Reliable imaging of ATP in living budding and fission yeast

Adenosine triphosphate (ATP) is a universal energy currency used by all living organism. In the human body, the half-life of ATP is estimated to be a few seconds (Mortensen et al., 2011), indicating high demand of this energy currency and suggesting that its synthesis and consumption rates are tightly regulated. Because of its importance, the molecular mechanism of ATP synthesis by glycolysis and mitochondrial respiration has been rigorously investigated (Berg et al., 2012; Lehninger et al., 2010). However, little is understood how an ATP concentration is maintained in a single cell under different conditions because our knowledge on ATP dynamics is largely based on biochemical analysis, which has poor time resolution compared with the rapid turnover of ATP. Biochemical analysis also precludes characterization of heterogeneity of ATP concentration within a population or a tissue.

The recent development of ATP-biosensors enabled us to monitor changes in ATP concentration in single living cells (reviewed in Dong and Zhao, 2016). ATeam, the first FRET-based ATP biosensor, has successfully been introduced into mammalian cells and is now widely used to visualize ATP dynamics in living cells (Imamura et al., 2009). The second generation ATP biosensor QUEEN (Yaginuma et al., 2014) uses a single green fluorescent protein (FP) , instead of the combination of cyan FP and yellow FP used for ATeam, and has substantial advantages, especially for the use in rapidly growing microorganisms, such as yeast. First, it has no maturation time lag between two FPs that can yield a dysfunctional sensor in rapidly dividing cells, such as bacteria and yeasts (see Discussion for details). Second, it is more resistant to degradation than the FRET-based sensor ATeam. Third, QUEEN has a 1.7 times wider dynamic range (the ratio between the maximum and minimum ATP concentration values) compared with that of ATeam sensors (Yaginuma et al., 2014).

By using QUEEN, it was reported that unexpectedly broad variations of ATP concentration exist within a clonal population of bacteria (Yaginuma et al., 2014). In addition to negative-feedback regulation of a metabolite concentration (Chubukov et al., 2014), eukaryotic cells harbor energy-sensing mechanisms, such as AMP-activated protein kinase (AMPK) (Hardie et al., 2016). Thus, it is expected that the ATP concentration is maintained at a specific concentration in eukaryotes.

Yeasts have provided an excellent model system for studying eukaryotic biology. Especially, the central carbon metabolism, including glycolysis and the tricarboxylic acid (TCA) cycle, of yeast has been extensively studied, and even engineered, because of its importance to the fermentation industry to produce useful metabolites, including ethanol (Borodina and Nielsen, 2014; Gibson et al., 2017). Yeast carbon metabolism also provides a tractable model for the energy metabolism of cancer cells since yeast and cancer cells are similar in that they both mostly synthesize ATP through glycolysis, even in the presence of oxygen, as long as glucose supply is high. This is known as ‘aerobic fermentation’ or ‘Warburg effect’ in cancer cells (Diaz-Ruiz et al., 2011). In addition to being an important indicator of cell energy, ATP itself is a common regulator of multiple glycolytic enzymes (Larsson et al., 2000; Mensonides et al., 2013). Therefore, to elucidate the cellular dynamics of ATP is essential also in order to decipher the regulation of glycolytic flux, but remained unaddressed for aforementioned reasons.

Here, we have applied QUEEN in budding and fission yeasts for the first time, and found that the ATP level showed little variation within a population, suggesting a robust ATP homeostasis in eukaryotic cells. We further found that the concentration of ATP is maintained at a constant level regardless of the carbon sources, and has no obvious correlation with the mitotic growth phase and growth rates. However, we found that ATP levels decline during fission yeast meiosis, suggesting that ATP homeostasis is controlled within a developmental context, not by the availability of sugar.

Taken together, visualization of ATP dynamics in yeast reveals the existence of robust ATP homeostasis. QUEEN-expressing yeast cells are useful tools to study metabolic activity in individual cells, and offer opportunities to test ATP dynamics under various environmental conditions and in various mutants.

To monitor the dynamics of ATP concentration in living single cells, we have recently developed several ATP indicators including ATeam (Imamura et al., 2009) and QUEEN (Yaginuma et al., 2014). To explore ATP homeostasis in budding yeast Saccharomyces cerevisiae, we chose QUEEN because it has several strongpoints (see Introduction). The unique feature of the fluorescent biosensor QUEEN is that the binding to ATP shifts its optimal excitation wavelength from 480 nm to 410 nm (Fig. 1A), which allows us to estimate the ATP level by quantification of the ratio between the fluorescence signal intensities excited at 410 nm and 480 nm.

We created a budding yeast strain stably expressing QUEEN-2m (K_d of QUEEN-2m with ATP is ∼4.5 mM at 25°C, comparable with estimated cellular ATP concentration in glucose grown yeast) under the promoter of translation elongation factor 1α (TEF1) from the HIS3 locus. The QUEEN protein was expressed at similar levels in a clonal population and evenly distributed both in the cytoplasm and nucleus due to its small size (42 kDa) but excluded from the vacuole (Fig. 1B). QUEEN has two excitation peaks (at 410 nm and 480 nm; hereafter referred to as 410ex and 480ex, respectively) and one emission peak (at ∼520 nm). QUEEN was sequentially excited by 480 nm and 410 nm light, and the emitted fluorescence signals at around 520 nm were imaged. The ratio of the emission intensity at the two excitation peaks (denoted 410ex:480ex) was calculated for each pixel. The mean of the QUEEN fluorescence intensity ratios (hereafter referred to as the QUEEN ratio) in the intracellular region reflects the ATP concentration of the cell (see Materials and Methods for details).

First, we tested if the QUEEN ratio, indeed, reflects the cellular concentration of ATP in living yeast cells. A previous biochemical study has demonstrated that ATP levels in budding yeast cells dropped to 15% upon glucose depletion and gradually recovered to 50% of the original level within 20 min (Xu and Bretscher, 2014). Moreover, replacement of glucose in medium with 2-deoxy-D-glucose (2DG), which strongly inhibits glycolysis, reduced cellular ATP levels to <1% at least for 30 min (Serrano, 1977; Xu and Bretscher, 2014). The QUEEN ratio in yeast cells rapidly dropped after glucose removal (Fig. 1C, 3 min) but partially recovered within 30 min (Fig. 1C). In addition, replacement of glucose with 2% 2DG resulted in rapid and prolonged reduction in the QUEEN ratio (Fig. 1C). Thus, the QUEEN ratio nicely reflects cellular ATP concentration analyzed by biochemical methods.

Second, we tested if QUEEN can reversibly monitor a change in ATP concentration. We took advantage of gph1Δ mutant cells that cannot catabolize glycogen and do not show any recovery of ATP after glucose depletion (Xu and Bretscher, 2014). The mean QUEEN ratios in glucose-depleted gph1Δ cells declined rapidly within 10-15 min, but re-feeding of glucose fully recovered these values within a minute (Fig. 1D), indicating that the reduction of QUEEN ratio after glucose depletion was not due to an irreversible damage caused to QUEEN. Taken together, these results suggest that the QUEEN ratio reliably reflects ATP levels in individual cells, allowing us to monitor the dynamicity of ATP concentration in living cells. Based on the QUEEN ratio, we were also able to estimate the actual ATP concentration in cells by fitting the acquired QUEEN ratio with the calibration curve (Yaginuma et al., 2014) (Fig. S1 and see Materials and Methods for details).

One of the advantages of using QUEEN compared with classic biochemical measurements is in its time resolution. We can now monitor the dynamic change of ATP concentration in seconds, which allows us to measure metabolic activity of individual living yeast cells. An example is shown in Fig. 1E, where we monitored the QUEEN ratio after the treatment with 2DG. This analysis suggests that the half-life of ATP under glycolytic conditions is <1 min in budding yeast.

We also have generated QUEEN constructs targeting the mitochondrial matrix (mitQUEEN) and the cytoplasmic surface of the endoplasmic reticulum (ER) (erQUEEN) to analyze the local ATP concentration (Fig. 1F). In the case of mitochondria, we found that the QUEEN ratio (ATP concentration) in mitochondria is less than that in the cytoplasm when cells were grown under glycolytic conditions containing 2% glucose (Fig. S2), which is similar to what has been reported in HeLa cells by using ATeam (Imamura et al., 2009). The detailed use of mitQUEEN and erQUEEN will be described elsewhere and we focus here on the levels of ATP in the cytoplasm and nucleus. Our results collectively demonstrated that QUEEN is a useful ATP reporter in budding yeast.

With the QUEEN system, we are now able to monitor ATP levels in single living cells under various conditions. First, to examine the use of carbon sources on ATP, we analyzed QUEEN ratios in cells grown in different hexoses. None of the hexoses tested (2% fructose, 2% galactose, 2% glucose, 2% mannose) significantly affected ATP concentration (Fig. 2A), suggesting that the ATP level is maintained stable, regardless of the carbon sources. We also examined the contribution of mitochondrial respiration to ATP levels by treatment with the respiration inhibitor antimycin A (Walther et al., 2010). Antimycin A treatment (2 µg/ml) only slightly reduced (∼20%) the ATP levels in cells growing in the presence of one of the hexoses (Fig. 2A), suggesting that respiration has a relatively minor role in ATP synthesis if there is enough carbon.

Next, we explored the effect of fermentative or non-fermentative carbon sources on ATP levels. Depletion of fermentative carbon source glucose resulted in a rapid drop of QUEEN ratio within 3 min, followed by a gradual recovery after 3 h (Fig. 2B). This recovery was due to activation of respiration because treatment of antimycin A suppressed the QUEEN ratio (Fig. 2B). When cells were grown in non-fermentative glycerol, the ATP level was comparable to that in a glucose-grown culture, and removal of the glycerol had small effect on the QUEEN ratio, even after 3 h. The high ATP level maintained in glycerol grown cells was due to active respiration, as treatment with antimycin A significantly suppressed it (Fig. 2B). Altogether, our visualization of ATP using QUEEN confirmed that yeast cells growing in fermentative carbon depend heavily on glycolysis to yield cellular ATP (Barnett and Entian, 2005). Respiration, however, is highly dominant in yeast cells grown in a non-fermentative carbon source (Kayikci and Nielsen, 2015; Shashkova et al., 2015) consistent with the observation that glycerol-grown yeast has highly developed mitochondria compared to glucose-grown yeast (Egner et al., 2002).

Next, we analyzed ATP concentration during mitotic cell cycle. The QUEEN ratio was followed over several generations using time-lapse imaging with Myo1-mCherry as a bud neck marker indicative of budded stages. In clear contrast to what has been observed in bacteria (Yaginuma et al., 2014), we observed little fluctuation of ATP levels in rapidly growing yeast in 2% glucose (Fig. 3A,B; Movie 1). We also quantified the QUEEN ratio in cells of different cell cycle stages, such as unbudded (G1), small budded (S) or large budded (G2 and M) but did not find significant differences between cell cycle stages (Fig. 3C). Little fluctuation of ATP concentration during the cell cycle in single cells and among a population suggest a mechanism that stably maintains the ATP concentration in yeast. Stable maintenance of ATP during the cell cycle was also observed in cells cultured in glycerol (Fig. 3D), suggesting that neither glycolysis nor respiration is significantly regulated during the cell cycle.

We then examined the use of QUEEN in fission yeast Schizosaccharomyces pombe. The QUEEN construct was integrated into the gene expressing 3-isopropylmalate dehydrogenase (Leu1) under the promoter of the translation elongation and termination factor eIF5A (TIF51). QUEEN protein was evenly distributed both in the cytoplasm and in the nucleus (Fig. 4A). The QUEEN ratio was calculated in the same manner as described for S. cerevisiae, and we found a relatively high QUEEN ratio in fission yeast cells grown in EMM medium containing 2% glucose. However, this ratio dropped within 5 min after replacing the medium with one containing 20 mM 2DG and 10 µg/ml antimycin A (Fig. 4A,B), suggesting that the QUEEN ratio is reflecting the intracellular concentration of ATP in fission yeast. We also noticed that the half-life of ATP under glycolytic conditions is ∼1-2 min in fission yeast, as judged by a rapid reduction in QUEEN ratio after 2DG treatment (Fig. 4C).

It is known that growth of fission yeast largely relies on glycolysis when there are high concentrations of glucose, but relies on respiration for survival when glucose is limited (Takeda et al., 2015). To examine the relative contribution of glycolysis and respiration in the ATP level, we monitored the QUEEN ratio of fission yeast grown in different glucose concentration. In the presence of 2% or 0.02% glucose, cells contain about 3 mM ATP. Even after 6 h of glucose depletion, cells are viable and maintained 2.3 mM ATP (Fig. 5A,B). The contribution of glucose and mitochondria to the level of ATP was clearly countercorrelated. Treatment of antimycin A had a minor effect on the ATP level in cells grown in 2% glucose but a larger effect in those grown in 0.02% glucose (Fig. 5A,B). Respiration was essential for ATP production in the absence of glucose. Glucose-depleted cells completely lost cellular ATP after the treatment with antimycin A for 10 min (Fig. 5A,B).

Our analysis of ATP concentration is in good accordance of a recent study by Takeda et al., showing that glucose has a dominant role in cellular energetics and that respiration plays main role when glucose concentration is severely limited (Takeda et al., 2015). Thus, QUEEN is a reliable and useful indicator of ATP concentration in fission yeast.

In the following, we examined whether ATP concentration fluctuates during the cell cycle in fission yeast. A unique feature of fission yeast is that the stage of the cell cycle is tightly correlated with cell length (Mitchison and Nurse, 1985). We quantified the QUEEN ratio in cells of various lengths but did not find a significant relationship between QUEEN ratio and cell length, suggesting that ATP concentration does not change during the mitotic phase of the cell cycle (Fig. 6A).

Time-lapse imaging of the QUEEN ratio in mitotically growing cells further confirmed that ATP levels fluctuate little during the cell cycle in fission yeast grown in 2% glucose (Fig. 6B,C and Movie 2).

We also examined the QUEEN ratio in sporulating fission yeast. Fission yeast undergoes the meiotic phase of the cell cycle when nitrogen is depleted. We first confirmed that depletion of nitrogen (in the presence of 2% glucose) did not affect cellular ATP levels for the first 30 min (Fig. 7A,C). After 16 h without nitrogen, a mixed population of fission yeast cells underwent various stages of meiosis (Yamashita et al., 2017). We found a partial decline in the QUEEN ratio in cells undergoing meiosis (Fig. 7A,B). Treatment with antimycin A (for 30 min) or 2DG (for 10 min) revealed that glycolysis is the main source of ATP during meiosis (Fig. 7A). After sporulation, ATP levels declined further in the remnant cytoplasm but were maintained at higher levels inside the spores (Fig. 7B,D). The ATP level in the spore also seemed to be due to glycolysis but not to mitochondria, as treatment with 2DG but not antimycin for 2 h had a significant effect (Fig. 7B). These results reveal that the ATP level is controlled both temporally and spatially during meiosis, and that glycolysis plays pivotal role in the ATP synthesis during meiosis.

In this study, we visualized ATP dynamics in both budding and fission yeast by using the ATP biosensor QUEEN for the first time in eukaryotic cells. In clear contrast to bacterial cells, there is little fluctuation/variation in the ATP concentration in yeast populations grown under the same culture conditions. Furthermore, we found that the ATP concentration in yeast is remarkably constant regardless of the carbon sources or the stage of the cell cycle stage, suggesting a robust ATP homeostasis in eukaryotic cells.

We show that QUEEN has a wide dynamic range that is optimal for the physiological concentration of ATP in yeast cells, and have established a reliable and sensitive assay to measure the ATP concentration in living yeasts. QUEEN-expressing yeast strains and plasmids are publically available from the Yeast Genetic Resource Centre Japan (YGRC, http://yeast.nig.ac.jp/yeast/top.xhtml). Using QUEEN to visualize ATP has many advantages. They include: 1) the reversibility of the QUEEN signal allows to monitor the dynamics of ATP concentration in a single living cell; 2) The time resolution when using QUEEN is much higher compared with standard biochemical assays; 3) Monitoring of the subcellular distribution of ATP is possible by using organelle-targeting QUEEN; 4) Variations in ATP concentration can be observed during the life of a single cell or in a lineage and among a cell population.

There have been several attempts to visualize ATP in living yeasts. One approach, using an aptamer binding ATP, was successful in real-time visualization of ATP in budding yeast (Özalp et al., 2010). However, the use of the aptamer is limited because its introduction to cells is an invasive and hard-to-control step. The FRET-based ATP biosensor ATeam is widely used in mammalian systems, but is unsuitable for rapidly proliferating yeasts cells (with a doubling time of 90–100 min) because the long maturation time of the sensor and the presence of malfunctioning sensors would lead to unreliable outcomes as observed when ATeam was used in growing bacteria (Yaginuma et al., 2014). We, thus, believe that QUEEN is the best ATP biosensor for yeasts currently available.

By using QUEEN, we found that ATP is maintained at a stable concentration (3–4 mM) regardless of the growth conditions, suggesting that a mechanism exists to maintain ATP concentrations at a certain level. We also noticed that the mean ATP concentration in budding yeast cells grown in the presence of 2% glucose (calculated using the QUEEN ratio 3.2±0.4 mM; Fig. 2) shows good agreement with that of intracellular ATP concentrations (estimated by biochemical analysis; i.e. 4 mM or 4.6 mM; Gauthier et al., 2008; Ljungdahl and Daignan-Fornier, 2012). This demonstrates reliability of QUEEN and the validity of our calibration method. By using QUEEN, we have previously reported a large variation in cellular ATP levels within rapidly proliferating bacteria (Yaginuma et al., 2014). In contrast to bacteria, we found little variation in the cellular concentration of ATP in yeast under different growth conditions, suggesting a mechanism that maintains ATP concentration in eukaryotic cells.

It has been proposed that concentrations of metabolites are generally regulated by negative feedback mechanism (Chubukov et al., 2014). Excessive levels of ATP can inhibit glycolysis by allosterically inactivating phosphofructokinases and pyruvate kinases (Larsson et al., 2000; Reibstein et al., 1986). In eukaryotic cells, a decrease of ATP (and increase of AMP) activates AMPK, which is thought to suppress ATP consumption (Hardie et al., 2016). It is of great interest to define the detailed molecular mechanisms leading to stable maintenance of ATP in yeast.

Because ATP is essential for many cellular functions, ATP concentration is often used as an indicator of cellular activity. It is surprising to find the cellular concentration of ATP to be similar in cells grown with or without glucose because both budding and fission yeasts proliferate significantly slower in the absence of carbon sources (Takeda et al., 2015; Tyson et al., 1979). Our study also revealed that the ATP concentration is stably maintained during normal cell cycle in both types of yeast. It is anticipated that the rate of cellular energy consumption is controlled via the cell cycle because the rate of cell growth is affected by the cell cycle (Goranov et al., 2009). Moreover, there are various energy-consuming events, such as DNA replication in S-phase and chromosome segregation in mitosis. Our findings suggest a mechanism that maintains the concentration of ATP at a certain level, regardless of the speed of growth or stage of the cell cycle. Thus, ATP concentration is unlikely to be a direct indicator of metabolic activity and we need to monitor the ATP turnover-rate in order to measure cell activity. The use of QUEEN offers an easy and reliable assay to measure the ATP turnover-rate compared with standard biochemical assays.

QUEEN-expressing yeast enabled us for the first time to monitor the ATP level during fission yeast meiosis. ATP levels partially decline when cells enter meiosis and are maintained at relatively high levels only inside the spores. The mechanism by which ATP in the spore is maintained requires further investigation. In addition, the ineffectiveness of antimycin A against ATP level in spores does not rule out the possibility that mitochondrial respiration also contributes to maintain ATP in spores because the spore membrane might be hardly permeable for the drug. The requirement of glycolysis but not respiration for ATP synthesis during fission yeast meiosis is somewhat confusing because fission yeast defective in respiration fails to sporulate (Jambhekar and Amon, 2008). This suggest requirement of mitochondria for sporulation involves steps other than ATP synthesis, such as lipid and amino acid metabolism. Alternatively, respiration might play a more-dominant role in meiosis of fission yeast under natural conditions where carbon sources can also be also limited (and in our sporulation assay, 2% glucose was added).

The mechanism on how the decrease of ATP level in meiotic cells is also unknown, but our analysis suggests a change in glycolytic activity not respiration as contributing factor. It has been reported that expression levels of metabolic enzymes, including glycolytic enzymes, are largely altered during meiosis in fission yeast (Mata et al., 2002; Yamamoto, 1996), suggesting that metabolic activity and ATP homeostasis are remodeled during this differentiation process.

Live cell imaging of metabolites is a powerful approach to discover sporadic events that only happen in a fraction of cells within a population, or that fluctuate within a single cell. The fact that biosensors can also be targeted to organelles or specific subcellular compartments, allows the measurement of the concentration metabolites at a certain location (Imamura et al., 2009). Further studies are needed to clarify the metabolic control of fission yeast meiosis but the use and application of QUEEN in yeasts is limitless. We are now able to monitor the dynamics of ATP with good spatial and temporal resolution in living yeasts, and can explore various unaddressed questions regarding ATP levels, such as levels during diauxic shift, starvation, stress response, differentiation and aging. Furthermore, the ease of using the QUEEN system to measure ATP consumption rates allowed us to analyze metabolic activity in single cells under various conditions. Therefore, the use of QUEEN in yeasts cells may bring our understanding of bioenergetics to another level.

Budding and fission yeast strains, and plasmids used in this study are listed in Tables S1, S2, and S3, respectively. Strains were constructed by using a PCR-based method (Janke et al., 2004) and genetic crosses. The yeast knockout strain collection was originally purchased from GE Healthcare (cat# YSC1053). Some plasmids were originally purchased from EUROSCARF (Oberursel, Germany). Construction of the fission yeast strain expressing QUEEN (KSP3769) is described elsewhere (Ito et al., 2019).

QUEEN-expressing budding yeast strains were constructed as follows. First, QUEEN-2m ORF was isolated by cutting pRSETb-QUEEN-2m (Yaginuma et al., 2014) with BamHI and HindIII, and ligated into BamHI/HindIII sites of pSP-G2 (Partow et al., 2010) to yield MTP3051. Next, a SacI/BamHI fragment of S. cerevisiae TEF1 (translation elongation factor 1α) promoter from pYM-N19 (Janke et al., 2004; Mumberg et al., 1995) and a BamHI/PvuII fragment of QUEEN-2m ORF flanked with CYC1 terminator from MTP3051 were inserted by trimeric ligation into the SacI-EcoRV sites of pRS303 to yield MTP3067. For expression of QUEEN-2m from the his3 locus, MTP3067 was linearized by PstI and integrated into the his3Δ1 locus of the wild-type strain MTY3015 to yield MTY3255 and 3261. Integrations and the copy numbers of QUEEN were confirmed by diagnostic PCR and western blotting.

A budding yeast strain expressing a QUEEN construct that localizes to the mitochondrial matrix was generated as follows. First, a DNA fragment encoding a two tandem copies of the mitochondrial targeting signal sequence of human cytochrome c oxidase subunit VIII and the first 20 amino acid residues of QUEEN-2m was artificially synthesized with codon optimization for yeasts (Eurofin Genomics, Ebersberg, Germany). Next, the DNA fragment was inserted into the XbaI-EcoRV sites of MTP3067 to yield MTP3079. This, in turn, was linearized and integrated into the his3Δ1 locus of MTY3015 to yield MTY3227.

A budding yeast strain expressing a QUEEN construct that localizes to the cytoplasmic surface of endoplasmic reticulum was generated as follows. First, a DNA fragment encoding SEC71TMD (Sato et al., 2003) and its neighboring sequences (40 amino acids), a five-glycine linker and the first 20 amino acid sequence of QUEEN-2m was artificially synthesized (Eurofin Genomics). Next, the DNA fragment was inserted into the XbaI-EcoRV sites of MTP3067 to yield MTP3080. MTP3080 was linearized and integrated into the his3Δ1 locus of MTY3015 to yield MTY3229.

All new strains and plasmids have been deposited to and are available from the Yeast Genetic Resource Centre Japan (YGRC, http://yeast.nig.ac.jp/yeast/top.xhtml).

Synthetic complete medium (SC) for budding yeast was prepared according to Hanscho et al. (2012). Complete yeast extract (YE) medium, synthetic Edinburgh's Minimal Medium (EMM) and malt extract (ME) medium for fission yeast were prepared according to the recipes by Forsburg et al. (Forsburg and Rhind, 2006). YE and EMM medium was supplemented with 100 µg/ml adenine, 20 µg/ml uracil, 20 µg/ml L-histidine, 30 µg/ml L-lysine and 60 µg/ml L-leucine. Cells were grown to mid-log phase at 30°C in synthetic medium before imaging unless otherwise indicated. 2-deoxy-D-glucose (2DG) was purchased from FUJIFILM Wako (Osaka, Japan) (cat# 046-06483) and dissolved in SC or EMM medium instead of glucose. Antimycin A was purchased from FUJIFILM Wako (cat# 514-55521) and dissolved in dimethylsulfoxide (DMSO) to make a stock solution (20 mg/ml).

To induce mating and meiosis of fission yeast cells, homothallic MTY1162 cells were grown to mid-log in liquid YE, washed twice with water, and then incubated on an ME plate for 14–20 h at 25°C. Zygotes and asci in the culture were suspended in EMM lacking a nitrogen source (EMM−N), and subjected to microscopy.

Budding and fission yeast cells were immobilized on a 35-mm-glass-bottom dish (#3971-035, 1.5 thickness, IWAKI, Shizuoka, Japan) coated with concanavalin A (C-7275, Sigma-Aldrich, St Louis) or soybean lectin (L-1395, Sigma-Aldrich), respectively, unless otherwise indicated. The dish was filled with excess amount of medium (4.5–5 ml) compared with the cell volume to minimize changes in chemical compositions of the medium during observation. In some cases (Figs 1B-D,F and 3C), cells were concentrated by centrifugation and sandwiched between a slide and a coverslip (1.5 thickness, Matsunami, Osaka, Japan). In the latter case, imaging was completed within a few minutes after the preparation. The immobilized cells were imaged using an inverted fluorescent microscope (Eclipse Ti-E, Nikon, Tokyo, Japan) equipped with Apo TIRF 100× Oil DIC N2/NA 1.49 objective lens and an electron multiplying charge-coupled device camera (iXon3 DU897E-CS0-#BV80, Andor-Oxford Instruments, Abingdon, UK) at around 25°C. QUEEN fluorescence emitted around 520 nm following excitation at 480 nm and 410 nm was assessed from a single z-plane using a FITC filter set (Ex465-495/DM505/BA515-555, Nikon) and a custom-made filter set (Ex393-425/DM506/BA516-556, Semrock, New York, US), respectively. We sometimes imaged QUEEN fluorescence signals by using an FV-1000 confocal laser scanning microscope (Olympus, Tokyo, Japan) (Figs 1B-D, 2B and 3C) equipped with UPLSAPO 100×O/NA 1.4 objective lens (Olympus) with excitation by 473-nm and 405-nm lasers using a dichroic mirror DM405/473 and a barrier filter BA490-540. QUEEN fluorescent signal from mitochondrial matrix was collected from stacks of eleven z-sections spaced by 0.5 µm. Images of cells were acquired from several fields of view for each experimental condition, providing a high enough sample size for quantitative analysis.

Acquired digital images were analyzed using a Fiji software (Schindelin et al., 2012). The fluorescence images of QUEEN upon excitation at ∼480 nm (ex480 image) or ∼410 nm (ex410 image) were collected as described above. The QUEEN ratio was calculated as follows. First, both the QUEEN images were converted to signed 32-bit floating-point grayscale and then corrected for background using the rolling-ball algorithm or by subtracting the mean pixel values in an area outside the cells. Next, the images were thresholded using the modified IsoData algorithm to set background (non-thresholded) pixels to the Not a Number (NaN) value. The pixel values of the ex410 image were divided by those of the ex480 image to calculate the QUEEN ratio at each pixel. The ratio images were expressed in pseudocolor with appropriate look-up tables and display range. The mean ratio in pixels corresponding to the inside of a cell was used to represent the ATP level of the cell.

Cell boundaries in the ratio image were determined as follows. QUEEN images were merged, and then thresholded to generate a binary image. Particles (corresponding to cells expressing QUEEN) in the binary image were analyzed and automatically outlined to draw regions of interest (ROIs) of the cells. Alternatively, ROIs of cells were manually drawn. Numerical data were plotted using the KaleidaGraph software ver. 4.5.1 (Synergy Software, PA, US) or the R studio software ver. 3.4.1 (R Core Team, 2017).

Based on the biochemical data (Serrano, 1977; Xu and Bretscher, 2014), we assumed depletion of ATP when cells were treated with 2DG in budding yeast. We quantified the QUEEN ratio in the cells treated with 2DG in the absence of glucose with our Eclipse Ti-E, Nikon microscope system (Fig. S1A,B) and adjusted the calibration curve previously published (Yaginuma et al., 2014) to the zero-point. This calibration curve was used to estimate the ATP concentration in yeast cells when the QUEEN ratio was analyzed with the Nikon system.

The QUEEN ratio in the budding yeast cells in 2% glucose medium was ∼2.7 times higher than that in the cells treated with 2DG. Therefore, the dynamic range of the QUEEN ratio is at least 2.7 below our experimental conditions. This is in good agreement with previous results, showing a dynamic range of QUEEN in bacteria of at least threefold (Yaginuma et al., 2014).

Means, SDs, and P-values were calculated using Excel software (Microsoft, WA). Significance between two sets of data was tested using unpaired one-tailed Welch's t-test and is indicated by the low P value (<0.05). The horizontal bar in dot plot indicates the average of each population. We plotted and compared data obtained from experiments carried out the same day and did not pool data from experiments carried out on different days because several factors vary slightly day-to-day and can affect QUEEN ratio (e.g. room temperature, batch of medium, intensity of excitation light, conditions of optical filters for fluorescence microscopy and cell density). All QUEEN ratio measurements were repeated at least twice.

We thank members of Yoshida/Takaine labs for their support. We also thank K. Ohashi, R. Chaleckis and F. Matsuda for valuable discussion and comments, and the Yeast Genetic Resource Center (Osaka City University, Japan) for providing plasmids.