Flp1, a fission yeast orthologue of the S. cerevisiae CDC14 gene, is not required for cyclin degradation or rum1p stabilisation at the end of mitosis

In the budding yeast Saccharomyces cerevisiae, the phosphoprotein phosphatase encoded by the CDC14 gene plays a central role in exit from mitosis and completion of the cell cycle (Hoyt, 2000). The CDC14 gene is essential (Wan et al., 1992), and heat-sensitive mutants arrest with an elongated mitotic spindle, separated chromosomes and high levels of CDC28-CLB kinase activity (Fitzpatrick et al., 1998; Jaspersen et al., 1998). During interphase, Cdc14p is sequestered in the nucleolus, where it is bound to Net1p as part of the RENT complex (Shou et al., 1999; Straight et al., 1999; Visintin et al., 1999). Release of Cdc14p from the nucleolus at the end of mitosis permits increased expression and stabilisation of the CDC28-CLB kinase inhibitor p40^sic1, and the activation of APC/C^Cdh1p-dependent ubiquitination of B-type cyclins, thereby promoting their destruction. Together, these events reduce CDC28-CLB kinase activity, permitting exit from mitosis, and cytokinesis (Visintin et al., 1998).

Release of Cdc14p from the nucleolus, and exit from mitosis requires the activity of the mitotic exit network (MEN) proteins (Visintin et al., 1999). Loss-of-function MEN mutants arrest at the end of mitosis with a phenotype similar to that of cdc14 mutants. Increased expression of CDC14 can rescue MEN mutants, but not vice-versa, leading to the suggestion that the essential function of the MEN is to mediate release of Cdc14p from the nucleolus (Jaspersen et al., 1998; Visintin et al., 1999). Functional homologues of Cdc14p have been identified in higher eukaryotes, although their role remains enigmatic (Li et al., 1997).

In the fission yeast Schizosaccharomyces pombe, the orthologues of the MEN genes control the onset of septum formation, and are collectively referred to as the septation initiation network (SIN) (Balasubramanian et al., 2000; Le Goff et al., 1999a; Sawin, 2000). Activation of the spg1p GTPase switch is central to signalling the onset of septum formation, which also involves the concerted action of four protein kinases (cdc7p, sid1p, sid2p and plo1p); cdc14p, which binds to sid1p; mob1p, which binds to sid2p; and sid4p, which acts as a scaffold for localisation of these proteins to the spindle pole body. (Note that fission yeast cdc14p bears no structural similarity to its S. cerevisiae namesake.) Cdc11p encodes the fission yeast orthologue of the S. cerevisiae nud1 gene product (A. Krapp and V.S., unpublished). Mutants defective in SIN signalling make a medial F-actin ring at the onset of mitosis to mark the division site but do not make a septum at the end of mitosis, becoming elongated and multinucleated (Chang and Gould, 2000; Fankhauser and Simanis, 1993; Fankhauser and Simanis, 1994; Guertin et al., 2000; Hou et al., 2000; Nurse et al., 1976; Salimova et al., 2000; Schmidt et al., 1997; Sparks et al., 1999). Failure to turn off SIN signalling at the end of mitosis results in multiple rounds of septum formation without cell cleavage (Cerutti and Simanis, 1999; Minet et al., 1979; Ohkura et al., 1995; Schmidt et al., 1997; Song et al., 1996). Signalling via the SIN is negatively effected by a two-component GTPase-activating protein comprised of the products of the byr4 and cdc16 genes (Furge et al., 1998). Both genes are essential (Fankhauser et al., 1993; Song et al., 1996). Loss of cdc16p function gives rise to two kinds of septated cell, depending upon the cell cycle stage of the inactivation: type I cells, which are binucleate, and undergo multiple rounds of septum formation after mitosis, and type II, which are mononucleate, and probably result from inactivation of cdc16p in G₁ (Minet et al., 1979). The ultimate target of the SIN remains enigmatic. However, the budding yeast paradigm suggested an attractive hypothesis, which we have tested, that it might be a phosphoprotein phosphatase analogous to Cdc14p. In this paper, we present the characterisation of flp1, a fission yeast orthologue of the S. cerevisiae CDC14 gene. Our data indicate that the flp1⁺ gene performs a different role in cell cycle progression to that of S. cerevisiae Cdc14p.

Standard methods were used for manipulation of DNA (Sambrook et al., 1989). Fission yeast were grown and manipulated according to standard protocols in either yeast extract (YE) or minimal (M) medium, containing appropriate supplements (Moreno et al., 1991). The S. pombe strains used in this study are from the Simanis, Moreno or Bueno lab collections. The following strains were obtained from other labs: sid4-SA1 (Kathy Gould, Nashville, TN), sid2-250 and sid1-239 (Dan McCollum, University of Massachusetts), rng2-D5 (Mohan Balasubramanian, IMA, Singapore). Since deletion of ste9⁺ or rum1⁺ causes sterility, all the crosses involving a deletion of these genes were done by transforming with pREP3X-ste9⁺, pREP3X-rum1⁺, and the double mutants were checked subsequently to ensure that the plasmid had been lost. Yeast transformation was carried out using the lithium acetate transformation protocol (Norbury and Moreno, 1997). Induction synchrony by cdc25-22 arrest-release was performed as described (Moreno et al., 1989). For induction of the nmt1 promoter, a culture growing exponentially in medium containing thiamine was washed twice and resuspended in medium without thiamine. The episomal REP vectors, pDW232 and the integrating nmt1 expression vector pINT5 have been described previously (Basi et al., 1993; Fankhauser and Simanis, 1994; Weilguny et al., 1991). Gene deletion and epitope tagging were carried out as described (Bahler et al., 1998b).

The full ORF was amplified by PCR from a genomic DNA library in pUR19 (Barbet et al., 1992), digested with SmaI and cloned into the SmaI sites of pDW232 and pREP vectors. Primers used for amplification: forward, ACTGCCCCGGGTTCGCAATTACTTGTCTGATGGA; reverse, GCTCGCCCGGGAACCAGTAATTACAGGTTTATTAAG.

The flp1 null and C-terminally tagged (GFP and 3HA) strains were constructed by direct chromosome integration of PCR fragments generated using plasmid pFA6a-kanMX6 as a template. Forward primer for deletion: CCACCAACACCCAGGTACACAATTTAGAACTCAACCATTACGGGTTTGACGAATATAGACGAGATTCGCAATTACTTGTCTGCGGATCCCCGGGTTAATTAA. Forward primer for tagging: GTGTTAGCATGTCATCACTTAACAATACTTCTAATGGCCGTGTTGCTAAACCTAAGCCTTCTAAAAGCCGGCTAATTTCTCGGATCCCCGGGTTAATTAA. Reverse primer for tagging and deletion GGTGCGCTAAATCAGGGAATATTTGTAAAGTTAATTAATGAAAAATTATGCAGGGTTGACACAGTATAATTCAAAGTTAGTGAATTCGAGCTCGTTTAAAC.

PCR fragments were gel-purified and introduced into a leu1-32 h^- strain following the protocol described previously (Bähler et al., 1998). Transformants were selected on YE G418 plates (100 mg/l). Correct integration was verified by PCR and Southern hybridization for the deletion and by PCR and western blotting for tagged strains.

For creation of flp1p (C286S) the following pair of oligonucleotides were used for PCR amplification of the gene: ATTGCTGTTCATTCTAAAGCAGGGCTC, GAGCCCTGCTTTAGAATGAACAGCAAT. The presence of the desired mutation was verified by sequencing.

Antisera and tagged strains permitting detection of cdc15, cdc7, spg1, cdc14, sid1, sid2, sid4, mob1 and ste9 have been described previously (Blanco et al., 2000; Chang and Gould, 2000; Fankhauser et al., 1995; Fankhauser and Simanis, 1994; Guertin et al., 2000; Hou et al., 2000; Moreno et al., 1990; Salimova et al., 2000; Sohrmann et al., 1996; Sohrmann et al., 1998; Sparks et al., 1999). Rabbit antiserum against S. cerevisiae nop1p (fibrillarin) was a gift from Susan Gasser (University of Geneva).

Approximately 10⁷ cells were collected by centrifugation, washed once with water, fixed in 70% ethanol and processed for flow cytometry or DAPI staining, as described previously (Moreno et al., 1991). A Becton-Dickinson FACScan was used for flow cytometry. To estimate the proportion of G₁ cells we determined the percentage of cells with a DNA content less than a value midway between 1C and 2C. The mitotic index was determined by counting the percentage of anaphase cells (cells with two nuclei and without a septum) after DAPI staining. The septation index was determined by counting the percentage of cells with septum after calcofluor staining. Cell number was determined using a Casy® cell number counter. DAPI-Calcofluor staining, and staining for F-actin and tubulin staining were done as described previously (Balasubramanian et al., 1997; Hagan and Hyams, 1988; Marks et al., 1986; Moreno et al., 1991). For examination of GFP-tagged proteins in living cells, TILLvisION software (v3.3; TILL Photonics GmBH) was used to analyse data captured with an IMAGO CCD camera mounted on an Olympus IX70 microscope. Deconvoultion was performed with BitPlane software. Images were assembled in Adobe PhotoShop 5.5 and PowerPoint 97. For immunofluorescence of flp1p-GFP, cells were fixed and processed as previously described (Salimova et al., 2000).

Protein extracts were prepared from 3-5×10⁸ cells in exponential phase, that had been collected by centrifugation and frozen on dry ice. All subsequent manipulations were done on ice or in the cold room (4°C). For immunoprecipitation, soluble protein extracts were prepared by vortexing with glass beads in HEN buffer (50 mM Hepes pH 8.0; 150 mM NaCl; 5 mM EDTA; 1 mM EGTA; 50 mM β-glycerophosphate with inhibitors: 0.1 mM sodium orthovanadate, 50 μg/ml leupeptin, 1% aprotinin; 1 mM DTT; 1 mM PMSF). Beads were washed by brief vortexing in the same buffer containing 1% NP-40. Cell extracts were clarified by two successive centrifugations. Protein concentration was measured using Bradford assay (BioRad). For each immunoprecipitation, 2-3 mg of soluble protein was incubated overnight with 10 μl of either 9E10 or 12CA5 monoclonal antibodies covalently coupled to the sepharose-Protein G beads (Sigma; P3296) (∼2 μg of Ab). Beads were washed three times with 1 ml of HEN-NP-40 buffer (by pelleting in a microfuge for 5 seconds), resuspended in FRB loading buffer.

Total protein extracts were prepared from 3×10⁸ cells collected by centrifugation, washed in Stop buffer (150 mM NaCl, 50 mM NaF, 10 mM EDTA, 1 mM NaN3 pH 8.0) and resuspended in 25 μl of RIPA buffer (10 mM sodium phosphate, 1% Triton X-100, 0.1% SDS, 10 mM EDTA, 150 mM NaCl, pH 7.0) containing the following protease inhibitors: 10 μg/ml leupeptin, 10 μg/ml aprotinin, 10 μg/ml pepstatin, 10 μg/ml soybean trypsin inhibitor, 100 μM 1-chloro-3-tosylamido-7-amino-L-2-heptanone (TLCK), 100 μM N-tosyl-L-phenyalanine chloromethyl ketone (TPCK), 100 μM PMSF (phenylmethylsulfonyl fluoride), 1 mM phenanthroline and 100 μM N-acetyl-leu-leu-norleucinal. Cells were boiled for 5 minutes, broken using 750 mg of glass beads (0.4 mm Sigma) for 15 seconds in a Fast-Prep machine (Bio101 Inc.) and the crude extract was recovered by washing with 0.5 ml of RIPA. Protein concentration was determined by BCA protein assay kit (Pierce).

For western blots, 75 μg of total protein extract was run on a 12% SDS-PAGE gel, transferred to nitrocellulose and probed with rabbit affinity purified anti-ste9-C-ter (1:200), SP4 anti-cdc13 (1:250) and anti-rum1 (1:250) polyclonal antibodies. Goat anti-rabbit or goat anti-mouse conjugated to horseradish peroxidase (Amersham) (1:3,500) was used as secondary antibody. Mouse TAT1 anti-tubulin monoclonal antibodies (1:500) and goat anti-mouse conjugated to horseradish peroxidase (1:2,000) as secondary antibody was used to detect tubulin as loading control. Immunoblots were developed using the ECL kit (Amersham) or Super Signal (Pierce). CIP treatment of immunoprecipitates was performed as previously described (Fankhauser et al., 1995EF20).

A BLAST search of the fission yeast database with the S. cerevisiae CDC14 gene open reading frame (ORF) revealed the presence of a single ORF, SPAC1782.09c, that had significant homology to Cdc14p outside the phosphoprotein phosphatase catalytic domain. A CLUSTAL-W alignment of the predicted proteins, together with putative orthologues from human, Drosophila and C. elegans is shown in Fig. 1. We have named the gene flp1 (cdc fourteen like phosphatase).

To determine whether the flp1 gene is essential for cell viability and proliferation, one copy of the gene was replaced in a diploid by the kanMX6 cassette (Fig. 2A), which was generated by PCR amplification using flp1-specific oligonucleotides (see Materials and Methods). Dissection of tetrads indicated that all gave rise to four colonies, of which two were resistant to Geneticin, and two were not. To confirm correct deletion of the flp1 gene, the haploid flp1::kanMX6 cells were checked for the absence of the flp1⁺ gene by Southern blotting (Fig. 2B). We conclude that the S. pombe flp1 gene is not essential for cell viability or proliferation.

Although viable, and capable of colony formation at all temperatures, flp1::kanMX6 cells are not phenotypically identical to wild-type (Fig. 2C,D). Measurement of the length of septated cells at 25°C indicated that flp1::kanMX6 cells divide at an average size of 11.5 μm, compared with 14 μm for wild-type. The cells are thus advanced into mitosis and may be considered `semi-wee'. At 36°C, multi-compartmented cells (Fig. 2D, cell 1), multinucleated postmitotic cells (Fig. 2D, cell 2, C, cell 1), anucleate compartments (Fig. 2D, cell 3), and `cut' nuclei (Fig. 2D, cell 4) were all observed. Staining with Rhodamine-conjugated Phalloidin indicated that no medial ring was present in the post-mitotic, multinucleated cells, indicating that this phenotype does not result from activation of the S. pombe morphology checkpoint (Liu et al., 2000). Together, these classes of aberrant cells represented approximately 8.5% of the population. Similar defects were also observed at 25°C, but these cells represented only 2% of the population. Thus, although they are viable, flp1::kanMX6 cells display a defect in septum formation and cleavage, and are slightly advanced into mitosis.

Staining of cells with TAT-1 revealed a normal pattern of interphase microtubules and mitotic spindles (data not shown). Staining with Rhodamine-conjugated Phalloidin revealed patterns of F-actin staining similar to wild-type cells (data not shown). Examination of the SIN components cdc7p, mob1p, plo1p, spg1p and sid2p showed that their localisation was normal in flp1::kanMX6 cells (not shown). The localisation of the medial ring components cdc15p and mid1p was also normal. Both cdc15p and mid1p undergo changes in phosphorylation during mitosis (Fankhauser et al., 1995; Sohrmann et al., 1996); these changes occurred normally in flp1::kanMX6 cells, indicating that they are not substrates of flp1p, or that another phosphatase can substitute for flp1p in these cells (not shown).

Northern blotting of total RNA extracted from synchronised cells indicated that the steady state level of flp1 mRNA does not vary significantly throughout the cell cycle (data not shown). To examine whether flp1p varies in level during the cell cycle, cells carrying the flp1-HA allele were synchronised by arrest release of cdc25-22 and protein samples were prepared as cells passed through mitosis and cytokinesis (Fig. 3A). Western blotting showed that, although the steady state level of flp1p-HA did not change significantly, there was a marked alteration in the apparent molecular weight of the protein at the time corresponding to anaphase, medial ring formation and septation (Fig. 3B, 30-80 minutes). As most cells completed cleavage and entered the next cell cycle, flp1p-HA returned to the faster-migrating form seen in G₂ arrested cells (Fig. 3B, 100-160 minutes). Flp1p-HA was immunoprecipitated from cell extracts prepared 45 minutes after release from the G₂ arrest, and the immunoprecipitate treated with alkaline phosphatase. This treatment shifted most of the protein back to the faster-migrating form (Fig. 3C), confirming that the slower-migrating forms are the result of phosphorylation.

To study the in vivo localisation of flp1p, the chromosomal copy of the gene was modified to add either Green Fluorescent protein (GFP) or the 12CA5 influenza virus hemagglutinin epitope tag (HA) to the C-terminus of flp1p. Both tagged proteins were considered to be functional since the tagged flp1 allele did not show any of the strong genetic interactions shown by the flp1::kanMX6. Observation of both living and fixed cells demonstrated that in interphase cells the flp1p-GFP was localised in the nucleolus and on the spindle pole body (Fig. 4Aa,B,C,D). The nucleolar staining was not uniform: dots and more intensely staining regions were observed (Fig. 4Aa). The nature of these structures is unknown. Although the nuclear signal of flp 1p-GFP was predominantly located in the non-DAPI staining (nucleolar) region of the nucleus, a weaker signal was also observed in the DAPI-staining region of the nucleus in fixed cells (Fig. 4B). The localisation to the nucleolus was confirmed by co-localisation with the nucleolar marker fibrillarin (nop1; Aris and Blobel, 1991; Henriquez et al., 1990; Potashkin et al., 1990) (Fig. 4C), while localisation to the SPB was confirmed by staining with antibodies against sad1p (Fig. 4D) (Hagan and Yanagida, 1995).

In early mitotic cells, flp 1p-GFP was observed on both spindle pole bodies (Fig. 4Ab), along the mitotic spindle (Fig. 4Ac,d,e), and in the medial ring (Fig. 4Ad,e). In some cells, the staining was continuous along the short spindle (Fig. 4Ad), while in others it was discontinuous (Fig. 4Ae). The reason for this difference is presently unclear. In early mitotic cells, the flp 1p-GFP signal was present throughout the nucleus. Previous studies (Hirano et al., 1989) of the localisation of nuc 1p (the large subunit of RNA polymerase I) have shown that in mitotic cells the nucleolar domain can be clearly distinguished from the DAPI-staining, chromatin region. We therefore conclude that flp 1p-GFP leaves the nucleolar region early in mitosis. This is in contrast to the result observed with S. cerevisiae Cdc 14p, which appears to leave the nucleolus only at the end of mitosis (Shou et al., 1999).

Treatment of exponentially growing flp 1-GFP cells with Latrunculin A resulted in the appearance of centrally located flp 1p-GFP dots in mitotic cells (Fig. 4G1,2). Interphase localisation to the spindle pole body and nucleolus was unaffected (Fig. 4G1i, 3). Staining of fixed flp 1::kanMX6 nda3-KM311 cells that had been arrested by incubation at 19°C, indicated that, in early mitosis, flp 1p-GFP co-localised with the F-actin ring (Fig. 4E), and was dispersed throughout the nucleus (Fig. 4F).

In late anaphase cells, flp 1p-GFP remained associated with both poles of the mitotic spindle, although the signals were weaker than in early mitotic cells (Fig. 4Af, and the medial ring (Fig. 4Af, 4Ag). The spindle staining became concentrated in the spindle mid-zone (Fig. 4Ag). During septum synthesis, flp 1p-GFP was seen in the contractile ring at the leading edge of the division septum, the spindle pole bodies, and the nucleolus (Fig. 4Ah). Upon completion of the division septum (which is coincident with late G₁/S-phase in terms of the nuclear cycle), flp 1p-GFP was observed uniquely on the spindle pole body and in the nucleolus, as was the case in late interphase cells (Fig. 4Ai).

Previous studies have demonstrated that, in S. cerevisiae, MEN function is required for Cdc 14p to leave the nucleolus at the end of mitosis (Shirayama et al., 1999; Shou et al., 1999; Visintin et al., 1999). To test whether SIN function is required for the cell cycle relocalisations described above, the flp 1-GFP allele was introduced into SIN null allele backgrounds. The use of heat-sensitive SIN mutants was precluded because the spindle and medial ring staining of flp 1p was found to be labile in cells fixed at high temperatures (>32°C), as has been described previously for some other SIN proteins (Salimova et al., 2000; Sparks et al., 1999). Nucleolar and spindle pole body localisation were unaffected. The flp 1-GFP allele was crossed to haploid SIN null strains in which the deletion is rescued by the presence of an episomal plasmid. Spores were germinated at 25°C selecting for the null allele and the flp 1-GFP allele (see Materials and Methods). The flp 1p-GFP structures present in multinucleated cells were examined. Interphase nucleolar and spindle pole body staining were observed in germinating cdc7 null (Fig. 5A1), cdc14 null (Fig. 5B2) and spg1 null (Fig. 5C3) spores. Likewise, all three mutants showed spindle (Fig. 5A2,B1,C1,2) and medial ring (Fig. 5A2,5B1,5C1) staining although, in the spg1 null cells, the rings appeared less well defined that those seen in cdc7 null cells. Note that some SIN mutants do not always form multiple medial rings in the later nuclear cycles after shift (see for example, Balasubramanian et al., 1998; Schmidt et al., 1997), explaining why in some cases only a single ring is seen. Similar results were obtained using thermosensitive alleles of sid2 and sid1 that die at 29°C (not shown). We therefore conclude that SIN signalling is not required for the changes in subcellular localisation of flp 1p-GFP.

The effect of deregulation of SIN signalling upon flp 1p-GFP localisation was also examined. Inactivation of cdc 16p, which is part of the GTPase activating protein (GAP) for spg1p (Furge et al., 1998) results in multiple rounds of septum formation without cell cleavage (Minet et al., 1979). In germinating cdc16 null spores, type II cells showed an interphase-like localisation of flp 1p-GFP (Fig. 5D5). By contrast, type I cells showed no discrete localisation of flp 1p-GFP (Fig. 5D1-4). Occasionally, one or more of the cell compartments contained what appeared to be aggregates of flp 1p-GFP (Fig. 5D2, arrows).

Since type I cells have entered mitosis, whereas type II cells have not (see Introduction), one explanation for these data could be that once cells have entered mitosis, SIN signalling must be attenuated for flp 1p-GFP to return to the interphase configuration once septation has been completed. To test this hypothesis, the localisation of flp 1p-GFP was examined in cells where the lethality of a cdc16 null allele is rescued by attenuation of the SIN (Fournier et al., 2001; Salimova et al., 2000). The double mutant sid2-1 cdc16::ura4⁺ is viable at 25°C, although some errors of septum formation still occur (Fournier et al., 2001). Normal interphase (Fig. 5E1-4), and mitotic (Fig. 5E5) distribution of flp 1p-GFP was observed, although occasional aggregates of the protein were also seen (Fig. 5E2, arrows). A normal interphase distribution of flp 1p-GFP was also seen in cells that had formed a second division septum (Fig. 5E1,2) indicating that the presence of a additional septa does not per se inhibit flp 1p-GFP localisation in a SIN-independent manner. Together, these data are consistent with the hypothesis that SIN signalling must be inactivated or attenuated for flp 1p-GFP to return to the nucleolus after septation is completed.

In S. cerevisiae, Cdc 14p is thought to be the essential target of the MEN (Jaspersen et al., 1998; Visintin et al., 1999). To investigate whether there were any interactions between flp 1 and the SIN, double mutants of flp 1::kanMX6 with heat-sensitive SIN mutants were constructed by tetrad dissection. It was found that the double mutants flp 1::kanMX6 cdc 7-24 and flp 1::kanMX6 mob 1-R4 were inviable at 25°C. A strong reduction of restrictive temperature was noted for other double mutant combinations (see Table 1). In most cases, the double mutants already showed a significant septation defect at the permissive temperature: the example of the mutant sid2-250 flp 1::kanMX6 is shown in Fig. 6A. In addition, at 36°C, the double mutants accumulated more nuclei than the single SIN mutant, and the nuclear cycles were no longer synchronised (Fig. 6A). The exception to this among the SIN mutants was the double mutant flp 1::kanMX6 sid1-239, which did not display any additive effects. Interestingly, many multinucleated, non-septated cells were observed when the double mutant cdc 16-116 flp 1::kanMX6 was grown at 25°C. This defect was alleviated partially at 29°C, and at 36°C cells showed the multiseptated phenotype characteristic of cdc 16-116 (Fig. 6B). No additive effects were observed when double mutants of flp 1::kanMX6 with a number of mutants defective in medial ring assembly or function were analysed (Table 1).

Increased expression of the S. cerevisiae CDC14 can rescue some of the MEN mutants (Jaspersen et al., 1998; Visintin et al., 1999). By contrast, increased expression of the flp 1 gene from a multicopy plasmid did not rescue any of the heat-sensitive SIN mutants tested (cdc7-24, cdc11-136, cdc14-118, spg1-B8, sid1-239, sid2-250, sid4-SA1, mob1-R4 and cdc16-116), whether expressed at low (from its own promoter in pDW232, or pREP3 non-induced), or high levels (pREP3, induced) (data not shown).

Increased expression of some SIN genes can induce septation independently of entry into mitosis (Ohkura et al., 1995; Schmidt et al., 1997). Expression of either plo1 (not shown) or spg1 (Fig. 6C) from the nmt1 promoter induced septation in flp1::kanMX6 cells. Likewise, increased expression of septation inhibitor byr4 (Song et al., 1996) blocked septum formation in flp1::kanMX6 cells, as it does in wild-type (Fig. 6D). Thus, we conclude that flp 1p is not an essential effector of spg1-mediated signal transduction. However, it was noted that increased expression of spg1 induced fewer multiseptated cells in a flp1::kanMX6 background (Fig. 6C), suggesting that the efficiency of SIN signalling was reduced in the absence of flp1p function.

S. cerevisiae Cdc14p promotes degradation of B-type cyclins at the end of mitosis, and also favours accumulation of p40^sic1 (Visintin et al., 1998). The effects of flp1::kanMX6 upon degradation of the B-type cyclin cdc13p and accumulation of rum1p were investigated by arrest-release of cdc25-22 and cdc25-22 flp1::kanMX6 strains, extraction of proteins, and western blotting. Western blotting with antisera recognising either cdc13p or rum1p indicated that both proteins accumulated and were then degraded in the flp1::kanMX6 background with similar timing to that seen in flp1⁺ cells (Fig. 7A). The kinetics of release from the cdc25-22 block were similar in the flp1⁺ and flp1::kanMX6 backgrounds (Fig. 7B). The timing of disappearance of cdc13p-myc13 was also examined in exponentially growing flp1⁺ and flp1::kanMX6 cells. Fixed cells were stained with 9E10 and the fluorescence due to cdc13p compared in cells at similar stages in mitosis. No significant differences were noted between the flp1⁺ and flp1::kanMX6 backgrounds (Fig. 7C). We conclude that flp1p activity is not required for accumulation or degradation of either cdc13p or rum1p. Consistent with this, the flp1::kanMX6 cells are capable of mating.

The APC/C accessory factor Cdh1p is thought to be dephosphorylated by Cdc14p at the end of mitosis, thereby allowing it to associate with the APC/C to promote B-type cyclin degradation (Jaspersen et al., 1999; Visintin et al., 1998). The S. pombe homologue of Cdh1p, ste9p (Kitamura et al., 1998; Yamaguchi et al., 1997), is also dephosphorylated and activated in G₁ (Blanco et al., 2000; Yamaguchi et al., 2000). Cells were synchronised by arrest-release of a cdc10 mutant, and the phosphorylation state of ste9p was monitored by western blotting. Ste9p was dephosphorylated normally in G₁ arrested cells, and then rephosphorylated upon S-phase entry in both flp1⁺ or flp1::kanMX6 backgrounds (Fig. 7D). The kinetics of entry into S-phase following release of the cdc10-129 block were similar in a flp1⁺ and flp1::kanMX6 background (Fig. 7E). We conclude that flp1p is not responsible for dephosphorylation of ste9p in G₁.

Strong expression of S. cerevisiae CDC14 promotes mitotic exit, B-type cyclin degradation, accumulation of the CKI p40^sic1, and G₁ arrest (Visintin et al., 1998). Increased expression of S. pombe flp1 from the thiamine-regulated nmt1 promoter produced elongated cells with a single nucleus (Fig. 8A). FACS analysis indicated that cells arrested predominantly with 2C DNA content (Fig. 8B). Staining with Rhodamine-conjugated Phalloidin indicated that F-actin patches were located at the tips of the cell consistent with a G₂ arrest (data not shown). Upon prolonged incubation, cells eventually entered mitosis and septated. Increased expression of the mutant flp1p(C286S), in which the conserved cysteine that is known to be essential for activity of S. cerevisiae Cdc14p (Taylor et al., 1997) is replaced by serine, did not produce a cell cycle arrest, indicating that the phosphatase activity of flp1p is essential for cell cycle arrest (Fig. 8C). Western blotting showed that the levels of the mitotic B-type cyclin cdc13p and the cdk-inhibitor rum1p did not change significantly in the arrested cells (Fig. 8D), indicating that the arrest does not result from degradation of B-type cyclins or stabilisation of rum1p. The mitotic inducer cdc25p was shifted to a faster-migrating form (Fig. 8D). This was confirmed by expressing flp1 in a strain carrying a myctagged cdc25p (Fig. 8E), and suggesting that the G₂ arrest might be mediated in part through the mitotic regulatory system.

A cell cycle block was also imposed by increased expression of flp1 in rum1::ura4⁺, ste9::ura4⁺ and rum1::ura4⁺ste9::ura4⁺ mutants, indicating that neither ste9p nor rum1p is necessary for the cell cycle arrest induced by increased expression of flp1p. Expression of flp1 in rad3^-, chk1Δ, and rad24Δ backgrounds also blocked cell division, suggesting that the G₂ arrest does not result from ectopic activation of the DNA structure checkpoint (data not shown).

The effects of increased flp1 expression were studied in the mutants wee1-6 (which divide at a reduced size at all temperatures (Fantes, 1981) and cdc25::ura4⁺wee1-6 (which lacks the normal size control over entry into mitosis (Sveiczer et al., 1999). Both strains continued dividing at times following induction of flp1⁺ expression when wild-type cells had arrested (Fig. 9A,B). This demonstrates that the G₂ arrest following increased expression of flp1p requires the mitotic inhibitor wee1p.

The effect of increased flp1⁺ expression in the presence of activated alleles of cdc2 was also investigated. The mutant cdc2-1w is less responsive than cdc2⁺ to inhibition by wee1p, but still requires cdc25p for activation (Fantes, 1983; Russell and Nurse, 1987b). Increased expression of flp1⁺ in cdc2-1w resulted in some cell elongation, but septated, dividing cells were still present, indicating that, as in the wee1-6 background, the cell cycle arrest was relieved. Moreover, anucleate compartments and cut nuclei were observed, indicating a loss of proper co-ordination between mitosis and septum formation (Fig. 9C). The mutant cdc2-3w is independent of cdc25p for activation, but still responds to inhibition by wee1p (Russell and Nurse, 1987b). In this background, increased expression of flp1⁺ arrested cell division. Approximately 50% of cells had a single nucleus, while the remainder had two, whose chromatin domains faced each other, indicating that these cells had completed mitosis without septating (Fig. 9D). Together, these data suggest that the G₂ arrest due to flp1 overexpression is imposed predominantly through the mitosis regulators cdc25p and wee1p.

The genetic interactions of flp1::kanMX6 with elements of the mitotic control system were also examined. Double mutants with wee1-6, cdc2-1w, and cdc2-3w were constructed. The mutants wee1-6 flp1::kanMX6, cdc2-1w flp1::kanMX6 and cdc2-3w flp1::kanMX6 showed strong additive effects. Multinucleate cells were frequently observed, indicating that cells had failed to septate. In addition, enlarged nuclei and nuclei of different sizes in the same compartment were observed, suggesting that aberrant mitoses had occurred. These phenotypes were more accentuated at 36°C than 25°C (Fig. 10A,B,C). The mutant cdc2-3w is defective in sensing the completion of DNA replication (Enoch and Nurse, 1990). Incubation of cdc2-3w flp1::kanMX6 in medium containing hydroxyurea produced `cut' phenotypes at both 25°C and 36°C, indicating that the flp1::kanMX6 mutation does not restore the DNA replication completion checkpoint function. Consistent with this, flp1::kanMX6 cells arrested normally in the presence of hydroxyurea, indicating that the DNA replication completion checkpoint is functional (data not shown). No additive effects were noted in the double mutants rum1::ura4⁺flp1::kanMX6 and ste9::ura4⁺flp1::kanMX6 (Table 1; data not shown).

Defects in septum construction can trigger a checkpoint that delays the resumption of tip-growth, and disassembly of the medial F-actin ring. This delay requires both wee1p and SIN function (Le Goff et al., 1999b; Liu et al., 2000). Since the flp1::kanMX6 allele showed strong genetic interactions with the SIN and mitotic control genes, the effect upon the morphology checkpoint was examined by constructing the double mutant cps1-N12 flp1::kanMX6. The cps1-N12 mutant is defective in β-glucan synthase function (Le Goff et al., 1999b). After shift to 36°C, more than 70% of cells arrest binucleate, with or without a division septum, having activated the morphology checkpoint (Fig. 11A,C). By contrast, although the aberrant septa were not cleaved, the double mutant cps1-N12 flp1::kanMX6 became elongated, and many cells underwent a second round of nuclear division during the incubation period (Fig. 11B,C). We conclude that flp1p function is required for the S. pombe morphology checkpoint.

In this paper we have presented the characterisation of flp1, the S. pombe orthologue of the S. cerevisiae CDC14 gene. Although flp1 has some features in common with its S. cerevisiae counterpart, its role in cell cycle control seems to be significantly different.

Why is flp1p function not essential ? At the time of writing, while the sequence of the fission yeast genome remains incomplete (though only about 50 gene's-worth of DNA remains to be sequenced), it is not possible to exclude formally that there is a second flp1p-like protein in fission yeast, with which flp1p is functionally redundant. Nonetheless, database searches do not identify any other closely related protein, and Southern blotting has not identified any closely related sequences.

Both proteins are nucleolar-located in interphase, from where they are released during mitosis. However, while release of S. cerevisiae Cdc14p from the nucleolus requires MEN function, the data presented in this paper indicate clearly that SIN function is not required to effect flp1p release from the nucleolus in S. pombe. In budding yeast, Cdc14p is tethered in the nucleolus by attachment to Net1p, which is a component of the RENT complex (Shou et al., 1999; Straight et al., 1999; Visintin et al., 1999). Database searches have failed to identify a Net1p orthologue in S. pombe to date. In this context, it is noteworthy that the closest S. pombe homologue of Sir2p (ORF SPBC16d10.07), which is another member of the RENT complex in S. cerevisiae, is also located in the nucleolus (M. Cockell, V.S. and S. Gasser, unpublished).

What is the signal for release of flp1p from the nucleolus? Flp1p is phosphorylated during mitosis, at the time when it is associated with the medial ring, the mitotic spindle, and is found throughout the nucleus. It is tempting to speculate that the phosphorylation is the trigger for the relocalisation of flp1p. It is noteworthy that there are two cdc2p consensus sites (S/TPXK/R) in flp1p. However, these are not conserved in the Cdc14p orthologues from other species. The role of phosphorylation in the regulation of flp1p will be the subject of future studies.

In S. pombe, return of flp1p to the nucleolus at the end of mitosis requires inactivation or attenuation of the SIN. It is possible that inactivation of SIN signalling marks the successful completion of a division septum and that. while it is active, it ensures that the flp1p remains available to dephosphorylate a substrate that is important to permit septation and/or cell cleavage. It is unlikely that initiation of DNA synthesis is the trigger for nucleolar localisation of flp1p at the end of mitosis, since a cdc16-116 mutant undergoes DNA synthesis at the non-permissive temperature (Minet et al., 1979).

Cells lacking flp1p function are advanced into mitosis, suggesting that flp1p is an inhibitor of mitosis. Consistent with this, increased expression of flp1⁺ arrests cells in G₂, in a wee1p-dependent manner. In the arrested cells, cdc25p is in a rapidly migrating, dephosphorylated form. Since cdc25p is activated by phosphorylation during mitosis (Creanor and Mitchison, 1996; Ducommun et al., 1990; Moreno et al., 1990), it is possible that this contributes significantly to the G₂ arrest. The fact that flp1 overexpression still marrests the cell cycle in a cdc25p-independent manner in the cdc2-3w mutant, suggests that it also acts upon either wee1p, or one of its regulators, such as nim1p (Russell and Nurse, 1987a; Wu and Russell, 1993), nif1p (Wu and Russell, 1997), or cdr2p (Breeding et al., 1998; Kanoh and Russell, 1998) to delay mitotic entry. In this context, it is noteworthy that cdr2p has been implicated in regulating septum formation, although at which level is not clear (Breeding et al., 1998).

In S. pombe, both cdc25p and wee1p are nuclear proteins, found predominantly, although not exclusively, in the DAPI-staining region of the nucleus (Aligue et al., 1997; Wu et al., 1996; Zeng and Piwnica-Worms, 1999). It is also noteworthy that some flp1p-GFP is observed in the non-nucleolar part of the nucleus. It is therefore possible that during interphase, cdc25p and wee1p phosphorylation states are modulated by flp1p to maintain them in an inactive and active state, respectively. Whether this is direct or indirect remains to be determined. Thus, although flp1p function is not essential for entry into (or exit from) mitosis, flp1p may play a `fine-tuning' role in regulating entry into mitosis. The fact that flp1 function is not essential either for mitotic onset or completion may be due to the presence of multiple control systems, which can compensate for each other.

It has been suggested that S. cerevisiae Cdc14p is the essential effector of the MEN. By contrast, flp1p is not an essential effector of SIN signalling. Nonetheless, SIN mutants that appear wild-type at 25°C in a flp1⁺ background show a strong septation phenotype in the absence of flp1p function. It is therefore possible that one of the roles of flp1p is to potentiate SIN signalling, for example, by activating one or more elements of the network. Two observations are consistent with this: first, flp1::kanMX6 cells show defects in septation signalling, and second, induction of septation by spg1 overexpression is less efficient in the flp1::kanMX6 background, suggesting that the absence of flp1p attenuates SIN signalling. In this context, it is noteworthy that in S. cerevisiae, Cdc14p may activate the MEN component Cdc15p providing the potential for a positive feedback loop to promote exit from mitosis (Jaspersen and Morgan, 2000). Whether this is the case in S. pombe will be addressed in future studies.

Septation is partially inhibited in the flp1::kanMX6 cdc16-116 mutant at 25°C. The single cdc16-116 mutant does not show any obvious septation defect at 25°C (Minet et al., 1979). However, increased expression of the cloned cdc16-116 allele in wild-type cells blocks septum formation, while cdc16⁺ does not (L. Cerutti and V.S., unpublished), suggesting that the mutant cdc16-116p may be a more effective negative regulator of SIN signalling. If the absence of flp1p also attenuates SIN signalling, a combination of these mutants may produce a synergistic effect. Growth at a higher temperature, such as 29°C, may partially inactivate cdc16-116p, reducing the inhibitory effect.

Previous studies have demonstrated that cdc2p kinase activity during mitosis is antagonistic to septum formation (Cerutti and Simanis, 1999; He et al., 1998; He et al., 1997). Thus, premature activation of cdc2p, and its delayed or incomplete inactivation during mitosis might account for the strong negative genetic interactions of flp1::kanMX6 with mutants that have reduced SIN signalling, and the greatly enhanced septation defects shown by flp1::kanMX6 in wee1^- and cdc2w backgrounds.

The involvement of flp1p in the S. pombe morphology checkpoint could be at several levels. One possibility is that it is an essential effector of the checkpoint, and that its activation (or release from the nucleolus) is required to delay the next cycle of nuclear division, and the resumption of tip-growth. Alternatively, since the morphology checkpoint requires both wee1p and SIN signalling for its activity, it is possible that the failure of the checkpoint results from reduced SIN signalling in the flp1::kanMX6 cells.

Our studies suggest that elements of the mitotic control system (or their upstream regulators) may be targets of flp1p. Flp1p is also located on the spindle pole body, mitotic spindle and the contractile ring. It is known that many regulators of mitosis and septum formation, are located on the spindle pole body (Alfa et al., 1990EF1; Bahler et al., 1998aEF4; Cerutti and Simanis, 1999EF13; Chang and Gould, 2000EF14; Eng et al., 1998EF17; Guertin et al., 2000EF28; Hou et al., 2000EF35; Mulvihill et al., 1999EF51; Sohrmann et al., 1998EF65; Sparks et al., 1999EF67), so these may be among the targets of flp1p. The diversity of phenotypes associated with loss of flp1p function may reflect the fact that it has more than one important substrate, and so its absence may reduce the fidelity, or efficiency, of several processes. Identification of the substrates and anchors of flp1p on the mitotic spindle, the contractile ring and the nucleolus will be the subject of future studies.

We thank Elena Cano for technical assistance, and Moira Cockell and Andrea Krapp for helpful comments about the paper. V.S. received financial support for this work from The Swiss National Science Foundation, The Swiss Cancer League, The Fondation Forez and ISREC. N.C. was supported by a grant from ARC. This research was also supported by CICYT and PGC grants from the Spanish Science and Technology Ministry to S.M. and A.B. We also thank Kathy Gould, Dan McCollum, Jürg Bähler, Paul Nurse, Susan Gasser and Keith Gull for strains and antibodies.