The fission yeast STS5⁺ gene is required for maintenance of growth polarity and functionally interacts with protein kinase C and an osmosensing MAP-kinase pathway

To understand how individual cell shape is regulated, we have initiated a genetic analysis in the simple unicellular organism, the genetically amenable fission yeast, Schizosaccharomyces pombe. Fission yeast cells are rod-shaped and cell growth occurs only at the cell tips. This offers great advantages for the use of this organism as a system to study cell morphogenesis and the control of polarity (reviewed by Nurse, 1994). The diameter of the fission yeast cell (3 µm) remains constant throughout the cell cycle and a newly born cell grows from an initial length of around 8 µm until it reaches a critical mass (at 14 µm), whereupon mitosis is triggered, followed by a medial septation and cytokinesis (Mitchison, 1970).

Like most other eukaryotes growth polarity in fission yeast is subject to cell cycle regulation. During the first third of the cell cycle, growth occurs only at the ‘old’ end which was already in existence in the previous cycle. Cortical actin localises exclusively to this growing old end. A drastic alteration of growth polarity occurs at ∼0.3 of the cell cycle when the cell completes DNA replication and reaches a certain cell size. Actin is then seen at both ends and growth at the new end is initiated. This phenomenon is called NETO (new end take off; Mitchison and Nurse, 1985). From this point on, the cell exhibits bidirectional growth. When the cell becomes committed to executing M-phase, cell growth ceases and this coincides with the transient disappearance of cortical actin from both ends. Actin now rapidly reassembles at the medial region of the cell body where the septum is to be formed. Following nuclear division and cytokinesis, actin moves from the middle of the cell, which is now the new end, to the old end where growth will be initiated (Marks and Hyams, 1985; Marks et al., 1986).

The sts5-7 mutant was originally isolated as one of eleven loci which are supersensitive to the protein kinase inhibitor staurosporine (Tamaoki et al., 1986; Toda et al., 1991) and which showed a round cell morphology even in the absence of the drug (Toda et al., 1993, and this study). Deletion of the ppe1⁺ gene, which encodes a type 2A-like protein phosphatase, shows a similar phenotype, namely, round cell morphology and staurosporine-supersensitivity (Shimanuki et al., 1993). The sts5⁺ and ppe1⁺ genes seem to coordinately act in the same pathway that determines cell shape as, in addition to the phenotypic similarities, a temperature sensitive mutation ssp1, which encodes a novel protein kinase, suppresses both sts5-7 and ∆ ppe1 mutant phenotypes (Matsusaka et al., 1995). The ssp1 mutant shows reduced mating efficiency and the mutant cell elongates like an early cdc mutant in which only monopolar growth continues without NETO. From these results, we have proposed that growth polarity and cortical actin localisation are regulated by two counteracting mechanisms; one is via the Ssp1 kinase which stimulates the relocalisation of cortical actin, the other is through the Ppe1 phosphatase and Sts5 which act to fix actin localisation and growth polarity (Matsusaka et al., 1995). However, physiological as well as genetic characterisation of the sts5 mutant and molecular analysis of the sts5⁺ gene have not been performed.

To understand the morphological defects in sts5 and their relation to cell cycle progression, we have characterised the mutant phenotype in detail. This analysis has led us to propose that the Sts5 protein is required for polarised growth during interphase. We have isolated the sts5⁺ gene and found that it encodes an evolutionarily conserved protein. In particular the predicted Sts5 protein has a significant similarity to two yeast proteins which are involved in cell morphogenesis and cell cycle control. Genetic analysis has indicated that the Sts5 protein functionally interacts with various protein kinases and phosphatases which are involved in cell morphogenesis and polarised growth.

The S. pombe strains used in this study are listed in Table 1. Complete medium, YPD (1% yeast extract, 2% polypeptone, 2% dextrose) and modified synthetic EMM2 (Gutz et al., 1974; Moreno et al., 1991) have been described previously. Plates contained 1.6% agar. Staurosporine (provided by Dr H. Nakano, Kyowa Hakko Co.) was used as described previously (Toda et al., 1991).

Standard procedures for S. pombe genetics were followed according to Gutz et al. (1974) and Moreno et al. (1991). The cell concentration and mass were measured with a Sysmex F-800 (TOA Medical Electronics, Japan). S. pombe cells were transformed using the lithium method (Ito et al., 1983). Gene disruptions are abbreviated as the gene preceded by ∆, such as ∆ sts5. Protein is denoted by the first letter with a capital, followed by two lower case letters, such as Sts5.

Several round mutants have been isolated through visual screening or by selecting for mutants which are resistant to an inhibitor of cell wall biosynthesis (Fukui and Yamamoto, 1988; Ribas et al., 1991; Snell and Nurse, 1994). Among a number of mutants tested, we found that sts5 is allelic to orb4 which was previously isolated by visual screening (Snell and Nurse, 1994). Consistent with this finding, orb4 mutant cells (obtained from V. Snell) exhibited staurosporine-super-sensitivity.

Tetrad analysis between the sts5 mutant and known genetic markers showed that sts5 is tightly linked to sds21 (PD:TT:NPD=18:0:0, <2.6 cM) and to wee1 (PD:TT:NPD=61:0:14, 9.3 cM), both of which map to the left arm of chromosome III (Mizukami et al., 1993). Fine mapping using three factor crosses by random spore analysis (Gutz et al., 1974) was performed to assign the precise order of the sts5 locus in relation to sds21 and cut1 which is also tightly linked to sds21 (Mizukami et al., 1993). The results showed that the order on chromosome III is sts5-cut1-sds21.

The procedures described by Shiozaki and Russell (1995a) were followed. β-glucanase (100 µg/ml, Zymolyase-20T, ICN) was used for digesting the cell wall.

Standard molecular biology techniques were followed as described (Sambrook et al., 1989). Enzymes were used as recommended by the suppliers (Takara Shuzo Co., TOYOBO Co. and New England Biolabs. Co.).

S. pombe genomic libraries constructed in a vector pDB248 (Beach et al., 1982) or in a cosmid sCOS-LEU2 (a gift from Dr Osami Niwa) were used for isolation of the sts5⁺ gene (TP40-5B, Table 1). Six different plasmids (designated pN2, pN15, pTC7-4, pTC10-2, pTC16-1 and pTC16-5) were recovered from transformants using a plasmid library which suppressed both staurosporine-supersensitivity and a cell shape defect. Subsequent physical mapping by hybridisation against an ordered cosmid library (Mizukami et al., 1993) showed that none of them contained the sts5⁺ gene as no clones hybridised to cosmids derived from chromosome III where the sts5⁺ gene resides (see above). Thus, these six plasmids are multicopy suppressors.

One transformant (TA101), which reverted to both the drug-sensitivity and the cell shape defect, was obtained using a cosmid library. The Leu⁺ phenotype in TA101 was mitotically stable, suggesting that the cosmid DNA was integrated into the genome by homologous recombination. Genetic crosses between TA101 and a marker (JY6 and cut1, Table 1) showed that the Leu⁺ phenotype segregated 2:2 and that the Leu⁺ marker was linked to the cut1 locus, suggesting that the insert DNA is integrated at the sts5⁺ locus and that the insert DNA of the cosmid contains the sts5⁺ gene. In order to recover an integrated cosmid DNA containing the sts5⁺ gene from TA101, the following method was undertaken. Genomic DNAs were isolated from TA101, completely digested with BamHI (which cleaved at only a single site in the vector, and was used to construct the library by ligating a genomic DNA which had been partially digested with Sau3AI into this site and therefore has a low probability of being regenerated in the clones), religated and transformed into Escherichia coli to obtain ampicillin-resistant colonies. By these procedures, it was expected that not the whole insert, but two extreme ends of the insert DNAs terminating in a BamHI site, would be isolated as a religated circular plasmid. As expected, a plasmid, designated pTA9, was recovered from E. coli and restriction mapping of pTA9 showed that it contained two fragments, of 1.4 kb and 0.3 kb, which were derived from the two ends of the integrated cosmid. Hybridisation using the above two fragments as a probe against an ordered cosmid library (Mizukami et al., 1993) showed that both of them were derived from a contig containing cut1⁺ and sds21⁺ both of whose genes are tightly linked to sts5⁺. The identity between the cloned fragment and the sts5⁺ gene was further confirmed by diploids constructed by crossing sts5-7 with ∆ sts5 (see below), which are still round and staurosporine-supersensitive.

The dideoxy method (Sanger et al., 1977) was performed using double-stranded plasmid DNA as template (Hattori and Sakaki, 1986). A 6.9 kb SacI/SalI fragment from cosmid 1817 was subcloned into pBluescript (KS+, Stratagene) to give plasmid pYN201. The nucleotide sequence of this fragment was determined using various restriction enzyme sites as well as synthetic oligonucleotides as primers to connect individual sequences. These sequence data are available from EMBL/GenBank/DDBJ under accession number D58421.

Two internal BamHI fragments (1.6 kb and 0.3 kb) were deleted from pYN201 and replaced with the 1.8 kb ura4⁺ fragment (Grimm et al., 1988), yielding pYN701. A 4.6 kb ClaI/SacI fragment containing this deleted version of sts5⁺ (∆ sts5; more than 50% of the coding region in the amino-terminal region is missing in this construct) was used to transform a diploid 5A/1D (Table 1). Stable Ura⁺ transformants were obtained and Southern hybridisation confirmed that the sts5⁺ gene had been deleted as expected.

The 6.9 kb SacI/SalI fragment that contains the entire coding region of the sts5⁺ gene was subcloned into pSK248, yielding pYN010. To ectopically express the sts5⁺ gene, the entire ORF of the sts5⁺ gene was amplified by PCR using two primer oligonucleotides. These are 5’AACGGTCGACGATGGATGAGGAGTGTGAA3’ for the 5’ primer and 5’AACGCCCGGGTCAAAACTCGACAATTGA3’ for the 3’ primer (SalI and SmaI sites are underlined). The 3.2 kb amplified fragment was subcloned into the SalI/SmaI sites of pREP1, which is a fission yeast expression vector under the control of the thiamine-repressible nmt1 promoter (Maundrell, 1990). The plasmid was designated pREP-sts5⁺. pREP-sts5⁺ could complement the sts5-7 mutant in the thiamine-containing repressing conditions, indicating that the PCR fragment of the sts5⁺ gene in pREP-sts5⁺ is functional.

Green fluorescent protein (GFP) from the jellyfish Aequorea victoria (Prasher et al., 1992) was inserted into an amino-terminal region of the sts5⁺ gene as follows. A DNA fragment encoding the GFP protein was amplified with PCR by using two oligonucleotide primers: GFP-N1, 5’AAAGTCGACGATGAGTAAAGGAGAAGAA3’ and GFP-C1, 5’AAAGTCGACGATTTGTATAGTTCATCCAT3’ where the SalI site is underlined. A 720 bp SalI fragment was inserted into the same site of pREP-sts5⁺, designated pREP-GFP-sts5⁺. Insertion of the GFP protein in the amino terminus does not interfere with the function of the Sts5 protein as pREP-GFP-sts5⁺ behaved in the same manner as pREP-sts5⁺. Transformants carrying pREP-GFP-sts5⁺ were grown in minimal medium containing 25 nM of thiamine and green fluorescence (GFP-Sts5) was observed under immunofluorescence microscopy without fixation.

Oligopeptides (GKLEKEENRRRKDPIS) corresponding to a region from the 361st to 376th amino acid residues were synthesised, coupled to Hemocyanin (Calbiochem) and used to raise polyclonal antibodies in rabbits. Antisera were affinity-purified using peptide-coupled gel as recommend by a supplier (Sulfolink, Pierce). Fission yeast cell extracts were prepared according to the method of Matsusaka et al. (1995). Protein samples were run on an SDS-polyacrylamide gel (Laemmli, 1970) and electrically transferred onto nitrocellulose filters (Towbin et al., 1979). Horseradish peroxidase-conjugated Protein A (Bio-Rad) and a chemiluminescence system (ECL, Amersham) were used to detect bound antibody.

For actin staining, TRITC-conjugated phalloidin (Molecular probes) or anti-actin monoclonal antibody (N350, Amersham) and a Cy3-conjugated sheep anti-mouse IgG secondary antibody (Sigma) were used to visualise actin as described by Alfa et al. (1993) and Matsusaka et al. (1995). A monoclonal anti-tubulin antibody (TAT1, a gift from Dr Keith Gull) was used for tubulin staining.

A comparison of the morphology of sts5-7 mutant and wildtype cells stained with Calcofluor is shown in Fig. 1. A typical cylindrical shape was evident in wild type with stripes consisting of dark division scars and bright growing tips (Fig. 1A, right; note that Calcofluor most brightly stains medial septa; Streiblová and Wolf, 1972). On the other hand, in the sts5-7 mutant grown at 32°C more than 90% of the cells were round or pear-shaped; the diameter being larger and cell length shorter (Fig. 1B). Nuclear chromatin regions stained with DAPI showed that chromosomal DNA structure was normal; the typical interphase hemispherical morphology was observed (Fig. 1A and B, left).

The effect of the temperature on the cell morphology of the sts5 mutant was examined. At 25°C cell shape was closer to that of wild type although ellipsoidal cells were often observed (Fig. 1C). In contrast, at 36°C the cell shape became almost completely round (Fig. 1D). The sts5-7 mutant is not temperature sensitive for growth but the generation time at the higher temperature is 50% longer than wild type. Flow cytometry analysis showed that no specific retardation of the cell cycle was observed except that the percentage of septated cells increased to more than 30% (data not shown).

To examine localisation and structures of actin in the sts5 mutant, exponentially growing wild-type or sts5 cells were stained with phalloidin. In wild-type cells cortical actin localised in either growing tips or in the septum as reported previously (Fig. 2A; Marks and Hyams, 1985; Marks et al., 1986). In contrast, in the sts5 mutant, cortical actin was dispersed within the cell (Fig. 2B). Next actin localisation in the sts5 mutant during cell cycle progression was examined. As shown in Fig. 3, it was clear that, in interphase cells; namely from the beginning of the cell cycle to just before mitosis (Fig. 3A-D), cortical actin did not localise to the cell tips but appeared dispersed within the round cell. In contrast, during the mitotic and post-mitotic stages (Fig. 3E-H), cortical actin localised in a manner analogous to that in wild type, i.e. it disappeared upon the onset of mitosis (Fig. 3E) and reassembled to a medial region after mitosis (Fig. 3F-H). These results suggested that the sts5⁺ gene is required for localisation of cortical actin to the growing tips before mitosis, but not to a medial region during and after mitosis.

Distribution of microtubules were also examined. It was found that cytoplasmic microtubules also appeared abnormal; they showed criss-crossed structures as reported in other round mutants (data not shown; Snell and Nurse, 1994).

Why do sts5 mutant cells lose polarity and become round? One defect could be in cell wall structure. It has been shown that cell shape of mutants defective in the protein kinase C-encoding gene sts6/pck2 (Toda et al., 1993) is altered and its cell wall composition is impaired (Shiozaki and Russell, 1995a). It was also reported that mutants defective in (1-3)β-D-glucan synthesis exhibit a round morphology (Ribas et al., 1991). To test whether or not cell wall composition is defective in the sts5-7 mutant, the kinetics of cell lysis upon β-glucanase treatment were examined. As shown in Fig. 4A, no difference between the sensitivity of wild type and the sts5-7 mutant was evident. Sensitivity to detergents, such as SDS, also did not noticeably differ between wild-type and sts5 mutant cells.

Defective phenotypes in some morphological mutants have been reported to be suppressed by the addition of osmotic stabilisers (Ribas et al., 1991; Shiozaki and Russell, 1995a). To examine the effect of a high osmolarity on the sts5 mutant, cell morphology was observed after growing cells in medium containing 1.2 M sorbitol. It was found that the mutant cells were still round (Fig. 4B). These results have strongly suggested that the morphological defect in the sts5-7 mutant is not due to a defect in cell wall biosynthesis or composition.

To address the relationship between loss of polarity and cell cycle progression in the sts5 mutant, double mutants were constructed between the sts5-7 and temperature-sensitive cdc mutants. As discussed by Snell and Nurse (1994), this type of analysis allows us to determine precisely a cell cycle point when the mutant cells lose polarity. The cdc mutants employed were cdc10 which blocks cell cycle progression in G₁, cdc25 which blocks at G₂ and cdc7 which bypasses septation without affecting other cell cycle events such as DNA replication and nuclear division (Nurse et al., 1976).

As shown in Fig. 5, all the double mutant cells of either sts5cdc10, sts5cdc25, or sts5cdc7 showed abnormal cell morphology, cells being either pear-like, dumbbell-shaped or round (Fig. 5B,D,F). In contrast, single cdc mutants were typically elongated (Fig. 5A,C,E). The number of nuclei was the same in single and double mutants, a single nucleus in sts5cdc10 and sts5cdc25, and multiple nuclei in sts5cdc7. These results show that the sts5-7 mutation does not affect the nuclear division cycle. Actin staining showed that in double mutants actin was dispersed as in a single sts5-7 mutant (data not shown). Thus the sts5 mutation causes a defect in the maintenance of polarity in both G₁ and G₂ phases of the cell cycle where normal cells elongate in mono- and bipolar manners, respectively. Therefore, the sts5⁺ gene is needed to maintain polarized growth.

As pck1 and pck2/sts6 mutants, which encode protein kinase C-like proteins (Toda et al., 1993), are also supersensitive to staurosporine, a multicopy plasmid containing the pck1⁺ gene was introduced into the sts5 mutant. This resulted in a reversal of the cell morphology of the sts5-7 mutant from a round to a rod shape (Fig. 6A). Staurosporine-supersensitivity of sts5-7 was also suppressed by the introduction of pck1⁺ or pck2⁺ (data not shown).

We have isolated six different multicopy suppressor plasmids of the sts5-7 mutant (see Materials and Methods). Based on restriction maps, none of them appeared to be pck1⁺ or pck2⁺. However, we found that one of them (pN15) contained the pyp1⁺ gene which encodes a protein tyrosine phosphatase (Ottilie et al., 1991). Fig. 6B shows the morphological suppression of the sts5-7 mutant phenotype by pN15 (upper panel in Fig. 6B). As reported previously, pN15 resulted in cell elongation when introduced into wild type (lower panel in Fig. 6B, Millar et al., 1992; Ottilie et al., 1992).

Recently it has been shown that pyp1⁺ acts, at least in part, via the dephosphorylation of a tyrosine residue of a novel MAP-kinase Sty1/Spc1, which is involved in both osmoregulation and mitotic entry (Millar et al., 1995; Shiozaki and Russell, 1995b). To address the question of how high copy pyp1⁺ gene suppresses the sts5 mutation, double mutants between sts5 and wis1, which encodes an upstream MAP-kinase kinase of Sty1/Spc1 (Warbrick and Fantes, 1991; Millar et al., 1995; Shiozaki and Russell, 1995b) was constructed. Consistent with the notion that Pyp1 acts through negative regulation of the Sty1/Spc1 pathway, the wis1 mutation also rescued the sts5 mutation: the sts5wis1 double mutant became drug-resistant (Fig. 7A). Different from a single wis1 mutant (Warbrick and Fantes, 1991), however, cells of the double mutant do not elongate (Fig. 7B), suggesting that the sts5 mutation also suppresses a G₂ delay phenotype of the wis1 mutation. These results demonstrate that Sts5 and the Sty1/Spc1 pathway regulate cell shape in an opposing manner.

Although a multicopy plasmid containing ppe1⁺ did not suppress the sts5 mutant and vice versa, the sts5 mutant showed synthetic lethality with the ∆ ppe1 mutant. This result is consistent with our previous results which showed that the ssp1 and ssp2 mutations, which are defective in the genes encoding novel protein kinases (Matsusaka et al., 1995), are capable of suppressing both sts5 and ppe1 mutants. The sts5 mutant is supersensitive to an immunosuppressive agent, cyclosporin A and showed synthetic lethality with ppb1 which encodes a catalytic subunit of calcineurin, a type 2B protein phosphatase (Yoshida et a., 1994). A summary of these genetic interactions between sts5 and other protein kinases and phosphatases is shown in Table 2.

We cloned the sts5⁺ gene by complementation of the sts5 mutation from a cosmid library and sequenced the gene (see Materials and Methods). The minimal complementing fragment contained a single uninterrupted open reading frame consisting of 3,198 bp which predicts a protein of 1,066 amino acid residues. Comparison with amino acid sequences in the data bases (EMBL and GenBank) showed that the predicted Sts5 protein shares significant homology with two known proteins, fission yeast Dis3 (Kinoshita et al., 1991) and budding yeast Ssd1/Srk1 (Sutton et al., 1991; Wilson et al., 1991) as well as Caenorhabditis elegans F46E8.6 (accession number: U23514). As shown in Fig. 8A and B, two regions exhibit the highest homology. One region ‘a’, consisting of 170 amino acid residues, shows a 47% identity only between Sts5 and Ssd1/Srk1. No homologous region was found in Dis3. In contrast, another region, ‘b’, consisting of 680 amino acid residues, shows 21% identity between Sts5 and Dis3, and 33% between Sts5 and Ssd1.

To test whether any of these three genes are functionally exchangeable, multicopy plasmids containing each gene were introduced into each mutant. No suppression was observed; neither sts5⁺ nor SSD1 could suppress the dis3 mutant and neither dis3⁺ nor SSD1 could suppress the sts5 mutant. Suppression of the ssd1 mutant by sts5⁺ or dis3⁺ has not been tested. Thus, these three genes are probably not functional homologues.

In order to examine the phenotypes arising upon complete disruption of the sts5⁺ gene, one step gene replacement (Rothstein, 1983) of sts5⁺ was carried out (Materials and Methods). We found that cells containing a disruption of the sts5⁺ gene were viable and showed almost identical phenotypes to those of the sts5-7 mutant, including round morphology, temperature-dependence of the morphology, β-glucanase resistance (Fig. 4) and staurosporine-supersensitivity. In addition, multicopy plasmids containing pck1⁺, pck2⁺, or pyp1⁺ were capable of suppressing drug-sensitivity of the deleted sts5 mutant. These analyses show that the sts5⁺ gene is not essential for growth in fission yeast but is required for maintenance of cell morphology.

To examine the phenotypic consequences of ectopic overexpression of the sts5⁺ gene, the entire ORF was placed under the control of the thiamine-repressible nmt promoter (designated pREP-sts5⁺). Wild-type cells were transformed with pREP-sts5⁺ and Leu⁺ transformants were selected on minimal plates containing thiamine (repressed condition). We found that transformants grew extremely slowly when they were restreaked on the minimal plates without thiamine (derepressed condition) and barely formed colonies (Fig. 9A), showing that ectopic overexpression of the sts5⁺ gene is toxic.

The morphology of sts5⁺-overproducing cells was observed in liquid minimal medium in the absence of thiamine. As shown in Fig. 9B, many cells had an ellipsoidal appearance; asymmetrically fat rather than rod-shaped. Actin staining of these cells showed that cortical actin did not localise properly to the cell tips, instead it was often dispersed within the cells. No specific cell cycle arrest was evident in these cells. We noticed that the fluorescence intensity of cortical actin in the overproducers was reproducibly weaker than in either wildtype or sts5-7 mutant cells. The level of the actin protein as judged by immunoblotting remained unchanged upon overproduction of the Sts5 protein (data not shown). Thus, ectopic expression of the sts5⁺ gene is toxic, leads to a disturbance of cell shape and actin localisation as well as possible alterations of the physical state of cortical actin. The level of the Sts5 protein upon induction was examined using anti-Sts5 polyclonal antibodies (see Materials and Methods). A specific band of around 130 kDa began to accumulate 12 hours after removal of thiamine and its level peaked at 16 hours and stayed at a high level after 22 hours (Fig. 9C). The kinetics of this increase in protein level is in complete parallel with the morphological alterations seen upon induction. These observations have confirmed that overexpression of the Sts5 protein leads to growth inhibition and alterations of cell morphology.

An attempt to determine a cellular localisation of the Sts5 protein with anti-Sts5 antibody has not been successful unless it is highly expressed, which causes morphological alteration of the cell as described before. To address this question, green fluorescent protein from jellyfish (Prasher et al., 1992; Chalfie et al., 1993) was fused in front of the Sts5 coding region which was driven under the thiamine-repressible promoter described before. This plasmid (designated pREP-GFP-sts5⁺) is toxic in the absence of thiamine and capable of suppressing the sts5 mutant in the presence of thiamine as in the case of pREP-sts5⁺ (data not shown). This result indicates that the insertion of GFP does not interfere with the function of the Sts5 protein. Using pREP-GFP-sts5⁺, we were able to detect green fluorescence in cells which were grown not only under toxic derepressing condition but also under partially repressed condition (in a low concentration of thiamine; see Materials and Methods) where cell growth was not affected and a normal cell morphology was maintained (Fig. 10A). Consistent with this finding, immunoblotting with anti-Sts5 antibody showed that in the low thiamine condition, the amount of the Sts5 protein was only mildly overexpressed (Fig. 10C). Under this growth condition, the Sts5 protein was found to localise in the cytoplasm as discrete punctate dots (Fig. 10B). These dots appeared not to correlate to those of cortical actin since it did not localise either in the cell tips or septa. It should be stressed that this dotted pattern could be observed without any preparative processes such as fixation or incubation with antibodies. It is therefore possible that these dotted bodies represent novel cytoplasmic structures which are involved in determination of growth polarity.

In this study we have demonstrated that the sts5⁺ gene is required for polarised growth when the cell elongates in defined directions. Cortical actin in the sts5 mutant did not localise to specific sites in the cell, instead it was randomly dispersed. The round phenotypes of the sts5 mutant could be due to an indirect effect, e.g. a defect in cell wall synthesis. However, the sts5 mutant appears not to be defective in cell wall integrity. Ectopic overproduction of the sts5⁺ gene also resulted in altered polarity, suggesting a direct role of the Sts5 protein in cell shape control. This notion is further supported by the fact that ssp1, an extragenic suppressor of the sts5 mutant, also showed polarity defect (Matsusaka et al., 1995). The sts5⁺ gene is not required for septum formation and cytokinesis when cortical actin moves from the cell tips to a medial region. It is important to note that during mitosis and in the post-mitotic period when septum formation and cytokinesis take place, cell length is constant and no polarised growth occurs (Mitchison, 1970). The sts5⁺ gene is, therefore, specifically required for polarised growth during interphase.

Either loss or overproduction of the Sts5 protein leads to similar, although not identical, morphological defects. Overproduction of the Sts5 protein might absorb interacting molecules that also regulate polarised growth, which would lead to a similar morphological phenotype to that caused by a loss of the sts5⁺ gene function. Analysis of the localisation of Sts5 protein by using GFP suggested that the protein exists in punctate structures in the cytoplasm. We believe that this punctate localisation is not a mere artifact as we could observe these structures in living cells. In addition, a similar punctate localisation of the Sts5 protein was also evident with anti-Sts5 antibody when the protein was overexpressed (unpublished results). It is interesting to speculate that these structures may represent novel vesicles or related intracellular bodies which are responsible for transport of proteins or other materials required for polarised growth.

In addition to sts5⁺, four other genes are required for maintaining polarised growth in the fission yeast cell cycle. These include orb1, 6,12, and cwg2 (Díaz et al., 1993; Ribas et al., 1991; Verde et al., 1995) where only the cwg2⁺ gene has been identified. The cwg2⁺ gene encodes the β subunit of a type I geranylgeranyltransferase (Díaz et al., 1993) and cwg2 mutant cells are supersensitive to staurosporine (D. Hirata and T. Toda, unpublished results). One intriguing possibility is that the Sts5 protein may be involved in a membrane-mediated signal transduction such as a GTPase pathway. At present, however, the exact molecular functions of the Sts5 protein in regulating polarised growth remain to be understood.

We have shown that sts5 mutant cells are as resistant as wildtype cells to treatment with a cell wall digesting enzyme. In contrast, pck2/sts6 mutant cells, defective in one of the two protein kinase C-like proteins, are hypersensitive to this enzyme treatment, leading to cell lysis (Shiozaki and Russell, 1995a). Although pck2/sts6 and sts5 mutants are defective in cell shape and supersensitive to staurosporine, a morphological defect could arise from two different mechanisms; one in which cell wall composition is defective and the other when polarised growth itself is defective. However, these two pathways are not functionally independent as we have shown that a multicopy plasmid containing the pck1⁺ gene suppresses the sts5 mutant as well as it does ∆ ppe1. It is, therefore, possible that the Sts5 protein regulates polarised growth in concert with Pck1, 2-dependent cell wall architecture. It is known that cell wall synthesis and polarised growth are closely coupled as there are differences in cell wall structure between the growing tips and non-growing parts of the cell body (Kobori et al., 1989, 1994).

Our genetic analyses suggest that the Sts5 protein and the Sty1/Spc1-dependent MAP-kinase pathway coordinately regulate cell shape in an opposing manner. We have shown that overexpression of the pyp1⁺ gene was capable of suppressing the sts5 mutation and that the wis1 mutation which is defective in an upstream MAP-kinase kinase also suppressed the sts5 mutation. It should be pointed out that the Sty1/Spc1 pathway is involved in cell shape control as the combination of ∆ pyp1∆ pyp2 in a double mutant resulted in spherical cell shape and a loss of viability (Millar et al., 1995). These results suggest either that the Sts5 protein negatively regulates the Sty1/Spc1-dependent MAP-kinase pathway or that it regulates cell shape in a parallel but antagonistic manner to the Sty1/Spc1 MAP-kinase pathway. Our results, in fission yeast, indicate that multiple pathways, involving Sts5 and the Ppe1 phosphatase, protein kinase C-like proteins and an osmosensing MAP-kinase, regulate polarised growth and maintenance of cell morphology.

The predicted amino acid sequence of the Sts5 protein contains regions similar to those in the fission yeast Dis3 and budding yeast Ssd1/Srk1 proteins. Interestingly a multicopy plasmid containing the dis3⁺ gene suppresses the morphological phenotype of ∆ ppe1 which is defective in a type 2A-like protein phosphatase (Shimanuki et al., 1993). Budding yeast SSD1/SRK1 is involved in a complex functional network with a number of genes including SIT4 which encodes a structural and functional homologue of Ppe1 (Sutton et al., 1991; Shimanuki et al., 1993), IPL2/BEM2 which encodes a GTPaseactivating protein (Kim et al., 1994), cAMP-dependent protein kinase (Wilson et al., 1991) and SPA2 whose product localises at sites involved in polarised growth (Costigan et al., 1992) and G₁ cyclins (Cvrcková and Nasmyth, 1993). The Sts5-related gene family seems to be evolutionarily conserved as a homologous protein also exists in higher organisms such as nematodes.

As shown in this study Sts5 closely functions with the Ppe1 phosphatase, as is true for budding yeast Ssd1/Srk1 and the Sit4 phosphatase (Sutton et al., 1991). The Sts5 protein, however, seems not to form physical complexes with Ppe1 in the cell (unpublished results). At present how Sts5 and Ppe1 functionally interact remains to be determined. It should be noted that defective phenotypes of ∆ ppe1 are more pleiotropic than those of sts5. These include cold sensitive growth, sterility and suppression of the pim1/dcd1 mutant which encodes a RCC1-like protein (Matsumoto and Beach, 1993). In contrast, sts5⁺ gene function seems to be limited to the maintenance of cell morphology as the mutant is fertile and does not suppress the pim1/dcd1 mutant (unpublished results). Further work will be required to identify molecules which physically interact with the Sts5 protein within the cell.

We thank Paul Nurse, Valery Snell, Fulvia Verde and Jonathan Millar for gifts of strains, Mizuki Shimanuki for help during the course of cloning of the sts5⁺ gene, Noriyuki Kinoshita for the stimulating discussion, Osami Niwa for providing an S. pombe genomic library constructed in cosmid sCOS-LEU2, Hirofumi Nakano (Kyowa Hakko Co. Japan) for staurosporine and Keith Gull for TAT1 monoclonal anti-tubulin antibody. We are grateful to Paul Nurse for constructive discussions, and Jonathan Millar and Paul Russell for communication of results prior to publication. We thank Iain Hagan for useful comments and critical reading of the manuscript. This work was in part supported by grants from the Ministry of Education, Science, and Culture of Japan and from Kyowa Hakko Co.