Cdk1 and transcriptional programs
Unidirectional movement through the cell cycle is critical for cell viability and well-being of the organism; reversal of the direction of the cell cycle can have devastating consequences for the cell, including genome instability. Therefore, cells have developed mechanisms that ensure that the cell cycle is irreversible. One major mechanism that promotes unidirectionality involves regulation of distinct transcriptional programs during the different phases of the cell cycle. Typically, each transcriptional program leads to expression of sets of proteins that carry out processes important for the next phase of the cell cycle, thereby promoting unidirectional movement through the cell cycle. Furthermore, as we will discuss below, feedback mechanisms have evolved that ensure that the cell cycle is irreversible; positive feedback loops make sure that cell cycle entry is robust and switch-like, while negative feedback loops inhibit transcriptional programs to prevent reversal of the cell cycle [
103-
105]. Regulation of the cell cycle's transcriptional programs is highly complex, and here we focus mainly on the Cdk1-dependent aspects of transcriptional regulation (Fig. ; for a recent review see [
106]).
Under physiological conditions, activation of transcription in G1 phase is primarily carried out by Cln3-Cdk1 complexes [
45-
47], although in absence of Cln3, either Cln1 or Cln2 is sufficient to induce Cdk1-dependent transcription [
48-
50]. Approximately 200 genes are specifically expressed in G1, and together they are referred to as the G1 cluster [
107,
108]. Two complexes exist that mediate expression of the G1 cluster: MBF (Mlu1-box binding factor), a complex between Mbp1 and Swi6, which binds promoters harboring the MCB (Mlu1 cell cycle box) promoter element; and SBF, a complex between Swi4 and Swi6, which binds promoters harboring the SCB element (Swi4/6 cell cycle box). Although there is overlap between the classes of genes that are controlled by MBF and SBF, it appears that MBF preferentially induces transcription of genes involved in control or execution of DNA replication and repair (such as
POL2, CDC2, RNR1, CLB5 and
CLB6), while SBF regulates transcription of genes involved in cell cycle progression, cell morphogenesis and spindle pole body duplication (e.g.
CLN1, CLN2, PCL1, PCL2, GIN4, FKS1 and
FKS2) [
106]. Recruitment of RNA polymerase II to the promoter region of these genes depends on Cdk1 activity [
109]. Furthermore, Cln3-Cdk1-induced cell cycle entry is dependent on Swi6 (which is shared by both MBF and SBF and which mediates transcriptional activation) [
110], suggesting that Cdk1 controls SBF/MBF. Indeed, Cdk1 controls SBF/MBF in multiple ways. During early G1, promoter-bound SBF is kept inactive by Whi5 [
111,
112]. In addition, Whi5 recruits the histone deacetylases Hos3 and Rpd3, thus further contributing to repression of transcription of G1 genes [
113,
114]. Efficient cell cycle entry requires phosphorylation of Whi5 by the CDKs Cdk1 and Pho85, which results in dissociation of the SBF-Whi5-Hos3/Rpd3 complex, thereby allowing SBF to activate transcription of its target genes [
111-
114]. In addition to Whi5, Cdk1 may directly control SBF, although mutating the Cdk1 sites in Swi4 and Swi6 had little effect on timing of transcriptional activation [
63,
110,
115] (also see below). However, combined mutation of Cdk1 sites in Whi5 and Swi6 results in cell lethality [
112,
116], indicating that redundancy exists in Cdk1-mediated transcriptional activation of SBF. The mechanism of Cln3-Cdk1-mediated transcriptional activation of MBF remains unknown and may involve a regulatory mechanism similar to Whi5. Interestingly, both MBF and SBF interact with Msa1, and this interaction contributes to proper timing of the G1 transcriptional program [
117].
Importantly, downregulation of Whi5 by Cln3-Cdk1 complexes results in enhanced expression of Cln1 and Cln2. Cln1/2-Cdk1 complexes can also activate SBF/MBF and inhibit Whi5, thus creating a positive feedback loop in which Cln1 and Cln2 boost their own expression, which is important for robust cell cycle entry [
104].
Several mechanisms have been described for switching off the G1 program as the cell enters S phase. For instance, phosphorylation of Msa1 by Cdk1 in its NLS sequence has been reported to result in its exclusion from the nucleus [
118], indicating that Cdk1 may target Msa1 to help shut off the G1 transcriptional program. However, the amplitude of transcriptional activation by SBF and MBF changes little in
msa1Δ mutants [
117], indicating that Msa1 is a relatively minor player in regulation of the G1 transcriptional program, and rather functions to fine-tune the timing of gene expression. Cyclin-Cdk1 complexes may directly target SBF and MBF to shut off the G1 transcriptional program. For instance, Clb6-Cdk1-mediated phosphorylation of Swi6 S160 results in its nuclear export [
63,
64]. However, binding of MBF to promoters is not regulated during the G1-S transition [
103], at which time Clb6 is degraded [
60], indicating that phosphorylation of Swi6 by Clb6-Cdk1 plays a relatively minor role in shutting off the G1 transcriptional program. Cdk1 may also target Swi4 to shut off the G1 program, because Clb2-Cdk1 directly interacts with Swi4 [
119], and this physical interaction inhibits the ability of Swi4 to bind promoters [
115,
120], which may be relevant to prevent expression of the G1 program during the later stages of the cell cycle when Clb2 is present. Stb1 may also be a target of Cdk1 during exit from G1. Stb1 is a protein that interacts with Swi6 to promote the activity of SBF and MBF [
121-
123], and phosphorylation of Stb1 by Cdk1 releases it from promoters, although it is unclear to what extent this contributes to shutting off the G1 program [
121-
123]. The major player in shutting off the G1 program appears to be the transcriptional repressor Nrm1, which binds and inhibits MBF complexes [
103]. Nrm1 acts through negative feedback, since Nrm1 expression is mostly dependent on MBF (although SBF can also activate
NRM1); thus, MBF activity leads to accumulation of Nrm1, which then binds and inhibits MBF to shut off the G1 program as cells enter S phase [
103].
A second transcriptional wave occurs when cells make the transition from G1 to S phase, resulting in expression of genes that make up the two S phase gene clusters, i.e. the histone cluster, consisting of all nine histone genes, and the MET gene cluster. Furthermore, it was recently discovered that a cluster of approximately 180 genes is induced during late S phase, nearly half of which function in chromosome organization and spindle dynamics, but this cluster also contains many genes encoding transcription factors that function later in the cell cycle, such as
FKH1,
FKH2 and
NDD1 (see below) [
124]. This cluster is controlled by the forkhead transcription factor Hcm1 [
124], and here we will refer to it as the Hcm1 cluster. Hcm1 expression itself is cell cycle regulated and peaks in late G1 [
124].
HCM1 expression is probably controlled by SBF and MBF because it has binding sites for both complexes in its promoter [
125]. Hcm1 induces the expression of Fkh1, Fkh2 and Ndd1 [
124], which function in the next stage of the cell cycle, which may contribute to robust cell cycle progression; Hcm1 also induces the expression of Whi5 [
124], which may provide negative feedback to prevent expression of the G1 transcriptional program outside of G1. Interestingly, constitutive expression of
HCM1 from the
GAL1 promoter did not completely abolish the fluctuation in the cell cycle-dependent expression of two Hcm1 targets (
WHI5 and
NDD1), suggesting that in addition to regulating its expression, the cell cycle may also control Hcm1 activity through post-translational modifications [
124]. It is tempting to speculate that Cdk1 is responsible for this regulation, because Hcm1 contains 12 potential Cdk1 sites and it is an efficient target of Clb-Cdk1
in vitro [
126].
From the end of S phase until nuclear division in M phase a set of approximately 35 genes, including
CDC5,
CDC20,
SWI5 and
ACE2, is expressed with similar kinetics as
CLB2, and is therefore referred to as the
CLB2 cluster [
106-
108]. The
CLB2 cluster was found to be controlled by the transcription factor called 'SFF' (SWI Five Factor), the identity of which was later shown to be the partially redundant forkhead transcription factors Fkh1 and Fkh2 [
127-
129]. Simultaneous deletion of
FKH1 and
FKH2 uncouples transcription of the
CLB2 cluster from the cell cycle, showing that Fkh1 and Fkh2 provide the link between the cell cycle and periodic expression of the
CLB2 cluster [
127]. Fkh2 occupies the majority of SFF sites due its interaction with the transcription factor Mcm1, which increases the affinity of Fkh2 for the SFF element about 100-fold, thus outcompeting Fkh1 (which does not interact with Mcm1). Cdk1 controls transcription of the
CLB2 cluster in multiple ways, creating a positive feedback loop in which Clb2 promotes its own synthesis [
119]. For instance, Clb-Cdk1 complexes phosphorylate Fkh2 on S683 and T697 (although additional sites may exist [
130]). In addition, Clb2-Cdk1 phosphorylates residue T319 on the rate-limiting transcriptional transactivator Ndd1 [
131,
132]; Ndd1 activates gene transcription upon recruitment by Fkh2 [
133]. Interestingly, phosphorylation of both Ndd1 and Fkh2 is thought to increase their interaction, thus stimulating transcription. Phosphorylation of Ndd1 on S85 by the polo kinase Cdc5 further enhances its transcriptional activity [
134]. Phosphorylation of proteins by Cdk1 can create a docking site for polo kinases [
135], and it is tempting to speculate that T319 phosphorylation of Ndd1 by Cdk1 serves as a priming site for Cdc5, which subsequently would phosphorylate S85. However, phosphorylation of Ndd1-T319 is not required for phosphorylation of Ndd1-S85 [
134]. Therefore, it remains unknown how Cdc5 is recruited to the Fkh2-Ndd1 complex. The key might be Fkh2, which is required for Cdc5-mediated phosphorylation of Ndd1 and which is also a target of Cdk1 [
130,
134].
Four clusters of genes are expressed between M phase and G1 phase: the MCM cluster, the
SIC1 cluster, the MAT cluster and the PHO regulon [
107,
108]. Expression of MCM cluster genes (including
MCM2-7,
CDC6,
SWI4, and
CLN3) is controlled by the Mcm1 transcription factor, which as mentioned above is also involved in expression of the
CLB2 cluster when it is complexed to Fkh2. However, throughout most of the cell cycle Mcm1 also binds the homeodomain repressors Yox1 and Yhp1, and genes that contain binding sites for Yox1 and Yhp1 in their promoter (the MCM cluster genes) are repressed by the Yox1-Mcm1 and Yhp1-Mcm1 complexes [
136]. Yox1 and Yhp1 are unstable proteins, and Yox1 is expressed in mid-G1 through early S, while Yhp1 is expressed later in the cell cycle [
108,
136]. During M-G1, when both repressors are not expressed, the promoters of the MCM cluster genes are de-repressed and transcription can occur. It is currently unknown whether Cdk1 directly controls the activity of Yox1 and Yhp1, but both proteins (especially Yox1) are efficient targets of Cdk1
in vitro [
126]. Expression of both these proteins fluctuates during the cell cycle [
108,
136], and the promoter regions of both
YOX1 and
YHP1 contain binding sites for SBF/MBF, while the
YHP1 promoter also contains multiple binding sites for Fkh1/2 [
137], suggesting that Yox1 and Yhp1 are at least indirectly controlled by Cdk1.
Expression of the
SIC1 cluster is controlled by the transcription factors Swi5 and Ace2, which bind the same DNA sequences
in vitro with similar affinities and whivh regulate an overlapping set of genes
in vivo [
138,
139]. However, in some cases the two proteins control distinct promoters, e.g. Swi5 activates transcription of the HO endonuclease gene whereas Ace2 does not; conversely, the
CTS1 gene encoding endochitinase is activated by Ace2 and not by Swi5 [
140]. Swi5 is negatively regulated by Cdk1, because Cdk1-mediated phosphorylation of the NLS of Swi5 results in its exclusion from the nucleus [
141,
142]. Presumably, when Cdk1 becomes inactivated at the end of M phase, Swi5 becomes dephosphorylated, allowing it to enter the nucleus and activate transcription of the
SIC1 cluster. Ace2 is also phosphorylated by Cdk1 on multiple residues including in the NLS [
143,
144], and similar to Swi5, phosphorylation of Ace2 by Cdk1 has been suggested to result in its nuclear exclusion [
143,
144].
Asymmetric cell division in budding yeast yields a bigger mother and a smaller daughter, and cell cycle entry is also asymmetric; mothers cells enter the cell cycle faster than daughter cells [
145-
148]. Interestingly, this cell cycle delay in daughter cells may be mediated by Ace2 [
149,
150]. Ace2 localizes to the cytoplasm during most of the cell cycle, presumably due to phosphorylation by Clb3,4-Cdk1 [
143,
144]. When cells exit from mitosis, Ace2 specifically localizes to the nucleus of the daughter cell, and this asymmetric localization of Ace2 requires the activity of the Mob2-Cbk1 kinase complex [
151-
153]. In addition, nuclear localization of Ace2 may require dephosphorylation of its Cdk1 sites [
143,
144], which likely occurs when Cdk1 is downregulated during mitotic exit (see section 'Cdk1 and exit from mitosis'). In the daughter cell, Ace2 represses the transcription of
CLN3, thus providing the daughter cell with the opportunity to properly control its cell size [
149,
150].
The MAT cluster is a set of genes (including
FAR1) normally induced by mating pheromone, but which is also expressed to a certain degree during M-G1 even in absence of pheromone. The rationale for basal expression of the MAT cluster in absence of pheromone could be that cells can respond quickly to arrest the cell cycle and to initiate mating once pheromone is detected. Expression of the MAT cluster depends on the aforementioned Mcm1 as well as the transcription factor Ste12, which binds to pheromone response elements (PREs) in the upstream activating sequences of its target genes [
154-
157]. Cdk1 has a profound effect on restricting the pheromone response (and thereby expression of genes with PRE promoter sequences) to the G1 phase of the cell cycle, which we will discuss later (see section 'Cdk1 restricts pheromone signaling to the G1 phase of the cell cycle').
The PHO regulon is also transcribed at the M-G1 boundary [
107,
108] and includes genes involved in scavenging and transporting phosphate [
158]. The expression of these genes might not necessarily be regulated by the cell cycle, but might rather be a result of depletion of cellular phosphate pools during the metabolic processes associated with cell duplication, thus triggering the phosphate starvation response [
158,
159]. Regardless, it was recently shown that Cdk1 can phosphorylate the transcription factor Pho2 on S230, resulting in increased binding of Pho2 to Pho4 [
160]. The Pho2-Pho4 complex is required for activation of
PHO5, which encodes an acid phosphatase that is secreted into the periplasmic space and scavenges phosphate by working in conjunction with high-affinity phosphate transporters [
161]. Pho2 also associates with the Myb-like transcription factor Bas1 to activate genes in the pyrimidine, purine and histidine biosynthesis pathways [
162]. Therefore, by activating the Pho2-Pho4 complex, Cdk1 may help replenish cellular phosphate pools and stimulate biosynthesis of basic building blocks for the next round of cell division. Pho85 and Cdk1 work together in this process, because upon phosphate starvation Pho85 phosphorylates the NLS of Pho4 resulting in nuclear import of Pho4 [
163].
Several other less well characterized transcription factors exist that show cell cycle-dependent expression and that are efficient targets of Cdk1
in vitro [
126], such as Plm2 (a putative transcription factor that is induced at Start and in response to DNA damage), Tos4 (putative transcription factor similar to Plm2; Tos4 expression peaks in G1) and Pog1 (a putative transcriptional activator that promotes recovery from pheromone-induced cell cycle arrest, presumably by relieving the repression of
CLN1 and
CLN2 [
164]). It will interesting to see how these proteins impact the cell cycle and whether they are controlled by Cdk1.
While Cdk1 regulates many aspects of transcription throughout the cell cycle, there is evidence that transcriptional programs are executed by a free-running oscillator independently of Cdk1 [
22]. Indeed, when Cdk1 was experimentally inactivated upon entry of cells into the cell cycle, about 70% of periodic genes continued to be expressed periodically and on schedule [
165], and therefore Cdk1 is unlikely to be the single determinant of global periodic transcriptional programs; rather, it may fine-tune coordination of the cell cycle with periodic transcription.
Finally, in addition to controlling transcription factors, Cdk1 has also been reported to affect the process of transcription in other ways. For instance, together with Cks1 it recruits the proteasome (which enhances efficient transcription elongation by RNA polymerase II [
166,
167]) to the
GAL1 ORF during galactose-induced transcription of the
GAL1 gene to promote transcription [
168]. Interestingly, this appears to be independent of its kinase activity, suggesting that Cdk1 may function as an adaptor protein [
168]. Cdk1 may also modulate transcription by regulating chromatin modifiers. For example, it was recently suggested that Clb2-Cdk1 is required for NuA4-mediated acetylation of Htz1 on Lys14 [
169], and Cdk1 has been speculated to exert this function through phosphorylation of Yng2 [
169], which is a component of NuA4 required for histone acetyltransferase activity and which may be phosphorylated on Cdk1 sites in vivo [
17]. Cdk1 may also affect histone acetylation by promoting dissociation of the repressive Sin3 histone deacetylase complex from the
CLB2 promoter, resulting in a local, transient increase in histone H4 acetylation, which facilitates transcription [
170]. The molecular target of Cdk1 in this process is not known, but could be Sin3 itself, because in proteomic studies it has been found to associate with cyclins [
144] and to be phosphorylated on Cdk1 sites
in vivo [
17,
171].
Cdk1 restricts pheromone signaling to the G1 phase of the cell cycle
The
S. cerevisiae pheromone signaling pathway is one of the best understood signaling pathways in eukaryotes (for a review see [
251]). While it is believed that most essential pathway components have been identified [
251], the modulation of the activity and specificity of these components during the cell cycle and during mating is less well understood; however, recent studies have identified an important role for Cdk1, which we will discuss in this section (see Fig. ).
The pheromone response is triggered by binding of mating pheromone to the seven-transmembrane, heterotrimeric G-protein-coupled receptor (Ste2 in
MATa cells and Ste3 in MATα cells) located on the cell surface. This induces a conformational change of the receptor, leading to GDP-to-GTP exchange by the associated G
α subunit Gpa1, thus releasing the Ste4-Ste18 complex (the G
βγ component of the heterotrimeric G protein) [
252-
257]. The Ste4-Ste18 complex, which is bound to the cell membrane because Ste18 is farnesylated and palmitoylated, recruits three effectors: (i) the Far1-Cdc24 complex, (ii) the Ste20 protein kinase, and (iii) the Ste5-Ste11 complex. Recruitment of the Far1-Cdc24 complex from the nucleus to the cell membrane results in localized activation of Cdc42 [
258,
259], which in turn binds and activates the PAK-like kinase Ste20 [
260,
261], which is membrane-bound through its interaction with Ste4-Ste18. Activation of Ste20 then results in reorganization of the actin cytoskeleton in order to form the mating projection (shmoo) that will ultimately fuse the
MATa and
MATα cells to form a diploid cell; reorganization of the actin cytoskeleton and subsequent shmoo growth is not unlike bud morphogenesis (discussed in section 'Cdk1 and cell morphogenesis') and makes use of similar mechanisms and components [
215]. Finally, the Ste4-Ste18 complex recruits Ste5, which serves as an adaptor for the kinases Ste11 (MEKK), Ste7 (MEK) and Fus3 (MAPK). Recruitment of the Ste5 complex brings Ste11 in close proximity to Ste20, which phosphorylates and activates it [
262,
263]. Ste11 in turn phosphorylates Ste7, which then phosphorylates the MAP kinases Fus3 and Kss1. Both MAPKs then phosphorylate the transcription factor Ste12, which induces expression of mating type specific genes that either have a positive feedback effect (
STE2,
FUS3,
FAR1) or a negative feedback effect (
SST2,
MSG5,
GPA1), probably to fine-tune the pheromone response. Ste12 also activates genes involved in the process of cell fusion (e.g.
FUS1,
FUS2,
FIG1,
FIG2,
AGA1). Targets of Fus3 include Bni1, a formin homologue the phosphorylation of which is required for actin polarization towards the site of shmoo growth [
264]; Sst2, which is involved in a negative feedback loop that attenuates pheromone signaling [
265]; and Tec1, which binds Ste12 to express genes required for cell differentiation, and phosphorylation by Fus3 targets it for SCF-mediated degradation, thus shifting the spectrum of Ste12-induced gene expression from differentiation genes towards pheromone response genes [
266,
267]. A key substrate of Fus3 is Far1, and phosphorylation of Far1 on T306 is essential for cell cycle arrest by inhibiting Cln-Cdk1 complexes [
74]. It is not entirely clear how phosphorylated Far1 inhibits Cdk1 signaling, because one study found that Far1 inhibits Cln-Cdk1 kinase activity [
69], while another study found that Cln-Cdk1 retains kinase activity in presence of Far1
in vitro [
74]. One mechanism for cell cycle arrest could be that Far1 blocks access of Cln-Cdk1 to at least some of its substrates, thus inhibiting cell cycle progression.
Mating of cells should only occur during G1 phase, because this is the only period in the cell cycle when cells have a single copy of their genome (1
n). Mating outside G1 would result in aneuploid cells with > 2
n DNA content, which could lead to genome instability. Cdk1 is inactive during G1 phase and this permits pheromone signaling and cell mating, while outside of G1 Cdk1 is active and inhibits the mating pathway (Fig. and ). One indication for a role for Cdk1 in regulating the pheromone response comes from the observation that in
fus3 deletion mutants the polarized localization of Bni1, Ste20 and Ste5 upon pheromone treatment is abrogated, but this polarized localization is restored upon inhibition of Cln-Cdk1 activity, suggesting that Cdk1 negatively affects pheromone-induced polarization of cells [
268]. One molecular target of Cdk1 in the negative regulation of pheromone signaling could be Ste20, which can be directly phosphorylated by Cln2-Cdk1
in vitro [
269,
270]. This is supported by the finding that mutation of all of the phosphorylation sites in Ste20 (Cdk1 consensus sites as well as non-Cdk1 sites) resulted in hypersensitivity of cells to pheromone, indicating that, under physiological levels of Cdk1 activity, phosphorylation of Ste20 negatively affects pheromone signaling [
271]. However, overexpression of
CLN2 was still able to overcome pheromone arrest in this
ste20 phospho-site mutant [
271], and therefore an additional target of Cdk1 must exist. Based on genetic data, Ste11 may also be a potential target of Cln-Cdk1 to suppress pheromone signaling [
272], but it has not been demonstrated that Cdk1 actually phosphorylates Ste11. More recently, Ste5 was identified as a target of Cdk1 [
273]; Cln-Cdk1 phosphorylates Ste5 on multiple residues flanking a membrane binding domain [
274], which blocks membrane localization of Ste5 and its associated proteins Ste11, Ste7 and Fus3, resulting in inhibition of pheromone signaling. Furthermore, phosphoryation of Ste5 may target it for degradation by the SCF [
275], further contributing to inactivation of the pheromone response pathway. It is not known whether Cdk1 phosphorylates Ste12; Ste12 controls the transcriptional program that is required for pheromone-induced cell cycle arrest and mating, and in absence of pheromone Cdk1 might be expected to inhibit Ste12 to prevent illicit expression of genes that mediate cell cycle arrest mating. Finally, Cln-Cdk1-mediated phosphorylation of the CKI Far1 on S87 targets it for degradation [
74]. Presumably, destruction of Far1 results in more active Cln-Cdk1 complexes, which in a feedback loop will phosphorylate and destroy more Far1, resulting in cell cycle entry and closure of the window of opportunity for cell mating.
Cdk1 and exit from mitosis
The final steps of mitosis encompass an ordered series of events referred to as mitotic exit, which mediates the inactivation of Cdk1 and the dephosphorylation of key Cdk1 targets to reset the cell cycle (Fig. , for recent reviews see [
400-
403]). It starts with the separation of sister chromatids during anaphase upon Esp1-mediated loss of chromosome cohesion and involves elongation of the mitotic spindle. Once chromosome segregation is complete, the cytokinetic furrow is formed at the future site of cell division, the spindle disassembles, and cell division is completed by cytokinesis and abscission. During the past decade, tremendous progression has been made towards unraveling the molecular mechanisms that mediate mitotic exit, although it should be emphasized that the picture is far from complete. Here we focus mostly on the function of Cdk1 in mitotic exit.
Anaphase is triggered by ubiquitination and thereby proteasomal degradation of Pds1 (securin) by the APC, relieving inhibition of Esp1/separase, which subsequently cleaves the cohesion complex that holds together the sister chromatids. Simultaneously, the APC targets mitotic cyclins for destruction, leading to downregulation of mitotic Cdk1 activity, and destruction of Clb2 is particularly important for mitotic exit [
404-
406]. Further inhibition of Cdk1 activity is mediated by expression of the Cdk1 inhibitor Sic1, which occurs at the M-G1 boundary [
404,
406,
407], and a feedback loop involving Sic1 ensures that mitotic exit is irreversible by preventing re-synthesis of mitotic cyclins [
408]. In addition, Cdc6 has been reported to have a similar function in inactivation of Cdk1 by directly binding and inhibiting Clb-Cdk1 complexes [
316,
317]. However, Cdc6 may modulate mitotic exit at least in part through a Cdk1-independent mechanism by affecting the activity of the APC [
314,
315], and in addition Cdc6 may be less important for mitotic exit [
316] than previously reported [
317]. Finally, the phosphatase Cdc14 reverses phosphorylation of Cdk1 targets to reset the cell cycle to a basic G1 state; the activity of Cdc14 is paramount to mitotic exit [
402,
403], and in absence of Cdc14 activity cells arrest before cytokinesis in a telophase-like state with long spindles and a divided nucleus [
409,
410].
Cdk1 induces mitotic exit - and thus its own inactivation - by affecting the activity of the APC. APC activity fluctuates throughout the cell cycle in response to differential association with the activating subunits Cdc20 and Cdh1 (Hct1): during mid-mitosis it associates with Cdc20, leading to the initiation of anaphase, whereas during late mitosis it associates with Cdh1, and the APC
Cdh1 complex stays active throughout the subsequent G1 [
411]. APC
Cdc20 and APC
Cdh1 have different substrate specificity; e.g. APC
Cdc20 targets Pds1 and APC
Cdh1 targets Ase1, while both APC
Cdc20 and APC
Cdh1 are required for full degradation of Clb2 [
405,
406,
412]. There is extensive interplay between Cdk1 and APC activity; APC
Cdh1 degrades mitotic cyclins to inhibit Cdk1 activity [
360,
413,
414], but upon entry of cells into S phase Cln1,2-Cdk1 and Clb5-Cdk1 phosphorylate Cdh1, blocking its interaction with the APC and thus allowing mitotic cyclins to build up again later in the cell cycle [
413-
415]. The interaction between Cdh1 and the APC is further inhibited by Cdk1-mediated phosphorylation and stabilization of Acm1, which inhibits Cdh1 by acting as a pseudosubstrate inhibitor [
416-
418]. Then at the end of mitosis Cdc14 dephosphorylates Cdh1, allowing it to interact with the APC again to destroy mitotic cyclins, thus completing the cycle [
414,
419]. Cdk1 is also required for activation of APC
Cdc20 during mitosis [
420,
421], which initiates the metaphase to anaphase transition by degrading Pds1 [
406,
422]. Cdk1 activates APC
Cdc20 by phosphorylating three components of the APC, Cdc16, Cdc23 and Cdc27, resulting in binding of Cdc20 to the APC [
421]. Activation of APC
Cdc20 results in degradation of Pds1, leading to activation of Esp1 and thereby dissolution of chromosome cohesion, but it also leads to activation of the so-called FEAR (Cdc fourteen early anaphase release) network which results in transient activation of the phosphatase Cdc14 [
422-
424] (Fig. ). Activation of the FEAR network is followed by activation of the mitotic exit network (MEN), which promotes sustained Cdc14 activity [
402].
During most of the cell cycle, Cdc14 is sequestered in the nucleolus by Net1 (also known as Cfi1) [
425,
426]. The FEAR pathway is triggered by Esp1/separase-induced downregulation of the phosphatase PP2A
Cdc55, which is apparently independent of the proteolytic function of Esp1 [
427,
428]. PP2A
Cdc55 keeps Net1 in a hypophosphorylated state, which promotes the interaction between Net1 and Cdc14 [
428,
429]. When Pds1 becomes degraded in early anaphase, Esp1 downregulates PP2A
Cdc55, resulting in a shift in the phosphorylation balance of Net1 to a hyperphosphorylated state due to the action of Clb1,2-Cdk1 and Cdc5 [
430-
432]. Phosphorylation of Net1 abrogates the interaction with Cdc14 [
430-
432], which is then released from the nucleolus into the nucleus and cytoplasm to dephosphorylate Cdk1 targets. The FEAR network also encompasses additional proteins, such as the Esp1-associated protein Slk19; Tof2, which bears homology to Net1 [
433]; Fob1, a nucleolar protein that localizes to rDNA and which interacts with Net1; and Spo12. Slk19 is a Cdk1 target, but the relevance of this is not well understood [
387]. Fob1 forms a complex with Net1 and Spo12, and phosphorylation of Spo12 by Cdk1 contributes to activation of the FEAR pathway [
423,
434,
435].
The initial release of Cdc14 is not sufficient for completion of mitotic exit, because when Cdk1 activity starts to drop during anaphase, Net1 could become hypophosphorylated again, which would then result in premature return of Cdc14 to the nucleolus before mitotic exit has been completed [
423,
428]. To circumvent this problem, cells activate the MEN to ensure sustained Cdc14 activity during late anaphase. The MEN pathway integrates information from the mitotic spindle with cell cycle progression [
436,
437]. A central component of the MEN is a small Ras-like GTPase named Tem1, which localizes to the daughter-bound SPB [
436-
438]. The MEN is thought to be activated when the daughter-bound SPB moves into the bud, which is the compartment where Lte1 is located, a protein with similarity to GTP-exchange factors that localizes only to the bud and which may induce the activity of Tem1 [
436,
437,
439]. Lte1 may not directly activate Tem1, but rather indirectly activates Tem1 by inhibiting Bfa1, which is an inhibitor of Tem1 [
440] (also see below); the asymmetric localization of Lte1 to the bud cortex is mediated by Cdk1 and Cla4 [
441-
443]. Active Tem1 then activates a signaling cascade by interacting with the kinase Cdc15, which in turn activates the Mob1-Dbf2 kinase complex [
444-
448]. Exactly how Mob1-Dbf2 then promotes Cdc14 release from the nucleolus is not well understood [
403], but it involves direct phosphorylation of Cdc14 on serine and threonine residues adjacent to a nuclear localization signal (NLS), thereby abrogating its NLS activity resulting in nuclear exclusion [
449]. This then promotes mitotic exit.
It is important that the MEN pathway is not activated before chromosome separation is complete, as this could result in missegregation of chromosomes. Premature activation of the MEN pathway is prevented by multiple means. Tem1 is kept inactive at the SPB by the GAP Bub2-Bfa1. Like Net1, Bfa1 is kept in a hypophosphorylated state by PP2A
Cdc55 during metaphase, but when PP2A
Cdc55 is downregulated by Esp1 during early anaphase, the balance shifts towards hyperphosphorylated Bfa1, which is mediated by Cdc5. Phosphorylation of Bfa1 inhibits its activity and therefore results in activation of Tem1 and hence mitotic exit. Bfa1 is also regulated by the spindle positioning checkpoint (SPOC), which delays mitotic exit when the anaphase spindle fails to extend toward the mother-daughter axis [
450]. When the mitotic spindle is misaligned, the kinase Kin4 phosphorylates Bfa1, which prevents phosphorylation and inhibition of Bfa1 by Cdc5 [
451-
453]. Because Cdc5 cannot phosphorylate and inhibit Bfa1, Bfa1 continues to block Tem1 activity, thereby preventing mitotic exit. However, during an unperturbed cell cycle - when the spindle is properly aligned - Kin4 localizes to the mother SPB, while Bfa1 localizes to the daughter SPB; as a result of this differential localization, Kin4 cannot phosphorylate Bfa1, which then becomes phosphorylated by Cdc5 instead, leading to inhibition of Bfa1, activation of Tem1, and mitotic exit [
453]. Interestingly, the asymmetric localization of Bfa1 was recently reported to be promoted by Lte1 [
440]. Thus, Lte1 may activate Tem1 indirectly by inhibiting Bfa1 rather than directly through its GEF domain [
440].
Full activation of the MEN pathway requires Cdc14-mediated dephosphorylation of Cdc15 and Mob1, both of which are targets of Cdk1 [
454,
455]. Phosphorylation of Cdc15 and Mob1 is inhibitory, and their dephosphorylation by Cdc14 may contribute to fine-tuning of MEN activity, but it may also ensure a right order of events, such that MEN does not take place before activation of the FEAR network (which releases Cdc14 which can then dephosphorylate Cdc15). Ultimately, when cells have successfully exited from mitosis, Cdc14 is downregulated by its return to the nucleolus, which is mediated by degradation of Cdc5 by the APC [
456], which results in a shift in the phosphorylation balance of Net1 and Bfa1 to a hypophosphorylated state.
Future directions and ramifications for cancer treatment
Many aspects of the cell cycle are directly controlled by Cdk1, and include regulation of cell polarity and morphology, DNA replication, chromosome segregation, and maintenance of genome stability. Many, if not all, facets of Cdk1 regulation involve positive and negative feedback loops, reflecting the need for tight control of the cell cycle. This is especially evident in regulation of processes that affect genome stability, because both an aberrant increase as well as a decrease in Cdk1 activity can lead to genome instability, with potentially disastrous consequences for the organism. While regulation of Cdk1 activity is relatively well understood, comparatively little is known about its downstream targets. As discussed in this review, approximately 75 Cdk1 targets have been described in
S. cerevisiae (See additional Table
1), but regarding the enormous complexity of cell duplication, we expect many more to be identified. While the use of classic yeast genetics has been useful in the discovery of upstream regulators of Cdk1, such as cyclins and Cak1, downstream components are rarely identified in suppressor screens, probably because Cdk1 activity is required for several essential cellular processes throughout the cell cycle, and no single Cdk1 target can compensate for loss of Cdk1 activity during all these different steps. More advanced genetic screens may be required to unravel the complete genetic network of the cell cycle that involve
CDK1, like e.g. synthetic genetic array (SGA) and synthetic dosage lethality (SDL) screens, which have been successful in identification of novel processes and targets controlled by the related CDK Pho85 [
191,
565-
568]. Furthermore, in a recent study, which combined specific chemical inhibition of Cdk1 with quantitative mass spectrometry, 308 potential Cdk1 substrates were identified [
17], many of which had previously been shown to be
bona fide Cdk1 targets. The functional consequences of phosphorylation of the vast majority of these potential Cdk1 substrates still needs to be determined.
Complexity to Cdk1 signaling is added by the fact that multiple enzymes can recognize Cdk1 phosphorylation sites to further modify those proteins; e.g. the proline isomerase Ess1/Pin1 can be recruited to phosphorylated SP/TP sites (potentially phosphorylated by Cdk1) to isomerize the proline residue, and this has been shown to affect diverse cellular processes, including growth factor-induced signal transduction pathways, cell-cycle progression, cellular stress responses, neuronal function and immune responses [
569]. Additionally, phosphorylation by Cdk1 can serve as a priming site for further phosphorylation by other kinases, such as the polo kinase Cdc5 [
135]. Furthermore, there exists extensive cross-talk between Cdk1 and Pho85 [
8]. Potential cross-talk between Cdk1 and the other CDKs (Ssn3, Kin28, Bur1 and Ctk1) remains largely unexplored, although cross-talk might be expected based on the fact that Cdk1 and the other CDKs all control various facets of transcription. Other aspects of Cdk1 signaling have remained obscure, e.g. Cdk1 has a kinase-independent role in regulation of transcription, but little more is known about this process than recruitment of the proteasome [
168], and it is not known whether Cdk1 (or its scantily studied interaction partner Cks1) has adaptor functions in other processes as well.
Because considerable attention has been focused on the function of Cdk1 in duplication of the genome (DNA replication, repair and chromosome segregation), the involvement of Cdk1 in other processes associated with the cell cycle is not as well studied, like for instance cell metabolism. When the cell enters the cell cycle, enormous changes take place in catabolic and anabolic processes to facilitate duplication of the genome and biosynthesis of cellular structures and organelles, and therefore one might expect Cdk1 to have a direct role in controlling enzymes required for biosynthesis. However, apart from a few Cdk1 targets, such as Tgl4 and Smp2, which are involved in fatty acid synthesis, and the transcription factor Pho2, which stimulates the expression of genes involved in purine and histidine biosynthesis pathways, little is known about the role of Cdk1 in cell metabolism. It seems likely that additional targets of Cdk1 exist that control metabolic pathways.
Finally, an important aspect of CDKs is their involvement in tumor growth. Like in
S. cerevisiae, a single CDK (Cdk1, also known as Cdc2) is sufficient to drive the cell cycle in higher eukaryotes, but additional CDKs (Cdk2,4,6) are required for proliferation of specialized tissues and development of the organism [
28,
570,
571]. While CDKs are crucial for growth and development of all eukaryotes, the aberrant activity of these CDKs is well known to underlie tumor growth [
28]. Numerous studies have shown that tumor cells evade antigrowth signals. One key inhibitor of the cell cycle is p53, which blocks the cell cycle by inhibiting CDK activity in several ways, one of which is inducing the transcription of p21 [
572,
573], which binds and inactivates cyclin-CDK complexes. Both p53 and p21 are frequently mutated in human cancers [
574], as well as other CKIs such as p16 and p27 [
28], and most human tumors aberrantly express cyclin D and cyclin E [
28], underscoring the importance of proper control of CDK activity. It is becoming clear that CDKs play an important role in the DNA damage response in
S. cerevisiae as well as mammalian cells, and treatment of cells with DNA damaging agents while simultaneously inhibiting Cdk1 activity results in extreme cell toxicity in
S. cerevisiae and human cells [
325,
515,
518,
575,
576]. Currently, several combination therapies are in clinical trial as cancer chemotherapy [
577]. The vast majority of current chemotherapies are based on drugs that induce DNA damage or that inhibit mitosis by targeting microtubules, and these therapies frequently result in serious side effects such as mucositis and myelosuppression, and increase the risk of secondary neoplasms. We believe that unraveling the genetic network of
CDK1 (i.e. the network of genes that become essential under conditions of reduced Cdk1 activity) might identify novel pathways that can be targeted by combination therapy with CDK inhibitors to induce synthetic lethality of cancer cells, thus contributing to more personalized, less toxic and more efficacious chemotherapy.