Introduction

The progression of the cell cycle is tightly regulated in order to maintain genetic integrity and to ensure that genetic information is correctly passed on to daughter cells. Surveillance mechanisms, known as checkpoints, have thus been discovered, that monitor the integrity of the cell cycle progression (Elledge, 1996). By definition, cell cycle checkpoints inhibit cell cycle progression until the specific cellular processes under surveillance are completed with high fidelity. In eukaryotes, both DNA replication and chromosome segregation are highly regulated and are monitored at several steps in the cell cycle. A loss of checkpoint function can result in infidelity of DNA replication or of chromosome segregation, and thereby predispose cells to genetic instability.

Reversible phosphorylation is a fundamental molecular mechanism that is critical for regulating cell division. Deregulated phosphorylation of the components important to cell cycle control frequently leads to abnormal cell division and/or transformation. To date, several families of protein kinases such as cyclin-dependent kinases (Cdks), polo-like kinases (Plks), and Aurora family of kinases have been characterized, which are important to cell cycle regulation. Extensive research in the past decade or so has demonstrated that Polo and Plks are involved in regulation of various cell cycle checkpoints that ensure the timing and order of cell cycle events such as DNA repair, bipolar spindle formation, chromosome segregation, and mitotic exit (Nigg, 1998; Dai et al., 2002; Barr et al., 2004).

In this review, we summarize the properties of Polo and Plks in various cell cycle checkpoint controls. We also present new lines of evidence supporting the role of mammalian Plks in regulating both an intra-S-phase checkpoint and a spindle checkpoint.

Plks' role in intra-S-phase checkpoint

Checkpoint activation pathway

During the S phase, the genome is most susceptible to damage caused by environmental stresses and/or internal perturbations. Thus, it is natural that cells have evolved sophisticated mechanisms to monitor DNA damage and initiate repair processes if the damage is not extensive. Extensive research in the past decade has shown that the signaling pathways that underlie the cellular response to DNA damage (or genotoxic stress) consist of sensors, signal transducers, and effectors (Zhou and Elledge, 2000). Although the identities of the damage sensors remain unclear, the molecular entities responsible for transducing the damage signals to specific effectors are relatively well characterized. ATM (ataxia telangiectasia mutated) and its homolog ATR (ATM-related) function early in the signaling pathways and are central to the DNA damage response (Sancar et al., 2004). Downstream mediators/transducers of ATM and ATR include the protein kinases Chk1, Chk2, and likely Plk3 (Xie et al., 2001b; Sancar et al., 2004). The effectors of the checkpoint include p53 and Cdc25 (Sancar et al., 2004).

Both CDC5 in the budding yeast and mammalian Plks are known to participate in the DNA damage response. An early study shows that the electrophoretic mobility of Cdc5 in denaturing gels is affected when the cells are exposed to DNA damage agents, and this modification of Cdc5 is dependent on Mec1, Rad53 (a yeast Chk1 homolog), and Rad9 (Cheng et al., 1998). In addition, a functionally defective Cdc5 mutant protein suppresses a Rad53 checkpoint defect, whereas overexpression of Cdc5 overrides checkpoint-induced cell cycle arrest (Sanchez et al., 1999), suggesting that Cdc5 acts downstream of Rad53. Several recent studies show that Plk1 activity is inhibited upon DNA damage (Smits et al., 2000; van Vugt et al., 2001; Ando et al., 2004). The DNA damage-induced inhibition of Plk1 appears to be mediated by blocking its activation because expression of activation mutants of Plk1 can override the G2/M arrest induced by DNA damage (Smits et al., 2000). Subsequent studies by this group reveal that DNA damage-induced inhibition of Plk1 is at least in part dependent on ATM or ATR because caffeine treatment blocks the inhibition of Plk1 after IR- or UV-induced DNA damage (van Vugt et al., 2001). The activity of p53 is greatly enhanced upon DNA damage checkpoint activation.

A recent study shows that Plk1 binds p53 and inhibits the transactivating activity of this tumor suppressor protein; the kinase activity of Plk1 is required for inhibition of p53 activity; besides, Plk1 also negatively affects the proapoptotic function of p53 in human lung tumor cell lines (Ando et al., 2004). Again, ATM is capable of attenuating Plk1-mediated suppression of p53 activity (Ando et al., 2004). Given the potential role of Plk1 in promoting cell proliferation and tumorigenesis (Dai and Cogswell, 2003), these observations are consistent with the negative role of Plk1 in DNA damage-induced cell cycle arrest. On the other hand, there are also reports suggesting that Plk1 may play a positive role in DNA damage checkpoint activation. Chk2 co-immunoprecipitates with Plk1, overexpression of which enhances phosphorylation of Chk2 at T68 (Tsvetkov et al., 2003), a site primarily targeted by ATM, and its phosphorylation is correlated with its activation (Ahn et al., 2000); in addition, Plk1 phosphorylates recombinant Chk2 in vitro at the same site and colocalizes with Chk2 to a certain mitotic apparatus (Tsvetkov et al., 2003). In a separate study, Drosophila Polo has also been implicated to be a substrate of Chk2 as well (Xu and Du, 2003).

Other mammalian Plks are also involved in the DNA damage checkpoint activation pathway. Expression of Plk2 mRNA is rapidly induced in human thyroid cells upon X-ray irradiation; a radiation-responsive element has been identified as p53RE, a p53-binding homology element, in the basal promoter region of this gene (Shimizu-Yoshida et al., 2001). Consistent with that observation, a separate study shows that Plk2 expression is partly mediated by p53 (Burns et al., 2003). A recent study by Swallow et al shows that Plk4 interacts with p53 and that this interaction is mediated by the polo box (this review issue). Several lines of evidence indicate that Plk3 is an important mediator in the DNA damage checkpoint response pathway. Plk3 kinase activity is activated upon oxidative stress and DNA damage induced by ionizing-radiation mimetic drugs, and its activation is ATM-dependent (Xie et al., 2001a, 2001b). Plk3 interacts with and phosphorylates p53, targeting serine-20 of p53 in vitro (Xie et al., 2001b). In response to DNA damage, the kinase activity of Plk3 is rapidly increased in an ATM-dependent manner (Xie et al., 2001a, 2001b). Peptide mapping, as well as in vitro phosphorylation followed by immunoblot analysis with antibodies specific for phosphorylated forms of p53, also indicates that Plk3 phosphorylates p53 on physiologically relevant sites. Immunoprecipitation and ‘pull-down’ assays reveal that Plk3 physically interacts with p53 and that the extent of this interaction is increased in response to DNA damage (Xie et al., 2001b). The role of Plk3 in regulation of serine-20 phosphorylation of p53 in vivo is further supported by our recent RNAi study. Downregulation of endogenous Plk3 through transfection of small-interfering RNA (siRNA) to Plk3, but not to luciferease, significantly compromises phosphorylation of p53 on serine-20 before and after oxidative stress (Figure 1). Plk3 also physically interacts with Chk2 (Bahassi et al., 2002; Xie et al., 2002), and there exists a functional connection as well between these two enzymes during DNA damage checkpoint activation. Plk3 phosphorylates Chk2 on a residue different from threonine-68, and it may contribute to the full activation of Chk2 although ATM is necessary for phosphorylation and activation of Chk2 in vivo (Bahassi et al., 2002). Together, these combined studies suggest that Plk3 functionally links DNA damage to the induction of cell cycle arrest or apoptosis.

Figure 1
figure 1

Silencing of Plk3 results in significantly reduced p53 phosphorylation on serine-20. GM00637 cells, which constitutively express high levels of p53, were transfected with siRNA duplexes to Plk3 (Plk3-siRNA) or to luciferase GL3 (Luc-siRNA) (SMARTpool, Dharmacon) for 72 h. The transfected cells were treated with H2O2 (100 μ M) for 15 min and then collected for preparation of lysates. Equal amounts of cell lysates were blotted for Plk3, serine-20-phosphorylated p53, and actin

Full size image

DNA damage repair

A possible role for Plks in DNA synthesis or repair is implied from several early studies. Expression of both mammalian Plk2 and Plk3 is rapidly induced by mitogen treatment (Simmons et al., 1992; Li et al., 1996). Induction of Plk3 mRNA by mitogens is protein synthesis-independent, suggesting that it is an immediate-early gene product (Li et al., 1996). Furthermore, microinjection of sense Plk1 mRNA, but not antisense RNA, into NIH3T3 cells that have been serum-starved results in incorporation of 3H-thymidine (Hamanaka et al., 1994). The latter observation suggests that Plk1 is required for DNA synthesis and overexpression of Plk1 appears to be sufficient for induction of DNA synthesis.

Our recent study suggests that Plk3 may directly participate in DNA repair by regulating the activity of DNA polymerase δ (pol δ). DNA pol δ is thought to be a major enzyme involved in the DNA damage repair process (Hubscher et al., 2002; Haracska et al., 2003). Pol δ has been identified as four subunits in Schizosaccharomyces pombe and four subunits in mammalian cells (p125, p68, p50 and p12) (Hubscher et al., 2002). Post-translational modifications by phosphorylation are well documented for DNA polymerases; Pol δ is a phosphorylatable protein whose phosphorylation peaks during the S phase (Zeng et al., 1994). In order to identify proteins that either are targets of Plk3 or regulatory components of Plk3, we have employed the antibody array approach to screen a potential interaction between Plk3 and cell cycle or signaling proteins. Our initial screen results reveal that p125, the major Pol δ subunit, is one of the proteins that interacted with Plk3 (data not shown). The physical interaction between these two proteins was further confirmed by co-immunoprecipitation (data not shown) and pull-down experiments. Ni-NTA resin, having a high affinity to proteins with a stretch of histidine residues, is capable of pulling down p125 antigen from lysates prepared from cells co-infected with His6-Plk3 and p125 baculoviruses, but not from parental cells with no viral infection or cells infected with p125 baculovirus alone (Figure 2a).

Figure 2
figure 2

Plk3 interacts and phosphorylates the p125 subunit of Pol δ. (a) Sf9 cells either infected with p125 baculovirus or both His6-Plk3 and p125 baculoviruses for 48 h. Cell lysates were prepared from the infected cells, as well as from uninfected control cells. Ni-NTA resin, which was capable of binding His6-Plk3, was added to each lysate. After thorough washing, proteins bound to Ni-NTA resin, along with cell lysates, were blotted for p125. (b, c) Various fragments of p125 were expressed as GST fusion proteins. Approximately equal amounts of purified proteins, as well as GST alone and casein, were used for in vitro kinase assays in the presence of purified His6-Plk3. Kinase buffer alone was also used as a negative control. Reaction mixtures were fractionated on denaturing SDS–polyacrylamide gels, followed by autoradiography. Each number denotes a peptide fragment encompassing the particular region of p125 protein. (d) The phosphorylated band (arrow GST-1-110) as shown in (c) was excised from the gel and eluted. The eluted protein was subjected to two-dimensional peptide analysis. The letters C and E correspond to chromatography and electrophoresis, respectively, and the letter o denotes the origin. (e) Each of five serine/threonine residues in the GST-1-110 fragment was replaced individually with alanine in five mutants through site-directed mutagenesis. Mutant proteins were expressed and purified and the purified proteins were analysed on an SDS–polyacrylamide gel, which was stained with Coomasie Brilliant blue (f). Roughly equal amounts of the purified mutant proteins as shown in e, along with casein, were used for in vitro kinase assays in the presence of His6-Plk3

Full size image

Plk3's kinase activity is high during the S phase (Ouyang et al., 1997). Therefore, in vitro kinase assays were performed to examine whether recombinant Plk3 directly phosphorylates p125. We first made a series of truncated forms of p125 expressed as glutathione s transferase (GST) fusion proteins. Purified proteins were subjected to kinase assays in vitro. We observed that a deletional form of p1251−365 encompassing amino acids 1–365 (of the total 1107 amino acids) is phosphorylated by His6-Plk3, whereas other truncated forms were not phosphorylated (Figure 2b). Additional deletions revealed that the phosphorylation site(s) resides in amino acids 1–115 (Figure 2c). Two-dimensional peptide mapping of this fragment shows that there is only one major phosphorylation site in the p1251−115 fragment (Figure 2d), although five serine residues could be potential targets of Plk3. To identify p125's Plk3 phosphorylation site, we constructed GST expression constructs coding for mutant peptide fragments with each of five serine/threonine residues replaced with alanines, respectively. Various phosphorylation site mutant proteins of p1251−115 were then expressed and purified (Figure 2e). Approximately equal amounts of the purified mutant proteins were used for in vitro kinase assays in the presence of His6-Plk3. The mutant p1251−115 with serine-60 substituted with alanine is not phosphorylated by His6-Plk3, whereas all other mutants are phosphorylated (Figure 2f), confirming that serine-60 is the site targeted by Plk3. The neighboring sequence of the identified phosphorylation site, L–Q–S*–V–L–E, partially conforms to the consensus motif, D/E–X–S/T*–Φ–X–D/E (X, any amino acid; Φ, a hydrophobic amino acid), for phosphorylation by vertebrate Plk1. The significance of physical interaction and phosphorylation of Pol δ by Plk3 remains to be elucidated. We speculate that the phosphorylation may regulate the subcellular localization of pol δ because the region of p1251−115 contains a nuclear localization signal.

Together with various observations reported in the literature, a simple model is proposed to illustrate the role of vertebrate Plks in an intra-S-phase checkpoint (Figure 3). Although downstream of ATM, Plk1 and Plk3 appear to be differentially regulated during the DNA damage response. Whereas Plk3 positively regulates p53 activity, Plk1 functions to inactivate p53 during DNA damage response. Plk2 is a transcriptional target of p53 and its function in the checkpoint response remains unclear. Both Plk1 and Plk3 interact and phosphorylate Chk2 in vitro, although their regulatory hierarchy in vivo remains to be elucidated. Plk3 may also participate in DNA damage repair by regulating Pol δ activity through phosphorylation.

Figure 3
figure 3

Major pathways where Plks may play a role in intra-S-phase checkpoint in mammalian systems. Arrows and bars denote the positive and negative regulations, respectively. Dotted lines denote where the regulatory role in vivo between these proteins remains to be elucidated

Full size image

Checkpoint adaptation

Adaptation to an intra-S-phase checkpoint is the molecular process that alleviates the cell cycle arrest even in the presence of a damaged genome or replication errors. It is best described in the budding yeast as the response to DNA strand breaks. CDC5, a Polo homologue in the budding yeast, is required for the adaptation (Toczyski et al., 1997). A recent study shows that a similar molecular mechanism also exists in vertebrates and Polo-like kinase is involved in this process (Yoo et al., 2004). Plx1, the Xenopus counterpart of mammalian Plk1, physically interacts with the checkpoint protein claspin that is phosphorylated on threonine-906; this interaction leads to additional phosphorylation of claspin on serine-934 by Plx1 (Yoo et al., 2004). Xenopus egg extracts are capable of adaptation after a long interphase delay and the adaptation is characterized by having incompletely replicated DNA, dissociation of claspin from chromatin, and inactivation of Chk1. Interestingly, egg extracts treated with the DNA replication poison aphidicolin supplemented with claspin mutants with threonine-906 and serine-934 residues replaced with alanines are unable to undergo adaptation (Yoo et al., 2004), strongly suggesting a role for Plx1 in alleviating DNA replication checkpoint response by inactivation of claspin through phosphorylation. For unicellular organisms, continued cell proliferation through adaptation is a better alternative than being permanently arrested in the presence of extensive DNA damage. From the evolutionary point of view, this mechanism may even offer certain selective advantages. However, the biological significance of checkpoint adaptation in multicellular organisms such as Xenopus remains a mystery because it is difficult to imagine why an organism would allow the survival of a cell with a damaged or incompletely replicated genome, a condition suitable for malignant transformation.

Plks' role in the G2/M checkpoint

The G2/M checkpoint prevents inappropriate initiation of mitosis when there exist abnormalities in the genome. One major mechanism of this checkpoint in vertebrates involves regulation of the activity of Cdc2/cyclin B, whose activation requires the combined action of the upstream phosphatase Cdc25 and a Cdk-activating kinase. Plks have been reported to phosphorylate and regulate two key molecular players involved in controlling mitotic onset, although their biological significance remains to be elucidated. Early work has shown that Plx1 indirectly regulates Cdc2/cyclin B, the activity of which is required for initiation of mitosis (Abrieu et al., 1998). Recombinant Plx1 is capable of phosphorylation of Cdc25, an activator of Cdc2, and stimulation of its activity in vitro; Cdc25 phosphorylated by Plx1 exhibits strong MPM-2 epitopes (Qian et al., 2001). The activation of Cdc25 by Plk1 also occurs in a mammalian system. Both Plk1 immunoprecipitates obtained from G2/M-arrested human cells and recombinant human Plk1 phosphorylate recombinant Cdc25C in vitro and this phosphorylation leads to the activation of the phosphatase (Roshak et al., 2000). Plk3 also interacts and phosphorylates Cdc25C in vitro (Ouyang et al., 1999; Bahassi et al., 2004) although the sites phosphorylated by Plk3 appear to be different from those of Plk1 (Toyoshima-Morimoto et al., 2002). The biological consequence of phosphorylation of Cdc25C by Plk1 or Plk3 in vivo remains largely unknown. However, a couple of studies suggest that Plk-mediated phosphorylation of Cdc25C may promote its nuclear translocation (Toyoshima-Morimoto et al., 2002; Bahassi et al., 2004). For example, Plk1 phosphorylates Cdc25C on serine-198, which is located in a nuclear export signal sequence (Toyoshima-Morimoto et al., 2002). Constitutively active Plk1 promotes the nuclear localization of this dual-specific protein phosphatase in vivo, whereas a mutant Cdc25C in which serine-198 is replaced by alanine, but not the wild-type one, remains in the cytoplasm during prophase (Toyoshima-Morimoto et al., 2002). Interestingly, a recent study shows that Plk3 phosphorylates Cdc25C mainly on serine-191, which is also located within the nuclear exclusion motif (Bahassi et al., 2004). Further analysis of ectopically expressed mutant forms of Cdc25 as well as Plk3 leads authors to conclude that Plk3 mediates Cdc25C nuclear translocation by phosphorylating it primarily on serine-191 (Bahassi et al., 2004).

Plk1 and its orthologue are reported to regulate the nuclear translocation of cyclin B, resulting in activation of Cdc2 as well. Plx1 appears to be the primary enzyme that phosphorylates serine-147 in vitro within the nuclear export signal sequence of cyclin B1, because antibody depletion abolishes the activity of Plx1 towards cyclin B1 (Toyoshima-Morimoto et al., 2001). Serine-147 is involved in both retention of cyclin B in the nucleus and maintenance of the activity of Cdc2, because a phospho-specific antibody to this residue detects cyclin B1 only during the G2/M phase and because a mutant cyclin B1 with serines 133 and 147 replaced with alanines remains in the cytoplasm (Toyoshima-Morimoto et al., 2001). Consistently, Yuan et al. (2002) demonstrate that human cyclin B1 contains four major serine residues (serines 126, 128, 133, and 147), phosphorylation of which is important for its nuclear translocation. Moreover, Plk1 is primarily responsible for phosphorylation of serine-133 (Yuan et al., 2002). On the other hand, Walsh et al. (2003) report that cyclin B1 is phosphorylated by Plx1 on serine-101 (equivalent to serine-133 of human cyclin B1), but not on serine-113 (equivalent to serine-147 of human cyclin B1), within the cytoplasmic retention sequence. Furthermore, Jackman et al. (2003) demonstrate that human Plk1 neither phosphorylates cyclin B1 in the nuclear export sequence nor causes translocation into the nucleus; using phospho-specific antibodies to cyclin B1, these authors show that centrosome-localized cyclin B is first phosphorylated during prophase, suggesting that Cdc2/cyclin B1 may be first activated in the cytoplasm and that the centrosomes may be the primary sites for activation of proteins that are essential for mitotic entry.

Together, these studies all indicate the importance of Plk1 or its orthologue in promoting mitotic onset by controlling the activity of Cdc2/cyclin B. Plk1 activity in turn is partly regulated through ubiquitination by Chfr (Kang et al., 2002), an E3 ubiquitin ligase commonly referred to as a checkpoint protein. Chfr contains a ring-finger domain which auto-ubiquitinates and is required for the ligase activity towards substrates such as Plk1; Chfr-mediated ubiquitination of Plk1 impairs both activation of Cdc25C and inactivation of Wee1, resulting in a delayed activation of Cdc2 (Kang et al., 2002). Chfr activity is negatively regulated through phosphorylation by PKB both in vivo and in vitro; a phosphorylation mutant of Chfr that cannot be phosphorylated by PKB causes significant reduction of Plk1 as well as inhibition of mitotic entry in the presence or absence of DNA damage (Shtivelman, 2003). In addition to DNA damage and incompletion of DNA replication, insufficient chromatid decatenation also triggers the activation of a G2/M checkpoint (Deming et al., 2002). Treatment of cells with ICRF-193, an inhibitor of topoisomerase II that does not cause DNA damage, suppresses Plk1 kinase activity, which is correlated with a reduction in cyclin B phosphorylation; the ICRF-193-mediated suppression of Plk1 appears to be ATR-dependent (Deming et al., 2002).

Plks' role in spindle checkpoint

The requirement of Plks in metaphase/anaphase transition is evident from the study of RNAi. Silencing Plk1 via RNAi alone induces a significant increase in the 4N cell population, as shown by the analysis of flow cytometry (Figure 4a and b). A further examination by fluorescence microscopy reveals that a majority of cells are arrested at metaphase when they are transfected with siRNA targeting Plk1, but not luciferease (Figure 4c and d). Similar observations are also reported in several independent studies (Liu and Erikson, 2002; van Vugt et al., 2004). Silencing of Plk1 by RNAi causes the accumulation of almost half of the cells with incompletely separated chromatids (Liu and Erikson, 2002). What is intriguing is that silencing of Plk1 apparently does not cause G2/M arrest because these cells exhibit a high level of Cdc2/cyclin B activity (Liu and Erikson, 2002). In addition, a separate study shows that Plk1 depletion via RNAi leads to prometaphase arrest (van Vugt et al., 2004). A detailed analysis of the timing of mitotic entry indicates that the majority of cells with Plk1 deletion initiate mitotic entry with kinetics similar to that of control cells, indicating that Plk1 is not required for mitotic entry (van Vugt et al., 2004). Furthermore, overexpression of kinase-dead Plk1 or Plk1 with an amino-terminal deletion also causes pre-anaphase arrest, which is characterized by an elevated Cdc2/cyclin B activity with DNA content ⩾4N (Seong et al., 2002). Together, these observations indicate that, whereas Plk1 is essential for progression of mitosis, its activity may not be essential for the onset of mitosis in mammalian cells.

Figure 4
figure 4

Silencing Plk1 results in metaphase arrest. (a) A549 or HeLa cells were transfected with Plk1 siRNA duplex (Plk1-siRNA, 5′-AAGGGCGGCUUUGCCAAGUGC-3′) or with a nonspecific siRNA duplex (CNTL-siRNA) for 3 days. Equal amounts of lysates prepared from the transfected and untransfected parental cells were blotted for Plk1 and tubulin. Recombinant flag-tagged Plk1 was used as a positive control for blotting. (b) HeLa cells were transfected with siRNA duplex to Plk1 (Plk1-siRNA) or to luciferase (Luc-siRNA, 5′-CUUACGCUGAGUACUUCGA-3′) for 3 days. Transfected cells, as well as the untransfected cells, were analysed for DNA content by flow cytometry. (c) HeLa cells seeded on chamber slides were transfected with or without siRNA duplex to Plk1 or to luciferase for 3 days. Cells were then fixed, stained with DAPI, and examined for their cell cycle status using fluorescence microscopy. The brightly stained cells are in various stages of mitosis. (d) The number of mitotic cells as shown in (c) were scored and the percentage of cells in various stages of mitosis was summarized. P, M, and A denote prophase, metaphase, and anaphase, respectively. (e) HeLa cells were fixed and stained with DAPI (blue), and the antibody with BubR1 (green). Representative cells of various cell cycle stages (interphase, prophase and metaphase) are shown. Arrows indicate the location of spindle poles. Bar, 5 μ M

Full size image

Bipolar mitotic spindle

The spindle checkpoint (also known as the spindle assembly/integrity checkpoint or mitotic checkpoint) ensures that cells with a defective spindle or with defective spindle–kinetochore interactions do not initiate anaphase entry. The accurate segregation of chromosomes during mitosis is a crucial cellular process that depends on the formation of intact bipolar spindles. Faithful transmission of chromosomes is at least partly regulated by the spindle checkpoint, a mechanism preventing the cell from prematurely entering anaphase until all replicated and condensed chromosomes have attached to functional bipolar spindles. Early studies have demonstrated the importance of Plks in the formation of bipolar spindle (Sunkel and Glover, 1988; Ohkura et al., 1995). In fission yeast, either disruption of the Plo1 or its overexpression results in the formation of monopolar spindles due to the failure of the spindle pole body to complete either its duplication or separation (Ohkura et al., 1995). Vertebrate Plks are also involved in regulating bipolar spindle formation. In HeLa cells, neutralizing Plk1 activity by antibody injection leads to a mitotic arrest, with a monopolar spindle formed around a centrosome that is reduced in size (Lane and Nigg, 1996). Moreover, human cells with Plk1 deficiency are incapable of formation of a bipolar spindle, resulting in mitotic arrest (van Vugt et al., 2004).

Although the molecular mechanism by which Plk1, or its orthologue, regulates mitotic spindle formation remains nebulous, several proteins appear to play a central role in this process (Budde et al., 2001; Yarm, 2002; Casenghi et al., 2003). (i) Plk1 phosphorylates NIp, ninein-like protein, and the phosphorylation disrupts the ability of NIp to be associated with ã-tubulin, as well as with the centrosome (Casenghi et al., 2003). As NIp is an important component of the centrosome and essential for stimulating microtubule nucleation, it is proposed that at the onset of mitosis Plk1 displaces NIp through phosphorylation, thus establishing a condition for assembly of the mitotic spindle (Casenghi et al., 2003). (ii) Plk1 interacts with and phosphorylates the microtubule-stabilizing protein TCTP; two serine residues of TCTP are apparently targets of Plk1 both in vivo and in vitro; ectopic expression of these TCTP phosphorylation site mutants impairs mitotic progression and chromosomal segregation, resulting in an increase in multinucleation (Yarm, 2002). Therefore, Plk1-mediated phosphorylation of TCTP may reduce its ability to stabilize microtubules, leading to increased microtubule dynamics needed during mitosis. (iii) Plx1 is capable of phosphorylating stathmin (also termed Op18), a microtubule-destabilizing protein, and thereby negatively regulates its activity (Budde et al., 2001). Phosphorylation of stathmin by Plx1 peaks at the onset of mitosis, promoting microtubule stabilization and mitotic spindle assembly. Thus, stathmin may be another substrate, whose phosphorylation by Plks plays a key role in mitotic progression.

Spindle checkpoint machinery

Extensive research in the past several years has delineated the molecular pathway by which spindle checkpoint activation leads to metaphase arrest (Figure 5). Spindle checkpoint is fully activated when there exists a defective spindle or a defective spindle–kinetochore interaction (Musacchio and Hardwick, 2002; Bharadwaj and Yu, 2004). Activated checkpoint components, including Bub1, BubR1, Bub3, Mad1, Mad2, and CENP-E, inhibit anaphase-promoting complex/cyclone (APC/C) via direct interaction with Cdc20, an APC/C activator (Bharadwaj and Yu, 2004). Besides, it has been shown that early mitotic inhibitor 1 (Emi1) also functions to block the activity of APC/C (Moshe et al., 2004). When the checkpoint is inactivated, APC/C, a ubiquitin E3 ligase, poly-ubiquitinates securin (a separase inhibitor) for degradation by the proteasomal pathway. After dissociation from securin, separase (a caspase-like protease) becomes activated, which cleaves cohesin subunit Scc1/hRad21 that connects sister chromatids, thus allowing mitotic progression from metaphase into anaphase. In other words, the cohesin complex must be removed by the proteolytic action of separase to allow the onset of anaphase. Both occupancy at the kinetochore and tension generated between the kinetochores and the spindle poles are enough to satisfy (inactivate) the spindle checkpoint (Musacchio and Hardwick, 2002). Plks appear to regulate the spindle checkpoint at least at four different levels (Figure 5). (i) Plk1 facilitates the degradation of Emi1 by SCFβ−TrCP ubiquitin ligase. This may be mediated through phosphorylation of a sequence in Emi1 that is required for polyubiquitin conjugation (Moshe et al., 2004). Thus, Plk1 promotes mitotic progression through removal of an APC/C inhibitor; (ii) Plks may directly regulate the activity of spindle checkpoint components. For example, Plk3 and Plk1 interact with BubR1 and Plk1 is capable of phosphorylating BubR1 in vitro (unpublished results). We have also observed that BubR1 exhibits spindle pole subcellular localization during metaphase but not in prophase or prometaphase (Figure 4e). Plks are known to reside at centrosomes and spindle poles (Arnaud et al., 1998; Wang et al., 2002). It is possible that the transit of BubR1 through spindle poles is a necessary step for its inactivation at the onset of anaphase. (iii) Plk1 is associated with APC components at the metaphase–anaphase transition, thus, indirectly regulating its ubiquitination activity (Kotani et al., 1998; Golan et al., 2002). Mouse Plk1 specifically phosphorylates at least three components of APC (subunits Cdc16, Cdc27, and Tsg24) and activates APC/C in vitro (Kotani et al., 1998). In fission yeast, Plo1 physically interacts with Cut23, a subunit of APC/C, through its noncatalytic domain; the physical interaction is compromised by a Cut23 mutation that is accompanied by a metaphase arrest, while overexpression of Plo1 can rescue this phenotype (Golan et al., 2002). (iv) A Plk1 homologue in yeast plays an important role in degradation of a cohesin subunit. In Saccharomyces cerevisiae, Cdc5 phosphorylates serine residues adjacent to the cleavage site of Scc1, a subunit of cohesin, thereby enhancing its cleavage by separase at the onset of anaphase (Alexandru et al., 2001). Similarly, Cdc5 is also required for phosphorylation and removal of cohesin from chromosomes during meiosis (Lee and Amon, 2003). In Xenopus, Plx1 destabilizes the linkage of sister chromatids because, when Plx1 is depleted from Xenopus egg extracts, the release of cohesin during prophase is blocked (Losada et al., 2002). Interestingly, this blockage is more pronounced when both Plx1 and aurora B are simultaneously depleted (Losada et al., 2002), suggesting that these two mitotic kinases may cooperate to promote mitotic progression.

Figure 5
figure 5

A model illustrating Plks' role in the spindle checkpoint response pathway. Four potential levels of regulation of the pathway by Plks are indicated

Full size image

Plks' role in the cytokinesis checkpoint

In low eukaryotes

The essential role of Plks in cytokinesis is first established from the study in yeast. Several studies report that CDC5 in the budding yeast and Plo1 in the fission yeast are involved in controlling mitotic exit (Heitz et al., 2001; Hu et al., 2001). (i) CDC5 is one of the components that form the machinery called the mitotic exit network, which consists of additional genes the GTPase TEM1, the phosphatase CDC14, CDC15, LTE1, MOB1, and DBF2 (Hu et al., 2001). (ii) Cdc5 is also part of the FEAR (Cdc fourteen early anaphase release) network, which also includes the separase Esp1 and the kinetochore-associated proteins Slk19 and Spo12 (Stegmeier et al., 2002). The FEAR network initiates Cdc14 release from Cfi1/Net1 during early anaphase, whereas the MEN maintains Cdc14 in the released state during late anaphase and telophase (Stegmeier et al., 2002). In fact, Cdc5 is capable of affecting the phosphorylation state of Net1, a nucleolar inhibitor of Cdc14, thus reducing its affinity with Cdc14 (Yoshida and Toh-e, 2002). Given that Cdc14 is required for the timely activation of MEN, Cdc14 releasing through the FEAR network is an important step for initiating mitotic exit. This notion is supported by several lines of additional evidence obtained from the study in low eukaryotes. By using temperature-sensitive mutants of CDC5 with carboxyl-terminal defects, Park et al. (2004) show that, at semi-permissive temperatures, certain mutants exhibit chained cell morphology and shared cytoplasms between connected cell bodies; a further examination reveals that septin assembly at the incipient bud site is delayed with a loosely organized septin ring at the neck of the mother bud, indicating a defect in cytokinesis. A similar requirement for Plk in the fission yeast has also been reported. Plo1 is a key regulator of cytokinetic actomyosin ring (CAR) formation and recruitment of Plo1 to the spindle pole body, as well as a functional spindle checkpoint, is required for CAR formation and septation (Mulvihill and Hyams, 2002).

Overexpression of wild-type Cdc5, or a catalytically inactive form, results in the formation of multinucleated cells in the budding yeast, which is apparently due to negative regulation of Swe1, a Wee1 kinase, by Cdc5 (Bartholomew et al., 2001). Moreover, ectopic expression of wild-type mammalian Plk1 complements the cell division defect associated with the CDC5-1 mutation in S. cerevisiae and the degree of complementation correlates closely with the Plk1 activity measured in vitro; expression of an activated Plk1 (Plk1T210D) also induces a class of cells with unusually elongated buds which develop multiple septal structures (Lee and Erikson, 1997). In addition, a loss of S. pombe Plo1 function leads to the failure of septation both in the formation of an F-actin ring and in the deposition of septal material, suggesting that Plo1 function is required in the regulatory cascade that controls septation (Schwab et al., 2001). The overexpression of Plo1 also induces the formation of multiple septa without nuclear division (Schwab et al., 2001). A more detailed analysis on Plo1 indicated that it plays a role in the positioning of division sites by regulating Mid1p, a known gene product regulating cytokinesis in the fission yeast (Bahler et al., 1998). A recent study on CAR has revealed that recruitment of Plo1 to the spindle pole body is substantially reduced in the presence of the microtubule-depolymerizing agent thiabendazole, causing an extensive delay in CAR formation (Mulvihill and Hyams, 2002).

In high eukaryotes

Cytokinesis in animal cells is achieved by the formation of an actin ring that contracts to divide the cytoplasm. Polo and vertebrate Plks are essential for this process. A Drosophila genetic study reveals the requirement of Polo in the regulation of cytokinesis (Carmena et al., 1998). In fruit flies harboring hypomorphic alleles of the Polo gene, a number of defects in cytokinesis are observed, which include a failure to form the correct mid-zone and mid-body structures at telophase and the frequent formation of binuclear or tetranuclear spermatids during meiosis. There are also defects in correct localization of the Pavarotti kinesin-like protein to function in cytokinesis, and incorporation of the septin Peanut and actin molecules into a contractile ring (Carmena et al., 1998). In Xenopus, it has been demonstrated that the addition of a catalytically inactive Plx1 mutant to M-phase-arrested Xenopus egg extracts inhibits the proteolytic destruction of several APC/C targets, the inactivation of Cdc2 protein kinase activity, and the entry into interphase induced by Ca2+ (Brassac et al., 2000).

Degradation of mammalian Plks such as Plk1 contributes to inactivation of this kinase, a necessary step for normal mitotic exit (Lindon and Pines, 2004), which may explain why overexpression of kinase-active Plk3 causes the accumulation of cells with cytokinetic failure (Wang et al., 2002). Putative destruction boxes have been identified in Plk1 (Lindon and Pines, 2004) as well as in the N-terminus of Cdc5 (Golsteyn et al., 1994), indicating that at least some Plks may function as the activator as well as the target of APC/C. Interestingly, Plk1 also interacts with 20S and 26S proteasome subunits and phosphorylates the 20S proteasome, leading to its enhanced proteolytic activity (Feng et al., 2001). Thus, Plk1 functions as a mitotic regulator of proteolytic activities through its effect on both APC/C and proteasomes.

Several mammalian Plks are found at the midbody region during late mitosis (Conn et al., 2000; Tsvetkov et al., 2003; Ruan et al., 2004). It has been shown that the functional Polo box domain of Plk4 localizes the enzyme to the cleavage furrow during cytokinesis (Hudson et al., 2001). It has also been demonstrated that EGFP–Plk3 fusion protein concentrates at the midbody and is associated with the cellular cortex (Conn et al., 2000), suggesting a potential role for Plk3 during mitotic exit. Our independent study confirms that ectopic expression of kinase-active Plk3, but not the kinase-defective one, causes a defect in cytokinesis, eventually resulting in apoptosis (Wang et al., 2002).

The exact molecular mechanism by which mammalian Plks regulate cytokinesis remains to be elucidated. However, a series of recent studies begin to shed light on the possible role of Plk1 in the regulation of cytokinesis. (i) Plk1 colocalizes with Pavarotti, a kinesin-related motor protein that is required for the organization of the central spindle, the formation of a contractile ring, and cytokinesis (Adams et al., 1998). Consistent with a role in regulating microtubule motor proteins, Plk1 also phosphorylates mitotic kinesin-like protein 2 (MKlp), the phosphorylation of which is essential for cytokinesis in mammalian cells (Neef et al., 2003). It is proposed that Plk1-mediated phosphorylation of MKlp2 may facilitate its spatial restriction to the spindle region during late mitosis, a necessary step for cytokinesis (Neef et al., 2003). Together, these studies suggest that motor proteins are targets of Plk1 during exit from mitosis. (ii) Plk1 physically interacts with nuclear distribution gene C (NudC) (Zhou et al., 2003). In fact, Plk1 phosphorylates NudC at conserved serine-274 and serine-326 both in vivo and in vitro; silencing NudC via RNAi causes multiple defects in mitosis, including arresting cells at the midbody stage, which can be rescued by wild-type NudC, but not NudC with serines 274 and 326 mutated into alanines (Zhou et al., 2003). As NudC, as well as Plk1, concentrates at the midbody during cytokinesis (Tsvetkov et al., 2003; Zhou et al., 2003), it is possible that NudC phosphorylated by Plk1 may generate signals that are essential for completion of cell division. (iii) Some recent studies on the role of Plks in Golgi dynamics may provide new insights into the mechanism of action of these kinases in cytokinesis (Lin et al., 2000; Litvak et al., 2004; Ruan et al., 2004; Xie et al., 2004). Plk1 is shown to interact with Golgi-specific protein GRASP65 and participate in regulation of fragmentation of the Golgi apparatus during mitosis (Lin et al., 2000). Similarly, human Plk3 is Golgi-localized during interphase and distributes in a manner similar to that of Golgi components during mitosis; moreover, during telophase, Plk3 and Golgi components exhibit two clusters, with one localized at the spindle pole and the other close to the midbody (Ruan et al., 2004). Furthermore, Nir2, a peripheral Golgi protein essential for cytokinesis (Litvak et al., 2002), is phosphorylated by Cdk1 during mitosis and this phosphorylation provides a binding site for Plk1 (Litvak et al., 2004). Given that phosphorylation of Nir2 during mitosis is an essential modification mediating normal cytokinesis (Litvak et al., 2004), it is tempting to speculate that Plk1 and perhaps other Plks may play a major role in coordinating Golgi fragmentation and assembly, mitotic spindle assembly and disassembly, and chromosome segregation during mitosis, deregulation of which would compromise cytokinesis, resulting in formation of multinucleated cells.

Conclusions

Cell cycle checkpoints monitor specific cell cycle-related processes and block cycle progression until these processes are completed with high fidelity. Plks participate in almost all known cell cycle checkpoints that protect cells against genetic instability during cell division. Although it remains unclear as to their exact mechanism of action in controlling the intra-S-phase checkpoint, spindle checkpoint, and cytokinesis checkpoint, an increasing amount of evidence suggests that protein kinases of the polo family are central players in the temporal and spatial coordination of cell division. As tumor cells invariably harbor defects in various cell cycle checkpoints, further characterization of the role of Plks in cell cycle checkpoint control may lead to the identification of new molecular targets for the development of drugs that are more effective and specific for killing tumor cells.