-dependent mechanism just described.^60,64,65 The function of this dual regulatory effect, in which Cdc28p both activates and inhibits its own inhibitor, is still not fully understood. It is possible that this mechanism allows for inhibition of mitosis until the cell reaches a critical level of Cdc28/Clb2, allowing for an “all or nothing” switch for entry into mitosis.⁶³

Similar to the mechanism described in S. pombe, after degradation of Swe1p, activation of the Cdc28p/Clb2p complex requires the removal of the inhibitory phosphate on T19 through the activity of Mih1p phosphatase.^49,51,53,66 In contrast to vertebrate cells and S. pombe, Mih1p in S. cerevisiae is hyper-phosphorylated in G₁ and becomes dephosphorylated as cells approach mitosis.⁶⁷ Dephosphorylation of Mih1p requires PP2A^Cdc55p.⁶⁷ This appearance of a dephosphorylated form of Mih1p as cells approach mitosis has led some researchers to hypothesize that, in contrast to the situation in vertebrates and S. pombe,^37,68 dephosphorylated Mih1p is more active toward Cdc28p.⁶⁷ However, formal biochemical proof of this idea is lacking.

Although the existence of a cell size checkpoint in G₁ is well established in S. cerevisiae,^69,70 controversy remains as to whether the G₂/M checkpoint regulated by Swe1p and Mih1p in S. cerevisiae monitors cell size. In contrast to the deletion of wee1⁺ in S. pombe, which results in a significant reduction in cell size,³⁰ deletion of SWE1 in S. cerevisiae yields only a small, though statistically significant, reduction in cell size.^41,71 In some strain backgrounds, SWE1 deletion does not have an obvious effect on cell size, morphology or cell cycle progression in logarithmically growing cultures in the lab.⁵³ Likewise, deletion of MIH1 does not lead to obviously larger cells,⁵¹ in contrast to the elongated cells that result from CDC25 deletion in S. pombe.^30,31 Although deletion of SWE1 does not significantly affect cell division, overexpression of Swe1p delays nuclear division, with cells pausing with short spindles.⁵³ This delay, if prolonged, results in a distinct hyper-elongated bud with reiterative constrictions along the length of the bud, indicative of arrest in G₂. In addition, failure to form a bud delays nuclear division in a manner dependent upon the inhibitory phosphorylation of Cdc28p.⁷² These findings led to the hypothesis that the Swe1p-dependent G₂/M checkpoint serves as a “morphogenesis checkpoint” that detects incorrectly formed buds and delays mitosis to allow for proper bud growth or size before nuclear division. Seeming to support this “morphogenesis checkpoint” hypothesis were the observations that SWE1 deletion rescued certain genetic manipulations that slow bud growth or cause aberrant bud morphologies.^26,73-77 However, it was later shown that nuclear division continues in a strain with a defective myosin (myo2–16^ts), which arrests at the small-budded stage when shifted to a non-permissive temperature during S phase.⁷⁸ This result showed that the Swe1p-dependent G₂/M checkpoint does not appear to monitor bud geometry.

Does the Swe1p-dependent G₂/M checkpoint in budding yeast monitor DNA integrity as it does in fission yeast and vertebrate cells?^38,39 The answer is no. Induction of DNA damage by irradiation causes S. cerevisiae cells to arrest in a SWE1-independent fashion, indicating that, in contrast to fission yeast and vertebrate cells, the Swe1p-dependent checkpoint in budding yeast does not directly monitor DNA damage.^79,80 DNA damage-induced arrest in budding yeast is controlled instead by the Rad9p/Rad53p/Mec1p pathway, which limits G₁/S transition and anaphase progression.^10,81,82

What the G₂/M Swe1p-dependent checkpoint actually monitors in S. cerevisiae remains unknown, though new ideas of its physiological significance are emerging. Wild type cells have an absolute requirement for polymerized actin for progression through G₂/M. Treatment of cells with the drug latrunculin, which leads to actin depolymerization, or the presence of specific mutations in ACT1, the only actin-encoding gene in S. cerevisiae, causes activation of the G₂/M checkpoint.⁷⁸ Deletion of SWE1 results in an inability to halt mitotic progression upon disruption of the actin cytoskeleton.^78,83 This observation led to the hypothesis that the G₂/M Swe1p-dependent checkpoint is an actin-sensing checkpoint that monitors the integrity of the actin cytoskeleton.⁷⁸ A similar cytoskeleton-checkpoint link exists in mammals. Margolis et al.⁸⁴ showed that keratin filaments act as a sink for a 14-3-3 protein, an inhibitor of the Cdc25p phosphatase (see refs. 85–87 and refs. therein), allowing for coordination between intermediate filament networks and mitotic progression. However, there are no known keratin genes in S. cerevisiae, and the 14-3-3 protein homologs (BMH1 and BMH2) have no obvious links to the S. cerevisiae Swe1p-dependent G₂/M checkpoint. Therefore, evidence that the Swe1p-dependent G₂/M checkpoint directly senses actin is lacking. An interesting modified version of the “actin-sensing” hypothesis is that progression through the Swe1p checkpoint merely requires an actin-dependent process, such as polarized exocytosis, rather than the checkpoint acting as an intrinsic actin sensor. An intriguing recent study by Kellogg and colleagues,⁸⁸ discussed below, supports this hypothesis.

Protein Phosphatase 2A (PP2A) and the G₂/M transition

PP2A catalyzes a large portion of the dephosphorylation events in the cell, including those that regulate the G₂/M checkpoint. PP2A is a highly abundant and ubiquitous heterotrimeric serine/threonine phosphatase (reviewed in ref. 89). The catalytic (C) subunit of this phosphatase is constitutively associated with the structural (A) subunit. Localization and substrate specificity is accomplished through binding to various regulatory B subunits.⁹⁰ In mammalian cells, there are at least 16 regulatory subunits comprising four classes: B, B′, B,′′ and B′′′ (PP2A structure reviewed in refs. 91 and 92). PP2A gained notoriety when it was discovered that suppression of PP2A, by treating mice with okadaic acid in a two-stage carcinogenesis experiment, led to an increase in tumors.^93,94 PP2A regulates the G₂/M checkpoint by modulating the phosphorylation state, and thus the activity, of Cdc25 and its homologs.^37,49,95 In addition, PP2A has been show to regulate the phosphorylation state of Swe1.⁹⁶ For these activities, PP2A utilizes the regulatory B55δ or 56 subunits in Xenopus,^23,84 the B subunit Cdc55p in S. cerevisiae^67,96 and most likely Pab1 in S. pombe⁹⁷ and B55 in humans.⁹⁰ Because PP2A regulates a checkpoint that is suppressed in many cancers, it is a target of several antitumor treatments (reviewed in ref. 29).

Although the mechanism by which PP2A regulates the Swe1p-dependent G₂/M checkpoint in S. cerevisiae has yet to be fully explored, loss of the B-class PP2A regulatory subunit Cdc55p has yielded insights. In wild type S. cerevisiae, Mih1p is dephosphorylated as cells enter mitosis.⁶⁷ Loss of Cdc55p causes persistent hyper-phosphorylation of Mih1p,⁶⁷ suggesting that Mih1p is a PP2A^Cdc55p substrate. Whether hyper-phosphorylation affects Mih1p phosphatase activity (either in vitro or in vivo), targeting, or localization is not known. Loss of Cdc55p also delays Swe1p hyper-phosphorylation and degradation,⁹⁸ in addition to promoting hyper-polarized bud growth reminiscent of Swe1p overexpression.²⁶ These results led to the proposal that PP2A^Cdc55p is a negative regulator of Swe1p, in contrast with the fact that PP2A is a positive regulator of Wee1 (Swe1p homolog) in vertebrates.⁹⁹ Inconsistent with the PP2A^Cdc55p -Swe1p negative regulation model in S. cerevisiae are data indicating that deletion of CDC55 suppresses the Swe1p-dependent G₂/M checkpoint.^100,101 Thus, the function of PP2A^Cdc55p in S. cerevisiae Swe1p-dependent checkpoint regulation remains debatable. However, these controversies may be explained by evidence that Cdc55p, presumably associated with other PP2A subunits, plays a more complex role in mitosis than was appreciated initially.

The Other Side of the Coin: Phosphatase Inhibition for G₂ to M Progression

For decades, the central dogma of mitotic control has revolved around the regulation of the mitotic Cdks and cyclins. However, recent evidence indicates that control of a phosphatase that opposes the kinase activity of mitotic Cdk-cyclin complexes is equally important. Prior to M phase, phosphorylation of mitotic substrates is counteracted by PP2A^B55δ, the activity levels of which are high in interphase and low in mitosis, as shown with cycling Xenopus extracts.²³ Therefore, in addition to the activation of mitotic Cdks, PP2A inhibition is needed for mitosis to begin. Four concurrent independent studies shed light on the mechanism of PP2A inhibition at the entry into mitosis. Two of these studies were in vitro, utilizing frog egg extracts,^102,103 and the other two were in vivo, utilizing S. cerevisiae.^101,104

In Xenopus, PP2A inhibition at G₂/M requires Greatwall (Gwl) kinase,^102,103 which is required for entry into mitosis.^23,25,105 Gwl kinase was first discovered in Drosophila, in which mutations of this kinase resulted in delayed G₂ to M progression.¹⁰⁶ Because Gwl kinase does not directly phosphorylate PP2A,^25,105 the existence of one or more additional signaling proteins to bridge Gwl kinase to PP2A was postulated. In 2010, Mochida et al.¹⁰³ and Gharbi-Ayachi et al.¹⁰² showed that cAMP-regulated phosphoprotein-19 (Arpp19) and α-endosulfine (Ensa), two proteins with high sequence homology, inhibited PP2A complexes containing a B55δ regulatory subunit,^102,103 and that this inhibition required phosphorylation of Arpp19 and Ensa by Gwl kinase.^102,103 It is inconclusive, however, whether both Arpp19 and Ensa inhibit PP2A upon mitotic entry. Mochida et al.¹⁰³ reported that Ensa-depleted Xenopus extracts never enter mitosis, whereas Gharbi-Ayachi et al.¹⁰² reported that only the depletion of Arpp19, but not Ensa, completely inhibited entry into mitosis. Nonetheless, these elegant and labor-intensive studies linked an upstream signaling event in vertebrates to the regulation of mitotic entry by PP2A. Moreover, the relevance to disease is direct; depletion of Arpp19 in HeLa cells leads to a 50% decrease in the number of mitotic cells.¹⁰²

A similar mechanism of PP2A regulation at G₂/M appears to exist in budding yeast, though it does not employ an obvious homolog of Ensa or Arpp19. In S. cerevisiae, Zds2p and its paralog, Zds1p, are inhibitors of PP2A^Cdc55p.¹⁰¹ The Zds proteins, which may have a role analogous to Arpp19 and possibly Ensa, were discovered in S. cerevisiae as regulators of calcium tolerance,¹⁰⁷ Cdc42p-mediated polarized cell growth¹⁰⁸ and cell cycle progression.¹⁰⁹ In all, the ZDS genes were identified in 20 independent studies (summarized in ref. 110), hence the name zillion different screens (ZDS). Deletion of both ZDS1 and ZDS2, but neither alone, results in elongated buds with multiple constrictions along the bud and a G₂ delay, reminiscent of the morphology of SWE1-overexpressing cells.^108,109 Suppression of this phenotype by loss of Swe1p suggested that the Zds proteins regulate the Swe1p-dependent G₂/M checkpoint.¹¹¹ In deciphering the mechanism of this regulation, Yasutis et al.,¹⁰¹ relying on a genetic approach that focused on Zds2p, and Wicky et al.,¹⁰⁴ relying on a more biochemical approach that focused on Zds1p, identified Cdc55p, a regulatory subunit of PP2A, as a Zds-interacting protein.^101,104 These data are consistent with high-throughput studies (see refs. in ref. 110) that showed protein-protein interactions between the Zds proteins and Cdc55p as well as with reports of co-immunopurification of Cdc55p and Zds1p.^104,112 Two observations position PP2A^Cdc55p downstream of the Zds proteins in the G₂/M checkpoint: (1) accelerated entry into mitosis upon Zds1p or Zds2p overexpression is Cdc55p-dependent, and (2) deletion of CDC55 suppresses the G₂ arrest phenotype (long buds) of zds1∆ zds2∆ cells.¹⁰¹ These observations indicate that the Zds proteins inhibit Cdc55p at G₂/M, allowing the transition into mitosis. In contrast to Arpp19 and Ensa,^102,103 data to show that Zds proteins inhibit the phosphatase activity of PP2A^Cdc55p in vitro are lacking, though an experiment by Queralt and Uhlmann¹¹² showed that PP2A^Cdc55p isolated from cells that overexpressed Zds1p was less active than PP2A^Cdc55p isolated from wild-type cells.

There are several interesting data that contradict the model of Zds proteins serving as negative regulators of PP2A^Cdc55p in the G₂/M transition. In vivo, the Zds proteins are strictly required for the PP2A^Cdc55p-dependent dephosphorylation of the Mih1p phosphatase, the Cdc25 homolog that opposes the phosphorylation of Cdk1p-cyclin complexes by Swe1p (Wee1).¹⁰⁴ Thus, the Zds proteins can promote PP2A^Cdc55p activity. Furthermore, work by Rossio and Yoshida¹¹³ showed that the Zds proteins, which localize to the cytoplasm and bud tip,^101,108,113 are necessary for the translocation of Cdc55p from the nucleus to the cytoplasm at G₂/M. A mutant form of Cdc55p containing a nuclear export signal rescued the G₂/M transition delay caused by deletion of the ZDS genes, whereas sequestration of Cdc55p in the nucleus delayed entry into mitosis. These observations provide a model in which the Zds proteins promote Cdc55p function in the cytoplasm, thereby inhibiting Swe1p and, in turn, Cdc28p phosphorylation.¹¹³ Inconsistent with this model, however, is the observation by Harvey et al.⁹⁵ that hyperphosphorylation of Swe1p, which promotes mitosis, is inhibited by PP2A^Cdc55p. This same study also showed that PP2A^Cdc55p inhibits the initial activating phosphorylation of Swe1p. Taking these data together, it appears that PP2A^Cdc55p has two opposing functions at G₂/M, perhaps regulated by localization, which resolves the conflict between what are viewed as mutually exclusive models for the regulation of PP2A^Cdc55p by the Zds proteins.

Although the S. cerevisiae and Xenopus models of PP2A-dependent cell cycle regulation contain non-homologous signaling proteins at some signaling steps, the overall architecture of these regulatory pathways shows commonalities (Fig. 1). First, in both species, an upstream inhibitor of PP2A promotes entry into mitosis in the absence of a phosphatase (Mih1p in S. cerevisiae; Cdc25 in Xenopus) that dephosphorylates mitotic Cdk-cyclin. This observation indicates that regulation of mitotic entry by Zds proteins or Arpp19/Ensa is through PP2A and not by direct regulation of mitotic Cdk-cyclins.^101,102 However, it remains possible that the Zds proteins, and possibly Arpp19/Ensa, partially contribute directly to mitotic Cdk-cyclin regulation because, as observed in S. cerevisiae, overexpression of ZDS1 and ZDS2 in mih1∆ cells induced mitotic entry, though at a rate much slower than measured in cells with wild-type MIH1.¹⁰¹ Wicky and colleagues¹⁰⁴ also noted that Zds1p and Zds2p regulate the dephosphorylation of Mih1p by PP2A^Cdc55p. A second commonality between Xenopus and S. cerevisiae is that the upstream inhibitor of PP2A is activated by phosphorylation in a cell cycle-dependent manner.¹⁰⁴ In Xenopus, the activating kinase is Gwl.^102,103 In S. cerevisiae, Protein Kinase C (Pkc1p) activates the Zds proteins.⁸⁸ It appears, therefore, that Gwl and Pkc1p are functional homologs with respect to PP2A inhibitor activation. As a third commonality, the PP2A inhibitors in each mechanism (i.e., Arpp19/Ensa or Zds) appear to be signaling nodes that integrate multiple upstream signaling pathways. This idea is supported by the observation that both Arpp19 and Ensa contain target consensus sequences for both Cdk and protein kinase A,¹⁰³ in addition to a target consensus sequence for Gwl kinase.^102,103 While much research has focused on the regulation of mitotic Cdk-cyclin pairs, the future lies in determining the signaling pathways that regulate the core checkpoint kinase, Cdk and its opposing phosphatase, PP2A. These studies represent the first exciting steps in this direction.

Figure 1. Cross-species comparison of opposing phosphatase and kinase signaling pathways that regulate G₂ to M progression, with predictions of proteins yet to be identified (question marks) or pathways yet to be validated (hatched arrows and T-bars). Shown are PP2A-dependent signaling pathways known to inhibit entry into M phase as well as the Cdk (green)-dependent pathways that promote entry into M phase. Not shown is Wee1 kinase (and similar kinases), which inactivates Cdk directly. Functional homologs are shown in the same color. Solid arrows and T-bars denote experimentally validated signaling events. For S. cerevisiae, gray and black arrows denote pathways discovered independently.

Cell Growth Lies Upstream of Zds Proteins

The signals that link cell growth to cell cycle progression are especially significant and have been a mystery for quite some time. In this regard, a recent study made a major stride in delineating a signaling pathway between a process that supports cell growth and one that allows entry into mitosis.⁸⁸ Kellogg and colleagues⁸⁸ found that inactivation of the Swe1p-dependent G₂/M checkpoint and entry into mitosis depends upon exocytosis (Fig. 1). In their work, mutational block of exocytosis, but not endocytosis, prevented Pkc1p-dependent phosphorylation of Zds proteins, the subsequent activation of the Mih1p phosphatase and, in turn, the dephosphorylation (activation) of Cdk1p. (To explain why the mutation used by Kellogg and colleagues⁸⁸ to block exocytosis triggered a Swe1-dependent arrest, while that used by Lew and colleagues^78,114 did not, requires further research). Pkc1p, a known effector of Rho1p, depends upon Rho1p for activation.^115,116 Prior to this biochemical study,⁸⁸ genetic interactions revealed a link between PKC1 and the ZDS genes.^117,118 What sensor signal Rho1p is actually transmitting is not known; however, genetics may again provide a clue. ZDS1 was identified as a multi-copy suppressor of a temperature-sensitive allele of FKS1, which encodes glucan synthase, an enzyme required for cell wall formation.¹¹⁹ In this pathway, Rho1p is not responding to cell size, because cell size does not appear to be coupled to the Swe1p-dependent G₂/M checkpoint in S. cerevisiae. It is more likely that Rho1p is transmitting a cell wall integrity signal. In this model, when the cell wall lacks integrity, Rho1p is not activated, preventing activation of Pkc1p and the Zds proteins, as well as inhibition of the G₂/M checkpoint. Consistent with this model, disruption of Zds and Zds-like proteins in the yeasts S. pombe and Cryptococcus neoformans increases cell wall thickness or mass concomitant with a reduction in cell wall integrity, implicating Zds proteins in a signaling pathway(s) that reports cell wall integrity.^120,121 Drawing firm conclusions about the relationship between cell wall integrity and checkpoint control will be challenging because extended activation of the Swe1p-dependent G₂/M checkpoint results in decreased cell wall integrity along the lateral bud cortex,¹²² making it unclear whether reduced cell wall integrity activates the checkpoint or is the result of checkpoint activation.

Perspectives

Since the discovery of cell cycle mutants in yeast and MPF in frog extracts over four decades ago,^16,30,123 we have learned much about the mechanics of the cell cycle. Milestones include the biochemical purification of MPF,¹²⁴ the landmark discovery that Cdks and their regulatory proteins are functionally conserved across species from yeast to vertebrates,^32,125-127 and the discovery of transcriptional, translational and post-translational regulatory mechanisms of cell cycle progression.^128-131 More recently, we have gained an appreciation that cell cycle progression relies not only on a core mechanism consisting of the activation of a master kinase but also on the concomitant inhibition of an opposing phosphatase. Together, this research has allowed us to formulate the first system-level models of cell cycle progression and checkpoint control.

Much work still lies ahead to continue the development of a model of cell cycle progression that utilizes insights gained from the four traditional model organisms of cell cycle research (Xenopus, S. cerevisiae, S. pombe and human cells) as well as from Drosophila, C. elegans andplants. Currently, the Xenopus and S. cerevisiae models are more developed with respect to delineation of signaling pathways upstream of Cdk and PP2A than those of other model organisms, but in unique ways. The Xenopus model has provided rich data that describe both the cell-signaling events that lead to the inhibition of a phosphatase that opposes Cdks and the positive feedforward regulation of Cdk-cyclin complexes. The S. cerevisiae model lags slightly on both points. The S. cerevisiae model, however, excels in connecting an upstream cellular process, i.e., exocytosis, to the G₂/M checkpoint. Whether cell types that divide without growth (e.g., Xenopus oocytes) share a similar connection between intracellular trafficking and checkpoint control remains to be seen. Regardless of which cellular process is generating a signal to the G₂/M checkpoint, the goal remains in each model to link these upstream signals to the core checkpoint mechanism, which consists of a master Cdk and an opposing phosphatase. This elucidation is especially important for cancer research, where it is now appreciated that cancer cells that have mutations that render them almost entirely dependent on the G₂/M checkpoint for division control are much like yeast cells.²⁹ A priori, it is not possible to assign greater value to one model organism for this G₂/M research than another, because what appears important is the architecture of the pathways that feed into the core checkpoint mechanism, not the identity of the proteins per se. For example, at G₂/M, Ensa and Arpp19 in vertebrates are likely functional homologs of the Zds proteins in fungi, even though they share no sequence similarity. Notably, phosphorylation of the bona fide Ensa/Arpp19 sequence homologs, Igo1p and Igo2p, in S. cerevisiae by the Gwl-like kinase Rim15p does not influence mitotic entry.¹³² The activity of Igo1p and Igo2p appears centered instead on entry into G₀, when, in the absence of nutrients, Igo1p (and perhaps Igo2p) binds to and inhibits PP2A^Cdc55.¹³³ Thus the value of different models of cell cycle progression lies not only in revealing similarities in the regulation of G₂/M progression, but also in revealing the variance and plasticity possible.¹³⁴

Irrespective of the model used, the challenges in building a systems level view of cell cycle progression will be to determine the following: (1) the stress signaling pathways that feed into the checkpoint mechanism and feedback regulation; (2) the physiological relevance of the specific phospho-states of proteins; (3) the binding constants of interacting proteins and (4) how these pathways are spatially regulated, considering the nucleocytoplasmic shuttling of checkpoint proteins.⁵⁴

Glossary

Abbreviations:

Arrp

cAMP-regulated phosphoprotein-19

Cdk

cyclin-dependent kinase

Ensa

α-endosulfine

Gwl

Greatwall kinase

MPF

maturation promoting factor

PKA

protein kinase A

PKC

protein kinase C

PP2A

protein phosphatase 2A