mph1, a member of the Mps1-like family of dual specificity protein kinases, is required for the spindle checkpoint in S. pombe

Proper mitotic spindle function is essential for the equal separation of chromosomes into two daughter nuclei during mitosis. The spindle is assembled when cells enter into mitosis in most species and disassembled when cells exit from mitosis. The duplication of the spindle organizing center, known as the centrosome in mammalian cells and the spindle pole body (SPB) in yeast, occurs before the entry into mitosis (Byers, 1981; Ding et al., 1997; Rattner and Phillips, 1973). In most species, when cells enter mitosis, microtubules connect the duplicated centrosomes/SPBs to form a short mitotic spindle that attaches to the condensed chromatin via the interaction between spindle microtubules and the kinetochores, a specialized region of the chromosome (Page and Snyder, 1993). During anaphase, the sister chromatids are separated into what will become the two daughter nuclei. The assembly of a fully functional mitotic spindle and the proper attachment of all kinetochores to the spindle is crucial for equal chromosome segregation. A regulatory pathway, known as the spindle assembly checkpoint, enhances the fidelity of this process, although it is not essential for passage through a normal mitosis (Rudner and Murray, 1996). The checkpoint does become essential for a successful cell division, however, when the cells encounter defects in spindle structure or kinetochore-spindle attachment. Under these conditions, the spindle checkpoint detects the defect and transduces a signal which arrests cell cycle progression, allowing the cells to assemble a proper mitotic structure before proceeding with chromosome segregation.

There are two models regarding the nature of the spindle defects that are detected by the checkpoint: the lack of physical tension on the chromosomes or the presence of kinetochores that are not attached to microtubules. The first model is supported by micromanipulation experiments performed in mantid spermatocytes (Li and Nicklas, 1995, 1997; Gorbsky and Ricketts, 1993; Nicklas et al., 1995). In this system, a single chromosome that was attached to only one spindle pole was sufficient to inhibit the onset of anaphase. When this mono-oriented chromosome was pulled with a microneedle so that physical tension was established, the inhibition of cell cycle progression was released. Furthermore, the binding of the phosphoepitope-specific monoclonal antibody 3F3/2 to the kinetochores of only the mono-oriented chromosome was eliminated when physical tension on the chromosome was established using a microneedle, suggesting that the signal generated by the lack of physical tension on the chromosomes was transduced by chemical modifications of the kinetochore. On the other hand, studies in other systems argue that the presence of kinetochores that are not attached to microtubules is crucial for the generation of the inhibitory signal. Budding yeast cells lacking the Cdc6 protein fail to initiate DNA replication (Piatti et al., 1995). Although these cells do not have sister chromatid pairs, the kinetochores are attached to microtubules and therefore all the chromosomes are mono-oriented. However, these cdc6 mutant cells fail to be arrested by the spindle checkpoint, and undergo a ‘reductional’ anaphase with nearly wild-type kinetics. Whether these apparently conflicting results indicate a more complex mechanism by which the inhibitory signal is generated or they simply reflect species differences is unknown. Despite the differences, these results all point toward the critical role of kinetochores in generating the spindle checkpoint activation signal.

The spindle checkpoint pathway has been most extensively studied, at a molecular level, in the budding yeast, Saccharomyces cerevisiae, in which eight genes have been identified that function in the spindle checkpoint pathway: MAD1, MAD2 and MAD3 (Hardwick and Murray, 1995; Li and Murray, 1991), BUB1, BUB2 and BUB3 (Hoyt et al., 1991; Roberts et al., 1994), MPS1 (Weiss and Winey, 1996) and CDC55 (Wang and Burke, 1997). Cells with a mutation in any of these genes fail to arrest the cell cycle in response to spindle defects and undergo an aberrant, lethal mitosis. Structural homologs of several S. cerevisiae spindle checkpoint proteins have been identified in other species suggesting that the spindle checkpoint is evolutionarily conserved. Mad2p homologs have been found in human (Li and Benezra, 1996), frog (Chen et al., 1996) and fission yeast (He et al., 1997); a Bub2p homolog, cdc16, has been found in fission yeast (Fankhauser et al., 1993); and a Bub1p homolog has been identified in mammalian cells (Taylor and McKeon, 1997). These homologous proteins all function in the spindle checkpoint in their own species. Novel checkpoint genes have also been identified, such as p44ERK2 in frog (Minshull et al., 1994) and dma1 in fission yeast (Murone and Simanis, 1996), although their possible interactions with the known checkpoint elements have not been established.

Biochemical studies in budding yeast suggest that phosphorylation of Mad1p protein plays a key role in checkpoint function: in wild-type cells, Mad1p is hyperphosphorylated specifically in response to spindle defects (Hardwick and Murray, 1995). Overproduction of Mps1p, a dual specificity kinase (Lauze et al., 1995), which mimics checkpoint activation and causes a metaphase-like arrest, also results in increased Mad1p phosphorylation (Hardwick et al., 1996). The fact that Mps1p can directly phosphorylate Mad1p in vitro further supports the hypothesis that Mps1p may be the kinase that phosphorylates Mad1p in vivo (Hardwick et al., 1996). However, MPS1 is not the only gene required for Mad1p hyperphosphorylation in vivo.

Mutations in BUB1, BUB3 or MAD2 prevent both Mad1p hyperphosphorylation and cell cycle arrest in response to spindle defects (Hardwick and Murray, 1995). Under the same conditions, mutations in two other components of the checkpoint, BUB2 or MAD3, do not interfere with Mad1p hyperphosphorylation, but do prevent cell cycle arrest (Hardwick and Murray, 1995). Taken together, these results suggested the following model for spindle checkpoint function (Hardwick et al., 1996): in response to spindle defects, Mps1p phosphorylates Mad1p in a Bub1p, Bub3p and Mad2p dependent manner, while Mad3p and Bub2p function downstream of Mad1p and are required for the phosphorylated Mad1p to impose a cell cycle arrest. Recently, it was found that frog Mad2p (Chen et al., 1996), mammalian Mad2p (Li and Benezra, 1996) and mammalian Bub1p (Taylor and McKeon, 1997) specifically associate with unattached kinetochores. Together with the observations that budding yeast Mad1p binds to Mad2p (Rudner and Murray, 1996) and Bub1p binds to Bub3p (unpublished result cited by Rudner and Murray, 1996), these results raise an intriguing possibility that these checkpoint proteins may form a large complex which specifically binds to the unattached kinetochore and sends a signal that arrests cell cycle progression. The possible functional consequence of Mad1p hyperphosphorylation on the formation or kinetochore binding of this putative protein complex, however, remains unknown.

Mps1p, the kinase which is likely to phosphorylate Mad1p in vivo, is unique among the checkpoint proteins identified in budding yeast in that it has an additional essential function in spindle pole body (SPB) duplication (Winey et al., 1991). Cells with the temperature sensitive mps1-1 mutation fail to duplicate the SPB at the restrictive temperature. Unlike other mutants that are defective in SPB duplication, such as mps2-1 (Winey et al., 1991) and kar1 (Rose and Fink, 1987), mps1-1 mutants do not arrest cell cycle progression because they also have a defective checkpoint (Weiss and Winey, 1996). All of the characterized mutations in MPS1 that eliminate its kinase activity cause both the SPB duplication defect and the checkpoint defect, suggesting that the kinase activity of the Mps1p is required for both functions (A. R. Schutz and M. Winey, unpublished observations). Nevertheless, when Mps1p is inactivated in mps1-1 mutant cells by a shift to the restrictive temperature after completion of SPB duplication, the spindle checkpoint is still impaired, indicating that the SPB duplication function and the spindle checkpoint function of the Mps1p are separable (Weiss and Winey, 1996).

In a fission yeast cDNA library screen, we isolated three cDNAs that cause a metaphase arrest when overexpressed (He et al., 1998). We have already characterized one of the cDNAs, mad2, and shown that it encodes a S. cerevisiae Mad2p homolog that functions in the spindle checkpoint system in fission yeast. In fission yeast mad2 overexpression mimics activation of the spindle checkpoint and causes a cell cycle arrest at the metaphase-to-anaphase transition (He et al., 1997). It has been shown that specific proteolysis under control of the anaphase promoting complex (APC) regulates cell cycle progression at this point of the cell cycle (Holloway et al., 1993; Irniger et al., 1995; King et al., 1995): the APC-dependent degradation of S. cerevisiae Pds1p (Cohen-Fix et al., 1996) and S. pombe cut2p (Funabiki et al., 1996) are required for sister chromatid separation, and the APC-dependent degradation of cyclin B is required for exit from mitosis in budding yeast (Irniger et al., 1995; Surana et al., 1993), fission yeast (Yamano et al., 1996) and frog egg extract (King et al., 1995). Our phenotypic and genetic analyses of mad2 overexpression in fission yeast suggest that the metaphase arrest resulting from spindle checkpoint activation is due to an inhibition of APC-dependent proteolysis (He et al., 1997).

Here we report the characterization of the other two cDNAs, dph1 and mph1, that were isolated from the same screen in which mad2 was identified. Our initial characterization indicates that S. pombe dph1 is functionally related to Dsk2p, a protein that is involved in SPB duplication in S. cerevisiae (Biggins et al., 1996). We demonstrate that S. pombe mph1 functions in the spindle checkpoint pathway, upstream of mad2. mph1 belongs to the same dual-specificity protein kinase subfamily as S. cerevisiae Mps1p but it is not essential for cell division, indicating that the checkpoint function of the Mps1p/mph1 kinase family is independent of SPB duplication.

S. pombe strains used were the haploid wild-type strain (h^-, leu1-32, ura4-D18, ade6-m216), diploid wild-type strain (h^-/h⁺, leu1-32/leu1-32, ura4-D18/ura4-D18, ade6-m210/ade6-m216), mutant strains cdc2-33^ts (h^-cdc2-33^ts, leu1-32, ura4-D18) (Carr et al., 1989), mad2^∆ (h^-, mad2::ura4, leu1-32, ade6-m210) (He et al., 1997) and nda3^cs (h^-, nda3-311, leu1-32, ade6-m210) (Hiraoka et al., 1984). The mph1 deletion strains, mph1^∆ and mph1-pd^∆ (partial) were generated by replacing the ORF of mph1 or a HindIII fragment that encodes 45% of the kinase domain (a.a. 368-487) with the ura4 gene (see DNA methods, below), and were confirmed by Southern blot analysis. Tetrad analysis of a mph1-pd^∆ heterozygous diploid showed co-segregation of ura⁺ with thiabendazole (Sigma) hypersensitivity at thiabendazole concentrations of 20-40µg/ml. A genetic cross between mph1^∆ and mph1-pd^∆ showed that these two null mutations were linked. The mph1^∆nda3^cs double mutant was generated by crossing mph1^∆ with nda3-311^cs and was identified by its cold sensitive and ura+ phenotypes. Yeast culture, transformation and genetic manipulations were performed by standard procedures (Moreno et al., 1991). dph1 null (dph1^∆) strain was generated by replacing one copy of dph1 with ura4 in a ura^- wild-type diploid strain (see DNA methods), and was confirmed by Southern blot. Yeast culture, transformation and genetic manipulation were performed by standard procedures (Moreno et al., 1991).

In a screen described in detail elsewhere (He et al., 1998), a S. pombe cDNA library in pREP3X vector (a gift from Bruce Edgar and Chris Norbury), in which cDNA expression is controlled by the thiamine repressible nmt1 promoter (Maundrell, 1993), was transformed into wild-type cells. Transformants were screened based on the toxicity and the mitotic defects caused by cDNA overexpression. Plasmids from S. pombe transformants were recovered by standard procedures (Moreno et al., 1991) and amplified in E. coli. The cDNA overexpression constructs pREP3X-mph1-1 and pREP3X-dph1 were directly isolated from the library screen following the above procedures. The cDNA inserts were released from pREP3X by SalI and NotI digestion and were subcloned into Bluescript KS^-(Stratagene) and sequenced using Sequenase Version 2.0 (USB). To clone the mph1 gene, cosmid ICRFc60H0216D (Lehrach et al., 1990) DNA was digested with EcoRI and the 5.8 kb fragment (Fig. 1B) which contains the mph1 gene was subcloned into Bluescript KS^-(Stratagene), generating pBS-mphl-g. The 5’-terminal EcoRI/PstI fragment (Fig. 1B) was used to replace the nmtl promoter and the 5’-end of mph1-1 in pREP3X-mph1-1 to generate pREP-mph1-g in which mph1 expression is controlled by its own promoter. To construct the full length mph1 cDNA, mph1-2, oligonucleotides 1 and 2 (oligo 1, 5’-TTAGTAACTGTCGACATGTCTAAG-3’; oligo 2, 5’-GTTATGCAGCTGCAGAAGG-3’) were used to PCR-amplify the genomic DNA fragment which encodes the N terminus of mph1 and to introduce a SalI site just 5’ of the Start codon ATG. The PCR product was digested with SalI and PstI and used to replace the 5’-end of mph1-1. pREP3X-mph1, pREP41X-mph1 and pREP-81X-mph1 were constructed by subcloning the SalI/NotI fragment of the mph1-2 cDNA into pREP3X, pREP41X and pREP81X respectively, to achieve different levels of mph1 overexpression (Forsburg, 1993). pREP3X-MPS1 and pREP81X-MPS1 were constructed by subcloning the XhoI/NotI MPS1 cDNA fragment from pAFS120 (Lauze et al., 1995) into the corresponding S. pombe expression vectors. To overexpress S.c. DSK2 in S. 01996) by BamHI digestion and subcloned into pREP3X. To construct the full length mph1 deletion, oligonucleotides 3 and 4 (oligo 3, 5’-CCAACTGTTTGTTAAGCTTAATC-3’; and oligo 4, 5’ AGCACAGAAATCTAAGCTTCTCAA-3’) were used to PCR amplify the flanking regions of mph1 and the intervening vector sequence from pBS-mph1-g and introduce HindIII sites 5’ of the Start codon and 3’ of the Stop codon. The PCR product was digested with HindIII and ligated with the 1.8 kb HindIII fragment containing the ura4 gene so that the entire ORF of mph1 was replaced by ura4. The mph1/ura4 hybrid DNA fragment (mph1^∆) was released by EcoRI digestion and used for gene deletion. To clone the dph1 gene, a 4.6 kb BglII fragment was recovered from cosmid SPAC26A3 (Hoheisel et al., 1993) which contains dph1 and subcloned into pSL1190 (Pharmacia Biotech). A XhoI site and a SpeI site were introduced at the 5’ and 3’ ends of the dphl ORF, respectively, by PCR amplification using the following oligonucleotides 5 and 6 (oligo 5; 5’-TTTCCACTGAGCGATGACGC-3’; oligo 6; 5’-GCCATGCTTACTAGTTTTGCC-3’). The 1.8 kb HindIII ura4 gene was first subcloned into pBluescript KS(+) and then released by XhoI/SpeI digestion so that it could be ligated with the dph1 PCR product described above to generate the dph1^∆ DNA fragment used for the gene deletion.

Wild-type cells or mad2^∆ cells transformed with either pREP3X-mph1 or pREP41X-mph1 were transferred from Edinburgh minimal medium (EMM) with 5µg/ml thiamine, in which transcription is repressed, to thiamine-free medium for 16 hours to induce mph1 overexpression at 32°C. The wild-type control cells, transformed with the vector pREP3X, were treated identically. Total yeast extracts were prepared using the glass bead method and total protein extracts were used for histone H1 protein kinase assay as described (Moreno et al., 1989). The amount of radioactivity in each band was determined by PhosphoImager (Molecular Dynamics). 10µg total extract of each sample was used to determine the amount of cdc2 by western blot (ECL, Amersham) using an anti-PSTAIRE antibody (Santa Cruz).

Equal numbers of single mutant cells, mph1^∆ or nda3^cs, and double mutant cells, mph1^∆nda3^cs, were spread on YE plates. The plates were then incubated at the nda3^cs restrictive temperature of 18°C for 0, 3, 6 or 9 hours and then returned to the permissive temperature of 36°C. The number of colonies was counted after three days.

Immunofluorescence procedures are as described (Hagan and Hyams, 1988). Cells were fixed with glutaraldehyde and paraformaldehyde and stained with anti-tubulin monoclonal antibody, TAT1 (Woods et al., 1989) (a gift from Keith Gull), and fluorescein-conjugated anti-mouse secondary antibody (Pierce) to reveal the microtubule structures. Cells were simultaneously stained with DAPI to visualize the DNA.

The strain W303 (kind gift from R. Rothstein) containing the mps1-1 mutation (a kind gift from P. Straight) were transformed with pRS314 (TRP) (Sikorski and Heiter, 1989), pRS314-MPS1 (kind gift from A. Castillo), 490-mph1, or 490-1-mph1 using standard techniques. 490-mph1 contains the mph1 gene under control of the gal1 promoter and was constructed by subcloning the 2.0 kb SpeI/NotI fragment from pREP-mph1-g into the corresponding sites in the S. cerevisiae expression vector 490 in which the gal1-10 promoter has been inserted into pRS314 (a kind gift from E. Siewert). Plasmid 490-mph1-1 was constructed by replacing the EcoRI/NotI fragment in vector 490 with a mph1 cDNA fragment from pREP3X-mph-1. The SalI/NotI fragment from pREP3X-mph-1 containing mph1-1 cDNA was first subcloned into the corresponding sites of pSL1190 (Pharmacia Biotech), then released by EcoRI/NotI digestion, subcloned into 490 vector at the EcoRI/NotI sites. In 490-mph1-1, the gal1-10 promoter was replaced by the mph1-1 cDNA but apparently the cDNA is expressed from an unidentified sequence in the vector, since 490-mph1-1 rescues the mps1 mutation regardless of the carbon source in the medium (see Results). Wild-type and mutant strains were grown on minimal or rich medium plates containing glucose, raffinose or galactose as a carbon source and phenotypes were checked at a variety of temperatures. Galactose-induced expression from neither 490-mph1 nor from 490-mph1-1 in wild-type cells causes cell cycle arrest, although expression from 490-mph1 does appear to be toxic over time. For the liquid culture experiments, strains were grown at the permissive temperature (25°C) in non-inducing raffinose medium lacking tryptophan, and synchronized in G₁ using alpha factor (10µmolar, Macromolecular Resources). Cultures were incubated for 2 hours (until >90% of the cells were in a schmoo state) and then spun and resuspended in either raffinose medium plus 4% galactose (inducing) or in raffinose medium alone (non-inducing) and incubated an additional 1.5 hours with alpha factor (0 time point). Cells were released from alpha factor arrest by washing twice with prewarmed medium (lacking alpha factor, plus 4% galactose) and then resuspended in the same medium at the nonpermissive temperature (37°C). Half of the mph1 culture was kept in raffinose medium for the duration of the experiment. Every hour for 4 hours, a sample was plated at the permissive temperature on complete glucose-containing plates to count viable (recovered) cells, and a sample was analyzed by flow cytometry (using a Becton Dickinson FacScan flow cytometer, San Jose, CA) to analyze DNA content (1C or 2C) (Hutter and Eipel, 1979). The DNA content of surviving cells was also analyzed by flow cytometry. This experiment was performed twice with similar results.

The mps1-pac8 strain was a kind gift from A. Hoyt (Geiser et al., 1997). The strain contains a deletion of the CIN8 gene and the pac8-1 allele of MPS1 in the chromosome and a plasmid containing the CIN8 gene, the LEU2 nutritional marker and a mutant allele of the L29 gene which causes the strain to be unable to grow on cyclohexamide containing plates (Geiser et al., 1997). This strain was transformed with pRS314-MPS1 or 490-mph1-1 and complementation was checked by ability of the strain to grow on cyclohexamide containing raffinose plates.

In a cDNA library screen aimed at identifying genes involved in the mitosis-to-interphase transition in S. pombe (He et al., 1998), we isolated three cDNAs that cause a metaphase arrest when overexpressed in wild-type cells. One of them, mad2, encodes an evolutionarily conserved protein that is an essential component of the spindle assembly checkpoint (He et al., 1997). Here we report the identification and characterization of the other two genes, each of which encodes a protein that is similar structurally and functionally to a corresponding S. cerevisiae protein: mph1 (Mps1p pombe homolog) is related to Mps1p; and dphl (Dsk2p pombe homolog) is related to Dsk2p.

The mph1-1 cDNA isolated in the library screen encodes a protein that contains a protein kinase domain approximately 50% identical to that of S. cerevisiae Mps1p (Fig. 1A) and the other members of the dual-specificity kinase family that are found in other species (Douville et al., 1992; Mills et al., 1992). Outside of the kinase domain, the protein does not have significant similarity to other identified proteins in GenBank. To clone the mph1 gene, an S. pombe genomic cosmid library filter (Lehrach et al., 1990) was probed with the mph1-1 cDNA and the following cosmids were identified: ICRFc60H0216D, ICRFc60G103D, ICRFc60G1011D, ICRFc60E072D, ICRFc60C1012D. Based on the chromosomal localization of these cosmids, the mph1 gene was mapped to chromosome III, between the genomic markers ade5 and adh1. The mph1 gene was subcloned from the cosmid ICRFc60H0216D as a 5.8 kb EcoRI fragment (Fig. 1B). Partial sequence analysis of the mph1 gene revealed that the mph1-1 cDNA was missing the 5’ end encoding N-terminal 159 amino acids. In order to express full length mph1, a complete mph1 cDNA (GenBank accession number AF020705), mph1-2, was generated by replacing the incomplete 5’ portion of mph1-1 with the complete 5’ portion from the genomic DNA and subcloned into the expression vectors pREP3X, pREP41X and pREP81X (see Materials and Methods). The phenotypes of wild-type cells overexpressing the full-length or the partial mph1 cDNA are identical (data not shown). In the following text, all descriptions of the mph1 gene refer to the full-length construct unless otherwise specified.

To further characterize mph1, the cDNA was overexpressed at different levels in wild-type cells. On a thiamine-free EMM plate that allows transcription from the nmt1 promoter (Maundrell, 1993) cells with high level (pREP3X-mph1) or medium level (pREP41X-mph1) mph1 overexpression did not grow while cells with low level overexpression (pREP81X-mph1) or the vector control grew well (Fig. 2A). Fifteen hours after the induction of mph1 overexpression in a liquid culture, cell division was completely inhibited in cells overexpressing mph1 from pREP3X or pREP41X (Fig. 2B). Up to 60% of these cells (Fig. 2C) had a short mitotic spindle and a single nucleus with hypercondensed chromosomes (Fig. 2D), that are the hallmarks of metaphase cells (Hagan and Hyams, 1988). Consistent with the accumulation of metaphase cells, mph1 overexpressing cells also had mitosis-promoting-factor (MPF) kinase activity 5 fold higher than asynchronous wild-type cells (Fig. 3C).

Morphologically, wild-type cells arrested by mph1 overexpression were identical to those arrested by mad2 overexpression (He et al., 1997), indicating that they were arrested at the same point in the cell cycle. mad2 overexpression arrests cells at the metaphase-to-anaphase transition. Subsequent inactivation of MPF by a temperature sensitive mutation in either the cdc2 kinase subunit or the cdc13 cyclin B subunit allows these cells to bypass anaphase and directly exit from mitosis (He et al., 1997). To test whether mphl overexpressing cells and mad2 overexpressing cells are arrested at the same cell cycle stage, we tested the effect of cdc2 inactivation on cells arrested at metaphase by mphl overexpression. In cdc2-33_ts cells at the permissive temperature, overexpression of mphl induced a metaphase arrest in approximately half of the cells (data not shown), as it does in wild-type cells (Fig. 2D). The arrested cdc2_t cells were then shifted to the non-permissive temperature of 36°C to inactivate the MPF kinase. Two hours after shift to the non-permissive temperature, 43% of the cells underwent septation (Fig. 2E) and had a single nucleus in only one half of the cell (Fig. 2F). This indicates that the cells arrested by mphl overexpression at the permissive temperature bypassed nuclear division and directly exited from mitosis upon MPF kinase inactivation.

Overexpression of mphl blocks wild-type cells at the metaphase-to-anaphase transition, an arrest that is identical to that caused by spindle checkpoint activation due to mad2 overexpression (He et al., 1997). To test whether the cell cycle arrest caused by mphl overexpression was spindle checkpoint dependent, we overexpressed mphl in wild-type cells and mad2 null (mad2^∆) cells, that are spindle checkpoint defective (He et al., 1997). Medium level mphl overexpression (pREP41X-mphl) in wild-type cells was sufficient to cause a metaphase arrest and to inhibit their growth (Fig. 3A). However, mad2^∆ cells grew normally with the same level of mphl overexpression indicating that the cell cycle arrest caused by mphl overexpression is dependent on the spindle checkpoint. Although neither wild-type cells nor mad2^∆ cells grew when mphl was overexpressed at the high level (pREP3X-mphl) (Fig. 3A), wild-type cells accumulated up to 60% metaphase cells (Fig. 2C,D) while only 3% of the mad2^∆ cells had a short spindle. However, approximately 20% of the mad2^∆ cells showed aberrant chromosome segregation with an elongated spindle (Fig. 3B) and the rest had interphase microtubule structures. The MPF H1 kinase activity of these cells was consistent with the percentage of the metaphase cells: wild-type cells overexpressing mph1 at the high (Fig. 3C, lane 1) or medium (Fig. 3C, lane 3) levels had high H1 kinase activity while the mad2^∆ cells overexpressing mph1 had relatively low H1 kinase activity (Fig. 3C, lanes 2 and 4). These results indicate that the metaphase arrest caused by mph1 overexpression is dependent on the presence of the spindle checkpoint protein mad2. At high level expression, the metaphase arrest is also alleviated in mad2^∆ cells, but the lethality, possibly the result of inappropriate phosphorylation events due to expression levels 60-fold greater than from pREP41X (Forsburg, 1993), is not.

mph1 protein is a structural homolog of the S. cerevisiae protein kinase Mps1p (Fig. 1A), which functions in the spindle checkpoint pathway. We tested whether mph1 was also required for spindle checkpoint function in S. pombe. Cells with a defective spindle checkpoint fail to arrest in the presence of spindle defects and undergo an aberrant, lethal mitosis.

Therefore, they are hypersensitive to structural defects in the spindle caused by microtubule depolymerizing drug treatment or mutant forms of tubulin (Fankhauser et al., 1993; He et al., 1997; Murone and Simanis, 1996). A mph1 null mutation was constructed by replacing the mph1 open reading frame with the ura4⁺ gene (see Materials and Methods). Unlike MPS1 null S. cerevisiae cells that were inviable (Weiss and Winey, 1996), mph1 null (mph1^∆) S. pombe cells were viable, demonstrating that mph1 is not an essential gene. To determine whether mph1 is required for spindle checkpoint function, we tested the sensitivity of mph1^∆ cells to the microtubule depolymerizing drug, thiabendazole (TBZ). Serial dilutions of mph1^∆cells and isogenic wild-type cells were spotted on plates containing 0-40µg/ml of TBZ (Fig. 4A). In the absence of TBZ, mph1^∆ and wild-type cells grew equally well. At TBZ concentrations of 15µg/ml, mph1^∆cell growth was dramatically reduced relative to that of wild-type cells, suggesting that mph1^∆ cells are hypersensitive to TBZ. The TBZ hypersensitivity of the mph1^∆ cells was rescued by the introduction of a plasmid containing the wild-type mph1 gene, suggesting that the TBZ hypersensitivity is specific to the mphl deletion. To test whether mphl^∆ cells are also hypersensitive to other types of spindle defects, we constructed a double mutant of mphl^∆and nda3-3ll^cs, which has a cold sensitive mutation in the β-tubulin gene (Hiraoka et al., 1984). At low temperature, nda3^cs cells arrest at metaphase with hypercondensed DNA due to the absence of a mitotic spindle (Kanbe et al., 1990). This cell cycle arrest is spindle checkpoint dependent (He et al., 1997; Murone and Simanis, 1996). nda3^cs cells retain high viability when transiently incubated at the restrictive temperature of 18°C and resume normal mitosis when shifted back to permissive temperature (Kanbe et al., 1990). mphl^∆nda3^cs double mutant cells at the low temperature, however, had a progressive decrease in viability when incubated at 18°C, compared with nda3^cs or mphl^∆ single mutant cells (Fig. 4B). DNA staining with DAPI showed that double mutant cells had decondensed nuclear DNA and heterogeneous nuclear size after a 9 hour incubation at 18°C (Fig. 4C). In 28% of cells, cytokinesis occurred without nuclear division suggesting that mphl^∆ cells failed to arrest at metaphase and proceeded with cytokinesis without nuclear division. In half of these septated cells one daughter had a nucleus and the other did not, whereas in the other half the nucleus was ‘cut’ by the septum (Fig. 4C, arrow) resulting in an unequal amount of DNA in the two daughter cells. Flow cytometric analysis confirmed that the cells had a heterogeneous DNA content (data not shown). These phenotypes of the mphl^∆nda3_cs double mutant cells are identical to those of mad2^∆nda3_cs cells (He et al., 1997). Taken together, these results show that mphl is required for spindle checkpoint function.

The observation that mphl overexpression requires a functional mad2 gene to induce a metaphase arrest (Fig. 3A), suggests that mphl is upstream of mad2 in the spindle checkpoint pathway. If this were true, the cell cycle arrest due to mad2 overexpression would not require mphl. To test this, mphl^∆ cells were transformed with the mad2 overexpression plasmid pREP3X-mad2 (He et al., 1997), and transcription was induced. After 16 hours, 60% of the cells were arrested at metaphase with a short spindle and hypercondensed DNA, and because of this cell cycle block are slightly longer than wild-type cells (Fig. 4D). The phenotype of mphl^∆ cells is identical to the phenotype of wild-type cells overexpressing mad2 (He et al., 1997), demonstrating that mad2 overexpression does not require mphl to induced a metaphase arrest.

Both mph1 and Mps1p function in their respective spindle checkpoint pathways, upstream of mad2/Mad2p. They also belong to the same dual specificity kinase subfamily (Lauze et al., 1995). To ask whether they are functionally conserved, we first tested whether S’. cerevisiae MPSl causes a metaphase arrest when overexpressed in wild-type S. pombe cells. An MPSl expression plasmid, pREP3X-MPSl, was constructed by subcloning the MPSl cDNA (Lauze et al., 1995) into the pREP3X vector (see Materials and Methods) and was then introduced into wild-type S. pombe cells. Under the same conditions as S.p. mphl overexpression, 55% of the S.c. MPSl overexpressing cells had hypercondensed DNA and a short mitotic spindle (Fig. 5A), indicating that MPSl overexpression causes a metaphase arrest in S. pombe cells.

We then tested whether MPSl rescued the spindle checkpoint defect of S. pombe mphl^∆ cells. Under the control of the weak nmtl promoter in pREP81X, MPSl rescued the TBZ hypersensitivity of mphl^∆at a level comparable to that of mphl under the same expression conditions (Fig. 4A), suggesting that MPSl complements the in vivo function of mphl.

We also tested whether S. pombe mphl can function in S. cerevisiae, by asking if it can complement the defect caused by mutations in MPS1. At the nonpermissive temperature, mpsl-l containing cells fail in both spindle pole duplication and checkpoint arrest (Weiss and Winey, 1996; Winey et al., 1991). This results in an aberrant monopolar mitosis leading to daughter cells with abnormally high DNA content (Winey et al., 1991) (Fig. 5B-1), and a decrease in viability (with most surviving cells being diploid). mpsl-l containing cells were transformed with plasmids containing either vector alone, MPSl, or mphl under control of the gal1 promoter. The presence of the mphl gene did not alleviate the temperature sensitive defect of mps1-1 cells. However, we had observed that in wild-type cells this level of mphl overexpression led to growth inhibition, suggesting that this analysis would be complicated by the requirement for a level of expression high enough for complementation but not so high as to cause toxicity. Therefore, we chose to do the experiment in liquid culture to look for complementation of the mpsl-l defect in the first cell cycle after mphl was expressed to avoid longer term toxicity. The cultures were synchronized in the G₁ phase of the cell cycle, under inducing conditions for the mphl gene (0 hour) and then released from arrest at the nonpermissive temperature for the mpsl-l mutation. Shown are samples taken at 0, 1 and 3 hours that were analyzed by flow cytometry for DNA content, and the number of colony forming units at the permissive temperature was determined. mps1-1 cells containing vector alone (Fig. 5B-1) or mphl under non-inducing conditions (Fig. 5B-4) showed a decrease in the 2C DNA peak and an increase in cells with >2C DNA content at the 3 hour time point. In contrast, MPSl containing cells showed a full rescue of these defects (Fig. 5B-2) and mphl, under inducing conditions, showed a partial rescue (Fig. 5B-3). The colony forming units correlated with the FACS data, dropping significantly for the vector alone and mph1 noninducing samples, increasing for the MPS1 sample and remaining the same for the mph1 sample (data not shown). In addition, flow cytometric analysis of the DNA content of surviving mph1 containing cells was consistent with a partial rescue: while only 5/20 survivors in the non-induced sample remained haploid, 17/20 survivors in the induced mph1 sample remained haploid, compared to 10/10 in the MPS1 sample. These data suggest that mph1 rescued the SPB duplication defect in S. cerevisiae because if only the checkpoint defect was rescued, mps1-1 cells with a monopolar spindle would be expected to arrest with a 2C DNA peak upon shift to the nonpermissive temperature. We cannot distinguish between rescue of the SPB defect alone, and rescue of both the SPB and checkpoint defects in this experiment.

Therefore, we tested the ability of mph1 to complement a second class of mutation in MPS1 distinct from the conditional lethal mutation, mps1-1. Mps1-pac8 was isolated as a mutation that is lethal in combination with a deletion of the CIN8 gene, which encodes a kinesin-related motor involved in spindle pole separation during mitosis, but has no phenotype on its own (Geiser et al., 1997). The mps1-pac8/cin8 null mutant strain is kept alive by the expression ofCIN8 from a plasmid that can be counterselected by growth on cyclohexamide containing plates, meaning that the non-transformed strain is inviable on cyclohexamide plates (Fig. 5C). Transformation of the strain with a plasmid containing MPS1 rescues viability, while transformation with a plasmid containing mph1 results in a partial rescue of viability (Fig. 5C).

Taken together, these results showing partial complementation by the mph1 gene of two different mutations in MPS1, in conjunction with the above experiments testing MPS1 function in S. pombe, demonstrate that mph1 and Mps1p show functional as well as sequence conservation.

Partial sequence analysis of the dphl cDNA showed that the corresponding genomic DNA sequence had been determined by the S. pombe genome project (SWISS-PROT accession number Q10169). The gene encodes a protein that has 46% identity to an S. cerevisiae protein, Dsk2p (Biggins et al., 1996), throughout its entire length (Fig. 6A) and has therefore been named dphl (Dsk2p pombe homolog). Like S. cerevisiae Dsk2p, S. pombe dph1 has a characteristic ubiquitin-like motif at the N-terminal end.

Like mad2 and mphl, dphl overexpression also caused a metaphase arrest in S. pombe. Complete inhibition of cell division required 22 hours of expression from the full strength nmtl promoter in pREP3X. At that time point, 15% of the cells had hypercondensed DNA and a short mitotic spindle (Fig. 6B), indicating that they were arrested at metaphase. To test whether the metaphase arrest was spindle checkpoint dependent, dphl was overexpressed in mad2^∆ cells, that are checkpoint defective. Unlike a wild-type culture, in which dph1 overexpression caused 15% of the cells to arrest with a short mitotic spindle, in mad2^∆ only 2-3% of dph1 overexpressing cells had a short mitotic spindle, which is identical to the percentage of such cells in a wild-type asynchronous culture. The observation that mad2^∆ cells overexpressing dph1 did not accumulate in metaphase and had low MPF kinase activity (data not shown) indicates that the metaphase arrest is dependent on mad2p. In addition, abnormal V-shaped and star-shaped spindle structures were observed in 2-3% of both wild-type and mad2^∆ cells overexpressing dph1 for 24 hours (Fig. 6D). These spindles are very similar to the monopolar spindles observed in sad1^tsS. pombe cells that have a temperature sensitive mutation in an essential SPB component (Hagan and Yanagida, 1995) and in S. cerevisiae cells overexpressing DSK2 (Biggins et al., 1996).

To study the in vivo functions of dph1, a null mutant (dph1^∆), was constructed by replacing the dph1 ORF with the ura4 DNA fragment (see Materials and Methods). dph1^∆ cells were viable at temperatures ranging from 18°C-36°C (data not shown), demonstrating that it is not an essential gene. dph1^∆ cells were not hypersensitive to TBZ, compared to wild-type cells (Fig. 6C), indicating that the gene is not required for the spindle checkpoint.

Because neither S. pombe dph1^∆ cells nor S. cerevisiae dsk2^∆ cells have any detectable morphological or growth defects, it is not possible to perform functional complementation analysis on these two genes. To test whether S. cerevisiae DSK2 and S. pombe dph1 have similar functions, we tested the effects of DSK2 overexpression in S. pombe. DSK2 was overexpressed in wild-type S. pombe cells under the control of the nmt1 promoter in pREP3X. Twenty-two hours after transcriptional induction, 14% of the DSK2 overexpressing cells had hypercondensed DNA and a short mitotic spindle (Fig. 6B), similar to the phenotype of dph1 overexpressing cells, demonstrating that overexpression of both genes caused a metaphase arrest in S. pombe cells.

We have described the isolation and characterization of the S. pombe mph1 gene. The following evidence shows that mph1 in S. pombe functions in the spindle checkpoint pathway: (1) mph1 protein is structurally and functionally similar to the S. cerevisiae spindle checkpoint kinase Mps1p; (2) mph1^∆ cells are hypersensitive to a microtubule depolymerizing drug thiabendazole (TBZ), which is one of the hallmarks of a spindle checkpoint mutant (Hoyt et al., 1991; Li and Murray, 1991); (3) mph1^∆ releases the metaphase arrest due to the absence of a mitotic spindle in nda3^cs cells, resulting in an aberrant mitosis and cell lethality, another hallmark of a spindle checkpoint mutant; (4) similar to MPS1 overexpression in S. cerevisiae, mph1 overexpression in S. pombe causes a metaphase arrest in a checkpoint dependent manner. The fact that overexpression of MPS1 and mph1 in fission yeast have the same phenotypes, that the TBZ hypersensitivity of the S. pombe mph1^∆ mutation was partially rescued by S. cerevisiae MPS1, and that the mps1-1 and mps1-pac8 defects were partially rescued by expression of S. pombe mph1 provides further evidence that these two kinases are functionally related. We also showed that the metaphase arrest caused by mad2 overexpression did not depend on the mph1 gene, demonstrating that within the spindle checkpoint pathway, mph1 functions upstream of mad2, in a fashion similar to MPS1 in S. cerevisiae. Taken together, these results suggest that mph1 and MPS1 share related functions in the spindle checkpoint.

In addition to its function in the spindle checkpoint, MPS1 is also essential for SPB duplication in S. cerevisiae, which is not true for S. pombe mph1, since deletion of mph1 does not cause any growth defect under normal conditions. The functions of MPS1 in SPB duplication and the spindle checkpoint can be temporally separated and do not seem to be inter-dependent (Weiss and Winey, 1996). One interpretation of these data is that the Mps1p kinase phosphorylates different substrates in these two independent biological processes. Since mph1 is not essential, it is possible that there are two separate protein kinases required for these two processes in S. pombe: an unidentified Mps1p-like kinase required for SPB duplication, and mph1 which performs the Mps1p-like functions only in the spindle checkpoint pathway in S. pombe. It is not yet clear why expression of S. pombe mph1 partially complements the SPB defect in the S. cerevisiae mps1-1 mutant, yet does not play an essential role in SPB duplication in fission yeast. One possibility is that two mps1-like kinases in S. pombe have partially redundant functions in SPB duplication. Another possibility is that the S. pombe homolog of the SPB duplication protein phosphorylated by Mps1p in S. cerevisiae is regulated independently of this family of kinases or that no such homolog exists. Our results strength the notion that the spindle checkpoint function of Mps1p/mph1 kinase family is independent of its functions in SPB duplication. S. pombe mph1 therefore provides a convenient model for the study of the Mps1p/mph1 kinase activity regulation in the spindle checkpoint pathway.

Studies in S. cerevisiae show that Mad1p is phosphorylated specifically in response to spindle checkpoint activation, which might be a key event in spindle checkpoint activation (Hardwick and Murray, 1995). Mps1p is likely to be the kinase that phosphorylates Mad1p, in a Mad2p, Bub1p and Bub3p dependent manner (Hardwick et al., 1996). In addition, cytological studies in frog egg extract and mammalian cells show that Mad2p and Bub1p specifically associate with unattached kinetochores (Chen et al., 1996; Li and Benezra, 1996). The combination of these results suggests that the binding of Mad2p and Bub1p to unattached kinetochores activates the spindle checkpoint, possibly by promoting the phosphorylation of Mad1p by Mps1p. This is consistent with the fact that in both S. cerevisiae and S. pombe, the metaphase arrest caused by Mps1p/mph1 overproduction is mad2-dependent (Hardwick et al., 1996; this study). Interestingly, in S. cerevisiae, Mad1p phosphorylation level is not affected by mad2^∆ when the checkpoint is activated by MPS1 overexpression (Hardwick et al., 1996) and yet, MAD2 is absolutely required for Mad1p phosphorylation when the checkpoint is activated by microtubule depolymerizing drug treatment (Hardwick and Murray, 1995). However, it has not been determined whether the phosphorylation site(s) on Mad1p is the same under these conditions and a putative mad1 ORF (on Cosmid SPBC3051), that has recently been identified by the S. pombe Genome Sequencing Project (Sanger Centre, Hinxton, Cambridge, UK) has not yet been characterized. It is difficult, however, to incorporate our observation regarding mad2 overexpression into this model. In S. pombe, mad2 overexpression causes a metaphase arrest independent of mph1 function. Therefore, if the two checkpoint pathways function similarly, it seems unlikely that excess mad2 protein could activate the spindle checkpoint in S. pombe only by inducing the mph1-dependent mad1p phosphorylation. It is possible that Mad1p phosphorylation induces downstream biochemical reactions, such as Mad1p and Mad2p complex formation (Hardwick and Murray, 1995), or the binding of Mad1p and Mad2p to other downstream checkpoint proteins, possibly Bub2p and Mad3p, which further induces the cell cycle arrest. Excess mad2 in S. pombe may directly induce these downstream events and activate the checkpoint without mad1 phosphorylation. At this time, we can not exclude the possibility that excess mad2 could stimulate mad1 phosphorylation by some non-specific kinase(s) thereby activating the checkpoint.

In the screen that led to the identification of mad2 and mph1, we also isolated the cDNA of dph1, which encodes a protein highly similar to S. cerevisiae Dsk2p. Both S. pombe dph1 protein and S. cerevisiae Dsk2p have a characteristic ubiquitin-like domain at the N terminus. In addition to their structural similarity, our initial characterization suggests that these two proteins have similar in vivo functions: both are non-essential; both have comparable cell cycle effects when overexpressed in their own species; and finally, S. cerevisiae DSK2 and S. pombe dph1 have identical effects when overexpressed in S. pombe. We have also demonstrated that the cell cycle arrest caused by dph1 overexpression is dependent on the spindle checkpoint. Budding yeast DSK2 was originally isolated as a dominant suppressor of kar1^ts, which is defective in SPB duplication (Biggins et al., 1996). The biological function of S.p. dph1 is unknown, however, the observation that overexpression of dph1 in mad2^∆ cells caused the formation of a monopolar spindle, suggests the possibility that dph1 is involved in SPB duplication. Studies of S. pombe SPBs using electron microscopy have revealed significant differences in SPB structure and the process of its duplication between S. pombe and S. cerevisiae (Ding et al., 1997). Nevertheless, some key protein components, such as γ-tubulin (Masuda and Shibata, 1996) and calmodulin (Moser et al., 1997), have been localized at the SPBs of both yeasts. More importantly, their function as the spindle organizing center is similar (Snyder, 1994). Our results show that S. cerevisiae DSK2 and S. pombe dph1 have similarities in both structure and function, which further supports the hypothesis that the two evolutionarily divergent yeasts may share common components of SPB structure and duplication.

In conclusion, we have isolated mph1, an Mps1p-like kinase gene in S. pombe, and found that it plays an essential role in the spindle checkpoint pathway, in a fashion similar to MPS1 in S. cerevisiae. The fact that mph1 is not essential strongly supports the hypothesis that the function of mph1/Mps1p kinase family in the spindle checkpoint is independent of its role in SPB duplication.

The authors thank N. Ong, P. Straight, and H. Chial for help with experiments, M. Rose for th S.c. DSK2 cDNA, RLDB for cosmids ICRF c60H0216D and SPAC26A3, K. Gull for TAT1 antibody, and the Sazer and Winey laboratories for helpful input. This work was supported by grants from the National Institutes of Health to S.S. (GM49119) and to M.W. (GM51312).