Fission yeast nucleolar protein Dnt1 regulates G2/M transition and cytokinesis by downregulating Wee1 kinase

Eukaryotic cells use a highly conserved mechanism to control entry into mitosis, which depends on the activation of cyclin-dependent kinase (CDK), a key enzyme consisting of the protein kinase Cdk1 and its regulatory subunit, cyclin B (Morgan, 1997; Nurse, 1990). The highly activated Cdk1 drives entry into mitosis by phosphorylating a wide range of targets (Lindqvist et al., 2009; Ubersax et al., 2003). In the fission yeast Schizosaccharomyces pombe, the machinery regulating entry into mitosis has been thoroughly characterized (Nurse, 1994). The activity of the fission yeast Cdk1 (encoded by the cdc2⁺ gene) oscillates throughout the cell cycle, peaking as cells enter mitosis (Morgan, 1997). During interphase, Cdk1 activity is held in check by the inhibitory phosphorylation on conserved tyrosine 15 (Tyr¹⁵) in the ATP binding domain by Mik1 and Wee1 kinases (Berry and Gould, 1996; Lundgren et al., 1991; Russell and Nurse, 1987b), and this modification interferes with the ability of Cdk1 to transfer phosphate to the target substrate (Morgan, 1997). Upon the G2/M transition and mitotic onset, a sharp reversal of this phosphorylation event is triggered by the Cdc25 family of phosphatases, which act as a major rate-limiting step in the activation of Cdk1 (Millar et al., 1991; Nilsson and Hoffmann, 2000; Russell and Nurse, 1986).

Both Wee1 kinase and Cdc25 phosphatase are in turn tightly regulated to provide an accurate control of the mitotic onset, and a variety of factors and signaling pathways converge upon Wee1 and Cdc25 to govern the G2/M transition (Callegari and Kelly, 2007; Calonge et al., 2010; Kuntz and O'Connell, 2009; López-Avilés et al., 2005; O'Connell et al., 2000; Suda et al., 2000). Among them, the MAP-kinase-mediated stress-nutritional response (SR), the cell-geometry sensing (CGS) and the G2 DNA-damage-checkpoint pathways, all regulate Tyr¹⁵ phosphorylation through Wee1 and Cdc25 (Callegari and Kelly, 2007; Martin and Berthelot-Grosjean, 2009; Millar et al., 1995; Moseley et al., 2009; Petersen and Nurse, 2007; Rhind et al., 1997; Russell and Nurse, 1987a; Shiozaki et al., 1998). The SR pathway connects the nutrient-responding TOR pathway with the recruitment of Polo kinase (Plo1) to the spindle pole body (SPB) and Cdk1 activation, and is responsible for nutritional modulation of mitotic entry (Petersen and Hagan, 2005; Petersen and Nurse, 2007). The CGS pathway controls mitotic entry in response to cell geometry through two related inhibitory kinases of Wee1, Cdr1/Nim1 and Cdr2, and it involves a spatial gradient of the protein kinase Pom1 along the long axis of the cell, coupling cell length to G2/M transition (Hachet et al., 2011; Martin and Berthelot-Grosjean, 2009; Moseley et al., 2009). Finally, the G2 DNA damage delays entry into mitosis primarily by inhibiting Cdc25, but it also has a secondary effect through the activation of Mik1 (Rhind et al., 1997; Rhind and Russell, 2001).

Wee1 has been shown to be destabilized during G2 and M phases of the cell cycle in diverse organisms ranging from yeasts to humans, and therefore degradation is thought to be a major mechanism triggering the decrease in Wee1 levels as cells enter mitosis (Aligue et al., 1997; Kellogg, 2003; McGowan and Russell, 1995; Sia et al., 1998; Watanabe et al., 2005). It has been demonstrated that a multicomponent E3 (ubiquitin ligase), the SCF (Skp1–Cdc53/Cullin-1–F-box) complex, is responsible for Wee1 degradation at the onset of M-phase in budding yeast, frog egg extracts and human somatic cells (Kaiser et al., 1998; Michael and Newport, 1998; Watanabe et al., 2004). However, whether SCF-mediated Wee1 degradation is conserved in fission yeast awaits direct evidence, although it has been shown that skp1 mutants display cell elongation and G2-delay phenotypes (Lehmann et al., 2004).

In fission yeast, a spindle pole body (SPB)-based regulatory network, called the septation initiation network (SIN), triggers exit from mitosis and cytokinesis. The SIN is homologous to the mitotic exit network (MEN) in budding yeast, although the two pathways have diverged in function (reviewed by Balasubramanian et al., 2004; Gould and Simanis, 1997; Guertin et al., 2002a; Krapp and Simanis, 2008). The SIN signaling cascade is initiated by activation of the GTPase Spg1 and promoted by three protein kinases and their associated subunits: Cdc7, Sid1-Cdc14 and Sid2-Mob1. Assembly of SIN signaling components and their regulators occurs at the SPB on a platform built by the SIN scaffolding components Cdc11 and Sid4, which recruit all the members of the SIN to the SPBs. The SIN controls the final stages of cell division, including actomyosin ring contraction and formation of the division septum. Loss-of-function mutations in sin genes result in elongated multinucleate cells as a result of multiple rounds of nuclear division, and cell growth in the absence of cell division.

SIN signaling needs to coordinate cytokinesis with completion of chromosome segregation and is negatively regulated by Cdk1 activity. Previous studies have shown that the transition from a symmetrical to an asymmetrical distribution of Cdc7 kinase on the SPBs during anaphase, the initial association of Sid1 kinase with only the new SPB at anaphase B, and maximum SIN activation, are all dependent upon Cdk1 inhibition (Chang et al., 2001; Dischinger et al., 2008; Guertin et al., 2000).

Previously, in a genetic screen for suppressors of the cytokinesis checkpoint defect in SIN mutants, we identified the nucleolar protein Dnt1 as an inhibitor of SIN signaling (Jin et al., 2007). Although the amino acid sequence of Dnt1 shows weak similarity to that of budding yeast nucleolar proteins Net1/Cfi1, it seems that Dnt1 functions in a distinct way from Net1/Cfi1. Our study revealed that, unlike Net1/Cfi1, which regulates the MEN through the Cdc14 phosphatase, Dnt1 inhibits the SIN independently of Clp1/Flp1 (hereafter referred to as Clp1), the fission yeast homologue of Cdc14 (Jin et al., 2007). However, a detailed mechanism explaining how Dnt1 antagonizes the SIN signaling in fission yeast remained unclear.

In this work, we show that S. pombe cells display a delay in G2/M transition during the cell cycle in the absence of dnt1⁺, and this delay correlated with persistent Tyr¹⁵ phosphorylation on Cdk1. Genetic and biochemical analyses revealed that the G2/M transition defect in dnt1Δ cells is due to elevated activity and protein level of the mitotic inhibitor Wee1 kinase, because removal of Wee1 completely abolished the rescuing effect of dnt1Δ on SIN mutants. Finally, our data also suggest that Dnt1 modulates Wee1 activity in parallel with SCF-mediated Wee1 degradation. Thus, in fission yeast Dnt1 negatively regulates SIN signaling by modulating the protein levels of Wee1 and therefore the activity of Cdk1. The above findings reveal the existence of a novel mode of control of SIN by a nucleolar protein.

While studying dnt1⁺ for its role in cytokinesis, we made several observations that indicate that dnt1⁺ might also regulate the cell cycle progression. First, the rescuing effect of dnt1Δ on temperature-sensitive SIN mutants cdc14-118, sid2-250 and actomyosin ring formation mutant cdc8-110 was sensitive to protein level or activity of Cdc25, because the enhanced growth at 30°C or 33°C of double mutants dnt1Δ cdc14-118, dnt1Δ sid2-25 and dnt1Δ cdc8-110 compared with single mutants cdc14-118, sid2-250 or cdc8-110, was abolished when higher levels of Cdc25 were induced by a gain-of-function version of cdc25⁺ (i.e. nmt1-cdc25-D1) (Daga and Jimenez, 1999; Tallada et al., 2007) (Fig. 1A and supplementary material Fig. S1). Second, dnt1Δ deletion was synthetically sick with the temperature-sensitive mutant of cdc25-22 (Fig. 1B). Third, dnt1Δ deletion was synthetic lethal with cdc25Δ in a cdc2-3w genetic background, which is largely insensitive to Cdc25 (Fantes, 1981). When cdc2-3w cdc25Δ::ura4⁺ cells were crossed with the dnt1Δ::kan^R cdc2-3w mutant, triple mutant spores germinated to form microcolonies with elongated branched cells that stopped dividing after several cell divisions (Fig. 1C). Fourth, we also noticed that dnt1Δ cells divided at a significantly longer length than wild-type cells, and this was also true for cdc25-22 dnt1Δ cells when compared with cdc25-22 cells (Fig. 1D,E). In S. pombe, there is a direct correlation between cell length at septation and the timing of mitotic commitment, and a longer cell length at division indicates mitotic delay (Mitchison and Nurse, 1985). Thus, the above data suggest that Dnt1 might be a Cdc25-independent positive regulator of G2/M transition in fission yeast.

Next, we examined whether levels of Cdc2 itself and its phosphorylation at Tyr¹⁵ (pY15) were affected by the absence of dnt1⁺. We found that whereas Cdc2 protein level remained unchanged in dnt1Δ cells, Cdc2 phosphorylation at Tyr¹⁵ was significantly increased in these cells (Fig. 2A,B). To more directly monitor the levels of Cdc2 phosphorylation at Tyr¹⁵ during the cell cycle, we performed a cdc25-22 block-and-release assay. Cells synchronized in G2 by the temperature-sensitive cdc25-22 mutation were released to permissive temperature to allow mitotic entry. As shown in Fig. 2C, the kinetics of Cdc2 dephosphorylation at Tyr¹⁵ in cdc25-22 dnt1Δ cells was notably delayed compared with control cells. The maximum percentage of binucleated (entering mitosis) and septated cells was also reduced for an extended period in the cdc25-22 dnt1Δ mutant, confirming a cell cycle defect at G2. Moreover, the duration of Cdc2 Tyr¹⁵ dephosphorylation during mitosis in dnt1Δ cells was extended compared with control cells (Fig. 2C). Taken together, these results suggest that, in fission yeast, Dnt1 positively regulates both G2/M transition and mitotic exit during the cell cycle by affecting the phosphorylation status of Cdc2 kinase.

Our genetic analyses suggested that the cell cycle defect at G2/M in dnt1Δ cells was not dependent on Cdc25. To dissect how Dnt1 facilitates the G2/M progression, we conducted epistasis experiments measuring the cell length at division of double mutants for dnt1Δ and a series of genes involved in G2/M transition control (Fig. 3B and supplementary material Fig. S2). Mostly these double mutants either showed additive effect on cell length or divided at an intermediate length. Notably, no increase in cell size at division was observed in cdc2-1w dnt1Δ, wee1-50 dnt1Δ and wee1Δ dnt1Δ double mutant cells compared with the cdc2-1w, wee1-50 and wee1Δ single mutants, respectively (Fig. 3A,B). Cell length measurements also revealed that the longer size at division in the dnt1Δ mutant was not dependent on enhanced activity of Clp1 phosphatase or Pom1 kinase, which have been shown to be involved in negatively regulating Cdc25 and Cdr1/2 activity respectively (supplementary material Fig. S2) (Martin and Berthelot-Grosjean, 2009; Moseley and Nurse, 2009; Wolfe and Gould, 2004; Wolfe et al., 2006). These findings suggest that Wee1 is required for Dnt1-mediated cell cycle control during mitotic commitment.

Previous studies have shown that in fission yeast, G2 DNA-damage-checkpoint activation causes G2/M delay and sustained Wee1-mediated Tyr¹⁵ phosphorylation on Cdc2 until completion of repair (O'Connell et al., 1997; Rhind et al., 1997). Moreover, deletion of the ribosome-associated RACK1 (receptor of activated C kinase) ortholog Cpc2 leads to a Wee1-dependent cell cycle delay at the G2/M transition (Núñez et al., 2010). To investigate the possible functional connection between Dnt1 and the function of either the G2 DNA-damage checkpoint or Cpc2, we examined the cell length of double mutants between dnt1Δ and G2 DNA-damage-checkpoint mutants (rad3Δ, chk1Δ, cds1Δ and tel1Δ) and cpc2Δ. As shown in supplementary material Fig. S3A, all double mutants divided at an intermediate cell length, suggesting that Dnt1 does not regulate these pathways. Consistent with these observations, we found that G2 DNA-damage-checkpoint mutants did not counteract the suppressive effect of dnt1Δ on SIN mutant cdc14-118. Moreover, cpc2Δ further elongated the length of dnt1Δ cells and enhanced the rescue of the cdc14-118 mutant by dnt1Δ, showing an additive effect (supplementary material Fig. S3B,C, and data not shown).

Recently, we have shown that during metaphase Dnt1 interacts with and inhibits Dma1 (Wang et al., 2012). Previous studies have shown that Dma1 localizes to the SPB through interaction with the SIN scaffold protein Sid4 and it ubiquitylates Sid4 to prevent recruitment of the Polo-like kinase and the SIN activator Plo1 to SPBs during a mitotic checkpoint arrest (Guertin et al., 2002b; Johnson and Gould, 2011). It is also known that the SPB component Cut12 is functionally linked with Plo1. The gain-of-function mutant cut12-S11 promotes Plo1 recruitment to the G2 phase SPB, boosts global Plo1 activity and drives advanced mitotic commitment (Bridge et al., 1998; Grallert et al., 2013). To examine whether Cut12 or Dma1-regulated Plo1 is involved in Dnt1 modulation of G2/M transition, we first measured the cell length of double mutants between dnt1Δ and cut12-S11 or dma1Δ, and found that neither of these mutants could reverse the longer size at division of dnt1Δ cells (supplementary material Fig. S4A). Moreover, neither cut12-S11 nor dma1Δ could rescue the synthetic lethality of dnt1Δ cdc25-22 double mutant (supplementary material Fig. S4B). Therefore, the G2/M transition defects in dnt1Δ cells are not linked to Cut12- or Dma1-controlled Plo1.

Our epistasis analyses suggested that the G2/M delay in dnt1Δ cells most likely acts through Wee1 to promote maintenance of Cdc2 Tyr¹⁵ phosphorylation. This Cdc2 inhibitory effect could be exerted by increasing Wee1 activity and/or Wee1 abundance. To gain insight into the mechanism by which Wee1 inhibits Cdc2 when dnt1⁺ is absent, we determined Wee1 protein levels in the wild-type and dnt1Δ strains expressing a genomic version of HA₃His₆-Wee1 (Raleigh and O'Connell, 2000). Although we occasionally observed an increase of HA₃His₆-Wee1 levels in dnt1Δ cells compared with wild-type cells (Fig. 3C), we experienced some technical difficulty with the reproducibility of these assays. Therefore, we turned to a newly developed strategy to monitor protein levels by tagging Wee1 with luciferase from the sea pansy Renilla reniformis (or R.luciferase) (Matthews et al., 1977), while using firefly-luciferase-tagged Ade4 (Ade4-F.luciferase) as an internal control (see Fig. 3D, top panel). By using this strategy we found a reproducibly significant increase of Wee1 protein levels in dnt1Δ cells compared with wild-type cells (Fig. 3D, bottom panel). Therefore, Dnt1 negatively affects the abundance of Wee1 kinase, which is a negative regulator of Cdk1.

To understand how Wee1 protein is affected by Dnt1, we examined the mRNA levels of the wee1⁺ gene in dnt1Δ cells by RT-PCR. As shown in Fig. 3E, our results showed that the increase in Wee1 protein levels observed in dnt1Δ cells did not result from enhanced expression of wee1⁺ mRNA. Thus, it is possible that Dnt1 might influence the stability and/or turnover of Wee1 protein. The observation that Wee1 is upregulated in dnt1Δ cells led us to explore the possibility that protein levels of various Cdk1 regulators, such as Cdc25, Cdc13 (cyclin B), Pom1, Cdr1 or Cdr2 might be altered in this mutant. However, none of these proteins was affected by deletion of dnt1⁺ (supplementary material Fig. S5 and data not shown).

We addressed the functional significance of the negative regulation of Wee1 by Dnt1 during activation of SIN signaling by exploring how wee1⁺ affects the rescue of SIN mutants by dnt1Δ. Maximal SIN activation depends upon Cdk1 inhibition and thus SIN mutants are very sensitive to high Cdk1 activity. Thus, it is plausible that the low Cdk1 activity promoted by elevated Wee1 levels in dnt1Δ cells might be sufficient to maintain the SIN signaling activity to certain degree. To test this possibility, we first examined whether elevated Wee1 levels alone could rescue SIN mutants. Intriguingly, we found that a slight elevation of Wee1 levels by using a P_81nmt1-GFP-Wee1 construct rescued the SIN mutant cdc14-118 to a similar extent as dnt1Δ (Fig. 4A). Next, we introduced the wee1Δ mutation into single SIN mutants (cdc14-118 and sid2-250) or double mutants with dnt1Δ (dnt1Δ cdc14-118 and dnt1Δ sid2-250), and monitored either growth at different temperatures or the accumulation of multinucleated cells after a shift to 37°C in liquid cultures (Fig. 4B,C). Strikingly, wee1Δ completely abolished rescue of SIN mutants by dnt1Δ, as indicated by the similar temperature sensitivity and rate of multinucleated cells between SIN and wee1Δ dnt1Δ SIN strains (Fig. 4B,C). We also observed that the wee1Δ completely abolished rescue of the actomyosin ring formation mutant cdc8-110 by dnt1Δ (supplementary material Fig. S6). Furthermore, rescue of SIN mutants by dnt1Δ was not affected by either dma1Δ or cut12-S11 (supplementary material Fig. S7), suggesting that Dma1- or Cut12-controlled Plo1 is not involved in SIN regulation by Dnt1. These results allow us to conclude that rescue of compromised SIN mutants by dnt1Δ is solely dependent on elevated Wee1 levels and the concomitant attenuation of Cdk1 activity.

The finding that the absence of dnt1⁺ leads to increased Wee1 levels prompted us to investigate the possible mechanism involved in this alteration. We anticipated that Dnt1 might influence the turn-over rate of Wee1. It has been shown that the SCF complex is responsible for Wee1 degradation at the onset of M phase in budding yeast, frog and human cells (Kaiser et al., 1998; Michael and Newport, 1998; Watanabe et al., 2004). Therefore, we first examined whether SCF is also required for Wee1 degradation in fission yeast. Wee1 protein stability was measured with Wee1-R.luciferase activity in the 26S proteasome subunit mutant mts3-1 (Seeger et al., 1996), and temperature-sensitive mutant skp1-A7, which carries a mutation (T454C) in the major SCF subunit Skp1 (Lehmann et al., 2004). We found that Wee1 protein was more abundant in mts3-1 and skp1-A7 cells than in wild-type cells at the restrictive temperature (Fig. 5A), suggesting that the SCF-mediated proteasome-dependent pathway is responsible for Wee1 proteolysis in fission yeast.

To further explore whether Dnt1 influences SCF-mediated Wee1 degradation, we measured Wee1-R.luciferase activity and cell length at division in skp1-A7 dnt1Δ double mutant cells. We observed that dnt1Δ further enhanced Wee1 protein stability and cell elongation in the skp1-A7 mutant (Fig. 5A,B). These data suggest that Dnt1 probably does not directly affect SCF-mediated Wee1 degradation.

We also noticed that the elevated Wee1 protein level caused by skp1-A7 mutation could rescue SIN mutants to a similar degree as dnt1Δ, and the triple mutant skp1-A7 dnt1Δ cdc14-118 showed an even lower rate of accumulation of multinucleate cells (Fig. 5C,D). It is noteworthy that the skp1-A7 mutation did not rescue cdc14-118 more effectively than dnt1Δ, although the skp1-A7 mutant had increased levels of Wee1. This is probably due to the thermosensitivity of the skp1-A7 mutation, which could make cdc14-118 cells sicker. Nonetheless, these data are consistent with the idea that Dnt1 works in parallel with SCF to regulate the rate of Wee1 protein turn-over.

Previously, we identified the nucleolar protein Dnt1 as an inhibitor of the SIN signaling in fission yeast (Jin et al., 2007), but its exact role and mechanism of action in the regulation of SIN remained unclear. Here, we show that deletion of dnt1⁺ leads to an increased abundance of the Cdk1 inhibitor Wee1 kinase and thus suppresses Cdk1 activity and causes a G2/M transition delay. The ability of dnt1Δ to lower Cdk1 activity through Wee1 is responsible for the rescue of SIN mutants, because deletion of the wee1⁺ gene completely abolished elevated SIN signaling caused by dnt1Δ. Hence, we have uncovered a novel mechanism by which Dnt1 controls SIN signaling during late mitosis through its effect on Cdk1 activity.

Initiation of cytokinesis is the final crucial event of the cell cycle, and it must be tightly coordinated with completion of chromosome segregation. In fission yeast, the timing of cytokinesis is regulated by the SIN signaling pathway. Studies performed with this organism have lent support to the notion that SIN signaling is negatively regulated by Cdk1 activity, and that maximum SIN activation depends upon Cdk1 inactivation. For example, an early study showed that inactivation of Cdk1 by cdc2-33 mutation promotes cytokinesis and septum formation even in metaphase-arrested cells induced by overexpression of the spindle assembly checkpoint protein Mad2 (He et al., 1997). Subsequently, another study confirmed that loss of Cdc2 activity in these metaphase-arrested cells was sufficient to allow SIN kinase complex Sid1–Cdc14 to localize at one SPB and to promote cytokinesis onset (Guertin et al., 2000). More recently, it has been shown that inactivation of Cdk1 using an analogue-sensitive mutant allele cdc2-as promotes the asymmetrical distribution of SIN proteins (Sid1 and Cdc7) to the spindle poles, and the recruitment of the most downstream SIN component Mob1 and β-(1,3) glucan synthase to the contractile ring (Dischinger et al., 2008). Conversely, high Cdk1 activity caused by increased expression of nondegradable Cdc13 (cyclin B) blocked cells in anaphase and prevented SIN kinase Cdc7 from becoming asymmetrically localized to one SPB during anaphase B; it also blocked all subsequent relocalization events in the SIN and therefore inhibited septation (Chang et al., 2001; Yamano et al., 1996).

The negative-feedback control of cytokinesis by Cdk1 activity is not restricted to fission yeast and is conserved in all eukaryotes from yeasts to vertebrates. In another model organism, budding yeast S. cerevisiae, inactivation of Cdk1 in cells arrested by expression of a non-degradable CLB2 promotes mitotic exit and cytokinesis (Ghiara et al., 1991). Inactivation of Cdk1 is also required for localization of the vesicle fusion machinery to the bud neck during mitotic exit and cytokinesis (VerPlank and Li, 2005). Recently, it has been shown that Cdk1 is asymmetrically localized to mother SPBs during early anaphase and inhibits Cdc15 (orthologous to S. pombe Cdc7) and thus MEN signaling activation (König et al., 2010). In dividing cultured human cells, inhibition of Cdk1 by selective small-molecule inhibitors also caused premature cytokinesis, indicating that Cdk1 activity is necessary and sufficient for maintaining the mitotic state of the cells and preventing premature cytokinesis (Vassilev et al., 2006). However, the detailed mechanism involved in this control is not entirely understood.

Thus, the data obtained from all model organisms studied to date point to a universal requirement for inactivation of CDK1 to promote cytokinesis. This control mechanism is regarded as the ideal coupling between nuclear and cell division. However, so far, only two budding yeast proteins involved in mitotic exit and cytokinesis, CDC15 and MOB1, have been confirmed to be phosphorylated by Cdk1 (Jaspersen and Morgan, 2000; König et al., 2010). Whether homologues of these proteins in other organisms (such as Cdc7 and Mob1 in fission yeast, or Mst2 and Mob1 in Drosophila and vertebrates (reviewed by Doxsey et al., 2005) are also phosphorylated by Cdk1 remains to be demonstrated. Further studies will be required to identify and define the complete repertoire of Cdk1 substrates influencing the timing of cytokinesis in fission yeast and higher eukaryotes.

Dnt1 is not the first nucleolar protein identified in fission yeast to be involved in regulating both Cdk1 activity and SIN signaling. Another nucleolar protein, the Cdc14 phosphatase family member Clp1, is a well-studied example of such regulation. At the end of mitosis, Clp1 reverses Cdk1-dependent phosphorylation of Cdc25 and destabilizes and inactivates Cdc25 by promoting recognition of Cdc25 by two ubiquitin E3 ligases, the anaphase-promoting complex/cyclosome (APC/C) (RING-type) and Pub1/Pub2 (HECT-type) (Esteban et al., 2004; Esteban et al., 2008; Wolfe and Gould, 2004). Thus, stabilized Cdc25 allows clp1Δ cells to delay mitotic Cdk1 inactivation at the end of mitosis and advance into G2 (Wolfe and Gould, 2004). Moreover, SIN signaling is probably antagonized by persistent Cdk1 activity. In addition to its role in affecting timely Cdk1 inactivation, Clp1 also directly contributes to robust and successful cytokinesis, particularly when the cell division machinery (such as the actomyosin ring) is slightly perturbed or damaged (Mishra et al., 2004). A recent study revealed that Clp1 is tethered at the contractile ring (CR) through its association with the anillin-related protein Mid1, and regulates CR protein dynamics and mobility by dephosphorylating CR components Cdc15 and possibly Myo2 (myosin II heavy chain) (Clifford et al., 2008).

Cdc14-family phosphatases are conserved in all eukaryotes examined so far, but have been best studied in both budding and fission yeasts. The negative regulation of Cdk1 at late mitosis represents a conserved feature of Cdc14 phosphatase family members (for reviews see D'Amours and Amon, 2004; Krapp et al., 2004; Mocciaro and Schiebel, 2010; Stegmeier and Amon, 2004). Budding yeast ScCdc14 and fission yeast Clp1 are regulated in part by their cell-cycle-dependent changes in localization, with both proteins thought to be sequestered and inactivated in the nucleolus during interphase. The activity of ScCdc14 is controlled by its association with the competitive inhibitor Cfi1/Net1, and only during anaphase is ScCdc14 released from its inhibitor to spread into the nucleus and cytoplasm, allowing the dephosphorylation of its substrates (reviewed by Mocciaro and Schiebel, 2010; Stegmeier and Amon, 2004). In contrast to ScCdc14, both Clp1 and mammalian Cdc14B are released from the nucleolus upon entry into mitosis (Cho et al., 2005; Cueille et al., 2001; Nalepa and Harper, 2004; Trautmann et al., 2001), but how this process is regulated is currently unknown. Although the amino acid sequence of Dnt1 shows weak similarity to that of budding yeast Net1/Cfi1 and it has been speculated that Dnt1 might function as a nucleolar inhibitor of Clp1 (Jin et al., 2007), the present study demonstrates that Dnt1 actually influences Wee1 stability and thus Cdk1 activity. Therefore, there are at least two parallel pathways emerging from the nucleolus which converge at Cdk1 to regulate SIN signaling: one including both Dnt1 and Wee1, and another involving Clp1-controlled Cdc25 (Fig. 6).

Our findings also suggest that SCF-mediated degradation of Wee1 is probably conserved in fission yeast, although the direct evidence of Wee1 ubiquitylation by SCF is currently lacking. The elucidation of the specific mechanism by which Dnt1 controls degradation of Wee1 remains a mystery and merits further investigation.

Fission yeast strains used in this study were constructed by standard techniques (Moreno et al., 1991) and are listed in supplementary material Table S1. S. pombe strains were grown in rich medium [yeast extract (YE)] or Edinburgh minimal medium (EMM) with appropriate supplements (Moreno et al., 1991). EMM with 5 µg/ml thiamine was used to repress expression from the nmt1 promoter. YE containing 100 mg G418 (Sigma) or nourseothiricin (clonNAT) (Werner BioAgents) per liter was used for selecting kan^R and nat^R cells, respectively. For serial-dilution drop tests for growth, three serial 10-fold dilutions were made and 5 µl of each was spotted on plates with the starting cell number of 10⁴. Cells were pre-grown in liquid YE or EMM at 25°C until they reached exponential phase and were then spotted onto YE or EMM plates at the indicated temperatures and incubated for 3 to 5 days before scanning. In block-and-release experiments performed with cdc25-22 mutant strains, the cells were grown in YES medium to an A₆₀₀ of 0.2 at 25°C (permissive temperature), shifted to 36°C for 3.5 hours, and then released from the growth arrest by transfer back to 25°C.

Firefly and Renilla Luciferase lacking start codons were cloned from the pGL3 and pRL-TK plasmids (Promega) into pFA6a plasmid carrying either kan^R or nat^R selection marker. These plasmids were used to add C-terminal epitope tags to the endogenous locus to create Ade4-F.luciferase (kan^R) and Wee1-R.luciferase (nat^R) alleles using PCR-based gene targeting (Bähler et al., 1998). Candidate transformants by homologous recombination were selected by growth on selective plates containing G418 or ClonNAT and subsequently confirmed by PCR analysis.

Yeast cell extracts were prepared in lysis buffer (120 mM Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, 8M urea, 0.6 M β-mercaptoethanol) by glass bead disruption using FastPrep homogenizer (MP Biomedical), and denatured in 2× SDS sample buffer (100 mM Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, 0.2% Bromophenol Blue, 0.6 M β-mercaptoethanol) before being resolved on 8–12% SDS-PAGE gels. For detection, mouse monoclonal anti-GFP (Roche), rabbit polyclonal anti-Myc (GeneScript), rat monoclonal anti-HA (Roche) and rabbit polyclonal anti-Cdc2 (Tyr¹⁵) (sc-7989-R, Santa Cruz Biotechnology) were used as the primary antibodies (1∶1000–1∶2000 dilutions). Cdc2 or GFP-atb2 was detected using rabbit polyclonal anti-PSTAIRE (sc-53, Santa Cruz Biotechnology) or mouse monoclonal anti-GFP (Roche), respectively, as loading controls (1∶1000 dilution). Goat anti-mouse, anti-rabbit or anti-rat conjugated to horseradish peroxidase (Pierce) was used as the secondary antibody at 1∶10,000 dilution. Membranes were developed with ECL western blotting reagents (Pierce). Protein levels were quantified using ImageJ software (National Institutes of Health, Bethesda, MD) with Cdc2 or GFP-atb2 as a loading control.

Yeast strains carrying both Ade4-F.luciferase (kan^R) and Wee1-R.luciferase (nat^R) were used. Exponential phase cultures were pre-grown in liquid YE at 30°C, 25°C or shifted from 25°C to 37°C for 4 hours, and 1–10×10⁷ cells were collected and washed once with sterile water. Cell lysates were prepared in 100–200 µl cold 1× passive Lysis Buffer (PLB) by glass bead disruption using FastPrep homogenizer. Firefly and Renilla luciferase activities in cell lysates were measured using the Dual-Reporter Assay System (Promega) following instructions provided by the manufacturer.

Total RNA was isolated from yeast cells grown in YES liquid medium to an A₆₀₀ of <0.4 and disrupted with FastPrep homogenizer by using the TRI Reagent Solution (TRIZOL) (TaKaRa Bio) and following the manufacturer's instructions. RT was performed with Prime-Script RT reagent Kit (TaKaRa) at 37°C for 15 minutes followed by treatment at 85°C for 5 seconds. Primer pairs to detect wee1⁺ and act1⁺ transcripts by PCR were as follows: wee1-2189 bp-F, 5′- GTGAATACATTGCGCGTAAG-3′ and wee1-2540 bp-R, 5′-CTACGCATTTCAACCCAGC-3′ (amplifying a 400 bp fragment); act1-F, 5′-GTATGCCTCTGGTCGTACCAC-3′ and act1-R, 5′-CAATTTCACGTTCGGCGGTAG-3′ (amplifying a 200 bp fragment). PCR products were then resolved by 5% Tris-acetate-EDTA PAGE (TAE-PAGE). The gels were stained with 0.25% AgNO₃ and scanned and images were processed with Adobe Photoshop software and the intensities of PCR products were quantified using ImageJ software with act1⁺ transcript as an internal control.

GFP-fusion proteins were observed in cells after fixation with cold methanol or in live cells. For DAPI (4′, 6-diamidino-2-phenylindole, Sigma) staining of nuclei, cells were fixed with cold methanol, washed in PBS and resuspended in PBS plus 1 µg/ml DAPI. To determine cell size at division the yeast strains were grown in either EMM2 or YES medium to an A₆₀₀ of <0.5 and treated with Calcofluor white (CW, Sigma) at 50 µg/ml final concentration, which specifically stains cell wall and septum. A minimum of 100 septated cells were scored for each mutant. Photomicrographs were obtained using a Nikon 80i fluorescence microscope coupled to a cooled CCD camera (Hamamatsu, ORCA-ER) and image processing, analysis and cell length measurement were carried out using Element software (Nikon) and Adobe Photoshop.

All experiments were repeated at least three times with similar results. In order to determine statistical significance of our data, two tailed Student's t-tests were performed, and P<0.05 was considered statistically significant.

We are grateful to Drs Kathleen L. Gould, Mohan K. Balasubramanian, Dannel McCollum, Jianhua Liu, Paul Russell, Iain Hagan, Takashi Toda, Janni Petersen, James Moseley and the Yeast Genetic Resource Center Japan (YGRC) for providing yeast strains or plasmids.