A Novel Protein, Psp1, Essential for Cell Cycle Progression ofSchizosaccharomyces pombe Is Phosphorylated by Cdc2-Cdc13 upon Entry into G0-like Stationary Phase of Cell Growth*

A novel gene, psp1 +, which functionally complements a temperature-sensitive mutant defective in cell cycle progression both in G1/S and G2/M has been isolated from the genomic and cDNA libraries ofSchizosaccharomyces pombe. Disruption of this gene is lethal for cell growth at 30 °C indicating that it is an essential gene for vegetative cell growth. Western analysis of the protein by polyclonal antibody made from glutathioneS-transferase-Psp1 fusion protein indicated that the Psp1 protein exists in two different molecular weight forms depending on the growth state of the cell. In vitro experiments with a phosphatase showed that this difference is due to phosphorylation. The dephosphorylated form of the protein is dominant in actively growing cells whereas the phosphorylated form becomes the major species when cells enter the stationary phase. The Cdc2-Cdc13 complex is shown to phosphorylate the GST-Psp1 fusion protein in vitro, and site-directed mutagenesis and phosphoamino acid analysis indicated that the serine residue at position 333 in the carboxyl-terminal region is required for phosphorylation. In situ fluorescein isothiocyanate-conjugated antibody staining showed that this protein tends to be localized to both ends of the cell upon entry into the stationary phase of cell growth. However, overexpression of the novel protein Psp1 in actively growing cells inhibits cell growth causing accumulation of DNA (4n or 8n). Thus we speculate that Psp1 can function at both G1/S and G2/M phases complementing the defect of the new mutant we have isolated. It is likely that Psp1 is required both for proper DNA replication and for the process of mitosis.

DNA replication (S) and mitosis (M) are the two major events in the eukaryotic cell division cycle and are proceeded by the two gap periods, G 1 and G 2 , respectively. Cell cycle transition from G 1 to S phase (G 1 /S) and from G 2 to mitosis (G 2 /M) occurs in a strict sequence. Blocking S phase prevents onset of the subsequent mitosis, which would be lethal if chromosome replication had not been completed, and blocking mitosis prevents initiation of the subsequent S phase which would otherwise lead to increase in ploidy. Dependences of S phase and mitosis are examples of checkpoints that ensure orderly progression through the cell cycle. Many genes are thought to be involved in this process, and tight regulation of function or expression of the relevant genes is required for proper cell cycle progression. The checkpoint control at the two transition points, G 1 /S and G 2 /M, ensures either progression or blocking of cell cycle according to the state of the cell. Several regulators such as cyclin-dependent kinases (CDKs) 1 function at these points (1,2). CDK activity is subject to regulation by association with positive regulatory subunits known as cyclins, negative regulators known as CDK inhibitors (CKIs), and by phosphorylation (reviewed by Pines (3), Morgan (4), Elledge and Harper (5), and Harper and Elledge (6)). The balanced function of these factors controls CDK activity and serves to integrate signals intended to coordinate cell cycle transitions. The levels of these proteins are tightly regulated both transcriptionally and posttranslationally as the cell cycle progresses. In yeast a single CDK, Cdc2 in fission yeast and Cdc28 in budding yeast, is required at the first checkpoint G 1 /S (7)(8)(9)(10)(11)(12), while in multicellular eukaryotes, several different CDKs, including CDK2, CDK4, and CDK6, are involved at different stages of G 1 and S phase (13)(14)(15)(16)(17). Different G 1 cyclins such as CLN1, CLN2, and CLN3 of budding yeast or cyclin D and cyclin E of mammalian cells and S phase cyclins such as CLB5, CLB6, or cyclin E and cyclin A are associated with these CDKs to ensure kinase function of the CDKs (18 -22) at G 1 to S phase progression. The phosphorylation state of the CDK itself and proper timing of destruction of the cyclins are important for regulation of CDK activities (4). Thus checkpoint control at G 1 /S by these multiple elements is important for the cell to commence DNA replication, to initiate the cell cycle, and to integrate the positive and negative signals of the cell cycle. At the second checkpoint, G 2 /M, p34 cdc2 , the prototypic member of cyclin-dependent kinase, is known to be the major regulator both in yeast and mammalian cells. Association with G 2 cyclins such as cyclin B is required to function as an active protein kinase and to induce active mitosis (23,24). The phosphorylation state of Cdc2 directly affects its kinase function and cell cycle progression to mitosis (4,25,26). Cdc25 dephosphorylates Cdc2 at serine/ threonine residues, and this activates the kinase function of the Cdc2-Cdc13 (cyclin B) complex (27)(28)(29). Meanwhile, phosphorylation of Cdc2 by Wee1 kinase results in inactivation of Cdc2 kinase function at G 2 /M (30 -33), and destruction of G 2 cyclin by the ubiquitin-associated cyclosome complex signals the cell to exit from mitosis and to prepare for the next round of G 1 to S phase progression (34). The CKIs such as p21, p27, or p15, which have negative function to inhibit CDK activity (5,35), counteract the cyclins and are required to prevent cell cycle progression when cells are damaged. Mutations in CKI genes cause unregulated cell cycle progression, and this results in abnormal cell growth. In particular, deregulation of START at G 1 /S may allow cell growth and division to become insensitive to external cues (36 -40). This can be a consequence of either the aberrant expression of positive regulators, such as the cyclins, or the loss of negative regulators, such as the CKIs. Thus proper regulation of cyclin-CDK-CKI complex formation is critical for normal cell cycle progression.
Even though the global cell cycle controls operating at the S (22,(41)(42)(43)(44)(45) and M phase transitions (46 -48) have been extensively analyzed, unknown relevant factors still remain to be identified. Fission yeast Schizosaccharomyces pombe has been a useful model organism to study cell cycle control in eukaryotic cells. The relative ease of genetic manipulation of yeast has allowed detailed analysis of gene function in vivo, which has produced a paradigm for cell cycle control applicable to higher organisms. In an attempt to isolate factors involved in control of cell cycle progression, we searched for new cell cycle-related genes in S. pombe. In this paper we describe isolation of a novel gene, psp1 ϩ (a gene encoding phosphoprotein at stationary phase of S. pombe cell), essential for progression of cell cycle. We used a strategy to isolate a new conditional mutant defective in cell cycle progression by chemical mutagenesis and to find a clone functionally complementing this mutant phenotype. Genetic and biochemical studies have permitted us to identify the function of the newly isolated psp1 ϩ in S. pombe cells.
Isolation of Temperature-sensitive Mutants of S. pombe Defective in Cell Cycle Progression-To isolate ts Ϫ mutants of S. pombe, ED665 cells grown in YEPD medium were mutagenized with ethylmethanesulfonate (EMS) to 40% survival as described previously (51,52). The mutagenized cells were spread on YEPD plates and incubated at 23°C for 4 -6 days. The cells grown on the plates were replica-plated onto YEPD plates containing phloxin B, and the plates were incubated at 23, 36, and 39°C for 3-4 days. The cells that grew at 23°C but did not grow at 36 or 39°C were selected. Out of 2.5 ϫ 10 5 cells mutagenized, 180 candidate colonies were first selected and examined for their morphological changes at the restrictive temperature (36°C) under a light microscope. The 51 colonies that showed defects in cell division were further characterized by back-crossing to the parent strain ED665 or ED668 and by genetic complementation test with the known cell division cycle mutants. Their DNA contents at 36°C were determined by flow cytometrical analysis as described in Jang et al. (52). One mutant, cyj92, that showed an elongated morphology at 36°C and a defect in cell cycle progression both at G 1 /S and after G 2 , was selected and used to clone a gene recovering the mutated phenotype.
Flow Cytometric Analysis-To measure DNA content in each cell cycle point, the cells were prepared by a procedure adapted from Lew et al. (53) and Costello et al. (54). The mutant cells were first grown in YEPD liquid medium at 23°C to midlog phase (A 580 ϭ 0.5) and transferred either directly to fresh YEPD medium or to nitrogen-deficient EMM (0.003% NH 4 Cl) which leads to cell cycle arrest at G 1 in S. pombe. The cells transferred to EMM were incubated at 23°C for 24 -36 h to induce cell cycle arrest at G 1 , and were then collected, washed, and resuspended in YEPD medium. The cells transferred to YEPD medium were incubated at 36°C for 4 -8 h to induce mutant phenotype. Aliquots of these cells (1 ϫ 10 7 cells) were collected, fixed with 70% ethanol for at least 12 h at 4°C, and treated with RNase A (Sigma, 0.1 mg/ml) for 2 h at 37°C. After staining the cells with propidium iodide solution (0.005% propidium iodide, 0.1% sodium citrate, and 0.03% Triton X-100), the DNAs in the cells were analyzed with Becton Dickinson FACScan System (FACScan TM ). Analysis was based on the accumulation of 8,000 -10,000 cells. As a reference, flow cytometric analysis of the mutant cdc2 Ϫ or cdc10 Ϫ was carried out simultaneously. To measure DNA content in cells overexpressing psp1 ϩ , the cells containing pnmt1-psp1 plasmid were first grown in EMM containing 100 M thiamine and then transferred to EMM devoid of thiamine and grown for 12 h. Aliquots of the cells were reinoculated into fresh EMM lacking thiamine to a cell density A 580 ϭ 0.1 and incubated at 30°C. Cell samples were collected thereafter every 3 h and subjected to flow cytometric analysis.
Synchronous Cell Cultures-To examine the expression pattern of the psp1 ϩ gene during cell cycle progression, synchronized cell cultures were used. The cells arrested at S phase by hydroxyurea or M phase by thiabendazole were used (55)(56)(57). ED665 cells were first grown in EMM to a cell density of A 580 ϭ 0.5 (1 ϫ 10 7 cells/ml) and then in the presence of 12 mM hydroxyurea or 100 M thiabendazole at 30°C for an additional 4 -5 h to arrest them at S or M phase, respectively. The cells were then collected by filtration, washed, transferred to fresh EMM, and incubated at 30°C to release them from S or M phase arrest. Aliquots of the cells were collected every 20 min thereafter and counted. Total RNA was prepared from these cell samples.
Cloning of a Gene Complementing the Elongated Mutant Phenotype of cyj92-To identify a gene responsible for blockage of cell cycle progression when mutated as shown in the mutant cyj92, a functional complementation test of the mutant with a S. pombe genomic library was carried out. The mutant cyj92, showing an elongated morphology and defect in cell cycle progression at 36°C, was transformed with a S. pombe genomic library prepared by ligating genomic DNA partially digested with Sau3AI into the BclI site of pWH5 vector (58,59). 2 The transformants showing Leu ϩ at 23°C were first selected, replica-plated onto the plates containing phloxin B, and incubated at 36°C. The pink colonies that grew up and showed wild type morphology at 36°C were selected. Individual plasmid isolated from these transformants was retransformed to test its ability to suppress the elongated ts Ϫ phenotype of cyj92 (50). Two such plasmids, containing 8.5-kb and 6.9-kb insert DNA, were isolated and characterized by restriction enzyme mapping and Southern hybridization. The 4.2-kb Sau3AI-PvuII fragment present in both of these plasmids and possessing the suppressive function was subcloned (Fig. 2a, pYJ4) and sequenced by the dideoxy method of Sanger et al. (60) with USB Sequenase (U. S. Biochemical Corp.). cDNA clones were also isolated by screening a S. pombe cDNA library constructed in the -ZAP vector (61) with the labeled 1.5-kb PvuII-BamHI fragment encoding most of the coding sequences. The DNA containing open reading frame sequences was designated as psp1 ϩ (phosphoprotein of stationary phase of S. pombe). The disruption construct of the psp1 gene, with the ura4 ϩ as the selectable marker, was made as follows (Fig. 2a, pYJ5). The PvuII-EcoRV fragment containing most of the Psp1 coding sequences, including 5Ј-nontranslating region DNA, was deleted and replaced with the 1.8-kb ura4 ϩ fragment. The resulting 3.5-kb BamHI-ApaI fragment was used to transform a diploid strain SP286. Transplacement of the genomic psp1 gene with the ura4 ϩ disrupted copy was confirmed by Southern analysis. The disrupted diploid strain was then sporulated according to the methods of Gutz et al. (49), and viability of the resulting meiotic products was analyzed (Fig. 2c).
Primer Extension Analysis-To determine the transcription start site of psp1 ϩ , the primer extension method was used (62). The antisense primer sequence that corresponds to the sequence from ϩ150 to ϩ169 (5Ј-GAGCAATTTCATTATCATG-3Ј) of the amino-terminal coding re-gion was synthesized and labeled with T4 polynucleotide kinase (New England Biolabs) and [␥-32 P]ATP. The labeled primer was then hybridized with 5-10 g of poly(A) RNAs isolated from the ED665 strain grown in EMM. Moloney murine leukemia virus reverse transcriptase (New England Biolabs) was added to the hybridization mixture and incubated at 37°C for 30 min. The position of transcription start site was determined by comparing the size of the extended fragment with the sequenced DNA using the same primer.
Preparation of Membrane and Cytosolic Fractions-To determine the location of the Psp1 protein in S. pombe cells, membrane and cytosolic fractions of the cells were prepared (63). Cells were broken by the glass bead method (64) in bead buffer (20 mM Tris-HCl, pH 7.5, 10 mM EGTA, 2 mM EDTA, 0.25 M sucrose, 20 g of leupeptin/ml, and 10 M phenylmethylsulfonyl fluoride) and centrifuged at 100,000 ϫ g for 1 h at 4°C. The supernatant was used as a cytosolic fraction. The pellet fraction was treated with 1% Triton X-100 for 30 min at 4°C and used as a membrane fraction.
Production of GST-Psp1 Fusion Proteins-To prepare the protein product of the psp1 gene, a coding region DNA of psp1 ϩ was fused to the isopropyl-1-thio-␤-D-galactopyranoside-inducible glutathione S-transferase (GST) of fusion vector pGEX-3X. The GST-psp1 fusion constructs that encode different portions of Psp1 were prepared as follows (Fig. 4c,  A). The DNA fragment containing the whole coding region (w), aminoterminal sequence 1-307 (Nw), or carboxyl-terminal sequence 323-408 (Cw) was cloned into the BamHI site of plasmid pGEX-3X. The DNA fragments encoding carboxyl-terminal sequences 323-408 (Cm1), 330 -408 (Cm2), and 338 -408 (Cm3), bearing substitutions of serine residues at 327, 333, and 341 with alanine, respectively, were fused into the same vector. Escherichia coli strain DH5␣ transformed with these fusion plasmids was grown in LB medium containing ampicillin, and production of GST-Psp1 fusion proteins was induced by growing cells in the presence of 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside at 30°C. Each GST-Psp1 fusion protein was purified on glutathione-agarose beads as described previously from E. coli crude cell extracts (65). For in vitro phosphorylation of Psp1 protein by Cdc2 kinase, GST and GST-Psp1 fusion proteins purified on glutathione-agarose beads were eluted with 10 mM reduced glutathione in 50 mM Tris-HCl, pH 8.0.
Immunochemical Analysis-To generate polyclonal antibody against Psp1 protein, affinity-purified GST-Psp1(w) protein was treated with factor Xa enzyme and separated on SDS-polyacrylamide gel (65). The Psp1 protein band on the gel was eluted, mixed with Freund's adjuvant solution, and used to inject a rabbit. After two more successive injections at 2-week intervals, serum was collected and used as a source of antibody for Western blot analysis and immunocytological experiments (74). For determining cellular localization of Psp1 protein, FITC-conjugated anti-rabbit IgG (Sigma, Catalog No. F7512) was used. Cells grown at log or stationary phase were collected, washed, and fixed with methanol (56,66,67). Antiserum prepared against Psp1 was mixed first with the fixed cell suspension and then treated with FITC-conjugated anti-rabbit IgG in a concentration of 1/50 dilution. Nuclei of the cells were counterstained with 4Ј,6Ј-diamidino-2-phenylindole (DAPI) (56).
Expression of psp1 Gene in S. pombe Cells-Transcriptional expression of psp1 ϩ during cell cycle progression was examined by Northern analysis of total RNA from cells grown synchronously by the arrestrelease method with hydroxyurea. Total RNA was prepared from the cells collected every 20 min by the method of Carlson and Botstein (68) and analyzed by hybridizing them with the 1.5-kb PvuII-EcoRV DNA fragment radioactively labeled by the random priming method. The expression level of Psp protein was also examined with the same synchronized cells released from the hydroxyurea arrest. Cell extracts were prepared, and the level of Psp1 protein in both soluble and particulate fractions was examined by Western analysis using the polyclonal antibody prepared from the GST-Psp1 fusion. The effect of overexpression of Psp1 protein in S. pombe cells was examined using the pnmt1-psp1 fusion construct. The coding region DNA of psp1 ϩ was amplified from S. pombe chromosomal DNA with the 5Ј-and 3Ј-end primer sequences and ligated into the SalI site of plasmid vector pREP1 containing thiamineregulated nmt1 promoter (69). The resulting plasmid pnmt1-psp1 was transformed into S. pombe cells, and expression of Psp1 protein was induced by growing cells in the absence of thiamine for more than 16 h.
Site-directed Mutagenesis Using PCR Methods-To introduce mutations substituting serine residues at the putative Cdc2 substrate sites, 324 LVQSPSC, 330 CPPSPKN, and 338 AHLSPGS to alanine, we used PCR methods with mutated primer sequences (70). Change of the first T of the triplet sequence encoding serine (TCA, TCC, TCT) to G allows substitution of serine with alanine. Thus the following oligonucleotides, corresponding to the amino acid sequences LVQAPSC (m1), CPPAPKN (m2), and AHLAPGS (m3), were synthesized: the sense strands of m1, 5Ј-CTTGTGCAGGCACCTTCTTGC-3Ј, m2, 5Ј-TGCCCTCCTGCCCCA-AAAAAT-3Ј, and m3, 5Ј-GCACATCTTGCCCCTGGATCC-3Ј and antisense strands of m1, 5Ј-GCAAGAAGGTGCCTGCACAAG-3Ј, m2, 5Ј-A-TTTTTTGGGGCAGGAGGGCA-3Ј, and m3, 5Ј-GGATCCAGGGGCAA-GATGTGC-3Ј. Also, the sense strand of the amino terminus, 5Ј-ATGC-CTTTGTCAACTCAATCG-3Ј, and the antisense strand of the carboxyl terminus, 5Ј-TCACCGACGTGGTGTATCTAC-3Ј were synthesized. To introduce point mutation at the sequence encoding serine 327 (Ser-327-Ala), the two primer pairs, sense strand of amino terminus and antisense strand of m1 and sense strand of m1 and antisense strand of carboxyl terminus were used for the first PCR. Then the second PCR was carried out using the first PCR product as a template and sense  4). The abscissa indicates intensity of fluorescence per cell, and the ordinate indicates cell numbers. The vertical line at the right side corresponds to 2n (G 2 ), and the one at the left side corresponds to n (G 1 ). c, back-cross test of the mutant. The mutant cyj92 was crossed to wild type ED668 cells, and the diploid was sporulated. The resulting tetrads were incubated at 23°C first and replica-plated and incubated at 36°C. strand sequence of amino terminus and antisense strand sequence of carboxyl terminus as the primers. Amplified DNA was then cloned into vector plasmid pTZ18U. The change in the sequence at residue 327 was confirmed by sequencing the amplified fragment (Fig. 4c, B). Mutations at Ser-333 and Ser-341 were introduced by similar PCR methods. The primer pairs of the sense strand of amino terminus and antisense strand of m2 or m3 were used for the first PCR and followed by the second PCR with the amino-and carboxyl-terminal sequence pairs. For overexpression of these mutated Psp1 in ED665, amplified DNAs were cloned into plasmid vector pREP1 containing the thiamine-inducible nmt1 promoter (pnmt1-m1, pnmt1-m2, and pnmt1-m3).
Phosphatase Treatment and Phosphorylation of Psp1 Protein by the Cdc2-Cdc13 Complex-To examine the phosphorylation state of Psp1 protein at different stages of cell growth, cellular proteins were treated FIG. 2. Sequence of psp1 ؉ gene complementing the mutant phenotype of a novel ts ؊ mutant cyj92. a, map of genomic clones and their subclones complementing the mutant phenotype of cyj92 and a disruption clone. ϩ indicates suppression of the mutant phenotype. b, nucleotide and deduced amino acid sequence of psp1 ϩ gene. Lowercase and uppercase letters indicate genomic and cDNA sequences, respectively. The arrowheads indicate the start sites and the end of two cDNAs. The short intron sequence at the 5Ј-nontranslating region is absent in the cDNA clone 1. The potential Cdc2 substrate sequences, 327 SPSC, 333 SPKN, and 341 SPGS were underlined. c, effect of disruption of psp1 on cell viability. Diploid SP286 strain containing psp1 gene disrupted with ura4 ϩ as evidenced by the Southern analysis of its chromosomal DNA (lane 2 in 1) produced two viable and two nonviable spores at 30°C on tetrad analysis (2). These sequence data are available from Gen-Bank TM under accession number L36906 and were deposited in August, 1994.
in vitro with phosphatase and analyzed. Cells were grown to log or stationary phase, collected, and broken in phosphatase buffer (50 mM Tris-HCl, pH 7.8, 5 mM dithiothreitol, 2 mM MnCl 2 , 100 g/ml bovine serum albumin) by the glass beads method (50). After removing the glass beads by centrifugation at 500 ϫ g, the whole cell extract was centrifuged at 100,000 ϫ g for 1 h. The supernatant was separated, and the pellet fraction was solubilized in phosphatase buffer containing 1% Triton X-100. 5-10 l of supernatant, solubilized pellet fraction, or whole cell extract was mixed with phosphatase buffer and incubated at 30°C for 30 min in the presence of 400 units of protein phosphatase (-PPase, New England Biolabs) (71,72). The reaction mixture was boiled in SDS-polyacrylamide gel (SDS-PAGE) loading dye for 5 min. The proteins were then separated on SDS-PAGE and electrotransferred to a nitrocellulose paper for Western analysis with the polyclonal antibody against Psp1. To determine whether Psp1 could be phosphorylated in vitro, 1-2 g of purified GST-Psp1 protein was mixed with 10 units of mitotic Cdc2-Cdc13 complex (New England Biolabs) and incubated at 30°C for 30 min in the presence of 40 Ci of [␥-32 P]ATP. The resulting protein was then resolved in SDS-PAGE and analyzed by autoradiography. As a standard substrate of Cdc2 kinase, 2 g/ml histone H1 (Boehringer Mannheim) was used (73). In vivo labeling of the protein was carried out with ortho[ 32 P]phosphate according to the procedure of Moreno et al. (50). Cells were first grown to late log phase (A 580 ϭ 2.5-3.0) in 5-10 ml of low phosphate medium (MMP plus 1 mM phosphate) and then transferred to the same volume of fresh MMP containing 1 mCi of ortho[ 32 P]phosphate (Amersham, Catalog No. PBS13). After incubation at 30°C for 5-6 h, cell extracts were prepared from the harvested cells as described above. For immunoprecipitation, 800 -1,000-l aliquots of the cell extracts were mixed with 20 l of antisera of Psp1 (final 1/40 dilute) and 50 l of protein A-agarose (Sigma). The immunoprecipitate was analyzed on SDS-PAGE and by autoradiography (74).
Phosphoamino Acid Analysis-To determine whether the phosphorylated amino acid is serine as expected for phosphorylation by the Cdc2-Cdc13 complex, phosphoamino acid analysis of Psp1 was carried out. The protein labeled in vivo by ortho[ 32 P]phosphate was immunoprecipitated with Psp1 antibody and separated by SDS-PAGE. At the same time, GST-Psp1 protein phosphorylated in vitro by the Cdc2-Cdc13 complex with [␥-32 P]ATP was also separated by SDS-PAGE. After being transferred to a polyvinylidene difluoride membrane (Millipore) and localized by autoradiography, the phosphorylated Psp1 band was excised and subjected to phosphoamino acid analysis (64,75). The protein on the membrane was eluted and hydrolyzed in 6 N HCl for 1 h at 110°C. After being dried in a Speed Vac evaporator, the sample was dissolved in 6 -10 l of water containing 100 g/ml each of phosphoserine, phosphothreonine, and phosphotyrosine. The sample was subjected to two-dimensional thin layer electrophoresis (HTLE 7000 apparatus, CBS Scientific). The first dimension was electrophoresed for 20 min at 1.5 kV in pH 1.9 buffer (50 ml of 88% formic acid, 156 ml of glacial acetic acid, and 1794 ml of water). The second dimension was electrophoresed for 16 min at 1.3 kV in pH 3.5 buffer (100 ml of glacial acetic acid, 50 ml of pyridine, and 1880 ml of water). 32 P-Labeled individual phosphoamino acids were identified by aligning the phosphoamino acid markers stained with ninhydrin to the signals obtained by autoradiography.

Characteristics of a Novel Temperature-sensitive (ts Ϫ ) Mutant Defective in Cell Cycle Progression and a Gene Complementing This ts
Ϫ Phenotype-Through EMS mutagenesis of S. pombe cells, several new temperature-sensitive mutants defective in cell cycle progression were obtained (52). Morphological analysis of one of the mutants, cyj92, showed a normal cell shape at 23°C but showed an elongated cellular morphology at 36°C (Fig. 1a). DNA content analysis of this mutant cell indicated a defect in cell cycle progression (Fig. 1b). When cyj92 cells were grown in rich medium at 23°C, most of the cells were at G 2 phase as was the case for wild type S. pombe cells (Fig. 1b,  4, panel 1). But when they were shifted to a restrictive temperature 36°C, DNA content increased and in the majority of cells the DNA content was 4n (Fig. 1b, 4, panel 2) indicating blockage of cell cycle after G 2 phase. When the mutant cells were grown in nitrogen-deficient medium at 23°C instead, cell growth was arrested at G 1 phase as shown in wild type cells (Fig. 1b, 3, panel 2). However, if these G 1 -arrested cyj92 cells were shifted to YEPD medium and incubated at 36°C, unlike wild type cells, they could not progress completely to G 2 phase, and cell cycle progression stopped at S phase (Fig. 1b, 3, panels  3 and 4). A known G 1 /S mutant, cdc10 Ϫ , and G 2 /M mutant, cdc2 Ϫ , showed cell cycle arrest at G 1 after nitrogen starvation at permissive temperature 23°C also (Fig. 1b, 1 and 2, panels 2) when they were tested simultaneously with cyj92. They   FIG. 2-continued showed their mutant phenotypes at the restrictive temperature (36°C); that is, blockage of cell cycle progression to G 1 /S phase and after G 2 phase, respectively, as expected (Fig. 1b, 1 and 2,  panels 3 and 4). Back-cross of the mutant strain cyj92 to wild type strain ED668 revealed 2 ϩ :2 Ϫ segregation of the mutant phenotype at 36°C which indicates that the mutation shown in cyj92 is a single gene mutation (Fig. 1c). Genetic complementation tests of cyj92 with known cdc mutants of G 1 and S such as cdc10 Ϫ , cdc20 Ϫ , or cdc22 Ϫ and G 2 /M phase mutants cdc2 Ϫ and wee1 Ϫ showed that the mutation in strain cyj92 was not allelic with any of them. Thus we concluded that the mutant we had isolated is a new ts Ϫ mutant possessing a defect in cell cycle progression after the G 2 phase at restrictive conditions and also at G 1 /S phase when cells pass the G 2 phase normally. It is likely that once the mutant cells progressed to G 2 phase, the mutation in cyj92 can cause blockage of further progression to M or cytokinesis. If mutant cells were in G 1 , this progression through the next S phase was also adversely affected. It is alternatively possible that the mutation carried in strain cyj92 causes blockage of cell cycle progression both at G 1 /S and G 2 /M. Functional complementation of this mutant with a S. pombe genomic library identified the two overlapping clones of 8.5 kb and 6.9 kb (pYJ1 and pYJ2). Subcloning and retransformation data confirmed that the 4.2-kb PvuII-Sau3A fragment (pYJ4) was sufficient to suppress the ts Ϫ phenotype (Fig. 2a). cyj92 mutant cells harboring plasmids pYJ1 or pYJ4 showed normal cell morphology at 36°C, and their DNA content analysis indicated normal cell cycle progression at the restrictive temperature, 36°C. Increased ploidy seen in cyj92 mutant cells (Fig.  1b, 4, panel 2) also disappeared in the transformed cells (Fig.  1b, 4, panel 4); transformants were mainly 2n like the wild type. Sequence analysis of this genomic DNA revealed an open reading frame sequence of 408 amino acids (Fig. 2b). Meanwhile, sequence analysis of several cDNA clones indicated the presence of two different transcripts depending on whether or not a short intron at the 5Ј-terminus was processed. As shown in Fig. 2b, one cDNA clone, which started at Ϫ90 (clone 1) did not contain sequences between Ϫ57 and Ϫ2 (intron). However, the other cDNA clone, which started at Ϫ88 (clone 2) contained these sequences. Primer extension experiments using mRNA isolated from cells grown in EMM also revealed two transcript start sites correlating with the two cDNA clones that differed by the presence or absence of the intron sequence (data not shown). Hydropathy plot analysis of the deduced amino acid sequence of psp1 ϩ showed that it is very hydrophobic and contains a putative transmembrane domain at its carboxyl terminus (amino acids 353-369). Sequence homology comparison of this protein showed that no known protein sequences in the GenBank TM data base exhibited high homology with it. Thus we concluded this gene to be a newly identified gene and designated it psp1 ϩ (phosphoprotein of stationary phase of S. pombe). Sporulation and tetrad analysis of a diploid strain containing one copy of psp1 transplaced with ura4 ϩ (Fig. 2a,  pYJ5) showed 2 ϩ :2 Ϫ segregation for cell viability at 30°C (Fig.  2c). All viable spores were Ura Ϫ indicating that psp1 is essential for vegetative cell growth at 30°C.
Expression of Psp1-Northern analysis of total RNA isolated from the wild type cells grown in EMM showed a 1.3-kb transcript, and its level did not fluctuate much throughout the cell cycle. The level of psp1 transcript in the synchronized cells released from the S phase arrest was almost same as that in the cells progressing through G 2 and M phase (data not shown). The level of Psp1 protein detected by Western analysis in the same synchronized cells was also constant throughout the cell cycle (data not shown). Thus, in actively growing cells the amount of Psp1 protein as well as its transcript is likely to be constant throughout the cell cycle. When the location of Psp1 protein in cells was examined by Western analysis using fractionated cell extracts and Psp1 antibody, Psp1 protein was found to be in the pellet fraction rather than in the soluble fraction (Fig. 3a). In addition, two different molecular weight forms of Psp1 protein were detected, and the ratio of these two forms differed depending on the phase of cell growth. In actively growing log phase cells (Fig. 3a, M-1, and Fig. 3b, 1-5), the lower molecular weight protein band was the major form. However, as soon as cells enter stationary phase, the higher molecular weight species became predominant (Fig. 3a, M-2, and Fig. 3b, 6 -8). This result suggests that Psp1 protein is modified to a higher molecular weight form upon entry into G 0 -like stationary phase.
Localization of this protein in situ by FITC-conjugated antirabbit IgG and Psp1 antibody demonstrated that Psp1 protein is present through out the whole cell in the actively growing stage (Fig. 3c, 1 and 3), whereas it is localized to each end of the cell at stationary phase (Fig. 3c, 6 and 8). We suggest that when cells are actively growing, Psp1 protein function may be re- quired, and it is located throughout the whole cell body. However when growth of cells almost ceases at stationary phase, this protein may function only in specialized areas of the cellular compartment or as a complex.
Psp1 Protein Is Phosphorylated at Stationary Phase of the Cell Cycle-Detection of the different molecular weight forms of the Psp1 protein indicated that the two proteins are either produced from two different size transcripts or modified after translation. Therefore we examined the size of psp1 transcript both by Northern analysis and determining the start site of the transcript using the primer extension method. Total RNA prepared from the cells grown in EMM showed only one major hybridizing band. However, the primer extension experiment indicated that there are at least two transcription start sites and the length of the 5Ј-nontranslated region is different depending on the processing of a short sequence from Ϫ57 to Ϫ2 (Fig. 2b). Even though these two transcripts differ by a 56-base pair sequence in the 5Ј-nontranslated region, they contained the same ATG codon, and the proteins translated from these two transcripts seem to be the same. Thus we examined the possibility of modification after translation by phosphorylation. We tested whether the high molecular weight Psp1 protein in stationary phase cells contained a phosphate residue(s). Western analysis of proteins in the membrane fraction treated with phosphatase in vitro showed a reduction in the amount of the high molecular weight Psp1 protein (Fig. 4a, lane 4) and a concomitant increase in the low molecular weight band instead. In contrast, the lower molecular weight form of Psp1 present mainly in the log phase cells remained unchanged after phosphatase treatment (Fig. 4a, lane 2). This result suggests that Psp1 protein can be phosphorylated. To test this possibility, GST-Psp1 fusion protein was prepared and tested in vitro with the mitotic Cdc2-Cdc13 complex or with mitogen-activated protein kinase. When the Cdc2-Cdc13 complex was incubated with GST-Psp1 protein in the presence of [␥-32 P]ATP in vitro, GST-Psp1 protein was phosphorylated as was histone H1 control which is a known substrate of Cdc2-Cdc13 (Fig. 4b, lanes 2 and  3). Phosphorylation did not occur in the GST portion of the fusion protein but only in Psp1. However, when mitogen-activated protein kinase was mixed with GST-Psp1 protein, phosphorylation did not occur at all (data not shown). This suggests that phosphorylation of the Psp1 protein by the Cdc2-Cdc13 complex was specific.
Cdc2-Cdc13 Complex Phosphorylates a Serine Residue at the Carboxyl-terminal Domain of Psp1-Phosphorylation of GST-Psp1 protein by the Cdc2-Cdc13 complex in vitro indicates the presence of a specific sequence(s) recognized by Cdc2 in Psp1. Thus the potential Cdc2 substrate sequences, (T/S)PX(K/R), (9,76) were searched. Three such sequences were found in the carboxyl-terminal region of Psp1 ( 327 SPSC, 333 SPKN, and 341 SPGS) (Fig. 2b). When we used a GST-Psp1 fusion construct devoid of these sequences (Nw), no phosphorylation of the truncated Psp1 was observed (Fig. 4c, A, lane 2). This suggests that the carboxyl-terminal region of Psp1 is important for phosphorylation by Cdc2 kinase. The GST-Psp1 containing only the carboxyl-terminal region sequence (Cw) including those potential Cdc2 kinase substrate sites was enough to serve as the FIG. 4. Phosphorylation state of Psp1. a, phosphatase treatment of Psp1 protein in vitro. Proteins in the membrane (M) and total fraction (T) of the cells from log and stationary phase (sta) were treated in vitro with -phosphatase (PPase), separated on SDS-PAGE gels, and analyzed by Western blotting with Psp1 antibody. Ϫ and ϩ indicate the untreated and phosphatase-treated proteins, respectively. b, in vitro phosphorylation of Psp1 with the Cdc2-Cdc13 kinase complex. The GST-Psp1 fusion protein purified from E. coli was mixed with the mitotic Cdc2-Cdc13 complex in the presence of [␥-32 P]ATP and analyzed by SDS-PAGE and autoradiography. The bands in lanes 1-3 showed the phosphorylated proteins, and those in lanes 4 -6 are the same amount of the proteins as in lanes 1-3 stained with Coomassie Blue. c, identification of the serine residue phosphorylated in vitro by Cdc2-Cdc13 (A) and in vivo (B). Three serine residues (327, 333, and 341) at the potential Cdc2 substrate sites in the carboxyl-terminal region of Psp1 protein were mutated to alanine and tested for phosphorylation by Cdc2-Cdc13. Lanes 1-6 in A showed the phosphorylated part of the GST-Psp1 fusion protein. The whole Psp1 protein (w), carboxyl-terminal protein (Cw), and carboxyl-terminal protein mutated at serine 327 (Cm1) showed phosphorylation by Cdc2, but the carboxyl-terminal protein mutated at serine 333 (Cm2, lane 5) did not show phosphorylation by Cdc2. Lanes 0 -4 in B showed the phosphorylated Psp1 protein produced in vivo from the pnmt1-psp1 plasmid. The entire psp1 gene mutated at the base encoding serine residue was fused to pnmt1 vector and was expressed in wild type ED665 cells. The cells grown to late log phase in the absence of thiamine were labeled in vivo as described under "Experimental Procedures." Cell extracts were prepared and immunoprecipitated with Psp1 antibody and analyzed on SDS-PAGE gels. The open arrowhead indicates the protein produced from the chromosomal copy of the psp1 ϩ gene, and the filled arrowhead indicates that from pnmt1-psp1. The asterisk indicates the mutation point. d, phosphoamino acid analysis of Psp1 protein labeled in vitro by Cdc2 (left) and in vivo in the presence of ortho[ 32 P]phosphate (right). S indicates phosphoserine detected by autoradiography; T and Y corre-spond to the positions of phosphothreonine and phosphotyrosine, respectively, detected by ninhydrin staining as shown in 2. e, direct interaction of Psp1 with Cdc2 kinase shown by the yeast two-hybrid method. psp1 ϩ fused to the Gal4 binding domain sequence (pGBT9) and cdc2 ϩ fused to the Gal4 activation domain sequence (pGAD) were introduced into yeast S. cerevisiae strain SFY526, and expression of the lacZ gene in yeast was examined. 1 and 2, cells containing pGBT9-psp1 and pGAD-cdc2 alone, respectively; 3, cells containing both pGBT9-psp1 and pGAD-cdc2; 4, cells containing pCL1 of the whole Gal4 gene. substrate molecule phosphorylated by Cdc2 (Fig. 4c, A, lane 3). When we substituted each serine residue at 327, 333, and 341 in the carboxyl-terminal region with alanine (Cm1, Cm2, and Cm3) and examined the phosphorylation state of the truncated Psp1 protein by Cdc2-Cdc13 in vitro, the serine at position 333 was the only critical one for phosphorylation by Cdc2 kinase (Fig. 4c, A-lanes 4 -6). Experiments with in vivo labeled protein also indicated that the serine residue at 333 is required for phosphorylation (Fig. 4c, B). PCR-mediated site-directed mutagenesis of psp1 generated a base change at the sequences encoding serines 327, 333, or 341, respectively, and resulted in their being changed to alanine. When this mutated full-length psp1 was introduced in wild type cells as pnmt1-psp1 fusions (pnmt1-m1, pnmt1-m2, pnmt1-m3) (Fig. 4c, B, m1, m2, m3) and the phosphorylation state of the mutated Psp1 by in vivo phosphate labeling at stationary phase was examined, serine 333 was shown to be essential for phosphorylation in stationary phase. The cells containing the unmutated pnmt1-psp1 (w) produced more phosphorylated Psp1 than cells containing only the chromosomal copy of psp1 (lanes 1 and 0). This indicated that the Psp1 produced from the overexpression plasmid pnmt1-psp1 is phosphorylated like that in the chromosome. However, when the cells containing pnmt1-m2 (mutation at serine 333) was used, in vivo phosphate labeling of the Psp1 greatly decreased (lane 3). This decrease was not observed when pnmt1-m1 and pnmt1-m3 (containing mutations at 327 and 341, respectively) were analyzed instead (lanes 2 and 4). This confirms the importance of serine 333 for phophorylation in vivo. Phosphoamino acid analysis of the Psp1 protein phosphorylated both in vivo and in vitro confirmed that phosphorylation actually occurred at a serine residue (Fig. 4d).
To examine whether phosphorylation of Psp1 by Cdc2 occurred through direct binding of Cdc2 to Psp1, a yeast twohybrid system was employed. When the plasmids containing Gal4 binding domain-cdc2 and Gal4 activation domain-psp1 were introduced into a S. cerevisiae strain SFY526 and ␤-galactosidase production was examined, ␤-galactosidase production was observed. This indicates interaction between Psp1 and Cdc2 (Fig. 4e). The level of ␤-galactosidase produced was not as great as that from GAL4 itself (lanes 3 and 4). However there is strong indication that Cdc2 and Psp1 interact with each other.
Growth-inhibitory Effect of Overexpressed Psp1 Protein on Actively Growing S. pombe Cells-Since a gene disruption study indicated that psp1 ϩ is essential for cell growth (Fig. 2c), we examined the multicopy gene dosage effect of this essential protein on cell growth. This was done by introducing multicopy plasmid pnmt1-psp1 into wild type cells and measuring their growth rate in the presence and absence of thiamine. A wild type culture grown to log phase in the presence of thiamine, which represses the nmt1 promoter function, was split in half. We continued incubating one half at 30°C in the same medium as before. The other half was transferred to medium devoid of thiamine which induces nmt1 promoter function. With the onset of induction psp1 at 12 h after transfer to thiamine-less medium, growth of the cells in thiamine-less medium became slower than those in thiamine-containing medium (Fig. 5a). When cells grown for 12 h in the above conditions were passaged into fresh medium at a lower cell density, the growth difference between the two cultures was greater in early log phase than that in late log phase (Fig. 5b). Cells overexpressing Psp1 protein in early log phase, as evidenced by Western analysis of the total protein with Psp1 antibody (Fig. 5b, inset, A  and a), showed half the growth rate of the uninduced cells (Fig.  5b, A and a). However differences in growth rate, caused by psp1 overexpression, decreased as cells entered stationary phase (Fig. 5b, B and b). The protein produced in the cells from pnmt1-psp1 in log phase was mainly the low molecular weight form whereas that in stationary phase was the phosphorylated high molecular weight form (Fig. 5b, inset, a and b) as for that produced from the chromosomal copy of psp1 ϩ (Fig. 5b, inset, A  and B). The cells with retarded growth following overexpression of Psp1 also exhibited an altered cellular morphology (Fig.  5c). These cells rounded up and DAPI staining of nuclear DNA revealed that the area occupied by nuclear DNA enlarged, that is, the compactness of the nuclear DNA was lost (Fig. 5c, 2,  panels 3 and 4). Flow cytometric observation of DNA content in the Psp1 overexpressed cells indicated doubled or tripled DNA contents (Fig. 5c, 3, panels 2-4). The changes in DNA content and cellular morphology were maximum at 13.5 h after transfer to fresh medium (4n or 8n, Fig. 5c, 3, panel 4). Growth rate at this hour was about one-fourth of that in uninduced cells (Fig. 5b, 4). However, by 24 h after transfer to fresh media, the growth difference became relatively small, and DNA content of the induced cells tended to return to amounts found in uninduced cells (2n or 4n). After 30 h, at which time cells entered the stationary phase (Fig. 5b, 5), DNA content and morphology of the cells overexpressing Psp1 were similar to those of the cells containing only one copy of the psp1 gene (Fig. 5c, 3, panel  5). Thus maximal growth inhibition and changes in DNA content on overproducing Psp1 was observed in the actively growing cells rather than in the slow growing cells. However, the doubling time of the cells in which Psp1 was overexpressed was prolonged.

DISCUSSION
Temperature-sensitive mutants of S. pombe defective in normal cell cycle progression provide a convenient system to identify the elements involved in cell cycle regulation. Random mutagenesis of S. pombe cells enabled one to isolate such a mutant. Using the same mutagenic approach as Nurse et al. (51), we isolated one mutant, cyj92, which possesses an elongated cellular morphology similar to G 1 /S mutants of Nurse et al. (51) at 36°C. Comparative DNA content analysis of the mutant in rich and nitrogen-starved media indicated that the mutation in cyj92 hindered progression of S. pombe cell from G 1 to S phase and perhaps also from G 2 to M phase. The terminal phenotype of the mutation is accumulation of abnormal DNA which is either incapable of completing DNA replication (Fig.  1b, 3) or overreplicated in the absence of separation into daughter cells (Fig. 1b, 4). It is likely that the mutation causes both blocking of S phase which leads to the prevention of onset of the subsequent mitosis and a defect in mitosis which prevents reinitiation of the subsequent S phase and leads to an increase in ploidy.
The characteristics of the mutant phenotype enabled us to isolate a novel gene, psp1 ϩ , which suppresses the elongated mutant phenotype of cyj92 at 36°C and made cell cycle progression normal. The finding that psp1 gene function is required for normal cell growth at 30°C suggests that this gene is essential for vegetative growth. However, the fact that its overexpression in the actively growing cells (log phase cells) is detrimental to cell growth indicates that excessive functioning of this gene also causes a defect in normal cell growth. From the flow cytometric DNA analysis data we speculate that this is due either to accumulation of abnormal DNA or to inhibition of production of the elements necessary for normal cell division. Since we were not able to observe the accumulation of undivided cells, that is the cells attached together without septum, we consider that overproduction of Psp1 protein does not affect the cytokinesis process itself but rather affects the normal process of DNA replication. The accumulation of the 2n or 4n DNA content in the actively growing cells by overproduction of Psp1 protein suggests that the signal for completion of DNA replication may not be transferred correctly to the elements working in G 2 or M. Alternatively, it may be possible that overexpression of psp1 may lead to bypass or delay of the mitosis process and thereby allow continued replication of DNA. The fact that Psp1 protein is required for normal DNA replication but excessive function of this protein causes delay of the cell cycle suggests that this Psp1 may act at two different points in the cell cycle. Identification of target molecules with which Psp1 protein interacts will be necessary to elucidate the detailed function of this protein. The finding that the inhibitory effect on cell growth becomes less severe upon entry into the stationary phase indicated that this protein is in its active form in actively growing cells but is inactivated when cell growth slows down. However, we do not know whether inactivation of this protein is a prerequisite for entry into the stationary phase or conversely entry into stationary phase leads to inactivation of this protein.
The results that Psp1 is a phosphoprotein and that its dephosphorylated form is the major one in actively growing cells indicated that the function of this protein depends on phosphorylation. If this protein functions as a positive element, required in the cell cycle progression, the dephosphorylated form is active in actively growing cells and the phosphorylated Psp1, which is the major form in stationary phase cells, is inactive. However, if Psp1 functions as a negative element, the dephosphorylated form is inactive in actively growing cells, and the phosphorylated form is its active form in the stationary phase. Although we don't know where in the cell cycle Psp1 acts, the result that this protein is phosphorylated by the Cdc2-Cdc13 complex in vitro suggests that this protein could be one of the substrates for the Cdc2-Cdc13 complex and hence Cdc2 kinase may be associated with its function.
The carboxyl-terminal domain of this novel protein is phosphorylated at serine 333 by Cdc2 kinase, and alteration of this serine to alanine abolishes phosphorylation by Cdc2 kinase in vitro. However, in vivo we were not able to detect a defect in phosphorylation in the cdc2 mutant at the restrictive temperature. Because it showed the mutant phenotype within 4 -5 h after being shifted to restrictive temperature, this is not long enough for cdc2 Ϫ cells to enter the stationary phase to show its kinase function. It is only in the stationary phase of the cell growth that Psp1 protein is phosphorylated and cdc2 Ϫ cells die after 4 -5 h at the restrictive temperature long before we were able to detect the defect in Cdc2 kinase function in the mutant. Thus it was not possible to determine whether mutation in Cdc2 kinase affected the function of Psp1 in actively growing cells. We consider that phosphorylation of Psp1 protein upon entering the stationary phase is one of the major events for slowing cell growth, and Cdc2 kinase controls this process. The yeast two-hybrid method confirmed that Cdc2 and Psp1 interact directly in vivo. Thus we can suggest that when cell cycle progresses actively in log phase, Psp1 protein is in its dephosphorylated state, and Cdc2 function is not required. However, when cell cycle progression slows at the stationary phase, Cdc2 kinase phosphorylates Psp1 protein, and the phosphorylated form of Psp1 facilitates entry into the G 0 phase of the cell cycle or inactivates elements required for active cell growth. Cdc2 protein function is required for modification of Psp1 protein at this stage and there must be phosphatase(s), which removes the phosphate from Psp1 protein when cells begins active growth. We did not find yet what gene product is involved in dephosphorylation of Psp1 upon entry into active cell growth cycle. It will be interesting to find a specific phosphatase for Psp1, and function of this phosphatase may be a key regulatory element for activation of Psp1 protein function. Localization of this Psp1 toward each tip of the cell upon entry into the stationary phase may also be related to phosphorylation of this protein. It could be suggested that the phosphorylated Psp1 may have the property to form protein complexes in the cells leading to translocation of the protein in a certain compartment of cells and the dephosphorylated Psp1 may exist in monomeric form and is localized throughout cells as an active monomeric form. The following model for Psp1 protein function can be drawn (Fig. 6). Psp1 protein is activated by dephosphorylation with a phosphatase when the cells enter G 1 to S phase of the cell cycle and is inactivated by phosphorylation with Cdc2 kinase upon entry into the stationary or G 0 phase. It is an essential protein for cell growth, and regulation of this protein function by phosphorylation could be the key feature in cell cycle progression from G 1 to S. In addition to Cdc2 protein, the other element(s) which interact directly with this Psp1 protein should exist, and finding this element(s) will be essential for further characterization of Psp1 protein function.