SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Calcium/calmodulin-dependent protein kinase (CaMK) is required for diverse cellular functions, and similar kinases exist in fungi. Although mammalian CaMK kinase (CaMKK) activates CaMK and also evolutionarily-conserved AMP-activated protein kinase (AMPK), CaMKK is yet to be established in yeast. We here report that the fission yeast Schizosaccharomyces pombe Ssp1 kinase, which controls G2/M transition and response to stress, is the putative CaMKK. Ssp1 has a CaM binding domain (CBD) and associates with 14-3-3 proteins as mammalian CaMKK does. Temperature-sensitive ssp1 mutants isolated are defective in the tolerance to limited glucose, and this tolerance requires the conserved stretch present between the kinase domain and CBD. Sds23, multi-copy suppressor for mutants defective in type 1 phosphatase and APC/cyclosome, also suppresses the ssp1 phenotype, and is required for the tolerance to limited glucose. We demonstrate that Sds23 binds to type 2A protein phosphatases (PP2A) and PP2A-related phosphatase Ppe1, and that Sds23 inhibits Ppe1 phosphatase activity. Ssp1 and Ppe1 thus seem to antagonize in utilizing limited glucose. We also show that Ppk9 and Ssp2 are the catalytic subunits of AMPK and AMPK-related kinases, respectively, which bind to common β-(Amk2) and γ-(Cbs2) subunits.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Calcium regulation is fundamental in many biological events ranging from cell cycle control to learning and memory formation in brain (Groigno & Whitaker 1998; Berridge et al. 2000). Ample evidence exists that the fission yeast Schizosaccharomyces pombe, an excellent model organism for eukaryotic cell regulations, has an extensive cellular system for calcium-mediated signaling. It has a single authentic calmodulin (CaM) gene cam1+ that is essential for viability (Takeda & Yamamoto 1987) and also for chromosome segregation (Moser et al. 1997). In the genome of S. pombe, 25 genes code for proteins having the calcium-binding EF motif (GeneDB Schizosaccharomyces pombe; http://www.genedb.org/genedb/pombe/index.jsp). They include myosin light chains and a phosphatase calcineurin regulatory subunit. Protein kinases, Pck1 and Pck2, which appear to be the counterparts of mammalian PKC, exist and are essential for viability, controlling lipid homeostasis, cell shape, cell wall morphogenesis and cell death (Toda et al. 1993; Kobori et al. 1994; Low et al. 2008). The budding yeast Saccharomyces cerevisiae has one essential calmodulin gene CMD1, and the detailed investigation of CMD1 shows the surprising result that mutant CaM that fails to bind to Ca2+ can still perform the essential functions in mitosis (reviewed in Cyert 2001). Schizosaccharomyces pombe has one calcium-dependent, calmodulin-stimulated phosphatase calcineurin Ppb1 that is implicated in cytokinesis and chloride ion homeostasis (Yoshida et al. 1994; Sugiura et al. 1998).

One of the target of CaM is calcium/CaM-dependent protein kinase (CaMK), which also exists in S. pombe, and is designated Cmk1 (Rasmussen 2000). It is non-essential for viability. The budding yeast has two CaMKs, CMK1 and CMK2, and the double deletion mutant is viable (Ohya et al. 1991; Pausch et al. 1991). Saccharomyces cerevisiae CaM, CaMK and calcineurin are involved in endocytosis, trafficking and Ca2+-dependent, stress-activated signaling pathways (Cyert 2001). In mammalian systems, CaMK has been intensely studied as the essential regulator for cellular calcium signaling. CaMK is known to function in synaptic and behavioral memory, and neuronal plasticity (Fink & Meyer 2002; Lisman et al. 2002). CaMK seems to control memory formation (Fox 2003), and is known to be a main regulator for cardiac calcium and sodium currents (Pitt 2007). Calmodulin and CaMK thus act as molecular switches for cardiac ion channels through controlling calcium homeostasis.

Previous studies on budding yeast and fission yeast did not establish the presence of CaM kinase kinase (designated CaMKK hereafter) that is shown to exist in mammalian cells as the upstream activating kinase of CaMK (Soderling 1999). These kinases are particularly abundant in brain. CaMKK is not only the upstream of CaMK, but also phosphorylates and activates AMP-activated protein kinase (AMPK) (Hawley et al. 2005; Hurley et al. 2005; Woods et al. 2005). AMPK is considered to control energy balance at both cellular and whole body levels, and is speculated to be closely implicated in obesity and type II diabetes (Hardie 2008). In S. cerevisiae, protein kinase Snf1 is similar to AMPK, and plays various roles including the controls on different carbon sources and is required for the adaptation of yeast cells to glucose limitation (Hedbacker & Carlson 2008). Three Snf1-upstream kinases are SAK1, ELM1 and TOS3 (Hong et al. 2003; Sutherland et al. 2003). The genome of S. pombe contains two sequences, Ppk9 and Ssp2, which resemble AMPK/Snf1 catalytic subunits (Matsusaka et al. 1995; Bimbo et al. 2005), but their properties have been little investigated.

In this study, we first looked for CaMKK in S. pombe as other members in calcium signaling have been found. The reason for such search was that CaMKK seemed to be a missing link in the calcium signaling. If a CaMKK orthologue does exist, we like to gain information on the degree of conservation between S. pombe and mammalian CaMKKs and also on the signaling properties of S. pombe CaMKK. We provide evidence that Ssp1 kinase previously implicated in cell cycle control and stress response (Matsusaka et al. 1995; Rupes et al. 1999; Robertson & Hagan 2008) is the putative CaMKK in S. pombe. We show that Ssp1 is functionally related to Sds23, as the elevated gene dosages of Ssp1 and Sds23 mutually suppress their mutant phenotypes. Ssp1 and also Sds23 confer the ability of tolerance to limited glucose. Mass spectroscopic analysis led us to a surprising finding that Sds23 forms the stable complex with type 2A protein phosphatase (designated PP2A hereafter) and PP2A-related phosphatase Ppe1 that resembles S. cerevisiae SIT4 and human PP6 phosphatase (Shimanuki et al. 1993). We show evidence that Sds23 is a PP2A-related phosphatase inhibitor, and discuss the possible opposed roles of Ssp1 and these phosphatases for cells to survive under limited glucose.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Amino acid sequence of Ssp1 kinase resembles calcium/calmodulin-dependent kinase kinase (CaMKK)

To look for an orthologue of CaMKK in S. pombe, YOGY (eukarYOtic ortholoGY, the Sanger Center) was used, and two datasets (KOGs, OrthoMCL) suggested that the orthologue of mammalian CaMKK1 and CaMKK2 is Ssp1 kinase that was previously shown to be required for the G2/M cell cycle control and stress responses independent of MAP kinase (Matsusaka et al. 1995; Rupes et al. 1999; Wilhelm et al. 2008). Ssp1 is implicated in calcium signaling as temperature sensitive (ts) ssp1 mutants were isolated by the ability to suppress sts5 and ppe1 mutants that interacted with PKC and calcineurin mutants. The amino acid sequences of human CaMKK1, CaMKK2, and S. pombe Ssp1 sequences are compared in Fig. 1A with frog CaMKK. Identical residues between the vertebrate and S. pombe sequences are boxed in red. We then examined whether the calmodulin-binding domain (designated CBD) is conserved in SpSsp1. By applying the CaM binding site search (Choi & Husain 2006; http://calcium.uhnres.utoronto.ca/ctdb/ctdb/home.html), a putative CBD was found at the position corresponding to the CBD present in mammalian CaMKKs with significant probability scores (Fig. 1B). Schizosaccharomyces japonicus (Sj) and S. octosporus (So) are evolutionary relatives to S. pombe. In addition, we found a short stretch (431–438) of SpSsp1 (consensus VEVS(T)XE(D)EV), which locates between the kinase domain and the CBD, and is commonly present in the equivalent region of vertebrate CaMKKs (Fig. 1B bottom, arrow). YOGY indicates that S. cerevisiae SAK1, ELM1 and TOS3 kinases are most similar to Ssp1 among the S. pombe kinases. The stretch is neither present in ELM1 nor TOS3, but found in SAK1 although its location is different from the corresponding region (Fig. 1B top). SAK1 contains a putative CBD but its location is between the kinase domain and the conserved stretch. SAK1 was reported to be functionally similar to mammalian LKB1 (Hong et al. 2003), but the short stretch and the CBD are not found in LKB1.

image

Figure 1. Ssp1 resembles CaMKK and interacts with Rad24 and Rad25. (A) Comparison of amino acid sequences. Sp, Schizosaccharomyces pombe; Hs, Homo sapiens; Xl, Xenopus laevis; Sc, Saccharomyces cerevisiae. The accession numbers in GenBank and the total number of amino acids are shown. (B) Location (top) and sequences (bottom) of the CBD and the conserved stretch are shown. Numbers shown on the right of diagram indicate the linker length between the kinase domain and the conserved stretch. The arrow on the aligned sequences indicates the conserved stretch. CBDs are underlined based on results of the CaM binding site analyses and assignments of database. Sj, Schizosaccharomyces japonicus; So, Schizosaccharomyces octosporus. (C) Abundant proteins immuno-co-precipitated with Ssp1-FLAG are shown, with their molecular weight (MW) and the number of peptides detected by mass spectroscopic analysis.

Download figure to PowerPoint

Ssp1 associates with 14-3-3 proteins, Rad24 and Rad25

Vertebrate CaMKK associates with 14-3-3 proteins (Davare et al. 2004; Ichimura et al. 2008). We examined whether Ssp1 is also bound to S. pombe 14-3-3 proteins, Rad24 and Rad25 (Chen et al. 1999; van Heusden & Steensma 2006). For this end, the FLAG-tag was chromosomally integrated in frame to the carboxy terminus of the ssp1+ gene, and resulting integrated Ssp1-FLAG gene was expressed under the native promoter. The integrant formed normal colonies at 26 °C and 36 °C (data not shown). The Ssp1-FLAG thus retains function and was used for immunoprecipitation.

Immunoprecipitated proteins were prepared using anti-FLAG antibodies, and run in SDS-PAGE. Ekc1-FLAG was used as control (see below). Sliced gels were applied to the mass spectroscopic analysis according to the procedures previously described (Hayashi et al. 2007). Results of mass spectroscopic analyses shown in Fig. 1C indicate that Ssp1 is indeed co-precipitated with Rad24 and Rad25. In addition, three less abundant proteins (SPBC8E4.01c, Cwf10, SPAC1751.03/eIF3m) implicated in inorganic phosphate transport, RNA splicing, and initiation of protein synthesis, respectively, were found in the immunoprecipitates.

Ssp1/CaMKK is phosphorylated

The phos-tag compound (Kinoshita et al. 2006) that causes the retardation of phosphorylated protein bands in SDS-PAGE was employed to determine whether Ssp1 is phosphorylated. To detect Ssp1 in the extracts, the strain described above that expressed Ssp1-FLAG under the native promoter was used. As shown in the immunoblot patters of S. pombe cell extracts (Fig. 2A), multiple upper bands of Ssp1-FLAG were produced in the presence of phos-stag (left panel) in the acrylamide gel, which were shifted to the lower positions after the treatment by phosphatase (PPase). If phos-tag was not added to the gel, the upper bands, though their separation was inferior to the presence of phos-tag, were still detected for Ssp1-FLAG, which were shifted to the lower positions after the treatment of PPase (right panel). Control Ssp2-FLAG protein (the catalytic subunit of AMPK in S. pombe, see below) chromosomally integrated and expressed under the native promoter showed the upper bands dependent on the presence of phos-tag, but another control PP2A regulatory subunit Paa1-FLAG (the regulatory subunit A of PP2A, see below) did not. These results show that Ssp1 and Ssp2 are hyper-phosphorylated.

image

Figure 2. Phosphorylation sites of Ssp1, and high copy suppressor, mutation sites and phenotypes of ssp1 mutants. (A) Extracts of cells expressing Ssp1-FLAG were run in SDS-PAGE in the presence (left panel) or the absence (right) of 50 µM phos-tag. Each sample was preincubated with phage lambda protein phosphatase (PPase, +) or buffer (–). Ssp2-FLAG and Paa1-FLAG are control samples (see text). (B) Seven phosphopeptides assigned are shown with possible phosphorylation sites in red. The dots indicate the cleavage sites R or K by trypsin. (C) The ts phenotype of ssp1-511 mutant was suppressed by multicopy plasmid pSDS23 (top). The ts phenotype of Δsds23 deletion mutant at 33 °C was rescued by multicopy plasmid pSSP1 (bottom). (D) Determined mutation sites of seven ssp1 strains are schematically shown. (E) Mutant cells cultured at the restrictive temperature (36 °C) are shown after DNA (DAPI) staining. The bar, 10 µm. Arrows indicate condensed chromosomes. (F) Movies of cell divisions for wild-type (right) and ssp1-412 mutant (left) were taken, and their cell length and generation time were measured.

Download figure to PowerPoint

Phosphorylation sites of Ssp1 were searched by the analysis of mass spectroscopic data as previously described (Hayashi et al. 2007), and seven phosphopeptides were assigned (Fig. 2B). Peptides 2 and 4 contain only one phosphorylatable S residue, which are the consensus for PKA, PKC and CaMK. The conserved stretch (Fig. 1B) is in peptide 5.

Temperature-sensitive ssp1 mutants and high copy suppressor sds23+

A collection of approximately 1000 temperature-sensitive (ts) strains that were randomly mutagenized (Hayashi et al. 2004) was previously made, and mass gene cloning for these strains has been conducted, using an S. pombe genomic DNA library. It was found that two classes of plasmids carrying the ssp1+ or sds23+ gene rescued the ts phenotype of seven strains (404, 412, 511, 617, 837, 860, 871). Subcloning of plasmids established that single gene sequence of ssp1+ or sds23+ suppressed these mutants. Tetrad dissection showed that all these strains are ssp1 mutants and sds23+ is multicopy suppressor. In Fig. 2C top, the suppression of ssp1 mutants by pSDS23 is shown. The genetic relationship between ssp1+ and sds23+ is further shown by the reciprocal suppression of the ts phenotype of Δsds23 by plasmid pSSP1 (Fig. 2C bottom). Sds23/Moc1 was previously reported to be high copy suppressors of mutants defective in PP1 phosphatase or APC/cyclosome E3 ubiquitin ligase mutants (Ishii et al. 1996), and also sterility caused by overproduced adenyl cyclase (Yakura et al. 2006).

To establish mutation sites of ssp1 mutants above described, we employed the PCR method to clone the mutant genes and determined their nucleotide sequences. As shown in Fig. 2D, strains 412 and 860 revealed the same substitution mutation (G340D) within the kinase domain, but remaining five strains showed distinct mutations, all of which were nonsense (stop codon) mutations at different positions (81, 110, 195, 289, 337th) of Ssp1. All these ssp1 mutants showed cell elongation after the temperature shift up to 36 °C (Fig. 2E), consistent with the previous finding that ssp1 defect caused the delay in G2/M progression at 36 °C (Matsusaka et al. 1995). Occasionally condensed mitotic chromosomes (arrows in the figure) are observed, suggesting that Ssp1 might also function in mitotic progression. To study the mode of cell division cycle, ssp1-412 mutant cells were grown at a semi-permissive temperature, 33 °C, in the synthetic EMM2 medium. Movies were taken and analyzed in comparison with the wild-type control. The increase of cell length and the generation time were measured (Fig. 2F). Mutant cells divided nearly normally at this temperature (the generation time, 181.5 ± 20.5 min, 10.5% longer than that of wild-type (164.3 ± 7.8 min), but their cell length 15.4 ± 0.5 µm at the timing of cell division was significantly (31.6%) longer than that of wild-type cell length (11.7 ± 0.7 µm). Ssp1 kinase may affect the size control for cell division.

ssp1 and Δsds23 mutants are sensitive to limited glucose

We found that the cell multiplication of ssp1 mutant was defective in limited glucose. Mutant cells were cultured in different concentrations of glucose at the permissive (26 °C) temperature. As shown in Fig. 3A, the cell number increase of wild-type (WT) and ssp1-412 at 26 °C was indistinguishable in the synthetic EMM2 medium that contains the regular 2% glucose. Reduction of the glucose concentration down to 0.5% did not affect the rate of cell number increase. In 0.1% glucose, however, the cell number increase of mutant ssp1-412 was significantly delayed (approximately 50%) in comparison with that of wild-type. Below 0.02% glucose, both wild-type and mutant strains did not increase the cell number at all. In higher concentration of glucose (3%–4%), the rate of cell number increase was approximately identical between wild-type and ssp1-412 at 26 °C (Fig. 3B).

image

Figure 3. Ssp1 and Sds23 are required for the cell number increase under limited glucose concentrations. (A) EMM2 synthetic medium containing 2%, 0.5%, 0.1% or 0.02% glucose concentrations were employed, and the cell number increase of wild-type (975, gray line), ssp1-412 (red) and Δsds23 (blue) mutants at 26 °C was measured. (B) The wild-type, ssp1-412 and Δsds23 were cultured in 3% and 4% glucose, and the cell number increase at 26 °C was measured. (C) The rates of cell number increase at a semi-permissive temperature (30 ˚C). ssp1-511, green line. (D) The cell number increase of ssp1-412 mutant at 30 °C was restored by plasmid pSSP1 (gray) and partly by pSDS23 (green), but not by vector plasmid (red). (E) Deletion and Ala substitution mutant strains were constructed (see text). In the culture medium containing 0.1% glucose, the rate of the cell number increase at 36 °C is reduced for both deletion (green) and Ala (blue) substitution mutants (ssp1-412, red; wild-type, gray). The short stretch mutant cells cultured in 0.1% glucose for 10 h were round-shaped. The bar, 10 µm.

Download figure to PowerPoint

Sds23 was also required for cells to tolerate the low glucose concentrations. Deletion mutant Δsds23 was slow in the cell number increase even in 2%–4% glucose at 26 °C (approximately 70% of the wild-type cell number increase) as seen in Fig. 3A,B. In 0.1%–0.5% glucose, cell division was virtually abolished.

At the semi-permissive temperature (30 °C), the rate of cell number increase was slightly reduced in ssp1-412 and ssp1-511 mutants even in 2% glucose (Fig. 3C). In 0.1% glucose concentration these mutant cells only very slowly increased the cell number. The cell divisions of Δsds23 mutant scarcely occurred in 0.1% glucose concentrations at 30 °C.

The sensitivity of ssp1 mutant to limited glucose appeared to be due solely to ssp1 mutations, because the introduction of plasmid pSSP1 into mutant strain ssp1-412 suppressed the phenotype of this slow cell number increase (Fig. 3D). Plasmid pSDS23 also suppressed the phenotype of ssp1-412 mutant in a lesser degree.

The conserved stretch is required for normal response to limited glucose

To examine whether the conserved stretch (Fig. 1B arrow) is required for normal Ssp1 function, we made two mutations: One is the deletion of eight residues 431VEVSTDEV, whereas the other is the Ala substitution mutant 431AAASTDAA. Resulting mutant genes were chromosomally integrated and expressed under the native promoter. Both deletion and Ala substitution mutants increased normally the cell number in 2% glucose at 36 °C, the restrictive temperature, like the wild-type (0.5% glucose too; data not shown), but in 0.1% glucose, the cell number increase significantly slowed down (approximately 50%) in both mutants (Fig. 3E upper panels). Cell shape of these slowly dividing mutant cells was rather round, differing from the wild-type ellipsoidal shape in 0.1% glucose concentration (lower panels). The ts mutant cells were arrested at 36 °C, and their cell shape remained to be rod in 0.1% glucose.

Mass spectroscopic analysis shows that Sds23 is co-precipitated with PP2A and PP2A-related phosphatases

To understand the basis of Sds23-promoted high-copy suppression, mass spectroscopic analysis was carried out to examine proteins that were physically bound to Sds23. For this purpose, the sds23+ gene was tagged with FLAG, chromosomally integrated, and expressed under the native promoter. Antibodies against FLAG were employed to obtain immunoprecipitates. Proteins immuno-co-precipitated with Sds23-FLAG are shown in Fig. 4A (no-tag extracts was used as control). Subunits of PP2A and PP2A-related phosphatase Ppe1 were obtained. Paa1 and Pab1 are the regulatory subunits, and Ppa1 and Ppa2 are the catalytic subunits of S. pombe PP2A phosphatase (Kinoshita N et al. 1993; Kinoshita K et al. 1996). Ekc1 and Ppe1, the regulatory and catalytic subunits, respectively, form a PP2A-related phosphatase (Shimanuki et al. 1993; Goshima et al. 2003). The amino acid sequence identity was 54.1% between Ppe1 and Ppa2, and the cs phenotype of Δppe1 deletion mutant was rescued by plasmid carrying PP2A and PKC genes. Ekc1 resembles S. cerevisiae and human SAPs, whilst Ppe1 is similar to S. cerevisiae SIT4 and human PP6 phosphatase. Several other proteins unrelated to phosphatase were also co-precipitated.

image

Figure 4. Sds23 physically interacts with PP2A and PP2A-related phosphatase Ppe1 and is hyperphosphorylated. (A) Mass spectroscopic analysis of Sds23-FLAG immunoprecipitates using anti-FLAG antibodies. The molecular weights (MW) and the number of peptides for proteins detected are shown. Control is no-tag cell extracts immunoprecipitated by anti-FLAG antibodies. Sds23 and six PP2A and PP2A-related phosphatase subunits are shown in the green box. (B) Yeast two-hybrid interactions between Sds23 and PP2A-related phosphatase subunits. GA and GB, galactose activation domains and DNA binding domains, respectively, which are fused to these proteins. Positive control used is p53 and large T antigen. (C) Mass spectroscopic analysis of Ekc1-FLAG immunoprecipitates (Ekc1 is the regulatory subunit of Ppe1). Control is Ssp1-FLAG cell extracts immunoprecipitated by anti-FLAG antibodies (see Fig. 1C). Sds23 and three PP2A-related phosphatase subunits are shown in the green box. (D) Sds23 produced hyperphosphorylated multiple bands in SDS-PAGE in the presence of phos-tag (100 µM). After the treatment (+) by phage lambda phosphatase, the multiple bands were diminished. (E) Mass spectroscopic analysis of phosphorylated residues in Sds23-FLAG isolated from exponentially growing wild-type cells. Six phosphopeptides were detected. The S residues in red color are unambiguously phosphorylated in the peptides, whilst S residues in orange color in peptide 1, 5 could not be determined unambiguously.

Download figure to PowerPoint

To obtain more information about the association of Sds23 with phosphatases, yeast two hybrid interactions were employed (Fig. 4B). Paa1, but not Ppa1 nor Ppa2, showed the interaction with Sds23, suggesting that the regulatory subunit rather than the catalytic subunits of PP2A are bound to Sds23. Ekc1 showed the interaction only when it was fused to the galactose DNA binding domain (GB, bottom panel), and Ppe1 showed no sign of interaction with Sds23. Furthermore, two-hybrid analysis using the amino- and carboxy-terminal truncated Sds23 mutants indicated that deletions of both amino and carboxy termini failed to interact with Paa1 (data not shown). The terminal domains of Sds23 might be implicated in the interaction with phosphatases.

Immunoprecipitates of Ekc1-FLAG contains Sds23, Ppe1, Tip41, Kap109 and elongator subunits

To further establish that Ekc1 physically interacts with Sds23 and also with Ppe1 phosphatase, Ekc1 was tagged with FLAG, chromosomally integrated and expressed under the native promoter. Proteins co-immunoprecipitated by anti-FLAG antibody were analyzed by mass spectroscopy, and results are shown in Fig. 4C. Many peptides of Ppe1 and Sds23 were detected, confirming that Ekc1 physically interacts with Ppe1 and Sds23. None of peptides derived from PP2A subunits was obtained, indicating that Ekc1 is specifically bound to Ppe1. Sds23 thus forms the complex with PP2A, which is independent of Ppe1 phosphatase. Ekc1 was co-precipitated with the six subunits of the elongator complex that plays intriguingly diverse functions (Svejstrup 2007). The homologous protein of Tip41 in S. cerevisiae is implicated in the TOR signaling pathway and activates Ppe1-like SIT4 phosphatase (Jacinto et al. 2001). Kap109 is an importin family protein.

Sds23 is highly phosphorylated

Sds23 is highly phosphorylated. Extracts of cells expressing Sds23-FLAG was run in SDS-PAGE in the presence of phos-tag by the procedures described in Fig. 2A. Multiple bands were seen and diminished by phosphatase treatment (Fig. 4D), which indicate that Sds23 is highly phosphorylated. Phosphorylated residues of Sds23-FLAG were then determined by mass spectroscopy as described above. Phosphorylated residues were verified manually and shown in Fig. 4E. One of the six phophopeptides found is located in the amino terminus (peptide 1) and the remaining peptides are in the carboxy-terminal domain (peptides 2–6). Some phosphorylated residues (red characters in peptides 2, 3, 4 and 6) were unambiguously determined, whereas other phosphorylated residues (orange characters in peptides 1, 5) were not clearly determined. The phosphorylation of 333S at stationary phase was previously reported (Jang et al. 1997). The phosphorylation sites in the carboxy-terminal ends (peptides 5, 6) are in the RxxS and KxS consensus for PKA and PKC, whereas the sites in other peptides are often in the PS or SP consensus. The coverage of peptides by the mass spectroscopic method was 92% so that these residues should represent most of the phosphorylated residues.

Sds23 inhibits protein phosphatase that is sensitive to okadaic acid

We examined whether Sds23 might inhibit or activate PP2A and PP2A-related Ppe1 phosphatase. For this purpose, the synthetic phosphorylated peptide substrate, RRA(pT)VA, was employed. This was appropriate to assay the activity of PP2A by color development. To assess that the activity was due to protein phosphatase, okadaic acid (OA), a typical PP2A inhibitor, was added to the reaction mixture. As shown in Fig. 5A, the activity in the whole cell extract of wild-type (red line) was relatively low and the phosphatase activity was diminished in the presence of okadaic acid (red broken line). In sharp contrast, the activity was high in Δsds23 mutant cell extracts (blue line), and this activity was abolished if okadaic acid (5 µm final) was added (blue broken line), suggesting that Sds23 might be an inhibitor against the PP2A and PP2A-related phosphatase.

image

Figure 5. Sds23 is an inhibitor for PP2A-related phosphatase. (A) Cell extracts of wild-type (WT) and Δsds23 deletion mutant were made, and the phosphatase activities were measured, using the phosphorylated peptide substrate RRA(pT)VA. The assay was carried out at 30 °C for 0–75 min in the presence or the absence of 5 µm okadaic acid (OA) that inhibits PP2A and PP2A-related phosphatases. The free phosphate generated by phosphatase activities forms a complex of molybdate-malachite green-phosphate, and develops a green color. Resulting absorbances at 620 nm were measured as the phosphatase activity. The amount of phosphate was estimated according to the absorbances of phosphate standards (right vertical axis). (B) Immunoprecipitations of two strains Ekc1-FLAG and Sds23-FLAG both of which were chromosomally integrated and expressed under the native promoter were carried out using anti-FLAG antibody, and resulting precipitates were run in SDS-PAGE (left). Immunoblot was carried out using antibodies against FLAG, Ppe1 and Sds23. Phosphatase activities of Ekc1-FLAG and Sds23-FLAG immunoprecipitates (no tag precipitate was control) were measured (right). (C) GST-fused Sds23 protein was expressed in Escherichia coli and purified by affinity chromatography. Ekc1-FLAG prepared above was immunoprecipitated, dissociated from the beads by FLAG peptide and mixed with a series concentration (0.01–10 pmol) of purified Sds23-GST. Resulting mixtures were then run in SDS-PAGE, and stained by Coomassie Brilliant Blue (left panel) and immunoblotted using antibodies against FLAG, Ppe1 and Sds23 (middle panel). The phosphatase activities were assayed at 30 °C for 0–75 min using the phosphorylated peptide substrate (right).

Download figure to PowerPoint

We then prepared immunoprecipitates (IP) of Ekc1-FLAG and Sds23-FLAG using anti-FLAG antibodies, and compared the OA-inhibited phosphatase activities in the immunoprecipitates (Fig. 5B). The proteins bound to IP beads were released by FLAG and used for the assay. The amount of Ppe1 catalytic subunit was made equal in the Ekc1-FLAG IP and Sds23-FLAG IP by diluting Ekc1-FLAG IP 4-fold (left panel). Note that the levels of Sds23-FLAG and Ekc1-FLAG detected by anti-FLAG antibodies were approximately equal in the two IPs of Sds23-FLAG and 1/4 diluted Ekc1-FLAG. Polyclonal antibodies against Sds23 also detected a high level of Sds23-FLAG in SDS23 IP but a low level in 1/4-diluted Ekc1 IP. The phosphatase activities of these Ekc1-FLAG and Sds23-FLAG IPs were assayed and compared (right panel). The phosphatase activity of Ekc1-FLAG IP (×1) was the highest, and was completely inhibited by the addition of OA. In sharp control, the activity of Sds23-FLAG IP was not detected like no-tag control. In Ekc1 IP × 1/4 in which the level of Ppe1 was approximately equal to Sds23 IP, the phosphatase activity was highly reproducibly detected and OA inhibited the phosphatase activity.

To further provide evidence that Sds23 is the inhibitor of Ppe1 phosphatase, we constructed the GST-fused Sds23 protein that was produced in bacterial cells and purified. A series of Sds23-GST concentrations (0.01–10 pmol) were mixed with the constant amount of Ekc1-FLAG immunoprecipitated proteins (Coomassie Brilliant Blue-stained proteins bands and immunoblot patterns are shown in Fig. 5C left and middle panels, respectively). Among proteins present in the mixtures, only Sds23-GST was variable in the level, whereas that of Ppe1 and Ekc1-FLAG was constant. The phosphatase activity measured (right panel) was abolished when 10 pmol purified Sds23-GST was added, whilst the activity was partly inhibited when 1 pmol Sds23 (similar to the level of Ppe1) was added. Taken together, Sds23 inhibits Ppe1-Ekc1 phosphatase in a near-stoichiometric fashion.

Cbs2, but not Sds23, is the γ-subunit of AMPK and AMPK-related kinase

Sds23 contains four CBS (cystathione β-synthase) domains, a pair of which is considered to form the module of nucleotide binding (Scott et al. 2004). Regarding the number and arrangement of CBS domains, Sds23 resembles the γ-subunit of AMPK (as S. cerevisiae Snf4, human HsAMPK-γ2 in Fig. 6A). However, S. pombe contains Cbs2/SPAC1556.08c, which is more similar to the γ-subunit of AMPK than Sds23. We therefore attempted to clarify whether Cbs2 is really the γ-subunit of AMPK in S. pombe. Cbs2 was hence chromosomally tagged with FLAG at the carboxy terminus and expressed under the native promoter. Mass spectroscopic analysis of immunoprecipitates by anti-FLAG antibodies is shown in Fig. 6B. Indeed two kinase catalytic subunits Ssp2/SPCC74.03c and Ppk9, which are similar to AMPK catalytic subunits, and other protein SPCC1919.03c similar to the β-subunit of AMPK (designated Amk2) were co-precipitated with Cbs2. Ade4 is also co-precipitated. Ssp2 contains the ubiquitin-associating UBA domain immediately adjacent to the kinase domain. This feature resembles mammalian AMPK-related kinases MARK or BRSK, which are the substrates of LKB1 kinase (Jaleel et al. 2006).

image

Figure 6. Cbs2, but not Sds23, is the γ-subunit of AMPK and AMPK-related kinases. (A) Sds23 and Cbs2 contain four CBS domains. Saccharomyces cerevisiae Snf4 is the γ-subunit of AMPK/Snf1. The fourth CBS domain of S. pombe and S. cerevisiae Sds23 contain the interrupting sequences. (B) Mass spectroscopic analysis of Cbs2-FLAG immunoprecipitates. Control is no-tag immunoprecipitates by anti-FLAG antibodies. (C) Mass spectroscopic analysis of Ssp2-FLAG immunoprecipitates. Control is Ssp1-FLAG immunoprecipitates by anti-FLAG antibodies (see Fig. 1C).

Download figure to PowerPoint

To determine whether Ssp2 forms the AMPK-like complex, we then constructed chromosomally FLAG-tagged strain that expresses Ssp2-FLAG under the native promoter. Mass spectroscopic results of immunoprecipitates made by anti-FLAG antibodies (Fig. 6C) indicate that Ssp2 is associated with Amk2, Cbs2 and Ade4. Neither Ppk9 nor Sds23 was present in the immunoprecipitates as a significant level, suggesting that Ssp2 kinase structure is independent of Ppk9 and Sds23. Schizosaccharomyces pombe thus has two AMPK kinase complexes; one UBA-containing AMPK-related Ssp2-Amk2-Cbs2, and the other is AMPK Ppk9-Amk2-Cbs2. Ade4 may be a specific subunit of these kinases or the major substrate, and this remains to be clarified. Sds23 was not detected in any of these AMPK-related complexes at a significant level.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

In this study, we show that S. pombe Ssp1 protein kinase and its multi-copy suppressor Sds23 are required for tolerance to limited glucose. As Sds23 is a phosphatase inhibitor, protein phosphorylation is important for the tolerance to limited glucose in S. pombe. Wild-type cells can multiply in the culture medium containing 0.1% glucose (instead of 2%), but ssp1 defective and Δsds23 deletion mutants do not so that Ppe1 (and PP2A) phosphatase may be inhibitory to the utilization of limited glucose. Figure 7 schematizes hypothetical relationships among the protein complexes described in this study. We postulate that Ssp1 is a conserved member of CaMKK. Firstly, it has the highest similarity score to CaMKK among the S. pombe kinases, and a putative CaM binding domain (CBD) is conserved at the site similar to mammalian CaMKK. Second, the short stretch outside the kinase domain is conserved. This stretch is required for normal Ssp1 function in response to limited glucose.

image

Figure 7. Ssp1/CaMKK and Sds23 are required to tolerate limited glucose. Wild-type S. pombe cells are able to multiply in the culture medium containing only 0.1% glucose (1/20 diluted from normal concentration). The present study shows that Ssp1, the putative CaMKK, and CBS-domain protein Sds23, an inhibitor of PP2A-related phosphatases, are required to tolerate 0.1% glucose for the cell number increase. See text for further explanation.

Download figure to PowerPoint

Third, Ssp1 binds to 14-3-3 homologues in S. pombe as mammalian CaMKK does. 14-3-3 usually binds to phosphorylated proteins. Indeed, Ssp1 is highly phosphorylated, and produces multiple bands in SDS-PAGE, which are diminished by the treatment with phosphatases. Mass spectroscopic analysis indicates that Ssp1 has eight phosphorylated peptides outside the kinase domain: phosphorylation may regulate protein–protein interactions rather than the kinase activity. Certain Ssp1 phosphorylation sites, R/KxS and RxxS, are the consensus for PKA, PKC and CaMK (Kennelly & Krebs 1991; Blom et al. 1999). Mammalian CaMKK is inhibited by 14-3-3 and PKA (Davare et al. 2004). Phosphorylation by PKA, PKC and CaMK may thus inhibit Ssp1 through association with Rad24 and Rad25. Cdc25, a CDK regulator required for the G2/M progression, is a critical target of PKA, resulting into sequestration by 14-3-3 in embryonic G2 cells (Duckworth et al. 2002); a drop of PKA activity is required for activation of Cdc25 and the entry into M-phase. Ssp1 also required for the G2/M transition may similarly be regulated by 14-3-3 and PKA, although its relationship to the activation of Cdc2-Cdc13 CDK is unknown.

Fourth, both Ssp1 and CaMKK are required for efficient glucose utilization. Mammalian CaMKK has been recently shown to regulate glucose utilization and uptake, and is proposed to be implicated in diabetes (Huang & Czech 2007; Witczak et al. 2007; Anderson et al. 2008). Schizosaccharomyces pombe Ssp1 becomes essential for normal cell multiplication in the medium containing 0.1% glucose at 26 °C and 30 °C. The structure, behavior and role of Ssp1 are hence consistent with the hypothesis that Ssp1 is CaMKK. Ssp1 will be an excellent model for mammalian CaMKK through the use of powerful S. pombe genetics, particularly in understanding its role for the glucose utilization. Note that 0.1% glucose in the culture medium is approximately equal to the blood sugar content (100 mg/dL) so that S. pombe multiplication under such glucose concentration should be physiologically relevant, regarding the situation of mammalian body cells to use glucose in blood vessels. Aspergillus nidulans was reported to possess a CaMKK-like kinase that is dependent on calmodulin (Joseph & Means 2000).

Both ssp1 and ssp2 mutants were initially isolated by the same isolation procedures as genomic suppressors for ppe1 and sts5 mutants (Matsusaka et al. 1995) that are implicated in calcium signaling: sts5 mutant is suppressed by PKC plasmid and synthetic lethal with calcineurin mutant Δppb1 (Yoshida et al. 1994; Toda et al. 1996), whereas Δppe1 mutant is synthetic lethal with ppa2 and pkc1 mutants, and suppressed by high copy plasmids carrying PKC and PP2A catalytic subunit genes (Shimanuki et al. 1993). The relationship between Ssp1 and Ssp2, putative CaMKK and AMPK-related kinase, respectively, is the subject in future studies.

In vitro phosphatase assay shows that Sds23 is a novel inhibitor of PP2A-related phosphatase Ppe1 that resembles S. cerevisiae SIT4, as the phosphatase incubated with Sds23 became inactive. This is an important finding to understand the role of Sds23. Mass spectroscopic analysis demonstrates that Sds23 forms the stable complex with the phosphatase subunits. Two hybrid interaction data suggest that Sds23 may directly interact with the regulatory subunits. We interpret this finding that Sds23 and Ssp1 may synergistically act, as the sds23+ gene is a multi-copy suppressor for ssp1 mutants, and the high copy ssp1+ gene reciprocally rescues Δsds23 deletion mutant. The PP2A and PP2A-related phosphatase activities may be antagonistic to Ssp1 kinase and the up-regulated phosphatase activities in Δsds23 cells may result into the sensitivity to limited glucose. Sds23 is highly phosphorylated. It remains to be determined whether phosphorylation affects the ability of inhibitor.

Sds23 was initially identified by its multi-copy suppressor role for the ts phenotype of PP1 regulator mutant and APC/cyclosome mutants (Ishii et al. 1996). The present finding that Sds23 binds to PP2A and PP2A-related phosphatase rather than PP1 was thus unexpected. However, if these phosphatases are antagonistic to PP1 and APC/cyclosome, multi-copy suppressions of PP1 and APC/cyclosome mutants by Sds23 may be explained. In S. pombe, PP1 phosphatase is required for the transition from mitotic metaphase to anaphase, whilst PP2A inhibits the entry into mitosis (Yanagida et al. 1992). As PP2A and PP2A-related phosphatases have diverse substrates, inhibitor Sds23 may have broad influences on cellular functions. Consistently, the multi-copy Sds23 rescued the sterile phenotype of over-expressed cyr1 adenyl cyclase (Yakura et al. 2006). Both Ssp1 and Sds23 are required for the maintenance of S. pombe quiescent cells under nitrogen starvation (Shimanuki et al. 2007; Sajiki et al. 2009), consistent with their housekeeping and nutritional roles in non-dividing cells.

Sds23 containing four CBS domains resembles the γ-subunit of AMPK. We at first suggested a possibility that Sds23 might be associated with AMPK. Mass spectroscopic results indicate, however, that this is unlikely. Other protein Cbs2 is the AMPK γ-subunit as it binds to two AMPK-like catalytic subunits Ssp2 and Ppk9, and Amk2, a protein similar to the β-subunit. Ssp2 containing the UBA domain is of considerable interest, as it resembles mammalian AMPK-related MARK or BRSK that is the downstream of LKB1. It remains to be determined whether S. pombe has LKB1-like kinase. Sds23-like proteins are present in all fungi examined, and possibly also in the cellular slime mold Dictyostelium (Y.K., unpublished), but its presence in higher animals and plants is unknown. The hetero-trimeric enzyme AMPK is activated when a pair of CBS domains in the γ-subunit binds to 5′-AMP. It is unknown whether the CBS domains of Sds23 also bind to AMP or other compound. It is of considerable interest to test whether Sds23 and AMPK may be regulated by identical or similar cellular signaling compound such as AMP.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Strains, media and plasmids

All Schizosaccharomyces pombe strains used were derived from the wild-type haploid h 972. Temperature-sensitive ssp1 mutants were newly isolated from the collection of 1015 ts strains (Hayashi et al. 2004). The Δsds23 deletion strain constructed previously was used (Shimanuki et al. 2007). Complete YPD (1% yeast extract, 2% polypeptone and 2% glucose) and minimal EMM2 (Moreno et al. 1991) media were used. Incubation in limited glucose (0.1%–0.5%) media was carried out as follows. Cells were grown in EMM2 to a concentration of 5 × 106 cells/mL at 26 °C. They were harvested by vacuum filtration using a nitrocellulose membrane (0.45 µm pore size, ADVANTEC), washed with EMM2–G (EMM2 lacking Glucose) once on the membrane, and then re-suspended in EMM2 containing indicated glucose concentration. The multi-copy plasmids pSSP1 (Matsusaka et al. 1995) and pSDS23 (Ishii et al. 1996) were previously described.

Construction of yeast strains

To construct deletion and alanine substitution mutants in the conserved stretch of Ssp1 kinase, site-directed PCR-based mutagenesis was carried out. In brief, complementary pairs of oligo DNA with mutations were used as PCR primers followed by two rounds of PCR. To construct chromosomally integrated strains, the mutated ssp1 gene was cloned and introduced into the endogenous ssp1 locus together with the drug-resistant kanMX6 marker. A similar strategy was used to tag the C-termini of ssp1+, ekc1+, ssp2+, paa1+, sds23+, ekc1+ or cbs2+. The FLAG tag sequence was inserted at the C terminus of the ORFs of these genes, followed by the kanMX6 marker, and introduced into endogenous locus. Correct integration was verified by PCR, and digestion of the PCR products with restriction enzymes.

Microscopy

DAPI staining was carried out as described before (Adachi & Yanagida 1989). Cells were fixed with 2.5% glutaraldehyde for 20 min on ice, washed three times with phosphate buffered saline (PBS), and stained with DAPI (25 µg/mL). Images were taken with the fluorescence microscope (BIOREVO BZ-9000, KEYENCE). For measuring the cell length in time course, live cell images were recorded using the DeltaVision microscopy system (Applied Precision).

Immunopurification

Exponentially growing cells (5 × 109) were lysed in the KB buffer (25 mm Tris–HCl at pH 7.5, 15 mm EDTA, 60 mmβ-glycerophosphate, 0.1% NP-40, 10% glycerol, 1 mm PMSF, 2 mg/mL pepstatin A and 1 mg/mL aprotinin). Extracts were centrifuged twice (5 min at 2300 g and 20 min at 17 700 g), and incubated with 100 µL (resin volume) of anti-FLAG M2 affinity gel (Sigma-Aldrich) for 2 h. The beads were then washed with the KB buffer. Eluates were obtained by incubating with 200 µL of 200 ng/µL 3 × FLAG peptide (Sigma-Aldrich).

Mass spectrometry

The procedures carried out were essentially the same as described before (Hayashi et al. 2007). Immunopurified samples were separated on a 12.5% SDS-PAGE, and visualized with Coomassie Brilliant Blue staining. The area from the top to the bottom of the separation gel was cut at approximately 1–2-mm intervals. After in-gel digestion with modified trypsin (Roche), the resulting peptides were analyzed by online LC-MS/MS on a Finnigan LTQ (Thermo Fisher). All MSMS spectra were searched against the S. pombe non-redundant protein database including common contaminants such as trypsin and keratin with the Mascot program (Matrix Science). The output data from Mascot was analyzed using in-house software to select reliable peptides. Phosphopeptides were identified by Mascot.

Phosphate affinity SDS-PAGE

Phosphate affinity SDS-PAGE was carried out using Phos-tag® Acrylamide (NARD institute) following the manufacturer's directions. The λ PPase (New England Biolabs) was used for phosphatase treatment. The PPase-treated or non-treated cell extracts were run on the polyacrylamide gel containing the Phos-tag acrylamide and MnCl2. To increase the transfer efficiency, the manganese ions were eliminated from the gel by soaking in the transfer buffer containing 1 mm EDTA. Immunoblot was carried out using the anti-FLAG M2 (Sigma-Aldrich).

Yeast two-hybrid analyses

The yeast two-hybrid analysis was carried out according to the procedures described in the manual for the two-hybrid analysis kit (MatchmakerTM, Clontech). Each cDNA was amplified by PCR and cloned into pGBT9 and pGAD424. The ß-galactosidase filter assay was carried out using the SFY526 strain that carried the GAL1-lacZ reporter as described by the manufacturer's directions.

GST protein purification

For the expression of GST fused Sds23 protein in Escherichia coli, the OvernightExpressTM system (Merck) and the plasmid pGEX-KG were employed. GST-affinity chromatography was carried out by the AKTA explorer system (GE Healthcare): the GSTrap-HP column was used. Binding of Sds23-GST in E. coli extracts was carried out in phosphate-buffer saline (PBS) with 1 mm dithiothreitol. A 3-mM Mg-ATP was used in the washing buffer. Sds23-GST bound was eluted by the buffer containing 50 mm Tris–HCl, 10 mm reduced glutathione and 1 mm dithiothreitol (pH 8.1).

Type 2A protein phosphatase assay

The phosphatase activities were measured using the Ser/Thr Phosphatase Assay System (Promega) following the manufacturer's directions. Cell extracts were prepared in the PP2A storage buffer (25 mm Tris–HCl, pH 7.5, 1 mm 2-mercaptoethanol, 2 mm EDTA, 0.1 mm PMSF, 0.1% triton X-100) containing a cocktail of protease inhibitors (Sigma-Aldrich). For measuring the phosphatase activities in the whole cell extracts, free phosphate was removed by fractionation on Sephadex G-25, and extracts containing 10 µg protein were used in the assay. For measuring the activities of immunoprecipitates, anti-FLAG immunopurification was carried out as above using the PP2A storage buffer, and 5 µL/reaction of the eluate were used. The phosphopeptide substrate RRA(pT)VA is a compatible with type 2A, 2B and 2C protein phosphatases. The assay was made specific for type 2A by employing the assay buffer without Mg2+ and Ca2+. The amount of free phosphate generated by the phosphatase activity was determined by measuring the absorbance of a molybdate: malachite green: phosphate complex.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

This work was supported initially by the COE specially promoted grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan and later by the CREST grant from the Japan Science and Technology Corporation. The work at OIST was supported by the fund of Initial Research Project for the Okinawa Institute of Science and Technology.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References