CDK actively contributes to establishment of the stationary phase state in fission yeast

Quiescence is the non-dividing but proliferation-competent state observed in all living organisms (Broach, 2012; De Virgilio, 2012; Gray et al., 2004; Herman, 2002; Marescal and Cheeseman, 2020; O'Farrell, 2011; Rittershaus et al., 2013; Roche et al., 2017; Sagot and Laporte, 2019; Sun and Buttitta, 2017; Sun and Gresham, 2021; Valcourt et al., 2012; van Velthoven and Rando, 2019; Velappan et al., 2017; Yanagida, 2009). Cells enter quiescence in response to various extracellular stimuli, including growth inhibitory signals and nutrient starvation, and quiescent cells restart cell division upon receiving appropriate stimuli or restoration of growth-suitable environments. In multicellular organisms, various non-differentiated stem cells and many differentiated cells including dermal fibroblasts, lymphocytes, and hepatocytes are in the quiescent state. A large majority of microorganisms in nature are in the quiescent state (Cole, 1999; Rittershaus et al., 2013). Cells of unicellular organisms enter the quiescent state, termed the ‘stationary phase’, upon nutrient exhaustion (Gray et al., 2004; Yanagida, 2009; Broach, 2012; Sagot and Laporte, 2019; Sun and Gresham, 2021). Quiescent cells are characterized by the substantial repression of their metabolic activity and their resistance to various stresses, and such characteristics are crucial for organisms to survive in stressful and/or nutrient-limited environments. In addition, the drug resistance of quiescent cancer cells or pathogens represents a major obstacle for treatment of several tumors and infectious diseases (Bojsen et al., 2017; Borst, 2012; Rittershaus et al., 2013). Therefore, understanding the molecular basis of quiescence is both biologically and clinically important.

Studies of stationary phase in yeast have demonstrated that the stationary phase state is markedly distinct from that of cycling cells (Gray et al., 2004; Klosinska et al., 2011; Rittershaus et al., 2013; Werner-Washburne et al., 1993). In stationary phase cells, numerous distinct types of granules, which contain mRNAs, signaling molecules, metabolic enzymes and/or proteasomes, form in the cytoplasm, (Brengues et al., 2005; Laporte et al., 2008, 2011; Narayanaswamy et al., 2009; Nostramo et al., 2016; Shah et al., 2014). Dynamic cytoskeletal filaments turn into immobile, non-dynamic filaments or aggregates (Laporte et al., 2013, 2015; Sagot et al., 2006), whereas chromosomes become condensed with a decrease in their Brownian-like fluctuation (hereafter called chromosome fluctuation) and change their intranuclear positioning, and the cytoplasm becomes solidified (Heimlicher et al., 2019; Joyner et al., 2016; Laporte et al., 2016; Piñon, 1978; Rutledge et al., 2015).

Stationary phase establishment is coupled with the cell cycle. Yeast cells become smaller in size by the modulation of the preceding cell cycle (Gray et al., 2004; Yanagida et al., 2011). In addition, although glucose starvation induces entry into stationary phase, acute glucose depletion, which immediately terminates the cell cycle, fails to induce some stationary phase-associated changes. For instance, after acute glucose depletion, budding yeast cells do not show resistance to cell wall-digesting enzymes, a common stationary phase characteristic, whereas in fission yeast, cell size reduction does not occur (Klosinska et al., 2011; Yanagida, 2009). Given that the decreased concentration of glucose in the growth medium induces some stationary phase-associated phenotypes in cycling cells (Kelkar and Martin, 2015; Klosinska et al., 2011; Pluskal et al., 2011), the gradual decline in environmental nutrients during cell proliferation probably induces modulation of the preceding cell cycle, which thus contributes to the stationary phase establishment.

Given the central role of cyclin-dependent kinase (CDK) in the cell cycle, CDK must actively contribute to cell cycle modulation in order to establish the stationary phase state. Indeed, in Schizosaccharomyces pombe, several signaling kinases, such as TOR, PKA, and MAPKs, which are responsible for the establishment of stationary phase (De Virgilio, 2012; Gray et al., 2004; Herman, 2002; Yanagida, 2009), regulate the sole S. pombe CDK, Cdc2, to reduce cell size in response to nutrient limitation (Kelkar and Martin, 2015; Petersen and Nurse, 2007; Yanagida et al., 2011). In addition, CDK and its regulators are important for survival of stationary phase S. pombe cells under nutrient-starved conditions (Masuda et al., 2016). However, given that cell cycle arrest is a hallmark of quiescence, many previous studies on nutrient-induced quiescence in yeast have focused on the role of CDK inhibition in quiescence establishment (Moreno-Torres et al., 2015; Simanis and Nurse, 1986; Werner-Washburne et al., 1993; Yanagida, 2009; Zinzalla et al., 2007). Therefore, the active contribution of CDK to the stationary phase establishment remains largely unstudied. In this study, we examine the stationary phase state of S. pombe and the contribution of CDK to its establishment, and show that CDK actively contributes to the establishment of the stationary phase state.

To understand how CDK functions in the establishment of stationary phase in S. pombe, we first characterized the stationary phase state by examining the morphology and size of the nucleus and chromosomes, which dynamically change during the cell cycle under CDK control and become smaller or condensed in quiescent cells of various organisms (Evertts et al., 2013; Faire et al., 2015; Guidi et al., 2015; Gupta et al., 2012; Piñon, 1978; Rawlings et al., 2011; Rutledge et al., 2015; Sajiki et al., 2009; Schäfer et al., 2008; Su et al., 1996; Swygert et al., 2019). A GFP-tagged N-terminal portion of cytochrome c reductase (Fig. 1A) (Ding et al., 2000; Nambu et al., 2022) and an mCherry-tagged histone H2A were used to visualize the nuclear envelope and chromosomes, respectively, and nuclear and chromosomal size was analyzed by measuring the area of the nucleus and of the chromosome-occupying space in the two-dimensional images or by calculating nuclear volume from the nuclear image. In this study, cells were grown in YES-rich medium and not minimal medium, as cells enter the stationary phase exclusively from G2 in the YES-rich medium (Fig. S1A), but from both G1 and G2 in the minimal medium (Yanagida, 2009). In addition, for evaluation of both exponentially growing (log phase) and stationary phase cells, only cells containing a single nucleus with non-condensed chromosomes, which are in G2 (Hayles and Nurse, 2018), were analyzed, to exclude cell cycle-associated nuclear and chromosomal changes.

In log phase, the majority of cells contained a single nucleus with non-condensed chromosomes (78.0%, n=118). In these cells, the nucleus was located at the center of the cell, and the chromosomes occupied a hemispherical intranuclear space (Fig. 1A), as reported previously (Toda et al., 1981). In stationary phase, cells were shorter and smaller in volume than in log phase (Fig. S1B,C,H,I), and most of them contained a single nucleus (91.3%, n=103), in which the chromosomes occupied a hemispherical space (Fig. 1A). However, area or volume analysis of the nucleus and the chromosome-occupying space showed that nuclear and chromosomal sizes were significantly smaller (Fig. 1B–D; Fig. S1D–F), as seen in quiescence induced by nitrogen starvation (Sajiki et al., 2009; Su et al., 1996). The nuclear and chromosomal sizes continued to decrease for at least 1 day after entry into stationary phase, and the nuclear size became less than one third of the original size. Furthermore, the nuclear to cell volume ratio (N/C ratio), which has been reported to be constant irrespective of cell size (Neumann and Nurse, 2007), was significantly smaller (Fig. 1E; Fig. S1G). Thus, the nuclear size reduction observed in stationary phase cells did not originate simply from the reduction in cell size.

In glucose-rich medium, glucose exhaustion is thought to induce entry into stationary phase. To explore the contribution of the preceding cell cycle to establishment of the stationary phase state, we next examined the cell state induced by acute glucose depletion, which causes immediate cell cycle arrest (hereafter, cells responding to acute glucose depletion and their state are called GD cells and GD state, respectively) (Yanagida, 2009). Log phase cells were transferred to YES medium lacking glucose (YES-Glc) and their nuclear and chromosomal sizes were examined. The immediate cell cycle arrest induced by glucose depletion did not cause cell size reduction (Fig. 1A; Fig. S1B,C,H,I). In these cells, the nuclear and chromosomal sizes became significantly smaller than in the log phase cells, but not as small as those in stationary phase cells (Fig. 1B–D; Fig. S1D–F). The N/C ratio was also reduced but not significantly different from that of stationary phase cells (Fig. 1E; Fig. S1G), suggesting that the observed nuclear size differences from stationary phase originate from the cell size differences. Furthermore, incubation of stationary phase cells in fresh YES-Glc medium immediately after entry into the stationary phase did not cause a further reduction in nuclear size or N/C ratio (Fig. 1B,D,E, Stat, 1 d, YES-Glc; Fig. S1D,F,G). This indicates that stationary phase cells can sustain nuclear size by utilizing nutrients.

The fact that S. pombe cells enter stationary phase exclusively from G2 prompted us to investigate the effect of entry into stationary phase on the association of sister chromatids. We examined the association state of three chromosomal loci on distinct chromosomes visualized using a GFP-tagged lac repressor (Fig. 2A) (Ding et al., 2004; Nabeshima et al., 1998; Yamamoto and Hiraoka, 2003). The nda3 β-tubulin gene promoter was used for expression of the lac repressor, thus enabling chromosome visualization in both stationary and log phase cells. We also avoided using the nuclear localization sequence (NLS) to direct the lac repressor to the nucleus, as the NLS was not essential for nuclear localization of the repressor and appeared to obscure changes in the association state of sister chromatids in stationary phase (see below), although the reasons for this are unclear. We simultaneously visualized microtubules to exclude spindle-forming log phase cells that were in the process of sister chromatid separation.

In contrast to the accepted view that sister chromatids are tightly associated with each other along their entire length before anaphase, all examined sister loci were separated in ∼10% of cells in log phase (Fig. 2B,C, Log). This result indicates that sister chromatids occasionally undergo partial separation. After entry into stationary phase, G2 cytoplasmic microtubules converted into one or more short, thick microtubule bundles, as reported previously (Laporte et al., 2015). In these cells, although sister locus separation occurred (Fig. 2B), the frequency with which these separations were observed significantly decreased at all assessed loci (Fig. 2C). This suggests that sister chromatids become more tightly associated with each other in the stationary phase. At 1 day after entry into stationary phase, the separation frequencies of the nhe1 and ade6 loci remained reduced, whereas that of the ade8 locus returned to log phase levels, indicating that the tight association was transient at some loci. Although we cannot completely exclude the possibility that sister locus separation and its repression are artifacts generated by the lacI/lacO chromosome visualization, these results suggest that the chromosomal association state is distinct in stationary phase. By contrast, separation frequencies did not decline in GD cells (Fig. 2B,C, GD), suggesting that acute glucose depletion does not alter the association sate of sister chromatids.

In budding yeast, chromosome loci exhibit Brownian-like fluctuation in the nucleus, and this fluctuation is decreased upon glucose depletion (Joyner et al., 2016). We therefore wished to examine whether chromosome fluctuation is similarly repressed in stationary phase in S. pombe cells. Images of the GFP-visualized loci were captured in three dimensions every 1.5 s and changes in their positioning over time were analyzed. The visualized loci changed their position over time in log phase cells (Fig. 2D, Log), indicating that chromosomes fluctuate as in budding yeast. In addition, the sister loci often underwent transient separation, indicating that separation of the sister loci seen in still images reflects their transient separation. Separation behavior varied among observations; separation was rare in some observations but very frequent in others (Fig. 2D, Log), suggesting that association varies between cells or stages. By contrast, in stationary phase, the visualized sister chromatid loci appeared to be less mobile, and their separation was rare (Fig. 2D, Stat). Plots of the mean square displacement (MSD) of the loci demonstrated that fluctuation mobilities varied among the loci (Fig. 2E; Fig. S2A). However, the fluctuation mobilities of all examined loci were significantly reduced in stationary phase compared to log phase, as demonstrated by a downshift of the stationary phase plots and an almost complete lack of an overlap of their 95% confidence intervals with those of the log phase plots during the 30 s observation (Fig. 2E). These results indicate that chromosome fluctuation is repressed in the stationary phase. Furthermore, the MSD plots of the loci in GD cells and their 95% confidence intervals showed that chromosome fluctuation was significantly repressed to the same level as in stationary phase (Fig. 2E). Considering the spatial scales of the fluctuation, this fluctuation repression cannot be attributed simply to spatial confinement. This conclusion is supported by simple numerical simulations of a particle diffusing in confined spherical spaces, which reflect the observed nuclear sizes in log and stationary phases (Fig. S2B). These results indicate that glucose depletion causes repression of chromosome mobility.

We next examined the intracellular localization of Cdc2, the sole CDK in S. pombe. Consistent with a previous report (Decottignies et al., 2001), during log phase, Cdc2, together with Cdc13 (the S. pombe B-type cyclin), was evenly distributed across the nucleus of a mono-nuclear cell and frequently accumulated at a point, which corresponds to the SPB (Fig. 3A–C, Log). However, upon entering stationary phase, Cdc2 accumulated at the periphery of the chromosome-unoccupied, nucleolar region with a convex localization pattern (Fig. 3A–C, Stat). At 1 day after entry into stationary phase, Cdc2 became localized at two distinct regions in approximately half of cells analyzed (54.2%, n=72) (Fig. 3A, Stat 1d). Nucleolar Cdc2 was colocalized with Cdc13 (Fig. 3C, Stat), suggesting that Cdc2 localizes to the nucleolus in a complex with a B-type cyclin. Cdc2 also colocalized with Clp1 phosphatase, which is localized in the nucleolus during most of the cell cycle and eliminates CDK-dependent phosphorylation (Fig. 3C, Stat 1 d), although Cdc2 nucleolar localization was not dependent on Clp1 (data not shown). By contrast, in stationary phase, Cdc2 was not colocalized with the conserved nucleolar proteins, Gar1 and Gar2 (Girard et al., 1993; Gulli et al., 1995), which accumulated in the central nucleolar region (Fig. 3C, Stat 1 d). These observations indicate that the stationary phase nucleolus has at least two distinct regions, a Gar1- and Gar2-enriched central region and a Cdc2- and Clp1-enriched peripheral region. The localization of these proteins was largely similar in GD cells (Fig. 3A–C, GD), although Gar1 and Gar2 often formed one or more extra dots in the chromosomal region (Fig. 3C, GD 4 h, arrowheads). Thus, glucose depletion is the cause of Cdc2 nucleolar accumulation.

The nucleolar localization pattern of Cdc13, Clp1 and Gar2 seen in stationary phase and GD cells was reminiscent of that seen in cells exposed to acute heat stress (Gallardo et al., 2020). We found that Cdc2 also frequently accumulated at the periphery of the nucleolus in cells exposed to 42°C for 20 min (75.8%, n=95) (Fig. S3). Furthermore, we found that incubation of cells for 2 h in YES medium containing 0.5 M Mg²⁺ resulted in a similar nucleolar accumulation of Cdc2 together with the cell cycle arrest, albeit with a reduced frequency (29.3%, n=150) (Fig. S3). These results suggest that Cdc2 nucleolar accumulation is a common cellular response to environmental stresses.

Our analysis of the GD state indicates that acute glucose depletion induces a state that is similar, but not completely the same, as the stationary phase state. We reasoned that if the purpose of the stationary phase state were to promote cell survival under nutrient-poor conditions, stationary phase cells should live longer in glucose-free medium than GD cells. To test this hypothesis, stationary phase and GD cells were incubated in YES-Glc medium and their viability over time was compared.

Without medium change, stationary phase cells mostly died within a few days, as reported previously (Fig. 4A) (Zuin et al., 2010). However, when they were transferred to fresh YES-Glc medium, they lived for a longer period, and dilution of the cell density further extended their survival period (Fig. 4A). Medium change- and dilution-dependent extension of cell survival probably results from an increase in the available essential nutrients for each cell given that addition of nutrients other than glucose into the stationary phase culture extended the survival period (Fig. 4B). Thus, stationary phase cells utilize nutrients other than glucose to survive under glucose-depleted conditions. As for stationary phase cells, GD cells lived for a longer period at lower cell density in YES-Glc medium (Fig. 4C). However, their survival time was shorter than that of stationary phase cells at the same cell density (Fig. 4A,C). These results indicate that GD cells lose viability more rapidly than stationary phase cells under glucose-depleted conditions and suggest that GD cells are less resistant to glucose-depleted environments or more rapidly consume essential nutrients compared with stationary phase cells.

Having characterized the stationary phase state, we next investigated the contribution of Cdc2 activity to its establishment by using a cdc2-L7 temperature-sensitive mutant (Moreno et al., 1989). We found that although cdc2-L7 cells did not grow at the higher temperature (34°C), they grew at 32°C at a growth rate that was comparable to that of wild-type (WT) cells (Fig. 5A,B). However, at 32°C, cdc2-L7 cells entered stationary phase at lower cell density than WT cells (Fig. 5B,C). Sizes of the cell, the nucleus and the chromosome-occupying space decreased, but were significantly larger compared with those in WT cells (Fig. 5D–F; Figs S4 and S5A). These differences were already evident in log phase and, despite the larger cell and nuclear sizes, there was no significant difference in the N/C ratio between log and stationary phases (Fig. 5G), suggesting that the larger nuclear size results from the larger cell size. Furthermore, the log phase chromosomes more actively fluctuated in cdc2-L7 cells than in WT cells, and despite a repression in chromosome fluctuation, fluctuation was still elevated in the stationary phase, as demonstrated by an upward shift of the cdc2-L7 MSD plot of the ade6 locus and a complete lack of an overlap of their 95% confidence intervals with those of the WT plot (Fig. 5H). The separation frequencies of sister chromatid loci in cdc2-L7 cells significantly increased in stationary phase compared with WT cells (Fig. 5I). These differences were not evident at 25°C (Figs S4, S5A,B,D–G). Therefore, in the mutant cells at the boundary temperature, Cdc2 activity was reduced to a level that did not inhibit cell cycle progression but caused changes in log and stationary phase states. These results indicate that Cdc2 activity contributes to the establishment of the stationary phase state and suggest that the preceding cell cycle is important for a proper stationary phase state.

We next examined contribution of Cdc2 to the GD state establishment. After acute glucose depletion, nuclear and chromosomal sizes were significantly larger in cdc2-L7 cells than in WT cells at 32°C, but not at 25°C (Fig. 5E,F; Figs S4, S5C–E,G). However, the N/C ratio was not significantly different (Fig. 5G), suggesting that the larger nuclear size in cdc2-L7 cells results from their larger cell size. The separation frequencies of sister chromatids were not significantly different from those of WT cells in the cdc2-L7 cells (Fig. 5I; Figs S4, S5H). To fully elucidate the contribution of Cdc2 to nuclear size reduction, we inhibited Cdc2 activity upon glucose depletion using an analog-sensitive Cdc2 variant (Cdc2-as), whose kinase activity can be inhibited by the ATP analog 3MB-PP1 (Dischinger et al., 2008). Cells containing mCherry-tagged Cdc2-as as the sole Cdc2 were grown, and their Cdc2 activity was inhibited by adding 3MB-PP1 to the culture. After a 10-min treatment, which did not cause any significant changes in nuclear size, cells were transferred to YES-Glc medium containing 3MB-PP1. As a control, cells were treated with DMSO, the solvent for the analog. In the presence of 3MB-PP1, despite some reduction, the nuclear size and the N/C ratio were significantly larger than those seen in the presence of DMSO (Fig. 5J,K; Fig. S4). This result indicates that Cdc2 contributes to nuclear size reduction in response to acute glucose depletion.

We next examined whether Cdc2 activity is required for its nucleolar accumulation. Cdc2 kinase activity is negatively regulated by its phosphorylation at tyrosine 15, and Cdc2 becomes constitutively active and inactive upon replacement of tyrosine 15 with phenylalanine (Cdc2-Y15F) or glutamic acid (Cdc2-Y5E), respectively (Fig. 6A) (Gould and Nurse, 1989). Cdc2 activity is also abolished by replacement of lysine 33 at the ATP-binding site of the catalytic domain with arginine (Cdc2-KD) (Fig. 6A) (Booher and Beach, 1986; Hanks et al., 1988). When introduced in the WT background, all of these Cdc2 variants were localized in the nucleus in log phase (Fig. 6B,C). In the stationary phase, Cdc2-Y15F accumulated in the nucleolus, whereas Cdc2-Y15E or Cdc2-KD often failed to accumulate in the nucleolus and eventually disappeared from the nucleus (Fig. 6B,C). At 1 day after entry into the stationary phase, Cdc2-Y15E mostly failed to localize at intracellular organelles, whereas Cdc2-KD was localized in the cytoplasm in a granular staining pattern (Fig. 6B). Signal intensities of Cdc2 in the nucleolar region relative to those in the chromosomal region confirmed these localization patterns (Fig. S6A–E). These results indicate that Cdc2 activity is necessary for the nucleolar accumulation of Cdc2 in stationary phase.

We next examined the localization of these Cdc2 variants in GD cells. In GD cells, although Cdc2-Y15F successfully accumulated in the nucleolus, Cdc2-Y15E and Cdc2-KD did not (Fig. 6B,C; Fig. S6A–E). Furthermore, upon acute glucose depletion, mCherry-tagged Cdc2-as, which was localized in the nucleus during log phase, accumulated in the nucleolus in the presence of DMSO but was mostly excluded from the nucleolus in the presence of 3MB-PP1 (Fig. 7). These results indicate that Cdc2 activity is required for Cdc2 nucleolar accumulation in GD cells. Finally, in both stationary phase and GD cells, overall protein levels of WT and mutant forms of Cdc2 were similar and remained unchanged under all conditions, indicating that the frequent lack or weakening of signal seen with inactive Cdc2 variants did not result from Cdc2 degradation (Fig. S6F,G).

We next investigated the requirement of Cdc2 activity for survival of stationary phase and GD cells under glucose-depleted conditions. At 32°C, the viability of cdc2-L7 cells was comparable to that of WT cells in log phase but decreased more rapidly than that of WT cells in stationary phase (Fig. 8A). The difference was even more striking when stationary phase cells were transferred to fresh YES-Glc medium (Fig. 8B). The difference in viability was marginal or not significant at 25°C (Fig. 8A,B). These results indicate that Cdc2 activity is required for survival of stationary phase cells under glucose-depleted conditions. We next addressed whether Cdc2 activity is required before and/or after stationary phase entry by temperature shift. No significant differences were observed when cells were incubated at 32°C after entry into stationary phase, whereas viability of cdc2-L7 cells decreased more rapidly than that of WT cells when cells were incubated at 32°C before and during entry into stationary phase (Fig. 8C). However, when temperature was downshifted after stationary phase entry, a less severe reduction in viability was observed (Fig. 8B,C). This probably reflects the lack of cdc2-independent viability reduction at 32°C (13 days after stationary phase entry, the mean±s.e.m. cell viability of WT cells was 93.3±1.3% at 25°C but 67.0±4.2% at 32°C). These results indicate that Cdc2 activity is crucial for cell survival before and/or upon entry into stationary phase, but dispensable after this point.

We then examined the requirement of Cdc2 activity for survival of GD cells. cdc2-L7 GD cells lost viability more rapidly than WT GD cells at 32°C, but not at 25°C (Fig. 8D). Furthermore, viability of cdc2-as GD cells decreased by one-third shortly after transfer to the 3MB-PP-containing YES-Glc medium, although no further decrease was observed, at least for 2 days (Fig. 8E). These results indicate that Cdc2 activity is also crucial for survival of GD cells.

In this study, we have characterized the stationary phase state in S. pombe. By examining the nucleus and chromosomes, we have found several stationary phase-specific cell properties. First, in addition to reduced cell size (Kelkar and Martin, 2015; Yanagida et al., 2011), volumes of the nucleus and chromosome-occupying space are significantly decreased. Second, the transient partial separation of sister chromatids, which occurs in log phase, is repressed. Third, fluctuation of chromosomes is repressed. Finally, Cdc2 accumulates in the nucleolus together with the B-type cyclin Cdc13.

Size reduction of the nucleus and chromosomes and repression of chromosome fluctuation are probably common characteristics of quiescence, as various quiescent cells exhibit similar changes (Evertts et al., 2013; Guidi et al., 2015; Gupta et al., 2012; Joyner et al., 2016; Laporte et al., 2016; Piñon, 1978; Rawlings et al., 2011; Rutledge et al., 2015; Swygert et al., 2019). It is currently unclear whether repression of sister locus separation is common among quiescent cells, as most other organisms enter quiescence from G1. However, it is possible that the transcription process induces sister locus separation, and that transcription silencing, a common characteristic of quiescence, represses their separation. The restoration of sister locus separation seen at the ade8 locus might reflect quiescence-associated transcription that occurs around the ade8 locus after adaptation to nutrient limitation. Likewise, it is unclear whether CDK nucleolar accumulation is a common feature of quiescence. However, nucleolar accumulation of Cdc2 and other regulators is also observed upon exposure to heat or a high concentration of metal ions (Fig. S2) (Gallardo et al., 2020). Furthermore, in vertebrate cells, upon inhibition of rDNA transcription, CDK2, together with various proteins, accumulates in the nucleolus to form a structure called ‘the nucleolar cap’ (Boulon et al., 2010; Ide et al., 2020; Liu et al., 2000; Reynolds et al., 1964; Scheer and Rose, 1984; Shav-Tal et al., 2005). By contrast, in budding yeast, Cdc28, the sole CDK in that organism, is colocalized with ribonucleoprotein-containing cytoplasmic granules during stationary phase (Shah et al., 2014). These results suggest that CDKs become commonly colocalized with the ribonucleoproteins in response to stress. Perhaps, this colocalization is crucial for the regulation of RNA metabolism. These stationary phase-associated changes are probably essential for cells to survive under glucose-depleted conditions, because when the changes are not fully induced by acute glucose depletion or reduced CDK activity, cells lose viability more rapidly than stationary phase cells in YES-Glc medium.

We have shown that the GD state resembles, but is not the same as, the stationary phase state. This result suggests that acute glucose depletion is not sufficient for the establishment of the stationary phase state and is consistent with previous results (Klosinska et al., 2011; Yanagida et al., 2011). Given that acute glucose depletion causes immediate cell cycle arrest, these data support the idea that modulation of the preceding cell cycle is required for the stationary phase establishment. Glucose levels gradually decrease in environments, and a low glucose-concentration induces stationary phase-associated phenotypes in cycling cells (Kelkar and Martin, 2015; Klosinska et al., 2011; Pluskal et al., 2011). In addition, Cdc2 nucleolar accumulation occurs upon entry into stationary phase. Taken together, we speculate that the stationary phase state is established in two steps; first, a gradual decline of environmental glucose induces stationary phase-associated changes in cycling cells, and subsequently, glucose depletion induces changes associated with stationary phase entry (Fig. 8F). However, depletion of nutrients other than glucose induces quiescence (Klosinska et al., 2011) and S. pombe cells secrete compounds that affect cell growth (Sun et al., 2016). Thus, although glucose depletion is a major trigger of stationary phase, we cannot completely exclude the possibility that depletion or decline of other nutrient(s) and/or the presence of unidentified secreted compounds additionally contribute to stationary phase establishment and that the GD state is different from the stationary phase state due to lack of these additional effects.

We have also found that stationary phase cells utilize nutrients in the medium for their survival and that if appropriate nutrients are supplied, their survival period extends. Therefore, the lifespan of a non-dividing cell, termed the chronological lifespan, changes in different nutritional environments. This suggests that the rapid loss of viability described in some of the previously identified chronological lifespan mutants results from impairment in nutrient utilization or rapid nutrient consumption. It will be important to examine survival periods for various mutants under the same, nutrient-unchangeable conditions to distinguish between death from nutrient starvation and that arising from impairment of stationary phase-specific housekeeping systems.

Our analysis of cdc2-L7 mutant demonstrates that CDK is crucial for a proper stationary phase state. In cdc2-L7 cells, the differences were already evident in log phase, and nuclear and chromosome size reduction occurred in stationary phase. These observations suggest that the observed improper stationary phase state originates from the improper state of cycling cells and not from impairment in stationary phase-associated changes. Even if so, it is obvious that CDK-dependent regulation of the preceding cell cycle is crucial for a proper stationary phase state. It is also apparent that CDK induces cell size reduction by modulating the preceding cell cycle (Kelkar and Martin, 2015; Petersen and Nurse, 2007; Yanagida et al., 2011). Furthermore, Cdc2 activity is required for Cdc2 nucleolar accumulation and nuclear size reduction upon acute glucose depletion. All these results support the idea that CDK plays a crucial role in stationary phase establishment before and upon stationary phase entry. Our viability analyses of stationary phase and GD cells also provided evidence to support this hypothesis (Fig. 8A–E).

How does CDK contribute to the stationary phase-associated changes? Nuclear size is determined by osmotic force balance, which depends on the ratio of the numbers of osmotically active molecules in the nucleoplasm to that in the cytoplasm (Lemière et al., 2022). The osmotic force balance is constant irrespective of cell size in cycling cells (Lemière et al., 2022), whereas it is markedly changed in stationary phase and GD cells, as shown by a reduction in the N/C ratio. Furthermore, solidification of the cytoplasm occurs in stationary phase and GD cells (Heimlicher et al., 2019; Joyner et al., 2016), which likely increases the osmotic pressure on the nucleus if the gelation is induced by increased hydrophilicity, expressed as a decrease in the Flory–Huggins parameter χ (Strobl, 2007). Therefore, a reduction in the intranuclear molecules and/or cytoplasmic solidification is likely to cause nuclear size reduction. In addition, given that lipid metabolism and membrane flow contribute to nuclear size regulation (Kume et al., 2017, 2019), alteration in the nuclear membrane homeostasis may further contribute to the nuclear size reduction. CDK might induce nuclear size reduction by regulating nuclear transport, cytoplasmic solidification and/or nuclear membrane homeostasis.

A decrease in the chromosome-occupying space might reflect a quiescence-associated chromosome compaction, which likely represses chromosome fluctuation. Upon nitrogen starvation, RNA interference-dependent histone modifications induce heterochromatin formation, resulting in inactivation of transcription (Joh et al., 2016; Roche et al., 2016), and in budding yeast, condensin-dependent intrachromosomal interaction at longer distances increases chromosome compaction and inhibits transcription in stationary phase (Rutledge et al., 2015; Swygert et al., 2019). Given that RNA interference-dependent heterochromatin formation and condensin activity are tightly coupled with the cell cycle (Kloc and Martienssen, 2008; Thadani et al., 2012), CDK might increase chromosome compaction and repress chromosome fluctuation by regulating histone modification and/or condensin activity. CDK might similarly regulate chromosome compaction and fluctuation in cycling cells, accounting for the increase in chromosome size and fluctuation seen in cycling cdc2-L7 cells. We also cannot exclude the possibility that the chromosome compaction level is coupled with the nuclear size by an uncharacterized molecular mechanism and that CDK changes the chromosome compaction and fluctuation by regulating nuclear size.

Given that the cohesin protein complex establishes sister chromatid cohesion through DNA replication (Blow and Tanaka, 2005; Koshland and Guacci, 2000), CDK likely increases sister chromatid association in response to nutrient decline by regulating cohesin in the preceding cell cycle, thus repressing sister chromatid separation in stationary phase. In cdc2-L7 cells, sister chromatid separation could be increased at an undetected level during log phase, resulting in an increase in sister chromatid separation in stationary phase. Supporting this idea, the functions of cohesin regulatory factors are altered in cycling cells under nutrient-limited conditions (Yanagida et al., 2011).

Given that nucleolar CDK accumulation occurs in response to various environmental stresses, CDK must have a common task at the nucleolus in stress responses. Inhibition of rDNA transcription is a common stress response in both quiescent and heat shock-responsive yeast cells (Claypool et al., 2004; Marguerat et al., 2012; Zhao et al., 2016). In addition, CDK inhibits rDNA transcription in mitosis (Heix et al., 1998; Sirri et al., 2002). Given these facts, we speculate that CDK accumulates in the nucleolus to inhibit rDNA transcription. The observed frequent Cdc2 localization to the two distinct regions might correspond to accumulation of Cdc2 at the two rDNA loci adjacent to both telomeres of chromosome III, which promotes transcriptional inhibition. As Gar1 and Gar2 are RNA-binding proteins (Girard et al., 1993; Gulli et al., 1995), it is also likely that these factors form an rRNA-containing stress granule-like structure, which represses rRNA metabolism at the nucleolar center. Alternatively, because cell cycle arrest is a common stress response, CDK might be sequestered at the nucleolus to promote termination of cell cycle progression. Given the requirement of Cdc2 activity for its nucleolar localization, Cdc2 likely executes its task through its phosphorylation-dependent interaction with a nucleolar factor or factors. However, phosphorylation of the nucleolar interactor(s) does not appear to be responsible for this localization, given that inactive Cdc2 variants failed to accumulate in the nucleolus even in the presence of endogenous intact Cdc2. One possibility is that the Cdc2–Cdc13 complex autophosphorylates, promoting the interaction of Cdc2 with the nucleolar factor.

Although CDK is crucial for stationary phase-associated changes, CDK inhibition did not completely prevent the nuclear size reduction and Cdc2 nucleolar accumulation in GD cells (Figs 5J and 7). Thus, CDK is not absolutely required for these changes, and other factors contribute. Many nutrient- and/or stress-responsive kinases including, TOR kinase, PKA, MAPK and AMPK, contribute to the establishment of quiescence and/or survival of quiescent cells (Forte et al., 2019; Gray et al., 2004; Petersen and Nurse, 2007; Yanagida, 2009; Zuin et al., 2010). It is likely that CDK collaborates with these kinases to establish the stationary phase, and even in the absence of CDK activity, these kinases might induce stationary phase-associated phenotypes.

We have shown that CDK actively contributes to the establishment of the stationary phase state in S. pombe. It is currently unclear whether CDK actively contributes to quiescence induced by distinct nutrient depletion or in other organisms. This is mainly because CDK inhibition is essential for the establishment of quiescence in all cases (Barbet et al., 1996; Daga et al., 2003; Marescal and Cheeseman, 2020; Matson and Cook, 2017; Moreno-Torres et al., 2015; Pajalunga et al., 2007; Sun and Buttitta, 2017; Tesio and Trumpp, 2011; Valcourt et al., 2012; van Velthoven and Rando, 2019; Velappan et al., 2017; Werner-Washburne et al., 1993; Yanagida, 2009; Zinzalla et al., 2007). However, considering the contribution of CDK to nutrient-dependent cell size control in budding yeast (Turner et al., 2012) and that there is a reduction of CDK activity in the cell cycle preceding quiescence in mammalian cells (Adikes et al., 2020; Arora et al., 2017; Barr et al., 2017; Fan and Meyer, 2021; Spencer et al., 2013), we speculate that the active contribution of CDKs to establishment of quiescence is common among eukaryotes. Furthermore, colocalization of CDKs with ribonucleoproteins seems to be a conserved stress response among eukaryotes. Understanding the functions of the CDK in stationary phase in S. pombe will shed a new light on the fundamental mechanisms underlying the establishment of quiescence in eukaryotes.

Strains and DNA primers used in this study are shown in Tables S1 and S2, respectively. YES and EMM media, and basic genetical techniques used in this study were described previously (Moreno et al., 1991). YES and EMM media contain 3% and 2% (w/v) glucose, respectively. Cells in log or stationary phase were prepared by growing cells to a cell density of ∼5×10⁶ or ∼1.2×10⁸ cells/ml, respectively, by monitoring cell growth every 2 h. For viability analysis, 110 individual cells were placed on YES solid medium using a thin microneedle under a microscope and incubated at 30°C, and viability was determined by their colony formation. To supply nutrients to the stationary phase culture medium, stationary phase cells were removed from the culture medium by centrifugation, and yeast extract was added to the culture medium at the concentration of 0.5% (w/v). The YE-supplied medium was filter sterilized and the removed cells were resuspended in it.

Intracellular DNA contents were analyzed by flow cytometry as previously described (Matsuhara and Yamamoto, 2016), but with the following modification. Intracellular RNA digestion was carried out using 400 µg/ml RNase A and intracellular DNA content was analyzed using a CytoFLEX flow cytometer (Beckman Coulter, Tokyo, Japan).

The nuclear membrane was visualized using an integration plasmid, which harbors the GFP-tagged N-terminal portion of the cytochrome C reductase gene, as described previously (Nambu et al., 2022). Chromosomes were visualized using the hta1-mCherry fusion gene, which was generated by inserting the mCherry gene at the hta1 chromosomal locus using a PCR-based method (Bähler et al., 1998; Krawchuk and Wahls, 1999). Cdc13 was visualized using the cdc13-GFP fusion gene as described previously (Tatebe and Yanagida, 2000). Clp1 was visualized using an integration plasmid pHM49, which was generated as follows. A genomic DNA fragment encoding the clp1⁺ gene together with its promoter and a portion of the pFA6a-GFP(S65T)-kanMX6 plasmid (Bähler et al., 1998), which encodes GFP and the ADH1 terminator, were amplified by PCR, and these amplified fragments were inserted at the multiple cloning site of pYC36 (Chikashige et al., 2004), yielding pHM49. Gar1 was visualized using an integration plasmid harboring the gar1-mCherry gene as described previously (Matsuhara and Yamamoto, 2016). Gar2 was visualized using the gar2-mCherry gene obtained from the Yeast Genetic Resource Center Japan. Microtubules were visualized as previously described (Yoshida et al., 2013).

The chromosome locus was visualized using the lacI/lacO recognition system, as described previously (Nabeshima et al., 1998; Yamamoto and Hiraoka, 2003), except that a different lacI construct was used. An integration plasmid, pMH1, which bears a lacI construct lacking the NLS, was generated by inserting a PCR-amplified DNA fragment encoding the GFP-tagged lac repressor at the cloning site of pTO2 (Yoshida et al., 2013) together with a PCR-amplified DNA fragment bearing the nda3 promoter. pMH1 was introduced into cells bearing the lacO repeats at the nhe1, ade8 or ade6 chromosomal locus.

WT Cdc2 was visualized using the endogenous cdc2⁺ gene fused with the GFP gene (a gift from Dr Hiroshi Murakami, Department of Biological Sciences, Chuo University, Japan) or using integration plasmids, pYN9 and pYN11, which bear cdc2⁺-GFP and cdc2⁺-mCherry fusion genes, respectively. pYN9 and pYN11 were generated as follows. Genomic and complementary DNA fragments encoding the cdc2 promoter and the cdc2⁺ gene, respectively, were amplified by PCR and inserted in front of the mCherry gene on an integration plasmid pHM4 (Yoshida et al., 2013), yielding pYN11. The PCR-amplified DNA fragments encoding the cdc2 promoter and CDS were also inserted at the cloning site of an integration plasmid, pTO2 (Yoshida et al., 2013) together with an PCR-amplified portion of a plasmid DNA, pFA6a-GFP(S65T)-kanMX6 (Bähler et al., 1998), encoding GFP(S65T) and the ADH1 terminator, yielding pYN9. The endogenous cdc2⁺ gene was deleted in cells expressing the cdc2⁺-GFP fusion using a PCR-based method (Bähler et al., 1998; Krawchuk and Wahls, 1999).

The cdc2-as mutant protein tagged with mCherry was introduced into cells using the integration plasmid pYN14. pYN14 was generated by introducing the corresponding mutation at the cdc2⁺-mCherry gene on pYN11 (see above), using the KOD-Plus-Mutagenesis Kit (TOYOBO, Osaka, Japan). The cdc2-Y15F, cdc2-Y15E, or cdc2-KD mutant protein tagged with GFP was introduced into cells using the integration plasmids, pYN7, pYN8 or pYN10, respectively. pYN7, pYN8 and pYN10 were generated by introducing the corresponding mutation at the cdc2⁺-GFP gene on pYN9 (see above), using the KOD-Plus-Mutagenesis Kit.

Given that placing cells between a slide glass and a coverslip caused rapid changes in the nuclear shape and intracellular localization of Cdc2 during observation, cells were observed using the following alternative method. Log phase cells were suspended in EMM medium, and stationary phase and GD cells were suspended in EMM-Glc medium. When cells were treated with chemicals, the cells were suspended in the medium containing the appropriate compound. ∼20 µl of the cell suspension was placed on a 40×50 mm coverslip (Matsunami Glass Ind., Ltd., Osaka, Japan) coated with 5 mg/ml lectin (Sigma-Aldrich Japan, Co., Tokyo, Japan). The cells were observed using an IX71 inverted microscope equipped with a cooled charge-coupled device camera (CoolSNAP-HQ2; Nippon Roper Co. Ltd., Tokyo, Japan) and a 60×/1.42 NA or 100×/1.40 NA Plan Apo oil-immersion objective lens (Olympus, Tokyo, Japan). Images of the nuclear envelope, chromosomes, or Cdc2 were taken at 11 or 20 focal planes spaced at 0.3 µm intervals. The resultant images were processed by deconvolution and analyzed using the MetaMorph (version 7) (Molecular Devices Japan, Tokyo, Japan) or ImageJ software (Schneider et al., 2012).

Cell volume was calculated using the Pombe Measurer ImageJ plugin (http://www.columbia.edu/∼zz2181/Pombe_Measurer.html; Pino et al., 2021). Nuclear area was determined by fitting an ellipse to the nuclear envelope on one section that showed the clearest envelope image. The nuclear volume was determined by regarding the nucleus as a spheroid with the semi-axis lengths, a, b, and (a+b)/2, where a and b are the lengths of the semi-axes of the fitted ellipse. Chromosome area was determined using the chromosomal region obtained from the binarized image of chromosomes in the two-dimensional projection formed from the 11 sections. Sister locus separation and intracellular localization of various molecules were analyzed in the two-dimensional projection.

The relative intensity of Cdc2 signal in the nucleolar region to that in the chromosomal region was determined as follows. Each deconvolved image set was combined to form a quantitative projection using an additive image projecting method. Then, the sums of the Cdc2 signal intensities in regions of the nucleus and chromosomes were obtained by subtracting the background signals. The total signal intensity in the nucleolar region was obtained by subtracting the total chromosomal signal intensity from the total nuclear intensity. The relative intensity of the nucleolar signal was determined using the obtained nucleolar and chromosomal signal intensities per area. Note that the relative nucleolar signal intensity was underestimated in this analysis given that the nucleolar region partially overlapped with the chromosomal region in the two-dimensional projection due to the shape and the positioning of the nucleolus.

Cells containing the GFP-visualized locus were transferred to EMM (log phase cells) or EMM-Glc medium (stationary phase cells or log phase cells exposed to acute glucose starvation). Approximately 20 µl of the cell suspension was placed on the bottom of 35 mm glass-bottom dishes (Matsunami Glass Ind., Ltd., Osaka, Japan) coated with 5 mg/ml lectin. Time-lapse images of the visualized locus in these cells were collected at three focal planes spaced at 0.4 µm intervals every 1.5 s using an IX71 inverted microscope equipped with a cooled charge-coupled device camera and a 100×/1.40 NA Plan Apo oil-immersion objective lens (Olympus, Tokyo, Japan). To minimize each acquisition time, the area captured was less than 800 pixels in size and an exposure time of less than 90 m second was used. During collection of time-lapse images, the cells were kept at 25°C. The resultant images were processed by deconvolution using MetaMorph software, and the three-dimensional coordinate of the GFP-visualized chromosome locus was then determined at each time point using a multi-dimensional motion analysis module. When the visualized sister loci were separated, the coordinate of their midpoint was used for analysis. The obtained coordinates were used to plot MSD plots and their confidence intervals were statistically evaluated. The basic bootstrap method (Davison and Hinkley, 1997) was used to evaluate the confidence intervals, in order to correct for differences in time length of the coordinate data, as previously described (Wakiya et al., 2021). A brief description of the statistical method is given below.

Cells containing mCherry-tagged Cdc2-as were grown in YES medium to log phase cell density at 30°C. Cdc2 kinase activity was inhibited by addition of 1:1000 culture volume of 3MB-PP1 stock solution (4 or 5 mM) (Funakoshi, Tokyo, Japan) to the culture and incubation for 10 min with shaking. The cells were then suspended in YES-Glc medium containing 4 or 5 µM 3MB-PP1 after washing twice with the same medium and further incubated at 30°C. As a control, DMSO was used instead of 3MB-PP1. Note that no differences were detected in cdc2-as cells treated with 4 or 5 µM 3MB-PP1.

A chromosomal locus was regarded as a particle diffusing in a space surrounded by a spherical wall representing the nuclear envelop. The time step of the simulation was taken to be Δt=0.5 s. Correlated random numbers were generated to make the displacement of the particle at each step. The statistics of the random displacement is such that MSD(t)∝t^α, where the exponent is set to be α=0.67 in order to reproduce the experimental results. Generation of correlated random numbers was carried out using the Fourier transformation method (Berkowitz et al., 1983). When the particle hits the wall, it is reflected elastically. The radius R_nuc of the spherical wall is set to be R_nuc=1.272 μm and 0.856 μm, obtained from the observed nuclear volumes for the log and stationary phases, respectively. For each nucleus size, 1000 trajectories each with 2¹⁶ steps were generated, and their statistical average was taken to calculate MSD.

For western blotting analysis of proteins, ∼5×10⁷ and ∼2×10⁸ cells were collected for log and stationary phase cells, respectively. The cells were washed twice with ice-cold STOP buffer (150 mM NaCl, 50 mM NaF, 1 mM NaN₃, 10 mM EDTA, pH 8.0) and stored at −80°C until use. The cells were suspended in 20 µl or 50 µl water for log phase or stationary phase cells, respectively, and the cell suspension was immediately heated at 100°C for 5 min. Subsequently the suspension was mixed with twice the volume of urea buffer (8 M urea, 4% SDS, 1 mM DTT and 120 mM Tris-HCl pH 7.4). The cells were then disrupted using glass beads and the bead cell disrupter (MagNA Lyser, Nippon Genetic). A portion of the cell extract was used for protein concentration analysis with a BCA protein assay kit (Thermo Fisher Scientific, Tokyo, Japan), and the rest of the extract was mixed with an equal volume of 2× SDS sample buffer and heated at 100°C for 3 min for protein extraction. After cell debris removal by centrifugation (22,140 g for 15 min), the cell extract containing 40 µg protein was loaded for each sample onto a 10% SDS-PAGE gel. After electrophoresis, the proteins were detected by western blotting using the following antibodies: mouse monoclonal anti-PSTAIR antibody (1:50,000; P7962; lot no: 090M4829; Sigma-Aldrich), mouse anti-GFP antibody (1:3000; clones 7.1 and 13.1; cat. no: 11 814 460 001; Roche Diagnostics), and mouse monoclonal anti-α-tubulin antibody (1:40,000; clone B-5-1-2; product no: T 5168; Sigma-Aldrich).

We are grateful to Tomomi Kato, Yukina Saitoh, Sayaka Yokota, Momoka Ishida, Kenta Sato, Aya Yamaguchi and Sadia Afrin for their contribution to the initial stage of this study; Hirotada Matsuhara for constructing plasmid pHM49; Hiroshi Murakami and the Yeast Genetic Resource Center for strains; and Takashi Ushimaru for the use of the bead cell disrupter. We are also grateful to Masahiro Uritani, Takashi Ushimaru and Yoko Kimura for helpful discussions, and Yoko Kimura and Hiroshi Murakami for critical reading of the manuscript and helpful comments. Intracellular DNA contents were analyzed using the flow cytometer at Shizuoka University Research Institute of Green Science and Technology.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260727.reviewer-comments.pdf