Nutrient availability and stresses impact a cell’s decision to enter a growth state or a quiescent state. Acetyl-CoA stimulates cell growth under nutrient-limiting conditions, but how cells generate acetyl-CoA under starvation stress is less understood. Here, we show that general stress response factors, Msn2 and Msn4, function as master transcriptional regulators of yeast glycolysis via directly binding and activating genes encoding glycolytic enzymes. Yeast cells lacking Msn2 and Msn4 exhibit prevalent repression of glycolytic genes and a significant delay of acetyl-CoA accumulation and reentry into growth from quiescence. Thus Msn2/4 exhibit a dual role in activating carbohydrate metabolism genes and stress response genes. These results suggest a possible mechanism by which starvation-induced stress response factors may prime quiescent cells to reenter growth through glycolysis when nutrients are limited.https://doi.org/10.7554/eLife.29938.001
Cell growth and proliferation are actively coordinated with extrinsic nutrient availability and intrinsic metabolic states (Broach, 2012). When nutrients are limited, cells enter into quiescent states to enhance survival (Gray et al., 2004). Repletion of nutrients stimulates quiescent cells back into growth (Dechant and Peter, 2008). Gene expression and metabolite profiles are dramatically remodeled during the transitions between quiescence and growth (Klosinska et al., 2011). Many transcription factors (TFs), enzymes and metabolites have been shown to regulate the transitions. However, gaps remain in understanding the crosstalk between transcriptional and metabolic activities during the transitions and the mechanisms by which cells make the decision.
Here, we exploited the yeast metabolic cycle (YMC) to study how transcriptionally regulated metabolism impacts cell growth program (Tu et al., 2005; Tu et al., 2007). In the YMC, cells are synchronized and exhibit respiratory oscillations under continuous, glucose-limited condition (Figure 1A). Thousands of transcripts and hundreds of metabolites are cycling as a function of the oxygen consumption oscillations, which divides the YMC into three phases: oxidative (OX), reductive building (RB) and reductive charging (RC). In the OX phase, respiration peaks and growth genes including ribosomal and amino acid biosynthetic genes are activated. Cell division occurs in the RB phase and cell cycle genes and mitochondrion genes are expressed. The RC phase is associated with minimal respiration and genes associated with stress/survival responses, glycolysis and fatty acid oxidation are up-regulated. OX and RB phases can be likened to the growth and proliferation phases based on the burst of respiration and protein synthesis, visualized cell division and the transcriptional and metabolic signatures. RC phase cells exhibit stationary and quiescent characteristics including increased cell density, accumulation of storage carbohydrates glycogen and trehalose and expression of stationary specific genes (Shi et al., 2010). Therefore, the YMC can be viewed as consecutive alternations of growth, proliferation and quiescent phases. By exploring the oscillations of transcription and metabolites, we may understand the mechanisms through which the cells switch the growth programs. Acetyl-CoA has been shown to drive the transition from RC/quiescence to OX/growth and RB/proliferation phase via the acetylation of histones at growth genes and G1 cyclin CLN3 (Cai et al., 2011; Shi and Tu, 2013). The role of promoting growth and proliferation by acetyl-CoA has further been shown in various mammalian systems including embryonic stem cells and cancer cells (Comerford et al., 2014; Moussaieff et al., 2015; Sutendra et al., 2014). However, how yeast cells accumulate acetyl-CoA in the quiescent phase to reach the critical level for growth is unclear.
Following this rationale, we developed a computational algorithm, DynaMO, to systematically predict which of the 175 yeast TFs that are linked to specific phases of the YMC (Kuang et al., 2017). Briefly, we grouped predicted binding sites for all TFs with similar dynamic binding patterns and for each pair of TF and binding pattern, we examined whether binding sites of the TF are enriched in the cluster/phase (Figure 1B). We identified 41 TFs that were specifically linked to each of the three phases in the YMC. We focused on two homologous TFs linked to the RC/quiescent phase, Msn2 and Msn4. They are known to regulate the general stress response in budding yeast, including starvation, thermal, osmotic and oxidative stresses (Estruch and Carlson, 1993; Estruch, 2000; Martínez-Pastor et al., 1996; Schmitt and McEntee, 1996). Deleting the two TFs caused a striking delay of transition from RC/quiescence to OX/growth phase in the YMC and slow accumulation of acetyl-CoA. Transcriptomic and cistromic analyses revealed that Msn2/4 control the expression of glycolytic pathway, which may promote the accumulation of acetyl-CoA and re-entry into growth.
Histone modification marks transcription regulatory regions and the intensity and dynamic nature of some modifications, such as H3K9ac correlates strongly with transcription activity. Therefore, we can predict TF binding sites by ‘foraging’ for TF consensus sequences associated with histone modification peaks and then predict the temporal activities at these binding sites in a highly dynamic process like the YMC. We can also identify important TFs associated with specific dynamic programs by enrichment analysis. The functionality has been incorporated in a computational tool, DynaMO (Kuang et al., 2017). We used the previously generated 16 time point H3K9ac ChIP-seq data across one round of the YMC (Kuang et al., 2014) to predict the binding activities of 175 yeast TFs. Three temporal binding patterns were captured, consistent with the three metabolic phases of the YMC (Figure 1B). 41 TFs were identified with their binding sites specifically enriched in one of the three phases, which were thus considered to be candidate regulators of those specific phases. Target genes of each TF were predicted and TFs were clustered by the similarity of targets between pairs of TFs (Figure 1C). GO term analysis of predicted targets genes of the 41 TFs reveals similar target specificities among TFs from the same clusters (Figure 1—figure supplement 1). In general, the functions of these candidate TFs were consistent with the transcriptional diagram in each phase (Kuang et al., 2017), such as ribosome biogenesis TFs Rap1 and Sfp1 enriched in cluster 1 (OX/growth), cell cycle TFs Mbp1 and Swi6 enriched in cluster 2 (RB/proliferation), and stress response TFs Msn2 and Msn4 enriched in cluster 3 (RC/quiescence) (Figure 1). Validation experiments performed by mutating individual TFs also show a significantly higher frequency of disrupted YMC oscillation (as measured by O2 consumption) in the candidate TFs (Arg80, Gcr1, Xbp1, Msn2, Msn4) than in control randomly selected TFs (Skn7, Arg81, Oaf1, Cin5, Ume6). (Various phenotypic defects were observed; the details can be found in Kuang et al., 2017) This indicates that we have a list of candidate TFs that help drive the YMC.
From the validation experiments, we observed a very unusual and specific ‘lengthened RC phase’ phenotype in the msn2Δ mutant (Figure 2A). The RC phase is longer than that in the WT strain and it gets longer after each cycle. msn2Δmsn4Δ showed a more severe defect than msn2Δ whereas msn4Δ was relatively normal, in support of partial functional redundancy in this instance (Estruch, 2000). This may be consistent with previous observations that some genes are completely turned off in the msn2Δ single mutant while other genes show reduced expression in the msn2Δ single mutant and are completely eliminated in the double mutant (Estruch, 2000). The timing of the OX/growth and RB/proliferation phases is normal in the mutants but the RC/quiescence phase is seemingly ever-lengthening, as though the cells await an ever-diminishing signal to proceed. Thus, the prolonged RC phase phenotype observed for MSN2 and MSN4 mutants can also be viewed as a delayed transition from RC to OX.
Two lines of evidence suggest that Msn2/4 function in the RC/quiescence phase. First, many stress response genes known to be the targets of Msn2/4 are activated in the RC phase (Gasch et al., 2000; Martínez-Pastor et al., 1996; Schmitt and McEntee, 1996). Second, Msn2/Msn4 motif sites were enriched in the predicted RC phase binding sites (Figure 1B, cluster 3). To test this hypothesis, we examined gene expression data for the YMC from the previous study (Kuang et al., 2014). MSN4 mRNA level is increased in the RC phase while MSN2 mRNA level is relatively constant (Figure 2B). This is in line with previous findings that MSN4 expression is activated by stress but MSN2 is constitutively expressed (Gasch et al., 2000). TFs function by binding to the promoters and enhancers of target genes so we performed time-course ChIP-seq of Msn2 and Msn4 across one YMC. Both TFs showed increased binding to the genome in the RC/quiescence phase regionally (Figure 2C) and genome-wide (Figure 2D). Msn2 and Msn4 shared a significant proportion of binding sites and target genes (Figure 2E, Figure 2—source data 1 and 2), supporting their largely redundant functions. Among the targets of Msn2 and Msn4, we observed a substantial proportion of genes expressed in the RC phase (Figure 2F), further supporting the hypothesis that Msn2/4 function in RC phase. Additionally, Msn2/4 target genes are enriched in classes of carbohydrate metabolism (Figure 2F), indicating that Msn2/4 may function through remodeling the metabolic pathways.
Next we investigated how Msn2 and Msn4 control the transition from RC/quiescent state to OX/growth state. We hypothesized that two types of signals may regulate this transition, stimulating signals such as nutrient and inhibitory signals such as stress and damage. Cai et al. (Cai et al., 2011) showed that adding acetate in the RC phase can directly induce an OX phase and they found that acetyl-CoA is the key metabolite that initiates cell growth and proliferation by promoting histone acetylation at growth genes. Acetyl-CoA is low at the beginning of RC phase and it is accumulated during the RC phase. When it reaches a threshold, it initiates the cell growth program. We asked whether acetate could induce an OX/growth phase in the RC/quiescence phase of a msn2Δ mutant. Although the RC phase is much longer in the mutant than that in the WT, acetate still stimulated the OX phase efficiently no matter when in the RC phase acetate was added (Figure 3A), consistent with the possibility that acetate or acetyl-CoA represented the limiting factor. Similar results were observed in the msn2Δmsn4Δ double mutant. The observations suggested that Msn2 and Msn4 might be involved in nutrient signaling, perhaps based on the accumulating intracellular acetyl-CoA level.
To test this hypothesis, we collected seven time points of WT cells across one YMC and the same seven time points of msn2Δmsn4Δ double mutant cells plus four additional time points in the RC phase to represent the relatively longer time of the RC phases (Figure 3B). We measured acetyl-CoA levels in WT and mutant cells and observed a dramatic delay of acetyl-CoA accumulation in the mutant cells (Figure 3C). Together, the results suggested that Msn2/4 control the transition from RC to OX phase by regulating the accumulation of acetyl-CoA.
We further asked how Msn2/4 regulate the acetyl-CoA level. Glycolysis and fatty acid oxidation are the major carbon metabolic pathways activated in the RC phase (Figure 1A) (Tu et al., 2005). DynaMO prediction analysis indicated that 20 of the 33 genes encoding glycolytic enzymes were predicted to be bound by Msn2/4 (Figure 3D) and the number of predicted binding sites was significantly higher than the number of sites from random sets of genes (p<1×10−5) (Figure 3—figure supplement 2E). We confirmed the observation by examining the ChIP-seq results. 27 out of the 33 genes were bound by Msn2/4 (Figure 3D and Figure 3—figure supplement 1A) and intriguingly, binding of Msn2/4 extends into the coding regions for some of the genes, which may suggest some unknown regulatory mechanisms. The coding regions of glycolytic genes are more conserved than the coding regions of other genes and the promoter regions of glycolytic genes in Saccaromyces strains (Figure 3—figure supplement 1B). The number of motif sites overlapping Msn2/4 peaks was significantly higher than the number of sites from random sets of 33 genes (p<1×10−7; Figure 3—figure supplement 2C,F). On the other hand, only 2 of 11 genes encoding enzymes in fatty acid oxidation were bound by Msn2/4 (Figure 3—figure supplement 2A). To further challenge the hypothesis that Msn2/4 regulate the level of acetyl-CoA through glycolysis, we examined the dynamics of mRNA levels of glycolytic genes in WT and msn2Δmsn4Δ double mutant by RT-qPCR. These genes were highly induced in the WT cells during the RC phase but were surprisingly dramatically reduced in the mutant (Figure 3E). Collectively, the results suggest that Msn2 and Msn4 control the transition from RC/quiescent state to OX/growth state by regulating the intracellular level of acetyl-CoA through glycolysis. Although we lack direct evidence for which intracellular pool of acetyl-CoA is regulated by Msn2/4, the pyruvate dehydrogenase (PDH) complex, which functions in mitochondria, was not targeted by Msn2/4, whereas multiple cytosolic enzymes were targeted by Msn2/4 (Figure 3D), suggesting that the nucleo-cytosolic pool of acetyl-CoA might be specifically regulated. It is also consistent with two other lines of previously obtained evidence: (1) that acetate, ethanol and acetaldehyde can induce the transition from RC to OX and (2) that histone acetylation is dramatically increased during such transitions (Cai et al., 2011). Interestingly, the oncogene MYC similarly regulates glycolytic genes in mammals (Dang, 2012).
Additionally, we asked how Msn2/4 affect the initiation of the cellular growth program from quiescence when we added 6 day stationary phase cells into fresh YPD medium. Glucose is at 2% in fresh YPD medium, which is much higher than the concentration in the continuous culture of the YMC. Interestingly, we still observed a delay of growth in the msn2Δmsn4Δ double mutant compared to the WT strain (Figure 3—figure supplement 3A). Inoculation in YP +2% galactose exhibited a bigger delay, probably because Msn2/4 also control the expression of genes in the galactose metabolism pathway (Figure 3—figure supplement 3B). The msn2Δmsn4Δ double mutant also showed a much lower saturation titer compared to the WT (Figure 3—figure supplement 3A). Deletion of MSN2 and MSN4 did not affect survival rate in glucose medium (Figure 3—figure supplement 3B) but decreased the size of stationary cells in glucose medium (Figure 3—figure supplement 3C).
To further characterize the functions of Msn2/4 in the YMC, particularly in the lengthy mutant RC/quiescence phase, we performed RNA-seq of 7 time points for WT cycling cells and 11 time points for msn2Δmsn4Δ double mutant cells across one cycle (Figure 3B). We first compared expression patterns of cycling genes in WT and mutant cells (Figure 4A and Figure 4—source data 1). The three clusters of gene expression patterns were still observed in the mutant cells. However, many RC genes were clearly down-regulated. Next, we identified 366 genes up-regulated and 426 genes down-regulated in mutant compared to WT cells at comparable time points. Whereas the 366 up-regulated genes were relatively equally distributed among OX, RB, RC and non-cycling groups, the majority (62.2%) of the 426 down-regulated genes were RC phase genes (Figure 4B and Figure 4—source data 2). For both up-regulated and down-regulated genes, the changes were primarily observed in the RC phase (Figure 4B). In the subsequent analyses, we focused on the down-regulated genes. Among these genes, we found significant enrichment of genes related to carbohydrate metabolism and response to various stresses (Figure 4B), and the GO terms were very similar to those detected by ChIP-seq analysis (Figure 2F). FAA1 and POX1, which are directly involved in fatty acid β-oxidation and bound by Msn2/4, did not show significant changes of mRNA level in the mutant cells (Figure 4—source data 2). This argues against the hypothesis that Msn2/4 promote acetyl-CoA production through fatty acid oxidation.
We next identified ‘core’ Msn2/4 target genes that were both bound by Msn2/4 and showed differential expression when comparing msn2Δmsn4Δ double mutant with WT cells. 112 genes, including 27 OX, 40 RB, 22 RC and 23 non-cycling genes, appear to be repressed by Msn2/4 in WT cells (Figure 4C; transcripts are up in mutant cells). 136 genes appear to be directly activated by Msn2/4 in WT cells (i.e., Down in mutant cells) and more than half of them (81 genes) were RC phase genes (Figure 4C). The timing of Msn2/4 binding is consistent with their activating function for target RC phase genes (Figure 4C). Carbohydrate metabolism genes and stress response genes are the two major classes in the core Msn2/4 targets (Figure 4D). As shown previously, many genes encoding glycolytic enzymes, such as HXK1, GLK1, ENO1, ENO2, PGK1, GPM1, TDH1 and TDH3 are core Msn2/4 targets. Heat-shock protein genes such as HSP12, HSP26, HSP78 and HSP82, oxidative stress response genes including CTT1, PRX1, SOD1, SOD2 are also core Msn2/4 targets. Certain metabolic genes are also already annotated as stress response genes. For example, TPS2 and GAC1, which are involved in the synthesis of trehalose and glycogen, two storage carbohydrates, are important for stress response. Interestingly, trehalose and glycogen have also been shown to be metabolized during the exit of quiescence to fuel the regrowth (Shi et al., 2010; Silljé et al., 1999). Msn2/4 may also facilitate these processes by activating NTH1, ATH1, GPH1, PGM1 and PGM2 (Figure 3D). Therefore, the regulatory network for Msn2/4 suggests parallel mechanisms for how yeast responds to limited nutrients. Stress response genes are activated to maintain survival by removing detrimental factors like mis-folded proteins and oxygen radicals and improving defenses. Carbohydrate metabolism genes are also activated so that cells can utilize the very limited nutrients in the environment for growth, as may often be the case in recovery from quiescence.
Given that Msn2/4 are involved in the transition from quiescence to growth, we next attempted to characterize the quiescent (RC) state of YMC by comparing it with multiple previously defined quiescent states, including starvation and limitation of one of the three essential nutrients, glucose, nitrogen or phosphate (Klosinska et al., 2011). We examined the transcription of cycling genes in the YMC and nutrient-scarce conditions (Figure 4—figure supplement 1A). The majority of the OX/growth genes, which function in biomass synthesis, are turned off when nutrients are scarce, as are the cell cycle genes which peak in the RB/proliferation phase. Interestingly, mitochondrial genes, which also peak in the RB phase, are relatively elevated upon nutrient scarcity. Genes encoding mitochondrial ribosome components are induced temporally at the beginning of glucose starvation or phosphate limitation but not under other conditions while genes encoding translation factors are activated modestly across all nutrient scarce conditions (Figure 4—figure supplement 1B). It will be interesting to explore the function of mitochondrial genes when nutrients are scarce. Not surprisingly, most RC genes are induced under nutrient scarce conditions too, reflecting the similarity between the RC phase and other quiescent states. We specifically examined genes encoding glycolytic enzymes, most of which are expressed in the RC phase (Figure 4—figure supplement 1C). The majority of these genes are also turned on under nutrient starvation or limitation conditions, suggesting a signal importance of in quiescent states. Details of the comparison are in the supplemental information.
In this study, we applied DynaMO analysis to the YMC, which led us to explore the function of two TFs, Msn2 and Msn4, in regulating the transition from a RC/quiescent state to an OX/growth state. Most previous studies on Msn2/4 were performed under various stress conditions such as starvation, heat shock, osmotic and oxidative stresses and used cell viability as readout (Estruch and Carlson, 1993; Estruch, 2000; Martínez-Pastor et al., 1996; Schmitt and McEntee, 1996). Although colony formation can be seen as a single round of regrowth, many details such as regrowth dynamics were ignored in such assays. In the YMC, in which nutrient is continuously limited, growth, proliferation and quiescence are temporally separated, providing a platform for mechanistic exploration of transitions between distinct cellular states. The unusual ‘lengthened RC phase’ phenotype in MSN2 and MSN4 mutants is consistent with a transcriptional program in the quiescent phase which could be critical for subsequent cell proliferation. Our analyses suggest that Msn2/4 modulate the cellular programs by changing the intracellular acetyl-CoA levels, probably by promoting glycolysis. Metabolism has strong effects on cell growth and proliferation. Metabolites such as acetyl-CoA have been shown to function not only as building blocks for the synthesis of fatty acids, amino acids and nucleotides, but also as signals for gene expression and enzymatic activities (Lyssiotis and Cantley, 2014). Acetyl-CoA has been proposed as a central metabolite that regulates cellular growth program from yeast to cancer cells (Cai et al., 2011; Comerford et al., 2014; Mashimo et al., 2014). Does a transcriptional program exist to control the acetyl-CoA level? Our analyses offer evidence that in budding yeast, transcription factors Msn2 and Msn4, well known as stress response factors, function as key transcription factors that activate genes encoding glycolytic enzymes. By controlling expression of these enzymes, Msn2/4 may support the generation of acetyl-CoA in preparation for cell growth and proliferation. This regulatory mode is not obvious when cells are grown in rich medium, probably because acetyl-CoA level is high enough to support continuous growth. But it becomes extremely important when cells are under nutrient-limited condition. Supposedly, nutrient limitation is sensed by TOR signaling and inhibition of TORC1 leads to nuclear translocation and activation of Msn2/4 and other stress response TFs (Beck and Hall, 1999; Loewith and Hall, 2011). That probably explains why many stress response TFs, such as Gis1, Mig1/2/3, and Msn2/4, that target metabolic enzymes were identified in the RC/quiescent phase of YMC by DynaMO (Figure 1). Therefore, this study seemingly uncover a critically important secondary mechanism for stress response – in the quiescent state, remodeling metabolic activities and activating stress responses and defenses pathways to maintain survival, and then enabling rapidly regrowth once nutrients are replenished (Ho and Gasch, 2015). It will be interesting to examine whether deletion of any of these TFs show similar effects on YMC and whether and how these TFs collaborate with each other in regulating the key metabolic genes and metabolites for recovery from quiescence.
Msn2/4-dependent glycolysis potentially provides fuels supporting cell growth under continuous nutrient limited conditions, similar to the situation in cancers, which are also associated with robust glycolysis and survive under nutrient limited conditions (Comerford et al., 2014). One speculative hypothesis is that a functional equivalent of Msn2/4 in mammalian cancer is MYC, which similarly targets all glycolytic enzymes and fosters tumor growth (Dang, 2012). It suggests that TF dependent activation of glycolysis to support cell proliferation under nutrient limited condition can be a prevalent biological motif.
Metabolic cycle experiments were performed as previously described (Kuang et al., 2014). A BioFlo 3000 from New Brunswick Scientific was used. YMC runs were operated at an agitation speed of 475 rpm (Bioflo 3000), an aeration rate of 1 L/min, a temperature of 30°C, and a pH of 3.4 in 1 L YMC medium. After the batch culture was saturated for at least 4 hr, fresh medium was added continuously at a dilution rate of ~0.09 ~0.1 h−1. three independent isolates of each mutant were tested for YMC and representative curves were presented in Figures 2 and 3.
YP + glucose or galactose medium contains 1% yeast extract, 2% bacto-peptone, 2% dextrose or galactose and 1.6 mM tryptophan. 200 µg/ml G418 or 100 µg/ml ClonNat or 300 µg/ml Hygromycin were supplemented in YPD for drug resistance selection. The YMC medium consists of 5 g/L (NH4)2SO4, 2 g/L KH2PO4, 0.5 g/L MgSO4•7H2O, 0.1 g/L CaCl2•2H2O, 0.02 g/L FeSO4•7H2O, 0.01 g/L ZnSO4•7H2O, 0.005 g/L CuSO4•5H2O, 0.001 g/L MnCl2•4H2O, 1 g/L yeast extract, 10 g/L glucose, 0.5 mL/L 70% H2SO4, and 0.5 mL/L Antifoam 204 (Sigma) (Kuang et al., 2014).
All strains were generated from the CEN.PK background and manipulated by standard budding yeast protocols:
|ZKY756||CEN.PK||MATa msn2Δ::hygMX, msn4Δ::KanMX6|
Gene knockout strains were generated by homologous recombination using PCR products containing a drug cassette (kanMX6 or hygMX) and 40 bp sequences flanking the target gene. Tagged-protein strains were generated similarly by integrating a cassette containing a protein tag and a drug resistance cassette at C terminus (Kuang et al., 2014). PCR products were transformed into a diploid strain and the heterozygous diploids were sporulated and dissected to select for haploids with drug resistance.
2 OD cycling cells were collected and flash frozen. RNA was extracted with the Qiagen RNeasy Mini kit (QIAGEN, 74104, Valencia, CA). First strand cDNA was synthesized by reverse-transcription using oligo(dT)20 primer from SuperScript III First-Strand Synthesis System (Invitrogen, 18080–051, Grand Island, NY). Fast SYBR Green Master Mix (Applied Biosystems, 4385612, Foster City, CA) was used for real-time PCR and experiments were done on the platform of StepOnePlus Real-Time PCR System (Applied Biosystems, 4385612, Foster City, CA). RNA-seq libraries were prepared in the New York University Genome Technology Center using Illumina Trueseq RNAseq v2 library kit (Illumina, San Diego, CA). PolyA beads were used for mRNA selection. 500 ng of RNA per sample was used as input and 12 cycles of PCR were run for amplification. Libraries were pooled together and sequencing was performed on Hiseq platform.
ChIP was performed as previously described (Kuang et al., 2014).~50 OD WT cycling cells per time point were collected for ChIP of Msn2 and Msn4. 6 time points were used to represent all three phases of YMC and they were relatively evenly distributed across the cycle. Antibodies are as following: Msn2 (y-300, sc-33631, RRID:AB_672215), Msn4 (yE-19, sc-15550, RRID:AB_672217). Validation is provided on the manufacturer’s website and the antibodies were further tested by western blots of WT, msn2Δ and msn4Δ lysates (Figure 2—figure supplement 1). 2.5 µg primary antibody was used per ChIP experiment. Briefly, cells were first fixed in 1% formaldehyde at 25°C for 15 min and quenched in 125 mM glycine at 25°C for 10 min. Cells were pelleted and washed twice with TBS buffer before freezing. The frozen pellet was resuspended in 0.5 ml ChIP lysis buffer (50 mM HEPES•KOH pH 7.5, 500 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% deoxycholate (DOC), 0.1% SDS, 1 mM PMSF, 5 µM pepstatin A, Roche protease inhibitor cocktail) and split into two tubes and lysed by bead beating. Lysate were combined and expanded into 1 ml and sonicated for 16 cycles (30 s on, 1 min off, high output) using a Bioruptor (Diagenode, Denville, NJ). The supernatant of the sonicated lysate was pre-cleared and incubated with 2.5 µg primary antibodies. After incubation overnight, 50 µl protein G magnetic beads (Invitrogen, Grand Island, NY, 10003D) were added and incubated for 1.5 hr at 4°C. Beads were washed twice with ChIP lysis buffer, twice with DOC buffer (10 mM Tris•Cl pH 8.0, 0.25 M LiCl, 0.5% deoxycholate, 0.5% NP-40, 1 mM EDTA) and twice with TE. 100 µl of TES buffer (TE pH8.0 with 1% SDS, 150 mM NaCl, and 5 mM dithiothreitol) was added to resuspend the beads at 65°C for 20 min. Reverse crosslinking was performed by incubation for 6 hr at 65°C. An equal volume of TE containing 1.25 mg/ml proteinase K and 0.4 mg/ml glycogen was added to the samples after reverse crosslinking and samples were incubated for 3 hr at 37°C. DNA samples were purified using ChIP DNA Clean and Concentrator (ZYMO RESEARCH, D5205, Irvine, CA). Library construction and sequencing were performed using KAPA Hyper Prep Kit (KAPABIOSYSTEMS, KK8502, Wilmington, MA). Briefly, DNA was end repaired and A-tailed. Barcoded adaptors were ligated and DNA was purified with Agencourt AMPure XP beads (Beckman Coulter, A63880, Indianapolis, IN) and amplified by 12–16 cycles. PCR products were gel-extracted and quantified on an Agilent Bioanalyzer. Sequencing was performed on Illumina Hiseq platform. Raw reads were mapped to the reference genome (sacCer2) by bowtie (Langmead et al., 2009) and peaks were visualized by the CisGenome Browser (Ji et al., 2008).
Acetyl-CoA was extracted with two methods. For the first method, Na azide was added (10 mM) and 5 OD cells were spun down and lysed in 200 µl of 10% perchloric acid by bead beating. The lysate was spun down and the supernatant was neutralized to pH 6–8 with 3 M K bicarbonate, with vortexing and cooling on ice for 5 min. K bicarbonate was spun down and supernatants were used. For the second method, 5 OD cells were resuspended in 4 mL Quenching Solution M (60% methanol, 10 mM Tricine pH7.4) and incubated at −40°C for 5 min. The cells were spin at 1000 g for 3 min at −10°C and washed once with Quenching Solution M. Pellets were resuspended in 1 mL Extraction Buffer M (75% ethanol, 0.5 mM Tricine pH 7.4) and incubated at 80°C for 3 min. Mix was cooled on ice for 5 min and spun down. Supernatants were dried down in speedvac and dissolved in 30 µl water. The concentration of acetyl-CoA was measured using Acetyl-Coenzyme A Assay Kit (Sigma, MAK039, St. Louis, MO).
Cells in log or stationary phase were counted microscopically in a hemocytometer. A fixed number of cells was plated on YPD plates and the colonies were counted. The survival rate was the number of colonies/number of cells plated. Three biological replicates were examined.
Log or stationary phase cells were harvested and washed with PBS twice. Forward scatter (FSC) was measured by BD Accuri C6 Flow Cytometer (BD Biosciences, Franklin Lakes, NJ) and used as the indicator of cell size. Two biological replicated were examined and plotted.
The DynaMO algorithm is described in detail in the manuscript (Kuang et al., 2017).
To evaluate the overlap between Msn2 and Msn4 binding sites (Figure 2), Msn2 and Msn4 peaks were combined across six time points using the ‘reduce’ function from the ‘GenomicRanges’ package. We then counted the numbers of overlapping Msn2 and Msn4 total or time-specific peaks and displayed in venn diagrams. To identify Msn2 and Msn4 target genes, we took the region of a gene from 700 bp upstream of the start codon to the stop codon and examined if the gene region overlapped a Msn2 or Msn4 peak. These genes were candidate targets. Candidate targets that are differentially expressed between WT and msn2msn4 mutant are declared as core target genes (see below). Gene ontology analysis was performed based on the SGD (RRID:SCR_004694) annotation file (http://www.yeastgenome.org/download-data/curation#.UQFqKhyLFXY) with Fisher exact test. P values were converted into FDR by p.adjust to adjust for multiple testing.
RNA-seq reads were mapped against SacCer2 genome using bowtie (Langmead et al., 2009). Read counts per gene was measured in R using the ‘CountOverlaps’ function. Differential expressed genes were identified using DESeq (Anders and Huber, 2010) package. We first compared pairs of WT mutant samples as follows: WT1-Mut1, WT2-Mut2, WT3-Mut3, WT4-Mut4, WT5-Mut5, WT6-Mut6, WT7-Mut7, WT4-Mut8, WT5-Mut9, WT6-Mut10, WT7-Mut11. In the first seven comparisons, we matched samples at the same absolute time in YMC. In the next four comparisons, we matched samples at the same relative time in the RC phase of YMC. For each pair, genes with adjusted p value less than 0.01 were identified as differentially expressed genes. We then compared the 5 WT samples in RC phase to the nine mutant samples in RC phase. Differentiated expressed genes from all above comparisons were grouped together as up- and down- regulated genes. Genes that were bound by Msn2/4 and showed differential expression in mutant cells were defined as ‘core’ targets. GO analysis was performed same as above.
To examine the enrichment of Msn2 and Msn4 motif sites in different processes, yeast genes were extended by 1000 bp upstream from the start codon. Glycolysis, fatty acid oxidation and galactose metabolism genes were selected based on SGD annotation and previous study (Tu et al., 2005). We examined total motif sites (T), predicted binding sites (Predicted) and motif sites overlapping TF ChIP-seq binding peaks. Numbers of total motif sites, predicted and ChIP-seq identified TF-bound motif sites were calculated in the gene regions of interested pathways and scrambled gene sets. To get scrambled gene sets, we randomly chose the same number of genes from the genome 100 times and examined if motif sites were located within the regions of these genes. To evaluate the distribution of Msn2 and Msn4 motif sites in human glycolytic genes, we first detected motif sites in hg19 genome using the yeast Msn2/4 motif sequences with CisGenome. We then identified the human homologs of yeast glycolysis genes using the BioMart tool from Ensembl and extended the gene regions by 20,000 bp upstream from the transcription start sites. Similarly, total motif sites or those overlapping with DNase I hypersensitive regions were examined in these genes and random selected genes. The human DNase-seq data for 57 cell types were downloaded from the ENCODE project through http://hgdownload.cse.ucsc.edu/goldenPath/hg19/encodeDCC/wgEncodeUwDnase. Then, the whole genome (chromosome Y excluded) was separated by 200 base pair bins (i.e. genomic loci). Bins with read counts larger than 10,000 in one or more cell types (abnormal loci) and bins with read counts smaller than 10 in all cell types (noisy loci) were excluded. After filtering, 1,108,603 genomic loci with unambiguous DNase-seq signal in at least one cell type were retained.
PhastCons scores for multiple alignments of the yeast strains to the Saccaromyces cerevisiae genome are downloaded from http://hgdownload.cse.ucsc.edu/goldenPath/sacCer2/phastCons7way/. Six strains were used for alignment, Saccharomyces paradoxus, Saccharomyces mikatae, Saccharomyces kudriavzevii, Saccharomyces bayanus, Saccharomyces castelli, and Saccharomyces kluyveri. Promoter regions were calculated as 0–500 bp upstream of start codons and ORF regions were regions from start codons to stop codons. Scores were averaged across each base pair and further averaged across interesting groups of genes.
Data of gene expression, nutrient-specific genes and genes required for specific nutrient starvation were downloaded from a previous study (Klosinska et al., 2011). Gene expression data in nutrient quiescence conditions and YMC were merged and displayed in a heat map. Mitochondrial genes in the RB phase were selected by intersecting RB phase genes and gene annotated as ‘Mitochondrion’ by SGD. Numbers of nutrient-specific genes and genes required for nutrient starvation intersected with OX, RB and RC phase genes were counted and enrichment levels were evaluated by Fisher exact tests. Adjusted p values were displayed in heatmap. Expression levels of nutrient-specific genes in YMC were examined by evaluating the max FPKMs across 16 time points of YMC. Msn2/4 target genes or randomly selected genes were intersected with nutrient-specific genes or genes required for nutrient starvation.
RNA-seq and ChIP-seq data have been deposited in the Gene Expression Omnibus database under accession code GSE72263 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE72263).
Two homologous zinc finger genes identified by multicopy suppression in a SNF1 protein kinase mutant of Saccharomyces cerevisiaeMolecular and Cellular Biology 13:3872–3881.https://doi.org/10.1128/MCB.13.7.3872
Genomic expression programs in the response of yeast cells to environmental changesMolecular Biology of the Cell 11:4241–4257.https://doi.org/10.1091/mbc.11.12.4241
"Sleeping beauty": quiescence in Saccharomyces cerevisiaeMicrobiology and Molecular Biology Reviews 68:187–206.https://doi.org/10.1128/MMBR.68.2.187-206.2004
An integrated software system for analyzing ChIP-chip and ChIP-seq dataNature Biotechnology 26:1293–1300.https://doi.org/10.1038/nbt.1505
Dynamic motif occupancy (DynaMO) analysis identifies transcription factors and their binding sites driving dynamic biological processesNucleic Acids Research, 10.1093/nar/gkx905.
High-temporal-resolution view of transcription and chromatin states across distinct metabolic states in budding yeastNature Structural & Molecular Biology 21:854–863.https://doi.org/10.1038/nsmb.2881
Trehalose is a key determinant of the quiescent metabolic state that fuels cell cycle progression upon return to growthMolecular Biology of the Cell 21:1982–1990.https://doi.org/10.1091/mbc.E10-01-0056
Naama BarkaiReviewing Editor; Weizmann Institute of Science, Israel
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]
Thank you for choosing to send your work, "Stress-response factors Msn2/4 control re-entry of quiescent cells into growth through glycolysis", for consideration at eLife. Your submission has been reviewed by three peer reviewers and the evaluation has been overseen by a Reviewing Editor and a Senior Editor.
This paper analyzes the role of the general stress-response master regulators, MSN2 and MSN4 in controlling the yeast metabolic cycle (YMC). Based on the results, the authors conclude that the two factors regulate key transition in this cycle by regulating the accumulation of acetyl-CoA.
As you will see below, all three reviewers appreciated your study in providing interesting new data concerning the regulation of the metabolic cycle. However, there was also a serious concern about your claim of direct relationship between Msn2/4 regulation and acetyl-CoA pool levels (and production of this metabolite). Two of the reviewers were more supportive of this conjecture, yet even these reviewers requested that you will further validate your claims by analyzing the consequences of MSN2,4 over-expression, either transiently or by using a constitutive system
The third reviewer, who is a metabolic expert, maintained that you should focus your study on the role of Msn2/4 in the shift between fermentation and respiration, and remove the claim about direct relationship between Msn2/4 regulation and acetyl-CoA pools. I would like to emphasize that since this reviewer is an expert in metabolism, we will not be accept the paper unless he/she accepts the arguments and claims presented.
In addition, all three reviewers requested that you better discuss previous studies, and in particular refer to previous work on carbohydrate metabolism during nutrient deprivation.
In this manuscript, Kuang and co-authors showed that the general stress responsive factors Msn2 and Msn4 play an important role in driving yeast metabolic cycle (YMC). They further revealed that it is through controlling the accumulation of acetyl-CoA and the same mechanism might generally contribute to cell growth regulation under prolonged nutrient limiting conditions.
I find this paper interesting in that (1) the effect of msn2/4 deletion to the progression of YMC is striking! (2) the proposed mechanism is based on careful genome-wide analysis and is convincing. (3) the potential relevance to cell growth regulation in general is intriguing. Therefore, I recommend the acceptance of this manuscript for publication in eLife provided the authors are able to address the major points below.
1) The conclusions are based primarily on loss of function mutant phenotypes (msn2/4 deletion) and the correlation analysis between metabolic enzyme expression or Msn2/4 genomic occupancy with YMC phases. It would significantly strengthen the paper if the authors can show some gain-of-function effects. For example, how does Msn2/4 overexpression affect YMC and the growth transition in general? This is especially important because most people in the field believe that activated Msn2/4 slow down cell growth.
2) Msn2 and Msn4 are activated by dephosphorylation and translocation from the cytoplasm to the nucleus, therefore nuclear localization can serve as a reporter for Msn2/4 activity. In the paper, the authors only monitored the expression changes during YMC. It would serve as the direct supportive evidence for Msn2/4 activation if the authors can observe nuclear localization of Msn2/4 specifically during the RC phase. And if so, what would be the authors' speculation on the mechanisms of Msn2/4 activation? This will provide some clues on why Msn2/4 are activated specifically during the RC phase.
3) The authors claim that Msn2/4 have a dual role in regulating carbohydrate metabolism genes and stress resistance genes. Under the conditions that they are focusing on (prolonged nutrient-limited condition), do the stress resistance genes also contribute to cell growth or they are unrelated by-products? An analysis or at least a discussion on this would be helpful.
This paper describes analysis of how the transcription factors Msn2 and Msn4 regulates expression of genes encoding enzymes of the central carbon metabolism in yeast, with focus on regulation during the metabolic cycle of yeast. These TFs are known to play a central role in stress regulation, and they regulate expression of a very large number of genes in yeast (>100). The authors claim that based on their findings it can be suggested that Msn2/4 control the transition from reductive charging phase to oxidative phase by regulating the accumulation of AcCoA. Regulation of shifts in yeasts metabolism is highly complex, and it tends to the naive to suggest such a statement, and it is by no means supported by the data. Thus, the authors focus their discussion on how Msn2/4 binds (and control expression) to genes of the glycolysis, but these genes are controlled by a large number of other TFs. Furthermore, Msn2/4 themselves control a large number of other genes, including several other TFs, which is not considered in their analysis. A main line of argument is measurements of AcCoA, but this metabolite is present in several cellular compartments – mitochondria, cytosol, peroxisomes and the nucleus (probably equilibrated with the cytosolic pool), and there is no free transport of AcCoA between these compartments. Measuring the total level and linking this to enzymes that are operating in different compartments is not correct (Figure 3 is wrong as ACS1 is forming acetyl-CoA in the cytosol and PDA1 is forming it in the mitochondria). The paper is interesting as the authors present a lot of data, but the biological conclusion they are trying to extract from these data is not supported by the data, and is likely wrong.
In Kuang et al., 2014 the authors examine the role of Msn2 and 4 to regulate glycolytic metabolism in order to promote acetyl-CoA and cell cycle entry. The authors utilize the very elegant YMC system to carefully monitor transitions between reductive and oxidative phases. They first introduce the DynaMO algorithm that identifies transcription factor motifs that influence periodic gene expression in the YMC. Using this method they identify Msn2 and 4, among others, as being involved in the reductive charging phase of the YMC. The authors perform ChIP experiments to demonstrate that Msn2 and 4 chromatin occupancy increases in the RC phase. They further demonstrate that deletion of Msn2/4 causes disruption of the respiration cycles in the YMC to one in which the reductive phase is dramatically lengthened. The authors reason that this lengthened cycle is due to a reduction in acetyl-CoA production. Indeed, ectopic addition of acetyl-CoA can rapidly induce OX phase transition (as has been previously shown by the Tu lab). Moreover, the authors show that RC phase gene expression closely resembles that of nutrient deprivation performed by other labs. Thus, data observed in the YMC is largely applicable to nutrient stress environments in general.
The findings presented in the manuscript are of high significance and broaden the knowledge of factors and mechanisms that influence cell survival. In particular, recent publications on acetyl-CoA demonstrate that this a key metabolite in promoting cell growth and proliferation. This data demonstrates that well-established transcription factors support the metabolism of this critical metabolite. The data in this manuscript is clear and the writing is concise. In general, the conclusions support the main findings of the paper.
The following are specific comments that can improve the manuscript:
1) The authors could do a better job in integrating current literature on the importance of carbohydrate stores, particularly trehalose and glycogen, in survival in nutrient stress conditions and re-entry into the cell cycle from quiescence.
2) Why did the authors focus on H3K9 acetylation, rather than gene expression, in the DynaMO prediction transcription factors that drive the YMC? (This may be detailed in the accompanying manuscript. However, some more details in this manuscript would be helpful to understand the outcome of the results.) Does the focus on H3K9ac provide a better assessment of gene expression that is responsive to acetyl-CoA?
3) The Introduction states that TFs were deleted to validate their role in the YMC. This is a bit vague. Can the authors state which ones? Is Msn2 and 4 unique in terms of its YMC disruption?
4) In the Introduction, the authors state that Msn2 and 4 are known to regulate stress genes. This statement should include relevant references, such as Gasch et al., 2000; Martínez-Pastor et al., 1996; Schmitt and McEntee, 1996).
5) The authors may also want to discuss upstream factors that regulate Msn2 function. In particular Tor signaling, which is influence by nutrient availability.
[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]
Thank you for submitting your work entitled "Stress-response factors Msn2/4 control re-entry of quiescent cells into growth through glycolysis" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Nan Hao (Reviewer #1); Ashby J Morrison (Reviewer #2).
Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.
We found the data on Msn2/4 regulating glycolysis convincing and believe it can be published as is. For eLife, however, we would have expected more solid insight into how one goes from altered glycolysis to changes in the regulatory molecule acetyl-CoA. This link, unfortunately, is not yet convincing enough and the revision did not help in solidifying this result.
In the revision, the authors did not examine the YMC progression or acetyl-CoA level (which are where the major debates centered on) in the overexpression strain, because "the YMC requires coordination of various parameters"(?). Instead, they just checked the regrowth rate and got a variety of inconsistent results. This is not surprising given that they transformed yeast with a 2 micro plasmid (pRS425) expressing Msn2 or Msn4. The variability of plasmid copy numbers is huge, and therefore the kind of colony variability is expected no matter which output phenotypes (growth rate or others) you will be looking at. A genomic integration (with pRS305) will work way better.
At this point, it seems my attempts to save the debated conclusion did not work out. However, given their striking findings summarized in my first review, I still support the publication of this manuscript, but would suggest the authors to tone down their conclusion and avoid claiming of direct causal or driving relationship.
The authors have addressed my previously stated concerns in the revised manuscript. I believe the manuscript is a significant advancement in our understanding of the Msn2/4 involvement in quiescence to growth transitions and energy metabolism.
The manuscript by Kuang et al., 2014 addresses transcriptional regulation during the yeast metabolic cycle. They attribute a key role of the transcription factors Msn2 and Msn4, well known for their role in the yeast stress response, to regulate the mRNA expression of glycolytic enzymes, during the metabolic cycle in the laboratory yeast strain CenPK (a popular model to synchronize yeast cells in their metabolic cycle). The authors conclude that transcriptional control of glycolytic enzymes though Msn2/Msn4 is causally linked to acetyl-CoA accumulation, triggering re-entry into cell growth.
In principle, the storyline of the manuscript has two parts. The first is the one in which the study authors identify Msn2 and Msn4 as transcriptional regulators of glycolysis during the recovery phase of the metabolic cycle. There may be some little bumps here and there as pointed out by the other reviewers, and these warrant to be addressed. But overall I find this part convincing, and in fact impressive.
Then there is the second part, the authors conclude from these transcriptional changes on how cells regulate metabolism via acetyl-CoA to drive cell growth. This part is not more than a (perhaps reasonable) speculation. It is intrinsically difficult to predict a change in the metabolome from transcriptional data, certainly one cannot conclude from the mRNA level of ~15 glycolytic enzymes on the acetyl-CoA concentration (that is no product of glycolysis) and its role in the recovery phase of the metabolic cycle.
In essence, the title of the manuscript should read something like 'Msn2 / Msn 4 regulate the expression of glycolytic enzymes during the recovery phase of the metabolic cycle'. I'm looking forward to discussing with the other reviewers whether this discovery on its own meets the bar set by eLife. I think it perhaps does; eLife did aim for very high in the past, but one needs to see this in the light of the recent and effectual situation. One might recommend to the authors however to tone down the manuscript to the part that they actually show.
[Editors’ note: following the rejection, the authors appealed. The appeal was assessed by the editors and revisions were requested prior to acceptance.]
Thanks for asking to re-consider the decision and sorry for the delay in getting back to you. The decision to reject the paper was based primarily on the worry, expressed by one of the reviewers and at least partly agreed to by the others, that, given the inability to establish the functional connection of MSN-related regulation to Acetyl-Co production, the significance of the paper may not reach the desired eLife standard.
In particular, please:
1) Tone-down the claims concerning acetyl-coA regulation. All reviewers were of the opinion that your study does not establish a functional connection from transcription regulation of glycolysis to the mechanism leading to altered acetyl-co levels.
2) Validate the antibodies used, as requested by the new reviewer.
3) Please see the other minor comments of the two new reviewers and perhaps emphasize some more the points that they emphasize as being mostly significant about your findings. These revisions are not essential but may improve your presentation.
[…] I do believe the authors present some important and novel findings worthy of reconsideration/publication. The data overall look pretty solid to me. It is not well-known that the stress response transcription factors Msn2/Msn4 also regulate glycolytic enzymes, and the evidence looks good that such genes may be induced by Msn2/4 to "prime" cells for rapid sugar utilization and re-entry into growth upon return to favorable nutrient conditions. I believe using the YMC system has enabled them to see this unanticipated relationship between Msn2/4 and glycolytic/carbohydrate metabolic enzymes, which may be obscured in traditional batch culture experiments. I also believe the RNA-seq and ChIP-seq could be high-quality datasets and will be of interest to those investigating both the overlapping and distinct roles for Msn2 and Msn4 in the regulation of stress-responsive gene transcription.
I would ask them to improve Figure 3—figure supplement 1, the legends/labels on the left side of the ChIP-seq views are hard to read, and also comment on the fact that there seems to be binding of these transcription factors throughout the entire coding region for some of these glycolytic genes, in addition to the promoter regions, which could be potentially interesting? Or artifactual? Also, they appear to use commercial antibodies from Santa Cruz for ChIP of Msn2/4, some of their own validation should be included to help demonstrate the specificity of these antibodies (e.g., a Western of WT vs. msn2∆, msn4∆ lysates).
I am sorry about the delay. I had to read this a couple times to appreciate what I think is the importance of this work, but let me emphasize that I have not been up to speed on Ben Tu's most recent work or that of Linda Breeden who has also worked in this area. Hence, there may be things that are more broadly known that were new to me here.
The importance of this paper to my mind is to broaden the appreciation of the roles of Msn2 and Msn4 from general stress resistance transcription factors to regulators of the metabolic cycle that have special roles in metabolism of cells that are coming out of quiescence that are not evident in standard growth conditions. The work is quantitative, incorporating multiple dimensions of analysis (though the magic sauce in DYNAMO was not described) and hence gives us a more comprehensive appreciation of the relevant targets of Msn2 and Msn4. As with any paper that provides a list of genes/proteins, there are some puzzles that are left unanswered such as why GIS1 is a target of this regulation. But on balance, the authors do a good job of drawing together logic behind the interactions described.
In searching for broader relevance, the authors speculate that Myc might be the human equivalent for the regulation of tumor growth since they assert that tumors are nutrient limited. That assertion may be true, but it certainly isn't widely known and raised my skepticism about this point, which is not that important to the results of the paper, but should be checked.
So why did I have to read it a couple times to get it? I think the authors could do a better job of emphasizing how different previous conditions used for the study of Msn2 and Msn4 are from the conditions encountered in the metabolic cycle. I would also advocate losing that damn acronyms for the three phases of the cycle as they are not commonly used and are disruptive to readers who are trying to understand this under explored region of metabolism.https://doi.org/10.7554/eLife.29938.024
- Zheng Kuang
- Jef D Boeke
- Zheng Kuang
- Hongkai Ji
- Zheng Kuang
- Hongkai Ji
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank B Tu and L Shi for suggestions on YMC. We thank Mordechai Choder for discussions on metabolic cycle and transcriptional control. We thank Adriana Heguy and her staff at the New York University School of Medicine Genome Technology Center for expert assistance with RNA-Seq and ChIP-Seq. This work was supported by the Technology Center for Networks and Pathways grant U54GM103520 and the research grant R01HG006841 and R01HG006282 from the NIH to JDB and HJ.
- Naama Barkai, Reviewing Editor, Weizmann Institute of Science, Israel
© 2017, Kuang et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.