Transition from proliferation to quiescence brings about extensive changes in cellular behavior and structure. However, the genes that are crucial for establishing and/or maintaining quiescence are largely unknown. The fission yeast Schizosaccharomyces pombe is an excellent model in which to study this problem, because it becomes quiescent under nitrogen starvation. Here, we characterize 610 temperature-sensitive mutants, and identify 33 genes that are required for entry into and maintenance of quiescence. These genes cover a broad range of cellular functions in the cytoplasm, membrane and nucleus. They encode proteins for stress-responsive and cell-cycle kinase signaling pathways, for actin-bound and osmo-controlling endosome formation, for RNA transcription, splicing and ribosome biogenesis, for chromatin silencing, for biosynthesis of lipids and ATP, for cell-wall and membrane morphogenesis, and for protein trafficking and vesicle fusion. We specifically highlight Fcp1, a CTD phosphatase of RNA polymerase II, which differentially affects the transcription of genes that are involved in quiescence and proliferation. We propose that the transcriptional role of Fcp1 is central in differentiating quiescence from proliferation.
Control of cell proliferation and quiescence is a central problem in biology. The transitions from proliferation to quiescence and vice versa have a broad range of implications for development, differentiation, cancer and longevity. Cells must have strategies for adapting to different conditions that will allow them to survive through division or arrest. Cellular quiescence may be defined as reversible or irreversible proliferation arrest, and the arrest is induced by withdrawal of the stimuli for either growth or proliferation. As cells become quiescent they undergo profound changes to their physiology and organization. In multicellular organisms, the majority of body cells do not divide and are maintained for their lifespan with occasional turnover. Identification of the gene functions that are required for the entry into quiescence and for its maintenance is of great importance. However, genes that are crucial for establishing and/or maintaining quiescence are largely unknown. Although the genes required for proliferation, such as the cell-division cycle (CDC) genes, have been identified and well integrated into the network of related gene functions, the genes required for quiescence have not.
The fission yeast Schizosaccharomyces pombe is easily amenable to genetic modification and its biology resembles mammals in many ways, such as in cell-cycle control, chromosome segregation, gene silencing, mitochondrial biogenesis and meiosis (Chikashige et al., 2006; Harigaya et al., 2006; Nurse, 1990; Schafer, 2003; Volpe et al., 2002; Yanagida, 2005). A quiescent cell state of S. pombe is obtained by incubating a heterothallic strain without a nitrogen source (Su et al., 1996), and offers opportunities for studying the core processes underlying quiescence. However, how similar the quiescence induced by such starvation in eukaryotic microbes is to the quiescence that is experimentally brought about by the absence of serum factors in mammalian cells is unknown. Quiescent cells are generally distinguishable from proliferating cells by their size, shape and homogenous cell state. Characterization of S. pombe quiescent cells under nitrogen starvation indicated that they are resistant to various stresses (Su et al., 1996), are highly efficient in DNA damage repair and are metabolically active (Mochida and Yanagida, 2006). Also, they contain transcripts that are distinct from proliferating cells and the cells greatly alter their constituents upon replenishment of the nitrogen source to restore proliferation (Shimanuki et al., 2007). In this study, a method has been developed to search for the genes required for both proliferation and quiescence using 610 temperature-sensitive (ts) mutants. Only about a quarter of the strains examined were found to be defective at quiescence entry and/or maintenance. We have identified 33 core genes that are crucial for entry into and maintenance of quiescence, which revealed well-defined subsets of cellular processes. The results of dissection on their roles in quiescence entry and/or maintenance are presented.
Temperature-sensitive mutants examined for quiescence-entry
The culture medium used in this study (EMM2, Edinburgh minimal medium, referred to as the +N medium hereafter) (Mitchison, 1970) contains glucose and NH4Cl as the respective carbon and nitrogen sources. This medium supports proliferative divisions of S. pombe. By contrast, EMM2-N (referred to as the -N medium hereafter), which lacks NH4Cl but is otherwise identical to +N medium, supports the quiescent non-dividing cells that utilize the intracellular nitrogen source. Heterothallic (h-) S. pombe ts strains previously made by random mutagenesis were used; their defective phenotypes were determined in the +N medium at 36°C (Hayashi et al., 2004). To observe the phenotype in the -N medium, the genetic markers, including the nutrient marker leu1-32, were eliminated from each strain by crossing with the wild type. The resulting 610 ts strains containing only the ts mutation were grown in the +N liquid medium at 26°C and then transferred to the -N medium at 26°C for 24 hours.
The wild-type cells, which were rod-like in the +N medium, became small and round after about two rounds of the division and were completely arrested with 1C DNA (Fig. 1A,B; refer to Table 1 for details). They retained a high viability (>80%) for at least a month, providing that the -N medium was refreshed at appropriate intervals. A movie of wild-type cells showing the transfer from the proliferate state in +N medium to the quiescent state in -N medium is shown in supplementary material Movie 1.
Genes identified by the loss of viability in entering quiescence
We examined cell number, percent viability and cell morphology by light microscopy for all the 610 mutants at 0 and 24 hours at 26°C cultured in the -N medium. Almost all of the mutants entered quiescence at this temperature (the temperature is permissive in the +N medium), with high viability (>70%). Twelve strains did not, however, enter quiescence, and their viability significantly decreased (<35%). Note that viability represents the ability of cells to exit from quiescence and to form colonies upon replenishment of the nitrogen source. These strains were thus hypersensitive to nitrogen starvation. We first attempted to identify the mutant genes by isolating the single plasmid gene that fully rescued the ts phenotypes. Nucleotide sequencing of the genes isolated by PCR from the mutant genomic DNA was then performed to determine the mutation sites. We established that the eight strains corresponded to seven distinct mutant genes (listed in Table 1).
As shown in Fig. 1C, two of the mutant genes are the mitogen-activated MAP kinase Sty1 (also known as MAP kinase Spc1 and MAPK) (Millar et al., 1995; Shiozaki and Russell, 1995) and the activator Wis1 MAP kinase kinase (MAPKK) (Warbrick and Fantes, 1991). The other genes corresponded to Ypt5 (RAB5 in mammals) (Armstrong et al., 1993), Vam6 (VPS39 in mammals) and SPAC823.12 (also known as Vps11 and PEP5), which are involved in vacuole (lysosome in higher eukaryotes) fusion, and to Wsp1 (Lee et al., 2000) (also known as LAS17 and WASP) and End4 (Iwaki et al., 2004) (also known as SLA2) proteins, both of which interact with actin. Genes for the remaining four mutants could not be determined because the plasmid genes (vas1+, cpc2+ and ptc1+) turned out to be high-copy suppressors.
Requirement of stress-responsive kinases for entry into quiescence
After shifting to the -N medium, the wis1 and sty1 mutants showed no change in cell shape, but undertook two rounds of divisions (Fig. 2A,B). As seen in supplementary material Movies 2 (wis1-982) and 3 (sty1-989), prior to arrest, cells of these mutants divided in the same way as proliferating cells. However, the nuclear chromatin region stained with DAPI was highly extended after 24 hours. The sharp decline of viability was initiated after two rounds of division in both sty1-989 and wis1-982 mutants (12 hours) (Fig. 2B, lower panel) (Warbrick and Fantes, 1991). The majority of mutant cells contained 2C DNA (see Table 1), and thus resembled S. pombe cells growing in the nutrient (Fig. 2C). Sty1 and Wis1 were not required for the divisions per se after the shift to -N, but were required for the arrest of cell growth under the stress of nitrogen starvation and for the development of characteristics of quiescent cells (round cell shape, small nucleus and 1C DNA), which is consistent with the previous reports (Kanoh et al., 1996; Shiozaki and Russell, 1996).
The mutation of wis1-558 resides in the kinase domain and that of wis1-982 resides 19 amino acids upstream of the kinase domain, whereas sty1-989 is a nonsense mutation in the middle of the catalytic domain. Therefore, the phenotypes were probably caused by the lack of kinase activity. The sty1 deletion showed a phenotype similar to sty1-989 (data not shown). The nuclear extension seen by DAPI was confirmed by electron microscopy (Fig. 2D): an unusually large nucleus was observed (nitrogen-starved wild-type control electron micrograph is shown in Fig. 3E). Otherwise, electron micrographs of mutant cells in the -N medium resembled the proliferating rod-like cells. The timing of nuclear extension was determined after the shift to the -N medium (Fig. 2E). The extension of the nucleus occurred around 12 hours after the cease of cell division and without an increase in cell length (Neumann and Nurse, 2007).
To understand what happened to cell-cycle regulation in sty1 and wis1 mutants after the shift to the -N medium, we examined the levels of some cell-cycle regulators by immunoblot (Fig. 2F). In the wild type, the level of mitotic cyclin (Cdc13) decreased after 6 hours in the -N medium, and the level of the CDK inhibitor Rum1 (homolog of S. cerevisiae Sic1) (Labib and Moreno, 1996; Moreno and Nurse, 1994) sharply increased at around 6 hours, suggesting that most CDK might become inactive. By sharp contrast, the level of Rum1 did not increase at all in the sty1-wis1 mutants, even 24 hours after the shift to -N, whereas the level of Cdc13 was relatively high. The Rum1 elevation was reported to be implicated in normal G1 arrest (Daga et al., 2003).
It was found, however, that the Rum1 deletion mutant (Δrum1), which grew normally in the +N medium, became small and spherical and contained 1C and 2C DNA when grown in the -N medium (24 hours, 26°C), as shown in Fig. 2G. The viability of Δrum1 was high (60%) in the -N medium so that Δrum1 arrested the cell growth principally in the post-replicative state, but did not mimic the phenotype of sty1-wis1 mutants in the -N medium. We also examined Atf1 (Takeda et al., 1995), another target of Sty1 and the transcription factor of Wis1-Sty1. The deletion mutant Δatf1 cells became round in shape and retained their high viability when grown in the -N medium (data not shown). Double deletion atf1 pcr1 and triple deletion atf1 atf21 pcr1 mutants (a gift of Kazuhiro Shiozaki, University of California, Davis, CA) also gave the small round cells, strongly suggesting that loss of Atf1 and related transcription factors (Davidson et al., 2004) was not directly involved in generating the cell shape phenotype observed in sty1 and wis1 mutants. Taken together, the results suggest that the Sty1-Wis1 pathway must play a major role in the entry into quiescence and that the target for arresting cell growth in the -N medium is something other than Rum1 and Atf1. Nitrogen starvation was shown to induce transient activation of Sty1 (Shiozaki and Russell, 1996), apparently via TOR signaling (Petersen and Nurse, 2007). The relationship between Sty1 MAPK and Cdc2-Cdc13 was unclear, although the increase of Rum1 definitely needed Sty1-Wis1.
Pre-quiescence division requires vacuole dynamics
In the three mutants, vps11-319, vam6-532 and ypt5-909, pre-quiescence division did not occur (Fig. 3A, upper panel; Table 2). Loss of viability occurred quickly in ypt5 and slowly in vam6 and vps11 mutants (Fig. 3A, lower panel). Mutant vam6 cells could undertake one division, probably because of the residual remains of nitrogen source on the agar used in supplementary material Movie 4. This was followed by swelling and apparent intracellular disorganization. The other common phenotypes of these three mutants were partly rounded cells with post-replicative 2C DNA content (Fig. 3B). In these mutants, the rise of the Rum1 level did not occur, but the amount of mitotic cyclin Cdc13 was high in comparison with the wild-type control in -N (Fig. 3C). This suggested that many features of proliferating cells remained, perhaps owing to the lack of cell division.
Ypt5, a RAB5-like G protein, is required for vesicle fusion (Armstrong et al., 1993). The mutation site in ypt5-909 resides in the C-terminal Cys209 (being altered to Arg). This residue is required for lipid modification (prenylation, geranylgeranylation) (Giannakouros et al., 1993; Newman et al., 1992), suggesting that the mutant phenotype was because of the diminished protein modification required for activation of the G-protein. Vam6 and Vps11 are, respectively, a GTP exchange factor GEF (Wurmser et al., 2000) and a RING-finger ubiquitin ligase (Rieder and Emr, 1997). Together these form the complex called HOPS (Starai et al., 2008), which is required for vacuolar fusion (Price et al., 2000). It was recently reported that RAB5 and HOPS form the maturation pathway for the phagosome that is required for autophagy proteolysis (Kinchen et al., 2008). FM4-64 (a fluorescent probe for vacuoles) (Vida and Emr, 1995) was used to stain vacuolar structures in these mutants (Fig. 3D). Numerous small vesicles were observed in vam6 and vps11 mutant cells, possibly owing to diminished vacuolar fusion. Normal vacuoles were found in the wild-type control. In thin-section electron micrographs of these two mutants (Fig. 3F,G; wild-type control, Fig. 3E), unknown cytoplasmic vesicles (probably corresponding to the small vesicles stained with FM4-64) were highly abundant. These might be non-fused vacuoles or their precursors or phagosomes that were not fused with vacuoles. By contrast, the vacuolar features of ypt5-909 (Fig. 3H; also see Fig. 3D, bottom panel) were more like those of vegetative cells.
Cortical actin-interacting endosome proteins might provide an osmo-environment
Two strains, wsp1-318 and end4-507, responded to the -N shift more normally than the above mutants: the resulting cells were small and round after 24 hours, and contained 1C DNA (Fig. 4A). The level of Rum1 increased while Cdc13 decreased (data not shown). The majority of the mutant cells divided twice and arrested like the wild type (Fig. 4B, upper panel), but their viability was low (14% and 1% for wsp1-318 and end4-507, respectively). The decrease occurred late for wsp1-318, but that of end4-507 was fast (Fig. 4B, lower panel). Surprisingly, however, the viability of end4-507 was fully restored after 144 hours. This remarkable recovery of end4-507 was highly reproducible and seemed to be caused by `adaptation' to the -N medium (evidence is given below).
The mutations of wsp1-318 and end4-507 reside at the N-terminal conserved regions, W44R in the EVH1 (WH1) domain (Higgs and Pollard, 2001) and G73D in the ANTH domain (Kay et al., 1999). Patients of WASP (Wiskott-Aldrich syndrome) (Notarangelo et al., 2005) have mutations in this same region of the gene. In Fig. 4C, the cortical actin visualized by rhodamine-conjugated phalloidin was scarce in wsp1-318, but isometric actin localization was seen in wild-type and in end4-507 cells. This finding is consistent with reports that, although both Wsp1 and End4 interact with actin and function in endosome formation (Kaksonen et al., 2003), actin needs Wsp1 to form the patch at the cortex (Sirotkin et al., 2005). End4, a SLA2 homolog that interacts with Huntingtin (Engqvist-Goldstein et al., 1999; Kalchman et al., 1997), is needed for the release of endosomes from the cortex (Kaksonen et al., 2003).
How do these mutants fail to restore cell growth and division upon the +N replenishment? supplementary material Movies 5 and 6, for wsp1 and end4 mutant, respectively, clearly demonstrated that they were physically disrupted upon the +N replenishment. These time-lapse images of wsp1-318 (Fig. 4D) and end4-507 (Fig. 4E, left) show that they failed to increase in cell size, and instead appeared to burst and/or shrink after 13-20 hours (wsp1-318) and 9.5-13 hours (end4-507) after +N replenishment. However, in the case of the end4-507 mutant `adapted' in the -N medium for 96 hours then shifted to the +N medium, cells could elongate and divide (supplementary material Movie 7; Fig. 4E, right). The growing end4-507 cells showed mono-polar growth, as previously reported (Castagnetti et al., 2005), in contrast to the bi-polar growth of wild-type cells (supplementary material Movie 8; Fig. 4F). These results indicated that the defective quiescent state already prescribed lethal events after the shift to conditions for growth.
To obtain information on intracellular structure, thin-section electron microscopy of wsp1 and end4 was done. No significant structural difference from the wild-type control cells was seen for wsp1-318 after 24 hours in -N medium at 26°C (Fig. 4G). By contrast, end4-507 cells revealed a conspicuous mushroom-like structure formed on the cortex (indicated by the red arrow in Fig. 4H and shown enlarged on the right). This mushroom-like structure was found in every cell after 24 hours but not after 96 hours in the -N medium. The head of the mushroom was often bound to vacuoles. It is probable that these aberrant structures were derived from defective endosome formation and were the cause of cell death.
We then examined whether these mutant cells might be osmo-sensitive. Indeed, the lethal phenotype of end4-507 and wsp1-318 in -N medium at 26°C was partly rescued by the addition of 1.2 M sorbitol (Fig. 4I). This is because end4 mutant cells pretreated in the -N liquid medium for 24 hours and then plated on the rich YPD produced colonies dependent on the presence of 1.2 M sorbitol, indicating that proliferating end4 mutant cells are osmo-sensitive for proliferation. If wsp1 mutant cells were treated in the -N liquid medium containing 1.2 M sorbitol before plating, they could then produce colonies on the YPD plate without sorbitol. This suggests that wsp1 mutant cells require conditions of high osmolarity to maintain viability during the entry into quiescence. If 1.2 M sorbitol was present throughout, both strains produced colonies.
Genes needed for quiescence-maintenance
In total, 610 strains were incubated at 37°C for 3 days in the -N medium after culture for 1-day quiescence-entry at 26°C. Of these, 164 strains showed significantly decreased viability (<50%) in this second-stage of culture for quiescence-maintenance. An unexpected conclusion was that approximately three-quarters of those ts genes required for proliferation might be dispensable for maintaining quiescence.
We attempted to identify the mutant genes for 34 strains by gene cloning, tetrad dissection and nucleotide sequence determination of the mutant genes. Twenty-six genes were identified from 34 strains, as shown in Table 2, and their predicted gene functions are schematized in Fig. 5. These functions covered a broad range of cellular activities involving the cell surface, cytosol, cytoplasmic organelles and nucleus. Four genes encode cell-cycle-related protein phosphorylation and dephosphorylation, and two are implicated in sister chromatid cohesion and chromatin remodeling. Two, eight and three gene functions are required, respectively, for ATP metabolism, protein trafficking and cell-wall morphogenesis, although six genes are implicated in RNA metabolism. Each one is required for protein synthesis and mitochondrial function.
None of the mutants showed the same mutation, although distinct mutations were found in the identical genes (ssp1+, prp4+, dhp1+; see Table 2). Micrographs of DAPI-stained mutant cells in the +N and -N media are shown in supplementary material Figs S1 and S2, respectively. After 3 days at 37°C, certain strains had a DNA content much less than the content of 1C, suggesting that chromosome DNA was degraded (indicated by + or ++ in Table 2). The presence of degraded DNA was indicative of cell breakage during the quiescence-maintenance stage, consistent with the fact that disrupted cells (seen as brightly fluorescent cells stained with DAPI) were abundant in these mutants.
Cell division cycle genes are also required for quiescence
It was unexpected for quiescent cdc2 and cdc13 mutants to show a loss of viability, even only moderately (29-35%). We found that the requirement of Cdc2 was allele-specific. Only cdc2-974 (I35N) lost viability; the other cdc2 alleles (cdc2-33, -3w) maintained viability and formed round cells in the -N medium. A mutant of the positive regulator of Cdc2, cdc25-22, also produced small spherical cells in the -N medium (data not shown). Cdc2-Cdc13 kinase might therefore be required for certain specific roles in quiescence. Ssp1 controls the G2-M progression, osmo- and salt-stress response, and localization of actin in the +N medium (Matsusaka et al., 1995; Rupes et al., 1999). The ssp1 mutants also showed the defect in cell shape in the -N medium. Ssp1 and Cdc2-Cdc13 might cooperate with Sty1-Wis1 in changing cell shape upon nitrogen starvation. All of the ssp1, wis1 and sty1 mutant alleles tested were rod-shaped in the quiescent culture medium.
Cohesion and chromatin remodeling proteins
The loss of viability of mis4-450 and mis16-33 in quiescence was also unexpected because these mutations were, respectively, defective in establishing sister-chromatid cohesion during S-phase and in recruiting a CENP-A homolog, centromeric histone H3 (Cnp1), to the centromere (Hayashi et al., 2004). Thus, we should consider possible non-cohesion and non-centromeric roles in quiescence for Mis4 and Mis16, respectively. Mis4 might be implicated in transcriptional regulation (Dorsett et al., 2005) in quiescence. The human homolog of Mis4 is the causal gene for Cornelia de Lange syndrome (Dorsett, 2007). Mis16 (also known as RbAp46) could play multiple roles through interacting with histone H3, besides its role in histone recruitment to centromere.
Two products that were identified are essential enzymes involved in the production of nucleotide triphosphates. Pyruvate kinase (Pyk1) catalyses the final ATP-producing step in glycolysis (the conversion of phosphoenolpyruvate to pyruvate with concomitant phosphorylation of ADP to ATP), while presumed CTP synthase, SPAC10F6.03c (designated Cts1), catalyzes the ATP-dependent amination of UTP to CTP with either L-glutamine or ammonia as the source of nitrogen. The substitution mutation D215N of pyk1-67 occurs in the conserved pyruvate kinase domain.
Protein trafficking and cell wall biogenesis
Six mutants (ypt1, apl2, hrf1, sec17, vps11 and ptb1) were linked to the pathway of protein trafficking and vesicle fusion, and two others (hcs1 and fps1) were related to lipid biosynthesis, leading to the activation of protein trafficking. Ypt1 (also known as RAB1), a small G-protein, is the target of C-terminal lipid modification. Three key vesicle transport proteins, Apl2, Hrf1 and Sec17, were found: Apl2 is similar to adaptin (also known as AP-1) that binds to clathrin (Le Borgne and Hoflack, 1998), Hrf1 and Sec17 are orthologs of COPII (Barlowe et al., 1994) and α-SNAP (Clary et al., 1990), respectively. Apl2 and Hrf1 are required for the formation of coated vesicles, but Sec17 together with Ypt1 and Vps11 acts in the fusion of vesicles. Three enzymes implicated in protein trafficking were also found: Hcs1 and Fps1 (HMG-CoA synthase and geranylgeranyltransferase, respectively) are enzymes that lead to C-terminal prenylation of Ypt1 and Ypt5, and Ptb1 is the putative β-subunit of geranylgeranyltransferase II that catalyzes the transfer of a geranyl-geranyl moiety from geranyl-geranyl pyrophosphate to small G-proteins having the C-terminal -XCC (Ypt1) or -XCXC (Ypt5), where both cysteines (C) might become modified. Lipid-modified G-proteins can be bound to the membranes of the ER and Golgi complex. All of the proteins described above are required for active protein trafficking and are functionally linked as depicted in Fig. 5, strongly suggesting that protein trafficking is vital in the maintenance of quiescence.
Three mutants, SPAC4F10.10c (designated mnn9), pps1 and SPBC30D10.17 (designated smi1), are defective in cell wall morphogenesis. Indeed, the cell integrity of these mutants was disrupted; cells were broken, shrunken and/or showed swelling at 37°C in -N medium, probably owing to the fragile nature of the cell wall.
Six mutant genes required for the maintenance of quiescence are involved in nuclear RNA metabolism. Rpc40 and Fcp1 (Archambault et al., 1997; Hausmann et al., 2004) are engaged in the regulation of RNA polymerases; Smd3 and Prp4 in RNA splicing; Dhp1 in transcription termination (Luo et al., 2006); and SPAC20G8.09c (designated Nat10) in ribosome biogenesis. A particularly severe loss of viability was caused by fcp1 and smd3 mutations. Fcp1 is a protein phosphatase acting on the CTD (C-terminal domain) of pol II, and the analysis of its mutant phenotype is described below. Smd3 is bound to the spliceosome complex.
Dhp1 is 5′-3′ exoribonuclease similar to S. cerevisiae RAT1 and human XRN1, and is required for the termination of transcription. Curiously, Dhp1 is also required for chromosome segregation in the +N medium (Shobuike et al., 2001), and dhp1 mutants isolated in the present study showed segregation defects (supplementary material Fig. S1).
Control of quiescence by Pol II CTD phosphatase
The mutation site (R223K) of fcp1-452 resides at one of the highly conserved seven residues (Hausmann and Shuman, 2003) in the FCPH domain that is essential for CTD phosphatase activity in vitro. We therefore employed this strain to examine the mutant phenotype of Fcp1 phosphatase on the transcriptomic patterns between proliferation and quiescence. Mutant cells increased their cell number normally at 26°C but not at 36°C in the +N medium. In the -N medium, mutant cells normally entered quiescence and kept 100% viability at 26°C; however, at 37°C viability decreased to 65% after 1 day and was down to 2% after 3 days. We undertook a phenotypic analysis in the -N medium after 1 day at 37°C, at which time mutant cells showed abnormalities but they were able to regain viability if shifted back to 26°C. Both wild-type and mutant Fcp1-GFP proteins in the -N medium were enriched in the nuclear chromatin region (Fig. 6A), and their levels estimated by immunoblot were in a roughly equal amount (Fig. 6B, the band indicated by the arrowhead).
Transcriptomic comparisons were performed between wild-type and fcp1-452 cells in the +N and -N media at 26 and 37°C. Transcriptomic results derived from eight culture conditions were reproducible; the details were deposited in the Gene Expression Omnibus database (GEO#GSE14319) and are shown in Fig. 6C (green and red represent the decrease and increase in expression, respectively; white, no change; see legend for further explanation). In the +N medium that permits proliferation, the transcriptomic profiles between the wild-type cells and the fcp1 mutant taken after 4 hours were very similar at 26°C, but became distinct at the restrictive temperature, 37°C. The data indicated that 249 (48) and 106 (14) genes showed respective decreases and increases in expression ≥ twofold (≥ fourfold) in the fcp1 mutant cells (Table 3; the number of genes showing ≥ fourfold changes are shown). In the quiescent cells, the transcriptomic profiles of wild-type and fcp1 mutant were similar at 26°C, but greatly different at 37°C. There were respectively, 647 (117) and 668 (142) transcripts showing the ≥ twofold (≥ fourfold) decrease or increase in levels at 37°C. Of the entire ∼5000 S. pombe genes, 26% changed transcriptional level by ≥ twofold in quiescent cells (7% in proliferative cells), suggesting that the influence of fcp1-452 might be stronger and/or broader in quiescence than in proliferation. Control transcriptomic results obtained here were basically consistent with previous results (Shimanuki et al., 2007).
A demonstration of the gene functions affected in fcp1-452 indicated them to be surprisingly selective. Many of the transcripts that decreased or increased in the mutant cells direct the synthesis of proteins involved in stress responses, or in trafficking and transport (identified because they contained sequences for signal peptide and trans-membrane domains). In addition, transcripts of cell-cycle-regulated genes and mitochondrial genes were significantly altered. Table 3 shows a summary of those transcripts displaying a fourfold decrease or increase in their levels (details of the gene names, transcript intensities and fold-changes are shown in supplementary material Tables S1-S3). As an example, among 14 transcripts showing ≥ fourfold increase in fcp1-452 in the +N medium at 37°C, the great majority (11 out of 14, 79%) belonged to stress-response genes, such as the heat shock proteins (Hsp16 and Ssa1) (supplementary material Tables S1 and S3). Among 142 genes showing increased expression in the -N medium at 37°C, 45 (32%) were stress-responsive. Representatives include oxidative-stress-response proteins, producing H2O2, and cadmium-responsive proteins (supplementary material Table S3). Note that the genome of S. pombe contains ∼700 stress-responsive genes. Remarkably, these ≥ fourfold changes in transcript levels in the four different categories defined in Fig. 6D are scarcely overlapped; only eight overlapped among 305 genes.
The changes in levels of cell-cycle-regulated transcripts (classified into four groups, clusters 1-4) (Rustici et al., 2004) were of considerable interest. Among ∼380 cell-cycle-regulated transcripts, 11 greatly diminished in fcp1-452 cultured in the nutrient +N medium. These all belonged to the cluster 2 that peaked at the anaphase, and they are all implicated in the S-phase control. The transcripts included Cdc18, Cdc22, Cdt1, Cdt2, Cig2 and Ams2, suggesting that Fcp1 might control the cell cycle by supporting fluctuation of the transcription of these genes. By contrast, 14 transcripts that either increased (10) or decreased (4) in quiescent (-N) fcp1-452 cells belonged the diverse clusters 1-4 (supplementary material Table S3). The clustered gene products (that peaked at early mitosis, anaphase, S-phase and G2) are mostly novel proteins. They include membrane proteins, putative SNARE proteins and transcription factors, strongly suggesting that Fcp1 controls quiescence through physiologically unknown aspects of cellular activities. Another notable feature regarding the nuclear chromatin is that gene products, such as Arb1, Chp2, Tas3, Raf1, Set1 and Pht1, that function in chromatin silencing and remodeling were strongly affected in quiescent fcp1-452 mutant cells.
Many transcripts encoding membrane and mitochondrial proteins that contained trans-membrane (TM) domains and signal peptides were also dramatically altered in fcp1-452. Note that the total number of proteins with a signal sequence, with a TM domain or located in mitochondria are, respectively, 512, 907 and 732 in the genome of S. pombe. In the nutrient +N medium, ten out of 48 decreased genes encoded proteins with the TM domains (seven of them are amine transporters), and 24 out of 48 decreased genes coded for proteins that localize at membranous organelles (signal peptide, TM, mitochondria). Indeed, electron micrographs of fcp1-452 in the +N medium at 37°C for 4 hours revealed diminished vacuoles and scarce ER-like structures (Fig. 7B) in comparison with mutant cells growing at 26°C (Fig. 7A) or wild-type cells grown at 37°C (data not shown). After 24 hours at 37°C in the quiescent -N medium, however, 41 genes encoding proteins with the TM domains and/or signal peptides were among the 142 showing increased transcripts. Consistently, membranous structures greatly increased in quiescent fcp1-452 in the -N medium at 37°C. Piles of membranous structures were seen in the vicinity of the plasma membrane or even within the nucleus (Fig. 7D). At 26°C, mutant cells looked identical to quiescent wild-type cells (Fig. 7C). Thus, Fcp1 CTD phosphatase seems to be a key transcriptional protein phosphatase that distinctly controls proliferation and quiescence, and shows remarkable selectivity for increasing and decreasing transcripts.
Taken together, the body of data we have assembled leads us to conclude that S. pombe provides an exciting experimental system for identifying genes that are essential for both proliferation and quiescence. Strikingly, only about one quarter (164) of 610 strains defective in proliferation are also defective in quiescence. Assuming that about 20% of the 5000 or so genes in the S. pombe genome are essential for proliferation, some 250 genes might be core or `super-housekeeping' genes that are necessary in both proliferation and quiescence. Although 33 mutant genes were successfully identified using 42 strains, screening of mutants defective in super-housekeeping genes is not complete.
The number (250) of these super-housekeeping genes might be an underestimate, however, as the quiescent cells were heat-resistant (Su et al., 1996), and ts proteins were possibly protected in quiescence by various compounds, such as trehalose (Inoue and Shimoda, 1981; Wiemken, 1990). Alternatively, some quiescence-maintenance genes might be of a type dependent on `longevity' and these become essential only after a long period of quiescence. In fact, the deletion mutant Δklf1 (Klf1, a transcription factor that is non-essential in the +N medium) lost cell viability after 2 weeks in quiescence (Shimanuki et al., 2007).
All of the 33 core genes identified in this study are conserved from fungi to mammals. Their functions range from the stress-responsive-signaling MAPKK-MAPK pathway to protein trafficking, actin-interacting endosome formation, lipid formation and ATP biosynthesis through sugar catabolism, RNA transcription, splicing, processing and protein translation metabolism, chromatin remodeling and dynamics, and cell-wall morphogenesis. Strikingly, quiescence-defective mutants displayed specific phenotypes in cell shape, DNA content, the levels of mitotic cyclin and CDK inhibitor, organelle formation and distribution, and osmo-regulation. The use of electron microscopy as well as light microscopy is essential for characterizing the basic phenotypes of individual mutants. These core genes are essential in the -N medium for cells to adapt to the change in nutritional environment, to metabolize external and internal sources for energy, to protect cells from various stresses and to keep their static and dynamic cellular structures for a long time. About 120 strains that showed a significant decrease of viability still remain to be investigated. As identification of the mutant genes by regular gene cloning and genetic mapping has been time-consuming, we have initiated an attempt to sequence the whole genome of mutant strains. This might greatly accelerate the process of identifying mutant genes.
Fcp1 plays a main role for the quiescence state
Our results indicate that the balance between proliferation and quiescence is controlled by RNA polymerase II through the regulation of the state of its CTD phosphorylation by Fcp1 phosphatase. Fcp1 plays a major role in transcriptional support for the quiescent state because its mutant lacks the principal intracellular features of quiescence. It is surprising that a single fcp1 mutation causes very large changes in the levels of mostly non-overlapping transcripts in both proliferating and quiescence cells at the restrictive temperature. It remains to be investigated how specific transcripts implicated in DNA replication, membrane transport, trafficking, stress responses, mitochondrial structure and function, and chromatin remodeling are regulated differentially in proliferation and quiescence. Because regulation of CTD phosphorylation of RNA polymerase II by cyclin-dependent kinases (e.g. CDK9-cyclin T1) contributes greatly to the regulation of gene expression between normal (quiescent) and tumor cells in humans (Wang and Fischer, 2008), and S. pombe seems to have the counterparts of these genes products, it becomes of interest to know the extent to which the global events we describe in fission yeast are conserved.
Materials and Methods
Strains and yeast cell culture
A collection of ts strains made by random mutagenesis was used (Hayashi et al., 2004). The strains grew at 26°C but not at 36°C in the nutrient medium (+N) (Mitchison, 1970). For many strains, plasmids that suppressed the ts phenotype were obtained by transformation, and plasmid-inserted genomic DNAs were partially sequenced (Hayashi et al., 2004). Using information from multiple suppresser plasmids obtained from many mutants, definitive identification of the mutant genes was done by determining the mutation sites using DNA sequencing and linkage analysis using tetrad dissection. To characterize the phenotypes in nitrogen-starved medium (-N), the auxotrophic marker (leu1-32) and the nuclear GFP marker (chromosomally integrated GFP-tagged SPAC664.02c/eat1+ gene) present in the original strains (both of which were inhibitory to the recovery process from the quiescent to the proliferate state) were removed by genetic crossing with the wild type. 610 Ts- Leu+ GFP- segregants were obtained. Their inability to form colonies at 36°C in the +N medium was verified. The 12 strains that failed to enter quiescence under nitrogen starvation were difficult to study as they mostly failed to mate owing to the lethality under nitrogen deficiency, which is required for normal meiosis: only wis1, end4 and wsp1 (indicated by + in Table 2) were confirmed by tetrad dissection for their tight linkage between the cloned gene and the ts allele.
Nitrogen starvation, Cell number and viability measurements and flow cytometry
For 610 mutant strains having only the single ts mutation, the cell number and viability were assayed for 0 and 24 hours in the -N medium. The condition for nitrogen starvation was previously described (Su et al., 1996). Briefly, cells exponentially grown in +N liquid medium to a density of 2×106 cells/ml at 26°C were harvested by vacuum filtration using a nitrocellulose membrane (0.45 μm pore size), washed in -N liquid medium twice on the membrane, and then re-suspended in -N liquid medium. The culture was incubated for 24 hours at 26°C in a water bath with shaking. Cell number was measured using a Sysmex CDA-500 cell counter. Cell viability was measured by diluting the culture in order to plate 300 cells on a YPD plate, incubating the plate with the cells at 26°C for 6 days, and counting the number of colonies formed. Viability was expressed as the percentage of the number of formed colonies against the total number of plated cells. To measure the DNA content, flow cytometry was done (Costello et al., 1986) using FACSCalibur (Becton Dickinson).
For DAPI staining, cells were fixed with 2% glutaraldehyde for 10 minutes on ice, washed three times with phosphate buffered saline (PBS), and observed under a fluorescence microscope after staining with DAPI (25 μg/ml). For vacuole staining, cells were labeled with N-(3-triethylammoniumpropyl)-4-(6-(4-(diethylamino)phenyl) hexatrienyl) pyridinium dibromide (FM 4-64; Invitrogen) (Higgs and Pollard, 2001) as described by Gachet and coworkers (Gachet et al., 2005). For actin staining, cells were fixed with 30% paraformaldehyde by shaking for 1 hour at room temperature. Cells were washed three times with PEM buffer (100 mM Pipes, 1 mM EGTA, 1 mM MgSO4), re-suspended in PEM+1% triton, and washed three times with PEM. Then, 1.5 μM rhodamin-phalloidin (Invitrogen) (Wieland, 1986) was added and incubated on a rotator at 4°C, for 30 minutes, in the dark, then visualized. For lipid droplet staining, cells were washed with deionized H2O twice, 1 μg/ml Nile red (TCI) was added, and the cells visualized.
Movies were taken using a DeltaVision Spectris restoration microscope (Applied Precision LLC) with a CH350L CCD camera (Photometrics). Cells were placed on a glass-bottomed culture dish (MatTek) and covered with a slip of appropriate agar solid medium. Images were taken at 10-minute intervals for 15 or 20 hours and edited by QuickTime 7 Pro software. The playback rate for each movie is 60 minutes/second.
Transmission electron microscopy
Cells were fixed with 2% glutaraldehyde in 100 mM phosphate buffer pH 7.2 for 2 hours at 26°C, post-fixed with 2% potassium permanganate overnight at 4°C, and embedded in Epon812 (TAAB). Ultra-thin sections were stained in 2% uranyl acetate and Raynold's lead citrate, and viewed with a TEM JEM1230R (JEOL) operating at 100 kV.
Total protein was extracted by the trichloroacetic acid precipitation method. The same amounts of protein were loaded into each lane. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed using 12% polyacrylamide gels, and samples were blotted on to nitrocellulose membranes. Anti-Rum1 (a gift from Sergio Moreno, Universidad de Salamanca, Salamanca, Spain), anti-Cig2 (a gift from Hiroyuki Yamano, Marie Curie Research Institute, UK), anti-Cdc13, anti-PSTAIR (a gift from Yoshitaka Nagahama, National Institute for Basic Biology, Okazaki, Japan), anti-α-tubulin (a gift from Keith Gull, University of Oxford, Oxford, UK) and anti-GFP were used as the primary antibodies. Horseradish-peroxidase-conjugated secondary antibodies and an ECL chemiluminescence system (Amersham) were used to amplify signal expression. An LAS3000 (Fuji Film) was used for signal detection.
For total RNA extraction, 100 ml of cell culture with 5 3×106 cells/ml was centrifuged and the pellet was immediately frozen using liquid nitrogen. From the pellet, total RNA was isolated by acid phenol methods described by Lyne and coworkers (Lyne et al., 2003), and the polyA-RNA purified by Oligotex-dT30<super>mRNA purification kit (Takara). The polyA-RNA (>2 mg) was reverse-transcribed to cDNA, then transcribed again to label with IVT using Oligo dT primer (GeneChip reagents; Affymetrix). The labeled targets were added to the GeneChip Yeast 2.0 (Affymetrix) and hybridized for 16 hours at 45°C with rotating. The GeneChips after hybridization were washed using the Fluidics Station 450 (Affymetrix). After washing, the GeneChip Scanner 3000 (Affymetrix) was used for scanning, and the acquired data was analyzed by GeneSpring 7.3.1 (Agilent). Measured fluorescent intensity was corrected by RMA (robust multi-array) normalization.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/122/9/1418/DC1
We gratefully acknowledge Kazuhiro Shiozaki, Sergio Moreno, Akira Ishihama (Hosei University, Tokyo, Japan), Yoshitaka Nagahama, Hiroyuki Yamano, Keith Gull and Minoru Yoshida (RIKEN, Japan) for strains, antibodies and plasmids. We thank Anthony Hyman (Max Planck Institute of Molecular Cell Biology, Dresden, Germany) for comments on the manuscript. This work was supported by a CREST research grant from the Japan Science and Technology Corporation (JST) and an Initial Research Grant from the Okinawa Institute of Science and Technology Promotion Corporation (OISTPC).
- Accepted January 6, 2009.
- © The Company of Biologists Limited 2009