Signaling Pathways that Regulate Cell Division

Abstract

Cell division requires careful orchestration of three major events: entry into mitosis, chromosomal segregation, and cytokinesis. Signaling within and between the molecules that control these events allows for their coordination via checkpoints, a specific class of signaling pathways that ensure the dependency of cell-cycle events on the successful completion of preceding events. Multiple positive- and negative-feedback loops ensure that a cell is fully committed to division and that the events occur in the proper order. Unlike other signaling pathways, which integrate external inputs to decide whether to execute a given process, signaling at cell division is largely dedicated to completing a decision made in G1 phase—to initiate and complete a round of mitotic cell division. Instead of deciding if the events of cell division will take place, these signaling pathways entrain these events to the activation of the cell-cycle kinase cyclin-dependent kinase 1 (CDK1) and provide the opportunity for checkpoint proteins to arrest cell division if things go wrong.

Two major transitions are required for cell division: the G2-M transition (regulated by Cdk1) and the metaphase-anaphase transition (regulated by APC). These are the main targets of signaling pathways that control cell division.

1. INTRODUCTION

The cell cycle (see Fig. 1) consists of DNA synthesis (S) and mitosis (M) phases separated by gap phases in the order G1–S–G2–M (Murray and Hunt 1993; Nurse 2000; Morgan 2006). Cell division involves two connected processes triggered at the end of G2 phase: mitosis itself (segregation of the chromosomes, which duplicate in S phase) and cytokinesis (division of the cell, per se). Mitosis can be subdivided into six distinct phases (see Box 1): (1) prophase, in which the spindle begins to assemble in the cytoplasm and chromosomes begin to condense in the nucleus; (2) prometaphase, in which the nuclear envelope breaks down and chromosomes attach to the spindle; (3) metaphase, in which chromosomes align at the spindle midzone; (4) anaphase A, in which chromosomes move to the centrosomes, which form the spindle poles; (5) anaphase B, in which the spindle elongates; and (6) telophase, in which the nuclear envelope reforms around the new daughter nuclei. Mitosis in yeasts differs in that the nuclear envelope does not break down; instead the spindle-pole body, the yeast equivalent of the centrosome, spans the nuclear envelope, allowing the spindle to access both the nucleus and the cytoplasm. Signals during telophase trigger cytokinesis, which separates the daughter nuclei into two daughter cells.

Figure 1.

The major events of the cell cycle. The major events of the cell cycle are regulated by successive waves of kinase and ubiquitin ligase activity. G1-cyclin–CDK activity is required to initiate the cell cycle and activate B-type-cyclin–CDK activity. Low levels of B-type-cyclin–CDK activity are sufficient to trigger S phase, but tyrosine phosphorylation by Wee1 prevents full activation, preventing premature mitosis. Full CDK activation triggers mitosis and activates APC, which triggers anaphase and feeds back to inactivate CDK activity. Inactivation of CDK allows exit from mitosis and the reestablishment of interphase chromosome and nuclear structure in G1 phase. See Box 1 for description of the stages of mitosis.

BOX 1: THE MAIN STAGES OF MITOSIS.

Prophase Triggered by activation of CDK1, CDK1–cyclin-B is imported into the nucleus, chromosomes condense from their diffuse interphase state to compact rods, and the centrosomes begin to separate, forming a spindle between them.

Prometaphase The nuclear envelope breaks down, the mature spindle is formed with centrosomes on either side of the cell, and so-called kinetochore microtubules from the spindle poles interact with the kinetochore protein complexes that form at chromosome centromeres, moving chromosomes to the spindle midzone. In yeasts, which do not break down their nuclear envelopes for mitosis, the spindle forms inside the nucleus, and the spindle poles span the nuclear envelope to connect nuclear and cytoplasmic microtubules.

Metaphase When each of the kinetochores on a pair of sister chromosomes is attached to microtubules from opposite spindle poles, the opposing force pulls the pair to the metaphase plate at the middle of the spindle in a “bi-oriented” configuration. Until all the chromosomes are bi-oriented, unattached kinetochores produce a checkpoint signal that prevents the metaphase/anaphase transition.

Anaphase Once the mitotic checkpoint has been satisfied, the APC ubiquitin ligase is activated. Ubiquitin-dependent proteolysis of the inhibitor securin leads to activation of separase—a protease that cleaves the cohesin proteins that hold sister chromatids together. This cleavage causes a loss of chromosome cohesion, allowing the chromosomes to separate. They do so in two movements: anaphase A, in which chromosomes are pulled toward the spindle poles by contraction of the kinetochore microtubules; and anaphase B, in which the spindles are pushed away from each other by the elongation of inter-spindle-pole microtubules.

Telophase After the chromosomes have been segregated to opposite sides of the cell, CDK1 activity is inhibited by APC-mediated destruction of cyclin B and activation of the CDK1-antagonizing phosphatase Cdc14. The reduction in CDK1 activity allows for reformation of the nuclear envelope, decondensation of chromosomes, and entry into G1.

The signaling pathways that operate in G2 phase to control the onset of mitosis, and those that operate during mitosis to control chromosome segregation and the initiation of cytokinesis, are somewhat different from other cellular signaling pathways. Instead of being involved in decisions about potential cell fates or responses to varying environmental conditions, they are involved in executing a decision that is made in G1 phase—to enter and complete another cell cycle. These signaling pathways have two key roles: to coordinate potentially independent events, such as mitosis and cytokinesis; and to provide quality-control checkpoints that arrest the process when things go wrong. The cell division signaling pathways are the archetypical checkpoints, defined as signaling pathways that ensure a dependency for the execution of later cell-cycle events on the successful completion of preceding events (Hartwell and Weinert 1989).

Two major transitions are required for cell division: the G2/M transition and the metaphase/anaphase transition. These are regulated by the protein kinase cyclin-dependent kinase 1 (CDK1) and the anaphase-promoting complex (APC, an E3 ubiquitin ligase), respectively (Murray and Hunt 1993; Nurse 2000; Morgan 2006). During G2 phase CDK1 is maintained at a low level of activity by mechanisms described below. Activation of CDK1 is necessary and sufficient to trigger entry into mitosis. Thus, CDK1 activation is the focal point of many signaling pathways that control the commitment to cell division. These pathways have both positive effects, such as activating CDK1 when cells reach a critical size, and negative effects, such as delaying CDK1 activation in the presence of DNA damage. In general, the negative regulatory pathways, known as checkpoints, are much better understood than the positive regulatory pathways. Activation of CDK1 and entry into mitosis lead to activation of the APC. The APC in turn causes separation of sister chromatids, which allows them to segregate to opposite spindle poles at anaphase and the complete inactivation of CDK1, which allows exit from mitosis and the resetting of the cell cycle to G1 phase. Being necessary and sufficient for the initiation of anaphase, the APC is the major target of the checkpoints that arrest cells in metaphase (Fig. 1).

Below we examine these G2 and mitotic signaling pathways and the checkpoints that regulate them, explaining how they ensure that events take place in the correct order even in the face of cell-cycle disruptions.

2. CDK1 AND ENTRY INTO MITOSIS

The onset of mitosis is brought about by the activation of CDK1, a 34-kDa proline-directed serine/threonine protein kinase (Nurse 1990). CDK1, also known as Cdc2 (cell division cycle 2), phosphorylates hundreds of different substrates at an S/TPx(x)R/K consensus, including downstream effector kinases, to initiate the events required for progression through mitosis. Unsurprisingly, it is tightly regulated at many levels. CDK1 is maintained at a constant concentration throughout the cell cycle; however, its periodic activation and inactivation is achieved by interactions with specific regulatory subunits and by dephosphorylation of critical residues.

As their name indicates, CDKs such as CDK1 require binding of a regulatory cyclin subunit (Murray and Hunt 1993; Morgan 1995; Bloom and Cross 2007), which activates the kinase by causing conformational changes near the active site. These structural changes allow the kinase to bind ATP and substrates in an orientation that promotes the transfer of the terminal γ phosphate of ATP to the target serine or threonine residue in the substrate. Different CDK–cyclin heterodimers regulate the different cell-cycle transitions (Morgan 2006). The G1 cyclins, in association with CDK4, CDK6, and CDK2, regulate entry into the cell cycle (Duronio and Xiong 2012), whereas S-phase events and the G2/M transition are primarily regulated by CDK1 (CDK2 also functions in S phase) bound to a members of the B-type family of cyclins (the CDK1–cyclin-B complex is also known as M-phase promoting factor [MPF]).

The periodic accumulation and disappearance of cyclins has a central role in regulating CDK1 activity during the cell cycle. B-type cyclin genes are generally expressed during S and G2 phases, leading to the accumulation of CDK1–cyclin complexes during these phases of the cell cycle. At the metaphase/anaphase transition, APC ubiquitylates B-type cyclins, which targets them for rapid destruction by the proteasome. The destruction of cyclins causes a precipitous drop in CDK activity, resetting the cell cycle as cells enter G1 phase.

B-type cyclins comprise a large and quickly evolving protein family (Table 1). Most species express multiple B-type cyclins, which are believed to confer different substrate specificities on CDK1–cyclin complexes. However, these differences do not appear to be crucial. For example, cyclin A (a mammalian B-type cyclin) is expressed during S phase and is responsible for triggering DNA replication in conjunction with CDK2, but in its absence cyclin B can provide this function (Kalaszczynska et al. 2009). More dramatically, in the fission yeast Schizosaccharomyces pombe, the cell cycle can be regulated by a single cyclin–CDK1 complex, demonstrating that differential B-type-cyclin–CDK1 specificities are not required for S-phase and M-phase events (Coudreuse and Nurse 2010). Likewise, a single B-type cyclin can drive S phase and M phase in budding yeast (Haase and Reed 1999).

Table 1.

Key proteins in cell division control

Protein	Budding yeast	Fission yeast	Human
Cyclin-dependent kinase 1	Cdc28	Cdc2	CDK1a
S-phase-expressed B-type cyclin	Clb5,6	Cig2	Cyclin A
M-phase-expressed B-type cyclin	Clb1,2	Cdc13	Cyclin B
CDK-inhibitory kinase	Swe1	Wee1, Mik1	WEE1, MYT1
CDK-activating phosphatase	Mih1	Cdc25	CDC25B, CDC25C
Checkpoint kinase	Tel1	Tel1	ATM
Checkpoint kinase	Mec1	Rad3	ATR
Checkpoint effector kinase	Chk1	Chk1	CHK1
Checkpoint effector kinase	Rad53, Dun1	Cds1	CHK2
MRN nuclease	Mre11	Rad32	MRE11
MRN scaffold	Rad50	Rad50	RAD50
MRN regulator	Xrs2	Nbs1	NBS1
ATR targeting subunit	Ldc1	Rad26	ATRIP
9-1-1 checkpoint clamp	Ddc1	Rad9	RAD9
9-1-1 checkpoint clamp	Rad17	Rad1	RAD1
9-1-1 checkpoint clamp	Rad24	Hus1	HUS1
Checkpoint mediator	–b	–	MDC1
Checkpoint mediator	Rad9	Crb2
Checkpoint mediator	Dpb11	Rad4/Cut5	TopBP1
Checkpoint mediator	Mrc1	Mrc1	Claspin
Fork protection complex	Tof1	Swi1	Timeless
Fork protection complex	Csm3	Swi3	Tipin
Mitotic kinase	Cdc5	Plo1	Plk1-4
Mitotic kinase	Ipl1	Ark1	Aurora A/B
APC regulator	Cdc20	Slp1	CDC20
APC regulator	Cdh1	Srw1	CCH1
Mitotic checkpoint regulator	Mad1	Mad1	MAD1
Mitotic checkpoint regulator	Mad2	Mad2	MAD2
Mitotic checkpoint regulator	Mad3	Mad3	BUBR1
Mitotic checkpoint kinase	Bub1	Bub1	BUB1
Mitotic checkpoint regulator	Bub3	Bub3	BUB3
MEN/SIN scaffold	Nud1	Cdc11	?c
MEN/SIN GTPase	Tem1	Spg1	?
MEN/SIN GAP	Bub2	Cdc16	?
MEN/SIN GAP cofactor	Byr4	Byr4	?
MEN/SIN GEF	Lte1	Etd1	?
MEN/SIN kinase	Cdc15	Cdc7	MST1/2
MEN/SIN kinase	–	Sid1	MST1/2
MEN/SIN kinase regulator	–	Cdc14	?
MEN/SIN kinase	Dbf2	Sid2	LATS1/2
MEN/SIN kinase regulator	Mob1	Mob1	MOB1A,B
Phosphatase	Cdc14	Clp1	CDC14

^aNames used in the text are in bold.

^b-, No ortholog is believed to exist.

^c?, An ortholog has not been identified.

These results have led to a quantitative model of cell-cycle regulation, in which relatively high CDK1 activity is required to bring about the onset of M phase, whereas a much lower level of CDK1 activity is sufficient to catalyze the events of S phase (Stern and Nurse 1996; Coudreuse and Nurse 2010). Nonetheless, different cyclins are expressed at different times and confer differential substrate specificities on CDK1, allowing layers of fine-tuning to the basic quantitative model (Uhlmann et al. 2011).

Cyclin binding is necessary but not sufficient for robust CDK1 activity. Full activation of CDK1 also requires phosphorylation of a threonine residue near the active site, which causes further realignment of active-site residues into an active conformation. This phosphorylation is catalyzed by a CDK-activating kinase (CAK), which requires the CDK to be cyclin bound (Morgan 1995). Curiously, the identity of CAK varies, depending on the species, and in some organisms there are multiple CAKs. CAK activity does not change during the cell cycle, nor is the dephosphorylation of the CAK-phosphorylated CDK–cyclin complexes regulated in a cell-cycle-dependent manner. Thus, phosphorylation of CDK1 by CAK is dependent on cyclin binding and is essential for cell division, but it does not control when CDK1 is activated.

Full activation of CDK1, and thus entry into M phase, is restrained by Wee1 and related protein kinases (Fig. 2), which inhibit the activity of CDK1–cyclin complexes that accumulate during S and G2 phases (Russell and Nurse 1987; Lundgren et al. 1991; Mueller et al. 1995). Wee1 family kinases resemble serine/threonine protein kinases, but actually phosphorylate CDK1 on a tyrosine residue (Y15), as well as the adjacent threonine residue (T14) once it is bound to a cyclin (Gould and Nurse 1989; Featherstone and Russell 1991; Parker and Piwnica-Worms 1992; Mueller et al. 1995). These phosphorylations probably inhibit CDK1 kinase activity by interfering with substrate binding (Welburn et al. 2007). Once CDK1 is bound to a cyclin and phosphorylated by CAK and Wee1, it is primed and ready to be activated through the dephosphorylation of T14 and Y15. Cdc25, a dual-specificity protein phosphatase, catalyzes this dephosphorylation reaction to bring about the G2/M transition (Gautier et al. 1991; Kumagai and Dunphy 1991; Strausfeld et al. 1991; Beausoleil et al. 2006).

Figure 2.

Signaling at the G2/M transition. The rate-limiting step for the transition from G2 to mitosis is the dephosphorylation of CDK1 on Y15, and in some organisms T14. This phosphorylation is catalyzed by the Wee1 family of dual-specificity kinases and the phosphate is removed by Cdc25 phosphatases. Most of the many signaling pathways that affect the G2/M transition regulate Wee1 or Cdc25. The DNA damage and replication checkpoints inactivate Cdc25; the morphogenesis and nutritional checkpoints activate Wee1. CDK1 regulates its own activation as part of a feedback loop by directly phosphorylating Wee1 and Cdc25 or doing so indirectly through Plk1.

Studies of fission yeast established that Wee1 and Cdc25 together determine when cells initiate mitosis (Nurse 1975; Russell and Nurse 1986, 1987). Mutants lacking Wee1 divide prematurely at about half the size of wild-type cells. Strains that have extra copies of wee1 divide at progressively larger sizes that directly correlate with increasing wee1 gene dosage. In contrast, elimination of Cdc25 activity generates cells that grow progressively larger and cannot initiate mitosis. Strains that overexpress Cdc25 look similar to wee1 mutants.

The opposing activities of Wee1 and Cdc25 underlie the rapid “switchlike” activation of CDK1 that occurs at the G2/M transition (Fig. 2). This behavior is controlled by feedback regulation from CDK1 (Pomerening et al. 2003). CDK1 can directly phosphorylate Wee1 and Cdc25. It can also activate Polo-like kinase 1 (Plk1), which in turn leads to the degradation of Wee1 and stimulates Cdc25 phosphatase activity (Kumagai and Dunphy 1996). These feedback loops create a bi-stable regulatory mechanism in which passage through a tipping point ensures the rapid and stable transition from a low-activity CDK1 state to a high-CDK1-activity state.

In addition to the feedback loops that activate CDK1, a feedback loop in animal cells inactivates the protein phosphatase 2A (PP2A), which antagonizes CDK phosphorylation (Wurzenberger and Gerlich 2011). CDK1 activates the Greatwall kinase, which in turn activates Arpp19 and α-endosulfine, two small inhibitors of the B55 isoform of PP2A (Fig. 3) (Castilho et al. 2009; Gharbi-Ayachi et al. 2010; Mochida et al. 2010). Inhibition of PP2A-B55 increases the effective activity of CDK1 by reducing the rate at which CDK1 substrates are dephosphorylated. These interacting feedback loops are believed to be critical for ensuring the commitment to mitosis that occurs at the G2/M transition (Domingo-Sananes et al. 2011).

Figure 3.

Signaling at the M/A transition. To establish metaphase, CDK1 directly, and indirectly through a network of kinases including Plk1 and Aurora A, phosphorylates substrates that trigger nuclear envelope breakdown, chromosome condensation, centrosome separation, and spindle assembly. In addition, in animal cells CDK1 activates the Greatwall kinase, which indirectly inactivates the CDK-antagonizing PP2A-B55 phosphatase via phosphorylation of the small phosphatase inhibitors Arpp19 and endosulfine. The rate-limiting step for the transition from metaphase to anaphase is the activation of the APC ubiquitin ligase. The APC is activated by binding of the Cdc20 regulatory subunit (and later Cdh1) and by CDK1 phosphorylation. Active APC targets securin for destruction, activating separase to release chromatid cohesion and trigger anaphase. In the presence of unattached kinetochores, anaphase is delayed by the mitotic checkpoint complex (MCC), a soluble inhibitor of the APC produced at unattached kinetochores. The APC also feeds back to inactivate CDK1 by targeting cyclin B for degradation and activates phosphatases (CDC14 in yeasts and PP1 and PP2A-B55 in animals) that oppose CDK activity, resetting the cell cycle to a CDK-free G1 state.

The regulation of protein subcellular localization, in addition to the regulation of protein activity, is important in the control of the G2/M transition. In particular, CDK1–cyclin-B complexes, which have critical nuclear targets, are maintained in the cytoplasm during G2 phase, where they can be inhibited by Wee1, but sequestered away from activation by nuclear Cdc25. One of the early events of the G2/M transition is the nuclear localization of CDK1–cyclin-B (Porter and Donoghue 2003). However, it is unclear whether nuclear localization of CDK1–cyclin-B causes its activation or is a consequence of activation that enforces the CDK1–Cdc25 positive-feedback loop. Subcellular localization of regulatory proteins also plays a crucial role in the control of cytokinesis, as described below.

3. REGULATION OF THE G2/M TRANSITION

The G2/M transition that commits cells to division is, in general, a default consequence of initiating the cell cycle at the G1/S transition; examples of cells voluntarily arresting in G2 phase are rare. However, what ultimately triggers the activation of CDK1 is unclear. Plk1 may do so (Pomerening et al. 2003; Barr et al. 2004), but such a model just begs the question of what activates Plk1. Moreover, the involvement of Plk1 in the feedback loops that regulate CDK1 makes it difficult to disentangle cause and effect.

One important input is cell size (Kellogg 2003; Tzur et al. 2009). In steady-state growth conditions, cell division is coordinated with cellular growth in such a way that a newly born cell doubles its mass before undergoing division. This coordination can be achieved by linking the activation of CDK1 at the G2/M transition to attainment of a particular cell size (Jorgensen and Tyers 2004). A number of potential mechanisms have been explored over the years (Turner et al. 2012), including size-dependent accumulation of activators (Tyson 1983), spatial gradients (Martin and Berthelot-Grosjean 2009; Moseley et al. 2009), and translational capacity (Polymenis and Schmidt 1997). Nonetheless, the mechanism by which cell size is measured is still unknown.

Another regulator, and one that is intimately entwined with cell size, is cell morphology. In budding yeast, defects in cell morphology or cytoskeletal architecture delay cell division (Howell and Lew 2012). This cell-cycle delay is enforced by the accumulation of Swe1, the budding yeast Wee1 ortholog. Swe1 is normally degraded at the end of S phase, which allows the dephosphorylation of CDK1 by Mih1, the budding yeast Cdc25 homolog. Swe1 degradation requires its localization to the bud neck; thus, disruption of bud-neck structure leads to Swe1 accumulation and cell-cycle arrest. In addition, disruption of the actin cytoskeleton also leads to Swe1 stabilization, further tying cell-cycle progression to cell morphology.

An unrelated signaling pathway required for proper morphology in fission yeast also affects the timing of division (Martin and Berthelot-Grosjean 2009; Moseley et al. 2009). Wee1 localizes at cortical nodes near the middle of cells. Pom1, a serine/threonine protein kinase, forms a concentration gradient, spreading out from the ends of cells, and indirectly activates Wee1 via inhibitory phosphorylation of Cdr2, a kinase that colocalizes with, and negatively regulates, Wee1 (Hachet et al. 2011). In newly born cells, which are short, the concentration of Pom1 near the center of the cells is reasonably high, which results in higher Wee1 kinase activity. As cells grow, the concentration of Pom1 decreases near the middle of the cells, decreasing Wee1 activity and tipping the Wee1–Cdc25 balance in favor of Cdc25 and activation of CDK1. This model explains how Pom1 can influence cell size at division; however, cells lacking Pom1 divide at a well-defined length, albeit about 15% shorter than that of wild type, which implies that Pom1 is not required to maintain accurate size control. Because Pom1-deficient cells frequently display asymmetrical division, instead of controlling cell size per se, Pom1 appears to function in a morphogenesis checkpoint that arrests cell-cycle progression if the incipient division site is too close to one of the cell tips.

The size at which cells initiate mitosis is also linked to nutrient conditions (Fantes and Nurse 1977). Although nutritional signaling via the PI3K and TOR pathways has profound effects on G1 cell-cycle progression (Ward and Thompson 2012; Duronio and Xiong 2012), it can also affect the G2/M transition. For instance, in fission yeast, poor nitrogen availability activates the Spc1/Sty1 mitogen-associated protein kinase (MAPK), which in turn activates Plk1, thereby inhibiting Wee1 and activating Cdc25 to delay CDK1 activation (Shiozaki and Russell 1995; Petersen and Nurse 2007). Note that this nutritional signaling may be considered part of the G2 checkpoints discussed below.

4. CHECKPOINT REGULATION OF THE G2/M TRANSITION

Although arrest in G2 phase in normal, unperturbed cells is uncommon, it occurs in response to activation of a variety of quality-control checkpoints. Because Y15 dephosphorylation of CDK1 is the rate-limiting step for entry into mitosis, these checkpoints target the regulators of this step: Wee1 and Cdc25. The best-understood checkpoints are those activated by DNA damage and problems with DNA replication.

5. THE DNA DAMAGE AND REPLICATION CHECKPOINTS

The purpose of the cell division cycle is to distribute complete and accurate copies of the genome to daughter cells. Genome instability arises if cells initiate mitosis when chromosomes are only partially replicated or are damaged by a double-strand DNA break (DSB). The consequences of genome instability can be cell death or neoplastic transformation. DNA damage and replication checkpoints ensure that the onset of nuclear division is delayed when chromosomes are broken or incompletely replicated. The cellular responses to DNA damage and replication-fork stalling are controlled by ATM (ataxia telangiectasia mutated), ATR (ATM and Rad3-related), and DNA-PK (DNA-dependent protein kinase) (Shiloh 2003; Cimprich and Cortez 2008). These protein kinases belong to the PIKK (phosphoinositide-3-like-kinase kinase) family, whose members include mTOR (Laplante and Sabatini 2012). PIKKs strongly prefer to phosphorylate serine and threonine residues followed by glutamine (S/TQ). Of the checkpoint kinases, ATM and ATR are conserved from yeasts to humans, whereas DNA-PK is found only in metazoans (but not all).

ATM, ATR, and DNA-PK are nuclear proteins that rapidly localize to sites of DNA damage through interactions with targeting factors. Acting with accessory proteins, these kinases set up signaling platforms on the chromatin that flanks DNA lesions. DNA-PK associates with the Ku70–Ku80 heterodimeric protein complex that binds to DNA ends and promotes repair by nonhomologous end joining. DNA-PK does not have a clearly established role in regulating cell-cycle progression. ATM and ATR, in contrast, control multiple responses to DNA damage, including regulation of DNA synthesis and repair proteins, regulation of transcription of the genes involved in DNA synthesis and repair, and regulation of cell-cycle progression. ATM and ATR regulate some of these activities directly, whereas others are controlled by their effector kinases: Chk1 and Chk2. ATM functions specifically in the signaling downstream from DSBs, whereas ATR is activated in response to formation of single-stranded DNA (ssDNA), which can occur through resection of DSBs, by endonucleolytic removal of other types of DNA lesions, or by the stalling or collapse of replication forks.

6. THE DNA DAMAGE CHECKPOINT TRIGGERED BY DSBS

DSBs are especially dangerous DNA lesions because they disrupt the physical continuity of chromosomes. Initiating mitosis with broken chromosomes results in gross genome instability, which is a feature of cancer in humans and is usually lethal in single-celled microorganisms (Hartwell and Weinert 1989; Harper and Elledge 2007; Jackson and Bartek 2009). DSBs therefore generate a potent checkpoint response. Indeed, in yeasts a single unrepaired DSB is sufficient to induce a robust checkpoint response that prevents the onset of mitosis even as cells continue to grow beyond the size at which they normally divide (Harrison and Haber 2006).

ATM initiates the checkpoint response to DSBs (Shiloh 2003). It associates with DSBs by interacting with the MRN protein complex, which consists of Mre11 (a nuclease), Rad50 (a dimeric scaffolding protein with ATPase activity), and Nbs1 (an adaptor protein) subunits (Fig. 2). Mre11 and Rad50, which are conserved in eubacteria and archaea, directly engage DNA ends. Nbs1, which is only found in eukaryotes, directly binds ATM (Stracker and Petrini 2011).

Once localized at DNA ends, ATM phosphorylates proteins involved in cell-cycle checkpoints, DNA repair, and chromatin structure (Fig. 2). Phosphoproteomic studies suggest that it has hundreds of substrates, but only a few have been studied in detail (Matsuoka et al. 2007). One of these is Chk2, which is activated by ATM. Another ATM target is an SQ motif in the exposed carboxy-terminal tail of the histone variant H2AX. Phospho-H2AX (also known as γH2AX) serves as a marker for DNA damage in large multi-kilobase domains of chromatin flanking DSBs (Bonner et al. 2008). In mammals, γH2AX establishes a recruitment platform for mediator of DNA-damage checkpoint protein 1 (Mdc1), which is critical for amplifying and maintaining the checkpoint signal (Stewart et al. 2003). The carboxy-terminal region of Mdc1 contains a pair of breast cancer susceptibility protein 1 (BRCA1) carboxy-terminal (BRCT) domains that form a phosphopeptide-binding pocket for the phosphorylated γH2AX tail (Stucki et al. 2005). Once Mdc1 binds to γH2AX, it recruits more MRN complex through an interaction involving phosphorylated motifs in Mdc1 and the amino-terminal forkhead-associated (FHA) domain of Nbs1 (Melander et al. 2008; Spycher et al. 2008). These constitutive phosphorylations of Mdc1 appear to be catalyzed by the kinase CK2. The MRN complex that binds to Mdc1 recruits additional ATM, thus establishing a checkpoint signal amplification loop that activates Chk2.

In addition to recruiting ATM to DSBs, the MRN complex functions with the DNA-end-processing factor CtIP (known as Sae2 in budding yeast and Ctp1 in fission yeast) to initiate the 5′-3′ resection of DSBs (Mimitou and Symington 2011). This resection generates ssDNA tails that are extended through further resection by Exo1 exonuclease and the Dna2–Sgs1 (also known as BLM) DNA endonuclease–helicase. These ssDNA tails are rapidly bound by heterotrimeric replication protein A (RPA). RPA is essential for homology-directed repair of DNA damage, but it is also a recruitment factor for ATR (Cimprich and Cortez 2008). ATR associates with RPA through its targeting subunit ATRIP (Fig. 2). ATM promotes CtIP recruitment to DSBs in mammals, thereby enhancing resection; hence, ATM is required for recruitment and activation of ATR at DSBs (You et al. 2009). Recruitment of the CtIP orthologs in budding yeast (Sae2) or fission yeast (Ctp1) does not require the ATM ortholog Tel1, which is a key reason why the ATR orthologs have a more dominant checkpoint role in these organisms.

DNA damage is recognized independently by a heterotrimeric ring-shaped protein complex known as the 9-1-1 (Rad9–Hus1–Rad1) checkpoint clamp. Rad17 and four subunits of replication factor C (RFC) form a protein complex that loads the 9-1-1 checkpoint clamp onto resected DNA ends. The 9-1-1 complex associates with TopBP1, which also interacts with ATR and enhances its kinase activity (Navadgi-Patil and Burgers 2009). Both ATR and 9-1-1 are required for checkpoint signaling. Furthermore, in budding yeast, ectopically targeting both complexes to the same locus is sufficient for checkpoint activation, which suggests that recruitment of the two complexes to sites of damage is necessary and sufficient for checkpoint signaling (Bonilla et al. 2008).

ATM and ATR share multiple substrates, including histone H2AX. In fission yeast the checkpoint mediator protein Crb2 has carboxy-terminal BRCT domains that bind to γH2AX. In addition, Crb2 also has Tudor domains that bind to histone H4 dimethylated on K20. Crb2 can also localize to DSBs by binding to the TopBP1 ortholog, which interacts with the 9-1-1 checkpoint clamp. Either Crb2-recruitment pathway is sufficient to mount a partial checkpoint response, but loss of both pathways abrogates the checkpoint because Crb2 localization at DSBs is required for activation of Chk1 by the ATR ortholog (Du et al. 2006).

7. THE DNA REPLICATION CHECKPOINT TRIGGERED BY STALLED REPLICATION FORKS

During DNA replication, ssDNA is exposed at the replication fork when DNA lesions, DNA-bound protein complexes, or limiting supplies of deoxynucleotide triphosphates (dNTPs) cause replicative DNA polymerases to slow down or stall. This ssDNA is bound by RPA, which elicits a checkpoint response by ATR/ATRIP and their orthologs (Bartek et al. 2004; Branzei and Foiani 2010). This checkpoint also requires the 9-1-1 checkpoint clamp and TopBP1, similarly to the DNA damage checkpoint activated by DSBs, but the checkpoint mediators are different. In budding and fission yeast Mrc1 (mediator of the replication checkpoint 1) travels with the replication fork and is required for activation of Chk2. In addition to other functions, the activated Chk2 arrests cell-cycle progression and maintains the S-phase pattern of gene expression (de Bruin et al. 2008; Dutta et al. 2008), which is critical for completing DNA replication. A similar checkpoint pathway operates in mammals, which use an Mrc1-related protein known as claspin to mediate ATR-dependent activation of Chk1 in response to stalled replication forks (Kumagai and Dunphy 2000). Another protein complex, the fork protection complex, comprising Tof1–Csm3 in budding yeast, Swi1–Swi3 in fission yeast, and Timeless–Tipin in mammals, also associates with stalled forks. The fork protection complex stabilizes stalled forks in a manner that promotes the replication checkpoint response (McFarlane et al. 2010).

8. CHK1 AND CHK2 TARGET CDC25

Chk1 and Chk2, the two (unrelated) key effectors of the DNA damage and replication checkpoints, share Cdc25 as a primary target (Fig. 2). This mechanism was first discovered in fission yeast, where Chk1 is essential for reducing the rate of CDK1 Y15 dephosphorylation in response to DNA damage (Rhind and Russell 2000). Chk2 similarly regulates Y15 dephosphorylation in response to replication fork arrest (in mammals this is done by Chk1). Both kinases directly phosphorylate Cdc25, the key effects being inhibition of its protein phosphatase activity and exclusion from the nucleus (Karlsson-Rosenthal and Millar 2006). The latter involves binding of 14-3-3 proteins to Cdc25.

9. THE MAPK-DEPENDENT STRESS CHECKPOINT

In addition to DNA damage, a variety of other cellular stresses, such as osmotic shock, oxidative stress, and microtubule depolymerization, can delay cells in G2 phase. Many of these stresses activate the p38 MAPK pathway (Davis 2012; Morrison 2012). Although the details are not well understood, p38 MAPK is believed to inactivate Cdc25 (Shiozaki and Russell 1995; Karlsson-Rosenthal and Millar 2006).

10. RESOLUTION OF G2 CHECKPOINT ARRESTS

G2 checkpoint arrests are generally reversible. Once the problem is resolved, the checkpoint is inactivated and the cell can proceed with mitosis. This inactivation appears to be a passive process in which resolution of the checkpoint-initiating event (e.g., the repair of the DNA damage) removes the signal that activates the checkpoint kinases, allowing constitutive phosphatases to remove the phosphates from their substrates and reset the system. In situations in which the problem cannot be resolved, cells can adapt to the checkpoint signal and enter mitosis in the presence of ongoing checkpoint signaling. Such adaptation generally appears to be a failure of checkpoint function, as opposed to a regulated attenuation of checkpoint signaling (Harrison and Haber 2006). As cells continue to grow during a checkpoint arrest, it is likely that the mitosis-promoting activities that link the onset of mitosis to attainment of a sufficient cell size eventually overcome mitosis-inhibiting activity of the checkpoint. Alternatively, in metazoans, transcriptional circuits involving p53 and p21 can be activated to drive cells into senescence or apoptosis in response to prolonged G2 arrest (Kastan and Bartek 2004). A failure to activate these circuits can lead to premature resumption of cell division in the presence of DNA damage (Bunz et al. 1998).

11. REGULATION OF MITOSIS

If the G2 checkpoints are not triggered, cells fully activate CDK1 and proceed through the G2/M transition into mitosis. This decision to commit to cell division is implemented by signaling pathways that regulate the various processes of cell division, in particular mitosis and cytokinesis. They also regulate organelles and other cytoplasmic components in ways that are less well understood.

The transition from G2 phase to mitosis involves reorganization of the nucleus, the condensation of the chromosomes, and the formation of the mitotic spindle (see Box 1 and Fig. 3). These events are triggered by CDK1 and culminate with the mitotic chromosomes aligned on the metaphase plate (Morgan 2006). How it coordinates these processes is a long-standing question in the field.

12. DISSECTING THE NETWORK

The challenge of dissecting this regulatory network is twofold. First, CDK1 has dozens, if not hundreds, of mitotic substrates, many of which are not essential for mitosis. This apparent redundancy is presumably because the transition to metaphase is driven by many phosphorylation events, each one of which slightly biases a single protein toward its mitotic state, but no one of which is necessary or sufficient to drive mitosis, per se. This complexity makes validating the importance of any single phosphorylation difficult.

Second, CDK1 activates a network of other kinases that are involved in various steps in the G2/M transition, and some of these feed back to ensure full activation of CDK1 (Fig. 3). The major mitotic kinases are the Polo and Aurora families (Ma and Poon 2011). In addition to functioning in feedback loops that stimulate CDK1, Plk1 is essential for executing mitotic progression, especially centrosome separation and formation of a bipolar spindle (Archambault and Glover 2009). Plk1 phosphorylates numerous substrates at the centrosome and kinetochore complexes that link them to spindle microtubules, many of which are also CDK1 substrates. Furthermore, CDK1 primes Plk1 substrates. The Polo-box domain of Plk1 binds to phospho-S/TP sites (the preferred CDK1 phosphorylation site), and therefore CDK1 phosphorylation of a protein can increase its affinity for Plk1, priming it for Plk1 phosphorylation.

The Aurora A and Aurora B kinases are also essential for mitosis. Aurora A acts earlier in CDK1 activation, centromere separation, and spindle formation. Aurora B acts later to detect and correct improperly attached kinetochores (see below). In addition to Plk1 and Aurora kinases, the NIMA and Greatwall families of kinases are involved in executing the G2/M transition and in the feedback loops that ensure and maintain the full activation of CDK1.

13. ANAPHASE ENTRY

Once chromosomes are properly aligned on the metaphase plate, anaphase is triggered via the activation of the APC (Fig. 3) (Morgan 2006). The APC is a large multisubunit E3 ubiquitin ligase that regulates the stability of a range of mitotic proteins by targeting them for ubiquitin-dependent proteolysis. It is regulated by the binding of one of two substrate-selectivity subunits: Cdc20 at the metaphase/anaphase transition and Cdh1 during telophase and into G1 phase (Pesin and Orr-Weaver 2008). APC–Cdc20 promotes the degradation of several key substrates that trigger the irreversible transition from metaphase to anaphase and the subsequent exit from mitosis. One key substrate is securin, a stoichiometric inhibitor of separase, the protease that cleaves the cohesin complexes that hold sister chromatids together during metaphase. Another key substrate is cyclin B, degradation of which inactivates CDK1. In addition, APC activation leads to the indirect activation of the Cdc14 phosphatase, which dephosphorylates many CDK1 targets, further enforcing the inactivation of CDK1 (Clifford et al. 2008). PP1 and PP2A also play a role (Wurzenberger and Gerlich 2011).

In yeasts, there appears to be an intrinsic delay between the activation of CDK1 and the activation of APC, which usually gives the cells enough time to set up the metaphase plate. APC is activated by CDK1 phosphorylation (Kraft et al. 2003). The lag may give chromosomes adequate time to align. If not, a checkpoint signal (described below) prevents APC activation until the problem is resolved. In most metazoans, metaphase is rarely achieved in time, presumably because the metazoan spindle is larger and more complicated, and the checkpoint is activated in most cell cycles to delay anaphase until the chromosomes are properly aligned. In effect, the regulation of anaphase has gone from being a quality-control checkpoint in yeast to being a central signaling pathway in metazoans that triggers anaphase upon the successful completion of metaphase.

14. CHECKPOINT REGULATION OF MITOSIS

The signaling pathway that delays initiation of anaphase until the successful alignment of metaphase chromosomes is called the spindle assembly checkpoint or mitotic checkpoint (Fig. 3). The checkpoint acts to prevent action of APC–Cdc20, thus preventing the degradation of securin and cyclin B (Musacchio and Salmon 2007). While securin remains stable, sister-chromatid cohesion is maintained, preventing separation of chromosomes; while cyclin B remains stable, CDK1 stays active, maintaining the mitotic phosphorylation state of the nucleus and preventing mitotic exit.

The spindle assembly checkpoint monitors chromosome alignment during the stages leading up to metaphase. In so-called bi-oriented attachment, the two sister chromatids are attached to spindle microtubules emanating from opposite poles; the checkpoint is triggered by sister-chromatid pairs lacking such attachments. The identity of the trigger that senses this lack of bi-orientation and activates the checkpoint has been controversial. Early experiments showed that a single unattached chromosome can activate the checkpoint, and that pulling on such a chromosome with a glass needle is sufficient to stop the checkpoint signal (Li and Nicklas 1995; Rieder et al. 1995). The interpretation of this result was that tension across the kinetochore is sufficient to stop activation of the checkpoint and thus that it monitors kinetochore tension. However, subsequent data were difficult to reconcile with the tension model and instead supported an occupancy model in which the checkpoint monitors proper binding of microtubules to the kinetochore, independently of the tension they generate (Khodjakov and Pines 2010).

These two models can be reconciled by the existence of an Aurora-B-dependent mechanism that recognizes chromosomes that are not bi-oriented (including those with kinetochores lacking tension) and destabilizes their kinetochore–microtubule attachments, allowing proper attachments to reform (Tanaka et al. 2002). In such a model, during reattachment cycles, lack of occupancy at the kinetochore activates the checkpoint. The checkpoint thus monitors kinetochore occupancy, but stable occupancy requires kinetochore tension (Khodjakov and Pines 2010).

A single unattached kinetochore can activate the spindle assembly checkpoint and inhibit all of the APC–Cdc20 in the cell, which suggests that unattached kinetochores produce a diffusible inhibitor of APC–Cdc20. That diffusible inhibitor is believed to be the mitotic checkpoint complex (MCC), which contains three checkpoint proteins—Mad2, Mad3/BubR1, and Bub3—as well as Cdc20 itself (Musacchio and Salmon 2007). The components of MCC interact dynamically with unattached kinetochores in a manner dependent on the more stable interaction of other checkpoint proteins, including Mad1 and Bub1. A crucial step in MCC formation is believed to be the conversion of the soluble open form of Mad2 to a closed form, which can bind stably to Cdc20. This conformational conversion of Mad2 is catalyzed by binding to Mad1 at unattached kinetochores and facilitates the assembly of MCC, which is then released to bind to and inhibit the APC. In metazoans MCC formation is inhibited in the absence of unattached kinetochores by p31-commet, which is believed to prevent open-Mad2 from binding to closed-Mad2 (Musacchio and Salmon 2007). The MCC-dependent checkpoint signal prevents anaphase until all kinetochores are properly attached. Once all kinetochores are occupied, the production of MCC ceases, the MCC-dependent checkpoint inhibition of the APC is relieved, and anaphase ensues.

Another signaling pathway that can block the metaphase/anaphase transition is the DNA damage checkpoint. In most organisms studied, the damage checkpoint arrests cells in G2 phase by preventing the activation of CDK1, as described above. However, in budding yeast the DNA damage checkpoint directly targets the metaphase/anaphase transition by preventing separase activity via Chk1 phosphorylation and stabilization of securin (Sanchez et al. 1999). Metazoan cells also appear to regulate metaphase progression in response to DNA damage (Rieder 2011). However, because these cells also display robust G2 damage arrest, the significance of the response is less clear.

Following the completion of anaphase, the CDK1 phosphorylation events that established the mitotic state are reversed and the cell returns to an interphase state in a process known as mitotic exit (Clifford et al. 2008; Rieder 2011; Wurzenberger and Gerlich 2011). CDK1 activity is reduced by APC-mediated proteolysis of cyclin B. In addition, the rate of CDK1 substrate dephosphorylation is increased by the activation of phosphatases that antagonize CDK1 phosphorylation. In yeast, Cdc14 is activated to reverse Cdk1-catalyzed phosphorylations, as described below. In animal cells, PP1 and PP2A appear to be the major phosphatases antagonizing CDK phosphorylation (Wurzenberger and Gerlich 2011). One mechanism for activation of PP2A-B55 at mitotic exit is the inactivation of Greatwall, but the discovery of other regulatory loops involving phosphatases seems likely. The process of mitotic exit is intimately connected to the final stage of cell division, cytokinesis.

15. REGULATION OF CYTOKINESIS

The proper coordination of cytokinesis with mitosis is essential to ensure faithful chromosome segregation and avoid aneuploidy or polyploidy. This coordination requires the septation initiation network (SIN) in fission yeast and the mitotic exit network (MEN) in budding yeast (McCollum and Gould 2001; Goyal et al. 2011; Meitinger et al. 2012). Cytokinesis is entrained to mitosis; thus, the signaling pathways that regulate cytokinesis are not involved so much in deciding when cytokinesis should occur as in coordinating cytokinesis with mitosis and providing opportunities for checkpoint regulation.

The MEN and SIN pathways are GTPase-triggered kinase cascades that culminate in the activation of the Sid2 kinase in fission yeast and the Dbf2 kinase in budding yeast (Goyal et al. 2011; Meitinger et al. 2012). Components of the MEN/SIN pathways are organized on the spindle-pole body, the yeast equivalent of the centrosome, making it a nexus for signaling pathways that control cell division. Sid2 is necessary and sufficient for initiating cytokinesis in fission yeast, although its exact targets are not known. Therefore, it is essential to restrain SIN signaling until after the successful completion of anaphase. Initiation of SIN signaling by the Spg1 GTPase is inhibited by the GTPase-activating protein Cdc16. Full activation of SIN signaling is antagonized by CDK1 activity (Guertin et al. 2000), which prevents cytokinesis until after activation of the APC and ensures that activation of the spindle assembly checkpoint will also delay cell division. In budding yeast, Dbf2 regulates cytokinesis by promoting localization of the chitin synthase Chs2 and the cytokinesis regulator Hof1 to the bud neck; this activity is antagonized by CDK1 activity (Meitinger et al. 2012).

In return, the MEN/SIN pathways antagonize the activity of CDK1 (Goyal et al. 2011; Meitinger et al. 2012). This reciprocal regulation allows cytokinesis errors, which prolong SIN signaling, to restrain CDK1 activity in the subsequent cell cycle, thus arresting cells in G2 phase and preventing the next mitosis until the previous cytokinesis is successfully completed. An important component of the MEN/SIN pathways is the Cdc14 (Clp1 in fission yeast) phosphatase (Clifford et al. 2008). Cdc14 directly dephosphorylates CDK1 targets, facilitating mitotic exit and resetting the cell to an interphase state at the beginning of G1 phase.

Many proteins in the MEN/SIN pathways are conserved in metazoans. In particular, the LATS kinases, relatives of Sid2/Dbf2 that also function in the Hippo pathway (Harvey and Hariharan 2012), appear to have roles in the regulation of cytokinesis (Yang et al. 2004). Although the details have yet to be established, similar signaling pathways probably coordinate mitosis and cytokinesis in metazoans.

In addition to coordinating the timing of cytokinesis, signaling during cell division is required to determine the location of cytokinesis. Although the details differ between organisms, the location and orientation of the cytokinesis furrow is generally determined by the mitotic spindle, except in fission yeast, in which the cleavage plane is determined directly by the location of the nucleus (Almonacid and Paoletti 2010). How the signal is transmitted from the spindle or the nucleus to the cortex to establish the site of cytokinesis has yet to be established. Budding yeast is unusual in this context because the cleavage plane (the bud neck) is established before mitosis. Therefore, instead of using the spindle to orient cell division, budding yeast uses cell division to orient the spindle. Specifically, the MEN GTPase Tem1 is localized to the spindle-pole body, whereas Tem1’s inhibitors, Bub2, Bfa1, and Kin4, are localized to the mother cell, and its activator Lte1, a guanine nucleotide exchange factor (GEF) relative, is localized to the daughter cell (Bardin et al. 2000). Thus, the MEN is only activated once the spindle is oriented such that one end of the spindle is through the bud neck and in the daughter cell. However, this strategy of triggering cytokinesis as a spatial consequence of spindle elongation may be general, as a similar mechanism functions in fission yeast (Garcia-Cortes and McCollum 2009).

16. CONCLUDING REMARKS

The major cell-cycle transitions that constitute cell division—the G2/M transition, the metaphase/anaphase transition, and cytokinesis—provide important decision points that are regulated by a number of signaling pathways. These pathways ensure that the critical events of cell division occur in the proper order and provide the quality controls that prevent cells from dividing with damaged DNA or misaligned chromosomes. As such, they are instrumental in maintaining genomic integrity and, in metazoans, preventing cancer.

These negative regulatory signaling pathways were the original inspiration for the checkpoint paradigm of active negative regulation of cell-cycle events (Hartwell and Weinert 1989). The initial model posited that checkpoints delayed cell-cycle transitions to allow time for checkpoint-independent processes to fix whatever problem had triggered the checkpoint. Since then, these same pathways have been shown to regulate many other aspects of cell metabolism, such as DNA repair, and the term “checkpoint” is now used much more broadly than originally intended. Nonetheless, these signaling pathways serve as prime examples of how cells reorganize their metabolism and cell cycle to damage and other perturbations.

Although the well-studied signaling pathways that regulate cell division inhibit transitions in response to signals of damage or other problems, there is evidence for at least one positive signaling pathway, the one that regulates cell size. One of the enduring mysteries of cell biology is how cells measure size and how that information is used to regulate cell-cycle transitions such as cell division. Notwithstanding the current lack of mechanistic insight, the way size is measured and the pathways that transduce that signal are poised to be areas of future progress in the field.

The signal transduction pathways that regulate cell division continue to be the focus of significant experimental effort. That effort looks set to continue as work continues on the discovery of new regulators of cell division, the increasingly detailed mechanistic understanding of the major checkpoint pathways, and the translation of our understanding of these pathways into diagnostic and therapeutic advances in fields such as fertility, cancer, and aging.